MAPPING

STRATEGIC
KNOWLEDGE
MAPPING
STRATEGIC
KNOWLEDGE

Edited by

Anne Sigismund Huff and Mark Jenkins

SAGE Publications
London • Thousand Oaks • New Delhi
© Anne Sigismund Huff and Mark Jenkins 2002

First published 2002

Decision Explorer is a registered trademark of Banxia

Software Ltd

Apart from any fair dealing for the purposes of research or

private study, or criticism or review, as permitted under the
Copyright, Designs and Patents Act, 1988, this publication
may be reproduced, stored or transmitted in any form, or by
any means, only with the prior permission in writing of the
publishers, or in the case of reprographic reproduction, in
accordance with the terms of licences issued by the
Copyright Licensing Agency. Inquiries concerning
reproduction outside those terms should be sent to the
publishers.

SAGE Publications Ltd

6 Bonhill Street
London EC2A 4PU

SAGE Publications Inc

2455 Teller Road
Thousand Oaks, California 91320

SAGE Publications India Pvt Ltd

32, M-Block Market
Greater Kailash - I
New Delhi 110 048

British Library Cataloguing in Publication data

A catalogue record for this book is available from

the British Library

ISBN 0 7619 6948 9

ISBN 0 7619 6949 7(pbk)

Library of Congress Control Number available

Typeset by SIVA Math Setters, Chennai, India

Printed in Great Britain by The Cromwell Press Ltd,
Trowbridge, Wiltshire
CONTENTS

List of Tables and Figures vii

Notes on Contributors xi

Introduction – Anne Sigismund Huff and Mark Jenkins 1

Part I: Methods for Directly Discovering

Managerial Knowledge
1 Mapping Successful Organizational Routines 19
Véronique Ambrosini and Cliff Bowman
2 Linking Content to Process: How Mental Models
of the Customer Enhance Creative Strategy Processes 46
Olaf G. Rughase
3 Enabling Strategic Metaphor in Conversation: A Technique
of Cognitive Sculpting for Explicating Knowledge 63
John R. Doyle and David Sims

Part II: Inferring Managerial Knowledge

4 Gaining Understanding in a Complex Cause–Effect
Policy Domain 89
Roger I. Hall
5 Entrepreneurial Narratives and the Dominant Logics
of High-Growth Firms 112
Joseph Porac, Yuri Mishina and Timothy G. Pollock
6 Spatializing Knowledge: Placing the Knowledge Community
of Motor Sport Valley 137
Nich Henry and Steven Pinch

Part III: Theoretic and Methodological Issues

7 A Mapping Framework for Strategy Making 173
Colin Eden and Fran Ackermann
8 The Individual in the Strategy Process: Insights from
Behavioural Decision Research and Cognitive Mapping 196
Gerard P. Hodgkinson and A. John Maule
vi CONTENTS

9 Facilitating Group Cognitive Mapping of Core Competencies 220

Phyllis Johnson and Gerry Johnson
10 Using a Knowledge-Based System to Study Strategic Options 237
Dale W. Jasinski and Anne Sigismund Huff

Annotated Bibliography 268

Heidi M. Neck and Nardine Collier

Index 300
L I S T O F TA B L E S
AND FIGURES

TABLES

1 Part I: Methods for directly discovering managerial

knowledge 7
2 Part II: Inferring managerial knowledge 10
3 Part III: Theoretic and methodological issues 11
1.1 Initial list of factors contributing to the organization’s
success used during the causal mapping session 30
2.1 Customer satisfaction with competitors 51
4.1 A summary of the behavioural rules and matching
AI policy inference procedures used to mimic group
policy-making behaviour 93
4.2 Assignment equations and documented explanations
representing the Axelrod-type mapping of the sports
club system of Figure 4.1 96
4.3 Results of the AI policy inference procedure: Trial 1:
Policy-to-goal cause–effect paths found with the
elementary comprehension level of cause–effect beliefs 99
4.4 Results of the AI policy inference procedure: Trial 1:
Desired policies for improving the achievement of
club goals derived with the elementary comprehension
level of cause–effect beliefs 99
4.5 Results of the AI policy inference procedure: Trial 2:
Policy-to-goal cause–effect paths with the higher
comprehension level of cause–effect beliefs 103
4.6 Results of the AI policy inference procedure: Trial 2:
Desired policies for improving the achievement of
club goals derived with the higher comprehension
level of cause–effect beliefs 103
viii LIST OF TABLES AND FIGURES

4.7 Results of the AI policy inference procedure:

feedback loops buried in the cause–effect system map
that can create unintended outcomes 104
5.1 Expansion path examples 120
5.2 Sample growth-strategy phrases 122
5.3 Correlations between proximity matrix and cluster
structure matrices 124
5.4 Cluster membership 125
5.5 Sample quotes 127
10.1 Primary text families 245
10.2 Initial coding scheme: responses to interview questions 246
10.3 ATLAS/ti relations 249
10.4 Examples of option-striking considerations 254
10.5 Document statistics 255

FIGURES

1 A basic visual statement of connection 4

2 Sales growth performance surface, 1970 5
1.1 Researching tacit routines: a summary of
the proposed method 29
2.1 A cognitive model for organizational change 47
2.2 The strategy learning process 49
2.3 Customers’ causal cognitive map: pattern I 57
2.4 Customers’ causal cognitive map: pattern II 58
2.5 The strategy learning process: linking content to process 60
3.1 Metaphor types as a function of the number
of metaphors tokens met for one-shot metaphors
and clichéd metaphors 67
3.2 Illustrating that typical scientific conventions are
often based on metaphors 69
3.3 Example Sculpture A 73
3.4 Example Sculpture B 76
3.5 Example Sculpture C 77
 LIST OF TABLES AND FIGURES ix

4.1 An Axelrod-type mapping of the system of

cause–effect of the operations of a sports club 95
4.2 User responses requested by the AI program to
solve indeterminacy found in the opposing
cause–effect path correlations 101
5.1 Growth logic: cluster 1 129
5.2 Growth logic: cluster 2 129
5.3 Growth logic: cluster 3 130
5.4 Growth logic: cluster 4 131
5.5 Growth logic: cluster 5 132
6.1 Motor Sport Valley: constructors 142
6.2 Motor Sport Valley: suppliers 143
6.3 Career history 147
6.4 Career history 148
6.5 Career history 150
6.6 Joint career path 151
6.7 Joint career path 152
6.8 It pays to be well connected 153
6.9 March family tree 154
7.1 A sample of a cognitive map 177
7.2 A strategy group beginning the OMT process 179
7.3 An example of a group using the computer-supported
system 181
7.4 The hierarchy of maps 184
8.1 Hypothetical value function accounting for framing 202
8.2 Example of a cognitive map elicited from a participant
exposed to the negatively framed version of the problem
in Study 1, whose decision choice (‘commitment to the
overseas market’) is in accordance with the predictions
of prospect theory 210
8.3 Example of a cognitive map elicited from a participant
exposed to the negatively framed version of the problem
in Study 1, whose decision choice (‘renew marketing effort
in the domestic market’) runs contrary to the predictions
of prospect theory 211
x LIST OF TABLES AND FIGURES

8.4 Example of a cognitive map elicited from a participant

exposed to the positively framed version of the problem
in Study 1, whose decision choice (‘renew marketing
effort in the domestic market’) is in accordance with
the predicitions of prospect theory 212
8.5 Example of a cognitive map elicited from a participant
exposed to the positively framed version of the problem
in Study 1, whose decision choice (‘commitment to
the overseas market’) runs contrary to the predictions
of prospect theory 213
9.1 An extract of a core competence map 222
9.2 Example of a cognitive map with only level one 225
10.1 Hierarchical view of codes 251
10.2 Initial network view of competition 252
10.3 Initial analysis of the role of competition 253
10.4 LAN Switch decision: positive evaluation model 259
NOTES ON
CONTRIBUTORS

Fran Ackermann is a Professor in the Management Science Department

at Strathclyde University, Scotland. She is interested in the role that group
decision support systems and cause mapping can play in supporting the
development and implementation of strategy. She has worked extensively
with both public and private organizations from a consultancy/action
research basis, helping groups work together more effectively, develop a
sense of shared meaning and begin to negotiate towards action. She has
written widely in strategy, operational research and information systems
arenas. With Colin Eden she is the author of Making Strategy: The Journey
of Strategic Management (Sage, 1998).

Véronique Ambrosini is a Research Fellow in the Strategic Management

Group at Cranfield School of Management, UK, working in association
with Professor Cliff Bowman. She joined Cranfield in April 1994, prior
to which she was an assistant manager working for McDonald’s
Restaurants. She holds an MBA from the University of Birmingham and
studied for her first degree at the ISEG, a French Business School. Her
research interests include the resource-based theory of the firm, tacit
knowledge, organizational routines and competitive advantage. Her PhD
considered the role of routines in organizational performance. Her most
recent journal articles can be found in the Journal of Management Studies,
British Journal of Management, European Management Journal, and Journal of
Management Studies.

Michel Bougon is Professor of Strategy at Bryant College, Rhode Island,

USA, a private undergraduate and graduate AACSB-accredited school of
business. His research centres on ontology and cognition applied to the
cognitive mapping of sense making by individuals and by congregations
of individuals. His published work includes the conceptual unification of
organization theory and strategy. He also researches the mapping of
strategizing by organizations and individuals. He has created several
techniques to support his research, such as the ‘self-Q interviews’. He has
also developed several conceptual tools to extend cognitive mapping,
such as the ‘cryptic labels’ and the ‘congregate map’ of small groups and
xii NOTES ON CONTRIBUTORS

teams. His publications include a chapter in Anne Huff (ed.), Mapping

Strategic Thought (Wiley, 1990).

Cliff Bowman is Professor of Business Strategy at Cranfield School of

Management, UK. He has published several books on strategy. His
research interests include resource-based theory, corporate strategy and
strategy processes. Recent publications include ‘Strategy in Practice’
(Prentice Hall, 1998) and ‘Competitive and Corporate Strategy’ (Irwin
1997, with D. Faulkner).

Nardine Collier is a Research Assistant in the Strategic Management

Group at Cranfield School of Management, UK. Nardine joined
Cranfield in 1998 and is currently undertaking a part-time PhD on the
relationship between strategy development processes and the innovation
performance of organizations. Recently, with Frank Fishwick and Gerry
Johnson, she published ‘The processes of strategy development in the
public sector’, in G. Johnson and K. Scholes (eds.) ‘Exploring Public
Sector Strategy’, Prentice Hall (2000).

John R. Doyle is Distinguished Senior Research Fellow at Cardiff

Business School, University of Cardiff, Wales. He is currently researching
the meanings that people extract from (and attach to) visual forms. Such
forms may be 3-D, as in cognitive sculpting, or flat, as in advertising mate-
rial or brand logos. He is eclectic in research method, ranging from con-
versation and ‘mere’ observation of people’s attempts to ‘cognitively
sculpt’ their thoughts and feelings, through to controlled psychological
experimentation in which the font, colour, name and image appearing in
presentational material may be manipulated.

Colin Eden is Director of the University of Stratchclyde Graduate School

of Business, Scotland, and Professor of Management Science and Strategic
Management. He has written over 200 articles published in management,
operational research and strategy journals, and published seven books in
the field of management science, managerial and organizational cogni-
tion, group decision support and strategy making. His most recent book,
with Fran Ackermann, Making Strategy: The Journey of Strategic
Management (Sage, 1988), introduces qualitative modelling to a strategic
management audience. Recently he has been a consultant to the senior
management teams from many settings including the Scottish Natural
Heritage, Bombardier Inc., the Northern Ireland Office, Elsevier Science
and the Royal Ulster Constabulary. As well as a continuing interest in
strategic problem solving and strategy making, he is currently conducting
research into strategic risk. With Fran Ackermann, he is the developer of
the cognitive mapping software ‘Decision Explorer’ and the group sup-
port software ‘Group Explorer’, each of which is now used worldwide by
researchers and practitioners.
 NOTES ON CONTRIBUTORS xiii

Roger I. Hall is a Senior Scholar at the I.H. Asper School of Business,

University of Manitoba, Canada. His prime research interest is in explain-
ing the rise and fall of organizations from the way policies are formed
through the collective decision-making behavior of managers who seek to
control a dynamic complex system, such as a business, that is only partially
understood. He holds degrees in physics, production management and
business administration and studied under Joan Woodward (a pioneer in
the field of organization theory) and Arnold Tustin (an important figure in
the founding of the system dynamics discipline). He received his practical
training in electronics and control theory in the Royal Naval Electrical
School (H.M.S. Collingwood), and also as an apprentice, and later a produc-
tion manager, in the British electronics industry (Siemens Edison Swan). He
is a founding member of the System Dynamics Society, a Past President of
the Administrative Sciences Association of Canada and won First Prize in
the 1984 International Competition for the Most Original New Contribution
to the field of Organizational Analysis, sponsored by the College on
Organization of the Institute of Management Sciences. Hall’s research has
been published in Administrative Science Quarterly, Management Science,
System Dynamic Review and the Journal of Business Research.

Nick Henry was recently appointed to a Readership in Urban and

Regional Studies in the Centre for Urban and Regional Development
Studies (CURDS), University of Newcastle, UK. Previously, he was
Lecturer in Economic Geography at the University of Birmingham. He
has been involved in a number of Economic and Social Research Council-
funded projects on regional development in the advanced economies. His
most recent publications include an interdisciplinary edited collection on
Knowledge, Space, Economy (Routledge, 2000, with J. Bryson, P. Daniels and
J. Pollard) and The Economic Geography Reader (Wiley, 1999, with J. Bryson,
R. Martin and D. Keeble). He is the author and joint author of numerous
academic articles on the motor sport industry and In Pole Position: Motor
Sport Success in Britain and Its Lessons for the World’s Motor Industry
(Euromotor Reports, Ludvigsen Associates Ltd, 1999).

Gerard P. Hodgkinson is Associate Dean (Research) and Professor of

Organizational Behaviour and Strategic Management at Leeds University
Business School, The University of Leeds, UK. A Fellow of the British
Academy of Management and a fellow of the British Psychological
Society, his principal research interests centre on the psychological analy-
sis of strategic management processes, especially cognitive processes in
strategic decision making and business competition. His work has
appeared in a range of scholarly international journals including Strategic
Management Journal, Journal of Management Studies, Journal of Occupational
and Organizational Psychology, and Human Relations. He is co-author (with
Paul R. Sparrow) of The Competent Organization: A Psychological Analysis
of The Strategic Management Process (Open University Press, forthcoming).
xiv NOTES ON CONTRIBUTORS

He has also published a number of chapters in edited volumes, including:

‘Cognitive Processes in Strategic Management: Some Emerging Trends
and Future Directions, in N. Anderson, D.S. Ones, H.K. Sinangil and
C. Viswesvaran (Eds.), Handbook of Industrial, Work and Organizational
Psychology (Sage, 2001). He is Editor-in-Chief of the British Journal of
Management and a Consulting Editor of the Journal of Occupational
and Organizational Psychology. A practising Chartered Occupational
Psychologist, he has conducted numerous consultancy assignments
across a wide range of private and public sector organizations.

Anne Sigismund Huff is a Professor in Strategic Management at the

University of Colorado, Boulder, USA, with a joint appointment at
Cranfield School of Management in the UK. Her research interests focus
on strategic change, both as a dynamic process of interaction among firms
and as a cognitive process affected by the interaction of individuals over
time. Her publications include the book Mapping Strategic Thought (Wiley,
1990) and co-edited volumes of Advances in Strategic Management
(JAI Press, 1990–7). In 1998–9 she was President of the Academy of
Management, an international organization of over 12,000 scholars inter-
ested in management issues.

Dale W. Jasinski is an Associate Professor of Management at Quinnipiac

University in Hamden, Connecticut, USA. He came to Quinnipiac to
launch the first Bachelor and Masters degree programs in entrepreneur-
ship in the state. A successful entrepreneur before earning his PhD from
the University of Colorado at Boulder, USA, he teaches both entrepre-
neurship and strategic management at the undergraduate, graduate and
executive level. His research interests include how individual and collec-
tive cognitive processes affect firm growth, especially as they relate to
the identification and recognition of strategic options. Current research
papers include the impact of strategic vision on employee performance,
and an exploration of the role of options in promoting firm growth.

Mark Jenkins is Professor of Competitive Strategy and Director of the

DBA programme at Cranfield School of Management, UK. Prior to joining
Cranfield he worked for the Lex Service Group and Massey Ferguson
Tractors Ltd. His teaching focuses on the areas of competitive strategy,
knowledge management and innovation. His consulting activities reflect
these specializations, where he has worked throughout Europe, the USA
and in parts of the Far East and Middle East. In addition to his work at
Cranfield he has been a visiting Professor in Strategic Management at the
University of Colorado and has contributed to the MBA programme at
Warwick Business School. He is currently researching the role of know-
ledge and innovation in the development of Formula One motor sport. He
has published and presented a wide range of work in the areas of strategy
and marketing. He is a founding editor of the Journal of Marketing Practice,
 NOTES ON CONTRIBUTORS xv

a member of the editorial review board for the European Journal of

Marketing and is author of The Customer Centred Strategy (Pitman, 1997).

Gerry Johnson is Professor of Strategic Management at the University of

Strathclyde Graduate School of Business, Scotland. After graduating from
University College London, he worked for several years in management
positions in Unilever and Reed International before becoming a manage-
ment consultant. He taught at Aston University Management Centre,
where he obtained his PhD, and Manchester Business School before
joining Cranfield School of Management in 1988, where he remained until
taking up his current appointment in 2000. Professor Johnson is co-author
of Europe’s best-selling strategic management textbook, Exploring
Corporate Strategy (Prentice Hall, 5th edn, 1999) and co-editor of a book
series that develops themes in that text. He is also author of Strategic
Change and the Management Process, editor of Business Strategy and
Retailing, The Challenge of Strategic Management and Strategic Thinking,
author of numerous papers on strategic management and a member of
the editorial board of the Strategic Management Journal. His research work
is primarily concerned with processes of strategy development and
change in organizations. He also works extensively as a consultant at a
senior level on issues of strategy development and strategic change with
UK and international firms and public sector organisations.

Phyllis Johnson is a Research Fellow at the Graduate School of Business,

University of Strathclyde, Scotland. She gained her MSc and PhD from
Cranfield University. Her PhD addressed shared thinking in top man-
agement teams and its impact on the strategic decision-making process.
She is a chartered psychologist and works as a psychotherapist for HM
Naval Base Clyde. Her areas of interest are top team interaction and
emotional life in organizations, in particular in the corporate board. She
has presented her work at many international conferences and con-
tributed chapters to several key research texts in the field. She acts as a
reviewer for the British Journal of Management, the Journal of Management
Studies and the Journal of Occupational and Organizational Psychology.

A. John Maule is Senior Lecturer in Management Decision Making at

Leeds University Business School, UK. His research focuses on how
people make judgements and take decisions in the face of risk and uncer-
tainty, focusing, in particular, on how individuals model strategic choice
and the effects of time pressure and stress on decision making. He has
recently been assisting UK government departments to develop strategies
for communicating food and health risk to the public.

Yuri Mishina is a doctoral student in organizational behavior with a

minor in strategic management at the University of Illinois at Urbana-
Champaign, USA. His research interests lie in the intersections of strategy,
xvi NOTES ON CONTRIBUTORS

entrepreneurship and cognition. Current projects explore the social

construction of capabilities, symbolic protests within an institutional
framework, the tactics of challenge and delegitimation, and the effects of
resources, strategies and environments upon rates of growth.

Heidi M. Neck is currently pursuing a PhD in Strategic Management and

Entrepreneurship from the University of Colorado at Boulder, USA. Her
research focuses on young, high-growth firms, and her dissertation seeks
to link strategic alternatives and organizational slack to the growth of
high-technology IPOs. She has published in Frontiers of Entrepreneurship
Research, the Journal of Developmental Entrepreneurship and the Journal of
Leadership Studies. Neck has recently been appointed as an Assistant
Professor of Entrepreneurship at Babson College, which commences in
the Fall, 2001.

Steven Pinch is a Professor in Geography at the University of

Southampton, UK. His primary research interests are in the changing
geography of the welfare state and in the growth dynamics of agglomer-
ating industries. His main books are Cities and Services: The Geography of
Collective Consumption (Routledge and Kegan Paul, 1985), Worlds of
Welfare: Understanding the Changing Geographies of Social Welfare Provision
(Routledge, 1997) and (with Paul Knox) Urban Social Geography: An
Introduction (Prentice Hall, 4th edn, 2000).

Timothy G. Pollock is an Assistant Professor in the Management and

Human Resources Department at the University of Wisconsin-Madison,
USA. Pollock’s research focuses on the role that social and political
factors, such as reputation, social networks and power, play in shaping
executive compensation, corporate governance activities and market
transactions. He is also interested in exploring how the cognitive
processes of managers interact with firm resources to affect the perfor-
mance and survival of high-growth firms. Pollock’s research has been
published in Administrative Science Quarterly, Human Communication
Research, the Journal of Organizational Behavior, the Corporate Reputation
Review, the British Journal of Management and the Academy of Management
Executive.

Joseph F. Porac is a Professor of Organization and Management and the

Senior Associate Dean for Faculty and Research at the Goizueta Business
School at Emory University, USA. His research interests centre on the
cognitive bases of markets and organizations, and he is currently pursu-
ing research on minivans, motorcycles, the US paper industry and the use
of information technologies. His research has appeared in a variety of
organizational journals. Recent papers include ‘Industry categories and
the politics of the comparable firm in CEO compensation’, with James
Wade and Timothy Pollock (Administrative Science Quarterly, 44: 112–144),
 NOTES ON CONTRIBUTORS xvii

‘Sociocognitive dynamics in a product market’, with Jose Antonio Rosa,

Jelena Runser-Spanjol and Michael S. Saxon (Journal of Marketing, 63:
64–77), Cognition, Knowledge and Organizations (Stamford, CT: JAI Press,
1999) with Raghu Garud and ‘Strategy and cognition’ with Howard
Thomas, in A. Pettigrew, H. Thomas and R. Whittington (eds), Handbook
of Strategy and Management (Sage, 2001).

Olaf G. Rughase is a partner with Panlogos GmbH, Offenbach, Germany,

a strategy consulting firm. Prior to his consulting activities, he worked
for Deutsche Bank Germany, Dresdner Bank Brazil and several other
companies in Germany and the United States. He is completing his
doctoral studies at the University of Witten-Herdecke, Germany. His
research interests include strategic change, issue management and cogni-
tive contributions to strategy theory.

David Sims is Professor of Management Studies, Brunel University, UK,

and Head of the School of Business and Management. He has an acade-
mic background in operational research and organizational behaviour,
and has been a consultant to organizations in the oil, power, computer,
publishing, airline, hotel and engineering industries, as well as in the
public sector. His research interests are in management thinking and
learning, in particular in agenda shaping, problem construction and man-
agerial story telling. He is editorial advisor of the journal Management
Learning, and is author or co-author of some sixty books and articles
(including the textbook Organizing and Organization [Sage, 2000]) and a
further forty or so international conference papers, though he cannot
remember what they all say.
INTRODUCTION

This book explores the connection between two interesting and important
conversations in the study of strategic management. The first involves
questions of how knowledge is generated and managed in organizations.
A central issue in this literature has been how informal, social mechanisms
affect knowledge development and use. For example, ‘communities of prac-
tice’ (Brown and Duguid, 1991) have been identified as important social
contexts sheltering knowledge development. ‘Knowing’ exemplified in
action has been distinguished from more rhetorical knowledge (Cook and
Brown, 1999), and found to be more important to practice. Both observa-
tions expand on the idea that tacit understanding is different from, and
often bolsters, explicit knowledge (Nonaka, 1994; Nonaka and Takeuchi,
1995; Polanyi, 1966).
Interest in the informal and less articulated aspects of knowledge is
closely connected with the resource-based theory of the firm (Conner
and Prahalad, 1996; Grant, 1991; Penrose, 1995; Peteraf, 1993; Wernerfelt,
1984), which emphasizes that competitive advantage comes from
resources that are difficult to imitate (Barney, 1991; Lippman and Rumelt,
1982). Organizational routines that embed knowledge in shared practices
have been particularly salient in this theory, because they are much more
problematic for outsiders to understand, buy or copy (Nelson and
Winter, 1977).
While many people believe that organizational knowing is particularly
valuable, analysts are left with the difficulties of identifying these more
obscure knowledge assets, and strategists are further encumbered by
significant problems managing them. It seems obvious that both exploita-
tion of current capabilities, and exploration that might lead to the devel-
opment of new capabilities (March, 1991), are hindered if informal, tacit
knowing cannot be united with more explicit, formal understanding.
We feel this juncture invites use of cognitive mapping, a tool developed
in the study of managerial and organizational cognition (Walsh, 1995). A
‘map’ makes conceptual entities more visible. While early maps were
sometimes taken as direct reports on cognitive processes, later studies
adopted a more modest, and more appropriate, position: maps are inter-
mediate tools – they facilitate the discussion of cognitive processes that
can never be directly observed (Eden, Jones and Sims, 1979). Mapping
tools are thus potentially as attractive to consultants and practitioners as
2 INTRODUCTION

they are to academics. In fact, we regard mapping work as an especially

strong vehicle for moving between theory and practice – a frequently
desired but rarely achieved goal in professional fields.
Neck and Collier’s annotated bibliography at the end of this volume
shows that cognitive mapping research and practice has significantly
grown in size and scope since the publication of Colin Eden’s pioneering
work (e.g. 1988, 1992) and Huff’s Mapping Strategic Thought (1990). Interest
in mapping has been fuelled by advances in mapping methods, especially
computer-based instruments that facilitate data collection and analysis.
Concurrently, experience with cognitive mapping has highlighted the
limitations of remaining rigidly within a cognitive perspective. Over time,
work on managerial cognition and cognitive mapping has become more
closely integrated with research on affect and social interaction. This
move makes it more compatible with the knowledge-mapping tasks that
interest us in this volume of original readings.
The purpose of our book is three-fold. We want to:

• demonstrate the range of strategically relevant knowledge questions

that can be approached from a mapping perspective;
• illustrate the varied mapping tools that are now available; and
• address methodological and theoretic issues that are raised by efforts
to map strategic knowledge.

WHAT IS A MAP?

Drawing on common language uses of the term, we suggest that a ‘map’

• is a visual representation that,

• establishes a landscape, or domain,
• names the most important entities that exist within that domain, and
• simultaneously places them within two or more relationships.

A more complex map has two further characteristics. It

• facilitates images of being ‘within’ the established domain, and

• encourages mentally moving among entities.

The last two characteristics clarify the importance of location and per-
spective and promote consideration of change in entities and relation-
ships – two issues of increasing importance as strategic management
comes to be seen as an ongoing, dynamic process.
In this book we are interested in a broad family of ‘maps’, which rarely
(with the exception of Chapter 6, by Henry and Pinch) have a geographic
referent. Each of the characteristics in our definition therefore deserves
further explication.
 INTRODUCTION 3

Represents Visually

At heart, a map is a visual way of thinking that many people, though not
all, find compelling. The image can be a useful ‘transitional object’ (Eden
and Ackermann, 1998) that provides perspective for the individual. It can
be equally helpful in a group context by creating distance between ideas
and their initial sponsors. A map can also be a mnemonic, which serves to
resurface connections considered in the past. Equally, maps often suggest
new connections.
Those who are able to use a map as a thinking tool find that it can
organize and simplify ideas, even complex ones, in much the same way that
an urban subway map clarifies complex underground connections. The
visual impact of a ‘good’ simplifying map makes it a useful means of com-
municating knowledge, and a useful repository for collecting knowledge.
Alternatively, however, current knowledge can sometimes be more eas-
ily elaborated once its skeletal form is revealed in visual form. The map
then becomes useful ground for making the simple more complex and
thus moving beyond what was known in the past. The map’s visual form
can reveal ‘holes’ in current information. Its structure also facilitates
comparison with other maps that can be used as possible templates for
modifying content. The map then becomes a helpful record of emerging
knowledge.

Defines Domain

A map does not exist without establishing a ‘domain’. While interpretive

and related academic arguments insist that everything is defined by context,
a map highlights this critical aspect of knowledge. By its very existence, a
map ‘frames’ a particular landscape, and makes it more clearly the subject of
consideration. What is included in the frame is much easier to identify, and
challenge. Alternative frames are easier to imagine once a first alternative is
made more explicit. Maps even make the absent more obvious.

Names Distinct Entities

The names or labels found on a map also deserve explication. The map-
maker is typically well advised to use some well-known names as an ori-
enting device, even if the map is a private thinking tool. Just the word
‘Paris’, for example, establishes a great deal in a geographic map. Almost all
of us would expect other names in France, or the rest of Europe if the scale
is large, though we might be wrong if the map were of Texas, which also
incorporates a place called ‘Paris’. This example indicates how much the
way knowing is ‘parced’ (Weick, 1979) affects subsequent sense–making.
Categorization is thus a major map-making task (Porac and Thomas, 1994).
4 INTRODUCTION

Again, what is not named can be as important to the course of

subsequent thought as what is named. Map makers must make choices.
But as consumers of maps in many different settings (driving an auto-
mobile, using a subway, trying to find an unfamiliar constellation), we
understand that choices have to be made, and we can imagine high-
lighting other aspects of the terrain.

Establishes Multiple Relationships among Entities

Perhaps the most distinctive attribute of a map is that it establishes rela-

tionships. Size distinctions are a common but important way of relating
map entities. Thus, in geographic maps we find rivers, streams and
creeks; highways and secondary roads; major metropolitan areas, smaller
cities and towns.

Figure 1 A basic visual statement of connection

A list could make the same distinctions among entities, but the infor-
mation conveyed on a geographic map establishes relationships in a two-
dimensional coordinate system. The simple box and arrows diagram
shown in Figure 1, for example, is a basic visual statement of connection.
This map can be understood in several different ways. Drawing on the
Western convention of ‘reading’ from left to right, it suggests that A pre-
cedes B and C in some ordering scheme. Often a map of this sort is
intended to read that A ‘causes’ B and C, while B and C are not causally
linked (except through A). But perhaps the relationship is merely tempor-
al, or spatial. Able map makers understand these alternatives, and make
their intentions clear in legends and accompanying text.
Design, colour, sound cues, computer links and other options can be
used for clarity and to add additional categorization schemes. As a rela-
tively simple example, consider the map of competitor relationships in
the pharmaceutical industry in Figure 2. The ‘north–south’ coordinate
indicates relative expenditures on R&D. Companies of different size
(the entities mapped) are made comparable by dividing their actual expen-
ditures by total sales revenue, though raw data might be more infor-
mative for some purposes. On the east–west coordinate, expenditures
on advertizing are similarly normalized. The vertical dimension, sales
 High-range contours

Mid-range contours W-L

Low-range contours

SYN
ML B-M
UP
GLX SCH C-W

E-L

MRK AH
NOV PF
20.00 ABB
RHO S-K ACY JJ
ROS 0.22
0.20
10.00
0.18
0.16
0.14
0.12
0.15 0.10
0.08 Advertising as a % of Sales
R&D as a % of sales 0.06
0.10
0.04
0.02

Figure 2 Return on Sales (RoS) performance surface, 1970 (Huff and Huff, 2000:188)
6 INTRODUCTION

growth, makes it possible to assess possible performance implications of

these two strategic investments.

Facilitates Images

We are particularly interested in cognitive maps because we believe their

visual form makes it easier to consider the implications of ‘standing’ at
different points on the map. Though this is particularly obvious with a
map such as that shown in Figure 2, where some firms’ performance posi-
tions are clearly better than other competitors, even a simple diagram like
the one shown in Figure 1 can be used to speculate about the differences
between interpreting a knowledge domain from point A, or point C.

Suggests Options for Movement and Change

Visual maps of this sort thus invite speculation about strategic moves,
which is why the map in Figure 2 was created. The map clarified the dif-
ficulty some firms in the industry are likely to have in ‘moving to’ more
profitable positions. Other maps might be compared with this map. For
example, how would the relative attractiveness of firm positions change
if a different performance measure were used, or if firms were placed
on coordinates defined by other strategic decisions? Even maps that are
less explicitly geographic invite such speculation, as the authors of the
following chapters make clear.

OVERVIEW OF THE BOOK

In defining and discussing the characteristics of a map, we have sought

both to emphasize the open architecture of the mapping concept, and also
to establish that we do not see maps as a direct portrayal of individual or
collective knowledge. Maps are tools that we can simplify or enhance. In
this volume we have brought together a diverse collection of empirical
studies that illustrate three uses of knowledge-mapping tools.

Part I: Methods for Directly Discovering

Managerial Knowledge

The first three chapters of the book illustrate the range of approaches
available to surface managerial knowledge. The chapters are united by
the practical motive of understanding knowing in context, and by reliance
on direct interaction with managers to collect information, as shown in
Table 1.
 TABLE 1 Part I: Methods for directly discovering managerial knowledge
Cognitive, knowledge and
mapping questions Data source/Level of Method of analysis/ Contribution/Key Knowledge management
Strategic issue(s) addressed analysis Purpose of map finding(s) implications
1. Ambrosini & Bowman How top management How can routines be Uses individual inter- Self-Q technique (Bougon) Proposes that organization- Describes the ‘digging
teams can identify and identified and examined, views, requests for and other more open- level tacit knowledge is process’ required to surface
Mapping successful explore the tacit routines if they are not readily metaphors and stories, ended techniques used to best understood as tacit rarely discussed activities.
organizational routines. that create competitive discussed, go beyond and causal mapping in elicit primary information routines. Provides an Provides practical advice
advantage. standardized procedures, an eight-hour interactive for workshop discussion. empirical example of one for other researchers (e.g.
and involve multiple workshop with the top organization’s attempt to helping participants focus
individuals? management team of Decision Explorer software identify the routines that on what is being done
a British mutual used to display and store underlie its success. rather than what should be
organization. data on the causal sources done).
of advantage in current
practice.

2. Rughase How to facilitate How do our customers Consultants draw SENSOR® process uses Change can be facilitated Illustrates value of a
customer-focused see us? cognitive maps from maps derived from cus- by connecting customer story-based methodology
Linking content to
strategic change. stories told by customers. tomers to challenge top thinking with current for sharing knowledge
process: how mental How do we build managerial assumptions. between key shareholders.
management thinking.
models of the customer consensus for strategic Discussion in the TMT Provides an empirical
enhance creative stra- action? (Top Management Team) example from a German
tegy processes. develops understanding warehousing firm.
of the maps and
uses them for further
strategizing.

3. Doyle & Sims Accessing highly How to understand deep Individual and group Physical objects placed in Chapter describes the use Everyday, unexamined
Enabling metaphor in embedded knowledge. knowledge? interaction with relationships are used as of ‘cognitive sculpting’ as physical metaphors can be
conversation: a tech- physical objects. an intervention tool. a tool for developing a powerful basis for
nique of cognitive metaphoric thinking. surfacing undiscussed
sculpting. knowledge.
8 INTRODUCTION

Chapter 1, by Véronique Ambrosini and Cliff Bowman, proposes that

routines should be examined as a particularly important repository of
tacit organizational knowledge. Resource-based theory supports the view
that organizations benefit from knowledge that is dispersed among par-
ticipants, and sustained by their joint efforts, without being widely codi-
fied. With this definition as a premise, Ambrosini and Bowman
individually interviewed top-level managers in a financial services firm,
then worked with the team to identify the tacit routines that managers
thought could be causally linked to their company’s competitive advan-
tage. Consistent with resource-based theory, some effort was required to
‘tease out’ beliefs about competitive advantage in a team context. In addi-
tion, the routines managers initially identified as the basis for their com-
petitive advantage rarely survived sustained conversation. The simple
mapping process used supported the identification of underlying, more
sustainable sources of advantage from the managers’ perspective.
Olaf Rughase, in Chapter 2, raises a potential problem with the man-
agerial perspective, however. This chapter examines the disparity
between what managers think they know about their customers and their
customer’s experience of product offerings. In the consulting engagement
reported, an initial customer survey by a German warehousing firm
was difficult for managers to interpret. Rughase describes how the tacit
knowledge of customers was discovered in the stories they told about
their experiences with the firm. Though these stories almost always
seemed trivial to the storyteller, and rarely directly covered the issues
found in the survey, the consultants were able to aggregate and map the
data gathered into two quite different accounts. Tape recordings from the
interviews then were used to introduce map relationships to the man-
agement team. In the case study reported, the conversation supported
reinterpretation of the previously obscure survey data, and led to
re-segmentation of the firm’s market.
Chapter 3 moves even more deeply into tacit knowledge by describing
a method of engaging managers with a ‘toy box’ of physical objects. John
Doyle and David Sims call their technique ‘cognitive sculpting’. It is
designed to elicit understanding so embedded that it is not accessible via
interactive query or storytelling (the data-gathering methods used in
Chapters 1 and 2). The authors demonstrate how rearranging evocative
physical objects enables a group of individuals to surface and debate
more deep-rooted issues arising from strategic change programs as an
example of their technique.
This work is an innovative and distinctive addition to the repertoire of
mapping techniques. It pushes the definition of a ‘map’ by making it more
transitory and plastic, which fits many obervers’ definitions of organiza-
tional knowledge very well. At the same time, using physical, visual tools
makes tacit knowing more explicit, which potentially facilitates knowl-
edge management.
 INTRODUCTION 9

Part II: Inferring Managerial Knowledge

Part I suggests the wide range of mapping methodologies that might be

used to directly interact with managers. Part II provides an equally wide
repertoire for mapping organizational knowledge from less direct
sources, as summarized in Table 2.
The first approach, taken by a pioneer in the cognitive mapping field,
Roger Hall, is reported in Chapter 4. Hall’s retrospective analysis of deci-
sion making by managers of the Old Saturday Evening Post (1976) was an
early indicator of the potential impact of managerial cognition. In Chapter 4,
he more completely illustrates how a highly structured map – in this case
an AI program – can be developed to represent the decision-making
process of a group of senior managers. Hall defines the behavioural rules
applied by a specific top management team, the leaders of a sports
club. Although managers are consulted at ambiguous junctures, the
researcher is surfacing and reflecting on managerial knowledge, then test-
ing the model against subsequent behaviour.
Joseph Porac, who made important contributions to the field of organi-
zational and managerial cognition with his work on the Scottish knitwear
industry (1989), reports on a new study of entrepreneurial growth strate-
gies in Chapter 5. This work, which is carried out with co-authors Yuri
Mishina and Timothy Pollock, uses a series of propositions developed
from resource-based theory to draw out different growth strategies across
industries. The data were collected by the Kaufman Foundation. Five dif-
ferent growth strategies, each with multiple components, are identified
from the data. This is particularly interesting because it suggests that
knowledge crosses industry boundaries.
The question of transorganizational knowledge is more completely
explored in Chapter 6, by Nick Henry and Stephen Pinch. This chapter
draws on the principles of ‘untraded interdependencies’ between firms
(Storper and Salais, 1997). The study more specifically considers a ‘tech-
nological knowledge cluster’, the motor sports industry in England, and
demonstrates how mapping approaches can be used to identify knowl-
edge flows at the inter-organizational level. The value of considering dif-
fering levels of abstraction (individual, firm and cluster) is emphasized,
and, once again, this is an approach to knowledge mapping that depends
on researcher framing and interpretation.

Part III: Theoretic and Methodological Issues

The third section of our book has a theoretic and methodological focus, as
summarized in Table 3.
Colin Eden and Ackermann, the authors of Chapter 7, have extensive
experience in using mapping techniques. They also developed ‘Decision
 TABLE 2 Part II: Inferring managerial knowledge
Cognitive, knowledge and
Strategic issue(s) and mapping questions Data source/Level of Method of analysis/ Contribution/Key Knowledge management
theoretic sources addressed analysis Purpose of map finding(s) implications
4. Hall Complex causes and Can an AI program mimic Sport club executives The author developed Policies resulting from the AI programs provide a
consequences of policy the decision-making logic provided information the AI program based AI program corresponded structure for discussing the
Mimicking collective decisions are rarely dis- of a policy group? about political entities on behavioural rules to executive decision consequences of alternative
interpretation of cussed by managers, even and variables actively from the research making in the sports club causal assumptions, along
complex cause–effect though dynamic systems Can managers use such a influencing decision literature. at two points in time. with the implications of
relations using simple cannot be adequately program to understand making within their adopting more or less
behavioural rules explained without them. and work with greater organization. At points of Group policy making is complex causal arguments.
embodied in an AI complexity? path indeterminancy in mimicked by running
program. the model they provided trials of the AI program on
additional input on causal the causal map developed
beliefs within the group. from users’ initial inputs.

5. Porac, Mishina & Explain ‘cognitive proposi- How can individual Open-ended narrative Content analysis with Study finds evidence of Evidence of distinctive
Pollock tion’ from resource-based strategies and an data from 54 ATLAS/tiTM software used multiple growth strategies yet dispersed knowledge
theory that firm growth overarching ‘growth logic’ entrepreneurial founders to identify growth strate- within firms. Five clusters and beliefs within a
Entrepreneurial and direction is deter- be identified in collected by the Kaufman gies; cluster analysis used or ‘growth logics’ that broad entrepreneurial
narratives and the mined by the interaction entrepreneurial firms? Foundation were used as to compare multiple combine different strate- community.
dominant logic of of available resources, the data for firm-level strategy patterns across gies are identified across
high-growth firms. capabilities and manager- Can similar growth logics analysis. founder narratives. the data set.
ial mental models. be discerned across firms?

6. Henry & Pinch Empirical investigation What are the sources of Specialist publications; Geographic maps and Describes how Provides an example of
by two economic shared meanings and 50 in-depth interviews; ‘genealogical trees’ used knowledge is continually the ‘microfoundation’ of
Spatializing knowledge: geographers of the understandings that site tours in the to locate and track transformed as players knowledge production
placing the knowledge proposition articulated support world-wide British Motor Sport movement of suppliers, observe, and more within a greater
community of Motor by Storper and Salais economic dominance of Industry. designers and services importantly move socio-political/
Sport Valley. (1997) and others that producers found within among, specific work economic system.
‘untraded interdependen- a specific geographic locations.
cies’ underlie economic region?
activity.
 TABLE 3 Part III: Theoretic and methodological issues
Cognitive, knowledge and
Strategic issue(s) and mapping questions Data source/Level of Method of analysis/ Contribution/Key Knowledge management
theoretic sources addressed analysis Purpose of map finding(s) implications
7. Eden & Ackermann The creation of a develop- How to use causal Shared and negotiated Decision ExplorerTM Illustrates ‘teardrop’ Summary of an extensive
ment processual (socio- mapping technologies perspectives within a computer software, method for connecting project using interactive
Mapping strategy: a political) model of to elicit and subsequently group process. computer groupware or strategic goals and and computerized mapping
framework for strategy formulation. structure involvement pencil-and-paper actions. tools to help top managers
facilitating strategy from multiple participants procedures described as and other stakeholders map
making. in the strategy alternatives for eliciting their environment, specify
development process? and relating concepts. mutually agreed goals and
define strategy.

8. Hodgkinson & Maule Experimental cognitive Can work associating Laboratory study using Participants completed a Study using more Preliminary analysis
psychology and behavi- positive decision frames undergraduate subjects causal map from a complex and strategically suggests there may be
The individual in the oural decision theory has with risk-averse choice replicated using 52 senior comprehensive list of relevant investment statistically significant
strategy process: insights generated a large and be replicated with more managers from a banking variables relevant to a decisions replicated and differences among mapping
from behavioural consistent body of labora- managerial-relevant organization. case vignette, justifying expanded on previous methods that provide the
decision research and tory evidence on ‘bounded materials and analytic investment decisions. studies of biased basis for overcoming some
cognitive mapping. rationality’ (Simon, 1956). tools? decision making. decision biases.
Research on managerial Subjects asked to
cognition has worked complete a cognitive
more closely with managers mapping task prior to
to develop more complex receiving the prompt did
but disparate models. not show the same bias.
(Contd.)
 TABLE 3 (Contd.)
Cognitive, knowledge and
Strategic issue(s) and mapping questions Data source/Level of Method of analysis/ Contribution/Key Knowledge management
theoretic sources addressed analysis Purpose of map finding(s) implications
9. Johnson & Johnson How to develop insight How can expert Group cause mapping Group mapping Draws on a large-scale Questions validity of
of unit core competences intervention improve use process developed from procedures facilitated by corporate attempt to unfacilitated mapping
Facilitating group that might support of mapping tools? Ambrosini and consultants. identify sources of advan- methods.
cognitive mapping of corporate strategizing. Bowman. tage to discuss theoretic
corporation core and practical difficulties
competencies. with causal mapping.
10. Jaskinki & Huff Identification and How to describe mental Study of strategic ATLAS/tiTM software used Chapter uses case study Suggests that computerized
development of strategic models? options used to as a knowledge-based of an entrepreneurial mapping programmes have
Using a knowledge-
options in a volatile and illustrate researchers’ system to code data, build organization seeking new the capacity to track
based system to study How to document
uncertain high-tech mental models. connections among codes, options to illustrate the researcher’s mental model.
strategic options and change in mental models
environment. visually document the use of ATLAS/ti as a
researchers, mental over time? conceptual structure and methodological tool.
models.
generate attending
propositions.
 INTRODUCTION 13

Explorer’, a widely used computer mapping program we offer to readers

on the last page of this book. In this chapter, Eden and Ackermann pro-
vide a very practical description of their mapping technology. A particu-
larly interesting aspect of the chapter, from our point of view, is the
authors’ sensitivity to the political aspects of mapping exercises. Their
account also illustrates the importance of identifying issues that are
salient to strategy development in a specific context.
Chapter 8 by Gerard Hodgkinson and John Maule, is theoretically ori-
ented in a very different way. These authors champion the cause of exper-
imental research methods to elicit and develop maps, and provide an
innovative approach to testing the effect of different mapping approaches.
An initial study of undergraduate decision making is replicated with
senior managers from the banking sector. The results suggest why maps
need to be carefully framed as a basis for managerial understanding to
avoid biases. In fact, the authors suggest there are some situations in
which mapping may be dysfunctional.
Care in testing methodologies must be matched by care when working
with managers to elicit knowledge maps; this issue is tackled in Chapter 9,
by Phyllis Johnson and Gerry Johnson. Their report focuses on theoretical,
methodological and practical difficulties of applying mapping techniques
to elicit managers’ views on the sources of competitive advantage. The
chapter explores some of the particular issues raised by using maps to
improve and aggregate local knowledge and raises some important ques-
tions regarding the nature of cognition itself.
The need to reflect specifically on the researcher’s mental models is then
discussed in Chapter 10, by Dale Jasinski and Anne Huff. This chapter
uses coding decisions made in Jasinski’s dissertation on option theory to
discuss issues of validity and reliability in qualitative work. The chapter
is also interesting because it uses Atlas/ti., a knowledge-based computer
system, to assist in the analysis of textual data.
Finally, the volume ends with an annotated bibliography constructed by
Heidi Neck and Nardine Collier. This bibliographic reference illustrates
the breadth and depth of the mapping field and provides a useful starting
point for those scholars and practitioners who wish to explore further.
The bibliography is followed by an order form for a CD by Michel
Bougon. Bougon discusses the rapidity of change in the new economy
and uses Decision Explorer – a tool intitially developed by Eden and
Ackermann and described most fully in Chapters 1 and 7 – to illustrate
the nested logic of several well-known firms from web and news sources.
A free copy of the program is available on the CD.

SUMMARY

We believe the chapters in this book make a strong case for mapping
strategic knowledge. The case is further elaborated in six interrelated
14 INTRODUCTION

points. First, maps can connect and organize dispersed organizational knowledge.
Organizations generate complex knowledge but much of it is difficult
to capture and communicate. As organizations push to deliver higher-
quality goods more quickly and more cheaply, they have less time for the
necessary management of these resources. Yet knowledge from past
experience is critical for a variety of purposes, from training new employ-
ees to updating current practices. The global imperatives of corporations
mean they are dealing with more complex processes, market segments
and product offers across geographic and cultural boundaries, which
makes the need for the connection and linking of organizational knowl-
edge even greater.
Second, knowledge maps can facilitate organizational activities by simplifying
inevitably complex domains. One of the enduring mantras of strategic
management is the need to achieve clarity and logic in the central ratio-
nale of the strategy. With increasing complexity, this becomes increasingly
challenging. Methodologies that distil complexity help strategists main-
tain coherence whilst, at the same time, grounding their ideas in the oper-
ational reality of the organization.
Third, maps also have the capacity to represent knowledge at various levels of
abstraction. They therefore can be used to connect the idiosyncratic detail of
one context with other conditions that either cannot or need not be repre-
sented in such detail. Computer-based maps, in particular, can collapse or
hide some portions of a knowledge domain while other portions are exam-
ined, developed and transferred. Thus a map might show how a ‘simple’
strategy connects with more ‘complex’ organizational activities, and vice
versa. Links between macro and micro phenomena are often vague or miss-
ing from the strategic logic of an organization and maps, potentially, can
bridge this gap. Different levels of detail in a family of maps similarly can
make the complex more simple, and the simple more complex.
A fourth compelling reason to pay attention to maps and mapping tech-
nology is that maps can surface and organize concepts and relationships that are
normally taken for granted. This agenda is a dominant one in the chapters that
follow. One of the most widely discussed ideas in the knowledge field is the
difference between explicit and tacit knowledge (Nonaka and Takeuchi,
1995). Recognizing and surfacing tacit knowledge is important (though
competitively risky) for further strategizing, yet embeddedness makes
clarity problematic. Maps are a tangible but flexible tool that can help reveal
that which has not been articulated in the past. Source data can come
from the organization, but about competitors, partners, suppliers and
other actors.
Fifth, maps also have the ability to facilitate communication in group settings
and help aggregate opinions within a group. This is particularly important as
the boundaries around individual tasks, specific formal and informal
groups, and larger organizational entities become more permeable and
transient. The engagement of many different individuals in the strategic
task has become more important, and attention is turning not only to
 INTRODUCTION 15

coordinating rational and creative thought but also to the impact of issues
previously seen as more peripheral, such as politics, emotion and values.
Maps are useful tools for transferring knowledge and developing new
knowledge among individuals in all of these areas.
Knowledge can only be transferred if those intended to receive it
are able to understand the way in which it is coded or structured. More
subtle knowing also requires shared coding schemes. Maps provide a
basis for dialogue on these issues.
While other benefits of mapping technology might be explored, we will
content ourselves with a final, especially critical, sixth point – maps have
the capacity not only to catalogue but also to generate knowledge. Mapping
exercises facilitate ‘what if’ revisions in a knowledge domain. When they
record beliefs about causal or other relationships, they extend the insights
available from other sources, including quantitative spreadsheets (which
are, of course, explicit, first-order maps in themselves). By providing a
basis to make knowledge more explicit and more likely to be shared,
debated and revised, maps become a dynamic basis for knowledge
creation. As organizations face increasingly demanding environments, we
believe the mapping tools we present for generating knowledge may be
the greatest contribution of the book.

REFERENCES

Barney, J. (1991) ‘Firm resources and sustained competitive advantage’, Journal of

Management, 17: 99–120.
Brown. J.S. and Duguid, P. (1991) ‘Organizational learning and communities of
practice: towards a unified view of working, learning and innovation’,
Organization Science 2: 40–57. Notes: cited in Brown & Duguid, 1998.
Conner, K.R. and Prahalad, C.K. (1996) ‘A resource-based theory of the firm:
Knowledge versus opportunism’, Organization Science, 7: 477–501.
Cook, S.D.N, and Brown, J.S. (1999) ‘Bridging epistemologies: the generative
dance between organizational knowledge and organizational knowing’,
Organization Science 10: 381–400. Notes: cited in EGOS 2000 paper.
Eden, C. (1992) ‘On the nature of cognitive maps’, Journal of Management Studies,
29: 261–265.
Eden, C. (1988) ‘Cognitive mapping’, European Journal of Operational Research
36: 1–13.
Eden, C. and Ackermann, F. (1998) Making Strategy: The Journey of Strategic
Management. London: Sage.
Eden, C., Jones, S. and Sims, D. (1979) Thinking in Organizations. London:
Macmillan.
Grant, R.M. (1991) ‘The resource-based theory of competitive advantage:
Implications for strategy formulation’, California Management Review,
33: 114–135.
Hall, R.I. (1984) ‘The natural logic of management policy making: Its implication
for the survival of an organisation’, Management Science, 30: 905–927.
Huff, A.S. (ed.) (1990) Mapping Strategic Thought. Chichester: Wiley.
Huff, A.S. and Huff, J.O. (2000) When Firms Change Direction. Oxford: Oxford
University Press.
16 INTRODUCTION

Lippman, S.A. and Rumelt, R. (1982) ‘Uncertain imitability: An analysis of

interfirm differences in efficiency under competition’, Bell Journal of Economics,
13: 418–438.
March, J.G. (1991) ‘Exploration and exploitation in organizational learning’,
Organization Science, 2: 71–87.
Nelson, R.R. and Winter, S.G. (1977) ‘In search of useful theory of innovation’,
Research Policy, 6: 36–76.
Nonaka, I. (1994) ‘A dynamic theory of organizational knowledge creation’,
Organization Science, 5: 14–37.
Nonaka, I. and Takeuchi, H. (1995) The Knowledge-Creating Company. New York:
Oxford University Press.
Penrose, E. (1995) The Theory of the Growth of the Firm. 3rd edn. Oxford: Oxford
University Press.
Peteraf, M.A. (1993) ‘The cornerstones of competitive advantage: A resource-
based view’, Strategic Management Journal, 14: 179–191.
Polanyi, M. (1966) The Tacit Dimension, Garden City, NY: Doubleday.
Porac, J.F., Thomas, H. and Baden-Fuller, C. (1989) ‘Competitive groups as cogni-
tive communities: the case of Scottish knitwear manufacturers’, Journal of
Management Studies 26: 397–416.
Porac, J.F. and Thomas, H. (1994) ‘Cognitive categorization and subjective rivalry
among retailers in a small city’, Journal of Applied Psychology, 79: 54–66.
Senge, P.M. (1990) The Fifth Discipline: The Art and Practice of the Learning
Organization. New York: Doubleday.
Simon, H.A. (1956) ‘Rational choice and the structure of the environment’,
Psychological Review, 63: 129–138.
Storper, M. and Salais, R. (1997) Worlds of Production: The Action Frameworks of the
Economy. Cambridge, MA: Harvard University Press.
Walsh, J.P. (1995) ‘Managerial and organizational cognition: Notes from a trip
down memory lane’, Organization Science 6: 280–320.
Weick, K.E. (1979) The Social Psychology of Organizing. Reading, MA: Addison-
Wesley.
Wernerfelt, B. (1984) ‘A resource-based view of the firm’, Strategic Management
Journal, 5: 171–180.
METHODS FOR
D I R E C T LY
DISCOVERING
I
MANAGERIAL
KNOWLEDGE
1
MAPPING SUCCESSFUL
O R G A N I Z AT I O N A L
ROUTINES

Véronique Ambrosini and Cliff Bowman

ABSTRACT

It is widely acknowledged in the strategy field and in the literature on the

resource-based view of the firm in particular that tacit knowledge can be a
source of sustainable advantage. However, despite the attention that has been
dedicated to resource-based theory and to tacit knowledge at the conceptual
level, there is little empirical research on the topic. In this chapter we explore
how cause mapping can help bridge this gap.
We first briefly define tacit knowledge and argue that organizational-level
tacit knowledge is best understood as tacit routines. Secondly, we suggest that
causal mapping is a simple but powerful technique that can help managers
elicit the routines that are central to their organization’s success. We argue that
causal mapping is a digging process that can facilitate the elicitation of less
explicit causes for success, especially when coupled with techniques such as
metaphors and storytelling, which have been argued to help express the inex-
pressible. In the last part of the chapter we report on an empirical study we
carried out in a British mutual organization, in which we used causal mapping
to uncover tacit routines. We describe the study and set out our findings con-
centrating on one of the issues that emerged from the study: the apparent lack
of management attention towards activities that are critical in delivering advan-
tage. We conclude by arguing that researching tacit routines is valuable and
may even be critical for managers who want to understand better how their
organization works in order to help them sustain advantage.

INTRODUCTION

In the strategy literature, and the resource-based view of the firm liter-
ature in particular, tacit knowledge has been argued to be a source of
20 DIRECTLY DISCOVERING MANAGERIAL KNOWLEDGE

competitive advantage largely because it is difficult to express, it generates

causal ambiguity, it is practical, and it is context-specific (Grant, 1993;
Nonaka, 1991; Sobol and Lei, 1994; for a more complete discussion please
refer to Ambrosini and Bowman, 1998). This implies that tacit knowledge
possesses all the requirements that a ‘resource’ needs to have to be a
source of sustainable competitive advantage (Barney, 1991): it is valuable,
rare, imperfectly imitable and imperfectly substitutable.
The main interest in strategy resides in organizational tacit knowledge
rather than individual tacit knowledge. However, when tacit knowledge is
being discussed, very often the distinction between individual tacit knowl-
edge and organizational tacit knowledge is not made, or it is blurred. We
believe that this confusion may be reduced by employing the expression
‘tacit routines’ rather than ‘organizational-level’ tacit knowledge.
Several authors (Grant, 1996; Nelson and Winter, 1982; Spender, 1996)
have argued that organizational knowledge resides in routines. They
have suggested that organizations are social systems, that is, that they are
more than the sum of their parts and that these parts are interdependent
(Gharajedaghi and Ackoff, 1994). This means that an organization is
conceived as a set of interrelated routines.
This view can be traced back to Nelson and Winter (1982). According
to them, it is the establishment of routines that codifies and preserves
memories and knowledge. Thus specialized knowledge becomes embed-
ded in ongoing organizational activities. They suggest that organizations
remember through informal and formal routines: they ‘remember by
doing’. Nelson and Winter’s position is unambiguous; they argue that ‘to
view organizational memory as reducible to individual memories is to
overlook, or undervalue, the linking of those individual memories by
shared experiences in the past, experiences that have established the
extremely detailed and specified communication system that underlies
routines performance’ (1982: 105).
However, Grant explains that routines are often tacit: ‘organizational
routines involve a large component of tacit knowledge which implies lim-
its on the extent to which organization’s capabilities can be articulated’
(1991: 110). These types of routines are firm-specific, taken for granted,
path-dependent and complex. This is precisely what gives them value –
they are unique and difficult to imitate.
Accepting these arguments, we are actually concerned with studying
tacit routines when we study tacit knowledge at the organizational level.
Tacit routines are not standard operating procedures; they are not codified
or prescribed ways of doing things. They are activities involving more
than one organizational member that cannot readily be talked about
and that are happening without having been deliberately and explicitly
established. The expression ‘tacit routines’ can be seen as synonymous
with ‘organizational tacit knowledge’ as both phenomena are difficult to
verbalize. They are about doing, and they are context-specific.
 MAPPING SUCCESSFUL ORGANIZATIONAL ROUTINES 21

Tacit

Once we specify what we mean by ‘organizational-level knowledge’, it is

important that we clarify what we mean by ‘tacit’. From the literature
(Nonaka, 1991; Polanyi, 1962; Reber, 1989; Spender, 1996), it can be
implied that there are various degrees of tacitness. Hence we could have
the following:

1 Tacit routines that are totally unavailable. They are not accessible
because they are deeply ingrained in organizational members.
2 Tacit routines that could be accessed but that cannot be expressed
through the normal use of words. They might be articulated differ-
ently, for example through the use of metaphors and storytelling.
3 Tacit routines that are unarticulated but that could be articulated
readily if organizational members were simply asked the question:
how do you do that? In this case routines are tacit simply because
nobody ever asked the right question.

We concentrated in an exploratory study on the last two cases of tacitness

(2 and 3): knowledge not yet articulated but that could be. As indicated
earlier, the claim that tacit knowledge occupies a central role in the devel-
opment of competitive advantage is widespread. However, despite the
attention that has been dedicated to the issue, there is little empirical
research to support these theoretical assertions. One of the main reasons
why there have been very few attempts to research tacit routines empiri-
cally is that research instruments such as surveys and structured inter-
views are likely to be inappropriate insofar as individuals cannot be asked
to state what they cannot readily articulate. The main challenge that may
have to be faced is finding ways of expressing what could be but has not
been expressed (Nonaka, 1991).

INVESTIGATING TACIT ROUTINES:

METHODOLOGY 1

In the next few pages, we set out the methodology we used to research
tacit routines empirically. We begin by describing a research proposal
based on our review of the literature. Then we outline what we did and
what we learnt from it.

Cognitive Maps

We shall not explain here what cognitive maps are about as this has been
done extensively elsewhere (Huff, 1990). Let us just say that cognitive
22 DIRECTLY DISCOVERING MANAGERIAL KNOWLEDGE

maps are the representation of an individual’s personal knowledge, of an

individual’s own experience (Weick and Bougon, 1986), and they are
ways of representing individuals’ views of reality (Eden et al., 1981).
There are various types of cognitive maps (Huff, 1990). One of them is
the cause map or causal map: ‘a cause map is a form of cognitive map that
incorporates concepts tied together by causality relations’ (Weick and
Bougon, 1986: 106).
Causal maps are probably the most pertinent type of cognitive maps to
use in this research primarily because causal mapping allows us to focus
on action (Huff, 1990). When assessing its appropriateness in this context,
we should restate that tacit routines are about doing things, they are goal-
oriented. Jenkins noted that ‘causality provides a potentially higher level
of procedural knowledge (how it works, and how to do it) than other sets
of relationships’ (1995: 53), which implies that causal maps are likely to be
appropriate for studying tacit routines. Another reason for using such
maps is that the question ‘Are tacit routines a source of competitive
advantage?’ is by nature causal.
Furthermore, causal maps can be particularly useful for eliciting factors
that are context-dependent, as tacit routines are, because, ‘by virtue of the
time spent in a particular department or function, managers develop a
viewpoint that is consistent with the activities and goals of that depart-
ment or particular function’ (Walsh, 1988: 875). As Bougon et al. (1977)
highlight, this knowledge is stored in the minds of managers in the form
of cognitive maps, and cause maps in particular. In short, causal maps
reflect what is understood to be happening in an organization. One of the
main benefits of using cognitive maps is that they ‘place concepts in
relation to one another, ... they impose structure on vague situations’
(Weick and Bougon, 1986: 107). Cause maps are therefore a way of order-
ing and analysing something that is ‘fuzzy’. These maps are also useful in
eliciting tacit routines because they allow us to study issues at a micro-
level; they can also represent multiple explanations and consequences,
show interrelationships between factors, and potential dilemmas (Eden
and Ackermann, 1998).

Researching Tacit Routines: proposed Method

Similarly to Eden et al. (1992), in this study causal maps are not assumed
to be models of cognition as such. We adopt Eden’s view that ‘the only
reasonable claim that can be made of cognitive maps as an artefact ... is
that ... they may represent subjective data more meaningfully than other
models’ (1992: 262). They are simply used as a technique that would allow
us to elicit tacit routines.
There are different ways of building so-called ‘group’ or ‘collective maps’.
They can be an average of individual maps, a composite of individual
 MAPPING SUCCESSFUL ORGANIZATIONAL ROUTINES 23

maps (Weick and Bougon, 1986) or they can be derived from group
discussion (Nelson and Mathews, 1991). However, because a group map
may encompass more than the common content of the individual maps
(Langfield-Smith, 1992), we suggest that construction of the causal map
should be a group activity, a group discussion. By looking at the views of
others and reconsidering their own views, group members should be able
to reflect on their own and others’ behaviours in the light of the group-
level interaction and discussion. Moreover, a group map ‘as a visual inter-
active model, acts in the form of a ... transitional object that encourages
dialogues’ (Eden and Ackermann, 1998: 71). This approach could be a
convincing element when appealing for organizations to participate in
research, as ‘the process of constructing a consensus around causal factors
influencing the organization can be a useful diagnostic exercise’ for the
organization (Nelson and Mathews, 1991: 381).
The group discussion could take the format of focus groups. To dis-
cover which tacit routines are valuable and to follow the resource-based
view of the firm’s line of questioning ‘Is X a source of sustainable com-
petitive advantage?’, the discussion should focus on a more straight-
forward, if less precise, question like: ‘What makes your organization
successful?’ Respondents could be asked to identify concepts and
express the relationships between them. The relations could be estab-
lished not only on what the participants ‘know’ but also in their atti-
tudes, reactions, feelings, and so on. The advantage of a focus group
format is that ‘the inherent group dynamics tend to yield insights that
ordinarily are not obtainable from individual interviews’ (Schiffman and
Kanuk, 1991: 52).

Eliciting Tacit Routines: The Process

The causal mapping system should be powerful in revealing tacit

routines because the process is about continuously asking the respon-
dents to reflect on their behaviours, on what they do. They would ordi-
narily not do so. During the mapping, they are encouraged to explain
what they do and in that process they reveal to themselves aspects of
their behaviour that up to that point were tacit (in terms of degree 2 and
3 above). The in-depth probing that allows the map to develop taps the
routines that go unspoken in the organization. While mapping, the
researcher should be able to catch instances where individuals say
things such as ‘Oh yes, that’s right’, ‘Aha! I hadn’t realized that’. The
‘Oh, yes’ experience suggests that something that was tacit has just been
made explicit by an individual. This shows that managers are gaining
insights from the session and are becoming aware of their, up-to-now
tacit, routines.
24 DIRECTLY DISCOVERING MANAGERIAL KNOWLEDGE

starting the causal map

As mentioned earlier, we start with the broad question: what causes
success in your organization? The first few answers are likely to be well-
known, general causes for that success. We describe how we move toward
a more specific map in the next few paragraphs.
There are several ways of constructing a map. For example, Walsh
(1988) uses a predefined list, Axelrod (1976) derives constructs from texts,
Markóczy and Goldberg (1995) draw material from interviews. As we are
looking at taken-for-granted tacit knowledge, prescribed checklists or
structured interviews are likely to be inappropriate. There is a need for
‘creat[ing] a situation in which cognitive maps can emerge as fully as pos-
sible with a minimum of influence’ (Bougon, 1983: 182), that is, we should
‘avoid suggesting anything to the individual that might become part of an
eventual cognitive map’ (Cossette and Audet, 1992: 332). So, we believe
that the map should ideally be built without predetermined constructs.
The constructs should be established during the mapping session itself.
However, we are aware that starting a causal mapping session from
scratch could be time consuming and hence we believe that preliminary
interviews are a way to elicit constructs that can be used as a basis for the
maps. From the literature we concluded that two different methods could
be appropriate to elicit those constructs. They could be uncovered
through self-Q interviews, a technique that limits the influence of the
researcher (Bougon, 1983), or through semi-structured interviews, with
storytelling. We designed a study in which we interviewed each partici-
pant once. Half of the group was interviewed using one technique and
half using the other. The results of both methods are included in what
follows. The motive for using two separate methods is to ‘make sure that
the ground is well covered’ (Markóczy and Goldberg, 1995: 310). The
interviews to elicit the constructs that are to be used to start the causal
map are to be carried out on an individual basis. These interviews should
provide us with an opportunity to establish rapport with the participants.
This is paramount as ‘close rapport with respondents opens doors to
more informed research’ (Fontana and Frey, 1994: 367).

self-Q
Self-Q is a non-directive mapping technique developed by Bougon (1983).
It is a self-interviewing technique that draws on the respondent’s account
of his/her beliefs to generate constructs. The reason why tacit routines
could be elicited through self-Q questioning is best given by this quote:

in self-Q interviews, participants essentially interview themselves. The first

key idea is that participants are the experts on the personal knowledge that
guides their social behaviour. The second key idea is that participants formu-
late their questions on the basis of their own personal knowledge ... and on
 MAPPING SUCCESSFUL ORGANIZATIONAL ROUTINES 25

the basis of their own thinking . . . about the situation they are questioning.
(Bougon et al., 1989: 328–329)

Bougon et al. also assert that with the self-interviewing technique ‘the
events, objects, and concepts [the participants] use to express their ques-
tions ... reveal their tacit and explicit knowledge’ (1989: 329).
As mentioned previously, this technique could be applied to elicit the
constructs used to start building the group cause map. The technique
involves ‘people ask[ing] themselves questions about whatever topic is
being mapped and the concepts are then extracted from the questions’
(Weick and Bougon, 1986:115). Here the questioning is to be focused on
the respondents’ views about what they do that makes the organization
successful. Practically,

the Self-Q technique uses a framing statement and a ... diagram. The framing
statement is read by participants and is intended to set the stage for self-
questioning and to provide the subject with enough information to begin the
self-questioning process. . . . The diagram is intended to be used by partici-
pants to cue themselves to ask additional questions. (Sheetz et al., 1994: 37)

This technique is appealing because it lowers the participants’ resistance

to respond: ‘people are not practiced in defending against questions that
they ask themselves and over which they have control. Furthermore,
since the person is asking questions rather than making assertions, the
questions themselves seem harmless’ (Weick and Bougon, 1986: 115).
Another benefit of this technique is that the researcher does not hinder the
production of constructs by his/her lack of knowledge of the organization
under observation. The fact that ‘often a researcher ... does not really
know enough to ask the right questions’ (Bougon et al., 1989: 353) is not
an issue when using the self-questioning technique.

semi-structured interviews
The second method we used to start uncovering constructs is semi-
structured interviews. The interviews are semi-structured in the sense
that their purpose and structure is pre-determined: for example, we
wanted to know what the causes of success were, and we wanted partici-
pants to give examples, to tell stories about this success.
We believe that it would be useful to encourage the interviewees to tell
stories because ‘stories are one of the many forms of implicit communica-
tion used in organizational contexts’ (Martin, 1982: 257). People ‘manage
the collective memory of the organization through storytelling’ (Boje,
1991b: 9). Martin explains that stories are used in organizations to ‘explain
“how things are done around here”’ (1982: 256). It is also an appropriate
device for studying tacit routines because people frame their experience in
stories (Wilkins and Thompson, 1991). Moreover, ‘stories are contextually
26 DIRECTLY DISCOVERING MANAGERIAL KNOWLEDGE

embedded’ (Boje, 1991a: 109), they ‘can reflect the complex social web
within which work takes place’ (Brown and Dugruid, 1991: 44). In other
words, through storytelling, participants can express what is done in the
organization, and hence some tacit routines may be uncovered. This
means that through stories people say more than they would normally:
‘stories permit researchers to examine perceptions that are often filtered,
denied, or not in the subjects’ consciousness during traditional inter-
views’ (Hansen and Kahnweiler, 1993: 1394).
The stories could be generated through interviews, with the partici-
pants being asked to tell maybe two stories, one positive and one nega-
tive, concerning what has in the past caused organizational failure and
organizational success. (This is based on the critical incident technique
developed by Flanagan [1954].) As recommended by Ford and Wood
(1992), the interviews should, if possible, take place in the participants’
organization as the familiar surroundings can serve as cues.

metaphors
Martin (1982) argues that both stories and metaphors can serve to trans-
mit tacit knowledge. Metaphors are interesting as a way of eliciting tacit
routines, for a variety of reasons. They may, for example, elicit tacit
routines because ‘metaphorical language gives tacit knowledge voice’
(Munby, 1986: 198). They ... ‘communicate meaning when no explicit
language is available, especially in regard to complex ambiguous experi-
ence’ (Srivastava and Barrett, 1988: 60).
Among the reasons behind the claim that metaphors can help express
what is not easily articulable is that metaphors can generate new mean-
ing. They can ‘render vague and abstract ideas concrete’ (Sackmann, 1989:
482). Because they allow different ways of thinking, people may be able
to explain complex organizational phenomena metaphorically (Tsoukas,
1991). Metaphors can ‘transmit an entire story visually using one image’
(Sackmann, 1989: 468). This idea of image is central in understanding the
argument concerning the articulation of tacit knowledge through
metaphors. Because metaphors are vivid images, they may substitute for
a large number of words (Sackmann, 1989) and they are ‘useful in coping
with a large amount of data’ (Hill and Levenhagen, 1995: 1068). Images
also allow us to speak about process because they are not discrete. This
matters because tacit routines are capabilities; they are a process (they are
about how to do things).
The procedural aspect of tacit routines is a factor in why it is difficult to
communicate routines through words. This is explained by Ortony:

language [is a] discrete symbols system . . . words partition experiences.

[However] experience does not arrive in little discrete packets, but flows,
leading us imperceptibly from one state to another. Thus the task we have
to perform in communication is to convey what is usually some kind of
 MAPPING SUCCESSFUL ORGANIZATIONAL ROUTINES 27

continuum by using discrete symbols. It would not be surprising if discrete

symbol systems were incapable of literally capturing every conceivable
aspect of an object, event or experience that one might wish to describe. ...
This deficiency is filled by metaphor. (1975: 46)

Metaphors are a means of capturing the continuous flow of experience,

hence they can be a means of capturing tacit knowledge. ‘They allow the
transfer of concrete bands of experience whereas literal discourse segments
experiences’ (Tsoukas, 1991: 581). ‘One can say through metaphor what
cannot be said in discrete, literal terms, especially when words are not
available or do not exist’ (Srivastava and Barrett, 1988: 37).
All the above suggests that it is worth considering metaphors when
attempting to elicit tacit routines. However, capturing metaphors may not
be trouble-free. One of the problems is obviously to make sure that a
metaphor is not used when more direct language could easily be avail-
able. It is necessary to ensure that the metaphor is employed where no
appropriate words are accessible. One way of making sure that a meta-
phor is used in such a manner is by asking the individual to express his/her
metaphor in another way. If another metaphor is used rather than literal,
usual, basic terms, then one may be inclined to think that the use of meta-
phor is appropriate. The second problem is that not every individual may
be ready or able to use metaphors during a discussion.

the mapping process

The interviews we have just described, whether based on self-Q or semi-
structured interviews, allowed us to know the participants better, raise
their confidence in the process and of course elicit a few ‘success’ factors.
This being done, the tacit-skill elicitation process can begin in earnest. The
map can start with ‘success’ and previously revealed factors. The goal of
the mapping is now to find the reasons for the success. The questions that
can help the participants to do so could be, for instance: How does that
happen? What causes that? Who is involved? What influences that? By
answering these questions, the participants can start eliciting more par-
ticular, precise reasons for success. The best image to describe the map-
ping process is an ‘onion’ metaphor. By peeling away layer after layer of
the reasons for success, participants get to the less explicit causes for
success. These are causes they could not have readily surfaced without
prompting and probing. It is important throughout the session to insist
that we are only interested in what they are currently doing, even if it is
perceived to be trivial or irrelevant, and not in what they believe they
should be doing, though this is an understandable concern of organiza-
tional members.
As the process moves from the explicit reasons for success to the point
where participants have to reflect on what they do not usually think about,
the flow of factors is likely to slow down. In this case we encourage
28 DIRECTLY DISCOVERING MANAGERIAL KNOWLEDGE

participants to think of examples of how they perform the factor they

have just elicited, or tell stories about the factor, or use metaphors to
explain how the factor works. In each case, as we explained previously,
answering questions helps express the inexpressible.
Throughout the mapping process participants are encouraged to speak
about what they do. The factors should be written as actions, if possible,
rather than abstract statements, because tacit routines are about doing
and not verbalizing what is done. It is not about ‘knowing about’ or reci-
procally ‘not knowing about’. Asking participants to use verbs and ‘I’ is
one useful way of making sure they are discussing what they are doing.
The map will stop when the respondents, despite being pressed for
more examples and encouraged to say more, cannot reveal more factors.
This was how we envisaged using causal mapping to elicit tacit knowl-
edge. Let us now examine how we applied this method and what we
learned from the experience.

Empirically Investigating Tacit Routines

It was with the definition and methodology just established in mind that
we began to elicit the tacit routines of a UK-based organization. Our aim
was to elicit the tacit routines that generate success in the organization,
that is, we wanted to explore whether theoretical claims that tacit routines
are a source of sustainable competitive advantage could be substantiated
empirically.
The organization we approached can be said to have achieved com-
petitive success. It is a UK-based mutual organization (called case A in
what follows). The full top management team (TMT) participated in the
study. We invited the TMT to take part in a one-day causal mapping
session aimed at uncovering the routines that make them successful. We
asked them to concentrate only on what they were currently doing and to
focus on what causes recent or current success.

Case A: Methodology

Figure 1.1 summarizes the methodology used in this study.

In order to start the process of uncovering the routines that mattered
for case A’s success, we first interviewed each TMT member about
what they believed the reasons for their organization’s success were. We
interviewed two members using Bougon’s self-Q technique. For the other
three members we used semi-structured interviews, including storytelling.
After analysis of these interviews, a list of factors that contribute
to the success of the organization was drawn up. The list contained
63 elements. This was too many to explore in a one-day workshop. We
 MAPPING SUCCESSFUL ORGANIZATIONAL ROUTINES 29

1. Preliminary interviews about what causes success

success in the organization to elicit constructs interviews
A C
to start the map (A, B and C).
B
2. Set up the map with the preliminary constructs
as starting points aa
a What causes A to
happen?
3. Begin the mapping process with questions
such as: How does (a) happen?
What causes that? b
How does it happen?
Could you tell us an
4. If the flow of constructs stops, ask questions example about (b)?
c
such as:
Could you give us an example of how Could you narrate a
that happened? d story about the
Could you tell us a story? occurrence of (c)?

Figure 1.1 Researching tacit routines: a summary of the proposed method

(adapted from Ambrosini and Bowman, 2001)

therefore decided to only examine the nine most frequently cited

factors (see Table 1.1).
Four weeks later we invited the whole top team to take part in a one-
day causal mapping session aimed at uncovering the routines that make
them successful. At the start of the workshop we provided the top team
with the nine primary causes of success that were elicited during the
interviews. The purpose of the workshop was to ask the team to uncover
the factors that caused these to happen – basically, to uncover what they
do that leads to success. We were only interested in what they were cur-
rently doing, even if it was perceived to be trivial or irrelevant. We
insisted that we were not interested in what they believed they should be
doing. In accord with the resource-based view of the firm, we also asked
them to concentrate on what they thought was unique to their organiza-
tion. Finally we asked them to establish links between the various routines
they identified.
In order to keep the mapping process going, we kept asking questions
of the participants when they were struggling to find reasons for success.
The questions we asked included, for instance: How does that happen?
What causes that? Who is involved? What influences that? Could you
please give us an example? The questions were kept simple and to the
point. We insisted that anything could (and should) be said, even it is was
perceived as ‘trivial’, as it was likely that what makes the difference
between firms was likely to reside not in obvious, generic factors but in
detail, in the idiosyncratic ways of doing things in the organization.
To help us with this process, we used the Decision ExplorerTM software
developed by Colin Eden, Fran Ackermann and their colleagues at the
30 DIRECTLY DISCOVERING MANAGERIAL KNOWLEDGE

TABLE 1.1 Initial list of factors contributing to the organization’s success

used during the casual mapping session
Big clients joining
CEO
Charity market
Customer focus
Ethos
Investment returns
Services offered
Staff commitment
Trustees

University of Strathclyde. The causal map being built was displayed on a

screen so that participants could see it evolving. Because of the amount of
data we collected, we could not manage to work on one whole map and
therefore we worked on only a few factors at a time. At the end of the
workshop each individual was given all the maps that were generated
(see Appendix).
Because whether a routine is tacit or not can only be assessed by the
individuals concerned, we needed more information from participants.
For instance, if during the group session it was surfaced that the routine
‘speaking to customers without recording the message’ is a source of
success, we needed to ask whether the managers explicitly ‘knew’ that
before the session. This implies that the maps generated should be coded.
Hence at the end of the causal map session, each TMT member was
requested to individually code the maps using three categories:

A routines that are well known to you;

B routines that are known about but that you find difficult to deal with;
and
C routines that were tacit to you up to the group session.

Discussion

case A: results
We did not receive all the coded maps from all TMT members. Because of
the amount of data, the TMT concentrated on only a few maps (see
Appendix for a couple of examples), and only three sets of maps were
returned to us: the Chief Executive’s, the Human Resources Director’s
and the Business Development Director’s.
Before carrying out any analysis all routines that were not positive (in
other words factors that were a hindrance to the organization’s success
or factors that were about what it wishes to do rather than what it does)
were removed.
As this study was totally exploratory and to our knowledge there is
no empirical research that has looked at the link between tacit routines
 MAPPING SUCCESSFUL ORGANIZATIONAL ROUTINES 31

and competitive advantage, we had very few hypotheses. Our view was
that a large majority of the factors to be coded would be found as
belonging to category A (well-known) and that only one or two of the
factors to be coded would be found as belonging to category C (tacit
before the group session).
The coding was different from one member to another. Two of them
coded the majority of the factors in category A (53.7 per cent and
54.4 per cent). One (the CEO) coded all the factors as A or B, and another
member coded 89.5 per cent of the factors in these categories. The results
of the third member were significantly different. Indeed, he categorized
only 28.6 per cent of the factors as belonging to category A and altogether
only 37.3 per cent as belonging to either A or B or both. In other words, for
him a large number of the factors were tacit (40.5 per cent vs 10.5 per cent
and 0 per cent for the others). The divergence in responses was to a certain
extent surprising, as the TMT is small and has been together for a while.
This would indicate that perceptions of tacitness are personal and not
organizational. As each TMT member has a different functional back-
ground and occupation, the differences could be functionally biased
(Walsh, 1988).
The analysis of the coding also revealed that a fair portion of the factors
that were perceived to cause success in the organization were categorized
as B, ‘difficult to deal with’. The Chief Executive of the organization classi-
fied 46 per cent of the causes for success in this category, the Business
Development Director 35 per cent and the Human Resources Manager
31 per cent. We explained to the participants that by this category we
mean the things they know are going on in the organization and that
they recognize as being important, but which tended to be left alone, or
unmanaged. This could be because no one really knows how to manage
or change the activity. Examples of statements that fell under the B category
include: ‘We can go on gut-feel; we don’t take a commercial view’; ‘We
accept different modes of behaviour in different areas of the organization’;
‘We write customers a ten-page letter instead of a two-page one’; ‘The
way managers and team leaders behave’.

comment
We would like to add that from the comments we received from the TMT
it seems that the experience has been useful for them as a way of uncover-
ing elements to their success factors that they do not know how to deal
with, how to manage. They believed that the experience was an ‘eye
opener’ as far as realizing that part of why their organization is success-
ful is beyond their immediate management. If anything, this study
revealed that researching tacit routines is indeed valuable and that it may
be critical for a management team to understand better how their organi-
zation works in order to help the organization sustain advantage.
32 DIRECTLY DISCOVERING MANAGERIAL KNOWLEDGE

methodological lessons
Case A was our first attempt at eliciting tacit routines. We drew a number
of lessons from the experience.

pre-causal mapping
• Self-Q is difficult to use. It is not a natural process, and we found that
even if the participants understood the idea, they had problems asking
questions spontaneously. We never reached any kind of flow in the
interviews and very few factors were generated in comparison with
the semi-structured interviews. This interview technique is also time
consuming, as the participants need to be briefed in-depth. Consi-
dering these problems, we decided to abandon self-Q in our subse-
quent studies.
• The interviews are a useful ‘ice-breaker’ for the subsequent workshop.
They allow the participants to know the researchers and the
researchers to know better the business the participants operate in.
They also allow the researchers to introduce the concept of tacit
knowledge, routines and causal mapping. However, we believe that
they are not essential. We think we could have managed without them.
This would have meant spending more time before the mapping
session to discuss the exercise and talk about the obvious reasons why
the business is successful, and then start the exercise as such, rather
than going straight into data from the interviews.

causal mapping
• Trust and confidence matter a lot during the mapping. Participants
must be relaxed and feel confident and comfortable expressing
themselves. They need to feel that they won’t be made to feel a fool.
Otherwise they are unlikely to air their intuitive understandings and
feelings and more likely to concentrate on tangible aspects of their
organization. Hence the exercise could be a waste of time, as only
well-known and well-defined aspects of the organization would be
discussed.
• As explained earlier, metaphors seem to be a powerful tool to capture
tacit knowledge, but their capture is in practice troublesome. We dis-
covered that they were very difficult indeed to capture ‘live’. In the
flow of the discussion during the mapping exercise they are difficult to
spot, and if spotted they are difficult to analyse. We believe that it is
still worth trying to capture them as they can be a powerful means of
expression. The researcher needs to take note of them and come back
to metaphoric language for further explanation. This can sometimes be
done straightaway.
• This first study made us realize that while we need to keep the causal
mapping session informal, insofar as we do not want to disturb the
 MAPPING SUCCESSFUL ORGANIZATIONAL ROUTINES 33

flow of ideas, we also need to be directive. It is especially important to

insist on individuals dealing with what they currently do. We also need
to ensure that, if we want to explore success, participants stay focused
on success. This is important. There was a tendency for some TMT
members, first, to mention what they thought should be done in the
organization rather than actual current or past behaviour, and, secondly,
to delve into what was not working in the organization and what
should be done to remedy that.
• We would like to stress that asking participants to think of particular
examples, to tell a story, is a powerful trigger. Many constructs were
revealed this way. Asking for examples pushes participants to explain
what is really happening, it forces them to give details, and also it trig-
gers other thoughts and stories and hence keeps ‘the ball rolling’.
• We do not think, in retrospect, that the Decision Explorer software
should be used ‘live’. It is a technical aid, and we found in our case that
it slowed the session, distracted the audience and researchers, and also
that we were too dependent on the technology working perfectly. Any
hitches were disruptive. We suggest instead the use of ‘Post-it’ notes
and large sheets of paper. Notes are easy to use, they can be easily
moved and should be reliable whatever the situation. We would advo-
cate the use Decision Explorer for analysis and presentation of the
maps. The maps can be keyed in after the session.
• We all found the mapping process very tiring. It went on for almost
eight hours, and the participants were quite exhausted at the end.
They did not expect this, as they thought that the exercise was going
to be quite straightforward.

coding
• Another conclusion we drew from Case A is that it is important to keep
the coding individual. From one manager to another there was a wide
difference in the number of factors coded and which factors were
coded as tacit. This highlights that tacitness is personal (Polanyi, 1962;
Ravetz, 1971) and not always shared across a team of managers, even
if they have been together for a long time.
• The previous point begs the question of whether the factors coded
as tacit were really tacit. It seems likely that some of the factors were
simply not known by respondents. This suggests that we may need to
add a new category: ‘routines you did not know about’. Not knowing
that something is done is not the same as not realizing that what is done
matters. We are returning here to the initial definition: tacit knowledge
is about doing and not verbalizing what is done; it is not about ‘know-
ing about’ or reciprocally ‘not knowing about’ (Nonaka, 1991). Hence
we need a coding category that would allow us to deal with ignorance,
rather than tacitness.
34 DIRECTLY DISCOVERING MANAGERIAL KNOWLEDGE

• At the start of the analysis we encountered one problem we had not

fully anticipated: a fair number of factors were not coded. This is
something we need to address. There are various reasons why all the
factors might not have been coded:
– The coding categories proposed may not have been adequate.
(This might mean we have to ask the respondent to tell us why
he or she chose not to code a particular factor.)
– Some of the participants simply refused to code the factors
because they did not want to admit that they did not understand
everything that is happening in the organization. To admit that
may have been uncomfortable. This appears to be what happened
in the case of one TMT member who refused to give the maps
back and who commented that he had learnt nothing from the
day, despite his displaying great enthusiasm to us at the end of
the causal mapping session itself about the value of the exercise.
Participants are justifiably afraid of losing some of their status.
We now believe we have to insist that the exercise is a challenge,
that there is no right or wrong. Our experience reinforces the
view that a high degree of trust is required within the team, and
between the team and researchers.
– The lack of coding, as we learnt from speaking to the team after-
wards, can also be due to the fact that it is time consuming.
– Participants may also see little value in the coding per se.
Creating categories is very much of an academic interest; it is of
little concern for managers. They do not need to know what the
nature of the constructs they have surfaced is. What matters to
them is what generates success in their organization, why it is so
and how it can be sustained and developed, and not who knew
what and what type of routine it was. Perhaps another round of
interviews would be more fruitful.

further insights into the causal mapping process

Following the session we discussed with the participants what they
thought of the process and what they learnt from it. We also used map-
ping with groups of managers attending short courses. These managers
engaged in a limited mapping exercise for their own firm and shared their
insights and experiences of the technique in small groups. Their thoughts
are also included in the following.
The benefits of the process of mapping organizational success, as seen
by managers, are as follows:
• It fosters understanding; it gives insights into what is happening in
the organization.
• It places value in the detail; it forces the respondents to deal with
specifics.
 MAPPING SUCCESSFUL ORGANIZATIONAL ROUTINES 35

• It highlights how difficult it is to know what causes success.

• It reveals the ‘hot buttons’, the key contributors to success.
• It helps in understanding the uniqueness of the organization; the suc-
cess factors identified are mostly context-specific.
• It shows the hidden strengths of the organization.
• It crystallizes connectivity; it shows interrelationships.
• It highlights that responsibility for what causes advantage in organi-
zations is not obvious.
• It is not attributable to one division, one function or one specific
person.
• It shows the importance of people as the bases of success, not equip-
ment, and so on.
• It focuses on the positive, which is not often done in organizations.
• It identifies critical processes that are unmanaged, that are not
planned, that participants did not realize mattered and that sometimes
are not well respected in the organization.
• It shows what the organization needs to keep doing.
• It gives new ways of discussing strategy, of discussing the business
process.

difficulties that emerged from the process

Similarly we discussed problems that were perceived to be attached to the
process. The main comments were as follows:

• Participants need to be honest for the session to be valuable.

• It is difficult to open up, it can be personally uncomfortable, because it
makes some participants realize they don’t know something poten-
tially important. They may not want to admit it, because it shows that
they are not in control.
• The map is very complex, there are a lot of interrelationships, and so it
is difficult to focus, and to find which connections really matter. You
may not see the wood for the trees.
• It is easy to get bogged down in the detail.
• It may get too complicated. Where do you end the map?
• It is a bit reductionist. It is all about the people involved in the process,
hence it may be biased, too subjective. Maybe a lot of people need to
be involved in identifying success factors.
• Who controls the mapping process? Is there a dominant personality?
Is the map a representation of the view of one specific group or
individual?
• Participants may have a preconceived idea of where the map is going
to go.
• How do you take the process forward as a researcher? What do you do
with the maps? How do you report back to the organization?
36 DIRECTLY DISCOVERING MANAGERIAL KNOWLEDGE

Considering the pros and cons, and considering that so far we have not
come across any method that seems to be as suitable and as fruitful in
eliciting tacit routines, we believe that it is a worthwhile effort to explore
causal mapping further as a means of uncovering the idiosyncratic
aspects of organizations.

TACIT AND UNMANAGED ROUTINES:

IMPLICATIONS FOR MANAGERS AND STRATEGIC
MANAGEMENT

The results showed that among all the factors that cause success in an
organization, some of them are often left ‘unmanaged’. This certainly
raises questions in terms of the fragility of competitive advantage and the
role of managers and strategic management in sustaining advantage.
If tacit and unmanaged routines are important success factors in an
organization, rather than, say, tangible resources, one can conclude that
the main sources of differential value between firms may be at the margin
of managerial understanding. A major implication of this is that sustain-
able competitive advantage based on those routines is most likely to be
vulnerable. We can express several hypotheses about why this may be so.
One of the first reasons comes from the nature of tacit routines. Tacit
routines generate causal ambiguity and, consequently, organizations
may not know precisely which routines or set of routines are at the
source of their advantage (Reed and DeFillipi, 1990). This means that
managers may inadvertently change or destroy a fundamentally impor-
tant routine. This appears to have happened in some organizations that
embarked on a business process re-engineering programme or decided
to downsize or delayer.
Another reason why competitive advantage may be vulnerable arises
from the managers’ disposition towards routines: they may not know that
routines matter or they may not care about them. Top managers may pay
scant attention to the detail of what is happening in the organization.
They are encouraged and expected to address the ‘big picture’, the strategic
agenda of the organization. They are not encouraged or expected to
engage in the detail of the organization.
Two implications may flow from this. First, the detail of how things are
done may not be recognized as having ‘strategic value’. Details tend to be
perceived by managers as peripheral or even trivial. They are seen as
things that are of concern for the shop floor, which senior managers do
not need to consider. The detailed running of the operation is usually not
seen as being ‘strategic’ because strategy is not understood as looking into
the routines of the organization, rather it is understood as ‘analysis and
strategy formulation’. In brief, this means that routines are of little con-
cern for senior managers and, consequently, one can easily understand
that superior performance based on routines may be vulnerable. If those
 MAPPING SUCCESSFUL ORGANIZATIONAL ROUTINES 37

routines are not perceived as being of value, they will not be protected
and nurtured, and hence can easily be inadvertently annihilated. The con-
sequence of this lack of attention to detail by managers is that what may
make the difference between organizations is very much at the margin of
managerial interest.
These points suggest that the role of the ‘manager’ and of the ‘strate-
gist’ and her/his vocabulary has to change. Maybe one needs to start to
understand ‘strategy’ as ‘recognizing organization routines’ and maybe
the word ‘manage’ needs to be replaced by words like ‘protect’, ‘nurture’
and ‘leverage’ rather than ‘control’, ‘direct’, ‘monitor’ or ‘plan’ (Bartlett
and Ghoshal, 1997). Indeed managers need to start understanding that
details matter, that the details, the activities that create value, need to be
nurtured and maybe leveraged to other parts of the organization where
they could be of even greater value.
However, the deliberate search for tacit routines may in itself be
dangerous, and that is where the paradox of tacit knowledge lies. If tacit
routines become manageable, then they may become imitable, even if we
can argue that the risk is limited because of the idiosyncrasy of tacit
routines: that is, they work in one specific context with specific people,
the interrelationships between them, their environment, the equipment,
and so on. We believe the risk is worth taking if it means that managers
can better understand what makes the firm successful today and have
better insights into what to protect when undertaking change.
More broadly, the critical role that routines may play in the generation
of organizational success leads us to question the role of the most ‘rational’
approaches to strategic management and strategic change. It is indeed an
extremely arduous, if not impossible, task to plan and analyse something
that one does not understand very well. Moreover, planning or control-
ling routines may destroy them as their informality may be part of what
makes them successful. Such actions may also render them imitable and
hence in turn nullify their effects.
It is also interesting that the Industrial Organization economics-based
approach of strategic management appears to be of limited use to
managers trying to find competitive advantage. This theoretic approach
focuses on what is happening outside the firm to explain performance
(Mosakowski, 1998). In contrast, the resource-based view of the firm
concentrates on what is happening inside (Barney, 1995) and what is unique
to each organization. If routines are a source of advantage, what causes
success in an organization is idiosyncratic to each organization: that is, one
cannot copy a ‘recipe’ (Spender, 1989) from one firm to another. ‘Generic
strategies’ (Porter, 1980) become the equivalent of tangible resources. They
are available to any organizations, and cannot lead to advantage.
Idiosyncrasy and the limitations of generic strategies may make us
realize that, for example cost cutting is not a tactic that is going to have
long-term benefits. Cost cutting is easy to understand, relatively easy to
implement, it lends itself to measurement and control and as such is a
38 DIRECTLY DISCOVERING MANAGERIAL KNOWLEDGE

tactic that managers like. Cost cutting does allow managers to avoid
uncertainty and ambiguity, however cost cutting is a well-trodden
‘recipe’. Everybody knows how to do it and hence it can be executed by
all, with benefits achieved by most firms. In other words, cost cutting will
only help an organization to stay in competition rather than help it move
ahead of competition.

CONCLUSION

In this chapter we have suggested that organizational tacit knowledge

could be understood as tacit routines. We have also suggested that there
are different degrees of tacitness. Routines may be unarticulable because
they are deeply ingrained in the unconscious; they may be only imper-
fectly articulable; or they may be articulable, if the researcher finds the
right trigger to allow organization members to express them. We have
suggested that, considering our lack of knowledge on tacit routines,
managers and their facilitators should concentrate on those routines that
are not yet articulated but could be, even if imperfectly.
We have proposed an avenue for empirical research that uses causal
mapping as a useful method to elicit tacit routines. This technique allows
participants to reflect on what they are doing. Continually asking managers
what they do that causes success teaches them to uncover skills that they
would not normally talk about. The discovery process is facilitated by
encouraging participants to tell stories and employ metaphors.
We would also like to acknowledge some of the limitations of our
research. Clearly, one cannot establish a unique and direct link between
tacit routines and competitive advantage. There is no single factor that
determines performance; tacit routines may be just one contribution of
many. Moreover, our broad aim is to map organizational success, but it
goes without saying that we can only map part of the routine processes
that might cause success in organizations.
There are quite a few reasons why this is so. First, the maps reflect
inputs by a certain group of managers. We can therefore only register the
routines they are able to access. Ideally everybody in the organization
should take part. Second, there is a time constraint: mapping a full organi-
zation would take weeks if not months. Outsider researchers are rarely
granted such access. Third, the principal reason why tacit routines have
been argued to be a source of sustainable advantage is that they are
immobile and inimitable. By surfacing them, organizations take the risk
of making tacit routines accessible and hence losing their advantage.
However, this argument can be reversed by saying that if an organization
is aware of the real source of its superiority, it can attempt to protect and
nurture it.
Next steps are also problematic. Recognizing the existence of a pheno-
menon is just a first step in moving to a resource that can be better
 MAPPING SUCCESSFUL ORGANIZATIONAL ROUTINES 39

managed and replicated. Factors that are difficult to articulate can also be
difficult to manage. In fact, do they need to be managed, or merely pro-
tected? Up to now, these factors had not been surfaced and hence nothing
was consciously done to manage them. The primary goal of the work-
shops was to identify factors associated with success to make sure that
they were not destroyed when implementing changes in the organization,
or rather, to make sure that the organization was well aware of why it had
so far been successful. Of course what has worked in the past is not guaran-
teed to work in the future. However, it is good to understand the real basis
for current and past success before trying to establish a new one.
 Extracts from the map generated during the causal mapping session.

APPENDIX
40 Feelgood factor 27 Herd instinct 43 Independent 41 Little financial
38 To become the 30 Gut feeling
obvious choice trustees knowledge
26 Standing
31 We are their in market 8 Different needs
best option
44 Lack of
6 Size of market 42 Non-profit making slime-balls
61 Staff retention
34 Defensible 32 Our financial
2 Charity market 9 Set up by/for
performance
charity
29 Trustees 39 Nicer people
5 Well known in
25 Reputation – 59 Driving costs
charity field
theirs and ours 10 Incestuous down
48 We grow
45 Customers like
with the client
7 Less direct 57 Not getting money our people
35 Encourage growth
competiton for what we do?

45 Lower unit cost

1 SUCCESS 4 Customer focus 58 Recruitment
3 Big clients
37 Morale criteria
joining
21 Good first 60 Recruiting
17 Knowing who the
33 Flexible impressions 11 Staff appropriate level
customer is
50 We don’t
lose clients 46 Recruitment
58 Committee
22 Caring
52 Whisky down throats –
committee relationships 55 Rank and file
51 Clients have low
(members)
financial knowledge
14 Systems
15 Empathy
54 Members ring
20 Talking to 18 Knowing the
the staff direct
customers freq/regularly customer

13 Understanding
12 Training
needs

19 Not thinking we are 16 Professionalism

always right – humility 24 Meeting needs

23 Getting it right
 5 We need 3 If it’s done for 16 Send an invoice for
professional help … charities it can’t everything we do …
give our money away be professional providing services
1 Balance between for goodwill
commercial and
voluntary sector

10 Voluntary sector 7 You consult and 18 Insurance

is becoming more advise … tell them companies treat you
professional – the new what to do badly
17 Call each other
school
by first names, more
4 Management friendly
12 Becoming
practices that
professional in the
embody the best
right way 2 Communication 6 Making staff
of both
through external 13 New business redundant is
literature profile incompatible with
8 Front-line charity ethos
professional
11 We have to be advisers, emphasis 20 Dealing with
professional in on professional … 14 Servicing people as human
order to get the flashy committees beings …
right people over-commercial 19 Internal
21 We encourage our competition
members to talk
to our staff (no concern
of cost) … billing clients
by time
23 Pressure created
by poor results
22 Not subject to
pay out dividends etc.

24 Positive
relationship with
trustees
 26 Without the long hours
30 Fear factor – not 39 Keep up with
5 Giving best advice we would need more staff
wanting to be a weak link service commitments
15 Similar values – and costs would increase 35 Like working in
no commission and London
not for profit 36 Pressure about
38 Coming in on not letting the side
11 Staff buy into 31 Most of the staff 29 Examples from the Saturdays to keep up down
objectives of the are not influenced top – commitment workload
client by the charity sector breeds commitment

37 We like being
6 Empathy with part of a successful 34 Bachelor status,
objectives of the organization 4 Staff commitment more commitment,
charity
33 Duality between they can ‘burn the hours’
12 Clients think we charity sector and
are different from professionalism
other providers

2 Ethos 25 Because they

41 Explaining to clients project right image
why things have worked to the clients
14 Identify with
32 Trust 1 SUCCESS out they way they have
the sector
28 Past training and
24 Good returns development for all staff
motivate the staff
3 Investment returns
20 Long-term returns
7 Participation of
compatible with
the trustees
pensions being long term 27 Benefits from long
18 Quality fund service of key staff
managers 16 More for
13 Client’s interests your money
are paramount
19 Money safe with 42 Move at the top, 23 Publicizing
fund managers not when we’re at the returns
22 Good historic
the bottom
track record 40 People have
confidence in us
21 Investment
committee monitoring –
and skills
 MAPPING SUCCESSFUL ORGANIZATIONAL ROUTINES 43

NOTE

1. This methodology section is adopted from Ambrosini and Bowman (2001).

REFERENCES

Ambrosini, V. and Bowman, C. (1998) ‘The dilemma of tacit knowledge: Tacit routines as a
source of sustainable competitive advantage’, paper presented at the British Academy of
Management Conference, refereed track, Lancaster.
Ambrosini, V. and Bowman, C. (2001) ‘Tacit knowledge: Suggestions for operationalization’,
Journal of Management Studies, November, 38(6): 811–829.
Axelrod, R.M. (ed.) (1976) The Structure of Decision: Cognitive Maps of Political Elites. Princeton
NJ: Princeton University Press.
Barney, J.B. (1991) ‘Firm resources and sustained competitive advantage’, Journal of
Management, 17 (1): 99–120.
Barney, J.B. (1995). ‘Looking inside for competitive advantage’, Academy of Management
Executive, 9 (4): 49–61.
Bartlett, C.A. and Ghoshal, S. (1997) ‘The myth of the generic manager: New personal
competencies for new management roles’, California Management Review, 40 (1): 92–119.
Boje, D.M. (1991a) ‘The storytelling organization: A study of story performance in an office
supply firm’, Administrative Science Quarterly, 36: 106–126.
Boje, D.M. (1991b) ‘Consulting and change in storytelling organisation’, Journal of
Organizational Change Management, 4 (3): 7–17.
Bougon, M.G. (1983) ‘Uncovering cognitive maps: The self-Q technique’, in G. Morgan (ed.),
Beyond Method. Beverly Hills, CA: Sage. pp. 173–188.
Bougon, M., Weick, K. and Binkhorst, D. (1977) ‘Cognitions in organizations: An analysis of
the Utrecht jazz orchestra’, Administrative Science Quarterly, 22: 606–639.
Bougon, M., Baird, N., Komocar, J.M. and Ross, W. (1989). Identifying strategic loops:
The self-Q interviews. In A.S. Huff [ed.], Managing strategic thought. Chichester: Wiley.
pp. 327–354.
Brown, J.S. and Dugruid, P. (1991) ‘Organizational learning and communities-of-practice:
Toward a unified view of working, learning and innovation’, Organization Science, 2 (1):
40–57.
Cossette, P. and Audet, M. (1992) ‘Mapping an idiosyncratic schema’, Journal of Management
Studies, 29 (3): 325–347.
Eden, C. (1992) ‘On the nature of cognitive maps’, Journal of Management Studies, 29 (3):
261–265.
Eden, C. and Ackermann, F. (1998) Making Strategy: The Journey of Strategic Management.
London: Sage.
Eden, C., Jones, S., Sims, D. and Smithin, T. (1981) ‘The intersubjectivity of issues and issues
of intersubjectivity’, Journal of Management Studies, 18 (1): 37–47.
Eden, C., Ackermann, F. and Cropper, S. (1992) ‘The analysis of cause maps’, Journal of
Management Studies, 29 (3): 309–324.
Flanagan, J.C. (1954) ‘The critical incident technique’, Psychological Bulletin, 51: 327–358.
Fontana, P.A. and Frey, J.H. (1994) ‘Interviewing: The art of science’, in N.K. Denzin and Y.S.
Lincoln (eds), Handbook of Qualitative Research, Thousand Oaks, CA: Sage. pp. 361–376.
Ford, J.M. and Wood, L.E. (1992) ‘Structuring and documenting interactions with subject-
matter experts’, Performance Improvement Quarterly, 5 (1): 2–24.
Gharajedaghi, J. and Ackoff, R.L. (1994) ‘Mechanisms, organisms and social
systems’, in H. Tsoukas (ed.), New Thinking in Organizational Behaviour. Oxford:
Butterworth and Heinemann. pp. 25–39.
44 DIRECTLY DISCOVERING MANAGERIAL KNOWLEDGE

Grant, R.M. (1991) ‘The resource-based theory of competitive advantage: Implications for
strategy formulation’, California Management Review, 33 (3): 114–135.
Grant, R.M. (1993) ‘Organizational capabilities within a knowledge based view of the firm’,
paper presented at the annual meeting of the Academy of Management Conference,
Atlanta, Georgia.
Grant, R.M. (1996) ‘Towards a knowledge based theory of the firm’, Strategic Management
Journal, 17: 109–122.
Hansen, C.D. and Kahnweiler, W.M. (1993) ‘Storytelling: An instrument for understanding
the dynamics of corporate relationships’, Human Relations, 46 (12): 1391–1409.
Hill, R.C. and Levenhagen, M. (1995) ‘Metaphors and mental models: Sensemaking and
sensegiving in innovative and entrepreneurial activities’, Journal of Management, 21 (6):
1057–1074.
Huff, A.S. (1990) ‘Mapping strategic thought’, in A.S. Huff (ed.), Mapping Strategic Thought.
Chichester: John Wiley. pp. 11–49.
Jenkins, M. (1995) ‘Subjective strategies for small business growth: An evaluation of the
causal maps of small independent retailers’, unpublished PhD dissertation, Cranfield
School of Management.
Langfield-Smith, K. (1992) ‘Exploring the need for a shared cognitive map’, Journal of
Management Studies, 29 (3): 349–368.
Markóczy, L. and Goldberg, J. (1995) ‘A method for eliciting and comparing causal maps’,
Journal of Management, 21 (2): 305–333.
Martin, J. (1982) ‘Stories and scripts in organizational settings’, in A.H. Hastorf and
A.M. Isen (eds), Cognitive Social Psychology. New York: Elsevier. pp. 255–305.
Mosakowski, E. (1998) ‘Managerial prescriptions under the resource-based view of strategy:
The example of motivational techniques’, Strategic Management Journal, 19 (12): 1169–1182.
Munby, H. (1986) ‘Metaphor in the thinking of teachers: An exploratory study’, Journal of
Curriculum Studies, 18 (2): 197–209.
Nelson, R.E. and Mathews, K.M. (1991) ‘Cause maps and social network analysis in organi-
zational diagnosis’, Journal of Applied Behavioral Science, 27: 379–397.
Nelson, R.R. and Winter, S.G. (1992) An Evolutionary Theory of Economic Change. Cambridge,
MA: Belknap Press.
Nonaka, I. (1991) ‘The knowledge-creating company’, Harvard Business Review,
69 (6): 96–104.
Ortony, A. (1975) ‘Why metaphors are necessary and not just nice’, Educational Theory, 25 (1):
45–53.
Polanyi, M. (1962) Personal Knowledge: Towards a Post-Critical Philosophy. London: Routledge
and Kegan Paul.
Porter, M.E. (1980) Competitive Strategy: Techniques for Analysing Industries and Competitors.
New York: Free Press.
Ravetz, J.R. (1971) Scientific Knowledge and Its Social Problems. Oxford: Clarendon Press.
Reber, A.S. (1989) ‘Implicit learning and tacit knowledge’, Journal of Experimental Psychology,
118: 219–235.
Reed, R. and DeFillipi, R.J. (1990) ‘Causal ambiguity, barriers to imitation and
sustainable competitive advantage’, Academy of Management Review, 15 (1): 88–102.
Sackmann, S. (1989) ‘The role of metaphors in organization transformation’, Human
Relations, 42 (6): 463–485.
Schiffman, L.G. and Kanuk, L.L. (1991) Consumer Behavior. Englewood Cliffs, NJ: Prentice
Hall.
Sheetz, S.D., Tegarden, D.P., Kozar, K.A. and Zigurs, I. (1994) ‘A group support systems
approach to cognitive mapping’, Journal of Management Information Systems, 11 (1): 31–57.
Sobol, M.G. and Lei, D. (1994) Environment, manufacturing technology and embedded
knowledge’, International Journal of Human Factors in Manufacturing, 4 (2): 167–189.
Spender, J.C. (1989) Industry Recipes: The Nature and Sources of Managerial Judgement. Oxford:
Blackwell.
 MAPPING SUCCESSFUL ORGANIZATIONAL ROUTINES 45

Spender, J.C. (1996) ‘Organizational knowledge, learning and memory: Three concepts in
search of a theory’, Journal of Organizational Change Management, 9 (1): 63–78.
Srivastava, S. and Barrett, F.J. (1988) ‘The transforming nature of metaphors in group
development: A study in group theory’, Human Relations, 41 (1): 31–64.
Tsoukas, H. (1991) ‘The missing link: A transformational view of metaphors in organiza-
tional science’, Academy of Management Review, 16 (3): 566–585.
Walsh, J.P. (1988) ‘Selectivity and selective perception: An investigation of managers’ belief
structures and information processing’, Academy of Management Journal, 31 (4): 873–896.
Weick, K.E. and Bougon, M.G. (1986) ‘Organizations as cognitive maps’, in H.P. Sims (ed.),
The Thinking Organization. San Fransisco: Jossey-Bass. pp. 125–135.
Wilkins, A.L. and Thompson, M.P. (1991) ‘On getting the story crooked and straight’, Journal
of Organizational Change Management, 4 (3): 18–26.
2
LINKING CONTENT TO
PROCESS

How Mental Models of the Customer Enhance the

Creative Strategy Processes

Olaf G. Rughase

ABSTRACT

We at Panlogos Consulting and the University of Witten-Herdecke consider

customers to be the most relevant source of information for guiding the strategy
process. For this reason, we have developed an analytical tool – called
SENSOR®1 – that helps managers explore the ‘world of the customer’. Using a
storytelling approach, this instrument expands the insights of research findings
from the Cognitive School. SENSOR®, introduced as a consulting tool in 1997,
has already been used in numerous business applications in engineering, finan-
cial services, publishing, consulting, trading, personnel services and manu-
facturing firms. In this chapter we provide one case example as an indication of
how this knowledge tool can be used to facilitate strategic change.

IDENTIFYING PITFALLS IN THE STRATEGIC

LEARNING PROCESS

The Cognitive School in strategic management has gained importance

in recent years with empirical findings and theory now widely recog-
nized by both academics and practitioners. One important branch of the
Cognitive School takes a subjective, interpretive view of strategy and
information processing (e.g. Daft and Weick, 1984). Recently, this branch
has become increasingly recognized in strategic management research
(Kemmerer and Narayanan, 2000; Walsh, 1995).
 LINKING CONTENT TO PROCESS 47

Attention to
issues/changes in
organizational
environment

Mental Change in top Organizational

models of top managers’ mental change/
managers models (‘learning’) strategic option

Interpretation of
issues/changes in
organizational
environment

Figure 2.1 A cognitive model for organizational change

Cognitive researchers using an interpretive approach have developed a

number of models to describe how managers and organizations deal with
highly significant information. In 1983, Dutton et al. proposed that an
interpretation (diagnosis) of strategic issues has three elements: inputs
(such as cognitive maps, political interests or issue characteristics), process
(recursiveness, retroductivity and heterarchy) and outcomes (such as
assumptions, cause–effect understanding, predictive judgements).
Interpretation is the key factor that links changes in strategic action to
changes in top managers’ reading of the environment (Dutton and
Jackson, 1987). Many researchers, including Barr et al. (1993), propose
that organizational renewal depends on learning, which means additions
to or changes of managers’ mental models (see Figure 2.1). These mental
models and interpretations may change in a recursive process, for exam-
ple by redefinition of meanings after receiving additional information. As
a result, the process of data (issue) interpretation is seen as less systematic
and unidirectional than it is when derived from purely rationalistic
schools (Dutton et al., 1983). Nevertheless, interpretation of strategic
alternatives affects what commitments are made in organizations.
Despite theoretical and empirical contributions, there has been little
effort to integrate analytical tools (e.g. environmental scanning, market
research, competitor analysis) into a cognitive strategy model with the
aim of combining content and process. For instance, Eden and Ackermann
(1998) developed techniques to facilitate strategic conversations in order
to enhance strategic action in organizations. But as with most techniques,
they tend to stick to the ‘inside-out’ perspective. Van der Heijden (1996)
similarly uses the scenario technique as a powerful learning tool for
managers to enable strategy creation by creative interpretations. He refers
mainly to external scenarios that are derived from managers’ shared
48 DIRECTLY DISCOVERING MANAGERIAL KNOWLEDGE

and agreed-upon mental models of the external world (the ‘inside-out’

perspective). A striking blind spot, theoretical as well as empirical,
involves the inattention to customers as an ‘outside’ source of information.
How can it be sufficient only to learn within a defined group, within
limited contexts? Learning in regard to strategy development is a
discovery-driven process that definitely needs additional information
(content) for an in-depth understanding of changes in the environment –
both from inside (e.g. core competences) and from outside of the organi-
zation (e.g. trends, markets, environment). Change of a manager’s mental
model or an addition to it is very unlikely without any specific external or
internal trigger of information (content).
At first glance, this challenge seems to be easy to cope with. But what
exactly does trigger the learning process? As shown in Figure 2.2, top
managers face many issues and symptoms. In almost every industry,
typical symptoms or issues might be that:

• purchase patterns of customers become unpredictable;

• target groups are disappearing after a short period of time;
• competitive advantages are non-persistent;
• revenue decreases dramatically;
• a new competitor has changed the ‘rules of the game’, and so on.

In responding to these issues and symptoms, each manager draws on

different personal experiences and knowledge, which in turn respond to
a diversity of contexts that exist in the minds of the managers. This diver-
sity of contexts generates different interpretations about issues, symp-
toms, events and situations. Multiple interpretations based on different
contexts are then being discussed within the group of top managers (and
beyond) by ‘calibrating’ their underlying assumptions. Do the contexts
lead to different meanings? Are these contexts related to each other? This
conversational ‘calibration’ process is not necessarily a formal one. ‘Often
much more important is the informal “learning” activity, consisting of
unscheduled discussions, debate and conversation about strategic ques-
tions that goes on continuously at all levels in the organization’ (van der
Heijden, 1996: 273). Following this important process, top managers build
consensus for strategic action based on an agreed in-depth understanding
of issues and symptoms.
In practice this learning process harbours three major pitfalls. First,
there is possibly no diversity of contexts among the top managers at all.
That does not mean that homogeneous contexts are counter-productive
for holding an appropriate strategic position in general, but homogeneity
is unlikely to lead to a learning process or to strategic change. From our
consulting experience, some currently successful companies get stuck in
this situation for too long, as they tend to read the environment only
through their (still) profitable products, processes and outcomes.
 LINKING CONTENT TO PROCESS 49

Issues
Symptoms

Diversity of Learning by Builds

contexts generates conversational consensus
different ‘calibration’ for strategic
interpretations process action

Issues
Symptoms

Figure 2.2 The strategy learning process

Second, there is perhaps a lack of contexts among the top managers. Very
often managers are surprised by unmanageable consumers, who show
unpredictable buying behaviour, and by societal demands, which were
formerly unknown. At that very moment, managers lack appropriate con-
texts in order to achieve an in-depth understanding of the issues.
Third, top managers may have no willingness or ability to drive a conver-
sational ‘calibration’ process. In fact, many discussions and debates are
limited to an exchange of superficial statements. Accordingly, managers
may persuade others, but do not convince them. On the contrary, a
change-producing conversation is based on an exchange of contexts and
related arguments in order to ‘understand’ your conversation partner and
to find solutions and/or a level of common agreement. If managers don’t
calibrate their contexts, strategy implementation is likely to fail, because
different interpretations (creating hidden agendas) allow inconsistent
strategic action by the management team.
As contexts play a major role in the learning process, the Cognitive School
typically views analytical tools with scepticism because many of these tools
are still embedded in an orderly strategic planning framework. Planning
models approach strategy development as a designed and conscious pro-
cess that must be kept simple (e.g. Andrews, 1987; Eden and Ackermann,
1998). However, as revealed by many researchers, this traditional strategic
planning concept has been proven to be inappropriate in times of hyper-
competition (e.g. Brown and Eisenhardt, 1998; Mintzberg, 1994). Moreover,
these analytical tools tend to provide lots of numbers (e.g. market share,
turnover, customer satisfaction measures). The quantitive data tend to be
seen as conclusions that present an ‘objective’ view of the world. Very
often, this information excludes contexts that are actually needed.
50 DIRECTLY DISCOVERING MANAGERIAL KNOWLEDGE

THE CASE STUDY: A WHOLESALE FIRM

In order to elucidate the shortcomings of many efforts at strategy

development, the following case study gives an example of a company
that initially lacked contexts to understand the data that were available to
them. The company is a German wholesale firm offering articles that can
only be differentiated to a certain degree. At the time we were first
involved with the firm it was faced with decreasing profit margins and
disappearing target groups. The customer base included large to small
businesses that needed the products for further processing. The competi-
tive environment in this market can be described as an oligopoly –
relatively constant over many years.
The top management team (TMT) were trapped in pitfall number two
of the learning process previously described: they were surprised by
symptoms and issues, and the team lacked appropriate contexts to make
sense of them. In order to cope with these demanding issues the TMT
wanted to (a) find new products and services for their existing target
groups and (b) differentiate themselves from their competitors. As an
increasing number of their customers were turning away, the TMT
decided to begin with their existing and lost customers by initiating a cus-
tomer satisfaction survey. The wholesale company interviewed about
2,500 representative customers with a standard questionnaire.
The results of this survey were extremely surprising to the TMT:
91.4 per cent of their customers were ‘always satisfied’ or ‘mostly satisfied’
with the company’s performance; 7.7 per cent of the customers rated the
performance ‘variable’; and only 0.9 per cent were ‘mostly unsatisfied’.
On the contrary, as can be seen in Table 2.1, 35.87 per cent of the customers
classified competitors ‘much better’ or ‘better’, 58.42 per cent classified
the competitor’s performance as ‘even’, and only 5.71 per cent considered
competitors ‘poor’ compared to their current supplier. In addition to these
irritating outcomes, the TMT received lots of ambiguous data about
specific topics. For example, in some cases, the frequency of field service
visits was too high. On the other hand, many customers were asking for
additional service and sales support. As a result, the TMT ended up with
more questions than they originally had – and were still lacking contexts
to understand the issues highlighted by the survey.
There was one piece of apparent good news: the TMT were reassured
that their company was often able to satisfy its customers. However, the
data seemed to suggest that it no longer had any sustainable competitive
advantages. Even more dramatic, conflicting information made a search
for new products and services almost impossible for the TMT.
How can an analytical tool be created to support the mental process of
strategy development and implementation under conditions like these?
The purpose of the cooperation between Panlogos Consulting and the
University of Witten-Herdecke is to develop analytical tools that reflect
 LINKING CONTENT TO PROCESS 51

TABLE 2.1 Customer satisfaction with competitors (per cent)

Competitor is … Much Better Even Poor Very
better poor
Competitor A 1.63 11.41 21.74 1.09 0
Competitor B 1.63 10.86 15.76 2.17 0
Competitor C 0.54 3.53 8.42 1.36 0
Competitor D 0.83 2.72 8.15 0.82 0
Competitor E 0.27 2.45 4.35 0.27 0
Total 4.90 30.97 58.42 5.71 0

the insights of the research findings of the Cognitive School. The key
seems to be to link content to process.

SENSOR®: A STORY-BASED TOOL

FOR STRATEGIC ANALYSIS

We developed an analytical approach that brings content into a strategy

process by considering the interpretation and learning process. We
consider the customer to be the most relevant source of information for
strategy processes. Firms need to annex the ‘thinking of the customer’
with their own thinking to strategically position products and services in
the future. Porter (1980) already claimed that uniqueness does not lead to
differentiation unless it is valuable to the consumer. In other words:
competitive advantages are only created in the minds of the customers. To
develop innovative products and services, companies need to find what
customers will value in the future. But there is a significant difference
between what customers want and what they need. In fact, companies
have to provide something that the customers would like to have, even
though they never knew that they were looking for it; afterwards telling
them (the companies) that they always wanted to have what they finally
got (Liebl, 2000). For this reason, it is necessary to understand the ‘world
of the customer’ in detail, in order to surprise customers with valuable
products and services they never thought of before.
Of course, managers already have a mental model about customers
and their ‘world’, according to their experiences with customers and other
information. But from our experience, these models are likely to be incom-
plete. Managers tend to focus only on those parts of the ‘world of the
customer’ that are directly linked with their existing business. In addition,
the mental map is often inaccurate, because managers are often left to
indirect information sources. In consequence, an analytical tool should
give managers food for thought in order to change their mental models
about the ‘world of the customer’ or enhance them with elements that
they ‘learn’ from the customer. Our analytical approach – which is called
52 DIRECTLY DISCOVERING MANAGERIAL KNOWLEDGE

SENSOR® – helps managers to explore the ‘world of the customer’ by

making patterns within this ‘world’ visible. In addition, it gives a deep
understanding for customer behaviour and customer attitudes.
A critical element in acquiring an in-depth understanding of customers
is the stories they tell (Liebl, 1999). Customers’ positive and negative
stories represent unique experiences with organizations, and thereby
form an organization’s strategic identity (Gabriel, 2000). Actually, there
are many techniques that describe customers’ perception and expecta-
tions, including buying criteria (Porter, 1980) and experiences of
customer value (Woodruff, 1997). But single techniques are barely able to
describe a complex competitive advantage based on a set of different
activities, services, attributes or values that are in interaction with each
other (Porter, 1996).
Storytelling, on the other hand, is the ‘preferred sensemaking currency
of human relationships’ (Boje, 1991). Among customers and organi-
zations it is an excellent candidate for connecting more specific techniques.
Empirical evidence shows that information has significant impact on
judgements and understanding when it is conveyed in the form of a
narrative. For example, Pennington and Hastie’s (1992) research focused
on the role of stories in making juridical decisions. They found out that
participants favoured the side (prosecution or defence) whose testimony
was provided in story-order. The participants were also more confident of
their decisions when the testimonies were in story-order than when they
were not. Adaval and Wyer (1998) made the same findings with con-
sumers who estimated the attractiveness of vacations described in two
different travel brochures. Customers generally preferred a narrative
form of events to a simple list of features. An example of using story-
telling in strategic management is given by 3M (Shaw et al., 1998), where
abstract business plans replaced by narrative texts further employees’
comprehension and sustainable commitment.
Stories matter not only in information processing, but also in remem-
bering (Bartlett, 1932). As Schank and Abelson (1995) claim, all of the
important knowledge that people acquire and retain in memory is based
on stories constructed around past experiences. What’s more, the ‘con-
tent of story memories depends on whether and how they are told to
others, and these reconstituted memories form the basis of the indivi-
dual’s remembered self’ (Schank and Abelson, 1995: 1). In other words,
‘stories provide the basis for (a) comprehending new experiences;
(b) making judgements and decisions about the persons, objects, and
events to which the story refer; and (c) developing general attitudes and
beliefs concerning these referents’ (Adaval and Wyer, 1998: 208). Stories
not only represent detailed descriptions of perceptions, they also elicit
values as well as the identities of storytellers (Conway, 1996; Meyer,
1995; Taylor, 1996). Frames and meanings can be identified as well as the
background of motivation and acting. This is particularly important in
order to understand customers’ perception in-depth. Further, stories are
 LINKING CONTENT TO PROCESS 53

easy to receive from customers. It is easier by far for customers to tell

stories than to transfer their experiences into the abstract value cate-
gories of questionnaires.
In order to absorb customers’ stories, our analytical instrument,
SENSOR®, contains qualitative elements that are heavily influenced by
cognitive science. An in-depth interview (narrative/problem-oriented)
represents the heart of the instrument (Patton, 1990; Witzel, 1985). This
interview contains elements of ethnographic interviews (Spradley, 1979;
Woodruff and Gardial, 1996) and of means–end theory (Claeys et al., 1995;
Reynolds and Gutman, 1988).2 During these interviews, customers are
given the opportunity to express their experiences in their own words
(stories). Within these stories, they talk about their perceptions, the way
they use the product or service, general beliefs and especially how they per-
ceived their experiences by describing related meanings and motivations.
For example, one customer of the wholesale firm we are using for our
case study told us the following story (extract):
Well ... [pause – 30 seconds] Yes, I can remember a remarkable story! Just three
months ago, we had severe problems with one of their products. We had – as
always – a time-critical order from one of our best customers. Only two days
before the day of delivery, we found out that 5 tons of the required material
in stock was bad quality and that it was impossible to process it with our
machines. These problems occur from time to time in our industry, but nor-
mally we immediately check the quality when we get the material. Now, we
really had a problem! First of all, the whole production plan was in danger –
our machines need to run 24 hours. But not only that, this customer especially
is always very pig-headed when it comes to delays. My boss put me under
enormous pressure to solve the problem. First thing I did, I called the whole-
saler and explained the situation to him. You know, we are not a big customer
and 5 tons is extremely difficult to get within one day, even for a professional
wholesaler. What is there to say – they got it fixed in only one day. What a job!
I can’t tell you how happy I was. [laughing] Thanks to Mr Lange, who took
care of it immediately, we didn’t lose one of our best customers. I mean, the
way he handled it was very professional; just a few questions and Mr Lange
knew what to do. In my opinion, that really makes a difference and really sur-
prised me. [pause – 1 minute] You know, talking is cheap in our industry –
everybody promises anything. But what really counts is that you can rely on
your supplier, who knows you and will help you in critical situations. Most
of the wholesalers just want to sell, I mean, I understand that, but afterwards
they don’t care anymore. Losing a good customer leads us into a tough situ-
ation. However, on that particular evening I bought a bottle of champagne,
went home a little bit early, and shared the rest of the evening with my
family. [pause – 15 seconds] Hey, talking about wholesalers that brings me to
another thing . . ..

Customers know many stories to tell that refer to experiences with cer-
tain products, services or companies. During these stories they change
topics quickly, ask themselves questions (and give answers), or jump from
beliefs to associations to other related stories. Normally, such an interview
54 DIRECTLY DISCOVERING MANAGERIAL KNOWLEDGE

takes from 45 minutes up to one and a half hours. Of course, all interviewees
are kept anonymous. There is no written guide or manual for the inter-
viewer, who needs to be well trained and experienced in passive interview
techniques, such as supportive or describing questions (‘Did I understand
you right ...?’ or ‘Could you describe to me what do you mean by ...?’), as
the interviewee should ideally talk for over 95 per cent of the time. All
interviews are tape-recorded, so the interviewer can focus on the customer
and the development of the narrative instead of making notes.
There are two important indicators for the success of these non-
standardized interviews. The first indicator is given by the customers
themselves. After describing detailed perceptions, beliefs, complex asso-
ciations, meanings and motivations, in many cases customers do not
know what has happened during the talk. A common feedback from cus-
tomers is: ‘It was real fun and somehow interesting, but all I did was talk.
There was no structure at all. I doubt that you will get anything concrete
from that!’ In fact, that is the best feedback we can get when aiming to
reveal the unconscious by means of an ethnographic interview. The
second indicator is given by the interviewers. After a certain amount of
interviews, they tend to say: ‘I am getting bored. Each time, I hear the
same stories over and over again!’ By then, you know that you are on the
right track as patterns have already become visible.

ANALYSING STORY-BASED DATA

During evaluation, each qualitative interview is analysed by an inter-

subjective, systematic and focused content analysis. During this process
each story is deconstructed into its elements. The elements of the story are
brought into relation to each other or into relation to other expressions,
meanings and motivations that were amplified by the customer.
In order to achieve this, every tape-recorded interview is analysed three
times by a specially trained analyst. The first time, the analyst tries to get
a general understanding of the basic structure of the interview. The
second time, the analyst begins to write down and draw all elements of
the stories (in the words of the customer) on a blank sheet of paper using
simple tools like a pencil and rubber. The analyst relates the elements to
each other as the customer creates linkages during the interview. This ses-
sion is very time-consuming, as the analyst needs constantly to evolve
and change an expanding network. The third time, the analyst verifies
all the elements and completes details which she/he might not have
heard during the second session. In order to get intersubjective results,
other analysts run random tests to compare results. Through this process,
a net of stories is created for each customer eliciting the contexts of his or
her ‘worldview’.
Having such a validated network for each interview, the analysts
begin to process a ‘qualitative content analysis’ (Mayring, 1997). When
 LINKING CONTENT TO PROCESS 55

comparing these single networks to each other, by creating categories

for similar or equivalent contexts, descriptions, meanings, attitudes or
perceptions patterns become visible very quickly. Usually, only 15 to 25
interviews per segment are sufficient to reveal detailed patterns (Bushko
and Raynor, 1997). These distinct patterns can be consolidated into
detailed cognitive maps (Huff, 1990; Laukkanen, 1998). Not only are the
elements within these cognitive maps important, but so are the relations
between the elements. The relations depict the customer’s associations
and complex perception of unique values (e.g. competitive advantages).
As a result, SENSOR® provides cognitive maps that are networks of stor-
ies within the ‘world of the customer’ – revealing customers’ values, atti-
tudes and reasons.
However, the instrument itself and its results are not sufficient to
enhance the strategy process. In order to link content and process, the
cognitive maps are not presented to the managers as traditional research
outcomes in a unidirectional way. Considering the use of stories for bet-
ter information processing, many examples and voice-recorded sequences
from different interviews are used to give managers the experience of
detecting patterns themselves during a workshop. Managers are first
asked to make their own pattern construction of the voice-recorded
sequences they have heard. Then, step by step, parts of the cognitive map
that resulted from analysis are revealed. During this process, managers
create a piecemeal understanding of linkages within the cognitive map. In
the course of the workshop, the complete cognitive map is (re)constructed
by the management. Moreover, they begin to tell stories of their own
experiences with customers, which are in accordance with some of these
linkages. By then, you know that the cognitive map’s information has
become useable knowledge for management.
By the end of the workshop, managers are able to ‘read’ and under-
stand the cognitive maps in detail. Of course, many managers have
heard a lot of stories and anecdotes from customers before, but
redesigning these fragments and constructing linkages between them
gives them a different perspective on the ‘world of the customer ’. This is
exactly the kind of information (content) that triggers an interpretation
and learning process.
In our consulting experience, the mental map of the customer is differ-
ent from the mental map of the manager about the customer. Often the
difference is considerable. We believe that important ‘outside-in’ infor-
mation is lacking if a strategy process only relies on ‘inside-out’ perspec-
tives. However, differences between the mental maps of the customer and
the mental maps of managers about the customer immediately bring
new creative ideas and new opportunities to the minds of managers. Our
SENSOR® approach provides detailed information about the customer’s
cognition to enhance this learning process. The ‘outside-in’ perspective
interacts with the ‘inside-out’ to create new knowledge from which new
strategy can be generated.
56 DIRECTLY DISCOVERING MANAGERIAL KNOWLEDGE

DISCUSSION OF RESULTS AND THE STRATEGIC

LEARNING PROCESS IN PRACTICE

To illustrate the outcomes of the analytical tool, we return to the case

study. After accomplishing the customer satisfaction survey, the German
wholesaler initiated a SENSOR® study in order to get contextual patterns
from stories its customers tell. To get a selection of interviewees from its
broad customer base, the company followed its existing segmentation by
customer’s company size and turnover. Thirty customers were chosen
within three different segments, plus five non-customers. Among cus-
tomers and non-customers, interviews were made with the chief execu-
tive or the head of purchasing department. As a result of the interviews,
we acquired two distinct cognitive maps (patterns) from the stories cus-
tomers told (Figures 2.3 and 2.4). In addition, non-customers told stories
in the same contextual patterns as did customers. In consequence, there
was no difference between customers and non-customers in regard to
their existing ‘worldviews’.
Usually cognitive maps are very complex. For the purpose of demon-
stration, the cognitive maps shown here are kept as simple as possible.
For instance, all arrows in the maps are only cause–effect relations.3 Some
arrows are bold, indicating that these cause–effect relations were named
by at least 85 per cent of all interviewees who created this contextual pat-
tern. The first pattern (Figure 2.3) is based on narratives of 13 customers,
and the second pattern (Figure 2.4) is derived from stories of 22 cus-
tomers. In order to ‘read’ the map, follow single paths of connected ele-
ments. For example, in Figure 2.3 take ‘quality of product’ as a starting
element. More than 85 per cent of the customers were telling stories
around these core cause–effect relations to the effect of (simplified):
The quality of the product is important, because we don’t want to have any prob-
lems in the manufacturing process with the product due to our machine-time
optimization. This optimization is necessary, because we always get extremely
time-critical orders from our customers. You need to know that our customers are
always under enormous time pressure to get the final product. Therefore, we
need to provide flexibility for our customers.

In accordance with our experiences, the TMT of the German wholesaler

had a far different mental map about their customers than SENSOR®
revealed. Most importantly, the TMT suddenly recognized that there were
two contextual patterns (Figure 2.3 and Figure 2.4) instead of one to be
taken into account. The first contextual pattern (Figure 2.3) was well
known to them. The customers in this pattern are under extreme competi-
tive pressure. Perceiving their own business as highly standardized, this
group of customers try to achieve cost leadership by machine-time opti-
mization and low purchasing costs in order to survive. They mainly
forced the wholesaler to lower prices – which raised the issue of decreas-
ing profit margins.
 Differentiation
Capacity
utilization
Customer Ruinous
orientation/ competition
Time flexibility Dialogue
pressure of
customer

Extremely Quick
time-critical problem
orders solving
Will act in
our behalf
Machine- Cost-push &
Optimize time low bids
time optimization Reliability It’s fun to
management deal with

Fair deal Trust

Make quick Aim of No Knows our
calculations wholesaler problems needs
with product Info. on Pricing Good
technical business
options relationship Informal
exchange

Fast & Short time Quality of Personal

flexible of delivery product service
service partner Events

Figure 2.3 Customers’ causal cognitive map: pattern I

 Partner for See trends Ruinous
Customer- other Differentiation
upfront competition
requested agencies
flexibility
Increasing time Customer High-quality
pressure of orientation & products
their customers loyalty important
Transfer to
my
Cost-push & Product customer
small consulting Actual
Extremely market
margins for customer
time-critical overview
orders Trust
Information
on new
Machine- Avoiding Specific
Quick products
time processing product demand
problem solving Good
optimization problems by agencies
business
relationship
Competent
consulting
Small/ Quick
Fair deal &
no storage reaction/
actual pricing
response Wide Info. on Pricing
product range technical
use options
Short time Good Knows
of delivery service & our needs
Innovation
flexibility
Quality of Personal Events/ Sample-service &
product service partner visits -collection

Figure 2.4 Customers’ causal cognitive map: pattern II

 LINKING CONTENT TO PROCESS 59

Taking a closer look at the patterns, the TMT found out that a major
part of their target group had not disappeared. In reality this target group
had developed a new contextual pattern (Figure 2.4), which was formerly
unknown to the TMT. The second group of customers perceive their
business quite differently. As shown in the figure, they are trying to get
out of ruinous competition by focusing on the use of innovative, high-
quality (high-price) products of the wholesaler and of competitors. In
consequence, these businesses reduced the turnover volume of standard
products with their wholesalers. Before knowing these contexts, the TMT
had the impression that many customers were just ‘disappearing’.
During a full-day workshop, the TMT calibrated their interpretations of
the contexts by using examples and stories, while referring to linkages in
the cognitive maps. Our experience is captured directly by Barry and
Elmes: ‘From a practitioner’s viewpoint, the narrativist stance can encour-
age people to explore strategic issues in more meaningful ways’ (1997: 431).
An immediate indicator of the workshops’ value was that by its con-
clusion the TMT were able to interpret every single piece of information
from the customer satisfaction survey they had earlier found so difficult
to understand. As mentioned, the frequency of field service visits was
perceived as too high by customers. The first group of customers
(Figure 2.3) needed fewer visits, because their interest was limited to pri-
cing and time of delivery. They can get this information by phone or
online. Only when problems arise do they require personal service to find
a solution as quickly as possible. The second group of customers
(Figure 2.4) need and ask for more visits, because their interest is focused
on information about new, innovative products and actual market situa-
tions. Intense personal support would help them to improve their own
service to end-users. In reality, however, current field service visits were
limited to information about existing standard products so that even this
group requested a reduction in the frequency of personal visits.

CONCLUSIONS FOR STRATEGIC REORIENTATION

The last task for the TMT was to build consensus for strategic action. They
decided to give up their existing customer segmentation and replace it
with the two segments that were detected by the storytelling approach.
The decision was made to focus primarily on the second group of
customers (Figure 2.4), because their knowledge-based core competences
allowed them to provide lots of new services and products, improving
customers’ professional aims. Several reasons supported this decision, for
example (a) higher profit margins, (b) no competitor in this field, (c) an
increasing number of customers in the future, and (d) ruinous competition
in the other customer group leading to a decreasing customer base.
This dramatic change in strategy was made possible by annexing
the ‘thinking of the customer’ to the knowledge of the managers. The
60 DIRECTLY DISCOVERING MANAGERIAL KNOWLEDGE

Customers’ context
=
Issues content as
Symptoms information by an
analytical tool

Diversity of Learning by Builds

contexts generates conversational consensus
different ‘calibration’ for strategic
interpretations process action

Issues
Symptoms

Figure 2.5 The strategy learning process: linking content to process

customers’ contexts were a critical content in enhancing a creative strategy

process (Figure 2.5). The wholesaler gained substantially more than it
would have by, for instance, just hiring a sales consultant to increase sales
power in traditional customer segments. It finally found a new sustain-
able strategic position.
In conclusion, bringing the content of analytical tools into a mental
strategy process is necessary in order to combine ‘inside-out’ and ‘outside-
in’ perspectives in strategic management. Analytical tools should be
carefully adapted to the insights of the research findings of the Cognitive
School. Experience has shown that in order to gain these contents, the
same methodologies and techniques (such as cognitive mapping) can be
used (see, e.g., Eden and Ackermann, 1998). These methodologies and
techniques already consider the interpretive view of information process-
ing and they are suitable to demonstrate differences of interpretations
from external sources, compared to those of managers inside the organi-
zation, thus triggering the learning process by conversational calibration.

NOTES

I would like to thank Franz Liebl for his help in clarifying these ideas and his thoughtful
comments on earlier drafts.
1. Registered trademark in Germany.
2. In specific cases the qualitative interview is followed by a quantitative questionnaire.
Sometimes the management would like to verify certain competitive advantages that they
think play a major role for the customer. The interviewer will then present only those ques-
tions to the interviewee that were not covered in the qualitative part of the interview. Using
 LINKING CONTENT TO PROCESS 61

a special form of questioning based on Kano (1984) and Berger et al. (1993), these results
indicate whether the described performance would be perceived as a ‘must-be’ or as an
‘attractive’ element. The quantitative results should only be interpreted together with the
results of the qualitative interviews.
3. Owing to the purpose of the SENSOR® study, the linkages may also describe non-
directive associations (A reminds me of B), time-order of elements (after A I make B) or
relations (A is better/worse than B).

REFERENCES

Adaval, R. and Wyer, R.S. (1998) ‘The role of narratives in consumer information process-
ing’, Journal of Consumer Psychology, 7 (3): 207–245.
Andrews, K.R. (1987) The Concept of Corporate Strategy. (3rd edn). Homewood, IL: Irwin.
Barr, P.S., Stimpert J.L. and Huff, A.S. (1993) ‘Cognitive change, strategic action, and organi-
zational renewal’, Strategic Management Journal, 13 (special issue: summer): 15–36.
Barry, D. and Elmes, M. (1997) ‘Strategy retold: Toward a narrative view of strategic dis-
course’, Academy of Management Review, 22 (2): 429–452.
Bartlett, F.C. (1932) Remembering: A Study in Experimental and Social Psychology. New York:
Macmillan.
Berger, R.B., Boger, D., Bolster, C., Burchill, G., DuMouchel, W., Pouliot, F., Richter, R.,
Rubinoff, A., Shen, D., Timko, M. and Walden, D. (1993) ‘Kano’s methods for under-
standing customer-defined quality’, Hinshitsu (Quality): The Journal of the Japanese Society
for Quality Control, Fall: 3–35.
Boje, D.M. (1991) ‘The storytelling organization: A study of story performance in an office-
supply firm’, Administrative Science Quarterly, 36: 106–126.
Brown, S.L. and Eisenhardt, K.M. (1998) Competing on the Edge: Strategy as Structured Chaos.
Boston: Harvard Business School Press.
Bushko, D. and Raynor, M. (1997) ‘Consulting’s future, game theory, and storytelling’,
Journal of Management Consulting, 9 (4): 3–6.
Claeys, C., Swinnen, A. and van den Abeele, P. (1995) ‘Consumers’ means–ends chains for
‘think’ and ‘feel’ products’, International Journal of Research in Marketing, 12 (3): 193–208.
Conway, M.A. (1996) ‘Autobiographical memory’, in E.L. Bjork and R.A. Bjork (eds),
Memory. San Diego: Academic Press. pp. 165–194.
Daft, R.L. and Weick, K.E. (1984) ‘Toward a model of organizations as interpretation
systems’, Academy of Management Review, 9 (2): 284–295.
Dutton, J.E. and Jackson, S.E. (1987) ‘Categorizing strategic issues: Links to organization
action’, Academy of Management Review, 12 (1): 76–90.
Dutton, J.E., Fahey, L. and Narayanan, V.K. (1983) ‘Toward understanding strategic issue
diagnosis’, Strategic Management Journal, 4: 307–323.
Eden, C. and Ackermann, F. (1998) Making Strategy: The Journey of Strategic Management.
London: Sage.
Gabriel, Y. (2000) Storytelling in Organizations: Facts, Fictions, and Fantasies. New York: Oxford
University Press.
Huff, A.S. (ed.) (1990) Mapping Strategic Thought. Chichester: Wiley.
Kano, N. (1984) ‘Miryoku-teki hinshitsu to atarimae hinshitsu’, Hinshitsu (Quality): The
Journal of the Japanese Society for Quality Control, 14 (2): 39–48.
Kemmerer, B. and Narayanan, V.K. (2000) ‘A cognitive perspective on strategic management:
Contributions and implications’, paper presented at Strategic Management Society (SMS)
20th Annual International Conference, Vancouver, Canada.
Laukkanen, M. (1998) ‘Conducting causal mapping research: Opportunities and challenges’,
in C. Eden and J.-C. Spender (eds), Managerial and Organizational Cognition. London: Sage.
pp. 168–191.
Liebl, F. (1999) ‘Was ist schon einmalig?’, econy, 2: 116–117.
62 DIRECTLY DISCOVERING MANAGERIAL KNOWLEDGE

Liebl, F. (2000) Der Schock des Neuen: Entstehung und Management von Issues und Trends.
Munich: Gerling.
Mayring, P. (1997) Qualitative Inhaltsanalyse: Grundlagen und Techniken. 6th edn. Weinheim:
Deutscher Studien Verlag.
Meyer, J.C. (1995) ‘Tell me a story: Eliciting organizational values from narratives’,
Communication Quarterly, 43 (2): 210–224.
Mintzberg, H. (1994) The Rise and Fall of Strategic Planning. New York: Free Press.
Patton, M.Q. (1990) Qualitative Evaluation and Research Methods. 2nd edn. Newbury Park, CA:
Sage.
Pennington, N. and Hastie, R. (1992) ‘Explaining the evidence: Testing the story model for
juror decision making’, Journal of Personality and Social Psychology, 62: 189–206.
Porter, M.E. (1980) Competitive Strategy: Techniques for Analyzing Industries and Competitors.
New York: Free Press.
Porter, M.E. (1996) ‘What is strategy?’, Harvard Business Review, November/December:
61–78.
Reynolds, T.J. and Gutman, J. (1988) ‘Laddering theory, method, analysis and interpretation’,
Journal of Advertising Research, 28 (1): 11–31.
Schank, R.C. and Abelson, R.P. (1995) ‘Knowledge and memory: The real story’, in R.S. Wyer
(ed.), Advances in Social Cognition: Vol. 8 Knowledge and Memory: The Real Story. Hillsdale,
NJ: Lawrence Erlbaum Associates. pp. 1–85.
Shaw, G., Brown, R. and Bromiley, P. (1998) ‘Strategic stories: How 3M is rewriting business
planning’, Harvard Business Review, May/June: 41–50.
Spradley, J.P. (1979) The Ethnographic Interview. New York: Holt, Reinhart, Winston.
Taylor, D. (1996) The Healing Power of Stories: Creating Yourself Through the Stories of Your Life.
New York: Gill & Macmillan.
van der Heijden, K. (1996) Scenarios: The Art of Strategic Conversation. Chichester: Wiley.
Walsh, J.P. (1995) ‘Managerial and organizational cognition: Notes from a trip down memory
lane’, Organization Science, 6: 280–321.
Witzel, A. (1985) ‘Das problemzentrierte interview’, in G. Jüttemann (ed.), Qualitative
Forschung in der Psychologie: Grundfragen, Verfahrensweisen, Anwendungsfehler. Weinheim:
Beltz.
Woodruff, R.B. (1997) ‘Customer value: The next source for competitive advantage’, Journal
of the Academy of Marketing Science, 25 (2): 139–153.
Woodruff, R.B. and Gardial, S.F. (1996) Know Your Customer: New Approaches to Customer
Value and Satisfaction. Cambridge, MA: Blackwell Business.
3
E N A B L I N G S T R AT E G I C
M E TA P H O R I N
C O N V E R S AT I O N

A Technique of Cognitive Sculpting for

Explicating Knowledge

John R. Doyle and David Sims

ABSTRACT

The chapter describes our way of operationalizing the concept of metaphor for
organizational intervention and research purposes. It describes cognitive sculpt-
ing, a technique we have developed to enable people to build physical
metaphors for whatever they are trying to talk about (Sims and Doyle, 1995).
The technique has turned out to be effective both for the expression of existing
metaphor and for the production of new metaphor. We discuss possible reasons
for this effectiveness and suggest some implications for surfacing and gene-
rating organizational knowledge. In particular, we are concerned with how
attention seems to be differently directed when discussion centres on a cogni-
tive sculpture, and how aspects of knowledge that are normally less honoured
in a propositional, entirely verbal and more digital conversation may come into
play when a sculpture is involved. We raise some of the questions that seem to
us to be begged by our work so far, and conclude by considering the potential
of facilitating strategic knowledge via sculpting.

We think in metaphors. This is true whether we are talking about thinking

that is abstract and creative, or about thinking that might appear mun-
dane and commonplace. At either of these extremes, and everywhere
64 DIRECTLY DISCOVERING MANAGERIAL KNOWLEDGE

between, our knowledge is built on metaphor and analogy. For example,

the idea that knowledge could be ‘built’ on something is metaphorical,
though you may not have registered it as such when you read the previ-
ous sentence. It is noticeable how often metaphorical expression arises
from primitive, concrete thinking using body-kinaesthetic and image
schemas to represent what we take to be our knowledge of the world,
both inner and outer.
The word metaphor is derived from the Greek ‘to carry’ or ‘transfer’. A
colleague tells me that in Greece the word for ‘house removals’ derives
from the same root. This image is reminiscent both of Monty Python
(removal men with a highbrow interest in metaphor) and of homunculi in
the head, for in the same way that household goods are carried from one
place to another, in metaphor a set of symbolic relationships is ‘carried’
from one place (technically known as the vehicle or, in research on
analogy, the source domain) to another (the topic, or the target domain
when applied to analogy). On arrival, this set of symbolic relationships
collects a meaning. (These terms are themselves a set of mixed metaphors:
topics, vehicles, domains, sources, targets and arrival.)

THREE KINDS OF METAPHOR

In this chapter it will be helpful to distinguish three kinds of metaphor.

In the first, the metaphor consists of a main theme with elaborations. TIME
IS MONEY is the main theme of a metaphor with which we are all familiar.
But since people have found the metaphor useful, we have a collection of
related metaphors around this same theme – these are the elaborations.
Money can be spent, wasted, saved; so too can time. Here metaphor is
much more than the metaphor heading (TIME IS MONEY); metaphor transfers
many of the uses associated with the source domain to the target domain.
We do not mean, by the isolation of a main theme, as distinct from elabora-
tions of the theme, that the former necessarily pre-dates the latter in the
evolution of a metaphor. Rather, we mean that the main theme acts as a
kind of category heading for the elaborations, and thus allows the basis of
the metaphor to be more explicitly seen in the language.
The uses that do or don’t get transferred via metaphor implicitly define
those aspects of the source domain that are relevant to the target domain.
For instance, the TIME IS MONEY metaphor makes no use of the physical
appearance of money, or the fact that different countries have different
currencies. The metaphor presumably originates in the fact that people
receive money for time worked. It is then a short step to equate time not
working with money not earned, which is one of the connotations of the
statement ‘Time is money’. The identification of money and time is very
close in this metaphor: we would not think twice if someone said ‘three
hours is £60’. It is worth reflecting that there is an underlying idea here:
time is a resource, and within our culture most resources are compared
 ENABLING STRATEGIC METAPHOR IN CONVERSATION 65

with each other in terms of one relatively quantifiable and convertible

resource: money. Hence the phrase ‘Time is money’ may be more a con-
venient statement of an underlying theme than a root theme in itself.
Another kind of metaphor is the one-shot metaphor, for instance
when a metaphor has just been coined for the first time (e.g. TIME IS A
COMPUTER PROGRAM). The new metaphor would probably first appear in a
single elaboration that picked out the particular sense in which the meta-
phor is to be understood (e.g. ‘Time’s algorithm will determine which of
us is right’). If the metaphor were to catch on, we might find elaborations
similar to the above (e.g. NIGHT IS DOWN-TIME) appearing in common usage.
But as we shall see, many metaphors are just one-shot metaphors.
The third category of metaphor is when there appears to be a family
resemblance of metaphors that lack an explicit statement of the main
theme. Uses such as:

• The prize giving is only a short way off.

• My partner’s birthday is close to my own.
• In the distant past...
• What length of time will it take?

These, and many more related usages, seem to be examples of a metaphor

TIME IS DISTANCE, which is not an explicit metaphor in its own right in English.
It is this third kind of metaphor that Lakoff and Johnson drew attention to
in their book Metaphors We Live By (1980), and it consists of elaborations
without a statement of the main theme. Because these metaphors are not
grounded in an explicit main metaphor that holds them together, we could
equally well describe them as implicit metaphors.

METAPHOR, ANALOGY AND SIMILE

Metaphor is sometimes contrasted with analogy (e.g. Miller, 1979;

Tsoukas, 1991) in that the latter focuses on the relation between objects
(A is to B as X is to Y), and that analogy may be made between objects
within the same domain. (E.g. it would be perfectly acceptable to draw
the analogy between the toes’ relationship to their feet and the fingers’
relationship to their hands, where both source and target are clearly from
the same domain – parts of the body.) Metaphor, on the other hand, applies
only when the domains are quite different (e.g. Spies are a nation’s ears). An
additional way that metaphor differs from analogy is that it is purely a
linguistic device, though it clearly implies the existence of sophisticated
cognitive models, to make the cross-domain transfer of meaning possible.
On the other hand, simile, which is also a linguistic device, differs from
metaphor in such an apparently insignificant way that many commenta-
tors have sought to conflate metaphor with simile as cognitively equiva-
lent (Ortony, 1975). At a surface level simile is signalled by the words ‘is
66 DIRECTLY DISCOVERING MANAGERIAL KNOWLEDGE

like’. To use a simile such as ‘Time is like money’ differs from metaphor,
first, because the verbal formula ‘is like’ draws attention to the figurative.
With simile we expect the speaker to carry on and tell us exactly how
time is like money.
Second, simile differs from metaphor (most notably from the third
kind of metaphor) in the degree of awareness invoked in the hearer
about the figurative nature of the utterance. By contrast, implicit
metaphors seem to go undetected by most speakers, because they are so
taken for granted. Whereas the simile emphasizes that meaning is being
transferred, metaphor hides this fact. In asserting that A is B, rather than
merely like B, a fusion of meanings is invited. The linguistic conventions
surrounding metaphor also make it possible for elaborations to be used
quite naturally. If A is B, then we may talk about A in much the same way
we would talk about B. In this way metaphor is allowed to become
embedded in our language.
Analogy, simile and metaphor are all primarily cognitive devices, and
the last two are realized in characteristic language conventions. Symbols
are also not to be taken literally, but usually convey their content in a more
overtly affective, rather than cognitive, manner, and are more likely to be
visual objects or enacted ceremony than linguistic devices. The relation
between the symbol and what it symbolizes may be totally arbitrary, as in
a country’s flag, or an exclamation mark. Saluting the flag or desecrating
the flag are both gestures to the essential meaning behind the symbol.

THE UBIQUITY OF METAPHOR

The various particular metaphors scattered throughout this chapter are

intended to give an impression of the prevalence of metaphor. In addition,
we now report on two studies that shed further light on this matter.
In a review of the neglect shown by cognitive psychology to figurative
language, Pollio et al. note that ‘speakers as varied as politicians enga-
ging in public rhetoric, to patients and therapists engaging in psycho-
therapy, produce an average of about 1.5 novel and 3.4 clichéd figures of
speech per 100 words spoken’ (1990: 143). (In our typology of metaphors,
above, only the one-shot metaphors could not be described as clichéd.)
They go on to estimate that speakers will produce an average of 15 million
or so figurative uses in a lifetime.
Martin (1994) used text from the Wall Street Journal (WSJ), which in
some ways must be closer to ‘business talk’ than Pollio et al.’s corpus. He
found overall frequencies consistent with Pollio’s for the incidence of
figurative language: 4.3 metaphors per 100 words of text. That is approx-
imately one metaphor per sentence.
But Martin also analysed the metaphors into clusters (themes), and
found that over half of all metaphor instances were accounted for by just
five metaphor themes. By far the most popular metaphor in the WSJ is
 ENABLING STRATEGIC METAPHOR IN CONVERSATION 67

40

clichéd metaphors

20

one-shot metaphors

Metaphor

Tokens

100 200 300 400

Figure 3.1 Metaphor types (instances) as a function of the number of

metaphor tokens met for one-shot metaphors (dashed line) and clichéd
metaphors (continuous line). NB: The sequence AABCBCA has three
types (A, B, C) but seven tokens

what Martin calls ‘Value-as-location/value-change-as-movement’, which

might occur in phrases such as house prices have fallen sharply, the index
finished down 12.3 at 2078.4, and turbulent trading on the floor. This elabor-
ated metaphor, by itself, accounted for a quarter of all uses.
Martin further divided the clusters into those that occurred once, and
those that occurred more than once. (This latter cluster are presumably
‘clichéd’, in Pollio et al.’s terminology.) We have schematically repro-
duced his findings in Figure 3.1. The number of clichéd metaphors in the
WSJ seems to asymptote at about 40. However, the rate of production of
one-shot metaphors increased constantly with the amount of text met.

METAPHOR TYPES

While the study was limited in the amount of text it analysed, it is

interesting to speculate that metaphors can be divided into two distinct
classes. For any given area of discourse we suggest there are a relatively
small number of metaphors that account for most of the instances – for
the WSJ it accounted for nearly 90 per cent – and that provide the basic
means of talking about the domain. These metaphors are part of the
background that goes largely unnoticed, as evidenced by the fact that the
five most popular metaphors were of the third kind listed above, namely
elaborations without explicit heading.
68 DIRECTLY DISCOVERING MANAGERIAL KNOWLEDGE

On the other hand, people continue to coin new metaphors at a rate that
does not seem to decline, and that enables them to talk creatively. By their
novel connecting of disparate domains, these, presumably, are the kinds
of metaphors we would most likely notice as metaphors.

METAPHOR AND SCIENTIFIC THINKING

In his ‘Catechism of Cliché’, Myles na Gopaleen (perhaps better known as

Flann O’Brien, author of The Third Policeman) makes fun of the apparently
arbitrary nature of prepositions in English.

What can one do with fierce resistance, especially in Russia?

Offer it.
But if one puts fierce resistance, in what direction does one put it?
Up.
In what direction does the meeting break in disorder?
Up.
In what direction should I shut?
Up. (1997)

Yet the use of prepositions in English is not as arbitrary as might

seem from this examples. Lakoff and Johnson (1980) suggest that the
metaphor MORE IS UP (as indicated by: a rise in salary, a higher rate of taxa-
tion, lower inflation) may be based in the everyday experience of adding
more to a pile or container, and observing that the level of the pile goes
up. They list a dozen or so other metaphors centred on the concept of
verticality (e.g. HAPPY IS UP, DISORDER IS UP, CONSCIOUS IS UP, etc.), each plausi-
bly grounded in physical experience.
Along with a renewed interest in metaphor has gone a parallel debate
about its usefulness (or lack of) in theorizing within the social sciences.
The traditional view (e.g. Miller, 1976; Pinder and Bourgeois, 1982)
sees metaphor as, at best, a useful starting point, but to be eliminated as
soon as possible in favour of scientific thinking. The other view is that
metaphor is inescapable.
We take the Lakoffian view that all thinking, whether scientific think-
ing or not, is inextricably bound up with figurative thinking. Scientific
thinking merely takes the basic ‘metaphors we live by’ more seriously,
more systematically, and perhaps more literally(!), than does non-
scientific thinking. As an example, we have already described two
metaphors TIME IS DISTANCE and MORE IS UP. Consider their appearance as
labels on a graph (Figure 3.2), typical of the conventions adopted in
scientific work. It is clear that the conventions implied by this labelling are
only a refinement of the two metaphors TIME IS DISTANCE (x-axis) and MORE
IS UP (y-axis), and not an escape from them.
Not only is metaphor fundamental to thought in general, but a particular
facility with analogy may have its own advantages. In a series of studies
 ENABLING STRATEGIC METAPHOR IN CONVERSATION 69

Salary

Time (years)

Figure 3.2 Illustrating that typical scientific conventions are often based
on metaphors. Someone’s salary has gone UP, then DOWN, then UP again
(just as the line does)

Klemp and McClelland (1986) used peer evaluation to identify ‘outstanding

managers’ in a number of organizations. They found that one of the features
that distinguished the outstanding manager from the merely very good
manager (these were all managers who by most people’s standards had
succeeded) was their ability to use unusual analogies.
This may be connected with the notion of analogue and digital thinking
(Watzlawick et al., 1967). Their distinction is probably best understood by
thinking of analogue and digital watches. The analogue communicates in
a way that is somehow recognizable as a picture of that which it commu-
nicates. The digital communicates without the use of such pictures. The
use of gesture, posture, voice tone and diagrams in most forms of com-
plex communication bear testimony to the idea that most people value
analogue communication, and that this still applies when the material
being communicated is scientific or technical. Metaphor is essentially an
extension of Watzlawick’s analogue communication to the level of ideas,
and Klemp and McLelland could as well have referred to ‘metaphor’ as
‘analogy’ in their work.

THE IMPORTANCE OF THE PHYSICAL

Why is the use of metaphor so important in our thinking? It seems that

there are understandings that we can form on the basis of our physical
relationships with objects that cannot be formed without making use of
these relationships. We can maybe understand more of the diversity of
thinking and why people might need metaphor to convey their meaning
if we consider one of the more extreme groups of metaphor makers,
70 DIRECTLY DISCOVERING MANAGERIAL KNOWLEDGE

dancers, talking about their work. Isadora Duncan said ‘If I could tell
you what it is, I would not have danced it’ – an elegant summary of the
sentiments of metaphor makers everywhere, but all the more convincing
from a dancer because it is so clearly ‘grounded’ in the physical world of
the body. Gardner quotes the dancer Martha Graham as saying: ‘I have
often remarked on the extreme difficulty of having any kind of conversa-
tion with most dancers which has any kind of logical cohesiveness – their
minds just jump around (maybe like my body) – the logic – such as it is –
occurs on the level of motor activity’ (1993: 225). By extension, the variety
of forms of expression that we engage in may come from a deep level, and
the restriction of thinking to abstract, digital, ungrounded language may
disempower most of what Gardner refers to as our ‘multiple intelligences’.

IMAGE SCHEMAS

The theoretical perspective that links the literature on metaphor with our
technique of cognitive sculpting (Sims and Doyle, 1995) is the role of
image schemas. We have already seen the conceptually basic notion of
UP/DOWN in metaphors that connected UP with increase and DOWN with
decrease. This would be an example of an image schema. Mandler (1992)
describes image schemas as condensed redescriptions of perceptual (and
to a lesser extent motor) processes. How many such image schemas do
humans typically have? Martin (1994) estimates a few hundred, Lakoff
(1989) describes a handful, and Johnson (1987) gives a ‘partial list’ of thirty
or so, some of which he analyses in depth. They include: container, linkage,
part–whole, blockage, balance, near–far, superimposition. It is not appro-
priate to go through the whole list; instead we shall briefly describe the
first two examples to give an idea of how image schemas and metaphor
may be related.
Containers fall into three types: those that typically contain us (such as
rooms, caves); those that typically do not contain us (such as cups, bags);
and our own bodies (clearly a special case). The main features of a con-
tainer are that it has an inside, an outside and a boundary, so that objects
can be in or out of the container, and can come into the container if we are
also in it or go into it if we are not, and so on. The boundary prevents free
movement of objects.
‘Container’ is used in very many metaphors. Here are some examples
of the metaphor MIND IS A CONTAINER:

• I have an idea in mind.

• I try to put thoughts of failure right out of mind.
• It suddenly came into my mind.
• He is out of his mind.
 ENABLING STRATEGIC METAPHOR IN CONVERSATION 71

These examples also rely on the related metaphor IDEAS ARE OBJECTS (Where
did you pick that idea up from?; I dropped the idea into the conversation; We
bounced ideas off each other; He stole her best ideas) and the source-to-target
mapping between ‘in mind’ = conscious of and ‘out of mind’ = not
conscious of, as in the saying: ‘out of sight, out of mind’.1
Our second example of an image schema is link. A link connects two
objects that are a distance apart. The connection is not rigid, so that if
I move one object the other is likely to move, but not necessarily.
Therefore linkage, or connection, is often used to denote causality or rela-
tionship between two entities, where the nature of the relationship is
somewhat imprecise:

• He was linked with the Watergate scandal wasn’t he?

• Smoking has been linked to lung cancer.
• Surely there is some connection between the two events.

Look again at the first two sentences after the heading ‘Image Schemas’ on
page 70. Metaphors are ubiquitous. We suggest that metaphors based on the
link schema will be popular with all those who are concerned with the non-
obvious relationships between events, things or people. This would be true
for academics, but also for the creators of business deals, the managers of
complex enterprises, and physicians working with poorly understood
diseases in the complexity of the human body, among others.

COGNITIVE SCULPTING

We have argued that much of our knowledge is based in (or on?)

metaphor, and much of our metaphorical thinking is based in or on our
representation of objects and their relationships with each other and with
ourselves. The facility to think metaphorically or analogically is highly
valued in many fields.
If metaphor is underpinned by an abstracted understanding of objects
and our bodily relationship with them, then it may make sense to use
objects explicitly to facilitate the use of metaphor and analogy. As we have
argued, this is the essence of metaphor: we wish to talk about an abstract,
complex and difficult-to-understand domain, where common meanings
are not necessarily shared; instead of doing so directly we talk about the
complex domain as if we were talking about a concrete, simple and well-
understood domain, where common meanings are shared. Such well-
understood domains are typically connected with everyday experience in
the world of bodies and objects – an understanding we built up in the first
few years of our lives. These understandings are extremely schematic
representations of simple relationships between objects: things can be
72 DIRECTLY DISCOVERING MANAGERIAL KNOWLEDGE

near or far away, things can be inside a container or not, things can balance
or not, and so on (these are image schemas). If one wants to think about a
complex issue, one will need to enable more of one’s abilities in thinking
to be used, and this will necessitate the use of metaphor.
It is obvious that it is easier to do complicated arithmetic on paper than
in your head, or that the minutes of a meeting are a more faithful record
than our own memories, or that a complex argument is easier to follow
and review if it is written down rather than spoken. In an analogous
manner we suggest it will be helpful to have an externalized representation
of people’s metaphors to work with. Similarly, it is often helpful for these
externalized representations to attempt to capture the metaphoric rather
than the literal; the drawn map is often more helpful to the driver than a
list of instructions, even if both are only hazy memories. Even for numeri-
cal calculations, a considerable proportion of the world’s population is
happier with the metaphoric abacus than the literal pencil and paper. Our
suggestion for an effective externalized, metaphoric representation of
knowledge is cognitive sculpting, which we now describe.
In cognitive sculpting, people are involved in a conversation or discus-
sion about issues that may be complex or emotionally charged – certainly
difficult to talk about in one way or another. We have a box (toy box) to
hand that contains a collection of objects that the participants are free to lay
out on a table in front of them, ostensibly to illustrate what they are saying.
Participants are also free to adjust the arrangement of the objects in order
to make a point, or even to remove objects from the table altogether. It can
be seen that we mean ‘sculpture’ in a very broad, modern sense. The sculp-
ture is shared by all participants, who may adjust any part of it.
Let us imagine a session in full swing. Three people sit around a 2’ × 4’
coffee table which at first glance already has a number of objects haphaz-
ardly strewn across it. A closer inspection reveals a small wooden person
looking out from inside an upturned glass tumbler. The wooden person
has a piece of string tied around its foot; the string is trapped under the
tumbler, so that the person is pressed up against the glass.
A light-bulb rests on the title page of a report. (The report is actually a
highly controversial internal report). A piece of electrical wire almost
reaches the bulb, but not quite. Tracing it back we see that it is trapped
under a large block of wood. On top of the wood sits a chess piece – not
a white knight but a black one. Elsewhere drawing pins surround a crowd
of pawns. A bunch of keys sit on a chair at some distance from the table.
A credit card and a calling card from a management consultant are
propped against the keys. This is the state of the sculpture so far. It has
been gradually added to, subtracted from and otherwise altered. Here is
a reconstructed snippet of a conversation among three managers that
takes place about and through the sculpture (see Figure 3.3).

X I get into this helpless state of mind [points to the upturned tumbler], so that
I can’t do anything about the problem. I’m tied so closely to this state of
 ENABLING STRATEGIC METAPHOR IN CONVERSATION 73

Figure 3.3 Example Sculpture A

mind that I don’t even see the glass. I mean, most of the time I don’t even
see that it is a state of mind I’m in. If I weren’t so tied to it, I could at least
look away.
Y But have you ever thought that your state of mind also protects you? If
you removed the tumbler – got into a different state of mind – you might
wander around and get trapped under this weight of authority [taps the
block of wood]. Or worse...
X Yes, but someone has to connect up the bulb to the wiring and show
people what this report is really about.
Z But if you’re not careful you might find it’s you that makes the circuit and
get fried for your pains!
Y No. FIRED!
[Laughter. One of the three rummages around and comes up with a knife and
fork, which he places on either side of the light-bulb, as if setting the table for a
meal. More laughter.]
Months later these three cognitive sculptors still share the joke (insight) about
‘making the circuit’.

This interchange is, of course, highly condensed and selective. It is none

the less quite typical of the kind of conversation that takes place.
Meanings are negotiated, and the symbolic world of objects is a fertile
arena for metaphor, symbolism and reconceptions, which sometimes
emerge as jokes. By acknowledging the element of risk (fried, fired, or just
74 DIRECTLY DISCOVERING MANAGERIAL KNOWLEDGE

trapped under the weight of authority) in being a visionary (i.e. outside

the glass), and by enacting in this small world some of the visionary
processes (i.e. re-visions of interpretation), the participants are legitimizing
to each other (particularly in the use of good-natured humour) a larger
visionary activity. Since one of the most effective blocks to creativity is
self-censorship for fear of ridicule, the group is signalling, with an imme-
diacy that words alone could rarely match, that it will support creativity/
strategic thinking/change, and thus making it more likely to happen.
One may keep tangible records of cognitive sculpting by video-taping
the session. In the absence of a video a couple of instant photographs to
chart the progress of the sculpture (certainly one photograph at the end of
the session), together with an audio-tape, should be sufficient to capture
most of the significant turns of the conversation, and to enable the partici-
pants to re-create some of the insights and the knowledge from the time
of the conversation.

Choice of Objects

We have participated in, and facilitated, cognitive sculpting sessions in

which we had access to our toy box, and also in sessions in which we did
not have such access, and instead made use of objects in the immediate
locale. We have compared groups working with objects from the toy box
with groups working from a box made up of a similar number of objects
of similar size gathered from the surfaces of our offices with little consider-
ation for their symbolic qualities, and we have found that having a care-
fully selected set of objects on hand is a benefit. In any case the toy box
can always be supplemented by local objects, which, like the knife and
fork in the above sketch, may come to have particular significance.
Our toy box is continually changing through both additions and losses
as objects are reclaimed for their ‘proper’ use. But there is a stable core of
objects that we have come to know and that our research participants
seem to love.
One criterion of choice was to select objects that seemed to support
ready (perhaps even stereotypical) associations. In this category we find:

• light-bulb – inspiration, an idea;

• dice – chance, fate;
• passport – credentials, identity;
• chess pieces – especially pawn (obvious), king (boss)
and knight (unpredictable);
• chain – lack of freedom, shackles;
• tangled string – a knotty problem, etc.;
• key – solution;
• pair of glasses – seeing something, insight;
• torch batteries – energy, power;
 ENABLING STRATEGIC METAPHOR IN CONVERSATION 75

• a credit card – a variety of financial associations;

• farmyard animals – (pig, cow, bull, sheep): the usual;
• toy people – people;
• calculator – computer, the accountant;
• red bill – itself, debt, arrears of various kinds.

There were also objects that seem less stereotypical, but are none the less
constrained:

• a ruler/tape-measure;
• a spirit-level;
• old coinage;
• foreign coinage;
• stone;
• block of wood (circa 10 cm × 15 cm × 25 cm);
• a small picture frame.

As an example of the greater variety in meanings that these objects have

supported, the block of wood was often used to represent a blockage
(stereotypical); but if placed upright it also came to signify a pedestal (see
Figure 3.4), and, especially for the authors, a pedestal that people could
get trapped on top of, which is the down-side of succeeding at something
when you find afterwards that you didn’t really want to succeed at it (see
Figure 3.5). Also, because of its relative size and weight, it sometimes
represented things that were threatening, and could crush; again, some-
times it represented a vantage point to look down from.
We also chose some objects that would defy neat categorization, and,
à la Rorschach, support a variety of different projections:

• metal lattice-like grill;

• driftwood;
• a clump of melted metal rescued from a fire.

Curiously, each were often stereotyped as ‘something amorphous, hard to

understand’. Some objects were chosen to support the representation of
vigorous action within the sculpture:

• hammer;
• hacksaw;
• scissors;
• screwdriver, etc.

Their very presence in our toy box may have suggested that these tools
were the least useful in someone’s toolbox – and if he or she did not want
to use them, why should the participants want to use them? Whether for
this reason, or because the vigorous actions these tools denote just do not
76 DIRECTLY DISCOVERING MANAGERIAL KNOWLEDGE

Figure 3.4 Example Sculpture B

‘fit’ with the style of cognitive sculpting, these objects have been used less
frequently by participants than we originally anticipated.
Finally, although many image schemas can be represented by their
relationship to other objects in the sculpture (e.g. near–far, contact,
part–whole, centre–periphery, blockage, etc.), others benefited from pur-
posefully ensuring enough props to support that schema. Examples are:

• Link – chain, string, electrical wiring;

• Container – cup, tin with lid (so things could be
hidden and/or kept), perspex box
(see-through).

Other schemas that we have not properly supported but could, and
clearly should, are:

• Balance – scales, or a set of pieces that could

readily be assembled into a balance;
• Attraction – magnet, sloping surface.

We do believe that some schemas are more problematic. Source–path–

goal, for instance, so strongly implies movement as to make natural
 ENABLING STRATEGIC METAPHOR IN CONVERSATION 77

Figure 3.5 Example Sculpture C

expression within cognitive sculpting difficult. Perhaps we should add a

diagrammatic arrow? Catapult? Similarly, there are a number of other
objects that would prove useful additions to the toy box (often because
participants told us so):

• magnifying glass – looking closer;

• mirror – seeing something about yourself;
• egg timer – limited time (‘time running out’);
• plug – being ‘plugged in’;
• phone – communication.

ATTENTION

The overt objective of cognitive sculpting, as described above, is to sup-

port metaphoric and symbolic productions in conversation. In the next
two sections we investigate additional aspects of conversations that use
cognitive sculpting that may not be shared by normal conversation. In
this section we deal with attentional effects.
First, cognitive sculpting does support a different view of the con-
versation, quite literally. The general course of the conversation may be
78 DIRECTLY DISCOVERING MANAGERIAL KNOWLEDGE

comprehended by all participants at-a-glance from the objects on the

table. Thus what was said half an hour ago can exert much more of an
influence on the present flow of ideas and expression than in a normal
conversation – the extant sculpture has a mnemonic effect. We have
noticed what we call the long loop of conversations: a theme may be aban-
doned and the objects associated with it left untouched only to return
much later. We have noticed that the accompanying feeling, rather than
one of exasperation (‘I have the feeling we have been going round in
circles’), is more like a musical theme returning at the right moment. Like
the musical theme, ideas in cognitive sculpting do not often literally
repeat; they are more likely to return varied and developed, or at least to
be repeated with a different interpretive emphasis. One might say that
cognitive sculpting tends to turn the normal serial conversation into more
of a parallel one; the wandering conversation into a focused one.
The second attentional effect concerns social self-presentation. In
normal conversation one spends a lot of time and mental energy in main-
taining one’s social self to others. One tries to maintain the right amount
of eye-contact, one inspects the other participants’ faces for reactions to
one’s words, one works to signal the right expression to their words. Part
of the reason for this activity is the lack of anything else to focus on. In
cognitive sculpting there is something else to focus on. Conversation takes
place through the sculpture. One speaks about the sculpture as much as
to anyone in particular. The other participants’ attention also tends to be
focused on the hands that are doing the manipulating, rather than the lips
that are doing the talking, or the eyes that are the ‘windows to the soul’ of
the other.
We suggest that there are two freedoms won by this deflection of atten-
tion from the private face to the public sculpture. The first of these is, to
use a crude information-processing metaphor, that the effort of social
self-presentation is diverted, and one has spare capacity to think about
what is being said and done. Anyone who has given a lecture that was not
going down well will be aware of the release felt by putting a slide on.
I am no longer talking to you the audience, I am talking about it, the slide,
and your attention can be on it rather than on my face and my persona.
For the time that the slide is up we share a common focus of attention; our
gaze is united, and we are not confronting each other.
Of course, in cognitive sculpting the participants do not always focus
on the sculpture. Sometimes the conversation will revert to normal for a
while. The participants will talk over the top of the sculpture, perhaps
sitting back away from the table, with a change of posture. One notices
this as a distinct change in ‘footing’ to the conversation (Goffman, 1981).
The second way in which focusing on the objects and not the person
frees the participants has to do with the attribution process. It is known
from attribution research (e.g. summarized in Hewstone and Antarki,
1988, or Ross and Nisbett, 1991) that if a speaker is made perceptually
 ENABLING STRATEGIC METAPHOR IN CONVERSATION 79

salient in a conversation (e.g. by allowing the observers to see the

speaker’s face full on rather than seeing the speaker from behind),
observers will make more dispositional attributions to that person’s
words (espoused opinions). That is, the observers will believe the
speaker’s words are a reflection of her personality rather than a reflection
of the situation she finds herself in. The less perceptually salient the
speaker, the more her words are attributed to the situation, and not to her
personality.
In cognitive sculpting, we have argued, each speaker is made less per-
ceptually salient since the participants focus on the sculpture rather than
the speaker. This means we should expect fewer dispositional attribu-
tions. In other words, as a participant in a cognitive sculpting session,
I will not imagine that the words you say necessarily reflect enduring atti-
tudes, opinions or traits, and so I will not try to hold you to your words.
Also, we have found that a cognitive sculpture on a table seems to permit
people to have the kind of conversation they might have in an art gallery –
a sharing of interested comments in response to a physical object, which
is often very different from the purportedly rational, linear comments of
purposeful discussion, where you are expected to ‘get it right’. If the
speaker becomes aware that she is not required fully to believe everything
she says, then we can see that the conditions are right for the kind of ‘let’s
imagine’ conversation that would otherwise have to take place in a
remote conditional or subjunctive tense, and thus be difficult to sustain.
In cognitive sculpting one can try out meanings in ways that would not
normally be sanctioned.
A final benefit is that associating ideas with objects is an effective
mnemonic technique that was known to the ancient Greeks (Norman,
1976), which entails remembering things by associating them with
memory of a physical arrangement. Normally people who want to
practise the technique have to make a conscious effort to do so. However,
in cognitive sculpting it comes as a by-product of the process itself. There
is no extra effort needed to make the conversation live longer in memory,
for the ideas have already been associated with objects.

SPEAKING WITH THE MUTE HAND

In the previous section we outlined some ways in which conversations

using cognitive sculpting might differ from normal conversations. In this
section we substantiate the hypothesis that cognitive sculpting gives a
medium of expression to kinds of knowledge that are often difficult to
access and communicate.
The human brain is divided into two hemispheres. Each hemisphere of
the brain is specialized to deal with different kinds of material (Springer
and Deutsch, 1989). In the normal right-handed person, the left hemisphere
80 DIRECTLY DISCOVERING MANAGERIAL KNOWLEDGE

(which also controls the right hand) specializes in logic, language, things
analytical. Right-handed lawyers, who need to argue cases eloquently and
rigorously, have well-developed left hemispheres. The right hemisphere
(which controls the left hand) specializes in things spatial, analogical and
metaphorical. Sculptors (at least those who are right-handed) have well-
developed right hemispheres. The right hemisphere is often described as
mute, in that it does not have any control of speech. People who suffer a
stroke to the right hemisphere (identified by loss of control of the left side
of the body) rarely lose the ability to speak. If damage is to the left hemi-
sphere, speech loss can be severe or total.
In the normal person the two hemispheres are connected by a massive
bundle of fibres called the corpus callosum, which allows information to
flow between centres. Despite this connexion, it appears that communi-
cation between the hemispheres is not always unimpeded. For instance,
in The Language of Change, Watzlawick (1978) describes ‘blockages’ to the
functioning of the right hemisphere, and methods to remove the block-
ages. His general approach may be described as blocking the left hemis-
phere (e.g. through paradox) in order to free up the right hemisphere.
Cognitive sculpting, on the other hand, relies on the co-operative,
rather than competitive, use of both right and left hemispheres. Our
approach is to facilitate the flow of information across the corpus callo-
sum in both directions. At the start of a cognitive sculpting session we
could say that the participant allows the right hemisphere the indulgence
of some kind of visual-metaphorical ‘commentary’ on the conversation.
But as the session progresses a different source of control sometimes takes
over, namely the mute right hemisphere. After all, spatial relationship is
the right hemisphere’s forte. It is also possible that, outside of our normal
locus of control and awareness, the right hemisphere arranges messages
in the sculpture (perhaps the ‘accidental’ but apposite juxtaposition of
two objects) that the speaking self incorporates into its commentary. Any
facilitator of cognitive sculpting should recognize the different feel that
the functioning of each hemisphere has, and understand that it is balance
and interplay between the hemispheres, rather than overthrow of one by
the other, that is desired.
Replaying the videos of a cognitive sculpting session, one notices
the pace of speech slowing, presumably as the left hemisphere begins to
give up some control and allows the apparently accidental work of the
right hemisphere to influence the arrangement and contemplation of the
objects. The pauses lengthen; each move, as in a game of chess, seems to
require long, silent thought. Then, quite suddenly, the mode may shift as
ordinary commentary again takes over.
Insofar as we experience blockage of direct communication across the
corpus callosum, cognitive sculpting works to enhance the co-operation
of both hemispheres (speech and space) in exploring complexity through
the medium of an external display to which they can both contribute.
 ENABLING STRATEGIC METAPHOR IN CONVERSATION 81

UNANSWERED QUESTIONS

We have presented cognitive sculpting as a method of helping people to

think through issues, for example strategic questions for their organization.
In other words, our orientation has been to produce something that is help-
ful to the participants in the immediate term in the progress of their discus-
sions. We also believe that participants take longer-term gains of other kinds
away with them. At a simple level there is the experience, and the symbol-
ism that the participants share, which can resurface months later. This may
be especially significant in the discussion of strategic issues, which have
enough longevity for the development of this new, metaphoric, visually ref-
erenced language to be of real significance. We also believe that participants
who engage in cognitive sculpting are able to internalize the process, even
to the extent of performing it in an imaginary world of objects in their heads,
thus enriching their metaphorizing. After all, it is not always possible, or
indeed appropriate, to pull out a bag of objets trouvés and get to work. Those
who have had explicit experience of cognitive sculpting might also be more
willing to illustrate a point in an otherwise normal conversation with refer-
ence to objects that come to hand (a kind of ‘barefoot’ cognitive sculpting).
Or since some people do this kind of thinking with physical props naturally,
once exposed to cognitive sculpting, someone who had previously been
unreceptive might be more willing to enter into his or her ‘object world’.
Another question is whether there are ‘good’ and ‘bad’ cognitive sculp-
tors. Certainly some participants take to the process more readily than
others. Why should this be? Can we distinguish them in advance?
A further question is whether what we have described is sculpture,
theatre or something else. Perhaps we should be pleased, rather than wor-
ried, that cognitive sculpting is more than merely cognitive, and includes
dramatic, symbolic, social and sometimes entirely whimsical elements.
But labels are important. Our name for the technique does justice to its
model-building aspect, but insufficient justice to the dramatic element.
The term ‘cognitive sculpting’ sets up expectations in the participants of
how they ‘should’ behave, which might be different had we adopted
alternative terms such as ‘dramatic sculpting’, ‘cognitive theatre’, or
‘strategic knowledge sculpting’. Part of our orientation is that the objects
do remain and are important, hence the focus on the sculpture. The
objects as configured at any time in a session are meant to represent a
significant part of the conversation to date, whereas in drama the props
are just props that come and go with the scenes.
We also wonder whether the metaphor of ‘sculpture’ emphasizes sta-
bility rather than change, statics rather than dynamics. Sculptures tend to
be relatively enduring, to imply something that expresses itself without
movement. How different would the participants’ experience have been
if we had worked instead on ‘kinetic cognitive sculpture’, and empha-
sized movement rather than the stasis of objects?
82 DIRECTLY DISCOVERING MANAGERIAL KNOWLEDGE

Having a tangible representation of the session is particularly useful if

two or more groups have worked independently on a common theme. At
the end of the session each group can explain the genesis of their sculp-
ture, and receive questions and other comments. For those who have
experienced an ‘art crit’ – the session in which students and staff at an art
college gather round to discuss the work of different students or groups
of students – the resulting ruminations have something of the same
atmosphere. Long silences are tolerated that give time for mature, and
probably non-verbal, reflection.
A final question is to what extent the two authors have developed an
idiom of usage that is peculiar to themselves. For instance, other users of
cognitive sculpting have found an important role for animals in the toy
box, which we have not emphasized. As one example, the psychologist
Tina Kotzé was working on team development with a multi-racial group
in South Africa, but had failed to get the energy and commitment she
usually got with other groups. Ready to give up, she decided to have one
more try, this time with cognitive sculpting. Because of the place of
animals (as she saw it) in the life and metaphors of South Africa’s indige-
nous peoples, she introduced quite a lot of animals into her toy box. They
had a superb session, with blacks enthusing about what they could demon-
strate with the animals; whites enjoying being shown a different way of
symbolizing; blacks enjoying the fact that whites were interested in their
symbolic language; and none of the issues about who was comfortable in
which language, or who was literate, which had plagued them previously.
In a second example from another user, the executive team of a UK
company were introduced to cognitive sculpting. Early on they charac-
terized one board member, a rather tall, aloof man who happened to be
absent that day, as a giraffe. In the course of the afternoon the giraffe came
to take on a number of interconnected meanings. The absent member
aspect of the giraffe came to represent the outsider (also their more serious
selves), looking in over the fence at some very strange goings-on. Yet the
giraffe itself is rather droll, with its ridiculously long neck. So it also came
to represent the entire wacky nature of what they were doing that after-
noon, a lightness in their dealings with each other that they wanted to
encourage. The giraffe was inside each of them, as both creator and
observer of the alternative behaviours.
For a third example, Baocheng Li has used cognitive sculpting in his
work with managers in a major international airline in discussing issues
to do with the environmental effects of the airline. His work has been
different from most of what we have described in that he has been work-
ing with individual sculptors, expressing themselves to themselves and to
him as interviewer, rather than in a group. Also, he describes the sculp-
tures we have been talking about so far as ‘complex’, while he has been
working with what he calls ‘simple’ sculptures. Participants in his work
have been picking up an object from the toy box, handling it, and com-
menting on the issues he wanted to know about, i.e., they play with just
 ENABLING STRATEGIC METAPHOR IN CONVERSATION 83

one or two objects. However, the effect of having a selection of toys to pick
up and use, and of being able to hold them and play with them while talk-
ing, seems to have been very similar to the effects we have talked about
with more complex structures and with groups of participants.
The first two examples illustrate that the right symbol(s) at the right
moment can have a dramatic effect in resolving opposites. In the former,
the opposites were inter-personal (the black/white divide, compounded
by differences in language, literacy and status). In the latter, the opposites
were intra-personal (expected serious behaviour/desired fun behaviour).
The third example illustrates a way in which the core ideas of cognitive
sculpting can be the basis for many different forms of practice in permit-
ting people to talk about issues, often strategic ones, where they might
have had no way of articulating their knowledge without the sculpting
devices. Also, for all three examples, some of the knowledge that they
articulated may not have pre-existed their words; the knowledge may be
new learning that only arose in the interaction between the hands and the
object(s) during play.

COGNITIVE SCULPTING AND STRATEGIC

KNOWLEDGE

We have discussed the rationale and practice of cognitive sculpting. We

and the others mentioned in the chapter who have used the technique
have done so without differentiating the kind of knowledge being
expressed or created. Having said this, Kotzé would not have taken the
trouble to enable lengthy conversations through the metaphors of animals
if the issues under discussion had not been seen as of major significance.
Similarly, Li’s conversations with airline managers would not have been
worth having had not both he and they seen the issues of a matter of
environmental strategy.
Strategic thinking consists of preparing for an uncertain future by
rehearsing a deep reflection on the present and on ways in which the
future might unfold from the present. It consists not so much of making
plans as of rehearsing and getting to know the current world and antici-
pated worlds so as to be able to respond as a skilled and competent
improviser in whatever world then unfolds. In working circumstances
where there is always something more urgent than reflection, how might
one achieve this?
We suggest that cognitive sculpting is uniquely appropriate here. Our
futures are controlled by the metaphors we bring to them, and we have
argued that cognitive sculpting is a good way of explicating and develop-
ing metaphor. The development of a shared metaphoric language within a
group may enable such explication and development of metaphor to
happen at a group level, thus preparing the whole group for the necessary
discourse to respond to new situations with shared strategic knowledge.
84 DIRECTLY DISCOVERING MANAGERIAL KNOWLEDGE

They will have formed metaphors together and built up a history of

discussion through them, which will prepare them particularly well for
responding opportunistically together to unanticipated events in a more or
less anticipated world.
The approach implies that strategic knowledge is not solely cognitive, but
employs what Gardner (1993) would term ‘other intelligences’. However,
no-one who has ever felt the urge to pace up and down, to move their
hands, or to draw a diagram while contemplating strategic issues would
dispute this; cognitive sculpting is perhaps practised more widely than is
often acknowledged, and part of our purpose in this chapter is to argue
for making more purposeful and explicit use of something that is often
used less deliberately and formally.
We have argued that metaphor is a way of talking about complex
domains in terms of simpler, better understood domains, which are
usually concrete. We have described and related our technique of cogni-
tive sculpting to metaphor, via the intermediate concept of image
schemas. We have examined work on the separation of functions within
the cerebral hemispheres, and related the activity of cognitive sculpting to
them. We suggest that cognitive sculpting is a device to help integrate the
two hemispheres, and to bring the metaphorical into a profitable dialogue
with the analytical. It also brings a more playful aspect to discussions that
can get too serious for their own good, and enable people to articulate
futures otherwise unspoken.

NOTE

1. It is interesting to note the change in sense when the related metaphor is PERSONS
ARE OBJECTS, and hence could be contained by their minds, or alternatively be regarded as
dangerously unbounded when they are ‘out of their mind’.

REFERENCES

Gardner, H. (1993) Frames of Mind. 2nd edn. London: Fontana.

Goffman, E. (1981) Forms of Talk. Oxford: Blackwell.
Hewstone, M. and Antarki, C. (1988) ‘Attribution theory and social explanation’, in
M. Hewstone, W. Stroebe, J.-P. Codol and G.M. Stephenson (eds), Introduction to Social
Psychology. Oxford: Blackwell.
Johnson, M. (1987) The Body in the Mind. Chicago: University of Chicago Press.
Klemp, G.O. and McClelland, D.C. (1986) ‘What characterizes intelligent functioning among
senior managers?’ in R.J. Sternberg and R.K. Wagner (eds), Practical Intelligence.
Cambridge: Cambridge University Press.
Lakoff, G. (1989) ‘Some empirical results about the nature of concept’, Mind and Language,
4: 103–129.
Lakoff, G. and Johnson, M. (1980) Metaphors We Live By. Chicago: University of Chicago
Press.
Mandler, J. (1992) ‘How to build a baby: II. Conceptual primitives’, Psychological Review,
99: 587–604.
 ENABLING STRATEGIC METAPHOR IN CONVERSATION 85

Martin, J.H. (1994) ‘MetaBank: A knowledge-base of metaphoric language conventions’,

Computational Intelligence, 10 (2): 134–149.
Miller, G.A. (1979) ‘Images and models, similes and metaphors’, in A. Ortony (ed.), Metaphor
and Thought. Cambridge: Cambridge University Press.
Miller, R.N. (1976) ‘The dubious case for metaphors in educational writing’, Educational
Theory, 26: 174–181.
na Gopaleen, M. (1977) The Best of Myles. London: Picador.
Norman, D.A. (1976) Memory and Attention. New York: Wiley.
Ortony, A. (1975) ‘Why metaphors are necessary and not just nice’, Educational Theory,
25: 45–53.
Pinder, C.C. and Bourgeois, V.W. (1982) ‘Controlling tropes in administrative science’,
Administrative Science Quarterly, 27: 641–652.
Pollio, H.R., Smith, M.K. and Pollio, M.R. (1990) ‘Figurative language and cognitive
psychology’, Language and Cognitive Processes, 5 (2): 141–167.
Ross, L. and Nisbett, R.E. (1991) The Person and the Situation: Perspectives of Social Psychology.
New York: McGraw-Hill.
Sims, D.B.P. and Doyle, J.R. (1995) ‘Cognitive sculpting as a means of working with
managers’ metaphors’, OMEGA, 23 (2): 117–124.
Springer, S.P. and Deutsch, G. (1989) Left Brain, Right Brain. 3rd edn. New York: W.H.
Freeman.
Tsoukas, H. (1991) ‘The missing link: A transformational view of metaphors in organiza-
tional science’, Academy of Management Review, 16 (3): 566–585.
Watzlawick, P. (1978) The Language of Change. New York: Norton.
Watzlawick, P., Bavelas, J.B. and Jackson, D.D. (1967) Pragmatics of Human Communication.
New York: Norton.
INFERRING
MANAGERIAL
KNOWLEDGE
II
4
G A I N I N G U N D E R S TA N D I N G
IN A COMPLEX
CAUSE–EFFECT POLICY
DOMAIN

Roger I. Hall

ABSTRACT

An artificial intelligence (AI) program that contains simple behavioural rules is

used in this study as a surrogate for the socio-cognitive and socio-political pro-
tocols employed by a group to make sense collectively of a complex system of
cause–effect. A formal corporate cause–effect map of the operations of a sports
club is used as a surrogate for the ‘real-world’ complex web of cause–effect that
converts policy decisions into outcomes affecting corporate survival goals.
Group policy making is mimicked by running trials with the AI program on the
causal map of the sports club to see what policies are formed and why.
The resulting policies compared favourably to those adopted by the executive
of the club at two different periods in its history. The resulting discussion focuses
on: (a) employing formal corporate cause–effect maps to improve group policy
making; (b) the missing feedback causal paths in the collective map of the group;
(c) whether managers can be trained to handle more causal complexity; (d) why
different starting beliefs might affect the future histories of identical organizations
in the same industry; and (e) the puzzle of exceptional organizations.

INTRODUCTION: CAUSE–EFFECT
COMPLEXITY AND POLICY MAKING

Speculating and arguing about the causal influences determining

successful outcomes comprises an important part of the work of policy
groups empowered to make weighty decisions committing the resources of
an organization. However, the true system of causal relations at work in
90 INFERRING MANAGERIAL KNOWLEDGE

determining the outcomes of policies is rarely known to the group in

complete detail; particularly if the causal relations perceived by the manage-
ment have been learnt from outcome-irrelevant learning structures (Einhorn
and Hogarth, 1978), or have been subjected to judgement biases (Kahneman
and Tversky, 1982) or conditioned beliefs (hypothesized cause–effects vali-
dated solely on the basis that policies based on them worked in the past), or
have been selectively biased to support the cause of the dominant coalition
for internal political reasons (Pettigrew, 1973), or have just gone unseen.
Even if the group has a good understanding of the causal relationships or
has commissioned experts to uncover the ‘true’ relationships, the map may
still be complicated by multiple chains of causality (many causal arguments)
from policy decision to outcomes, indeterminacy (opposing arguments about
the effects of the causal chains) and feedback loops (confounding circular
chains of causality imparting dynamic properties to outcomes).
Even quite simple organizational operating systems can give rise to a
relatively large number of chains of cause–effect paths from policy vari-
ables to organizational goals and many potentially confounding feedback
loops. For example, the rudimentary formal map to be used in this study
was found to contain 24 direct causal influences among 15 active variables
representing chains of cause–effect in the operations of a sports club.
Two of the active variables were identified by club policy makers as
beginning policy variables under their aegis, and two other variables
were likewise identified as ending outcome survival-goal variables. On
analysis with a Path-Finding programme, the map was found to contain
21 chains of influences (paths) from beginning policy to ending goal
variables (13 chains containing arguments for increasing the value of the
policy variables and 7 arguing the opposite) and 8 feedback loops.
The executive committee of the club was comprised of volunteer
members enamoured of the sport, but with limited comprehension of the
causal influences and limited resources to unravel the causal system.
Under such circumstances, how could they gain an understanding of the
cause–effect system on which to base their policy decisions?

HOW MANAGERS COLLECTIVELY DEAL

WITH CAUSAL COMPLEXITY USING
SIMPLE BEHAVIOURAL RULES

In order to make the policy process manageable, policy-making groups

will resort to activities of a socio-political nature for collectively negotia-
ting what is to be believed and accepted (Eden, 1990), including the ostra-
cizing of apostates (Steinbrunner, 1974). A kind of administrative natural
logic (Hall, 1981, 1984, 1989 and 1990) is applied by the policy-making
group to simplify the emerging collective map so as to avoid equivocality
(Weick, 1969), uncertainty (Cyert and March, 1992) and conflict (Coser,
1964), and to attempt to correct unsatisfactory performance of the corporate
 A COMPLEX CAUSE–EFFECT POLICY DOMAIN 91

survival goals and, at the same time, satisfy subunit goals (Cyert and
March, 1992). This process usually results in some kind of a simple shared
cognitive map of the major causes and their effects thought to be active
within and between the organization and its significant environments;
complete with attendant biases and misattributions. Policy problems are
then inferred from the movements of key performance indices that result in
identifiable performance gaps. Policies can then be invoked from the shared
cognitive map to rectify or take advantage of the perceived problems.
The notion that programs based on simple rules can be built to think
and act like people is borrowed from the field of artificial intelligence
(Newell and Simon, 1966; Schank and Colby, 1973). From this perspective
it is argued that the group processes for simplifying the complex cause–
effect relations for policy making can be represented by a limited set of
interlocking behavioural rules, such as the following:

• Policy elites will usually agree on the key policy variables, such as
prices and promotion expenditure, directly affecting policy outcomes,
such as revenue growth and profit (after Roberts, 1976).
• Policy makers readily accept pairs of influence relations (simple argu-
ments of direct cause and effect of one variable on another – e.g. pro-
motion expenditure influences sales positively [Axelrod, 1976]).
• Policy makers formulate arguments from the pairs of causal relations to
connect policy variables to their effects on goals (after Roberts, 1976).
• Policy makers’ cause–effect maps are characterized by a few short
determinate causal paths. For example, higher prices lead to higher
revenue – the more complicated causal argument linking loss of cus-
tomers to higher prices is ignored (after Axelrod, 1976).
• The political entities within an organization, such as the departments
of marketing, operations and finance, will each identify and pursue
their own status-enhancing goals, such as the number of customers,
volume of production and total revenue (after Pettigrew, 1973).
• The top political entity of the organization, such as the governing body
or board of directors, will focus its policy-making efforts primarily on
corporate survival goals, such as revenue growth and net revenue or
profit.
• Each political entity will form its own preferred policies – the marketing
department may seek to increase the promotion budget (after
Pettigrew, 1973).
• The top political entity will form its preferred corporate policies
(increased promotion expenditures to increase revenues and increased
prices to increase profits).
• As corporate problems arise, each political entity will seek to use the
situation to pursue its preferred policies or will be faced with defending
its status (after Pettigrew, 1973). Corporate problems are defined by
unsatisfactory performance of the survival goals, usually associated
with growth, profit or both together, or by their over-achievement
(how to absorb the slack in the system; after Cyert and March, 1992).
92 INFERRING MANAGERIAL KNOWLEDGE

• Where there is consensus of all political entities on a policy action to

overcome a corporate goal shortfall (such as increasing prices to shore
up a sagging profit), it will be implemented.
• Where there is not consensus, the dominant entity or coalition of enti-
ties will unilaterally implement policies to protect the corporate goals.
This will give rise to tension and conflict within the organization.
• Where there is an impasse (conflicting corporate policies), a ‘sequen-
tial attention to goals’ rule is applied – for example, first increase pro-
motion expenditures to bolster revenue growth and, second, increase
prices to shore up profit (after Cyert and March, 1992).

These rules for group policy making can be likened to decision-making

protocols stored in the brain to interpret its physical domain and make
decisions on how to move the body to a new location. However, the mind
requires the body to enact its decisions and some knowledge of how the
body can be expected to react to its decisions. In like fashion, the group
‘mind’ requires policy-making protocols to interpret its policy domains
and some knowledge, such as a cause–effect map, of how it can be
expected to react to its policy decisions. The artificial intelligence (AI) that
I have programmed to mimic group policy making similarly requires a
corporate system cause–effect map or model. The AI treats the cause–
effect map as its policy territory upon which it applies its protocols. In
effect, the map acts as a surrogate for the real-world complex of causal
relationships, and the AI acts as a surrogate for the socio-cognitive and
socio-political rule-based processes employed by a group of policy ma-
kers as it attempts to fathom the causal system and make decisions to
move the organization to some new state. The question is whether this
rule-based approach will generate policies that are similar to those
observed in the real organizations, and what insights this might afford us.

HOW THE BEHAVIOURAL RULES EMBEDDED IN

AN ARTIFICIAL INTELLIGENCE PROGRAM
EMULATE MANAGEMENT POLICY MAKING

The prototype AI is programmed in HyperCardTM and HyperTalkTM as

a Graphic User Interface program for the Macintosh computer.
Reprogramming in VisualWorksTM for the MacOS, Windows and Unix
environments is currently in progress. The processes programmed in
the AI are designed to read in the corporate system cause–effect map
in mathematical representational form and to work interactively with a
user conversant with the organization that supplies information about
departments, their goals and their current beliefs in policy-to-goal causal-
ity. The AI program then synthesizes a simplified group cause map and
analyses it to form policies for certain problem situations such as poor
revenue growth or financial loss, but in a way that emulates the manage-
ment group decision processes. The behavioural rules listed above are
 TABLE 4.1 A summary of the behavioural rules and matching AI policy
inference procedures used to mimic group policy-making behaviour
Behavioural rules AI procedures
[What cause-effect map of an organization is to be studied?] Prompt the user to select the file containing the equations representing the organization’s
cause-effect map. Import the file. Parse the equations.
Policy elites can usually agree on the key policy variables (e.g. prices and promotion Form an alphabetic list of all variables in the equations representing the corporate
expenditure) directly affecting financial results (Roberts, 1976). system. Prompt the user to select those to be categorized as policy variables.
Organizations are made up of political entities (e.g. departments of marketing, Prompt the user to name the dominant political entities (departments, divisions, groups,
operations and finance) each pursuing its own status-enhancing etc.) and to select the goal variable for each entity from the alphabetic list.
goal (after Pettigrew, 1973).
The governing body of the organization (board of directors) will be concerned primarily Prompt the user to select the corporate survival goal variables from the alphabetic list.
with the survival goals of growth or decline and profit or loss.
Policy makers readily accept pairs of influence relations (simple arguments of direct Parse the equations representing the corporate system into bivariate correlational influence
cause and effect of one variable on another; e.g., promotion expenditure relations of the form: variable x influences variable y positively or negatively.
influences sales positively [Axelrod, 1976]).
Policy makers formulate arguments from the pairs of causal relations to connect policy Launch a path-finding algorithm to find all paths of causal relations from each policy
variables to their effects on goals (after Roberts, 1976). variable to each goal by concatenating the bivariate relations and computing their total
path correlations.
Policy makers’ maps are characterized by a few short determinate causal paths (e.g. Select the two shortest paths of opposing policy representing opposing arguments of the
higher prices leads to higher revenue – the more complicated causal argument effect of each policy variable on each goal. The user is prompted to select one
linking higher prices to loss of customers is ignored [after Axelrod, 1976]). of these argument as the current belief.
Each political entity will form its own preferred policies (e.g. the marketing Form a table of the desired movement of each policy variable (increase, decrease or
department may seek to increase the promotion budget [after Pettigrew, 1976]). indifferent) to favour the goal of each political entity.
As corporate problems arise, each political entity will seek to use the situation Form a table of the movements of each policy variable (increase, decrease, indifferent,
to pursue its preferred policies or will be faced with defending its status impasse) for solving the corporate problems. Produce an index of dissension (e.g.
(after Pettigrew, 1973). Corporate problems are defined by unsatisfactory number of dissenting departments) for each policy. Choose the policies where
performance of the survival goals of growth, profit or both together, or by their consensus is found (no dissent), otherwise select the policies most favourable
over-achievement (i.e. how to absorb the slack in the system; after or least damaging to the dominant group (political entity).
Cyert and March, 1992).
Where there is an impasse (conflicting corporate policies), use a sequential Select a revenue growth policy first and a profit-correcting policy next.
attention to goals (after Cyert and March, 1992).
Source: Reprinted from Hall, R.I., ‘A Study of Political Formation in Complex Organizations’, Journal of Business Research, 45(2): 162 (1999) with permission from Elsevor Science.
94 INFERRING MANAGERIAL KNOWLEDGE

incorporated in the AI as matching procedures, as shown in Table 4.1 and

described next. These procedures constitute the ‘guts’ of the AI program.
The user is first prompted to select the corporate cause–effect system to
study. The map must be in a suitable mathematical equation form of
either a set of assignment equations representing an Axelrod (1976) or
Eden-type (Eden and Huxham, 1988) cognitive map, or a set of time-
difference equations representing a System Dynamics model (Hall,
1976/1996).1 A parsing routine in the AI converts the equations into (a) a
list of variables in alphabetic order, and (b) a list of bivariate links (pairs
of variables) in the form: variable x influences variable y positively or nega-
tively in a direct causal sense.
Next, the user is prompted to name the political entities (e.g. depart-
ments, divisions, influential coalitions, boards of directors and executive
groups) and the dominant goal of each. The user then is asked to identify
the policy variables and the goals of each political entity from the alpha-
betic list of variables provided by the program. The AI program then
searches the cause–effect relations to find paths made up of the bivariate
links starting from those variables named by the user as policy variables to
those named goals. A depth-first with backtracking path-finding algorithm
(after Tarjan, 1972) is used to identify all the paths (representing lines of
arguments) from the named policy variables to the named goals.2 It also
finds the feedback loops embedded along the paths, but since these are
usually oblivious to the management (Axelrod, 1976), they are not used in
the subsequent AI process for forming policies.3
The AI program then identifies the two shortest paths of opposing
cause–effect as representing the simplest counter-arguments for how each
policy variable effects each goal variable. These are candidates
for policies to remedy unsatisfied goals. Where two paths of opposing
influences exist, the user is asked to choose one as representing the
current causal belief of the policy-making group, perhaps learnt from
prior environment conditioning, or based on current preference
for socio-political reasons.4 From these paths, the desired policies of each
department are formed (which policy variables to change and in which
direction). In this way, the AI mimics, albeit in a simplistic fashion, the
socio-psychological and socio-political processes that groups employ to
make sense of complex and potentially conflictful situations. In the next
section, a simple corporate system cause–effect map of a sports club will
be described. It will be used to illustrate how the AI program works and
to compare simulated policies with those adopted by the club’s executive.

A CAUSE–EFFECT SYSTEM MAP OF A SPORTS CLUB

The corporate cause–effect relations are represented in three different, but

corresponding, ways: (a) a directed digraph arrow diagram; (b) a set of
assignment equations; and (c) a verbal description.
 A COMPLEX CAUSE–EFFECT POLICY DOMAIN 95

Policy
Intervening variables Goal variables
variables causal
belief +
Quality Application
0 or +
of + rate +
Quality service
of causal − No. of
No.
management belief employees + of
0 or + Cancellation per member
rate Members
+
−
causal − Budget
belief
Quality for
0 or +
of wages
ice + +
Bar and + Raffle
causal belief
food net
0 or + profit
revenue
causal
belief Costs
+
0 or + +

Annual Revenue −
−
fee from
− renting
+ Total
profit
+ Membership + +
revenue

Figure 4.1 An Axelrod-type mapping of the system of cause–effect of the

operations of a sports club (adapted from Hall and Menzies, 1983/1990)

(a) Directed Digraph Representation

The relations are depicted diagrammatically in Figure 4.1 by an Axelrod

arrow-type of cause–effect system map. The map is a simplified version
taken from a previously published study of a failing prestigious curling
club approaching its 100th anniversary (Hall and Menzies, 1983/1990).
Figure 4.1 was constructed from interviews with the club’s policy
elite backed up by club documentation such as membership and
accounting statements. (For a detailed description of the corporate
system of influences, see Hall and Menzies, 1983/1990.) Referring
to Figure 4.1, the survival goals of the curling club, as defined by the
club’s executive, are to maintain the NumberofMembers and make
enough TotalProfit to remain viable. The policy decision variables under
the aegis of the club executive were identified as AnnualFee and
QualityofManagement.
96 INFERRING MANAGERIAL KNOWLEDGE

TABLE 4.2 Assignment equations and documented

explanations representing the Axelrod-type
mapping of the sports club system of Figure 4.1
Policy Variables
AnnualFee=−TotalProfit
{annual membership fee is negatively influenced by the total profit, i.e. when profit declines,
fees are increased}
QualityOfManagement=1
{quality of club management is a variable changed by executive fiat and is not influenced by
other variables, i.e. it is an exogenous variable in this conceptualization on the club opera-
ting system}
Intervening Variables
ApplicationRate=+NoOfMembers
{application rate of new members is positively influenced by the number of current
members, i.e. club members recruit more new members}
CancellationRate=[0 or +]AnnualFee−QualityOfIce−QualityOfService
{the number of members cancelling their membership is influenced negatively by quality of
ice and services and [not to be influenced or to be influenced positively] by fees, i.e. lower
club quality will cause more members to quit and higher fees [will or will not] cause more
members to quit}
QualityOfService=[0 or +]QualityOfManagement+NoOfEmployeesPerMember
{quality of services is influenced [not at all or is positively] by quality of club management
and is influenced positively by the number of employees relative to members, i.e. quality of
services is perceived to have [either no effect or to increase] with better management and/or
a better ratio of service employees to number of club members}
QualityOfIce=[0 or +]QualityOfManagement
{quality and consistency of the curling rink ice surface [is not or is] directly related to the
quality of club management, i.e. good management [will not or will] ensure higher-quality
conditions for playing the game of curling}
NoOfEmployeesPerMember=−NoOfMembers+BudgetForWages
{the ratio of employees to club members is reduced by a bigger membership and/or
increased by a larger budget for wages}
BudgetForWages=+BarAndFoodNetRev
{monies available for wages are directly related to revenue from the bar and food services}
BarAndFoodNetRev=[0 or +]QualityOfManagement+NoOfMembers
{net revenues from bar and food services [are not or are] directly influenced by the quality of
the club management, and/or are directly influenced by the number of members, i.e better
management [will or will not] increase the net profit from the bar and food facility, and
more members will result in higher usage and hence a higher profit from the food and
bar services}
RaffleProfit=+NoOfMembers
{more members will increase the profit from the annual raffle}
RevenueFromRenting=−NoOfMembers
{revenues from renting out the facilities to outsiders are negatively influenced by the current
number of members, i.e. fewer members leads to less use of the facilities and more time
available for leasing it out, and vice versa}
MembershipRevenue=+NoOfMembers+AnnualFee
{revenues from membership fees are directly related to the current number of members and
the level of fees they are paying}
(Contd.)
 A COMPLEX CAUSE–EFFECT POLICY DOMAIN 97

TABLE 4.2 Contd.

Costs=[0 or +]QualityOfManagement
{costs of running the sports side of the club are mostly overhead expenses [not related or
directly related] to the quality of management}
Goal Variables
NoOfMembers=+ApplicationRate−CancellationRate
{the current number of members is increased by new members joining and decreased by
those quitting}
TotalProfit=+RaffleProfit+BarAndFoodNetRev+RevenueFromRenting+MembershipRevenue−
Costs
{total profit is increased by profits from the annual raffle and the bar and food services, and
by revenues from renting out facilities and membership fees, and decreased by costs for
running the sports facilities}
Legend
y=x+z denotes x & z both influence y positively (in a correlational sense).
y=+x−z denotes x influences y positively & z influences y negatively.
y=1 denotes that y is not influenced by any other variables in the map.
y=[0 or +]x denotes that x can either have no (zero) influence on y or x can influence y
positively. The sign ‘0’ will, in effect, remove (cut) the the influence of x on y.

(b) Assignment Equations Representation

Table 4.2 shows the same relationships as a set of documented symbolic

assignment equations. These assignment equations are in the form of:

the variable on left of the ‘=’ sign is influenced by the variable(s)

on the right side either positively (+) or negatively (−).

That is, if the influence is ‘+’, this means that an increase in the value of
the variable on the right side will bring about an increase in the value of
the variable on the left, and vice versa. And if the influence is ‘−’, this indi-
cates the reverse, namely an increase in the value of the variable on the
right will cause a decrease in the value of the variable on the left, and vice
versa. To simulate the management’s blindness to a particular influence,
a ‘0’ sign is used to zero it; in effect, to cut the cause–effect link. The AI
requires these equations in order to compute the paths from policy vari-
ables to goals.

(c) Verbal Description Representation

The verbal descriptions of the cause–effect relations are contained in the

documentation following each equation in Table 4.2. The documentation
details the causal arguments for each equation and its associated directed
arrow of influence in Figure 4.1 and is intended to aid the reader in
following the skein of causal influences throughout the map from policy
variables through intervening variables to goals.5
98 INFERRING MANAGERIAL KNOWLEDGE

TRIALS WITH THE AI AND CORPORATE SYSTEM

MAP OF THE SPORTS CLUB

The AI program imports the data file containing the assignment equations
of Table 4.2, after prompting the user for its name and location. By
following the routines outlined in Table 4.1, the AI produces an alphabetic
listing of all variables and prompts the user to identify the policy and goal
variables from this list. AnnualFees and QualityOfManagement were chosen
for the policy variables and TotalProfit and NumberOfMembers were chosen
for goals in accordance with Table 4.2. Trials with the AI program were then
run for two levels of comprehension by the club executive of the complexity
of the influence relations represented in Table 4.2 and Figure 4.1.

Elementary Level of Comprehension

At first, because of the past success of the club, the executive committee
demonstrated little understanding of the direct effect of the AnnualFee and
the indirect effect of QualityOfManagement on the CancellationRate of
members. Traditionally, when a financial loss was incurred, the manage-
ment raised the annual fee. The missing influences are treated as causal
beliefs to which the club’s executive was blinded in the first instance. In
Table 4.2 and Figure 4.1, these causal beliefs affect: (a) the influence of the
AnnualFee on CancellationRate of club members, and the influences of
QualityOfManagement on (b) QualityofServices, (c) the QualityOfIce and
(d) the Costs of running the club. When asked by the AI program whether
any changes to the causal relations are required, the above influences
were given a zero assignment value that, in effect, negates their effects
and severs the links. From the memory of former executive committee
members, it appears that the concept of QualityOfManagement and its pos-
sible effects never came up for conscious discussion at executive commit-
tee meetings at that time.
When prompted by the AI program to enter the names of political enti-
ties within the club, there was only one political entity identified – namely
that of the president and executive committee of the club – due to the
small size of the organization.6

Results of the Trial with the Elementary

Level of Comprehension

The results of running the AI program with the modified cause–effect

relations are shown in Tables 4.3 and 4.4.
In Table 4.3 only one causal path was identified by the AI program,
namely AnnualFee influences MembershipRevenue, which in turn influences
TotalProfit. It is a path of positive influence from the policy variable
 A COMPLEX CAUSE–EFFECT POLICY DOMAIN 99

TABLE 4.3 Results of the AI policy inference procedure:

Trial 1: Policy-to-goal cause–effect paths
found with the elementary comprehension
level of cause–effect beliefs
to corporate Perceived effect of No. of links in
Causal path from survival goal policy on goal along causal path
policy variable variable causal path
AnnualFee NoOfMembers Zero N/A
AnnualFee TotalProfit Positive 2
QualityofManagement NoOfMembers Zero N/A
QualityofManagement TotalProfit Zero N/A

TABLE 4.4 Results of the AI policy inference procedure:

Trial 1: Desired policies for improving the
achievement of club goals derived with the elementary
comprehension level of cause–effect beliefs
Corporate survival goals variables
Desired direction of policy Desired direction of policy
variable to increase goal of variable to increase goal of
Policy variables NoOfMembers TotalProfit
AnnualFee No connection seen Increase
QualityOfManagement No connection seen No connection seen

AnnualFee to the goal variable of Profit; that is, an increase in AnnualFee will
cause an increase in TotalProfit (ceteris quietus – all other influences remain-
ing dormant). There are no causal paths from QualityOfManagement to
TotalProfit or NoOfMembers. Cutting all the links emanating from the policy
variable QualityOfManagement in effect made them unseen in the emulated
collective map of the club executive committee.7
These relationships are translated by the AI program in Table 4.4 into a
desired policy to increase AnnualFees to improve the goal TotalProfit, since
the causal path from AnnualFee to TotalProfit has a ‘positive’ correlation.
There is an indifference to all other available policies. This is exactly the
policy adopted by the club executive at that time to correct financial losses.
These simple and straightforward results might be considered intuitively
obvious. However, one has to take into account the comprehension of the
committee at that time (its collective map) of the complexity of the ‘real’
cause–effect system. For larger organizations with several divisions or
departments and more convoluted and complex maps of causal relation-
ships, the results are not so intuitively obvious. The simple club example,
however, does serve to illustrate how the AI program works.

Trials with a Higher Level of Comprehension

Later, in response to the declining membership, brought about by high

fees and a critical financial loss, a new executive committee of the club
100 INFERRING MANAGERIAL KNOWLEDGE

undertook a study of the situation. It led to a better understanding of the

effects of QualityOfManagement and AnnualFee on membership decline
and financial viability. To reflect this improved understanding, the causal
beliefs previously cut were restored. The AI program now found multiple
paths from policy variables to goals. Many of the paths exhibit indetermi-
nacy, representing counter-arguments for raising or lowering the values of
the policy variables to enhance the goals.

Resolving Path Indeterminacy

It is not now apparent whether a policy variable influences a goal variable

positively or negatively, in a causal correlational sense. The AI program
deals with this by listing the two shortest paths of opposite cause–effect
linking each policy variable to each goal. The pairs of paths are listed in
Figure 4.2 and represent the two simplest causal arguments of the effect
of each policy variable on each goal: one arguing to increase the policy
variable to increase the goal variable and the other to decrease the policy
variable to increase the goal variable. This is where the user’s knowledge
of the particular situation is required.
To resolve this indeterminacy, the user is requested to choose
which one of each pair of paths will best represent the current beliefs
in causality of the policy-making group. For the #1 causal paths (in
Figure 4.2) from AnnualFee to TotalProfit, the shortest path of positive cor-
relation (AnnualFee, MembershipRevenue, TotalProfit) is now countered by a
path of negative correlation (AnnualFee, CancellationRate, NoOfMembers,
BarAndFoodNetRev, TotalProfit). The latter path is predicated on the argu-
ment that higher fees will increase member cancellations, leading to fewer
club members and reduced revenues from the bar and food services and
lower profit. It has more causal links and so might be more difficult to
argue in an executive meeting without strong supporting evidence to
counter the more direct and persuasive accounting logic of higher fees
bringing in more revenue and profit. The study conducted by the club
executive indicated that higher fees had increased the cancellation rate
of members to the point where it exceeded the application rate of
new members, thus causing the observed decline in the number of club
members. This leads us to the choice of the second (negative) path as rep-
resenting the new belief of the group.
For the #2 causal paths (Figure 4.2) from AnnualFees to NoOfMembers,
the AI could find no paths of positive correlation since increasing
AnnualFee leads only to an increase in CancellationRate, which, in turn, has
a negative causal correlational effect on NoOfMembers. So, by default, the
path of negative causal correlation is chosen.
For the #3 causal paths (Figure 4.2) found from QualityOfManagement to
TotalProfit, the shortest path of positive correlation (QualityOfManagement,
BarAndFoodNetRev, TotalProfit) is now countered by a path of negative
 A COMPLEX CAUSE–EFFECT POLICY DOMAIN 101

#1 Causal Paths: from AnnualFee to TotalProfit Select one

either
Path: AnnualFee, MembershipRevenue, TotalProfit
No. of links: 2 Path causal correlation: positive

or
Path: AnnualFee, CancellationRate, NoOfMembers, BarAndFoodNetRev, TotalProfit
No. of links: 4 Path causal correlation: negative

#2 Causal Paths: from AnnualFee to NoOfMembers Select one

either
Path: none
No. of links: 0 Path causal correlation: positive

or
†
Path: AnnualFee, CancellationRate, NoOfMembers
No. of links: 2 Path causal correlation: negative
†
chosen by default

#3 Causal Paths: from QualityOfManagement to TotalProfit Select one

either
Path: QualityOfManagement, BarAndFoodNetRev, TotalProfit
No. of links: 2 Path causal correlation: positive

or
Path: QualityOfManagement, Cost, TotalProfit
No. of links: 2 Path causal correlation: negative

#4 Causal Paths: from QualityOfManagement to NoOfMembers Select one

either
Path: QualityOfManagement, QualityOfService, CancellationRate, NoOfMembers
No. of links: 3 Path causal correlation: positive

or
Path: QualityOfManagement, Cost, TotalProfit, AnnualFee,
CancellationRate, NoOfMembers
No. of links: 5 Path causal correlation: negative

Figure 4.2User responses requested by the AI program to solve

indeterminancy found in the opposing cause–effect path correlations
The shortest policy-to-goal paths of opposing correlation, found by the AI
program, are displayed above. These represent the simplest arguments for
raising or lowering the value of each policy variable to enchance
each goal.
For each pair of policy-to-goal paths, select the path that best reflects the
group’s belief in the causal correlation (either positive or negative).
102 INFERRING MANAGERIAL KNOWLEDGE

correlation (QualityOfManagement, Cost, TotalProfit). The executive

committee came to understand that a good club manager could guaran-
tee to manage the club’s services efficiently, to contain costs and turn the
loss on Bar and Food operations into a profit. So the positive path corre-
lation was chosen to represent the new belief in the effect of
QualityOfManagement on TotalProfit.
For the #4 causal paths (Figure 4.2) found from QualityOfManagement
to NoOfMembers, the shortest path of positive correlation (QualityOf-
Management, QualityOfService, CancellationRate, NoOfMembers) is now
countered by a path of negative correlation (QualityOfManagement, Cost,
TotalProfit, AnnualFee, CancellationRate, NoOfMembers). The executive com-
mittee also came to the understanding that good club management could
provide both improved services to members and consistent ice quality for
the game of curling. It was argued that these improvements would fur-
ther stem the loss of members. So the path of positive correlation linking
QualityOfManagement to NoOfMembers was chosen to represent their
belief. As mentioned previously, these choices to resolve the indetermi-
nacy in the AI program’s causal map of the operations of the club require
a person with local knowledge of the situation.

Results of the Trial with the Higher Level

of Comprehension

Once all the policy-to-goal paths have been cleared of indeterminacy, the
AI program summarizes in Table 4.5 the now deterministic paths as nega-
tive influences of policy variables AnnualFee and positive influences of
QualityOfManagement on the TotalProfit and NoOfMembers of the club.
These relationships are translated by the AI program (shown in Table 4.6)
into a desired policy to decrease AnnualFees to improve the goals TotalProfit
and NoOfMembers, and, likewise, to increase QualityOfManagement to
improve the goals TotalProfit and NoOfMembers. There is, also, no impasse
since the policies for increasing membership and profits are the same. As
described in Hall and Menzies (1983/1990), policies identical to these were
adopted and vigorously pursued by a new executive that took over the
running of the club, which adds further to the credibility of the AI program
in emulating the logic used by a group in policy making.

FEEDBACK LOOPS OF CAUSALITY

Lastly, in Table 4.7, the AI lists the feedback (circular) paths of cause–effect
buried in Figure 4.1. These loops can create unintended outcomes if they
go unnoticed. The positive loops can lead to uncontrolled growth or
decline resulting in instability and loss of control. For example, for Path
#1 in Table 4.7, any increase in the NoOfMembers, through its effect on
 A COMPLEX CAUSE–EFFECT POLICY DOMAIN 103

TABLE 4.5 Results of the AI policy inference procedure:

Trial 2: Policy-to-goal cause–effect paths with the higher
comprehension level of cause–effect beliefs
Causal path from policy to corporate Effect of policy on No. of links in
variable survival goal variable goal along causal path causal path
AnnualFee NoOfMembers Negative 2
AnnualFee TotalProfit Negative 4
QualityOfManagement NoOfMembers Positive 3
QualityOfManagement TotalProfit Positive 2

TABLE 4.6 Results of the AI policy inference procedure:

Trial 2: Desired policies for improving the achievement
of club goals derived with the higher comprehension level
of cause–effect beliefs
Corporate survival goal variables
Desired direction of policy Desired direction of policy
variable to increase goal of variable to increase goal of
Policy variables NoOfMembers TotalProfit
AnnualFee Decrease Decrease
QualityofManagement Increase Increase

increasing the ApplicationRate, will, in turn, further increase NoOfMembers,

resulting in an exponential escalation of members over time. The opposite
can also happen. If NoOfMembers decreases, resulting in a decrease in
ApplicationRate and, in turn, a further decrease in NoOfMembers, a down-
ward spiral of NoOfMembers ensues. One can infer from this that ‘positive’
feedback cause–effect paths are critical to growth, but can also lead to
uncontrolled decline if not monitored and quickly acted upon. There are
five ‘positive’ feedback paths listed in Table 4.7, any one of which could
take the club operations for a roller-coaster ride!
Path #4 in Table 4.7 provides an example of a ‘negative’ feedback path. It
indicates that if the NoOfMembers increases, the NoOfEmployeesPerMember
(an index of service) declines, reducing the QualityOfService perceived by
club members, thus increasing the CancellationRate and decreasing the
NoOfMembers. It forms a self-regulating loop that, ceteris quietus, offsets an
increase in NoOfMembers with a decrease at a later time, and vice versa. It
suggests that any increase in membership will be limited if the service
capacity of the club is not improved. There are three ‘negative’ feedback
paths in total in Table 4.7. Any one of these could limit the growth of the
club or, in unison with other feedback paths, create unstable behaviours
such as oscillations of operations. Yet these feedback paths go largely
unseen because the group policy-making protocols are fixated on finding
linear policy-to-goal paths and do not include the search for circular
paths. Also, circular paths so confound the simple cause–effect predictions
as to require a more rigorous dynamic analysis beyond what could be
expected from a group meeting in committee.
 TABLE 4.7 Results of the AI policy inference procedure:
feedback loops path buried in the cause–effect system map that
can create unintended outcomes
No. of
causal
# links correlation Feedback loop paths found in sports club cause–effect map
1 2 Positive NoOfMembers ApplicationRate NoOfMembers
2 3 Negative AnnualFee MembershipRevenue TotalProfit AnnualFee
3 4 Negative NoOfMembers NoOfEmployeesPerMember QualityOfService
CancellationRate NoOfMembers
4 5 Negative NoOfMembers RevenueFromRenting TotalProfit AnnualFee
CancellationRate NoOfMembers
5 5 Positive NoOfMembers MembershipRevenue TotalProfit AnnualFee
CancellationRate NoOfMembers
6 5 Positive NoOfMembers BarAndFoodNetRev TotalProfit AnnualFee
CancellationRate NoOfMembers
7 5 Positive NoOfMembers RaffleProfit TotalProfit AnnualFee
CancellationRate NoOfMembers
8 6 Positive NoOfMembers BarAndFoodNetRev BudgetForWages
NoOfEmployeesPerMember QualityOfService CancellationRate
NoOfMembers
 A COMPLEX CAUSE–EFFECT POLICY DOMAIN 105

DISCUSSION

Limitations of the Study

The causal system map of the operations of the sports club may not
trap all the requisite richness of the real-world cause–effect system it is
intended to represent. As in any qualitative research study, the researcher
has to avoid pitfalls such as interviewing the wrong person and asking
the wrong questions, leading to the building of a map of how the corpo-
rate system of cause–effect is thought to work or is supposed to work
rather than how it does work (Hall, 1978).
Similarly, the AI procedures are based on generic behavioural protocols
garnered from the literature. Some may be situation- or era- or national-
culture-dependent, or just may not represent all the complexity of human
group decision making. Nevertheless, given the interwoven complexity
of causal relations within the system map and the relatively parsimonious
set of simple generic behavioural protocols built into the AI, the similari-
ties between the predicted and adopted policies by the club’s executive
are quite startling. The explanations generated by the analysis into how
the policies became adopted adds further to the credibility of the model.
Bearing in mind the above limitations and the fact that we have only a
sample of two corporate operating systems (this study of a club and
another study of a magazine publishing firm [Hall, 1999]) to compare the
polices adopted with the AI program predictions, the following discus-
sion is put forward.

Can Formal Methods of Cause–Effect Mapping

Improve Group Policy Making?

The detail and precision required in the building and analysis of corpo-
rate system cause–effect maps is not to everyone’s liking. The subsequent
analysis can alienate policy makers because it potentially confounds their
relatively simple linear maps and may question their policies, without
necessarily posing any better policies. Extracting more reliable policies
from the analysis of a formal corporate system map can take considerable
time and effort. This is not the stuff for action-oriented managers who
wish to make quick decisions and get on with other things. Yet in my
experience, there are a growing number of managers (in some organiza-
tions, at least) with the patience and interest to follow the convoluted
logic of causal diagrams without getting too upset!
As an example, the effect of walking a class of senior managers from a
large furniture manufacturing firm through an elaborate cause–effect
system map was quite unexpected. It concerned a complex map built
around employee commitment that showed graphically why some pro-
ductivity improvement programmes, such as Total Quality Management
106 INFERRING MANAGERIAL KNOWLEDGE

(TQM), can lead to bankruptcy.8 The presentation concentrated on

sensitizing the class to the feedback cause–effect paths that usually go
unseen yet bring about the unintended side-effects of layoff decisions.
Immediately after the class, the managers convened a meeting and
reversed a policy decision made previously concerning laying off surplus
employees resulting from a successful company TQM programme. One
might draw the conclusion that something in the map enriched the
group’s comprehension, thus tipping the balance from one policy to
the other.

The Implications of the Unseen Feedback

Cause–Effect Paths

Positive and negative feedback loops working in unison can form confi-
gurations giving rise to behaviours such as the limits to growth and the
quick fix (Senge, 1990: App. 2). For instance, the limits to growth scenario
could arise if the membership of the club declines due to the lack of
service capacity and/or quality of management and/or high fees. The fre-
quent raising of fees to rectify the profit position without realizing the
deleterious side-effects it was having on membership could be labelled a
quick fix. It would appear that both scenarios were involved in the long-
run decline of membership and financial viability of the sports club,9 yet
the feedback cause–effect paths causing it went unseen by the club execu-
tive committee.
The study provides insight into how feedback cause–effect paths
become overlooked in the organizational behaviour processes of building
a group cause–effect map; a characteristic noted by other researchers
(Axelrod, 1976; Forrester, 1970). It means that a group would be unlikely
to perceive the system dynamics giving rise to growth and decline and to
unintended side-effects of their policies. It is akin to forecasting the
weather without an understanding of the thermal feedback dynamics that
can suddenly turn an unstable air mass into a raging storm. The absence
of recursive causal paths in the thinking of managers and policy makers
is of particular concern to members of the System Dynamics and
Organizational Learning movement (European Journal of Operational
Research, 1992; Senge, 1990). Without the inclusion of feedback relations,
even improvements leading to more accurate group cause–effect maps
would not necessarily provide any better policies. How can policy
makers fathom which of the numerous feedback loops hidden in a formal
corporate systems map could possibly destabilize the system? Even with
this knowledge, it is rarely immediately obvious what policies to invoke.
A rigorously developed System Dynamics simulation model (see, for
examples: Hall, 1976/1996; Hall and Menzies, 1983/1990; Morecroft and
van der Heijden, 1992) is capable of providing this analysis.
 A COMPLEX CAUSE–EFFECT POLICY DOMAIN 107

Can Managers be Trained to Handle More

Causal Complexity?

A relevant question concerns how trainable managers are in cause–effect

mapping. Jacques and Carson (1994) argue that managers fall naturally
into categories in their mental ability in handling complexity. They iden-
tify four types of complexity of mental processing (declarative, cumulative,
serial and parallel) working on either symbolic or conceptual information;
the highest order obtainable by managers being conceptual parallel-level
reasoning. Is system cause–effect thinking a special kind of conceptual
parallel reasoning or is it of a yet higher universal order? Should we con-
centrate only on those managers who exhibit the potential to reason at
this level? Can system thinking be moved down to the symbolic informa-
tion level so that managers with the capacity for this type of reasoning
could benefit? This seems to be the approach taken by Senge (1990) when
describing the archetypal feedback loops present in organized systems
using vignettes, such as the fix that fails and the limits to growth,
mentioned before, that might help such managers perceive certain con-
figurations of feedback paths, appreciate their effects and know what to
do or what not to do about it. In this regard, the AI program used in this
study is capable of exposing the hidden feedback loops to facilitate this
kind of learning.

Do Different Starting Beliefs Lead to Different

Future Histories?

As explained before, the AI program finds the two shortest paths from
policy-to-goal of opposing cause–effect path correlation. The user is
asked to choose which one (the ‘positive’ or ‘negative’ path) represents the
current belief of the group. Suppose that instead of choosing, say, the
‘positive’ sign of correlation, a ‘negative’ sign is chosen as the belief,
and supposing the resulting policy is just as successful with the ‘nega-
tive’ sign under the current circumstances. Would the organization
applying the policy experience a quite different future? That is, different
starting beliefs could explain why different organizations in the same
industry, with similar cause–effect policy domains, might evolve in quite
different ways. This is consistent with the theory of ‘Order Out of Chaos’
(Prigogine and Stengers, 1984), which postulates that the kind of organi-
zation evolving in closed-feedback cause–effect systems is very depen-
dent on the initial starting conditions. This suggests that the method
employed in this study might be used by a policy group to check out
its current beliefs to see if it is missing out on more attractive future
histories!
108 INFERRING MANAGERIAL KNOWLEDGE

The Puzzle of Exceptional Organizations

The logic encoded in the AI program aids the selection of the shortest
paths from policy-to-goal representing the cause–effect arguments with
the most immediate effect. This is argued to be the most realistic situation
in group policy making, where simple arguments would more likely carry
the day. There is growing evidence that exceptional organizations adopt
policies that are premised on more complex arguments that relate less
tangible policy variables, such as training the workforce, to relatively
long-run goals, such as developing the potential of the organization, that
run counter to the conventional wisdom of their industries at the time
(Doman et al., 1995; Starbuck, 1993). How different are the collective
policy maps in such organizations, and how does this difference come
about? Is it culturally determined by a founder and passed down to suc-
ceeding generations of managers to apply blindly? Or by chance did the
group start with different beliefs to other organizations in the same indus-
try? Or is there some higher level of group policy map comprehension
responsible for the phenomenon? The answers to these questions could be
most illuminating!

SUMMARY

In this study I have presented a process model of group policy making

based on a relatively small number of behavioural rules garnered from
the literature on how a policy-making group makes sense of its complex
cause–effect policy domain. These rules were embedded in an AI program
to mimic the collective decision-making behaviours of policy-making
groups. A cause–effect system map of a club was used as a surrogate
policy domain. The map was interpreted by the AI for two levels of com-
prehension of the cause–effect system of relationships among the concept
variables in the map. The resulting policies were compared with those
adopted by the club executive committee at two separate periods in the
club’s history that corresponded to the two levels of comprehension. The
identical policies observed add credence to the AI program emulating
the collective cognition of the policy makers. The AI program affords an
explanation for the missing feedback paths of cause–effect in the policy
maps of managers observed by others. Questions are addressed of
whether formal cause–effect maps can help policy groups enhance their
maps to make better policy decisions, whether it might be possible to
train managers in this kind of system thinking, and what the implications
are of the missing feedback paths in group cause–effect maps. The way
policies are chosen and the closed-feedback nature of the cause–effect
system of the policy domain suggest that organizations are subject to
‘Order out of Chaos Theory’. This might explain why different starting
 A COMPLEX CAUSE–EFFECT POLICY DOMAIN 109

beliefs in cause–effect path correlations result in quite different time paths

or histories. It might also cast some light on the phenomenon of excep-
tional organizations.

NOTES

This is a revised and expanded version of working papers presented at the Symposium on
Strategic Implementation and Assessment Research, Montreal, 1996, the Colloquium of the
European Group for Organizational Studies on Contrasts and Contradictions in Organization,
Istanbul, 1995, and the International Workshop on Rationality and Organization of the
European Institute for Advanced Studies in Management, Brussels, 1993.
I wish to acknowledge the support received for this study by a Strategic Research Grant
No. 804-94-0007 from the Social Sciences and Humanities Research Council of Canada, and
by a Visiting Fellowship from the University of Warwick Business School.
I am indebted to my colleagues Peter Aitchison, Department of Applied Mathematics, and
Bill Kocay, Department of Computer Science, University of Manitoba, for providing the
path-finding algorithm used in this study.
1. For a discussion on methods for eliciting formal cause–effect maps from managers, see
Hall et al. (1994).
2. See Hall et al. (1994) for a more detailed description of the path finder.
3. Axelrod (1976) notes that the cognitive maps of managers are characterized by a few
short determinant causal paths (arguments) with few or no feedback loops (circular argu-
ments) that would complicate the map and confound the analysis.
4. Pettigrew (1973) notes that group status enhancement or defence can have a powerful
influence on a group’s preferences in decision making.
5. Different modes of presentation, it would seem, appeal to different people. For instance,
a corporate lawyer attending a company in-house seminar pleaded with me to use words
rather than diagrams to explain a complex cause–effect system map. The other participants,
on the other hand, much preferred the visual display in arrow diagram form.
6. For a more complex organization with several political entities each pursuing its own
goals, see the analysis contained in Hall (1999).
7. This does not mean that all the members of the committee were blind to the missing
relations, but that those who had insight may have chosen not to challenge the group’s
implicit understanding, for whatever reasons.
8. Taken from Sterman et al. (1997).
9. For a more detailed explanation of the effect of the feedback loops on the performance
of the club, see Hall and Menzies (1983/1990).

REFERENCES

Axelrod, R.M. (ed.) (1976) The Structure of Decision: Cognitive Maps of Policy Elites. Princeton,
NJ: Princeton University Press.
Coser, L.A. (1964) ‘The termination of conflict’, in W.J. Gore and J.W. Dyson (eds), The Making
of Decisions. Glencoe, IL: Free Press.
Cyert, R.M. and March, J.G. (1992) A Behavioral Theory of the Firm. 2nd edn. Englewood Cliffs,
NJ: Prentice-Hall.
Doman, A., Glucksman, M., Mass, N. and Sasportes, M. (1995) ‘The dynamics of managing
life insurance companies’, System Dynamics Review, 11: 219–232.
Eden, C. (1990) ‘Strategy development as a social process’, Journal of Management Studies, 29:
799–811.
110 INFERRING MANAGERIAL KNOWLEDGE

Eden, C. and Huxham, C. (1988) ‘Action-oriented management’, Journal of the Operational

Research Society, 39: 889–899.
Einhorn, H.J. and Hogarth, R.M. (1978) ‘Confidence in judgment: Persistence of the illusion
of validity’, Psychological Review, 85: 395–416.
European Journal of Operational Research (1992) Special edition on Modelling for Learning.
59 (1).
Forrester, J.W. (1970) ‘Counterintuitive behavior of social systems’, Technology Review,
3: 52–68.
Hall, R.I. (1976) ‘A system pathology of an organization: The rise and fall of the Old Saturday
Evening Post’, Administrative Science Quarterly, 21: 185–211. Reprinted (1996) in
G.P. Richardson (ed.), Modelling for Management I: Simulation in Support of Systems Thinking
(The international library of management series). Aldershot, UK: Dartmouth Publishing.
pp. 251–277.
Hall, R.I. (1978) ‘Simple techniques for constructing explanatory models of complex systems
for policy analysis’, Dynamica, 3: 101–144.
Hall, R.I. (1981) ‘Decision making in a complex organization’, in G.W. England,
A.R. Negandhi and B. Wilpert (eds), The Functioning of Complex Organizations (selected
papers from the specialized conference of the International Institute of Management of
the Science Center, Berlin). Cambridge, MA: Oelgeschlager, Gunn & Hain. pp. 111–144.
Hall, R.I. (1984) ‘The natural logic of management policy making: Its implications for the
survival of an organization’, Management Science, 30: 905–927.
Hall, R.I. (1989) ‘An artificial intelligence approach to building a process model of manage-
ment policy making’, in M.C. Jackson, P. Keys and S.A. Cropper (eds), Operational Research
and the Social Sciences (selected papers from the 1989 IFORS specialized conference,
Queen’s College, Cambridge University). New York: Plenum Press. pp. 439–444.
Hall, R.I. (1990) ‘Building an artificial intelligence model of management policy making:
A tool for exploring organizational issues’, in M. Marsuch (ed.), Organization, Management
and Expert Systems: Models of Automated Reasoning (selected papers from the international
workshop of the Management Center, University of Amsterdam). Berlin: de Gruyter.
pp. 103–121.
Hall, R.I. (1999) ‘A study of policy formation in complex organizations: Emulating group
decision making with a simple artificial intelligence and a system model of corporate
operations’, Journal of Business Research (special edition on Strategy Implementation and
Assessment Research), 45: 157–171.
Hall, R.I. and Menzies, W.B. (1983) ‘A corporate system model of a sports club: Using simu-
lation as an aid to policy making in a crisis’, Management Science, 29: 52–64. Reprinted
(1990) in R.G. Dyson (ed.), Strategic Planning: Models and Analytical Methods. Chichester:
Wiley. pp. 183–197.
Hall, R.I. Aitchison, P. and Kocay, W.L. (1994) ‘Causal policy maps of managers: Formal
methods for eliciting and analysis’, System Dynamics Review, 10: 337–360.
Jacques, E. and Carson, K. (1994) Human Capability. Falls Church, Victoria: Carson Hall & Co.
Kahneman, D.S. and Tversky, A. (1982) Judgment Under Uncertainty: Heuristics and Biases.
Cambridge: Cambridge University Press.
Morecroft, J.D.W. and van der Heijden, K.A.J.M. (1992) ‘Modelling the oil producers:
Capturing oil industry knowledge in a behavioral simulation model’, European Journal of
Operational Research, 59: 102–122.
Newell, A. and Simon, H.A. (1966) ‘Computer simulation of human thinking’, Science, 134:
2011–2017.
Pettigrew, A.M. (1973) The Politics of Organizational Decision Making. London: Tavistock.
Prigogine, I. and Stengers, I. (1984) Order Out of Chaos. London: Heinemann.
Roberts, F.S. (1976) ‘Strategy for the energy crisis: The case for commuter transport policy’,
in R.M. Axelrod (ed.), The Structure of Decision: Cognitive Maps of Policy Elites, Princeton,
NJ: Princeton University Press. pp. 414–426.
Schank, R.C. and Colby, K.M. (1973) Computer Models of Thought and Language. San Francisco:
W.H. Freeman.
 A COMPLEX CAUSE–EFFECT POLICY DOMAIN 111

Senge, P.M. (1990) The Fifth Discipline: The Art and Practice of the Learning Organization.
New York: Doubleday.
Starbuck, W.H. (1993) ‘Keeping a butterfly and an elephant in a house of cards: The elements
of exceptional success’, Journal of Management Studies, 30: 886–921.
Steinbrunner, J.D. (1974) The Cybernetic Theory of Decision. Princeton, NJ: Princeton
University Press.
Sterman, J., Repenning, J.N. and Kofman, F. (1997) ‘Unanticipated side effects of successful
quality programs: Exploring a paradox of organizational improvement’, Management
Science, 43: 503–521.
Tarjan, R. (1972) ‘Depth-first search and linear graph algorithms’, SIAM Journal of Computing,
1: 146–160.
Weick, K.E. (1969) The Social Psychology of Organizing. Reading, MA: Addison-Wesley.
5
ENTREPRENEURIAL
N A R R AT I V E S A N D T H E
DOMINANT LOGICS OF
HIGH-GROWTH FIRMS

Joseph F. Porac, Yuri Mishina and Timothy G. Pollock

ABSTRACT

Our focus has been on extracting information from narrative data about the
‘dominant logics’, or mental models, of entrepreneurial teams. It is based upon
the assumption that open-ended narrative data about firm competences,
founder characteristics, corporate values and future plans represent a rich
source of information about the mental models of a company’s management. In
this study we focus on what managers view as the feasible growth strategies
available to their firms and the linkages among these strategies. We show that
there are inter-industry patterns in the clustering of growth strategies that sum-
marize distinctive cognitive representations of management regarding firm
expansion. The patterns are evidence of macro-level entrepreneurial knowledge
with both public policy and strategic implications.

INTRODUCTION

Uncovering the mental models of managers is a topic of primary interest

for managerial cognition researchers. Examining the cognitive frame-
works of managers can provide insights into the motivations behind
managerial actions and subsequent firm performance (Fiol, 1995; Penrose,
1959). However, ‘thoughts are, by definition, unobservable through any
direct means’ (Fiol, 1995: 522). As a result, researchers face the difficult
task of attempting to deduce managerial beliefs by examining the verbal
communications of managers. To this end, a variety of cognitive mapping
 ENTREPRENEURIAL NARRATIVES 113

techniques have been used by researchers to assess the implicit mental

models of managers. Most research on cognitive mapping has examined
causal beliefs (Bougon, 1983; Cossette and Audet, 1992; Eden et al., 1992;
Langfield-Smith, 1992; Laukkanen, 1994; Pitt, 1998) and/or categories of
competitors (Calori et al., 1994; Porac et al., 1995; Reger and Palmer, 1996).
While research of this type has been interesting and valuable, cognitive
mapping is applicable to more than causal beliefs and environmental
categorizations. One area in which cognitive mapping has not previously
been used is in understanding the linkage between firm growth and
entrepreneurial cognitive frameworks.
Entrepreneurs are cognitive agents who operate in turbulent, equivocal
environments such as new industries or new industry segments (Hill and
Levenhagen, 1995). In these uncertain competitive environments, changes
in technology shift knowledge streams so that knowledge gaps are con-
stantly being opened and closed. Whenever knowledge gaps are created,
uncertainty and ambiguity are perceived. Entrepreneurs reduce this per-
ceived uncertainty by ‘bridging and closing gaps between different
streams of knowledge’ (Levenhagen et al., 1993: 77). In doing so, they are
able to create potentially sustainable competitive advantages. However,
aside from their ability to implement their ideas and erect barriers to imi-
tation, there are several things that entrepreneurs must do in order to
succeed in finding and filling gaps. First, they must stay abreast of fun-
damental changes in various knowledge streams in order to be able to rec-
ognize the opening and closing of knowledge gaps (Levenhagen et al.,
1993). Second, they must challenge pre-existing assumptions and beliefs
in order to be able to generate the new conceptualizations or recombina-
tions of existing knowledge and resources that are required to fill these
gaps (Levenhagen et al., 1993). Finally, entrepreneurial leaders must
create compelling ‘stories’ and arguments to reduce ambiguity and
uncertainty for their stakeholders and thereby gain credibility and legit-
imization for their new conceptualizations (Levenhagen et al., 1993). As a
result, entrepreneurs’ mental models are instrumental in determining
their success in discovering and bridging gaps in knowledge in order to
create value for their organization.
This suggests that one difference between entrepreneurs and non-
entrepreneurs may be in their respective cognitive structures. Palich and
Bagby (1995), for example, found that while entrepreneurs do not have a
higher risk propensity than non-entrepreneurs, the former perceive situa-
tions as being more positive and filled with opportunities than do the
latter. Similarly, Jenkins and Johnson (1997) found that entrepreneurial
success was associated with specific types of causal attributions that
entrepreneurs make.
While researchers have examined entrepreneurial characteristics and
behaviours associated with the successful start-up of a business (Gatewood
et al., 1995), differences between entrepreneurs and non-entrepreneurs
(Chen et al., 1998; Jenkins and Johnson, 1997; Palich and Bagby, 1995) and
114 INFERRING MANAGERIAL KNOWLEDGE

corporate entrepreneurship (Russell, 1999), research on entrepreneurial

cognition has generally stopped at the point a business is started. Therefore,
little attention has been given to the way in which the perceptions and
beliefs of entrepreneurs affect the continued growth of the firm. In addition,
cognitive mapping techniques have been applied to the study of the entre-
preneurial mind in only a few studies (e.g. Jenkins and Johnson, 1997;
Russell, 1999). Therefore, not much is known about the linkages between
cognitive mapping, entrepreneurial cognition and growth strategies. To
address this deficit in the literature, we examine entrepreneurial beliefs
using the resource-based view of the firm and its underlying cognitive
proposition as a theoretical lens and present a new method for mapping
managerial logics for organizational growth.

The Cognitive Proposition Underlying

Resource-Based Views of Growth

The resource-based view of the firm provides a useful framework for

understanding the role of managerial cognitive structures in shaping
entrepreneurial strategies for growth (Mahoney and Pandian, 1992;
Peteraf, 1993). Tracing the origins of their work to Penrose (1959),
resource-based theorists argue that a firm’s unique portfolio of tangible
and intangible resources influences the rate and direction of its growth
and diversification. The rate of growth is influenced by how the manage-
ment team conceptualizes and uses the firm’s resource base. The direction
of expansion is controlled by path dependencies inherent in imperfect
resource substitution, the indivisibility of the firm’s assets, and what man-
agement considers to be the firm’s feasible growth strategies. All of these
‘internal’ forces interact with the competitive environment to determine
economic performance. A firm can capture sustainable profits and grow
to the extent that its key value-creating activities, or core competences
(Hamel and Prahalad, 1994), cannot be easily imitated by competitors
(Barney, 1991; Lippman and Rumelt, 1982).
The ability to sustain profitability in the face of competition has
received the bulk of attention in the resource-based literature (Peteraf,
1993). No doubt, this has been due to the theoretical and practical impor-
tance of understanding firm performance within competitive markets.
However, as Penrose (1959) noted, the way in which competitive forces
impact a firm’s profit stream depends in part on the ‘image’ (in Penrose’s
terms) that its managers have of both the firm itself and the competitive
environment in which it operates. Managers will pursue competitive
actions and deploy resources in ways that are consistent with their images
of their firms’ capabilities and the competitive threats that they believe
the firm faces. In fact, Penrose suggests that this is the key entrepreneurial
role of the management team. Thus, the Penrosian competitive environ-
ment is a subjective one as management teams from different firms
 ENTREPRENEURIAL NARRATIVES 115

survey the competitive landscape from their own idiosyncratic vantage

points, use their market knowledge to define firm-specific productive
capabilities, and then shift their firm’s activities toward unique market
opportunities that they believe their productive capabilities make possible.
It is the combination of these managerial images of the firm and the com-
petitive environment with what Penrose calls the ‘entrepreneurial ambi-
tion’ of a management team (its risk propensity and desire for growth)
that defines the set of growth strategies that the management team deems
feasible and attempts to enact.
One of the key insights of the Penrosian view is that it is not the actual
resources themselves that determine the growth and direction of a firm.
Rather, growth is a function of the productive capabilities that are engen-
dered by resources interacting with managerial cognitive frameworks. To
our knowledge, the ‘cognitive proposition’ (Mahoney and Pandian, 1992)
described by Penrose, which underlies the resource-based view of organ-
izational growth, has not been the subject of any systematic, scholarly
inquiry in the entrepreneurship literature. Although a resource-based
view of entrepreneurship has begun to receive some attention (Brush and
Radha, 1997; Brush et al., 1997; Chandler and Hanks, 1994; Greene and
Brown, 1997; Hart et al., 1995), the study of growth strategies from a cog-
nitive perspective has not been the subject of much quantitative empirical
research. The purpose of this study is to begin to investigate the cognitive
bases of entrepreneurial dominant logics among a group of high-growth
and successful firms. In particular, the research attempts to determine the
extent to which underlying patterns of dominant growth logics exist
across a sample of entrepreneurial firms from widely varying industries.

Dominant Growth Logics as Clusters

of Growth Strategies

The probability of adopting a given set of growth strategies will vary

given the unique conditions of a firm. Each combination of strategies can
differ in its cost, a firm’s ability to implement them, and the time they will
take to provide a return on the company’s investment. At the same time,
however, research in strategic management has shown that there are
systematic patterns of strategic similarity within and across industries
(McGee and Thomas, 1986). These results have been generalized to the
entrepreneurship arena by Carter et al. (1994), who found clear clusters of
competitive strategies among new ventures across six different industries.
These results make it clear that firms are not completely unique in the
strategies that they enact, and uncovering similarities in strategies is an
important line of research in the strategic management field.
Our study expands this line of research in two ways. First, most
research on strategic similarity has focused on competitive strategies
rather than explicit strategies for expanding a firm’s business. In our
116 INFERRING MANAGERIAL KNOWLEDGE

research, we seek to uncover patterns of similarity among firms in their

logics and strategies for growing the business. Second, our approach is
explicitly cognitive in nature in the sense that we focus on the belief
systems of entrepreneurs regarding their growth plans, what we have
called growth ‘logics’. Although some research has been conducted on
strategic similarity from a cognitive perspective (Porac et al., 1995; Reger
and Huff, 1993), this work was not focused on growth logics or on entre-
preneurial firms per se. Putting these two contributions together, the goal
of the present research is to determine the extent to which inter-industry
similarities exist in the beliefs of entrepreneurs regarding their firms’
feasible expansion paths, such that clusters of differing growth logics can
be discerned across firms. To answer this question, we coded open-ended
statements by entrepreneurs regarding their plans for future expansion
and then used cluster analysis techniques to uncover inter-industry
patterns in these growth statements.

RESEARCH METHOD

Data

Our statistical sample of entrepreneurial firms consisted of 54 firms

drawn from the Kauffman Foundation’s recently developed database of
high-growth firms. This database contains information on regional final-
ists and winners in the annual Entrepreneur of the Year Competition
sponsored by Ernst and Young, the Kauffman Foundation and Inc. maga-
zine. Of the 54 firms, 31 per cent are in manufacturing, 48 per cent are
service companies, and 11 per cent are retail companies. The remaining
10 per cent of the firms are in construction and real estate. At the time the
narratives were written, 54 per cent of the companies in the sample were
public. The average number of employees per company was 742,
although the numbers ranged from 12 to 5640. The majority of the narra-
tives (72 per cent) were written in 1994 and 1995. The remaining 28 per
cent were written in 1993. The average company in our sample had
$7.98 million in cash, $11.63 million in fixed assets, $6.71 million in inven-
tory and an average net income, in the year the narratives were written,
of $6.08 million. The average company’s cash reserves grew by 76 per
cent, fixed assets nearly doubled, and after-tax net income grew by almost
80 per cent from the year before the company was nominated for the
Entrepreneur of the Year award through the year after the company was
nominated. The average company also increased its labour force by
33 per cent from the year before the narrative was written to the year the
narrative was written. This suggests that the companies in this sample are
indeed high-growth firms.
Of particular significance for this study, the Kauffman database contains
narrative information, written by the CEO or a top management team
 ENTREPRENEURIAL NARRATIVES 117

representative, describing both the company and the individual being

nominated for the award. The narratives include background information
on the nominee and his or her accomplishments, the company’s history,
its primary products and services, its business practices and its plans for
the future. We developed a cognitive map of entrepreneurial dominant
logics by content analysing these narratives to identify each firm’s
intended growth strategies.

Overview of Methodology

Cognitive mapping is a technique that is recognized as a useful way of

depicting managerial mental models (Fiol and Huff, 1992). A cognitive
map is a depiction of the hidden assumptions and beliefs of managers
that represent concepts and the relationships among them (Fiol and Huff,
1992). Managerial cognitive structures must be inferred through manage-
rial verbal or written statements (Fiol, 1995). While most studies on man-
agerial cognition seem to use interviews, other researchers have collected
public written statements in order to uncover managerial belief systems.
For example, executives’ statements in annual reports as well as internal
planning documents have been used as sources of data to map the
managerial mind (Fiol, 1989, 1995). In this study we continue in this vein
by examining open-ended written statements from entrepreneurs.
Entrepreneurial narratives are ideal for studying the thoughts and sub-
jective beliefs that a manager may have about his or her environment
because they contain ‘implicit, personal theories of managerial action’
(Pitt, 1998: 387). The statements were then content analysed using a
computerized text analysis package, focusing on the sentence as the unit
of analysis.
There are several benefits to using computers for qualitative data analy-
sis. Computers can reduce the amount of labour required when organizing
and coding ethnographic data without fundamentally changing the nature
of the process (Dohan and Sánchez-Jankowski, 1998). However, there are
many different types of software packages that can be used to assist
researchers in qualitative data analysis, and each type has unique benefits.
Dohan and Sánchez-Jankowski (1998) suggest that there are three cate-
gories of qualitative analysis software packages: document processors, data
organizers and symbolic manipulators. Document processors are used to
create and manage ethnographic databases as well as to search and retrieve
text strings from these databases. Data organizers, on the other hand, allow
the researcher to structure ethnographic databases by organizing and
annotating or by coding. In this way, researchers are able more easily to
place data in context, maneuvre through large databases and retrieve
information based on theoretical mark-ups of the databases. Finally,
symbolic manipulators assist researchers in the development and/or test-
ing of theories about the relationships present in the ethnographic database.
118 INFERRING MANAGERIAL KNOWLEDGE

The software package that we used, ATLAS/ti™, falls into this last category
of symbol manipulators, and was chosen for its applicability to cogni-
tive mapping.
ATLAS/ti is a package that appears to be uniquely suited for the task of
cognitive mapping, for three reasons. First, it allows the source data to be in
almost any form. Unlike some other packages, which can be used to analyse
only textual data, ATLAS/ti allows for the analysis of text, audio, video,
graphics or a combination of all four. Second, there is great flexibility in
the types of linkages that can be created in ATLAS/ti. For example, both
quotations and codes can be annotated, and various types of linkages can
be created between the quotations, codes and memos (Dohan and Sánchez-
Jankowski, 1998). As a result, several types of cognitive maps, including
taxonomies and causal maps, can be drawn from the data using this pro-
gram. Finally, ATLAS/ti contains a feature that allows for the graphical rep-
resentation of the codes and linkages. Therefore, it makes the process of
drawing a cognitive map less labour-intensive as well.
The content analysis of textual data is an analytic technique that can be
used to make inferences regarding the sender of the message, the message
itself and/or the intended audience (Weber, 1990). Content categories are
developed and used to make analysis of the text more tractable by grouping
units of text that have similar meanings. These units may be words,
phrases, sentences or other lexical units that the researcher deems appro-
priate. Past research (Abrahamson and Park, 1994; Wade et al., 1997) has
used the frequency with which top management discusses a concept in
written narratives as an indicator of the concept’s importance. In this
study we used the mention of a particular growth strategy as an indicator
of the salience of that strategy to the management team of the firm. We
assume, therefore, that mentioning a particular strategy at least once
implies that the strategy is a component of the entrepreneurial team’s
growth logic.
The narrative statements from the Kauffman Foundation’s database of
high-growth firms provide rich and informative interpretations that
entrepreneurs have of their environments and the resources available to
their firm, as well as perceptions of their ability to influence the perfor-
mance of their firm (Pitt, 1998). However, there are several potential prob-
lems in using such communications as cognitive proxies. The major
problem is that there may be a self-serving bias such that the narratives
may reflect impression management more than the true feelings and
beliefs of the managers (Fiol, 1995; Wagner and Gooding, 1997).
None the less, researchers have demonstrated that even narratives that
may have a self-serving bias are useful sources of information on man-
agement cognition and strategy (Barr, 1998; Clapham and Schwenk, 1991;
Huff and Schwenk, 1990). For example, non-evaluative statements, as
opposed to those containing positive or negative evaluations, may be rela-
tively accurate indicators of managerial beliefs (Fiol, 1995). The use of
narrative statements from the entrepreneurs is appropriate in this case
 ENTREPRENEURIAL NARRATIVES 119

since we are interested in their mental models. We first analysed the

content of each statement made by entrepreneurs in their narratives to
separate those statements that referred to future plans for growth from
statements that were not about growth. We then generated a map of
different firm-growth logics by performing cluster analyses on the growth
statements to classify firms into groups that used similar combinations of
intended growth strategies. An advantage of content analysis is that the
growth strategies and strategy combinations can be inductively derived
rather than imposed in an a priori way by the researcher (Weber, 1990).

Content Analysis of Narratives

In order to identify the growth strategies that each company considered

viable, we conducted two rounds of content analysis on each of the nar-
ratives. Two of the researchers performed the content analysis using
ATLAS/ti. During the first round of content analysis, we coded each sen-
tence of each narrative into one of six categories: CEO Characteristics,
Company Characteristics, Company Capabilities, Growth Strategies,
Image of the Market and Other. Less than 1 per cent of the sentences fell
into the Other category. Interrater reliability was determined using
Cohen’s Kappa (Cohen, 1968), which adjusts for random chance in deter-
mining interrater agreement. The Cohen’s Kappa for this stage of the con-
tent analysis, which was calculated on a 20 per cent subsample, was .87.
Any disagreements among the raters were resolved.
In the second stage of the content analysis, each growth-strategy sen-
tence was coded using six growth-strategy categories. These categories
were derived by moving iteratively between prior theory and the data.
The categories that were ultimately identified were consistent with six
strategies that the literature on growth suggests can be used to increase
revenues and profits (Leonard-Barton, 1995; Lieberman, 1989; Penrose,
1959; Powell et al., 1996; Thompson, 1986). These strategies are: growth
through mergers, acquisitions and expansion of the firm’s production
capacity (Lieberman, 1989; Penrose, 1959); developing alliances or part-
nerships with other firms (Leonard-Barton, 1995; Powell et al., 1996);
expanding into new geographic and customer markets (Leonard-Barton,
1995); developing new products or services to be offered in existing or
new markets (Leonard-Barton, 1995); expanding the firm’s labour force
and/or increasing the capabilities of its employees (Penrose, 1959); and,
finally, increasing earnings by reducing costs and/or increasing the pro-
ductive capacity of existing assets through technical and process improve-
ments (Thompson, 1986). Table 5.1 provides examples of the types of
growth strategies that fell into each of these categories.
The first category of growth strategies are those that are ‘capital-
intensive’. The strategies in this category include building new plants or
facilities, expansion of the firm’s production capacity, strategic acquisitions,
120 INFERRING MANAGERIAL KNOWLEDGE

TABLE 5.1 Expansion path examples

Expansion path category Activities included
Capital-intensive Capital expenditures for new plants or facilities
Acquisitions and mergers
Joint ventures
Alliances and partnerships Alliances, partnerships, franchising and
licensing agreements
Market expansion Enter new markets (product or geographic)
Increase share in existing markets
New real estate developments
Access new customer base
Product/service New products and/or services
Human resources Increased hiring
Management team and employee training,
education or quality of work life improvements
Technical improvements Production and administrative process improvements
Technological innovations and developments

mergers and joint ventures (Lieberman, 1989; Penrose, 1959). All of these
strategies require large capital expenditures, and growth-strategy
sentences were coded as ‘capital-intensive’ when they mentioned such
things as ‘joint venture’, ‘construction of ... [new facilities]’ and ‘strategic
acquisitions’.
The second category of growth strategies are those that are ‘non-capital-
intensive’. The strategies in this category include the development of
strategic alliances, partnerships with other firms, franchising and licens-
ing agreements (Leonard-Barton, 1995; Powell et al., 1996). Unlike the
capital-intensive strategies, these strategies are based upon sharing costs
with other firms or otherwise reducing the amount of down-side risk
faced by the firm. Growth-strategy sentences in this category mention
‘strategic alliances’, ‘strategic partnerships’ and ‘licensing agreements’.
The third category of growth strategies are those that involve ‘market
expansion’. The strategies in this category include entry into new product
or geographic markets, increase of market share in existing markets, and
accessing new customer segments (Leonard-Barton, 1995). These strate-
gies involve increasing the number of customers served, whether by cap-
turing higher market share or by geographic or product diversification.
‘Market expansion’ sentences mention ‘international expansion’, ‘seizing
opportunities to enter new markets’ and ‘expansion and diversification ...
into new fields’.
The fourth category of growth strategies are those that involve ‘product
or service development’. The strategies in this category include the devel-
opment of new products as well as services (Leonard-Barton, 1995).
Growth-strategy sentences were coded as ‘product or service develop-
ment’ when they mentioned things such as ‘developing new products’,
 ENTREPRENEURIAL NARRATIVES 121

‘increasing the number of product lines’ and ‘bringing to market new ...
products and services’.
The fifth category of growth strategies are those that involve ‘human
resource improvements’. The strategies in this category include hiring
new employees, improving the quality of work life for employees, and
enhancing the skill of the management team and employees through
training and/or education programmes (Penrose, 1959). Growth-strategy
sentences were coded as ‘human resource improvements’ when they
mentioned such things as ‘expanding [the company’s] workforce’, attract-
ing ‘additional outstanding employees’ and an ‘incentive reward system’.
The final category of growth strategies are those that involve ‘process
improvements’. The strategies in this category include production and
administrative process improvement and technological innovation
(Thompson, 1986). These strategies are aimed at increasing earnings
through enhanced productivity or through cost reduction. Growth-strategy
sentences were coded as ‘process improvements’ when they mentioned
things such as ‘productivity-enhancing technologies’, ‘automate numerous
processes’ and ‘development of adaptive process control techniques’.
Table 5.2 provides sample phrases from the growth-strategy sentences
that indicated which type of growth strategy was being discussed by the
entrepreneur. The Cohen’s Kappa for this stage of the analysis, which was
calculated using all 274 growth strategy phrases, was .93.

Cluster Analysis

Cluster analysis has been used in various disciplines for the purposes of
classification (Aldenderfer and Blashfield, 1984; Bailey, 1994; Romesburg,
1984). In this study, dummy variables were created for the six growth
strategies. Since we were interested in the presence or absence of each
strategy, as opposed to the relative importance of each growth strategy, a
firm’s narrative was coded as a ‘1’ if the firm’s entrepreneur mentioned
his/her intent to use a particular strategy and a ‘0’ if he/she did not. The
presence/absence of each of the six strategies was coded in this way. The
firms in the study were then clustered along these six binary growth-logic
scores using the average-between-groups method (Romesburg, 1984).
The average-between-groups method is one of the most widely used
clustering techniques. Romesburg (1984) recommends the use of this tech-
nique since there is less distortion in transforming similarities between
objects into a tree and because it can be used with any resemblance
coefficient. Since we wished to cluster using binary data, the average-
between-groups method was the most appropriate choice.
The resemblance coefficient used in our cluster analysis was a simple
matching coefficient. The simple matching algorithm calculates similarity
based upon the proportion of 1–1 and 0–0 matches along the different
attributes between any two cases (Romesburg, 1984). This particular
122 INFERRING MANAGERIAL KNOWLEDGE

TABLE 5.2 Sample growth-strategy phrases

Growth-strategy category Sample phrases
Capital-intensive strategies ‘Potential plans for further growth include the acquisition
of other companies or technology …’
‘Strategic planning for the Company includes
acquisitions and mergers …’
‘The city facility will increase manufacturing
capacity …’
‘A newly formed joint venture … should go on stream
within twelve months …’
‘Fboss also plans continued growth through internal
expansion and by the acquisition of businesses …’

Non-capital intensive strategies ‘These alliances … strongly position Fcomp for a bright
future.’
‘Another new division of Fcomp will license in and
distribute Pproducts …’
‘… development of strategic alliances with partners in
complementary industries …’
‘With this in mind, one of Fcomp’s future goals is to
build “Strategic Partnerships” with original equipment
manufacturers …’
‘Future growth will be seen in the use of licensing
arrangements now in hand with Ccomp, Ccomp for
Pproduct and Ccomp for Pproduct.’

Market expansion strategies ‘Fcomp plans to expand geographically …’

‘… the company intends to expand into related
markets …’
‘Efforts to penetrate or expand business in the emerging
Rregion, Sregion and Tregion markets are a priority and
will produce a larger customer base.’
‘… growth in existing Fcomp markets …’
‘Fcomp’s Plans for the Future include careful, reasoned
expansion and diversification, where needed by clients,
into new fields and geographical regions.’

Product and service ‘We will soon expand our services to include …’
development strategies
‘Potential plans for further growth include … developing
new products from within.’
‘Its vision of the future includes the development of ever
more capable versions of the Pproduct and Pproduct xx
products.’
‘Development continues on these products and
associated peripherals.’
Human resource improvement ‘To facilitate future growth Fboss has recently added
strategies several members to the senior staff.’
‘To accomplish this expansion will require a major effort
to attract additional outstanding employees to the Fcomp
fold.’
(Contd.)
 ENTREPRENEURIAL NARRATIVES 123

TABLE 5.2 (Contd.)

Growth-strategy category Sample phrases
‘Fboss believes that employees should be rewarded for
good work and plans to structure compensation and
career paths at all levels to be closely aligned with
performance.’
‘… improvement and expansion of training through
in-house and supported training programs …’
‘… an incentive reward system that will improve staff
motivation …’

Process improvement strategies ‘Because [xx] is an ongoing quality system of process

improvement, we will undergo a rigorous schedule of
external and internal audits to prove we are maintaining
and improving our quality.’
‘Eliminating and/or reducing re-work through the
implementation of Total Quality Management (TQM)
techniques …’
‘Adaptive process control is one of the many paths Fboss
is exploring.’
‘In the mid-19[xx]s Fboss plans to invest in productivity-
enhancing technologies and quality improvement.’
Note: All identities have been disguised and refer to companies, products and individuals
using a letter notation, e.g. Ccomp, Pproduct and Fboss.

resemblance coefficient was chosen for several different reasons. First, it

was an association measure and therefore appropriate for binary data.1
Second, it provides a meaningful proportion since it is the proportion of
agreements. Third, based on the nature of the formula used to calculate
this coefficient, unlike some of the other association measures, it can be
calculated even if there are only 0–0 matches. Finally, simple matching
weighs 1–1 and 0–0 matches equally. This was important since we were
interested not only in which strategies were mentioned, but also in which
ones were not mentioned.
One concern that arises when using cluster analysis is in choosing the
appropriate number of clusters. Determining the number of clusters has
been problematic and has often been subjective (Aldenderfer and
Blashfield, 1984; Bailey, 1994; Milligan and Cooper, 1985; Porac et al., 1995;
Romesburg, 1984). One method for determining the optimal number of
clusters is to compute correlations between the proximity matrix used to
form clusters and a series of structure matrices that correspond to various
cluster solutions (Milligan and Cooper, 1985; Porac et al., 1995). The opti-
mal solution occurs when the correlation between the proximity matrix
and the structure matrices reaches a maximum. Upon running the cluster
analysis, we created structure matrices for two-through ten-cluster solu-
tions and calculated correlations between each structure matrix and the
original proximity matrix using the quadratic assignment procedure
(QAP) in UCINET IV (Borgatti et al., 1996). From Table 5.3 we can see that
the five-cluster solution has the highest correlation and therefore is the
124 INFERRING MANAGERIAL KNOWLEDGE

TABLE 5.3 Correlations between proximity matrix

and cluster structure matrices
Cluster solution
2 3 4 5 6 7 8 9 10
.326 .513 .520 .535 .533 .527 .514 .511 .507

optimal solution. Once the cluster memberships were determined, the

narrative statements were qualitatively analysed to search for any
commonalties among cluster members that may help explain why entre-
preneurs chose a particular set of intended growth strategies. Particular
attention was paid to mentions of entrepreneurs’ experiences, company
experiences and descriptions of their businesses and their environments.

RESULTS

Table 5.4 presents the results of the cluster analysis with five clusters.
Figures 5.1 through 5.5 present the five clusters identified in our analysis
graphically. Dotted lines indicate that some, but not all, members of a clus-
ter pursued a particular growth strategy. Each cluster has distinct charac-
teristics. The firms in the first cluster focus (Figure 5.1, see page 129) on
process improvements and human resources development. The second
cluster (Figure 5.2, see page 129) has a focus on market and product expan-
sion, with some of the firms mentioning alliances and partnerships as well.
The third cluster (Figure 5.3, see page 130) is focused on human resources
and capital-intensive strategies with some market and product expansion.
The firms in the fourth cluster (Figure 5.4, see page 131) focus on market
expansion, product expansion and capital-intensive strategies. Finally, the
firms in the fifth cluster (Figure 5.5, see page 132) mention all six growth
strategies. The qualitative analysis of the narrative statements provided
some interesting results.
Table 5.5 lists selected sample quotes from the narrative statements that
indicate that there are regular patterns within each cluster that reflect the
combination of strategies that comprise a cluster’s typical growth logic.
The firms in the first cluster (Figure 5.1) seem to focus on expanding
product offerings through process improvements and training of
employees. The narratives for these firms frequently mention ‘quality’,
‘efficiency’ and ‘Total Quality Management’, and the entrepreneurs’
descriptions of their businesses and environments seem to indicate that
quality and efficiency are the key ingredients that lead to flexibility and
growth. Therefore, these firms evidence a dominant logic of ‘expansion of
products via continuous improvement’.
The second cluster (Figure 5.2) includes two distinct types of firms. The
narrative statements indicate that both types of firms see opportunities
 ENTREPRENEURIAL NARRATIVES 125

TABLE 5.4 Cluster membership

Clusters Company DUMMKT DUMOUT DUMPROC DUMHUM DUMCAP DUMNO
ID#
1 1 1591 0 0 1 1 0 1
2 1885 1 1 1 1 0 0
3 2102 0 1 1 1 0 0
4 2213 0 0 1 1 0 0
5 2311 0 1 1 1 0 0
Total 5 5 5 5 5 5 5

2 1 1652 0 0 0 0 0 0
2 1808 1 1 0 0 0 0
3 1814 1 1 0 0 0 0
4 1818 0 0 0 0 0 0
5 1859 1 0 0 0 0 0
6 1868 1 1 0 1 0 0
7 1888 1 0 0 1 0 0
8 1940 1 0 0 0 0 0
9 2066 0 0 0 0 0 0
10 2100 0 1 0 0 0 1
11 2109 1 0 0 0 0 0
12 2112 0 1 0 0 0 1
13 2120 0 0 0 0 0 1
14 2140 0 0 0 1 0 1
15 2141 1 1 0 0 0 0
16 2152 1 1 0 0 0 1
17 2155 1 1 0 0 0 1
18 2226 1 1 0 0 0 0
19 2227 0 0 0 0 0 0
20 2233 1 1 1 0 0 0
21 2244 1 0 0 0 0 1
22 2297 1 0 0 0 0 1
23 2506 0 0 0 0 0 0
Total 23 23 23 23 23 23 23

3 1 1836 1 0 0 1 1 1
2 1884 1 0 1 1 1 0
3 1889 1 1 0 1 1 0
4 1999 0 0 0 1 1 0
5 2083 0 0 0 1 1 0
6 2098 0 0 1 1 1 0
7 2121 1 0 1 1 1 0
8 2149 0 1 0 1 1 0
9 2181 0 1 0 1 1 0
10 2295 0 0 0 1 1 0
Total 10 10 10 10 10 10 10

4 1 1870 1 0 0 0 1 0
2 1934 1 1 1 0 1 0
3 1979 1 1 0 0 1 0
4 2046 0 0 1 0 1 0
5 2096 1 1 1 0 1 0
6 2103 0 1 1 0 1 0
7 2125 1 1 0 0 1 0
(Contd.)
126 INFERRING MANAGERIAL KNOWLEDGE

TABLE 5.4 (Contd.)

Clusters Company DUMMKT DUMOUT DUMPROC DUMHUM DUMCAP DUMNO
ID#
8 2136 0 1 0 0 1 0
9 2144 0 0 1 0 1 0
10 2209 1 0 0 0 1 0
11 2212 1 1 0 0 1 0
12 2266 1 0 0 0 1 0
13 2327 1 1 0 0 1 0
Total 13 13 13 13 13 13 13

5 1 2090 1 1 1 1 0 1
2 2271 1 1 1 1 1 1
3 2314 1 1 1 0 1 1
Total 3 3 3 3 3 3 3
Total N 54 54 54 54 54 54 54

for growth into new product and geographic markets, but only one type
has identified how to accomplish this growth. The firms that know how
they want to accomplish their growth in these new markets generally
mention franchising, alliances and partnerships as their intended modes
of entry. Since all of the firms see growth opportunities, but only some
know how they want to accomplish it, these firms have a ‘market and
product expansion’ logic.
The firms in the third group (Figure 5.3) need to build capacity to serve
increasing demand in their current segment or to expand where they have
identified ‘unmet’ market needs and ‘under-served’ segments. Therefore,
these firms have a logic of ‘dealing with capacity deficits’.
The fourth group (Figure 5.4) is preparing for future growth through
capital expenditures in plant and equipment, mergers, acquisitions and
joint ventures. These firms anticipate future growth and are preparing for
it by building plants and acquiring other firms. These firms have a logic
of ‘anticipatory growth’.
The firms from the fifth group (Figure 5.5) follow a ‘scattered growth’
logic. Each of these firms has experienced growth through a number of dif-
ferent strategies. As a result, they cannot seem to identify which particular
growth strategy or combination of strategies is most appropriate for
them – everything seems to work. Therefore, the firms in this group state
that they are going to follow every single one of the growth strategies. A
typical statement will include expansion into new geographic markets,
development of new products, improvement of processes, recruitment,
strategic acquisitions, partnerships and construction of new plants.
While not all firms in a given cluster choose exactly the same combina-
tion of intended growth strategies, it is interesting to note that the narra-
tives indicate that there were dominant logics that led to their choice. For
each of the clusters, prior successes with a particular growth strategy
seemed to lead to a belief that it was the appropriate strategy to pursue
 ENTREPRENEURIAL NARRATIVES 127

TABLE 5.5 Sample quotes

Cluster Quotes
1. Expansion of products via ‘total quality control’
continuous improvement
‘Total Quality Management (TQM)’
‘be the most cost-effective, quality leader’
‘advocate of total quality management and continuous
improvement’

2. Market and product ‘find an unfilled niche in the marketplace, create a retail
expansion concept to fill that niche, and then grow the business
through franchising’
‘expand globally by seizing opportunities to enter
new markets’
‘Future growth will be seen in the use of licensing
arrangements now in hand with Ccomp, Ccomp for
Pproduct and Ccomp for Pproduct.’
‘expanding largely with the money of the franchisees’
‘the development of strategic alliances, with partners in
complementary industries’

3. Dealing with capacity ‘Fboss and Fcomp are at the forefront of the fastest
deficits growing segment of the employment market’
‘increasing our market share’
‘In my opinion the need to provide this service will
continue to grow during the next [xx].’
‘We are constantly spending time and money on
educational seminars to maximize current job
competence and readiness for future opportunities.’
‘Our group is constantly looking for investment
opportunities that we feel are “under”-managed.’
‘[The newly built plant] … enhances the company’s
ability to develop and serve its authorized builder
organization in those regions.’
‘These capacity increases were necessary for Ccomp to
meet the demand of its rapidly growing customer base.’
‘the company intends to expand into related markets. …
Each of these share some common characteristics: they
are under-served by their present systems vendors, their
requirements are largely met by Fcomp products, and
there is no system available today that meets all
their needs.’

4. Anticipatory growth ‘strategic acquisitions’

‘To support this projected growth, annual capital
expenditures at Fcomp have risen from approximately [xx]
in 19[xx] to over [xx] today.’
‘to keep the organization and its resources focused on
future challenges and opportunities’
‘Fcomp plans to expand geographically by opening new
offices or through mergers or acquisitions.’
‘Potential plans for further growth include the acquisition
of other companies or technology, as well as developing
new products from within.’
(Contd.)
128 INFERRING MANAGERIAL KNOWLEDGE

TABLE 5.5 (Contd.)

Cluster Quotes
5. Scattered growth ‘continued growth through internal expansion and by the
acquisition of businesses with complementary markets or
technologies’ and ‘collaborative efforts with large prime
contractors and equipment manufacturers’ and ‘maintain
the high quality of the company’s workforce’
‘introduction of new products’ and ‘development of adaptive
process control techniques [that will] reduce rework,
increase yields and improve quality’ and ‘eventual
acquisition of related companies’ and ‘strategic alliances
with the remaining major equipment [manufacturers]’
‘expansion and diversification, where needed by clients,
into new fields and geographical regions’ and ‘focus on
recruiting, management training, and developing
management information systems’

in the future (Tyler and Steensma, 1998). There are several possible
explanations for this bias. If the firm had past experiences with the strategy,
the entrepreneur may believe that it is merely the appropriate strategy for
the environment. Alternatively, the entrepreneur may believe that there
are several appropriate strategies for the environment, but that the firm
may have differential ability to implement specific growth strategies. A
third explanation is that the entrepreneur believes him/herself to have
the appropriate capabilities to implement a particular growth strategy
based on past success with it.

DISCUSSION

The resource-based view of the firm argues that the growth and direction
of organizations are determined by a complex interaction between avail-
able resources, capabilities and managerial mental models (Mahoney and
Pandian, 1992; Penrose, 1959). These mental models, or dominant logics,
provide subjective constraints to firm actions by limiting the amounts
and types of opportunities that a firm will consider. Penrose (1959) has
suggested that research on entrepreneurial mental models can be of
help in explaining or predicting the behaviour of firms. By qualitatively
analysing entrepreneurial narrative statements about growth, we have
begun to discern some of the commonalties among firms that share simi-
lar dominant logics.
Research in strategy has rarely examined the existence of multiple
strategies within an organization, and there appears to be an underlying
assumption that a firm will pursue only one strategy at a time. In fact, the
only way in which researchers appear to examine multiple strategies is in
the context of firm diversification and maintaining coherence among
the different lines of business (Grant, 1988; Prahalad and Bettis, 1986).
Therefore, while firms may use different competitive strategies in different
 ENTREPRENEURIAL NARRATIVES 129

Expansion of
Human
Process products via
resource
improvements continuous
improvements
improvement

Figure 5.1 Growth logic: cluster 1

Non-
capital-
intensive

Market and Product/

Market
product service
expansion
expansion development

Figure 5.2 Growth logic: cluster 2

businesses, following more than one strategy is seen as generally a bad

idea (Porter, 1985). As a result, there has been little research on whether or
not organizations will follow multiple strategies or which strategies are
likely to co-occur if they do.
However, competitive positioning is based upon the combination of
activities that a firm engages in across various markets and lines of busi-
ness (Porter, 1996). Therefore, it may be more appropriate to think of firms
as following a cluster of different types of strategies (growth, marketing,
etc.), guided by a single dominant logic about the competitive advan-
tages, resources and capabilities of the firm relative to the competition, as
opposed to a single strategy. Since a dominant logic is defined as ‘the way
in which managers conceptualize the business and make critical resource
allocation decisions’ (Prahalad and Bettis, 1986: 490), dominant logics are
clusters of strategies that the managers see as appropriate for the firm.
Since entrepreneurs attempt to reduce uncertainty and create competitive
advantage by bridging gaps across multiple streams of knowledge
(Levenhagen et al., 1993), dominant logics may be capturing these
130 INFERRING MANAGERIAL KNOWLEDGE

Market
expansion

Dealing with Human

Capital-
capacity resource
intensive
deficits improvements

Product/
service
development

Figure 5.3 Growth logic: cluster 3

attempts to close gaps. The results of this study are consistent with this
perspective, and suggest that firms intend to follow multiple strategies for
growth, and that these strategies cluster in somewhat predictable ways.
As such, dominant logics can be seen as strategy ‘portfolios’.
In our study we found five distinct clusters of growth logics, illustrated
in Figures 5.1 through 5.5. Each of these growth logic clusters differed in
the number and types of strategies that were included. An examination of
the narratives suggest that in most of these logic clusters, the entrepre-
neurs are actively trying to combine multiple strategies to obtain a desired
result. Since there has been a wide-spread assumption that multiple
strategies are bad and that strategies are inherently unrelated, this may
represent attempts to bridge multiple streams of knowledge. However,
this is not necessarily the case for all clusters. For example, the market and
product expansion logic (Figure 5.2) does not seem to involve an explicit
linkage between strategies. Similarly, there does not appear to be any link-
age across the strategies in the scattered growth logic (Figure 5.5). Still, it
is not entirely clear from the data available in this study if the use of
multiple strategies is a way of closing gaps.
 ENTREPRENEURIAL NARRATIVES 131

Market
expansion

Capital- Anticipatory
intensive growth

Product/
service
development

Figure 5.4 Growth logic: cluster 4

Not all of the entrepreneurs discuss whether or not these are multiple,
independent strategies or if they anticipate synergies/links among them
and are therefore a part of a true ‘logic’. One might inquire into the
performance effect of following multiple strategies. As stated earlier,
strategy researchers have long operated under the assumption that lack of
focus is detrimental to firm performance (Porter, 1985). Returning to the
resource-based view, growth is constrained by available resources, includ-
ing managerial capabilities, and attempting to follow too many different
opportunities will result in the outstripping of managerial resources
(Penrose, 1959). Therefore, growth without focus leads to setting oneself
up for failure. As a result, we would expect the number of different strate-
gies included in a firm’s dominant logic to be negatively related to
performance, with more focused firms having higher performance than
less focused firms.
On the other hand, too much focus may leave a firm unable to meet the
requirements of the competitive environment (Miller, 1993, 1996; Miller
and Chen, 1996). While some amount of focus leads to the development
of distinctive competencies, too much focus leads to a skill set that is
not capable of keeping up with variation in environmental demands
132 INFERRING MANAGERIAL KNOWLEDGE

Market
expansion

Non-
Process
capital-
improvements
intensive

Scattered
growth

Human
Capital-
resource
intensive
improvements

Product/
service
development

Figure 5.5 Growth logic: cluster 5

(Miller, 1993, 1996; Miller and Chen, 1996). This may limit a firm’s ability
to compete in all but one small segment of the environment. However, the
opportunities available in any one area may be limited, so it may not be
possible for firms to maintain a steady rate of growth unless multiple
strategies are followed at some point. Also, some entrepreneurship
researchers have stated that entrepreneurs search for opportunities
beyond their current resource configurations, with the assumption that
the required resources will be discovered and acquired along the way
(Kirzner, 1985). This suggests that the current resource configurations
need not necessarily constrain the number and type of growth strategies
chosen, but there may be an optimal level of focus. As a result, we would
expect an inverse U-shaped relationship where firms with both low or
high numbers of strategies would have lower performance than firms
that have an intermediate number of strategies included in their strategy
configuration.
While we are unable to clarify the link between the number of dimen-
sions included in a firm’s growth logic and their subsequent performance
based upon this data set, we are currently pursuing several different
directions in our study of entrepreneurial mental models. First, future
 ENTREPRENEURIAL NARRATIVES 133

research will examine the link between growth logics and performance.
There are a number of ways in which this could be done. We will try to
determine if there are appropriate dominant logics based upon a firm’s
resource configurations or the environment; if there is ‘one best’ dominant
logic that leads to superior performance; or if there is a link between the
number of dimensions included in a firm’s dominant logic and its subse-
quent performance. Second, we seek to study the question of intention
versus implementation regarding entrepreneurial growth logics. Argyris
and Schon (1974) make distinctions between ‘espoused theories’ and ‘theo-
ries in use’. By examining sources of secondary data, we may be able to
discover to what extent these entrepreneurs followed the strategies that
they discussed in their narratives. Finally, we wish to examine changes in
dominant logics. Different theories offer various explanations for why a
manager’s mental models may change, and we wish to test several expla-
nations against one another to determine which accounts seem to provide
valid explanations for strategic change.
While we believe that this study contributes to the resource-based
literature by identifying commonalties among entrepreneurs in their
dominant logics for growth, there are two limitations of our research that
must be recognized.
First, the size and nature of the sample may constrain the ability to
make generalizations. With a sample size of only 54, it is difficult to find
other systematic relationships among the firms following each dominant
logic. For example, it was not possible to see if there was a systematic
relationship between industry type and dominant logic. In addition, since
the Kauffman database is limited to firms identified as being successful, it
is not possible to generalize these findings to firms that are not as suc-
cessful. This also leaves open the question of how unsuccessful experi-
ences can influence dominant logic development. As a result, larger and
more diverse samples should be examined in future research.
Second, the use of narrative statements written by the CEOs or top
management team members of the sample firms presents a partial picture
of the types of experiences that the firm has truly had. In addition, it is
unclear if the sample firms intended to implement the growth strategies
they mentioned, or if there may have been some impression management
involved on the part of the entrepreneurs. Therefore, it may be worth-
while for future studies to obtain more complete data on firm experiences
as well as the actual growth strategies implemented by each firm.

NOTE

1. Association measures are appropriate for binary data (e.g. Aldenderfer and Blashfield,
1984; Bailey, 1994) while Euclidean distance measures are appropriate for data measured on
interval or ratio scales (e.g. Romesburg, 1984).
134 INFERRING MANAGERIAL KNOWLEDGE

REFERENCES

Abrahamson, E. and Park, C. (1994) ‘Concealment of negative organizational outcomes: An

agency theory perspective’, Academy of Management Journal, 37: 1302–1334.
Aldenderfer, M.S. and Blashfield, R.K. (1984) Cluster Analysis. Newbury Park, CA: Sage.
Argyris, C. and Schon, D.A. (1974) Theory in Practice: Increasing Professional Effectiveness.
San Francisco: Jossey-Bass.
Bailey, K.D. (1994) Typologies and Taxonomies: An Introduction to Classification Techniques.
Thousand Oaks, CA: Sage.
Barney, J. (1991) ‘Firm resources and sustained competitive advantage’, Journal of
Management, 17: 99–120.
Barr, P.S. (1998) ‘Adapting to unfamiliar environmental events: A look at the evolution of
interpretation and its role in strategic change’, Organization Science, 9: 644–669.
Borgatti, S.P., Everett, M.G. and Freeman, L.C. (1996) UCINET IV Version 1.64. Natick:
Analytic Technologies.
Bougon, M.G. (1983) ‘Uncovering cognitive maps: The Self-Q technique’, in
G. Morgan (ed.), Beyond Method: Strategies for Social Research. Beverly Hills, CA: Sage.
pp. 173–188.
Brush, C.G. and Radha, C. (1997) ‘Resources in new and small ventures: Influences on per-
formance outcomes’, paper presented at the Babson–Kauffman Conference on
Entrepreneurship, Boston, MA.
Brush, C.G., Greene, P.G., Hart, M.M. and Edelman, L.S. (1997) ‘Resource configurations
over the lifecycle of ventures’, paper presented at the Babson–Kauffman Conference on
Entrepreneurship, Boston, MA.
Calori, R., Johnson, G. and Sarnin, P. (1994) ‘CEOs’ cognitive maps and the scope of the organi-
zation’, Strategic Management Journal, 15 (6): 437–457.
Carter, N.M., Stearns, T.M., Reynolds, P.D. and Miller, B.A. (1994) ‘New venture strategies:
Theory development with an empirical base’, Strategic Management Journal, 15: 21–41.
Chandler, G.N. and Hanks, S.H. (1994) ‘Market attractiveness, resource-based capabilities,
venture strategies, and venture performance’, Journal of Business Venturing, 9: 331–349.
Chen, C.C., Greene, P.G. and Crick, A. (1998) ‘Does entrepreneurial self-efficacy distinguish
entrepreneurs from managers?’, Journal of Business Venturing, 13: 295–316.
Clapham, S.E. and Schwenk, C.R. (1991) ‘Self-serving attributions, managerial cognition,
and company performance’, Strategic Management Journal, 12: 219–229.
Cohen, J. (1968) ‘Weighted kappa: Nominal scale agreement with provision for scaled dis-
agreement or partial credit’, Psychological Bulletin, 70: 213–220.
Cossette, P. and Audet, M. (1992) ‘Mapping of an idiosyncratic schema’, Journal of
Management Studies, 29: 325–347.
Dohan, D. and Sánchez-Jankowski, M.S. (1998) ‘Using computers to analyze ethnographic
field data: Theoretical and practical considerations’, Annual Review of Sociology, 24:
477–498.
Eden, C.E., Ackermann, F. and Cropper, S. (1992) ‘The analysis of cause maps’, Journal of
Management Studies, 29: 309–324.
Fiol, C.M. (1989) ‘A semiotic analysis of corporate language: Organizational boundaries and
joint venturing’, Administrative Science Quarterly, 34: 277–303.
Fiol, C.M. (1995) ‘Corporate communications: Comparing executives’ private and public
statements’, Academy of Management Journal, 38: 522–536.
Fiol, C.M. and Huff, A.S. (1992) ‘Maps for managers: Where are we? Where do we go from
here?’, Journal of Management Studies, 29: 267–285.
Gatewood, E.J., Shaver, K.G. and Gartner, W.B. (1995) ‘A longitudinal study of cognitive
factors influencing start-up behaviors and success at venture creation’, Journal of Business
Venturing, 10: 371–391.
Grant, R.M. (1988) ‘On “dominant logic”, relatedness and the link between diversity and
performance’, Strategic Management Journal, 9: 639–642.
 ENTREPRENEURIAL NARRATIVES 135

Greene, P.G. and Brown, T.E. (1997) ‘Resource needs and the dynamic capitalism typology’,
Journal of Business Venturing, 12: 161–173.
Hamel, G. and Prahalad, C.K. (1994) Competing for the Future. Cambridge, MA: Harvard
Business School Press.
Hart, M.M., Stevenson, H.H. and Dial, J. (1995) ‘Entrepreneurship: A definition revisited’,
Frontiers of Entrepreneurship Research, 75–89.
Hill, R.C. and Levenhagen, M. (1995) ‘Metaphors and mental models: Sensemaking and
sensegiving in innovative and entrepreneurial activities’, Journal of Management, 21:
1057–1074.
Huff, A.S. and Schwenk, C.R. (1990) ‘Bias and sensemaking in good times and bad’, in
A.S. Huff (ed.), Mapping Strategic Thought. New York: Wiley. pp. 89–108.
Jenkins, M. and Johnson, G. (1997) ‘Entrepreneurial intentions and outcomes: A comparative
causal mapping study’, Journal of Management Studies, 34: 895–920.
Kirzner, I.M. (1985) Discovery and the Capitalist Process. Chicago: University of Chicago Press.
Langfield-Smith, K. (1992) ‘Exploring the need for a shared cognitive map’, Journal of
Management Studies, 29: 349–368.
Laukkanen, M. (1994) ‘Comparative cause mapping of organizational cognitions’,
Organization Science, 5: 322–343.
Leonard-Barton, D. (1995) Wellsprings of Knowledge. Cambridge, MA: Harvard Business
School Press.
Levenhagen, M., Porac, J.F. and Thomas, H. (1993) ‘Emergent industry leadership and the
selling of technological visions: A social constructionist view’, in J. Hendry and G. Johnson
(eds), Strategic Thinking: Leadership and the Management of Change. New York: Wiley.
Lieberman, M. (1989) ‘The learning curve: Technology barriers to entry and competitive sur-
vival in the chemical processing industries’, Strategic Management Journal, 10: 431–447.
Lippman, S. and Rumelt, R.P. (1982) ‘Uncertain imitability: An analysis of interfirm differ-
ences in efficiency under competition’, Bell Journal of Economics, 13: 418–453.
McGee, J.T. and Thomas, H. (1986) ‘Strategic groups: Theory, research, and taxonomy’,
Strategic Management Journal, 7: 141–160.
Mahoney, J.T. and Pandian, J.R. (1992) ‘The resource-based view within the conversation of
strategic management’, Strategic Management Journal, 13: 363–380.
Miller, D. (1993) ‘The architecture of simplicity’, Academy of Management Review, 18: 116–138.
Miller, D. (1996) ‘Configurations revisited’, Strategic Management Journal, 17: 505–512.
Miller, D. and Chen, M.J. (1996) ‘The simplicity of competitive repertoires: An empirical
analysis’, Strategic Management Journal, 17: 419–439.
Milligan, G.W. and Cooper, M.C. (1985) ‘An examination of procedures for determining the
number of clusters in a data set’, Psychometrika, 50 (2): 159–179.
Palich, L.E. and Bagby, D.R. (1995) ‘Using cognitive theory to explain entrepreneurial risk-
taking: Challenging conventional wisdom’, Journal of Business Venturing, 10: 425–438.
Penrose, E. (1959) The Theory of the Growth of the Firm. Oxford: Oxford University Press.
Peteraf, M.A. (1993) ‘The cornerstones of competitive advantage: A resource-based view’,
Strategic Management Journal, 14: 179–191.
Pitt, M. (1998) ‘A tale of two gladiators: “Reading” entrepreneurs as texts’, Organization
Studies, 19: 387–414.
Porac, J.F., Thomas, H., Wilson, F., Paton, D. and Kanfer, A. (1995) ‘Rivalry and the industry
model of Scottish knitwear producers’, Administrative Science Quarterly, 40: 203–227.
Porter, M.E. (1985) Competitive Advantage. New York: Free Press.
Porter, M.E. (1996) ‘What is strategy?’, Harvard Business Review, 74: 61–78.
Powell, W., Koput, K. and Smith-Doerr, L. (1996) ‘Interorganizational collaboration and the
locus of innovation: Networks of learning in biotechnology’, Administrative Science
Quarterly, 41: 116–145.
Prahalad, C.K. and Bettis, R.A. (1986) ‘The dominant logic: A new linkage between diversity
and performance’, Strategic Management Journal, 7: 485–501.
Reger, R.K. and Huff, A.S. (1993) ‘Strategic groups: A cognitive perspective’, Strategic
Management Journal, 14: 103–124.
136 INFERRING MANAGERIAL KNOWLEDGE

Reger, R.K. and Palmer, T.B. (1996) ‘Managerial categorization of competitors: Using old
maps to navigate new environments’, Organization Science, 7: 22–39.
Romesburg, H.C. (1984) Cluster Analysis for Researchers. Malabar, FL: Krieger Publishing.
Russell, R.D. (1999) ‘Developing a process model of intrapreneurial systems: A cognitive
mapping approach’, Entrepreneurship Theory & Practice, 23: 65–84.
Thompson, R. (1986) ‘Understanding cash flow: A system dynamics analysis’, Journal of
Small Business Management, 24: 23–30.
Tyler, B.B. and Steensma, H.K. (1998) ‘The effects of executives’ experiences and perceptions
on their assessment of potential technological alliances’, Srategic Management Journal, 19:
939–965.
Wade, J., Porac, J. and Pollock, T. (1997) ‘Worth words and the justification of executive pay’,
Journal of Organizational Behavior, 18: 641–664.
Wagner, J.A. and Gooding, R.Z. (1997) ‘Equivocal information and attribution: An investi-
gation of patterns of managerial sensemaking’, Strategic Management Journal, 18: 275–286.
Weber, R.P. (1990) Basic Content Analysis. 2nd edn. Newbury Park, CA: Sage.
6
S PAT I A L I Z I N G
KNOWLEDGE

Placing the Knowledge Community

of Motor Sport Valley

Nick Henry and Steven Pinch

ABSTRACT

Within the contemporary world of ‘the knowledge economy’, understanding

the spatial organization of knowledge production has become a key issue.
Drawing on recent emphases in economic geography on the cultural construc-
tion and social embeddedness of economic success, this chapter outlines an
attempt to understand one form of spatial organization of knowledge – the
agglomeration or cluster – using the analytical concept of ‘the knowledge com-
munity’. Using the example of Motor Sport Valley, the chapter provides an
empirical and methodological example of how one might place a knowledge
community. The British motor sport industry dominates its world of production
with a regional agglomeration, Motor Sport Valley, centred on Oxfordshire and
stretching into East Anglia and down into Surrey. The chapter provides empiri-
cal demonstration of the processes of knowledge generation and dissemination
constituting the space (and knowledge community) of Motor Sport Valley.

PROLOGUE

The Role of Geography

The sub-discipline of economic geography conveys essentially two major

propositions. First, that a simple mapping of economic activity – sectors,
trade flows, multinationals, R&D expenditure, labour migration, and so
138 INFERRING MANAGERIAL KNOWLEDGE

on – reveals the geographically uneven development of economic life. It

highlights the realities and interconnections of economic systems at
various spatial scales: for example, what Dicken (1998) calls the ‘global
triad’, the ‘sunbelt’ and ‘rustbelt’ of the USA (Sawers and Tabb, 1984) or
inner-city decline and the rise of commuter belts and dormitory villages.
Second, it argues that these changing geographies matter economically,
politically and socially. Moreover, they matter much more than just
shaping (governmental) responses to the uneven processes of economic
development. For geographers, space and place matter because the objects
and processes of our investigations – the (multinational) firm, markets
(e.g. NAFTA), globalization, fair trade, the information highway, com-
muter belts, sovereignty, and so on – are constituted in, and through,
spatial relations (measured in various ways such as distance, proximity,
connectivity, presence or absence, borders, etc.). In other words, the out-
comes of economic systems and our responses to them are shaped by
processes that are inherently geographical in character.

Making ‘Maps’ in Contemporary Economic Geography

The following chapter on the economic agglomeration or industrial clus-

ter of Motor Sport Valley is an example of the spatial constitution of an
economic system. Essentially, the chapter argues for a number of eco-
nomic processes whose geographies (sometimes referred to in the jargon
as ‘spatialities’) both overlap and intertwine within a particular territory.
Labour market and firm formation, technology transfer and knowledge
flows have become centred on a geographical region of the UK little
known to those outside of the industry who comprise this economic
space. Together they constitute, and are partially constituted by, Motor
Sport Valley.
Ironically, considering our disciplinary background as geographers, it
has needed the editors of this collection to highlight, and reflect upon, the
processes of ‘mapping’ we have undertaken to represent and demonstrate
this economic system. The disciplinary history of geography, of course,
originates from the first systematic attempts to map the physical features
of the world (Livingstone, 1992). Nevertheless, remarkably, the following
chapter is distinctive within our sub-discipline by virtue of its attempt to
‘track’ a set of processes that are arguably growing in significance with the
move to centre-stage of ‘knowledge’ as a key resource of the contemporary
economy (Bryson et al., 2000; Castells, 1996; Coyle, 1997; Leadbetter, 1999).
The importance of knowledge as a factor of production has emphasized
the social embeddedness of economic activity (Grabher, 1993; Granovetter
and Swedberg, 1992; Hodgson, 1999). Within economic geography, the
challenge is to represent (and map) the variety of systemic elements
(‘untraded interdependencies’, in the language of Storper and Salais,
1997) that comprise the ‘intangibles’ of embeddedness. Thus far, as an
 SPATIALIZING KNOWLEDGE 139

empirical activity, this mapping has tended towards a concentration on a

series of regional economic success stories – for example, Silicon Valley
(Saxenian, 1994), the Third Italy (Pyke et al., 1990) and the City of London
(Amin and Thrift, 1992) – whose economic dynamism has been attributed
to their ability to embed (or make ‘sticky’ [Markusen, 1996]) regionally
distinctive forms of economic activity in the face of economic globaliza-
tion. Other disciplines, too, recognize the territorial basis of competitive
advantage (Castells, 1996; Clark, 1999; Krugman 1991, 1995; Porter, 1990),
although fierce debates are ongoing as to the structural basis for such
advantage and the level and type of empirical evidence required to
deepen understanding (see, e.g., Economist, 1999; Martin, 1999). Clearly
here, too, the potential of mapping techniques is great.

Mapping Strategic Knowledge:

a Geographical Perspective?

In recent years, and as evidenced in our chapter below, geographers

have adopted a strongly ‘relational’ perspective in the conceptualization
of the object of study (Allen et al., 1998; Massey, 1993; Massey et al., 1999;
Thrift, 1996b). If we can be criticized for often having taken the firm and
its decision-making processes as a ‘relative given’ (notwithstanding an
earlier ‘behavioural’ period in the discipline, see Hayter and Watts, 1983),
for stopping at the entrance to the black box if you like, current concep-
tualizations are likely to highlight the extremely porous and relational
nature of the firm in today’s economy. Of course, firm boundaries still
have important implications for many factors related to legal status, profit-
ability, balance sheets, labour relations, and the like. Nevertheless, firms
can increasingly be viewed as mere nodes in a network or web of diverse
sets of (spatially constituted) economic relationships; relationships that
push and pull and are multi-scalar, diverse, heterogeneous and uneven.
In this context, the resource base of the firm could be taken to be its
uniqueness as the locus of a particular set of economic flows and relation-
ships, and its strategic knowledge rests with its ability to manage
these flows and their internal combination. In this sense, the distributed,
differential, partial and complex character of knowledge is emphasized
and mapping becomes a key technique for identifying, tracking and moni-
toring this complexity. Moreover, this mapping becomes focused on the
chains (even knowledge chains [ESRC, 2000]) of which the firm is but a
part (business to business, value, commodity networks, organizational
alliances, labour, technology, etc.) in a greater systematic blurring of
environment and firm.
Research themes centred on factors such as knowledge, competitive
advantage, clusters, organizational geography, networks, technology
transfer, embeddedness and mapping are increasingly oblivious to, if not
positively seeking to transcend and refashion, disciplinary boundaries.
140 INFERRING MANAGERIAL KNOWLEDGE

We thank the editors for the opportunity they have provided to make a
small contribution to this process of development.

INTRODUCTION: KNOWLEDGE,
ECONOMY, SPACE

There is now an array of descriptive labels on offer that attempt to capture

the heightened role of ‘knowledge’ within contemporary economies.
Initially encompassed within theories of transition such as post-Fordism
(Lipietz, 1992), flexible accumulation (Harvey, 1989), flexible specialization
(Piore and Sabel, 1984) and disorganized capitalism (Lash and Urry, 1987),
other entrants on to the scene include reflexive accumulation and economies
of signs and space (Lash and Urry, 1994), knowledge-based capitalism (Florida,
1995), the informational age (Castells and Hall, 1994), the learning economy
(Lundvall and Johnson, 1994), the digital economy (Pratt, 1998), the demate-
rialized economy (Quah, 1996), the network society (Castells, 1996), soft capi-
talism (Thrift, 1996a) and the weightless world (Coyle, 1997). With varying
emphases, all highlight how ‘knowledge’ in a variety of overlapping
forms (e.g. aesthetic, cognitive, scientific, discursive, digital, information,
tacit, etc.) is central to the operation of the contemporary advanced
economies. And with varying emphases again, all recognize that the spa-
tial basis of production is being transformed under the imperatives of the
‘knowledge economy’. Moreover, it is increasingly clear that this spatial
basis is not leading to spatial homogeneity but is generating a distinctive,
yet different, pattern of uneven economic development – a new geogra-
phy of knowledge production. Furthermore, as MaCleod (2000) high-
lights, and as part of an emergent ‘regional political economy’, a strong
component of this new geography has been an argument for the regional
scale as the territorial expression of the spatial organization of knowledge
production (Castells and Hall, 1994; Cooke and Morgan, 1993; Morgan,
1997; Storper, 1997).
This chapter examines one regional expression of knowledge production
in the form of Motor Sport Valley. Elsewhere, drawing on elements of the
‘cultural turn’ in economic geography, we have argued that a greater
understanding of the growth of Motor Sport Valley can be gained from
understanding this region as a ‘knowledge community’: a geographically
concentrated node of knowledge production on how to construct the
‘best’ racing cars in the world (Henry and Pinch, 1997; Pinch and Henry,
1999a, 1999b; Pinch et al., 1997). The aim of this chapter is to demonstrate
the processes of knowledge generation and dissemination critical to the
constitution of the knowledge community known as Motor Sport Valley. In
achieving this aim, the chapter outlines a methodology for tracking how
‘knowledge’ circulates within a regional world of production. Whilst the
regional knowledge network is the main focus, attention is drawn also to
the political economy of the world of production of which this network is
 SPATIALIZING KNOWLEDGE 141

but a part, albeit an essential one. Drawing on this analysis, the chapter
concludes with some thoughts on how the example of Motor Sport Valley
might influence our understanding of (a) the ‘region’ as an organizational
mode of production within (b) ‘the knowledge economy’.

EXPLAINING MOTOR SPORT VALLEY

Motor Sport Valley (see Figures 6.1 and 6.2),1 the name given to the
regional agglomeration of the British motor sport industry, dominates the
world’s production of racing cars. It represents a classic example of a
world-leading regional agglomeration of small firms (Henry, 1999; Henry
and Pinch, 1997; Henry et al., 1996; Pinch et al., 1997). The industry
employs well in excess of 30,000 people and consists of scores of small and
medium-sized firms clustered in a 50-mile radius around Oxfordshire in
southern England. Approximately three-quarters of the world’s single-
seater racing cars are designed and assembled in this region, including
the vast majority of the most competitive Formula One, Championship
Auto Racing Teams and Indy Racing League cars. The region is also the
base for a large number of the world’s rallying teams.
In economic geography, one of the most influential explanatory frame-
works for regional agglomeration was the ‘Early Californian’ ‘Approach’,
which interpreted spatial concentration as an attempt to minimize the
transaction costs associated with inter-firm linkage (Scott, 1988; Scott and
Storper, 1987). Recently, Storper (1995, 1997) criticized the inadequacy
of theories such as these based as they are on analysing the transfer of
physical inputs and outputs. Drawing upon what he terms the ‘new
heterodoxy’ of regional analysis, he focuses upon the socially and institu-
tionally embedded character of economic activity, the role of habits and
customs in forging technological and production technologies, and the
critical and problematic nature of knowledge acquisition in contemporary
economies (Storper, 1997; Storper and Salais, 1997). Storper argues that a
more effective approach to the enigma of regional concentration may lie
in the notion of untraded interdependencies. These consist of various con-
ventions, rules, practices and institutions that combine to produce ‘frame-
works of economic action’. These frameworks structure ‘possible worlds
of production’, that is, frameworks of foreseeable action for economic
actors (Storper and Salais, 1997: 20). Possible worlds of production
become ‘real worlds of production’ when they exhibit ‘coherence’. For
Storper and Salais, coherence in the contemporary advanced economies
‘means creating the absolute advantages that propel economic specializa-
tion’ (1997: 21). Regional agglomerations such as Motor Sport Valley that
have come to the fore in recent years are, in Storper and Salais’ eyes,
successful examples of coherence, real worlds of production identified
most especially by their profitable firms and international share of market
traded products (1997).
142 INFERRING MANAGERIAL KNOWLEDGE

26
35

33,36 39
Birmingham 8
22 30
17,20 18
42

21 Cambridge
1 41
27
Silverstone Circuit
23
28,43 11 9
2,5
3
14,29,31,34
37 13
Oxford 40
4 24 10 16

London
19
Reading 32
7
12
38
6

Southampton
15

25

0 50 km

FORMULA ONE
1 Benson and Hedges 4 Rothmans Williams Renault
Total Jordan Peugeot 5 Stewart Ford
2 Danka Arrows Yamaha 6 Tyrrell Racing Organization
3 Mild Seven Benetton Renault 7 West McLaren Mercedes

OTHER FORMULAE and TOURING/RALLY CARS

8 Argo Cars 21 Marrow-Jon Morris Designs 34 Spice Racing Cars
9 Audi Sport 22 Mitsubishi Ralliart 35 Spider
10 BMW Team Schnitzer 23 Motor Sport Developments 36 Total Team Peugeot
11 Bowman Cars 24 Nissan Motorsport Europe 37 TWR Racing
12 Elden Racing Cars 25 Penske Cars Ltd. 38 Valvoline Team Mondeo
13 Ford Motorsport 26 Pilbeam Racing Design Ltd. 39 Van Diemen International
14 Galmer Engineering 27 Prodrive 40 Vector Racing Car
15 G Force Precision Engineering 28 Pro Sport Engineering Ltd. Constructors
16 Hawke Racing Cars 29 Ralt Engineering 41 Vision
17 Jedi 30 Ray Mallock 42 Volkswagen – SBG Sport
18 Lola Cars Ltd. 31 Reynard Racing Cars Ltd. 43 Zeus Motorsport
19 Lyncar 32 Ronta Engineering
20 Magnum 33 Rouse Sport

Figure 6.1 Motor Sport Valley: constructors

 SPATIALIZING KNOWLEDGE 143

111,121

23
6
102 Norwich 115,151
144 7,138,139 31,62
50 34
4,64 51,59 95 19
163
1 46
68 132 15 30
152 44 150
37,105,164 Birmingham 93
76
55,74,89 80,129,131
9
13,14,32 69 85
113 28,123
33,54,90 107
133 79 37 110,142,154 Cambridge 42,134
11,25,45,
Silverstone Circuit 70
72,99,104
38
98,147 57
27,117,153,160,161 26,125,149
17
22 92,108,120,112 81

20 63,100
104
24 106 Oxford 86
51
56,71,
156
118 87 91
65,124,146 136
67,73
78
135 London 49
43 41,97,143 47,126
Reading 18
39 75, 101,148
162,140 112 127 66,94
21 58 157 48 35,82
2 96
61 130
83,145 5 43,157 128
10 16 109
161 40
77 12 84
139
158
119
Southampton 36
114 88
52
3
29

60 0 50 km

SUPPLIERS/DESIGNERS/SERVICES
1 Accralite Piston 33 Colledge and Morley 65 Griffin Motorsport 97 MM Competition 129 Rollcentre
Division 34 Composite Wings 66 Guy Croft Tuning Systems Ltd. 130 RT Quaife
2 Activa Technology 35 Connaught 67 Hewland Engineering 98 Mo Tec (Europe) 131 Safety Devices Ltd.
3 Active Sensors Ltd. Competition Engines 68 Hi-Tech Motorsport 99 Motorsport 132 Samco Sport
4 ADS 36 Corbeau Seats 69 HJS Motorsport Catalysts Distribution Ltd. 133 SAS Motorsport
5 Aerospace Metal 37 Cosworth Engineering 70 Holbay Racing 100 Mountline 134 Scholar Engines
Composites 38 Cranfield Impact Engines Ltd. Racing Ltd. 135 Serck Marston
6 AES Engineering Centre 71 Icore International 101 MSAS Motorsport 136 Serdi (UK)
7 Alcon Components 39 DCL Components 72 Ilmore Division 137 Servo and
8 Aldon Automotive 40 Dove Composite Engineering Ltd. 102 MTEC Electrical Sales
9 Aleybars 41 DPS Composites 73 Impact Finishers 103 Mugen (UK Base) 138 Spa Aerofoils Ltd.
10 Alfred Bull 42 Dunnell Engines 74 Induction 104 Nimbus Motorsport 139 Spider Engineering
11 Aluminium Radiators 43 Dymag Racing UK Technology Group 105 Omega Pistons 140 Spot On Control Cables
and Cooling Systems 44 Dynamic Suspensions 75 International Spares 106 Oselli Engineering 141 Stack Ltd.
12 Alresford Tectonics 45 Earl’s UK 76 James Lister and 107 Pace Products 142 Steve Bunkhall
13 Andy Rouse 46 Eibach Suspension Sons Ltd. 108 Pallas Connections 143 Stone Foundaries
Engineering Ltd. Technology Ltd. 77 Janspeed 109 Piper Cams 144 Staubli Unimation
14 AP Racing 47 Elaborazione Colasuna 78 Janus Technology 110 PI Research 145 Strain Gauging Co
15 Arrow Race Engine 48 Electron Beam Processes 79 John Morris Designs 111 Premier Fuel 146 Swindon Racing
Components 49 Engine and 80 Kartronix Systems Engines Ltd.
16 Astratech Dynomometer 81 KDM Motorsport 112 Prisma Design 147 The Monogram Co
17 ATL Competition 50 Engineering and 82 Kent Aerospace 113 Proctor and Chester 148 Think Automotive
Equipment Motorsport Supplies Castings Measurements 149 Tickford
18 Auriga 51 En Tran 83 Kulite Sensors 114 Quantum Racing 150 Tony James
19 Aurora 52 ERL 84 Lane Electronics Service Component Wiring
20 Automotive 53 Eurotech Motorsport 85 Langford and Peck Ltd. 115 Quentor Cases 151 Trick Machinning Ltd.
Developments 54 Farndon Engineering 86 Leda Suspensions 116 Racam Precision 152 Trident Racing Supplies
21 Avon Racing Division 55 FF Developments 87 Lee Products 117 Race Logic 153 Turbo Technics
22 AWF 56 Flexible Hose Supplies 88 Lemo UK 118 Raceparts UK 154 TWI
23 Aztek Racing Division 89 Lifeline 119 Race-Tec NAK 155 Uniclip Automotive
24 Berkeley Prowse 57 Fuel Safe UK Ltd. 90 Lifeline Fire and 120 Ralt Engineering 156 Universal Air Tool
25 BM Motorsport Ltd. 58 Genesis Electronic Safety Systems 121 Ramair Filters 157 Universal Grinding
26 Bob Sparshott Systems 91 McDonald Race UK Ltd. Services
Engineering 59 GKN Motorsport Engineering 122 Rapid International 158 Warrior Automotive
27 Brembo 60 GKN Westland 92 Mag Think 123 Ravic Engineering Research Ltd.
28 BTB Exhausts Aerospace Automotiveneco 124 Rayfast 159 Willans Racing Harness
29 C & B Consultants 61 Gomm Metal TWIr UK 125 Ray Mallock 160 WP Competition
30 Champion Developments 93 Mark Dunham Racing 126 RESB Systems
Manufacturing 62 Graham Goode Racing 94 Minister Motorsport International Ltd. 161 WP Suspension (UK) Ltd.
31 Clarendon Eng. & 63 Graham Hathaway 95 Mira 127 Ricardo Consulting 162 X Trac
Motorsport Supplies Racing 96 Monk Aeroelectronic Engineers 163 Zephyr Cams Ltd.
32 Colinton Engineering 64 Grainger and Worrall Design Ltd. 128 Risbridger 164 Zeus Ltd.

Figure 6.2 Motor Sport Valley: suppliers

144 INFERRING MANAGERIAL KNOWLEDGE

In arguing for untraded interdependencies, frameworks of economic

action and worlds of production, Storper and Salais highlight the concep-
tual frameworks needed to produce knowledges of labour, capital, mate-
rials, markets, and so on, and the successful combination required to
produce a good. In addition, they highlight the current territorialization of
such frameworks of knowledge in certain spaces and places. Arguing that
this approach bypasses the distinction between notions of ‘the economic’
and ‘the non-economic’, Storper and Salais’ work, nevertheless, is one of
a recent series of attempts at unravelling the ‘non-economic’ social rela-
tions and structures that ‘embed’ growth (see numerous chapters in Lee
and Wills, 1997, for example). Other influential strands of thought within
this ‘new heterodoxy’ or emergent ‘regional political economy’ include,
for example, the concepts of institutional thickness (Amin and Thrift, 1994),
learning regions (Morgan, 1997), governance in a variety of forms (Goodwin
et al., 1993; Jessop, 1990; Jones, 1998; Peck, 1996; Peck and Tickell, 1994),
the network paradigm (Castells, 1996; Cooke and Morgan, 1993) and the
associational economy (Cooke and Morgan, 1998), to name but a few.
Whatever the label, all of the above perspectives are part of a broader
conceptual shift in economic geography that recognizes the socially
embedded character of production systems; the cultural and discursive
aspects of production systems; and the critical importance of knowledge,
as a cultural construction, in modern economies. While there has been a
great deal of discussion about the role of knowledge in gluing together
agglomerations of firms, there have been few demonstrations of the
processes through which this knowledge is spread. To some extent, this
lack of work is hardly surprising since it is acknowledged that many of
these processes of knowledge circulation are untraded, intangible, covert,
sometimes immoral (and even illegal). Understanding these processes
therefore requires a great deal of work to get ‘inside’ a particular ‘world
of production’.
Elsewhere we have argued that a focus upon the process of knowledge
generation provides a convincing explanation for the origins and develop-
ment of Motor Sport Valley (Pinch and Henry, 1999a, 1999b; Pinch et al.,
1997). For example, the crucial impetus to the dynamic growth of Motor
Sport Valley – the element that glued the contingent factors of club racing,
airfields, reduced market competition, tobacco sponsorship, and so on,
together – was a radical shift in technological regime or knowledge
system. Just as the shift from valves to transistors and semi-conductors
was reflected in a spatial shift in the electronics industry from East Coast
USA to Southern California (eventually to be labelled Silicon Valley), so the
shift to an aerospace-inspired technological trajectory ultimately saw the
locus of the world’s motor sport industry shift from northern Italy to
southern England (eventually labelled Motor Sport Valley). Furthermore,
a knowledge-based approach is also important in accounting for the main-
tenance and growth of the British motor sport industry and Motor Sport
Valley as well as its origins. Knowledge generation and dissemination
 SPATIALIZING KNOWLEDGE 145

have been crucial to maintaining the dominance of the region and have
taken place, we argue, through the medium of ‘the knowledge commu-
nity’. Here we define a knowledge community as a group of people (prin-
cipally designers, managers and engineers in this case), often in separate
organizations but united by a common set of norms, values and under-
standings, who help to define the knowledge and production trajectories
of the economic sector to which they belong. In the case of Motor Sport
Valley, the region is synonymous with the knowledge community of motor
sport; the core of the community’s spatial organization is that of the Valley.
The key processes that construct and maintain this community occur on
both a traded and untraded basis.

THE METHODS OF PLACING A KNOWLEDGE

COMMUNITY

The evidence presented here was collected during the course of a two-year
research project which comprised two main parts. Part One was an investi-
gation into the scale, character and geographical location of the British
motor sport industry (for example, see Figures 6.1 and 6.2). This was
achieved through analysis of specialist literature and interviews with ‘key
informants’. The specialist literature on the industry is available in large
quantities and is of a high quality. For example, there are numerous weekly
and monthly magazines on the industry, and dozens of books are produced
each year on the sport and the industry that have been collated in several
specialist libraries. Although many of these publications are hagiographies
and should be interpreted with caution, they provide a wealth of back-
ground detail and technical information that few other industries can
match. In addition, key informant interviews were undertaken with
10 individuals, including academics, consultants, industry journalists and
representatives of the major regulatory bodies of the industry.
The aim of Part Two of the project was to gain knowledge on the nature
and geography of both traded and untraded interdependencies in the
industry. Part Two therefore involved 50 in-depth interviews with com-
pany executives, managers, designers and engineers. Remarkably, we
received no refusals to be interviewed. Face-to-face semi-structured inter-
views took place between June 1996 and January 1997. Each interview, on
average, lasted two to three hours and was taped and subsequently tran-
scribed. In many cases, interviews were followed by tours around the
premises. The choice of sample followed a ‘purposive’ sampling approach
(Layder, 1998) combining a sample of the most dominant companies in the
industry and coverage of the different levels of competition/construction.
The aim was to use qualitative methods to cross-check answers in depth
and undertake ‘triangulation’ of findings. In the case of component sup-
pliers, the interviews were continued until ‘saturation’ was achieved (i.e.
the main patterns became established and further interviews were felt
146 INFERRING MANAGERIAL KNOWLEDGE

unnecessary [Cresswell, 1994]). The level of access achieved meant that the
final sample included virtually every major company in the British motor
sport industry with every Formula One team in the UK bar one inter-
viewed, every UK-based Indycar manufacturer, five touring car designer/
manufacturers, 10 leading supplier/subcontractor firms, three engine
builders, one rally car manufacturer/team and a Formula Ford manufac-
turer. The project was particularly successful in gaining access to the lead
personnel in the industry, including Managing Directors, Technical
Directors and Chief Designers as well as Aerodynamicists, Gearbox
Engineers and Composites Specialists.
Overall, whilst the methodology was based on well-established quali-
tative methods, it was innovative in its use of a variety of methods to try
to track ‘knowledge processes’. The methodology was driven by an
attempt to make graphic the ‘intangibility’ of the world of motor sport
production, that is, an attempt to try to delineate knowledge (ideas, infor-
mation, etc.) and the processes that create, sustain and circulate this
through space. Thus, this search meant that as researchers we held upper-
most in our minds a particular version of ‘culture’, in terms of shared
meanings and understandings (of say a piece of information) within the
industry, and continually investigated the issue of how this was formed
as an element of production.

PLACING A KNOWLEDGE COMMUNITY

The methodology produced a number of ‘traces’ of knowledge processes

and their geographies, which are the subject of this major section of the
chapter. For the sake of organization, these traces are divided into two
major sub-themes on the basis of a rather crude dichotomy between
‘knowledge’ as embodied literally in people who ‘travel’ and disembodied
ideas and knowledges travelling across space.2

Knowledge, Process, Community and Space (1)

A first approach to tracking ‘knowledge’ was to recognize its embodi-

ment. What is meant here is not the normal view as in embodiment in a
piece of machinery, paper or other artefact (although see the recent work
on actor-network theory), but literally embodiment as in a thinking,
breathing body such as the engineer (see, e.g., Allen and du Gay, 1994;
Boden, 1994; Du Gay, 1996; McDowell, 1997; McDowell and Court, 1993;
Thrift, 1994, for work that similarly recognizes embodiment in this sense).

staff turnover
One of the most important ways in which knowledge is spread within
the motor sport industry is by the rapid and continual transfer of staff
(bodies) between the companies within the industry. Especially at the end
 SPATIALIZING KNOWLEDGE 147

Rory Byrne – R&D and design – South African

(D.o.B. 10.1.44)

Witwatersrand University 1960s

(Degree in Industrial Chemistry – South Africa)

Polymer manufacturing plant – late 1960s

(Chief Chemist – Johannesburg, South Africa)

Set-up own business – 1967–73

(Importing & selling motor spares & then engine
development & build – South Africa)

Royale (Formula Ford) – 1973–7

(‘I was so involved in racing that there was only one place to
go – the UK.’ – Designer – UK)

Toleman/Benetton – 1977–90
(Co-designed with John Gentry first Toleman car, an
F2, in 1980 & then designed, or co-designed with Gentry, F1 cars
from 1981 to 1990 – Witney, Oxon.)

Reynard – October 1990–2

(Involved in design of aborted X project – Bicester, Oxon.)

Benetton F1 – 1992–December 1996

(Responsible for R&D – Enstone, Oxon.)

Ferrari F1 – January 1997–present

(Chief designer – Maranello, Italy)

Figure 6.3 Career history

of every racing season, there is an intense period of negotiation as design-

ers, engineers, managers and, of course, drivers move between teams. As
part of the research, the career histories of 100 leading designer/engineers
within the industry were mapped (see Figures 6.3 and 6.4 for examples).3
This mapping revealed a move, on average, once every 3.7 years and an
average total of eight moves in a career in the industry. Similarly, a study
of advertisements for technical posts in Formula One between 1996 and
1997 revealed vacancies for 93 posts, more than 10 per cent of this for-
mula’s total employment at that time.

career trajectories
Whilst the above evidence highlights the mobility encapsulated within
the labour market, its significance also stems from the geographical
148 INFERRING MANAGERIAL KNOWLEDGE

Frank Dernie – Aerodynamicist/Designer

(D.o.B. 3.4.50)

Imperial College late ’60s/early ’70s

(Mechanical Engineering Degree – London)

David Brown Industries 1970s

(Working with gears)

Garrods 1970s
(Record players – Swindon, Wilts)

Hesketh/Williams 1977–9
(Consultancy work – co-designed 1978 Hesketh with
N. Stroud – Towcester, Northants/Didcot, Oxon.)

Williams 1979–88
(Aerodynamicist & R&D – co-designed 1986–8 cars with
P. Head – Didcot, Oxon.)

Lotus 1988–90
(Chief Engineer – co-designed 1989 car with
M. Coughlan – Norfolk)

Ligier 1990–2
(Technical Director – France)

Benetton 1992–July 1994

(Race engineer & then chief engineer
in 1994 – Enstone, Oxon.)

Ligier July 1994–April 96

(Went back to Ligier after team purchased by Briatore – Technical
Director – France then Oxon.)

Arrows April 1996–7

(Technical Director – moved with T. Walkinshaw’s
take-over – Oxon.)

Lola Cars 1998 – present

(Huntingdon)

Figure 6.4 Career history

 SPATIALIZING KNOWLEDGE 149

expression of this labour market. As the sample career histories of

Figures 6.3 and 6.4 show, Motor Sport Valley acts as the epicentre of a global,
yet highly spatially defined, labour market. In particular, whilst moves occur
on an international scale into, and out of, the Valley, overall, the largest
number of moves occur between sites within the Valley. Moreover, interna-
tional moves out are invariably followed in quick succession by a return to
the Valley, as can be seen in the career history in Figure 6.5. Indeed, these
particular career histories introduce a constant and overriding theme of the
world of production of Motor Sport Valley. As an individual in motor sport,
it is rare that you do not, at some point in your career, spend some time
within Motor Sport Valley (and most likely the majority of your career).
Essentially, this is about joining the knowledge community of motor sport
production. Yet opportunities exist in this highly mobile industry to sell your
knowledge elsewhere, and many do so. However, the knowledge of the
Valley is continually being reconfigured and advanced (see below): it is in
effect, a knowledge pool. To move out of the Valley is also to stretch your
relationship within the community, to risk your position within the knowl-
edge loop. Thus, few stay away from the Valley for long, returning once
more to refigure their position within the community of knowledge. An
analogy for the labour market of Motor Sport Valley is that of the dating
agency’s dinner dance. With each change of course (‘racing season’) comes
a change of seating arrangement with, by the end of the evening, occupants
often returning after one or two musical interludes to a seat previously occu-
pied. Furthermore, and skirting the dangers of taking this analogy further,
such mixing does not take place only at the level of the individual: joint career
paths may also be mapped with individuals often moving together, or con-
tinually recombining, throughout their careers (see Figures 6.6 and 6.7).
Thus, ultimately, as was so graphically illustrated in an article in a trade
journal, the industry labour market is made up of people networks such that
‘it pays to be well connected’ (see Figure 6.8).

firm births and firm deaths

It is important to note, however, that this mobility of the labour market is
both voluntary and non-voluntary and it is driven not only by movements
of individuals but also by the continual ‘movement’ of firms. Another
significant feature of Motor Sport Valley is the high firm turnover involving
both high rates of new firm formation and high firm death rates. Within
an industry of several hundred firms, only a handful employ more than
200 people, and there is a widespread feeling within the industry that even
this size is dangerous to the competitive edge of the company. Indeed, in a
process of growth reminiscent of Silicon Valley (Saxenian, 1994) and the
Cambridge Phenomenon (Segal, Quince and Wichsteed, 1985), the
hundreds of firms in Motor Sport Valley today can be traced to a few
pioneers. For example, Figure 6.9 provides an incomplete family tree for
March, a spin-off from Cosworth (Northampton) in 1969.
150 INFERRING MANAGERIAL KNOWLEDGE

Sergio Rinland – designer

(from Argentina)

PRS
(Formula Ford)

Ralt early/mid-1980s?
(Surrey)

RAM F1 – early/mid-1980s
(Bicester, Oxon.)

Williams GPE 1985

(Didcot)

Brabham F1 1986–7
(Surrey)

Dallara 1987–9
(Designed first Dallara F1 car – Italy)

Scuderia Italia 1988

(Chief designer – Italy)

Brabham 1989–91
(Technical director – Surrey)

Astauto – December 1991

(Set up own company – many staff ex-Brabham – Designed 1992
F1 car for Fondmetal – Surrey)

BMS Lola Scuderia Italia 1993

(Temporary work as development engineer – Italy?)

Consultancy: designed AAR/Eagle IndyCar gearbox 1994/95?, technical consultant for

Forti F1 1995, technical director for Opel Team DTM touring car 1995

Benetton April 1996–present

(Composites designer – Oxon.)

Figure 6.5 Career history

 SPATIALIZING KNOWLEDGE 151

Harvey Postlethwaite, Gary Thomas & Patrick Head link

Hesketh 1973–5 Frank Williams Racing

Harvey Postlethwaite early/mid-1970s
Gary Thomas, Patrick Head

Wolf/ Williams 1975/6

H. Postlethwaite, G. Thomas, P. Head

Wolf/Fittipaldi 1976–80 Williams GPE 1976–9

H. Postlethwaite G. Thomas, P. Head

Williams 1980–2
P. Head

Ferrari 1980–8 Wolf/Fittipaldi 1980–2

H. Postlethwaite G. Thomas

Williams 1982–90
G. Thomas, P. Head

Tyrrell 1988–91
H. Postlethwaite
Williams 1990–present
P. Head
TWR /Jaguar 1990–3
G. Thomas

Sauber 1991–2
H. Postlethwaite

Tyrrell 1993
G. Thomas

Ferrari 1992–3
H. Postlethwaite

Tyrrell 1994–98
H. Postlethwaite, G. Thomas

Figure 6.6 Joint career path

March located in nearby Bicester and began a process that has led this
small market town to become the world centre for motor sport production.
Bringing the tree up-to-date, one of March’s spin-offs, Reynard Racing
Cars, has itself sponsored a number of internal spin-offs in the town:
152 INFERRING MANAGERIAL KNOWLEDGE

John Barnard, Mike Coughlan & Giorgio Ascanelli link

Barnard, ‘I also try to instil my design philosophy in the key development engineers at Ferrari, like
Georgio Ascanelli, with whom I have worked in the past at Benetton and Ferrari. I get lots of feedback
from him on the practical aspects of operating the cars at the track and on set-up issues.’ (Wright, 1996)

Giorgio Ascanelli John Barnard Mike Coughlan

Ferrari 1985–6 McLaren 1980–6 Lotus late 1980s?–90

Abarth rally 1987

Ferrari 1987–9
J. Barnard (chief chassis designer) &
G. Ascanelli (race engineer)

Benetton 1990–91
J. Barnard (R&D & designer), G. Ascanelli (race engineer chief) & M. Coughlan
(Barnard’s design assistant)
McLaren F1 1992–4
G. Ascanelli
Ferrari 1992?–5
J. Barnard (tech. director), M. Coughlan (chief designer)

Ferrari 1995–present
J. Barnard (tech. director), G. Ascanelli (chassis/track manager),
M. Coughlan (chief designer)

Figure 6.7 Joint career path

...the critical mass is about 25 to 30, so when our original company got to
25 to 30 we then bought another factory and we started Reynard Manufacturing
and Reynard Racing Cars. … When that grew to about 60 people, we then
started Reynard Composites … and when that group of companies grew to
90 people we started another company which was designs, and then when
that company grew to 120 people we started Reynard’s Special Vehicle
Projects and that has now grown to 20 people. So we have now got 152 people
in the company, but it is made up of separate buildings, some in Bicester and
some in Brackley [a nearby small town] (42)4

The company opened another large site just outside of Bicester. This site is
the production centre for the latest entrant to Formula One, British
American Racing, a combination of Reynard Racing Cars and the Tyrrell
Formula One team (also within the Valley), which has ended the latter
team’s twenty-year presence in Formula One racing. In fact, this is a clas-
sic example of the process of reconfiguration so characteristic of Motor
Sport Valley. One of the longest-running teams in Formula One, Tyrrell
will not so much die as be reborn in a modified and reinvigorated version.
Thus Tyrrell is one in a long line of Formula One constructors who have
come and gone; this includes Pacific, Forti, Simtek and Lola in a field that
rarely exceeds a dozen constructors a year. Lola, a major Indycar producer,
did not last a season in Formula One, with many commentators suggest-
ing that the company had failed to learn from its previous foray into
 It pays to be
well connected
F1 is a small world. Everyone knows
everyone, and most have worked with
each other too. We trace the top careers.
SAUBER

SAUBER HARVEY WOLF WILLIAMS LIGIER

LEO FRANK LOÏC ERIC
POSTLE-
RESS WOLF WILLIAMS DERNIE BIGOIS BROADLEY
THWAITE
Sauber TWR Ligier Lola
Tyrell

WILLIAMS
LOLA
WILLIAMS
BENETTON HAAS HAAS WILLIAMS
BENETTON

PATRICK PAT ROSS NEIL

HEAD SYMONDS BENETTON BRAWN BEATRICE OATLEY
Williams Benetton FERRARI Ferrari McLaren

BENETTON
HAAS
HAAS
HAAS
LOLA
WILLIAMS
GARY ADRIAN JOHN McLAREN ALAN
ANDERSON INDYCARS NEWEY FERRARI BARNARD JENKINS
Jordan ex-Williams ex-Ferrari Stewart

LOLA
HAAS

Figure 6.8 It pays to be well connected (adapted from Cooper, 1997)

 MARCH FAMILY TREE & SPIN-OFF COMPANIES

COSWORTH 1958 ROY WINKELMANN RACING – F2 1960s

Robin Herd (designer) leaves to set up constructor Alan Rees, team mngr, leaves & sets up constructor with school friend R. Herd March also spawned F1 cars – Hesketh 1973 (designed
by Postlethwaite ex-March) Penske (designed by
G. Ferris ex-March) 1974 & Merzari 1978
R. Herd leaves (taking Cosworth employees with him) & co-founds own constructor with Max Mosley
(University contemporary), Alan Rees (school friend) & Graham Coaker (friend of Rees)
Harry Mendel sets up own MENDEL AUTOSPORT
firm building F1 show cars
1980s – Northants
MARCH 1969 – Bicester Jim Brady sets up own
fabrication firm
Dave Bembo BRADY
A. Rees (co-founder) John Thompson sets Bill Stone leaves & sets sets up F3 team FABRICATIONS
leaves & helps set up up sub-contractor up Sabre Automotive with Dave Reeves 1981–present–Bicester
Shadow’s UK F1 for March Adrian Reynard making sets up brake
DB Nick Wirth leaves
operation with J. Oliver sub-contracted parts firm March & sets up Simtek
for March MOTORSPORT Research with Max
ATS buy March’s Alan Mertens sets
1980s Banbury F1 assets & up motor sport
Mosley in 1989
TC take most of engineering firm
AVS SHADOW PROTYPES RAPID March F1 team Galmer
SABRE
1972–80 (UK F1 1970 – Northants MOVEMENTS SIMTEK
Rub Gustafson members with
AUTOMOTIVE (Transmissions it in 1978
operation) – 1977 RESEARCH
1973 Bicester dept) & Nick
Northampton J. Wickham (F2 1989–95 Wasyliw leave
Thompson sets up to set up Roni
GALMER
Bill Stone manager) & Banbury ATS F1 1980s–present
A. Rees, J. Oliver, & fabrication/ A. Reynard sets G. Coppuck Developments
sets own
T. Southgate leave & design business up constructor,
constructor (design) leave 1978–84 Bicester
set up own F1 with T. Southgate Reynard, after March & set up Simtek Grand Prix
(F1 constructor)
Bicester
constructor working on as constructors
other projects in F2 then F1 formed in 1993 by
N. Wirth RONI
ARROWS AUTO RACING BILL STONE DEVELOPMENTS
1977 – Milton TECHNOLOGY RACING 1985–92
SIMTEK
Keynes Circa 1983 1980s Bicester Key
GRAND PRIX
Direct spin-off
1993–5 Banbury
REYNARD B. Stone rejoins SPIRIT RONI Indirect spin-off
RACING CARS Reynard 1982–85 – MOTORSPORTS Merger/take-over
1982–Bicester Slough F3000 circa 1992 Car design spin-offs

Figure 6.9 March family tree

 SPATIALIZING KNOWLEDGE 155

Formula One a decade ago. This time its entry bankrupted the company,
including its Indycar production, only for it to be resurrected once again
by a new Irish buyer. Its production of Indycars continues unabated and it
still holds ambitions to return to Formula One once again. And it does so
from its continued location within Motor Sport Valley, a location held
throughout its chequered history.
In similar vein, the recent Formula One bankrupt Simtek Grand Prix
provides clear evidence of the entwining of labour movement with firm
‘churn’ through high rates of firm births and deaths. On its demise, of the
top eight individuals employed by Simtek, six were traced to other motor
sport jobs within the Valley inside of six months.

The Churning of Knowledge Within

the Valley Community (1)

In summary, the evidence above highlights that a key characteristic of the

industrial organization and labour market of motor sport is a set of
processes leading to the continual ‘churning’ of people. Fundamental to
the argument of this chapter is, first, that this process of ‘churning’ is
centred on, and within, Motor Sport Valley. Second, this churning is a
process of circulating and producing embodied knowledge within the
knowledge community and regional production centre of Motor Sport
Valley. As personnel move, they bring with them knowledge and ideas
about how things are done in other firms, helping to raise the knowledge
throughout the industry. As one interviewee commented:

Whoever comes, it doesn’t matter from where, he [sic] has to change his ways
and adapt, and then once he changes his ways and adapts and he gets part of
the system, then someone else comes and learns from him. Then he moves to
another team and he takes ways of one team into another, you see. And you
go from [leading F1 constructor] to McLaren and say ‘at [leading F1 con-
structor] we did it this way’ and they say ‘oh really, let’s try it’. And then vice
versa, someone comes from McLaren to here and someone goes to Williams
and you get all these exchanges of people and then we learn from each other
all the time. Not trade secrets, obviously, but ways of doing things. If some-
one used to draw wings at McLaren or Williams or whatever and he comes to
draw wings here, he is not going to bring the drawings of McLaren under his
arm, but he is going to bring the ways of doing it in here, so he is allowed to
explain. He may improve because he is going to see what we have got here,
he is going to see the way he did it before and he will say ‘ah, I can mix these
two and get something even better’, and you improve that little bit, and it
keeps happening. (43)

The crucial point is that whilst this process may not change the pecking
order within the industry, this ‘churning’ of personnel raises the knowl-
edge base of the industry as a whole within the region. The knowledge
156 INFERRING MANAGERIAL KNOWLEDGE

community is continually reinvigorated and, synonymous with this, so is

production within Motor Sport Valley.

Knowledge, Process, Community and Space (2)

The second approach is to recognize how ideas/knowledge move rather

than people per se. Thus, this interpretation does move closer to more
commonly used concepts of embodiment of knowledge in machinery, for
example, although, ultimately, the split between the two interpretations
will be seen to be more of a continuum.

suppliers
In this context, an important process of knowledge transfer is through the
links constructors and manufacturers have with the numerous component
suppliers. Many component suppliers work for more than one constructor
making bespoke items of the client’s design (but usually with a substantial
contribution of their own expertise). Clearly, it is in the interests of these
component suppliers not to disclose secrets to other constructors for
whom they also undertake work or they would soon be out of business.
Nevertheless, as many of our respondents acknowledged, over time, there
is a gradual assimilation of the advantages of particular approaches. This
knowledge exchange might occur, for example, because a component sup-
plier subtly steers a constructor away from a sub-optimal way of doing
something; other examples of the fine lines of this process of knowledge
transfer are evident in the quotes from interviewees below:

There is a difference between confidentiality and the knowledge that one

builds up within an environment. (16, Supplier)
We did have an innovative idea which would solve a problem or improve
performance of a particular component, and we did tell the supplier that, and
it ended up being on their standard products and sold to everyone else
(14, Constructor).5
It is much easier if we have problems to phone somebody and they say, ‘yes,
we do that for somebody else’. (11, Constructor)

This means that there will inevitably be some ‘leakage’ of knowledge

throughout the industry as a whole. The reason why this is an advantage
rather than a problem is because of the very rapid rate of innovation in the
industry – as long as a team knows it can keep ahead of its rivals, it is not
a major problem if the ideas eventually get transferred throughout the
industry. But perhaps more importantly, any individual team probably
has more to gain than lose through the skills that components suppliers
learn by servicing a number of different racing car companies. The major-
ity of the leading suppliers are based within Motor Sport Valley and these
firms and personnel are part of the knowledge community.
 SPATIALIZING KNOWLEDGE 157

gossip, rumour, observation

As economic geographers have come to recognize the importance of ‘dis-
cursivity’ in both the philosophical underpinnings of their work (Barnes,
1996; Bryson et al., 1999; Gibson-Graham, 1996; Martin, 1994; Thrift,
1996b) and its empirical application, the importance of gossip, rumour
and observation in defining productive knowledge has been highlighted.
In the case of the motor sport industry, particular sites and mediums of
this process include the pit lane, test track and race meetings, and the
frenzy of speculation encouraged by the large number of specialist maga-
zines and extensive press and television coverage. The role of ‘gossip and
rumour’ in shaping individual careers and, indeed, the uptake of innova-
tions should not be underestimated (see Thrift, 1994, for similar argu-
ments on the City of London):

F1 mechanics are generally chosen by reputation. Few are picked cold from a
business outside racing. They need to know what F1 is about, not simply
because of the proficiency that will bring but because it means the mechanic
will be aware of precisely what he is letting himself in for. This is not a 9 to 5
job and the motor racing grapevine will let a team owner know whether or
not his prospective employee can cope. (Hamilton, 1994: 162)
I get several calls a week from companies in the same industry asking us if
somebody is good for the money or what’s happening here or what’s hap-
pening there. (41)

In terms of observation, teams keep a close eye on their rivals’ cars and
monitor their performance in test periods. Moreover, all the senses are
involved in such monitoring, with, for example, engine noises, gear
changes, lap times and the behaviour of the cars when cornering provid-
ing important clues – for some – as to the design and construction of a car
and its components:
Spying I do a lot of, I think that’s all fair in love and war. So if I get a chance
to photograph someone’s wing profile or something like that I’m there, and
I walk with a camera permanently attached to me. That’s so important. (7)
If you call spying trying to get a look under the bonnet of a car or when the
wheels are off quickly ducking your head down, if you call that spying, yes
there is, but we all do that. (56)

the discursive aspect of technological innovation

Once analysis moves into the realms of gossip, rumour and observation –
understandings between groups of individuals of the meaning of certain
things – the understanding of discourse becomes significant. Shared
beliefs and understandings comprise, in the terms of Storper and Salais,
some of the discursive conventions that make up the framework of action.
In turn, in an industry dominated by technical innovation, this highlights
the discursive aspect of technological innovation.
158 INFERRING MANAGERIAL KNOWLEDGE

Elsewhere, we have discussed the insights the sociology of scientific

knowledge (SSK) literature provides in understanding the social con-
struction of technological innovation (Pinch and Henry, 1999a). Thus, for
example, concerning the role of discourse, SSK argues that crucial to the
generation and dissemination of scientific and technological advances are
shared sets of understandings – discourses that help to interpret the
world and whose uncertain construction are encapsulated within SSK
through the concepts of interpretive flexibility and social closure. In effect, a
version of truth becomes ‘winnowed’ from the various interpretations
that are promulgated (Bijker et al., 1990). Ambiguities are resolved,
usually only partially for a limited period, and certain development tra-
jectories are put in place.
Typically in motor sport, considerable uncertainty abounds about the
value of any new approach. This uncertainty, and period of interpretive
flexibility, can be illustrated by the history of British racing car design in the
late 1970s and early 1980s (see also Pinch and Henry, 1999a). This was a
period in which British-based teams such as Lotus and McLaren were intro-
ducing new approaches at a remarkable rate. Some of these schemes –
which in retrospect look ‘speculative’ and even ‘wild’ – fell by the wayside,
such as four-wheeled drive, six-wheeled vehicles and the double chassis.
There were many reasons for their rejections: some approaches were in-
effective; some were too expensive; some were too good and were banned
by the Fédération Internationale de L’Automobile (FIA), which governed
the sport; or the approaches simply did not have enough support from the
community of engineers to make them work effectively. Thus even some-
thing that today is widespread – carbon fibre – was inevitably greeted
with some suspicion in the early days. The properties of carbon fibre had
been known for some years previously through work in the aerospace
industry and following specialist articles in the motor sport industry
press extolling its virtues. Nevertheless, the material was used in a very
limited fashion in a few designs. Carbon fibre implied a radical departure
from traditional construction metals and a series of questions arose con-
cerning, for example, cost, strength, reliability, weight and, ultimately,
would it help win races? Eventually, the leading designers adopted the
material, although in different ways and from different supplier sources.
Thus it is possible to map the spread of the idea for almost simultaneously
Gordon Murray at Brabham, John Barnard at McLaren and Peter Wright
of Lotus began using carbon fibre. Barnard’s carbon fibre car, MP4,
appeared only a few days before the rather different Kevlar-enhanced
Lotus 88. Eventually, Barnard’s approach became the standard.
Interestingly, at about the same time, some designers were experimen-
ting with another aerospace-inspired technology – ‘ground effects’. These
involve changes to the aerodynamic profile of the lower part of the racing
car through the use of ‘sliding skirts’ to make the car hug the ground and
corner more effectively at high speed. This technology was ‘discovered by
accident’ and virtually ignored in its early days. Later, all the British
 SPATIALIZING KNOWLEDGE 159

teams copied the technology, although only a few did so successfully. As

with carbon fibre, the spread of this concept can be mapped, but in this
case outside of Motor Sport Valley. Ferrari attempted to buy-in the tech-
nology (and knowledge of carbon fibre) through the appointment of
Harvey Postlethwaite from the (British) March–Wolf team. It is widely
acknowledged that Ferrari never fully mastered the technology of ground
effects but, instead, were heavily instrumental in the decision to ban the
technology in 1981.
Such uncertainty leads to a great deal of blind copying. As one senior
engineer commented:

...a car might have a particular style or component which sets it out from the
rest, but that may not be the reason why it is doing well, but people just
blindly copy it because they think… they must have found something, some
way of measuring a gain in performance that we can’t find, so we’ll do it any-
way. (33)

Another designer graphically explained:

...if painting a race car purple gives it a second a lap, I mean next week they
will all be purple. (23)

The outcome is designs led by strong beliefs and discourses about how
cars should look and behave as well as the results of testing.
Similarly, it is SSK that can provide insights into the ‘embodiment in
artefact’ of knowledge, as discussed earlier, in examples like supplier
components. Knowledge consists of more than materialized ideas (knowl-
edge embedded in physical artefacts, technologies) and includes dis-
courses, understandings and ways of doing things, which highlights the
relationship of embodiment as artefacts and embodiment as understand-
ings, meanings and readings of artefacts (see also Gertler, 1994).
Furthermore, SSK highlights the spatiality of these knowledges; how
these factors reside in knowledge production centres such as scientific
laboratories, research centres and factories and how their transportation
is a fraught process. Scientific experiments and technical innovations are
often dependent upon knowledges of accepted ways of doing things that
are seldom written down (Gertler, 1994; Polanyi, 1967). These forms of
knowledge often emerge from sets of people who have undertaken long
periods of apprenticeship and are integrated into networks of contacts.
For this reason, early scientific experiments are often difficult to replicate
in other laboratories. There are so many detailed sets of understandings
involved in the formulation of the experiment or in the construction of
technological artefacts that not all of these can be incorporated by people
attempting replication.
The point is, of course, that both beliefs and discourses about knowl-
edges are debated and determined within the knowledge community, just
160 INFERRING MANAGERIAL KNOWLEDGE

as they are learnt, and the knowledge community is based in Motor Sport
Valley (MSV). Indeed, the (double-faceted) embodied nature of knowl-
edge in the MSV was a theme that was further reinforced on numerous
occasions in our interviews. One example was a claim by a designer that
although some of the important machine tools of the industry were pro-
duced in another country,
They manufacture the tool but they don’t know how to use it for that particu-
lar use. (35)

Or as a foreign national working in the Valley put it:

To make racing cars in southern California was so difficult because the indus-
try is not prepared. So you may find engineers but their mentality is different.
They are aerospace so they think different, they organize different, they are
not used to our racing mentality. (43)

These examples highlight knowledge as a (community) process, which

was summed up by this statement by an English engineer who, whilst
based in Motor Sport Valley, was working for a foreign-based constructor
whose site overseas he visited regularly:

We are trying to educate them about the English way of doing things. (11)

Indeed, Ferrari have run courses for their engineers in Maranello, Italy,
led by engineers from the UK.

the seamless web of science and technology

Thus, SSK helps us to understand how agreements are forged upon
understanding the signs, symbols and conventions of racing car produc-
tion; the theory underlines how technological trajectories may be put in
place as part of worlds of production. Yet a striking element of racing car
production is how some of these trajectories end abruptly, or, more pre-
cisely, are banned through regulation. One of the key insights from SSK –
the seamless web of society and technology – is relevant here. Whilst the
above discussion has highlighted the processes of technological discur-
sivity embodied in the designers and engineers of Motor Sport Valley,
they are only one of the set of actors involved in the process of creating a
‘fast’ car. They are not the only members of the knowledge community.
The message of the ‘seamless web’ is that technologies are political in
the sense that they can be designed, either explicitly or implicitly, to open
some options and close others. Motor sport illustrates these points very
well, for the design of racing cars has been shaped over the years by many
competing interest groups: not least the teams, their sponsors, the televi-
sion companies and the regulators of motor sport. The major sponsors are
keen to ensure that racing is exciting, but it must also be relatively safe or
the publicity resulting from death or injury to drivers and/or spectators
 SPATIALIZING KNOWLEDGE 161

will generally have a negative impact upon the sponsor’s image and
commercial prospects. Sponsors are all too aware of the historic with-
drawal of Mercedes Benz in 1955 after a crash that killed 83 spectators,
and there has been an increased emphasis upon enhanced safety features
in motor racing in recent years due to untimely (and unexplained) deaths.
For example, following the death of Formula One driver Ayrton Senna in
1994, it was suggested that:

We are much more bothered about all this than we used to be, even as little
as quarter of a century ago. When Jim Clark, the British driving hero and two
times World Champion, was killed in a Formula Two race at Hockenheim in
1968, the event was front page news, to be sure. But if you look at the Daily
Telegraph of those times, you do not find the prolonged analysis of the crises
in motor-racing or leader page articles agonizing over the future of the sport,
or enthusiasts pleading for its continued autonomy. (the Sunday Telegraph, 8
May 1994, quoted in Twitchen, 2000 pp. 123–155)
Clark’s death was largely regarded as simply a legitimate consequence of com-
peting in a hazardous sport. The post-mortems were limited and, to a large
extent, shrugged aside. A generation later, we live in a very different environ-
ment where the freedom of the individual to make a personal choice as to how,
or indeed whether, he risks his life is very much hemmed in by wider social
constraints. (Henry, 1994, 6; quoted in Twitchen, 2000 pp. 123–155)

These pressures get worked out also through the Technical Panels that
regulate the various racing formulae. Within motor sport, the Technical
Panels achieve a consensus about the types of cars the various teams can
produce. Yet, importantly, the Technical Panels include the key designers
in the industry, so reinforcing their presence within the knowledge com-
munity (and the presence of Motor Sport Valley within the regulatory
framework of racing car production). This means that there is a very strong
co-operative, as well as fiercely competitive, element in racing car design
and within the Technical Panels. For example, as one engineer noted:

... if a team asks FIA [regulatory body in charge of panels] a technical ques-
tion, the FIA sends a copy of the question and answer to every team. (5)

Yet in competition terms this is significant also because you gain

some element of understanding of what other people are trying to do by the

questions that they ask. (1)

Ultimately:

I know some of the other designers and, you know, obviously we are all in
competition with each other, but you still might well do that [ring them up]
and say, ‘do you know a good way to. . .?’ or ‘do you know a source of this on
that information? (27)
162 INFERRING MANAGERIAL KNOWLEDGE

The Churning of Knowledge Within

the Valley Community (2)

On the one hand, then, the travelling of people is about the dissemination
of knowledge, which in turn generates innovation in one of the most
rapidly, incrementally, innovating industries in the world. On the other
hand, the travelling of ideas involves their translation into productive
knowledge in particular sites (Amin and Thrift, 1992). As one begins
to track shared understanding and meaning – the basis of productive
knowledge – so the processes of knowledge production and transfer start
to become apparent, although their representation is still problematic. In
tracking and representing such processes, so do their geographies become
apparent. In the productive world of motor sport the sites, the people and
the knowledge are first, and foremost, the knowledge community that is
Motor Sport Valley.
Furthermore, the message of the ‘seamless web’ is that to understand a
phenomenon such as Motor Sport Valley one needs to be aware of the
broader relations of political economy that enable this regionalized
knowledge network to flourish.6 Over many years, and due in large part
to the efforts of Bernie Ecclestone – acting as impresario, entrepreneur,
political lobbyist and deal maker – the British racing car companies have
worked collectively to promote their interests.7 They have thereby
ensured that their competitive position is not threatened by sudden
changes in the regulations surrounding motor racing that would under-
mine their technological trajectory or their economic base of corporate
sponsorship.8 As many have observed, the dissemination of knowledge is
intimately bound up with power relations. It involves the capacity of one
group to preclude others – or impose upon them a vision or ‘worldview’
(in this case the acceptable parameters of automotive racing).

CONCLUSIONS

The agglomeration of Motor Sport Valley as a form of spatially constituted

knowledge community provides one example of the spatial organization of
knowledge production in the contemporary economy. Clearly, many other
forms exist, such as the global intranets of the management consultants
(Bryson and Daniels, 1998), the array of technopoles of the informational age
(Castells and Hall, 1994) and the telecommunications networks of knowl-
edge transmission (Warf, 1995). Concerning Motor Sport Valley, we have
argued that its formation and reproduction as a knowledge community has
been key to understanding its world-wide economic dominance. As a ‘real
world of production’ (Storper and Salais, 1997), Motor Sport Valley has:

successful systems of conventions, that is, successful real worlds, ‘invent’ the
de facto best products and processes in the sectors in which they operate.
Hence, they play a strong role in describing the evolutionary path of those
 SPATIALIZING KNOWLEDGE 163

sectors tout court. They produce the supplies of certain objects that lead to the
‘social construction of markets’. The economic tests of those sectors are thus
themselves path-dependent and subject to strong influence by the generation
of products, technologies, and practices from specific conventional systems.
It is in this sense that actions, as embodied in conventions, produce the path-
way of development of the economy. (Storper and Salais, 1997: 299)

In this sense, Motor Sport Valley should be recognized as the current ‘pio-
neer’ district in what is a global industry (Amin and Robins, 1990),9 or an
example of one of Amin and Thrift’s (1992) neo-Marshallian nodes (Henry
and Pinch, 2001). Motor Sport Valley is the site of massive investment by
global companies (Ford, DaimlerChrysler, Honda, Mitsubishi, to name but
a few), but, in the main, into a network of small and medium-sized inde-
pendently owned British companies (Henry et al., 1996; Turner, 1998).
As a neo-Marshallian node or real world of production, Motor Sport Valley
is a centre of ‘representation, interaction and innovation in a global
filiere’ (Amin and Thrift, 1994, 1995; but see also Henry and Pinch, 1998) –
a world of convention setting, untraded interdependencies and a frame-
work of action.
In the context of the knowledge economy, we have argued in this chapter
that this centre or ‘world’ is first and foremost constructed through the
medium of the knowledge community. Clearly, the concept of ‘knowl-
edge’ used in this chapter, and in much work by economic geographers on
the knowledge economy, is a crude one, yet the aim has been to demon-
strate empirically the processes by which some versions of knowledge do
get spatialized, and, in this case, spatialized around very particular places.
In this regard, the research is a tentative, yet important, step towards the
demonstration of concepts of economic geography and agglomeration
such as ‘industry in the air’, ‘milieu’, ‘untraded interdependencies’, ‘local
industrial systems’, ‘learning regions’ and ‘institutional thickness’. These
are commonly recognized to be an important facet of agglomeration and
yet there are relatively few concrete examples of such concepts in action.
In summary, the empirical analysis undertaken in this chapter shows that
Motor Sport Valley is a close-knit knowledge community characterized by
a complex mixture of co-operation as well as fierce competition. This
‘world of production’ (Storper and Salais, 1997) includes ‘ways of think-
ing’ and ‘ways of doing’ that transcend individual firms but not the
British industry as a whole, and that contribute significantly to its world-
beating status. Thus, this chapter, we believe, is an important exemplifier
of how to understand the increased salience of agglomerations within the
globalizing economy. Furthermore, it opens up further intriguing possi-
bilities about understanding contemporary capitalism and its geographi-
cal growth dynamics.
For example, in the British motor sport industry, the paradox of highly
secretive companies located close to each other is explained by their vital
need to dip into the deep stream of knowledge that keeps being generated
and circulated throughout the region largely on an untraded basis.
164 INFERRING MANAGERIAL KNOWLEDGE

Recruitment of staff, observation of rival designs, new firm formation, the

assimilation of gossip and rumour and discussions with extensive per-
sonal contacts, are all facilitated by geographical proximity. In effect, the
region constitutes a ‘knowledge pool’ whose internal configuration is contin-
ually changing but that, overall, is on a constant learning trajectory. To
leave the region is to risk your position within the ‘knowledge loop’, and
it is clear that whilst many British engineers/designers sell their knowl-
edge overseas, very few leave the Valley for any great length of time.
Similarly, just as the software engineers of Silicon Valley discuss how they
can change firms without changing car parks (Saxenian, 1994), so in the
high firm-death industry of British motor sport, engineers know that a
location within the Valley is likely to ensure continued employment with-
out the need for a house move, even if the name of the employer is likely
to change relatively frequently.
But, as others have argued concerning other regional production
systems (Cooke and Morgan, 1993; Saxenian, 1994), this implies that the
competitive advantage of the British motor sport industry is integrally
bound up with its regional concentration, thus it could be stated that the
essential analytical focus should be on the region as the unit of analysis and
not the firm or industry per se (Storper and Walker, 1989). And in this case,
the knowledge community is the critical constituent of that region. If one
is to capitalize on knowledge, it must be recognized that the motor sport
industry is characterized by a high turnover of firms, yet the industry goes
from strength to strength. In the extreme case, the asset base of the firms
themselves is often minimal, involving, for example, some specialist
machinery, a (rented?) building and a car that becomes virtually worth-
less in months, and, critically, the knowledge of its employees. It is this
knowledge base, and the reputation that this base can command in the
eyes of sponsors, that is the key element of the industry. The industry is
therefore perhaps best represented as continually shifting networks of
people in, and outside of, continually shifting bounded entities called
firms. However, it is a knowledge base that is territorialized, and this is the
crucial competitive advantage of Motor Sport Valley. The question that
then arises is whether or not current measures of economic activity, and
economic performance, can accurately portray this emergent form of
(soft) capitalism. Or is there a need to rethink the objects of study in the
economy (e.g. firms, assets, profits, region) to reflect accurately its emer-
gent dynamics?
For example, if the significant unit of competitiveness is the network
and region, how do you invest in companies? Furthermore, if that com-
petitiveness is based on ‘signs’ as well as spaces, exactly what are the
assets of any firm/network/region? As the British government recently
recognized in its White Paper on competitiveness (DTI, 1998), if the bank-
ing system, for example, has always been ‘conservative’ in its investment
policies (i.e. regarding investment in UK small firms rather than ‘emerg-
ing markets’ from Latin America to the Pacific Rim and Eastern Europe),
 SPATIALIZING KNOWLEDGE 165

how will it now do its sums when faced with economic growth
characterized by a rapid churn of firms and serial entrepreneurs with both
firm failures (and successes) to their names?
Finally, a key issue for future studies of agglomerating industries is
the relationship between knowledge generation and power relations.
Although some of the most innovative racing car companies, and espe-
cially the early pioneers such as Coopers, survived on ‘shoestring’ budgets,
there is little doubt that in the contemporary world, competitive success in
most racing formulae is dependent upon securing substantial sums of
sponsorship money. This funding facilitates the creation of knowledge
necessary to produce cars that win races. This competitive success, in turn,
enables firms to gain political power to influence the various actors that
regulate motor sport: the FIA, various governments, the television compa-
nies that transmit the races around the world, and the various sponsors of
the sport. As indicated by this chapter (see also Pinch and Henry, 1999a),
knowledge is not therefore produced in a vacuum but is an integral part of
political economy/society. Industrial agglomerations cannot therefore be
understood without appreciating the wider political economy within
which they operate. Similarly, relations of power and domination enable
certain knowledge communities to construct the technological and pro-
duction trajectories of particular economic worlds.

NOTES

This is a revised version of a paper published in Geoforum, 31 (2), N. Henry and S. Pinch,
‘Spatialising knowledge: Placing the knowledge community of Motor Sport Valley’, 2000,
published with permission from Elsevier Science.
1. The maps in Figures 6.1 and 6.2 represent a snapshot of Motor Sport Valley at a particu-
lar point in time – late 1998/early 1999. As discussed later in the chapter, these maps are
likely to become outdated in a very short period of time due to the high rate of firm births
and deaths endemic to the industry.
2. This is, indeed, a crude distinction as in order for any ideas or knowledge to travel it must
do so through some medium or other (i.e. be embodied in some human/non-human artefact).
3. The ability to achieve this came from the vast array of specialist literature within the
industry. In addition, a number of these career histories were of individuals interviewed. In
these cases, pre-constructed histories were used as part of the interview.
4. The numbers refer to the transcript code. All respondents were assured of anonymity
to enable them to speak freely.
5. A constructor is the company that assembles the car using in-house and external sup-
plies. A constructor is, normally, the racing team (e.g. Williams Grand Prix Engineering runs
the Rothmans Williams Renault Formula One team) but constructors may also run more
than one team (e.g. Prodrive has run a Subaru team and a Nissan team in the World Rally
Championship).
6. As another referee commented, the ‘cultural-social’ perspective of the chapter initiates
a certain exclusion of the ‘economic’ in the sense of, for example, the cohesiveness, competi-
tiveness and employment characteristics of this regional production system. The point is
well taken (although see Henry, 1999; Henry et al., 1996; Henry and Pinch, forthcoming)
and pleasing in relation to the ultimate aim of the chapter as space is limited to develop fully
both strands.
166 INFERRING MANAGERIAL KNOWLEDGE

7. We should, nevertheless, bear in mind that this is a fiercely competitive industry, and in
the words of Henry ‘solidarity between F1 teams is largely built on quicksand’ (1990: p. 34).
8. See, for example, the recent, high-profile, ability of the ‘industry’ to gain exemption
from the European ban on tobacco sponsorship of sport for eight years by persuading the
UK government to oppose full implementation.
9. As an early paper in the ‘geography of post-Fordism/FS (flexible specialization) debate’,
Amin and Robins (1990) argued that many of the empirical canons of the industrial districts
did not provide reproducible paths partly because they included ‘pioneer’ regions who had,
in effect, ‘invented’ new industries and there can only be one pioneer. Interestingly, Motor
Sport Valley wrenched ‘pioneer’ status in motor sport from the Third Italy.

REFERENCES

Allen, J. and du Gay, P. (1994) ‘Industry and the rest: The economic identity of services’,
Work, Employment and Society, 8 (2): 255–271.
Allen, J., Massey, D., Cochrane, A. with Charlesworth, J., Court, G., Henry, N. and Sarre, P.
(1998) Rethinking the Region. London: Routledge.
Amin, A. and Robins, K. (1990) ‘The re-emergence of regional economies? The mythical geog-
raphy of flexible accumulation’. Environment and Planning D: Society and Space, 8: 7–34.
Amin, A. and Thrift, N. (1992) ‘Neo-Marshallian nodes in global networks’, International
Journal of Urban and Regional Research, 16 (4): 571–587.
Amin, A. and Thrift, N. (1994) ‘Living in the global’, in A. Amin and N. Thrift (eds),
Globalization, Institutions and Regional Development in Europe. Oxford: Oxford University
Press, pp. 1–19.
Amin, A. and Thrift, N. (1995) ‘Globalisation, institutional “thickness” and the local econ-
omy’, in P. Healey, S. Cameron, S. Davoudi, S. Graham and A. Madani-Pour (eds),
Managing Cities: The New Urban Context. Chichester: Wiley.
Barnes, T. (1996) Logics of Dislocation: Models, Metaphors and Meanings of Economic Space.
New York, Guilford Press.
Bijker, W.E., Hughes, T.P. and Pinch, T. (1990) The Social Construction of Technological Systems.
Cambridge, MA: MIT Press.
Boden, D. (1994). The Business of Talk. Cambridge: Polity.
Bryson, J.R. and Daniels, P.W. (1998) ‘Business link strong ties, and the walls of silence: Small
and medium-sized enterprises and external business-service expertise’, Environment and
Planning C: Government and Policy, 16: 265–280.
Bryson, J., Henry, N., Keeble, D. and Martin, R. (1999) The Economic Geography Reader:
Producing and Consuming Global Capitalism. Chichester: Wiley.
Bryson, J., Daniels, P., Henry, N. and Pollard, J. (2000) Knowledge, Space, Economy. London:
Routledge.
Castells, M. (1996) The Rise of the Network Society. Oxford: Blackwell.
Castells, M. and Hall, P. (1994) Technopoles of the World: The Making of 21st Century Industrial
Complexes. London: Routledge.
Clark, P. (1999) Organizations in Action: Competition Between Context. London: Routledge.
Cooke, P. and Morgan, K. (1993) ‘The network paradigm: New departures in regional develop-
ment’, Environment and Planning D: Society and Space, 11: 543–564.
Cooke, P. and Morgan, K. (1998) The Associational Economy. Oxford: Blackwell.
Coyle, D. (1997) The Weightless World: Strategies for Managing the Digital Economy. Oxford:
Capstone.
Cresswell, J.W. (1994) Research Design: Quantitative and Qualitative Approaches. London: Sage.
Department of Trade and Industry (DTI) (1998) Our Competitive Future: Building the
Knowledge Driven Economy. London: HMSO.
Dicken, P. (1998) Global Shift: Transforming the World Economy. 3rd edn. London: Paul
Chapman.
 SPATIALIZING KNOWLEDGE 167

du Gay, P. (1996) Consumption and Identity at Work. London: Sage.

Economic and Social Research Council (ESRC) (2000) ‘The evolution of knowledge: Interaction
of research and practice’, <http://www.esrc.ac.uk/knowconc.htm>. Accessed 29 March.
Economist (1999) ‘Knowing your place’, 11 March.
Florida, R. (1995) ‘Toward the learning region’, Futures, 27, 527–536.
Gertler, M. (1994) ‘“Being there’’: Proximity, organisation and culture in the development
and adoption of advanced manufacturing technologies’, Economic Geography, 71 (1): 1–25.
Gibson-Graham, J.K. (1996) The End of Capitalism (As We Knew It): A Feminist Critique of
Political Economy. Oxford: Blackwell.
Goodwin, M., Duncan, S. and Halford, S. (1993) ‘Regulation theory, the local state, and the
transition of urban politics’, Environment and Planning D: Society and Space, 11: 67–88.
Grabher, G. (1993) The Embedded Firm: On the Socioeconomics of Industrial Networks. London:
Routledge.
Granovetter, M. and Swedberg, R. (eds) (1992) The Sociology of Economic Life. Boulder, CO:
Westview Press.
Hamilton, M. (1994) Race Without End: The Grind behind the Glamour of the Sasol Jordan Grand
Prix Team. Sparkford: Patrick Stephens.
Harvey, D. (1989) The Condition of Postmodernity. Oxford: Blackwell.
Hayter, R. and Watts, H.D. (1983) ‘The geography of enterprise: A re-appraisal’, Progress in
Human Geography, 7: 157–181.
Henry, A. (1990) The Turbo Years: Grand Prix Racing’s Battle for Power. Swindon: Crowood
Press.
Henry, A. (1994) The Darkest Hour. Autocourse, Vol. 44. London: Hazelton Publishing.
Henry, N. (1999) In Pole Position: Motor Sport Success in Britain and Its Lessons for the World
Motor Industry. London, Euromotor Reports.
Henry, N. and Pinch, S. (1997) ‘A Regional Formula for Success? The innovative region of
Motor Sport Valley’, School of Geography, University of Birmingham.
Henry, N. and Pinch, S. (1998) ‘“Motor Sport Valley” and institutional thickness: Thick or
thin?’, paper presented at the Reflections on the Institutional Turn in Local Economic
Development Conference, University of Sheffield, 9–10 September.
Henry, N. and Pinch, S. (forthcoming) ‘Neo-Marshallian nodes, institutional thickness and
Britain’s Motor Sport Valley: Thick or thin’? Environment and Planning, 33, 1169–1183.
Henry, N., Pinch, S. and Russell, S. (1996) ‘“In pole position?” Untraded interdependencies,
new industrial spaces and the British motor sport industry’, Area, 28 (1): 25–36.
Hodgson, G. (1999) Economics and Utopia: Why the Learning Economy is Not the End of History.
London: Routledge.
Jessop, B. (1990) State Theory: Putting Capitalist States in Their Place. Cambridge: Polity.
Jones, M. (1998) New Institutional Spaces: TECS and the Remaking of Economic Governance:
London, Jessica Kingsley.
Krugman, P. (1991) Geography and Trade. Leuven: Leuven University Press.
Krugman, P. (1995) Development, Geography and Economic Theory. Cambridge, MA: MIT Press.
Lash, S. and Urry, J. (1987) The End of Organized Capitalism. Cambridge: Polity.
Lash, S. and Urry, J. (1994) Economies of Signs and Space. London: Sage.
Layder, D. (1998) Sociological Practice: Linking Theory and Social Research. London: Sage.
Leadbetter, C. (1999) Living on Thin Air: The New Economy. London: Viking.
Lee, R. and Wills, J. (eds) (1997) Geographies of Economies. London: Arnold.
Lipietz, A. (1992) Towards a New Economic Order: Post-Fordism, Ecology and Democracy.
Cambridge: Polity.
Livingstone, D. (1992) The Geographical Tradition. Oxford: Blackwell.
Lundvall, B.-A. and Johnson, B. (1994) ‘The learning economy’, Journal of Industry Studies,
1: 23–42.
McDowell, L. (1997) Capital Culture: Gender at Work in the City. Oxford: Blackwell.
McDowell, L. and Court, G. (1993) ‘The missing subject in economic geography’, South East
Programme Occasional Paper Series No. 5. Faculty of Social Sciences, Milton Keynes: Open
University.
168 INFERRING MANAGERIAL KNOWLEDGE

MaCleod, G. (2000) ‘Re-inventing a region in the age of austerity: Placing knowledge,

entrepreneurialism, and reflexive capitalism in Lowland Scotland’, Geoforum, 31: 29–236.
Markusen, A. (1996) ‘Sticky places in slippery slopes: A typology of industrial districts’,
Economic Geography, 72: 293–313.
Martin, R. (1994) ‘Economic theory and human geography’, in D. Gregory, R. Martin and
G. Smith (eds), Human Geography: Society, Space and Social Science. Basingstoke: Macmillan.
pp. 21–53.
Martin, R. (1999) ‘The new “geographical” turn in economics’, Cambridge Journal of
Economics, 23: 65–92.
Massey, D. (1993) ‘Power-geometry and a progressive sense of place’, in J. Bird, B. Curtis,
T. Putnam, G. Robertson and L. Tickner (eds), Mapping the Futures: Local Cultures, Global
Challenge. London: Routledge. pp. 59–69.
Massey, D., Allen, J. and Sarre, P. (eds) (1999) Human Geography Today. Cambridge: Polity.
Morgan, K. (1997) ‘The learning region: Institutions innovation and regional renewal’,
Regional Studies, 31: 491–503.
Peck, J. (1996) Work-Place: The Social Regulation of Labor Markets. New York: Guilford Press.
Peck, J. and Tickell, A. (1994) ‘Searching for a new institutional fix: The after-Fordist crisis
and the global–local disorder’, in A. Amin (ed.), Post-Fordism: A Reader. Oxford: Blackwell.
pp. 280–315.
Pinch, S. and Henry, N. (1999a) ‘Discursive aspects of technological innovation: The case of
the British motor sport industry’, Environment and Planning A, 31: 665–682.
Pinch, S. and Henry, N. (1999b) ‘Paul Krugman’s geographical economics, industrial clus-
tering and the British motor sport industry’, Regional Studies, 33 (9): 815–827.
Pinch, S., Henry, N. and Turner, D. (1997) ‘Explaining the supremacy of Motor Sport Valley’,
School of Geography, University of Birmingham, Working Paper Series 62.
Piore, M. and Sabel, C. (1984) The Second Industrial Divide. New York: Basic Books.
Polanyi, M. (1967) The Tacit Dimension. London: Routledge and Kegan Paul.
Porter, M. (1990) The Competitive Advantage of Nations. London: Macmillan.
Pratt, A. (1998) ‘Making digital spaces: A constructivist critique of the network society’,
mimeo, available from the author, London: Department of Geography, London School of
Economics.
Pyke, F., Becattini, G. and Sengenberger, W. (eds) (1990) Industrial Districts and Inter-Firm
Cooperation in Italy. Geneva: International Institute for Labour Studies.
Quah, D.T. (1996) ‘Growth and dematerialisation: Why non-stick frying pans have lost their
edge’, Centre Piece. London: CEP, London School of Economics.
Sawers, L. and Tabb, W. (1984) Sunbelt/Snowbelt: Urban Development and Regional
Restructuring. Oxford: Oxford University Press.
Saxenian, A. (1994) Regional Advantage: Culture and Competition in Silicon Valley and Route 128.
Cambridge, MA: Harvard University Press.
Scott, A.J. (1988) New Industrial Spaces. London: Pion.
Scott, A.J. and Storper, M. (1987) ‘High technology industry and regional development:
A theoretical critique and reconstruction’, International Social Science Journal, 112: 215–232.
Segal, Quince and Wicksteed (SQW) (1985) The Cambridge Phenomenon: The Growth of High
Technology Industry in a University Town. Cambridge: SQW.
Storper, M. (1995) ‘The resurgence of regional economies ten years later: The region as a nexus
of untraded interdependencies’, Journal of European Urban and Regional Studies, 2: 191–221.
Storper, M. (1997) The Regional World: Territorial Development in a Global Economy. New York:
Guilford Press.
Storper, M. and Salais, R. (1997) Worlds of Production: The Action Frameworks of the Economy.
Cambridge, MA: Harvard University Press.
Storper, M. and Walker, R. (1989) The Capitalist Imperative: Territory, Technology and Industrial
Growth. Oxford: Blackwell.
Thrift, N. (1994) ‘On the social and cultural determinants of international financial centres:
The case of the City of London’, in: S. Corbridge, R. Martin and N. Thrift (eds), Money,
Space and Power. Oxford: Blackwell. pp. 327–355.
 SPATIALIZING KNOWLEDGE 169

Thrift, N. (1996a) ‘Knowledge economies: The rise of Soft Capitalism’, paper presented to
the Institute of British Geographers Annual Conference, University of Stratchclyde,
Glasgow, 3–6 January.
Thrift, N. (1996b) Spatial Formations. London: Sage.
Turner, D. (1998) ‘A study of the relationships between the small firms of Motorsport Valley
and multi-nationals’, unpublished MPhil., School of Geography, University of
Birmingham.
Twitchen, A. (2000) ‘The body’s second skin: Forming the protective community of grand
prix motor racing’, in L. McKie and N. Watson (eds), Organising Bodies. New York:
Macmillan.
Warf, B. (1995) ‘Telecommunications and the changing geographies of knowledge transmis-
sion in the late 20th century’, Urban Studies, 32: 529–555.
Wright, P. (1996) ‘The designers – John Barnard’, Racecar Engineering, 5 (6): 41–2.
THEORETIC
AND III
METHODOLOGICAL
ISSUES
7
A MAPPING FRAMEWORK
F O R S T R AT E G Y M A K I N G

Colin Eden and Fran Ackermann

ABSTRACT

When embarking upon a strategy development process, organizations are often

faced with the question of where to start, whom to involve and how to manage
the contributions that are surfaced. This chapter provides details of an approach
that allows organizations to reflect upon what they have been doing hitherto in
order to understand the nature of the organization, its capabilities and aspira-
tions. Our approach helps them explore this understanding, and negotiate an
agreement towards a strategic direction. It is, in short, a method for surfacing
and sharing current knowledge, and collectively creating new knowledge.
The approach presented is essentially a participative one (rather than being the
sole domain of a team of planners), although the extensiveness of participation
depends on the culture of the organization, the time available for the exercise,
and the resources accessible. It is part of the JOURNEY-making approach to strat-
egy (see Eden and Ackermann, 1998a, for a more detailed explanation), focus-
ing predominantly on the use of cognitive and cause mapping to elicit and work
with strategic possibilities.

SURFACING AND EXPLORING CURRENT

STRATEGIZING MODELS

Detecting ‘emergent strategizing’ (Eden and van der Heijden, 1995;

Mintzberg and Waters, 1985) is an appropriate starting point for strategy
making. Emergent strategizing is what will determine the strategic future
of the organization when no deliberate choices are made about that
future. What this process entails is asking the group members (whether
they be the management team or Board of Directors, or some other group)
to reflect on unwritten rules of thumb, taken-for-granted ways of thinking
174 THEORETIC AND METHODOLOGICAL ISSUES

about the strategic issues facing the organization. These are the structures
and procedures that will shape the strategic future of the organization
should they persist in current activities. Through this part of the
JOURNEY-making approach, current concerns as well as implicit or
embedded aspirations, knowledge, wisdom and beliefs are surfaced and
participants provided with the means to reflect on and debate them. We
want to surface that which currently drives the organization’s direction (the
‘theories in use’) rather than any espoused and/or published strategy.
The group may, as a consequence of this consideration, agree to let the
organization continue ‘muddling through’ (Lindblom, 1959), resulting in
deliberate agreement to this approach as a strategy. Alternatively, they may
wish to develop a different strategic future and design the strategic intent
to guide the change. Whether or not they are self-conscious of the process,
organizations develop and reinforce emergent strategy – as illustrated
through the ‘recipes’ or patterns they adopt (Spender, 1989). Through
detecting the underlying recipe(s) and reflecting upon it/them, any stra-
tegic management undertaken by the group can be thought of as ‘deliberate
emergent strategizing’, as opposed to non-deliberate emergent strategizing.
This conscious choice process provides an understanding of the organiza-
tion’s history, capabilities and aspirations, and is a key element to the
JOURNEY-making approach. Cognitive and cause mapping – the structur-
ing technique adopted by the process – provides some assistance both in
the elicitation and subsequently in the structuring of the contributions sur-
faced. The resultant model of strategic thinking then facilitates negotiation
(strategic conversations) to take place and change becomes more possible
because it is politically feasible. We can see our future clearly and wisely
only when we know the path that leads to it from the present.
The most powerful conversation that can drive emergent strategic
change is that focused on crisis, challenge and ‘firefighting’ (Mintzberg
et al., 1976). Managers attribute cause and effect relationships in their
attempts to make sense (Kelly, 1955) of such crises. Their conversations are
usually the starting points for ‘issue selling’ (Dutton and Ashford, 1993),
where each manager seeks to influence his or her colleagues’ view of what
is important. In crisis, managers fight harder for their views and the drive
for action means they reveal – often implicitly – more of their embedded
vision, surface more of their assumptions, and take one another more
seriously than on ‘special’ occasions designated for strategy making (e.g.
‘away-days’). However, the urgency involved in dealing with strategic
crises can mean that whilst each manager surfaces more of his or her
strategic assumptions, there is less time for other managers to check out
their understanding of these assumptions. Consequently assumptions can
be surfaced and missed, or can be severely misunderstood. These mis-
understandings occur because the important stage of sharing meaning and
making sense may have been bypassed, and each participant is forced to
make untested assumptions about the implicit assumptions embedded in
proposals for action coming from colleagues.
 A MAPPING FRAMEWORK FOR STRATEGY MAKING 175

Thus a more purposeful strategy development approach must recognize

the need for a strategy model to be transparent and understandable to
participants (so as to enable them to reflect on one another’s assumptions)
as well as being appropriate to the task of strategy making. It must capture
the subtleties and nuances of meaning, allowing new material to surface
and comprehension to build.
The modelling technique – our mapping approach – is a particular form
of cognitive mapping (Eden, 1988; Eden and Ackermann, 1998b; Eden
et al., 1979) and its developed extensions are cause or strategy mapping.
A cognitive map is intended to be a model/representation of a particular
person’s thinking about the strategic issues facing the organization. The
amalgamation of a series of cognitive maps (from a group of individuals)
into a computer-based model provides the source for the creation of a
strategy map as a cause map. This strategy map or model contains the
aggregation of the thinking of many people, including conflicting views,
subtly different slants on the same issues, and different perspectives on
similar views.
The cause- or strategy-mapping model therefore provides a useful
device/means for group work through providing a holistic representation.
It supports relevant group processes, including the importance of negotia-
tion and the significance of anonymity (each of these will be addressed
below). The model, which is typically a set of embedded maps, provides
the means for strategic conversations flushing out the values, assumptions
and beliefs of managers (and thus the emergent strategizing), and enables
reflection/examination and development of strategic direction. The maps
may be created directly by groups (using the ‘oval mapping technique’
[OMT], or using networked laptop computers and special purpose group
support software – ‘Group ExplorerTM’ (www.Phrontis.com). They may
also be created during and following interviews with individuals, where
the output from the interview is a cognitive map.
Cognitive mapping is used when working with individuals and helps in
the elicitation and structuring of each of the participant’s distinctive pers-
pectives – how he or she perceives the issue(s) being discussed. The maps
strike a balance between (a) providing a device for exploring meaning –
through illustrating not just what the facts/issues are but also how they fit
together – and (b) ensuring, through their qualitative fuzziness and impre-
cision, a sufficient degree of equivocality to facilitate negotiation. This duality
is particularly important when subsequently working in groups.
Cognitive maps are typically used for capturing and exploring the
thinking of each member of a senior management team, allowing tacit
knowledge, assumptions, assertions, aspirations, values, beliefs and con-
cerns to surface. They are idiographic in nature (Eden and Ackermann,
1998b), reflecting the particular thinking of an individual. However, these
individual maps are then woven together to form a group representation,
and, following a process of analysis, used to form a group map or model
(a ‘negotiative device’) that can subsequently support the process of
176 THEORETIC AND METHODOLOGICAL ISSUES

developing agreement (the strategic conversation). This process is usually

undertaken using software that mirrors the mapping technique (‘Decision
Explorer’TM [www.Banxia.com]), providing the facility to manage infor-
mation more effectively through rapid retrieval processes, analytical pro-
cedures and flexible displays.
So what actually is a cognitive map and how is it constructed and used
during an interview with a senior manager? A cognitive map is essentially
a directed graph or causal (option/outcome) network with aspirations/
goals, values and beliefs at the apex, issues/concerns in the centre and
detailed actions or assertions supporting these issues (or concerns) at the
base. (Figure 7.1 provides an example of a small part of a cognitive map.)
In a cognitive map (and later in the strategy map developed by the
group), explanations as implicit actions are those statements that are
taken to cause a given outcome. However, each explanation or action in
turn is informed by other possible actions that support them (explana-
tions), placing the former action as an outcome. Therefore, each node on the
map is both an explanation and an option for action, and an outcome
depending upon the level of abstraction required. As part of the mapping
process, the maps are deliberately coded with an active verb so that each
of the statements/constructs constitutes a call for action as if it were to
be taken as an option. The coding process also aims to ensure that the
language of the interviewee is captured. Each statement includes an actor
as well as action, and, where volunteered, the statement reflects the oppo-
site, contrasting or bipolar element (using the ‘rather than’ clause).
The ‘call for action’ also provides an example of statements (‘personal
constructs’, in Kelly’s term1) as contributors to the process of ‘manage and
control’. Cognitive maps help by prompting managers to focus on con-
trasting the current with the past, the current with the aspired future direc-
tion, or the past with the future. By focusing on strategic issues rather than
the development of idealized scenarios or preferred goals, the process is
deliberately designed to get as close as possible to ‘theories-in-use’ (Argyris
and Schon, 1974; Bartunek and Moch, 1987). Doing so helps reduce the
possibility of participants discussing ‘espoused theories’ derived from
attendance of management courses or previous strategy documents! This
process helps avoid strategy being constructed as ‘motherhood and apple
pie’. In this chapter we outline several steps that move from individual
cognitive maps to the construction of a group ‘strategy map’.

METHODOLOGICAL NOTE: THREE WAYS OF

MAPPING WITH GROUPS

Alongside the use of individual interviews, group mapping allows

members to have a forum for discussing and reflecting upon the organi-
zation and its direction (particularly useful where consultation is not a
familiar process, or there are no interpersonal issues). Cognitive maps
can be constructed from the following two methods for working with
 18 Reducing 16 Asset base of 13 Regular customers are
turnover customers for the getting lost as they 171 Competitors
future is deteriorating are introduced to others becoming stronger

112 Customers are no

19 Costs of manufacture longer turning to us
1 Aggressive price
increasing ... as a 1st call
cutting by some
staying at the same rate
competitors
11 Losing breadth of
96 Losing control product coverage
of product quality 79 Inexperience of
15 Not getting
doing business in
enough new product
China 107 missing too many 20 New entries don’t
ideas coming through
new business understand their
opportunities costs
9 Motivation of 83 Difficulty in 122 New entries
staff dropping retaining good staff – prepared to make
8 Overheads rising 72 New entries using
particularly the losses to gain market
faster than turnover a different way of
trouble shooters entry
accessing the market

5 Too much staff 25 Vision not 99 Head-hunters 24 IT expanding out

turnover among projected well enough concentrating of control
middle managers by senior team efforts on our best 21 E- business entries from
people those who don’t yet
23 IT part of understand the business model
business is not well
4 Danger of understood by senior mgt
acquisition by ABC plc

Figure 7.1 A sample of a cognitive map

Note: For clarification, the numbers at the front of the ‘concepts’ are simply numeric tags provided by the software to aid manipulation of the data.
Where there are three dots (an ellipsis), this is shorthand for ‘rather than’. Finally, the dotted arrows highlight the fact that there is more material
around the concept in question.
178 THEORETIC AND METHODOLOGICAL ISSUES

mapping in groups. The first, the Oval Mapping Technique (OMT),

provides a manual forum for eliciting contributions and structuring them
according to the mapping conventions. The second, using computer-
supported mapping, allows the group to interact with the map visually
and literally. In both cases the techniques support the strategy-making
process through aiding conversation and negotiation. However, there are
a number of issues that must be addressed in designing such a process:
for example, difficulties keeping participants from contributing; the signi-
ficance of synthesis; enabling enough creativity and fun; and developing
strategy maps in real time. We have also used a third, more anonymous
method with networked computers, which has its own strengths and
weaknesses.

Using the Oval Mapping Technique (OMT)

One approach to producing group maps encourages group members to

record and publicly display their views on large ‘oval’ cards (Bryson et al.,
1995; Eden and Ackermann, 1998a). Unlike the production of cognitive
maps in a more private environment (with only the interviewer/mapper
and one organizational participant), members in a group are able to see
the perspectives of the others, add their own contributions, and watch the
strategy map unfold. During the process the ovals will be organized by
a facilitator following the same guidelines used for building individual
cognitive maps – thus concepts will be identified, clusters formed, and
inter- and intra-cluster links elicited. Through the process of both sur-
facing and more significantly structuring the material, participants are
able to move towards a model that has the ownership of most of the
group. This is achieved as they begin to understand better the different
points of view by considering them in their context, frequently adding
new, subtler, points that can be woven into the overall picture. Through
carrying out rough public analyses, identifying clusters of concepts
within the context of a structured hierarchy, participants are given the
chance to identify and explore the emergent properties of the map.
This public procedure enables group members to begin to change their
mind as they examine the rationale of alternative viewpoints. They are not
required to defend their own viewpoint. It is expected that their own
ideas might change as they are seen within the context of the views of
others in the group. Moreover, the mapping process promotes synthesis
by enabling linkages between and within clusters of inputs. Group maps
also encourage creativity through the ability to see alternative points of
view, and from this position identify and develop new options and strate-
gies. By encouraging multiple perspectives, providing an environment
with enough equivocality to avoid the need for ‘face saving’, and by struc-
turing the knowledge and insights, the process attends to procedural
rationality. It makes sense – which in turn promotes emotional and
 A MAPPING FRAMEWORK FOR STRATEGY MAKING 179

Figure 7.2 A strategy group beginning the OMT process

cognitive commitment to the outcomes. The process is designed to

encourage social negotiation.
To illustrate, Figure 7.2 shows a picture of a small group working on
their strategy. The group, who are responsible for providing public trans-
port and information pertaining to its schedule/cost, are in the process of
surfacing material on oval cards and beginning to structure it. As they
teased out the differences between the various forms of communication
they were providing, they also began to negotiate a set of actions to
improve and facilitate communication. The group were also responsible
for liaising with other transport operators – with the intention of provid-
ing an integrated public transport network. Therefore not only did they
have to communicate with their own operators but they also provide the
public with information on services, including connection possibilities,
and so on. Both of these were seen as different requirements from the
communication that had to take place within the organization. The debate
raised resulted in the organizations goal system being revised.

Using Computer-Supported Mapping

A second way of using mapping with groups is to work with an electronic

version of the strategic map – a computer model (Ackermann and Eden,
2001). Although often a starting model has already been generated either
180 THEORETIC AND METHODOLOGICAL ISSUES

through the interviews or through the Oval Mapping Technique, it can

also start from scratch. Using the software, either the facilitator enters in
participants’ comments as they are made or alternatively a group decision
support system is used enabling participants to enter their ideas (and
links) directly into the model – in a manner similar to writing on the ovals
(Ackermann and Eden, 1997). Using the computerized form of mapping
(rather than the ovals) allows easy manipulation of the model (e.g. edit-
ing, deleting and adding) and encourages participants to play with ideas,
explore their context and begin to build up a common understanding.
Each of these activities is a precursor to negotiation.
Computerization also supports fast retrieval of material from other
parts of the model, should they be seen as relevant to the topic being dis-
cussed, and enables rapid analysis to be undertaken, allowing exploration
of the model’s structural properties. However, computer maps also con-
strain the amount of material seen at any time to the size of a computer
screen – much less material than can be viewed on an OMT wall.
As can be seen from the photo presented in Figure 7.3, the computer-
projected model is also attractive because it enables participants to work
with the material at their own pace, search the screen for their contribu-
tions and, subsequently, read the statements of others. From this process
of allowing members to locate their own views in addition to seeing the
perspectives of others in context – because the map depicts the surround-
ing information – a degree of equivocality is possible from the multipli-
city of contextual data. Participants are not forced to make immediate
responses and so the process of psychological negotiation can proceed
without undue influence of emotion. The screen acts as a buffer between
a proponent and other contributors, allowing easier negotiation.
In one company this process went a long way to resolving a difficult
emotionally charged situation where two coalitions of staff were in dis-
pute about the future. The model’s ability to reassure members that their
views were captured and taken seriously, along with the opportunity to
provide further explanation, was critical to moving beyond deadlock. The
strategic problem was that whilst marketing was a central service, the
company was understaffed and not providing sufficient service to some
of the smaller products seen as significant for the future. Meanwhile
because the smaller products still generated overheads, their managers
planned to recruit their own marketing staff. The workshop commenced
with a clear division between roles, but after both sides had began to
appreciate one another’s perspectives/difficulties, then it was possible to
move to a more collaborative stance.
Whilst each form of mapping has its own specific purpose, both oval
mapping and computer modelling (Ackermann and Eden, 2001) are
grounded in the same theoretical standpoint – Personal Construct Theory
(Kelly, 1955). Regardless of whether they are individual- or group-
oriented, the maps provide models that are designed to reflect the think-
ing, beliefs and aspirations of those whose views have been incorporated.
 A MAPPING FRAMEWORK FOR STRATEGY MAKING 181

Figure 7.3 An example of a group using the computer-supported system

Consequently, they provide a means for enabling group members to begin

jointly to understand the perspectives of others, reflect on the emergent
issues that are surfaced from them, and start to negotiate an agreed strate-
gic direction. Moreover, they promote synthesis through the mapping
technique’s ability to pull together viewpoints. Both mapping techniques
promote creativity that ensues from synthesis and from multiple perspec-
tives. The model itself becomes a socially negotiated reality – ‘a transi-
tional object’ (de Geus, 1988) – with new meanings being discovered, a
common language being developed and an overall understanding and
agreement being forged.
Most teams find the process fun as they begin to learn more about their
organization and see progress being made. For one group within Shell –
who were struggling with developing their own strategic direction – the
use of the model to surface, develop and negotiate agreement helped
them achieve something that they had not previously been able to do
(Galer, 1990). Through the mapping process participants often gain a
sense of emotional commitment to the outcome as they find the process
both procedurally just (allowing all to have their say) and procedurally
rational. In addition, we have found on numerous occasions the act of
working together and developing a map and strategic direction helps in
the process of building teams (Ackermann, 1992).
The participative aspect of the mapping methods we will describe is
critical. First, the process of including staff from around the organization
ensures some degree of representation, allowing staff to feel part of the
182 THEORETIC AND METHODOLOGICAL ISSUES

process through influencing or informing the strategic direction taken.

From this they are more likely to begin to own the outcomes and under-
stand the final product, thus increasing the chance of effective implementa-
tion. Second, by developing a map in a group and thus exploring different
perspectives, it becomes possible to persuade the key players to also buy
into the outcomes, thus ensuring that the outcomes are politically feasible.

Working in Groups using a Networked System of

Laptop Computers for the Direct Elicitation of Maps

Not all organizations are comfortable or familiar with participative methods.

When working in a part of a Health Authority, it was necessary to start
the mapping process using interviews as staff were unused to being asked
their viewpoint and apparently unwilling to contribute in a group setting
because they were wary of exposing themselves. Finding different ways
of eliciting material in a group setting, but also ensuring anonymity,
became important. The group support system ‘Group Explorer’ can pro-
vide anonymity (through the direct entry of contributions via a personal
laptop computer that is networked to the public screen). Networked com-
puters also allow productivity gains: a group can surface, structure and
negotiate towards an agreed direction within half a day, should that be
necessary. However, unlike the interviews, starting with a networked
system does not promote the same level of deep and reflective thinking at
a personal level.

BUILDING THE INITIAL FRAMEWORK FOR

STRATEGIC ACTION WITH THE
MANAGEMENT TEAM

Step 1: Individual Interviews

All three of the group processes just described here may begin with initial
individual interviews. These help group members surface contributions that
might otherwise not emerge in a group setting, and also help in furthering
reflection and deeper thinking. For example, one organization, unused to
participative approaches, used interviews to encourage members to raise
their issues, concerns and aspirations because the Chief Executive strongly
believed that working directly in a group environment would reduce the
knowledge and beliefs elicited and thus diminish the potential of the team
and process. The interviews were used to allow members to build up
confidence both in the mapping technique and to take more responsibility
for decisions and strategic decision making. In other organizations,
characterized by more open sharing, initial interviews are still useful
because they surface deeper wisdom and so allow for the identification of
 A MAPPING FRAMEWORK FOR STRATEGY MAKING 183

more of the distinctiveness of the organization during the group sessions

and also elicit a wider range of possible strategic options.

Step 2: Developing a Model of Strategic Intent from

Individual Interviews

As noted earlier, the mapping process commences with surfacing the

organization’s emergent strategizing, detecting the patterns of beliefs and
values within individual cognitive maps. This effort usually forms the basis
for the first part of JOint Understanding and Reflection – an important part
of the JOURney-making approach. Using the maps as a vehicle for debate,
a social process evolves in which team members explore one another’s
concerns, assumptions of the world and aspirations for the organization.
The facilitators structure this map as a ‘backroom’ activity. The compo-
site map takes a hierarchical form similar to the cognitive maps. The possi-
ble aspirations or goals are at the top of the model; these are supported by
(or derived from) emergent strategic issues, which in turn are impacted
upon by strategic problems and possible strategic options. This structure is
shown by Figure 7.4. The categorization of each of the statements that
make up the map into each of these levels is initially a rough one. Typically
a first pass will take place following individual interviews, whilst weaving
together different cognitive maps. This effort involves carrying out various
forms of analysis (explained more fully in Eden et al., 1992, and Eden and
Ackermann, 1998a) to tidy the model and subsequently begin to explore its
characteristics and thus gain clues for the categorization.
Tidying the model normally takes the form of exploring the ‘heads’ of
the model (statements with no consequences – out-arrows) to check
whether these statements, as outcomes, are likely to be taken by the group
as ‘good or bad in their own right’ and therefore as goals or ‘negative
goals’. They may alternatively be statements requiring further linking
into the main body of the model. ‘Orphans’ (statements currently uncon-
nected) are also identified and woven into the model if appropriate. The
weaving process is assisted by the ability of the mapping technique to
show not only the concept (statement) but also its context (i.e. the material
surrounding it, adding richness and helping determine meaning). Word
searches can be conducted to ensure full integration of the different per-
spectives. This single network/map then provides the basis of a vehicle
for strategy negotiation – both psychological and emotional.
Following the tidying process, further analyses are conducted. These
include:

• examining the density around the statements (looking for the ‘busiest’);
• calculating each statement’s centrality within the overall structure of
the model; and
• searching for feedback loops (showing possible dynamic behaviour).
 ‘the aggregation of cognitive maps’ the structure of the
or group map final strategy

Aspiration System
EMERGENT
GOALS
ASPIRATIONS/GOALS

supports

STRATEGIC STRATEGIES
ISSUES
constructs or
statements supports

STRATEGIC
STRATEGIC PROGRAMMES
PROBLEMS

supports

STRATEGIC
OPTIONS ACTIONS
‘tear-drops’ of portfolios
explanatory statements

STEP 2 Developing the ‘Strategy Map’ STEP 4

Developing a model of strategic Mapping strategic
intent from individual interviews intent – the Strategy Map

Figure 7.4 The hierarchy of maps

 A MAPPING FRAMEWORK FOR STRATEGY MAKING 185

Various forms of clustering are also carried out, slicing the model in
different ways and enabling emergent patterns to be detected. For example,
hierarchical sets – one of the clustering analyses available in the software –
provides valuable assistance during both the ‘backroom’ work and the
subsequent group working. The analysis presents the information as a
‘tear-drop’ of argument and explanation supporting the statement that acts
as the label at its apex – usually a key strategic issue. By cycling around
these analyses the modeller is able to gain insights into the structure of the
model and develop some sense of confidence in the contents of the various
categories. A similar process is undertaken following the generation of a
group cause map using either the Oval Mapping Technique or computeri-
zed modelling processes, discussed below.
The model of strategy such as the one shown in Figure 7.4 and its asso-
ciated categories/insights will be further refined by the management
team as they work with the model/map. The composite map formed by
aggregating individual cognitive maps becomes a ‘transitional object’
(de Geus, 1988) facilitating conversation. Participants confirm or change
the judgments made by the modellers, adding new material and editing
existing material. Thus, the group map will either contain the core belief
statements that are supported/explained by other statements (options
etc.) or provide the rationale for more superordinate goals. From this
exploration an emergent aspirations system appears.
These infant candidate goals are the data used by the management
team as the basis for the first steps in exploring strategic intent through an
analysis of emergent competencies and their coherence in relation to aspi-
rations (Eden and Ackermann, 2000). In some cases these aspirations may
appear to be similar to the aspirations of other organizations in the par-
ticular industry/area. However, their structure (i.e. the linkages between
them and the means of realizing them) is expected to be substantially dif-
ferent. For example, Govan Initiative, an organization in Glasgow
(Scotland) dedicated to managing social and economic regeneration,
surfaced many of the same goals as have similar organizations dealing
with other parts of Glasgow. However, as they negotiated their own
strategy, it was through the particular emphasis taken on specific goals/
aspirations and their interaction as a network of goals that caused the
organization to be differentiated from the others.

REACHING AGREEMENTS

Step 3: Agreeing on the Goal System:

the Defining Strategic Intent

This early stage of the JOURNEY-making approach often focuses on

enabling the group to begin to explore their emergent aspirations system
and to identify the core beliefs that may represent distinctive competencies.
186 THEORETIC AND METHODOLOGICAL ISSUES

As noted above, the most hierarchically superordinate concepts will

provide the first draft of the aspirations or goal system, as in principle the
most hierarchically superordinate constructs are assumed to be goals.
Following this, each level below is considered in turn – on the basis of
whether the concept expresses a desired outcome that at least one member
of the management team sees as good ‘in its own right’. To be a goal, the
concept will not be treated as an option by the individual but will rather be
assumed to be a clear aspiration for the group and organization. In addition
to surfacing and working with positive statements, these aspirations may
be expressed as ‘negative goals’ – that is, ‘We will not allow this outcome to
occur. To keep it from happening is as important as attaining other positive
goals.’ For example, a publishing company seeing one of its key goals as
growth in particular aspects of its business, and considering both an
increase in product line and marketing effort, also noted that a negative
goal could be staff experiencing burn-out. At this time the negative word-
ing is retained.
Following the first identification of goals is an analysis of the group
map. Encompassed within this group map are likely to be emergent
competencies and distinctive competencies (Eden and Ackermann, 1998a,
2000). The group map evolves by asking the management team to
identify distinctive competencies within the model (and add new com-
petencies, if necessary) and subsequently to evaluate them carefully to
establish whether they are indeed distinctive or not. Distinctiveness is
always a relative ‘measure’, and so a critical issue, resolved with reference
to all nodes in the goals/aspirations system, is to determine the ‘anchor
points’ for relativeness. If an organization aspires to have the largest
market share in Europe, then distinctiveness is likely to be referred to a
European marketplace, whereas if the aspiration relates to France only,
then distinctiveness will likely be easier. When considering what is a dis-
tinctive competence, questions focus upon (a) whether it is difficult to
emulate, (b) whether it can easily be bought, (c) whether there is a high
cost of entry, and so on. In addition to agreeing on whether a particular
competency is distinctive in its own right, a distinctive competency may
also be a particular pattern of competencies that are unique to the organiza-
tion (either as a linked portfolio or as a feedback loop). Thus a loop of
competencies may provide competitive advantage as the group believes
that even if another organization could acquire the individual competen-
cies, they would not be able to put them together in the same way.
Once a first draft of the process has been undertaken, the results are
then viewed alongside the goal system and checks made to ensure that
each goal is supported by at least one distinctive competency. Where
this is not the case, the group then work on identifying new competencies
that must be developed in order to meet the aspiration/goal, or spend
effort reviewing the goal and possibly refocusing it. Where competencies
are found that do not support the goal system, discussion ensues as to
 A MAPPING FRAMEWORK FOR STRATEGY MAKING 187

whether they should be dropped, as the energy and resources required to

sustain them can be channelled to another area.
The important aspect of this process is recognizing that it is identifying
the systemicity of competencies (i.e. their relationships with one another)
and their relationship (support) with the goal system that provides added
value. The mapping process, through its ability to represent the system
graphically, can show how a Business Model, or Livelihood Scheme (for
public and not-for-profit sector), can be realized. Core distinctive compe-
tencies will not emerge until the cycle of exploration for coherence
between distinctive competencies and goals (the Business Model) has
been undertaken. This cycle usually leads to the identification of strate-
gies for the development of new competencies and diverts attention away
from useless distinctive competencies. (For a more detailed description of
the process, see Eden and Ackermann, 2000.)
Jointly understanding, reflecting upon and representing beliefs about
distinctive competencies and goals/negative goals is the first part of the
management team’s journey. But which is to be explored first? Our expe-
rience suggests that the best starting point is wholly dependent on the
particular nature of the organization and its management team. Some
teams are issue-focused and can hardly become involved in consideration
of strategy without discussion strategic issues/crises first. For many other
management teams, starting the discussion by focusing upon developing
the draft aspirations/goal system from surfacing issues can be lacking in
focus and too abstract. Therefore, for them, the first mapping episode is
usually more successful and rewarding if it focuses on an examination
and exploration of distinctive competencies, and then use the implicit
notion they have of distinctiveness as the basis for developing a draft goal
system. Some organizations already have a draft strategy statement –
goals and values – and so mapping this draft can act as a starting point for
further mapping of issues and competencies.

Step 4: Mapping Strategic Intent: the Strategy Map

We suggested earlier that strategy can be represented as a hierarchical net-

work of elements, or layers, which can become confirmed or redesigned
through the journey taken by the organization (Figure 7.4). At the apex is
the aspirations (goals) system, supported in turn by the strategies as a
system of interconnected statements about future direction. These in turn
are supported by strategic programmes as identifiable deliverables,
which in turn require the support of portfolios of strategic actions. The
important characteristic of the strategy map is that it represents strategy
as a hierarchical systemic network of interconnected statements of strategic
intent. This use of mapping to produce a network provides a range of
benefits, including the following:
188 THEORETIC AND METHODOLOGICAL ISSUES

• It ensures that the final strategic direction is coherent.

• It breaks down the task of agreeing and monitoring the different action
portfolios into manageable chunks.
• It assists those working to deliver strategic actions to appreciate how
their contribution relates to the strategic programmes and so to strate-
gies and onwards to the organization’s goals. Understanding ends and
means enables staff to become more engaged in the organization’s
strategic progress.
• It attains greater leverage through fully appreciating and exploiting
the multiple outcomes from each area of effort.
• It provides a focus on the ‘arrows’ – demonstrating that actions are
taken to achieve desired outcomes, rather than taken for their own sake.

The level of detail or elaboration of the strategy (i.e. how comprehen-

sively it is articulated, how extensive the network) is dependent on the
organization and its environment. Where there is considerable turbu-
lence, then a clearly articulated set of goals, associated distinctive compe-
tencies and attendant strategies may well be sufficient to provide a
substantive framework for strategic direction without stifling the need to
be contingent, opportunistic and flexible – but within a strategic frame-
work. For more stable environments, and in particular for the public
sector, where the need for accountability is a significant concern of power-
ful stakeholders, the development of programmes and action portfolios
becomes more important – particularly in light of each action’s evalua-
tion. This positioning on a contingency spectrum from deliberate emergent
strategy to detailed strategic planning (Eden and Ackermann, 1998a) will
influence the extensiveness of the network and is therefore a decision the
management team must make.
As noted above, developing agreement about strategic intent requires
the management team to (re)visit the goal/aspiration system, not only
ensuring that all potential goals have been considered, but subsequently
examining the goals in relation to the organization’s distinctive compe-
tencies to develop the Business Model. Where goals are not supported
and new competencies need to be grown, the key issues already surfaced
(along with any new ones) can be evaluated and the best candidate cho-
sen. These make up the strategies for the organization – each supporting
the achievement of either distinctive competences or aspirations. As the
team begin to refine the interconnected systems, new insights and a
deeper level of understanding are achieved. The network is given extra
meaning through the context of each statement, rather than relying on an
overview with broad ‘motherhood and apple pie’ statements.
Attending to the connections among goals and between the goal system
and the pattern of distinctive competencies often forces the team to con-
front ambiguities. This leads to developing clarity about what is actually
meant by each goal (the context or ‘tear-drop’ of associated material sup-
porting it). Considering it often forces the management team to take a
 A MAPPING FRAMEWORK FOR STRATEGY MAKING 189

deeper look at what they are aspiring towards. On a number of occasions

this process has brought the management team to a temporary standstill.
Their previously held belief that they were all wanting to pursue the same
ends, and only wanted assistance as to how to achieve these ends,
becomes unravelled and unanticipated differences become apparent.
Early stages of agreeing goals usually involve a draft of ‘too many’
goals – perhaps over 25 interlinked goals. At the beginning of the negotia-
ting process some form of categorization can be helpful. For example,
when Scottish Natural Heritage – a large not-for-profit organization –
examined their goal system, the management team found it helpful to
begin to weight goals allocating either one, two or three stars to each goal
depending on its significance. Another category used was to separate
short-term goals from longer-term ideals. Balancing the mix in this way
can often provide valuable motivation assistance, as it is easier to demon-
strate progress towards the short-term goals. A further useful form of
categorization is one that distinguishes facilitative goals from core goals.
Here the group seeks to identify those goals that distinctively determine
the purpose of the organization and those that exist only to support these
outcomes and yet are so important that they must be labelled as goals.

Step 5: Developing Action Programmes

Where it is appropriate to begin to develop action programmes, the strat-

egy map is used to guide action. The aim is to ensure that each strategy is
supported by a tear-drop of strategic programmes where any strategic
programme may act in support of more than one strategy, and so be
potentially more potent. Likewise, each of the strategic programmes rep-
resents a ‘tear-drop’ of options (potential actions), and some of the actions
will be particularly potent because they impact the outcomes of several of
these strategic programmes. The strategy map as a network therefore pro-
vides the basis for analysis of action plans through the exploration of the
following:

• The potency of actions within the network of causality – that is, those
actions that are expected to help in the fulfilment of many strategies or
goals. This can be undertaken in relation to programme delivery, and
thus ultimately in relation to aspirations. From this exploration it is
possible to indicate rough priorities, as relative levels of leverage are
calculated. This can act as a first stab at option evaluation – however,
one that must then be tempered by good judgement.
• The degree to which a set of actions work together as a portfolio and
can be used to cluster together responsibilities for delivery. For example,
the management team may create interdisciplinary teams to involve
staff from several divisions, departments or disciplines. This process
might commence with having the teams examine the current options,
190 THEORETIC AND METHODOLOGICAL ISSUES

add or amend them and begin to work on ways forward, taking

ownership of the outcomes through the process. An alternative
approach is to keep responsibility together in one part of the organi-
zation. In some cases organizations have been redesigned to reflect the
structure of the strategy map, so that, for example, particular tear-
drops become the responsibility of a single division or department.
• Alongside providing insights into the mode of delivery, the degree to
which sets of actions work together may give clues as to possible
synergies where the combined effect of the proposed actions yields an
outcome greater than the sum of the actions.

Finally a map can suggest the structure of actions within a network.

One method is to allow the resource demands of each strategy (as if the
strategy levels in the strategy map were ‘resource gates’ [Eden and
Cropper, 1992]) to be checked by tracking down the hierarchy. From this
point, it is possible to explore all actions that support the strategy and
their requirements.

Step 6: Using Maps to Test the Coherency of Strategy

Using mapping as the means of capturing and structuring group contri-

butions to strategy not only supports the process of negotiating and
agreeing strategic intent and developing action programmes, but also
makes possible coherence checks. The first and most superordinate
coherency check focuses upon examining the fit between the desired
strategic intention (goal system) and the organization’s current and future
competencies – formulating the Business Model/Livelihood Scheme.
Potential strategies, developed to achieve the goals, can be scrutinized in
the light of resources and rationale, increasing their viability and ensuring
their relevance. Strategic direction can be further developed by examining
its relevance against a number of possible alternative futures for the organi-
zation’s environment (Eden and Ackermann, 1998a: Chapters C8 and P4;
Eden and Ackermann, 1999). Not only do future scenarios provide a valu-
able means of testing options and strategies (Schwartz, 1991; van der
Heijden 1996; Wack 1985 a & b: 85), but the process of creating and dis-
cussing scenarios also gives rise to new options. These new options can
subsequently be incorporated into the map.
Another form of analysis for increasing the coherency of strategic direc-
tion is to examine it against the responses of different stakeholders inter-
ested in and influential over its delivery. Testing how different
stakeholders might respond to the various goals, strategies and actions
facilitates responses to lock in those who are positively inclined and
ensure their supportive nature, as well as manage those who are likely to
cause difficulties (Eden and Ackermann, 1998a: Chapters C7 and P4). For
example, within one organization, whose members were from a series
 A MAPPING FRAMEWORK FOR STRATEGY MAKING 191

of other organizations and mandated to promote economic and social

regeneration, the group deliberately considered how the various associ-
ated organizations would respond to the strategy direction proposed and
in the light of their findings revised certain aspects of the initial map to
ensure support.
Each of these episodes of analyses – testing the Business Model/
Livelihood Scheme, exploring alternative futures, and stakeholder
analysis – forces coherence within the strategy. As stakeholders are identi-
fied, particularly those with power as well as an interest/stake, their iden-
tification forces questions about the efficacy of strategies. ‘Role thinking’
their response to apparently effective strategies sometimes helps in
thinking through the potential dynamic disruption stakeholders might do
to the effective delivery of strategic outcomes. Futures are identified as
important and yet their importance may be difficult to comprehend when
seen in the light of the currently agreed system of aspirations.
The strategy map is the continuing reference point for all these consid-
erations. As a network, any conflicting causality will be identified easily
and quickly. The map also ensures that the strategy is made up of a coher-
ent set of individual discrete actions in support of a system of goals that
are supported as a portfolio by a self-sustaining critical mass or momen-
tum of opinion in the organization. It provides the ability to examine the
systemic network and check for its overall, holistic, logic.

DELIVERING STRATEGY: – THE MAP AS A MEANS

OF GAINING AGREEMENT AND GIVING MEANING
TO ACTION

Step 7: Getting ‘Buy in’ to Strategic Change

For senior managers and the top management team involved in strategy
making, maps can make clear what actions need to be taken in relation to
a strategic declaration. However, there may be resistance from the other
levels in the organization. This may be as much to do with a lack of clar-
ity about what is expected when, rather than being an outright disagree-
ment with the strategy. Thus, what is an actionable statement for one
person may remain ambiguous or meaningless for another. The confusion
can be further exacerbated when senior managers, who believe they are
acting consistently within a world of complex multiple goals, are per-
ceived to be acting inconsistently by others who are more singularly
focused in their tasks. Double messages seem to abound, particularly
where a senior manager demands one thing from her subordinates, but
appears to pay lip service to it herself by doing the opposite. Providing
clarification by presenting extracts of the strategy network – thus redu-
cing ambiguity and demonstrating the multiplicity of objectives – can
192 THEORETIC AND METHODOLOGICAL ISSUES

often alleviate this difficulty to some extent. The network shows what is
to be achieved, why and how – this is the meaning.
In contrast, one of the reasons why strategic change may be successful
is because as strategic conversations unfold as the logic of the map devel-
ops and actions are put into place, new ways of doing things appear
obvious and old ways are forgotten. Thus, the more successful the change
process, through being persuaded by rhetoric and logic, the less easy it is
to recognize it as change.
Facilitators face a similar situation when they successfully manage the
decision-making process of a team of managers by genuinely changing their
minds (rather than compromising them into agreement). A genuine change
of mind is often largely imperceptible for the person who has changed his
mind, and so the role of the facilitator in the process will be difficult to
acknowledge. As a consequence the facilitator is not rewarded and so not
used again! The facilitator who is acknowledged as successful is often
forced to use performance measures that can be counterproductive to the
reality of success, where hidden and subtle objectives may be paramount
(and only sometimes known to the single client rather than to the manage-
ment team as a whole). In the same way an organization and its managers
may feel pressured to create success measures for strategy that can be, in
many ways, counterproductive to successful embedded strategic change.

Step 8: Closure – Using the Map to Build a Strategy

Delivery Support System (SDSS) and to Measure
Progress

There are two important aspects of closure:– (a) the need (or not) for a rou-
tine symbol of strategy, such as the ‘mission statement’; and (b) the need
for action plans (and their potential for reducing flexibility). The mission
statement – often a textual version of the goal system, albeit with a focus
on touching the hearts and minds of those who read it – provides an exter-
nal portrait against which the organization can be compared and judged
by those outside of the organization. It is thus often a part of stakeholder
management strategy.
Action plans (individual contributions) can be realized through assign-
ing responsibilities and delivery dates (for start, progress or completion)
to particular members of staff – often agreed during computer-supported
workshops. As these actions are allocated and their time-scales settled, the
map allows loads to be ascertained, ensuring that staff are able to under-
take the tasks without endangering their current tasks (thus balancing
change with continuity). The structure provides ways of analysing the
implementation loads on individual staff, departments or divisions.
Colour coding can illustrate the amount of progress achieved (e.g. red for
no progress, blue for up to half-way, light blue for greater than half-way,
and silver for completion).
 A MAPPING FRAMEWORK FOR STRATEGY MAKING 193

The use of special computer software for mapping strategy becomes a

Strategy Delivery Support System (SDSS). This is a system that can be
routinely locked into strategy reviews with individual managers, teams
or the senior management team (Ackermann et al., 1992, 1993). The
system can also be used with great effect as a Decision Support System
(DSS) in the annual performance reviews for staff, so explicitly locking
reward systems into strategy delivery. Where progress is not being
achieved, further examination is possible in order to detect whether a
prior action is outstanding and thus incompletion is someone else’s
responsibility. Checks can also be made to ensure that the action being
pursued (and the progress monitored) matches that being achieved at the
level of strategies. Thus, once again, a coherency check is made possible.
Finally, the system acts as an organizational memory allowing regular
reference checks to be made as well as informing new members of the
organization of the strategic direction.

CONCLUSIONS2

Not only can mapping be a powerful means of capturing the knowledge,

ideas, beliefs and wisdom of the organization’s members embarking on
the journey of strategy making, but it can also do so in a manner that
structures the resultant material. Maps provide a means of facilitating con-
versation and negotiation in a way that seeks to amplify emotional and intel-
lectual commitment. Moreover, as context is provided along with content,
participants are able to reflect upon the material being explored, consider
alternative perspectives and thus provide the basis for the strategy
process to absorb social and political processes endemic in an organiza-
tion. Through the underlying procedural rationality (the logic of the maps
[Simon, 1976]) and procedural justice ([McFarlin and Sweeney, 1992;
Thibaut and Walker, 1975] enabling a wider participant set to be
involved), the resultant direction becomes something that those involved
find themselves becoming committed to.
Maps provide a means of differentiating one organization’s purpose and
strategic future from another through their ability to reflect the management
team’s deliberate choice in relation to the structure – the linking together
of different goals, strategies and actions. Many of the goals of an organi-
zation may be similar to those of others. For example, it is likely that for
most, if not all, private sector companies, the goal ‘increase profit and
growth’ is evident. None the less, the goals supporting this aspiration and
their relationships are unique. It is the pattern that drives the distinctive-
ness of the strategy.
As noted above, mapping also helps with ensuring the coherency of the
proposed strategic direction, enabling checks to be made of the underlying
logic. These checks are made not only across the different elements of stra-
tegy (goals, competencies, strategies, etc.) but also with the organization’s
194 THEORETIC AND METHODOLOGICAL ISSUES

context – its external and internal environment. Since the map provides
context, it is more likely to promote understanding of underlying purpose.
Through the sense of involvement that the maps engender, they are
more likely to promote ‘buy in’ to the subsequent results.
Finally, the mapping processes presented in this chapter can provide
assistance to implementation. In the first instance, this is through ensuring
that those responsible for actions are clear about what it is that they are
mandated to undertake. The map provides the rationale for action, thus
directing the effort more effectively. Second, the map – a form of model –
provides the management team with the ability to ensure that the resources
(in terms of people and time as well as more tangible assets) are available.
Finally, through seeing the strategy as a systemic network, not only can
individual items be assessed in terms of their progress, but triangulation
can take place where progress on detailed actions is compared with
progress on the higher-level concepts they were developed to support.

NOTES

1. The mapping process is substantially informed by Kelly’s Personal Construct Theory

(Kelly 1955).
2. The proposals summarized here are from Making Strategy, (Eden and Ackermann,
1998a), where theory, concepts, practice and sources are brought together in greater detail.

REFERENCES

Ackermann, F. (1992) ‘Strategic direction through burning issues: Using SODA as a strategic
decision support system’, OR Insight, 5: 24–28.
Ackermann, F. and Eden, C. (1997) ‘Contrasting GDSSs and GSSs in the context of strategic
change: Implications for facilitation’, Journal of Decision Systems, 6: 221–250.
Ackermann, F. and Eden, C. (2001) ‘Contrasting single user and networked group. Using
decision support systems for strategy making’, Group Decision and Negotiation, 10: 47–66.
Ackermann, F., Cropper, S. and Eden, C. (1992) ‘Moving between groups and individuals
using a DSS’, Journal of Decision Sciences, 1: 17–34.
Ackermann, F., Cropper, S. and Eden, C. (1993) ‘The role of decision support in individual
performance review’, in P.W.G. Bots, H.G. Sol and R. Traunmuller (eds), Decision Support
in Public Administration. Amsterdam: Elsevier Science Publishers BV. pp. 43–55.
Argyris, C. and Schon, D.A. (1974) Theories in Practice. San Francisco: Jossey-Bass.
Bartunek, J.M. and Moch, M.K. (1987) ‘First-order, second-order, and third-order change and
organization development interventions: A cognitive approach’, Journal of Applied
Behaviourial Science, 23: 483–500.
Bryson, J.M., Ackermann, F., Eden, C. and Finn, C. (1995) ‘Using the “Oval Mapping Process”
to identify strategic issues and formulate effective strategies’, in J. Bryson (ed.), Strategic
Planning for Public and Nonprofit Organisations. 2nd edn. San Francisco: Jossey-Bass.
de Geus, A. (1988) ‘Planning as learning’, Harvard Business Review, March/April: 70–74.
Dutton, J. and Ashford, S. (1993) ‘Selling issues to top management’, Academy of Management
Review, 18: 397–428.
 A MAPPING FRAMEWORK FOR STRATEGY MAKING 195

Eden, C. (1988) ‘Cognitive mapping: A review’, European Journal of Operational Research,

36: 1–13.
Eden, C. and Ackermann, F. (1998a) Making Strategy: The Journey of Strategic Management.
London: Sage.
Eden, C. and Ackermann, F. (1998b) ‘Analysing and comparing idiographic causal maps’, in
C. Eden and J.-C. Spender (eds), Managerial and Organizational Cognition. London: Sage.
pp. 192–209.
Eden, C. and Ackermann, F. (1999) ‘The role of GDSS in scenario development and strategy
making’, in the 6th International SPIRE/5th International Workshop on Groupware Proceedings.
Cancun Mexico, 22–4 September. IEEE Computer Society Los Alamitos, California.
pp. 234–242.
Eden, C. and Ackermann, F. (2000) ‘Mapping distinctive competencies: A systemic
approach’, Journal of the Operational Research Society, 51: 1–9.
Eden, C. and Cropper, S. (1992) ‘Coherence and balance in strategies for the management of
public services: Two confidence tests for strategy development, review and renewal’,
Public Money and Management, 12: 43–52.
Eden, C. and van der Heijden, K. (1995) ‘Detecting emergent strategy’, in H. Thomas,
D. O’Neal and J. Kelly (eds), Strategic Renaissance and Business Transformation. New York:
Wiley.
Eden, C., Jones, S. and Sims, D. (1979) Thinking in Organisations. London: Macmillan.
Eden, C., Ackermann, F. and Cropper, S. (1992) ‘The analysis of cause maps’, Journal of
Management Studies, 29: 309–324.
Galer, G. (1990) ‘A client’s perspective’, International Journal of Strategic Management,
23: 44–47.
Kelly, G.A. (1955) The Psychology of Personal Constructs. New York: Norton.
Lindblom, C.E. (1959) ‘The science of muddling through’, Public Administration Review,
19: 79–88.
McFarlin, D.B. and Sweeney, P.D. (1992) ‘Distributive and procedural justice as predictors of
satisfaction with personal and organizational outcomes’, Academy of Management Journal,
35: 626–637.
Mintzberg, H. and Waters, J.A. (1985) ‘Of strategies, deliberate and emergent’, Strategic
Management Journal, 6: 257–272.
Mintzberg, H., Raisinghani, H. and Théorêt, A. (1976) ‘The structure of unstructured deci-
sion processes’, Administrative Science Quarterly, 21: 246–275.
Schwartz, P. (1991) The Art of the Long View. New York: Doubleday.
Simon, H.A. (1976) ‘From substantive to procedural rationality’, in S.J. Latsis (ed.), Method
and Appraisal in Economics. Cambridge: Cambridge University Press.
Spender, J.-C. (1989) Industry Recipes: An Enquiry into the Nature and Sources of Managerial
Judgment. Oxford: Blackwell.
Thibaut, J. and Walker, J. (1975) Procedural Justice: A Psychological Analysis. Hillsdale, NJ:
Erlbaum.
van der Heijden, K. (1996) Scenarios: The Art of Strategic Conversation. Chichester: Wiley.
Wack, P. (1985a) ‘Scenarios, shooting the rapids’, Harvard Business Review, November/
December: 131–142.
Wack, P. (1985b) ‘Scenarios, uncharted waters ahead’, Harvard Business Review, September/
October: 73–90.
8
THE INDIVIDUAL IN THE
S T R AT E G Y P R O C E S S

Insights from Behavioural Decision Research

and Cognitive Mapping

Gerard P. Hodgkinson and A. John Maule

ABSTRACT

This chapter draws on the preliminary findings of a project entitled ‘Navigating

an uncertain world: Strategic cognition and risk.1 The primary objectives of this
work are two-fold: (a) to develop a more comprehensive understanding of the
cognitive processes underpinning strategic decision making; (b) to investigate,
under relatively controlled conditions, the prescriptive validity of mapping tech-
niques as a basis for helping executives and other organizational actors to over-
come cognitive bias and inertia.
We report findings from two separate experimental studies, using business
students in the laboratory and practising managers in an organizational field
setting, that demonstrate the efficacy of causal mapping techniques for attenu-
ating the well-documented framing bias. The framing bias occurs when trivial
changes to the way in which a problem is presented crucially affect preferences
for choice alternatives. A preliminary analysis of our participants’ maps illus-
trates the potential of causal cognitive mapping for gaining key insights into
mental representations underpinning this phenomenon. Our findings are inter-
preted in the context of a dual-process model of strategic decision making,
which differentiates situations in which cognitive mapping is likely to be bene-
ficial from those where it is likely to be dysfunctional.

A considerable volume of the burgeoning literature on strategic cognition

that has emerged over the past two decades or so has drawn on concepts
and theories from the field of behavioural decision making. It has become
 THE INDIVIDUAL IN THE STRATEGY PROCESS 197

almost mandatory for cognitively oriented strategy researchers to begin

their articles and books with the basic observation that, owing to funda-
mental information-processing limitations, or ‘bounded rationality’
(Simon, 1955, 1956), a variety of heuristics, or ‘rules of thumb’, such as the
‘availability’, ‘representativeness’ and ‘anchoring and adjustment’ heuris-
tics (Kahneman et al., 1982), are deployed by managers in an effort to
simplify reality, thus rendering their worlds manageable. Typically,
researchers who subscribe to this view go on to observe that an unfortu-
nate, latent consequence of the deployment of these heuristics is that they
can give rise to cognitive biases such as ‘hindsight bias’ (the tendency to
over-estimate the predictability of past events), ‘logical reconstruction’
(the logical reconstruction of events that cannot accurately be recalled)
and ‘wishful thinking’ (the tendency to over-estimate the probability of
desirable outcomes), which in turn may result in inappropriate/sub-
optimal decisions.
For a growing number of writers in the strategy field this stream of work
has laid the central conceptual foundations for developing cognitivist
accounts of strategic decision processes (Barnes, 1984; Das and Teng, 1999;
Schwenk, 1984, 1989, 1995) and there has been a steady accumulation of
evidence, within both laboratory and field settings, that suggests that
many of the phenomena identified by behavioural decision researchers in
other contexts are highly applicable in the context of strategic decision
making (Bateman and Zeithaml, 1989a, 1989b; Lant et al., 1992). Others,
however, have questioned the appropriateness of this body of knowledge
as a basis for advancing understanding of managerial and organizational
cognition (MOC) on the grounds that it is essentially behavioural in terms
of the observational methods and techniques employed and requires clear,
logical outcomes against which the observed behaviours of research par-
ticipants can be compared in order to assess the nature and extent of bias
in their reasoning. As Spender and Eden (1998: 2) observe:

For us, managerial and organizational cognition must go well beyond what
can be deduced from such observations because neither managers nor those
who research their behaviours have the logical answers available as reference
points. In practice managers make their decisions under conditions of infor-
mation inadequacy and other forms of uncertainty.

Thus, an alternative approach to the cognitive analysis of strategic decision

processes has arisen over recent years that explores the structure and con-
tent of actors’ mental representations of strategic problems in a relatively
direct fashion. This stream of work, predicated on the assumption that
actors construct a simplified representation of reality (or ‘mental model’)
that they use in an effort to navigate an uncertain world, has led to con-
siderable innovations among MOC researchers in the development and
application of cognitive mapping procedures as varied as repertory grid
techniques (Reger and Huff, 1993), card-sort methods (Daniels et al., 1995),
198 THEORETIC AND METHODOLOGICAL ISSUES

multidimensional scaling and clustering techniques (Hodgkinson, 1997a;

Hodgkinson et al., 1991, 1996), and the taxonomic interview procedures
pioneered by Porac and his associates (Porac et al., 1989). The fruits of
this approach are exemplified by the works contained within the present
volume and other edited collections published over the past decade (Eden,
1992; Eden and Spender, 1998; Hodgkinson and Thomas, 1997; Huff, 1990;
Meindl et al., 1996; Porac and Thomas, 1989).
While this second stream of research has enriched understanding of the
nature and significance of MOC in the strategy process, a central con-
tention of this chapter is that the wholesale rejection of the concepts,
theories, methods and empirical findings of behavioural decision research
at this stage in the development of the MOC field would be profoundly
misguided and lead ultimately to its impoverishment. The purpose of this
chapter is to argue that there is much to be gained by bringing together
these hitherto largely disparate streams of theory and research. The MOC
field is entering a period where the timely application of cognitive map-
ping techniques has the potential to enhance understanding of a variety
of phenomena identified originally in the laboratory by behavioural deci-
sion researchers. Similarly, scholars whose substantive research focus has
been concerned with developing the theory and practice of cognitive
mapping potentially have much to gain at this juncture by the judicious
application of concepts, methods and techniques from the field of behav-
ioural decision making.
In order to demonstrate the potential mutual enrichment that can be
accomplished by bringing these two streams of work closer together, the
chapter draws on the preliminary findings of a series of studies that
have been conducted by the authors in collaboration with several of their
colleagues at the University of Leeds (Hodgkinson and Bown, 1999;
Hodgkinson et al., 1998, 1999a, 1999b). Our primary objective in conducting
this research is to move beyond the application of experimental methods
and cognitive mapping procedures per se, in order to develop a more
comprehensive understanding of the cognitive processes underlying
strategic decision making. Our programme of research is focused on the
individual decision maker as the primary unit of analysis, while recog-
nizing fully that the act of strategizing takes place in the context of a socio-
political arena (Johnson, 1987; Mintzberg, 1983; Pettigrew, 1973, 1985;
Schwenk, 1989) and that strategies are the product of a negotiated order
(Walsh and Fahay, 1986), the consequence of which is that the conflicting
cognitions of differing stakeholders and stakeholder groups must some-
how be reconciled (Forbes and Milliken, 1999; Hodgkinson and Johnson,
1994). Our reasoning is that in order better to understand the cognitive
and behavioural dynamics of the strategy process at higher levels of
analysis (primary work group/team, the organization as a whole, or inter-
organizational network), it is first necessary to illuminate the mechanisms
underpinning the judgements and beliefs of the individual strategist as
 THE INDIVIDUAL IN THE STRATEGY PROCESS 199

he or she enters the various socio-political arenas in which he or she

must operate. This is in much the same way that social cognition scholars
have analysed the cognitive processes of individual actors as a necessary
precursor to the development of better theory and research on group and
social influence processes (see, e.g., Fiske and Taylor, 1991; Tesser, 1995).

DUAL INFORMATION-PROCESSING STRATEGIES

IN STRATEGIC DECISION MAKING: TOWARDS
A COGNITIVE MODEL

Drawing on previous research within the fields of strategic management

(Barnes, 1984; Hodgkinson, 1997a, 1997b; Reger and Palmer, 1996;
Schwenk, 1984, 1985, 1986, 1989, 1995), social cognition (Chaiken, 1980,
1987; Petty, 1995; Petty and Cacioppo, 1986) and behavioural decision
making (Kahneman et al., 1982), much of the early effort of the Leeds
team has been spent developing a dual-process model of strategic deci-
sion making (Hodgkinson and Bown, 1999; Hodgkinson et al., 1998,
1999a). A detailed consideration of our emerging theoretical framework
and its underlying rationale lies beyond the scope of the present chapter.
Briefly, however, we contend that the way in which individual actors
internally represent decision problems and evaluate alternative courses of
action in the process of strategizing is accomplished by means of two
different, mutually reinforcing information-processing strategies. Type I
or heuristic processing, is said to occur when individuals engage in a
relatively superficial analysis of stimuli. This is a largely automatic pro-
cess that lies beyond conscious control. Type II or elaborative processing,
by contrast, entails detailed analysis through conscious control.
Faced with complex decision choices under risk and uncertainty, the
accumulated body of theory and empirical research evidence suggests
overwhelmingly that strategic decision makers employ a range of heuris-
tics in an attempt to reduce the burden of information processing that
would otherwise ensue. Drawing on this body of evidence, it is hypothe-
sized that under Type I processing, actors have only limited attentional
resources at their disposal and are, therefore, more likely to deploy a vari-
ety of heuristics, such as the anchoring and adjustment, availability and
representativeness heuristics, in an effort to reduce the burden of infor-
mation processing upon them. Unfortunately, however, as noted earlier,
the deployment of heuristic processing strategies can have a deleterious
effect on decision making. To the extent that an individual is reliant on
Type I processing strategies, therefore, there is a much greater likelihood
that he or she will exhibit biased reasoning. It does not necessarily follow,
however, that Type II processing strategies per se will result in better-
quality decisions. Indeed, a key danger associated with Type II processing
strategies is that actors may become overwhelmed by the sheer quantity
200 THEORETIC AND METHODOLOGICAL ISSUES

of information at their disposal. It is precisely for this reason that Type I

processing strategies have evolved.

Research Implications

Utilizing this basic distinction between Type I and Type II processing

strategies, it is possible to raise a number of non-trivial, descriptive and
prescriptive issues that need to be addressed by the cognitive mapping
community. At the descriptive level, to what extent do the mental repre-
sentations of individuals engaged in elaborative processing strategies dif-
fer from those adopting heuristic processing strategies? At the very least,
if our distinction between Type I and Type II strategies is meaningful, it
should be possible to detect differences in the structure and content of
actors’ cognitive maps elicited under differing experimental conditions
designed to induce these alternative processing strategies.
At the prescriptive level, to what extent does the application of cogni-
tive mapping procedures, deployed as an intervention technique, result in
changes in processing strategy and, hence, lead to refinements in the
judgements of strategic decision makers? In the context of intervention
work, one possible outcome of the application of cognitive mapping
procedures is that they may cause actors to switch from heuristic to
elaborative processing strategies, which in turn may result in significant
reductions in cognitive bias. To the extent that our reasoning in this
respect is correct, by utilizing experimental procedures it should be pos-
sible to demonstrate, under controlled conditions, that the deployment of
cognitive mapping techniques leads to significant reductions in a number
of cognitive biases, such as those outlined above (see also Barnes, 1984;
Bazerman, 1998; Das and Teng, 1999; Goodwin and Wright, 1998; Maule
and Hodgkinson, in press; Schwenk, 1984, 1995).

EXPLOITING SYNERGIES BETWEEN COGNITIVE

MAPPING AND THE EXPERIMENTAL METHOD:
THE CASE OF FRAMING

In an effort to address these broad research questions the Leeds team has
embarked on a series of laboratory and field studies that have utilized the
experimental method in conjunction with cognitive mapping techniques.
To demonstrate the potential of our approach, preliminary findings from
two related studies are reported. As will be seen, encouraging progress has
been accomplished in respect of both research questions outlined above.
The studies in question were designed to gain further insights into a
particular form of cognitive bias known as ‘the framing bias’ (Kahneman
and Tversky, 1984; Tversky and Kahneman, 1981), a factor known to
 THE INDIVIDUAL IN THE STRATEGY PROCESS 201

reduce the quality of decisions in a wide variety of situations (Bazerman,

1998; Goodwin and Wright, 1998) and that previous research has sug-
gested is likely to be applicable in the context of strategic decision making
(Bateman and Zeithaml, 1989a, 1989b). This bias arises when trivial
changes to the way in which a decision problem is presented, emphasiz-
ing either the potential gains or potential losses, lead to reversals of
preference, with decision makers being risk-averse when gains are high-
lighted and risk seeking when losses are highlighted (Kahneman and
Tversky, 1979, 1984; Tversky and Kahneman, 1981), as illustrated by the
following problem (taken from Bazerman, 1984: 333–334):

A large car manufacturer has recently been hit with a number of economic diffi-
culties, and it appears as if three plants need to be closed and 6,000 employees
laid off. The vice president of production has been exploring alternative ways to
avoid this crisis. She has developed two plans:
Plan A: This plan will save one of the three plants and 2,000 jobs.
Plan B: This plan has a one-third probability of saving all three plants and
all 6,000 jobs, but has a two-thirds probability of saving no plants and
no jobs.
Which plan would you select?

Now reconsider this problem, replacing choices A and B, above, with the
following choices:

Plan C: This plan will result in the loss of two of the three plants and 4,000
jobs.
Plan D: This plan has a two-thirds probability of resulting in the loss of all three
plants and all 6,000 jobs, but has a one-third probability of losing no
plants and no jobs.
Which plan would you select?

An analysis of each pair of decision choices reveals that they are objec-
tively identical (Plan A is the same as Plan C, and Plan B is the same as
Plan D). Nevertheless, when presented with the first pair of choices, an
overwhelming majority of individuals express a preference for Plan A,
whereas Plan D becomes the favourite when presented with the second
set of choices. Thus, the way in which problems are presented can
dramatically alter the way in which outcomes are evaluated.
Kahneman and Tversky (1984) have attempted to account for these
systematic departures from rationality by means of ‘prospect theory’.
According to this theory, actors evaluate potential outcomes as gains and
losses relative to a neutral reference point, situated mid-way along an
202 THEORETIC AND METHODOLOGICAL ISSUES

Value units

200

150

100

Losses 50
Gains

Plants −3 −2 −1 1 2 3
Jobs − 6000 − 4000 − 2000 2000 4000 6000
− 50

− 100

− 150
Risk neutrality line
Hypothetical value curve
− 200

Figure 8.1 Hypothetical value function accounting for framing (adapted

from Bazerman, 1998: 49 [© John Wiley & Sons Inc.], based on an
original figure by Kahneman and Tversky, 1979 [© The Econometric
Society]). Adapted by kind permission of both publishers.

S-shaped value function, as indicated in Figure 8.1. In this figure the units
gained or lost relative to the neutral point are plotted along the x-axis,
while the units of utility associated with these varying levels of gain or
loss are plotted along the y-axis. The relationship between gains/losses
and utility, described by the S-shaped value function, suggests that deci-
sion makers are risk-averse when choosing in gains, but risk seeking
when choosing in losses. Hence when people build a mental representa-
tion of the problem in situations where the outcomes are described in
terms of potential gains (number of plants and jobs saved: plans A and B),
the potential disaster of losing everything becomes the neutral reference
point and outcomes are evaluated along the upper portion of the value
function. In consequence, the majority of decision makers express a pre-
ference for the safer outcome (a risk-averse strategy). In situations where
the problem is framed in terms of job and plant losses (plans C and D), by
contrast, the current position of having three plants open becomes the
 THE INDIVIDUAL IN THE STRATEGY PROCESS 203

neutral reference point. Under this scenario, the two choices involving
job and plant losses are evaluated along the lower portion of the value
function in Figure 8.1, and as a result, a risk-seeking strategy becomes the
preferred outcome.
Prospect theory not only enables us systematically to identify the ways
in which the framing of problems causes decision-making behaviour to
deviate from rationality; it also suggests that our response to loss is more
extreme than our response to gain. The value function in Figure 8.1 indi-
cates that, ordinarily, the pain of losing $x is generally greater than the
pleasure of winning the same amount.
Potentially, prospect theory and the framing-bias phenomenon have a
number of important implications for the field of strategic management,
both at the level of descriptive theory building and from an interven-
tionist perspective. For example, a key potential danger confronting
strategists faced with complex decision choices under risk and uncer-
tainty is that the way in which problems are presented, or framed, may
bias their judgements, which in turn may give rise to sub-optimal deci-
sions. Simply stated, the choices adopted by strategists may ultimately
be determined by the way in which the options have been framed, or pre-
sented, rather than by their objective utility. In situations where riskier
options are presented as a glass half full (positively framed), decision
makers are likely to adopt a risk-averse strategy and opt for the safer
alternative, whereas the same proposal presented as glass half empty
(negatively framed) will increase the chances of a risk-seeking strategy
being selected. Previous research in the field of strategic management
lends credence to this hypothesis.
In a pair of laboratory experiments designed to explore ‘the psychologi-
cal context of strategic decision making’, Bateman and Zeithaml (1989a)
investigated the impact of decisional frame (positive versus negative
future outlook), together with feedback on a past decision (success versus
failure) and perceived organizational slack (high versus low) on a re-
investment decision. In the first experiment, involving undergraduate
students, all three variables were found to have main and interactive
effects on the dependent variable (the amount in US dollars participants
were prepared to reinvest). The second experiment, involving business
executives, combined the decision feedback and organizational slack vari-
ables (i.e. failure feedback/low slack versus success feedback/high slack)
and investigated the impact of this combined variable and decisional
frame on the amount participants were prepared to reinvest. A significant
main effect was obtained for decisional frame, together with a significant
interaction effect between decisional frame and the combined feedback
and organizational slack variable. Although in the predicted direction, the
main effect for the combined feedback and organizational slack variable
was found to be non-significant.
Taken together, these studies show that the way in which information
concerning strategic decisions is presented, or framed, has an impact on
individual choice preferences. Merely demonstrating that this phenomenon
204 THEORETIC AND METHODOLOGICAL ISSUES

occurs in the context of laboratory studies involving strategic decision

scenarios, however, does not enable us to answer two vital questions:
(1) What is the precise cognitive mechanism underpinning this pheno-
menon? (2) What steps might be taken to minimize its effects? In the
following sections we report two of our ongoing studies that address these
concerns directly.
Using a variant of causal cognitive mapping (Axelrod, 1976) we have
begun to explore actors’ mental representations of relatively complex
decision scenarios, elicited under controlled, experimental conditions
designed to induce a framing bias. In so doing, our aim is to gain insights
into the strengths and weaknesses of this particular mapping technique,
both as a basis for better understanding why the framing bias occurs and
as a means for limiting the potential damage accruing from this pheno-
menon. Causal mapping techniques are particularly attractive for both of
these purposes because, as observed by Huff (1990: 16), they are action-
oriented, requiring individuals not only to reflect on events occurring
prior to the current situation, in an effort to derive plausible explanations,
but also to anticipate future changes.

METHOD

Participants

The first study was an extended pilot study involving a total of N = 88

undergraduate students of business and management, nearing the end
of their degree programmes. This study examined the feasibility of
our methodological approach in general, and tested some of our key
hypotheses, in a relatively controlled environment. The second study was
a follow-up in the field, involving a sample of senior managers.
It is worth noting at this point that previous investigations into the
effects of framing in strategic management (Bateman and Zeithaml,
1989a) have failed to reveal substantive differences in findings between
studies using undergraduate students as participants and those using
experienced practitioners completing MBA degrees. Nevertheless, in
order to ensure that our findings are generalizable to practitioners,
hence avoiding the charge of conducting research that is rigorous, but
none the less irrelevant (Anderson et al., 2001; Hodgkinson and
Herriot, in press; Huff, 2000; Tranfield and Starkey, 1998), our second
study was designed to replicate and extend Study 1 in an organi-
zational setting; a banking organization based in the UK. Fifty-two
senior managers from a total of 204 initially contacted agreed to take
part in this follow-up investigation, a response rate of approximately
25 per cent.
 THE INDIVIDUAL IN THE STRATEGY PROCESS 205

Case Materials

The stimulus materials employed in both studies were based on real-life

decision episodes, and in both studies participants were required to draw
cause maps in an effort to uncover their mental representations of the
strategic decision scenario in question. Both sets of materials comprised a
case vignette (circa 500–750 words) culminating in a decision involving
two choice alternatives.
In both studies elaborate arguments were presented concerning the
pros and cons of each alternative. The participants were required to make
an investment decision. As per the Bazerman (1984) plant closure problem
outlined earlier, two different versions of each problem were developed:
a positively framed version, highlighting the potential gains; and a nega-
tively framed version, highlighting potential losses.
The vignette employed in Study 1 described a strategic investment
decision facing a manufacturer and distributor of fast paint-drying sys-
tems used in the repair of automotive vehicles. A ten-year history of the
company, from its inception to the present day, culminating in the deci-
sion dilemma, was presented and participants were required to assume
the role of a member of the Board and state which of two investment alter-
natives – ‘continue in the domestic market’ (the ‘safe’ alternative) or
‘invest overseas’ (the ‘risky’ alternative) – they would recommend. Thus,
in the positively framed version, participants were presented with the
following choices:

A Developing a new marketing effort within the domestic market and not
attempting to export overseas. Market research indicates that this option
would certainly lead to profits of £1 million.
B Halting new developments within the domestic market but a commitment
to the export market overseas. Market research indicates that this initiative
would lead to profits of £3 million with probability one-third, and no profits
with probability two-thirds.

In the negatively framed version, by contrast, they were required to

choose between:

C Developing a new marketing effort within the domestic market and not
attempting to export overseas. Market research indicates that this option
would certainly lead to profits £2 million below target.
D Halting new developments within the domestic market but a commitment
to the export market overseas. Market research indicates that this initiative
would lead to profits at target level with probability one-third, and profits
£3 million below target level with probability two-thirds.
206 THEORETIC AND METHODOLOGICAL ISSUES

Apart from these particular variations, the stimulus materials presented

to participants were identical in all respects. In this way we were able to
impose experimental control on the decision task.
The materials employed in our first study were derived from a careful
analysis of documentary sources, chiefly newspaper accounts. In order
to further enhance the ecological validity of our work, the materials employed
in the follow-up study were based around a ‘live issue’ and were developed
and pilot-tested on the basis of a series of briefing meetings held with
members of the senior management team, in conjunction with documen-
tary sources both internal and external to the organization concerned. None
of the managers who assisted with the development of the stimulus mate-
rials subsequently acted as participants in the main exercise.
The case materials employed in Study 2 were centred on a strategic
issue currently facing many financial organizations, namely the extent to
which they should continue investing in conventional delivery channels
(chiefly high-street branch networks) and/or experiment with new forms
of delivery (smart cards and telephone and Internet banking). While there
is considerable agreement that present forms of banking have a limited
future, there is still great uncertainty regarding which of the various tech-
nologies currently under review will ultimately come to dominate the
industry. Since each of these technologies is highly capital-intensive,
investment in the ‘wrong’ technology could prove disastrous for the organi-
zations concerned. Building directly on this scenario, the case entailed a
fictitious bank (with a similar customer profile to the participants’ actual
organization) contemplating an investment in two alternative technolo-
gies: a telephone banking system involving the use of machine-readable
swipe cards; and a networked PC banking system. As in Study 1, elabo-
rate arguments were presented concerning the pros and cons of each alter-
native and participants were required to make an investment decision.
However, following the briefing meetings, it was apparent that a decision
involving strictly dichotomous choices would lack realism for the partici-
pants. In order further to enhance the ecological validity of this study,
therefore, we developed a decision task that required participants to allo-
cate a fixed sum of money (£15 million) to each of the alternatives directly
in proportion to their strength of preference. Thus participants in this
second study had the option of allocating all the money to one or other of
the two alternatives, should they wish to do so. Alternatively, they could
split the money across both options, in any proportion, with the proviso
that the total amount allocated summed to the full £15 million.

Cognitive Mapping Procedure

The cognitive mapping task entailed the participants completing a cause

map, which was drawn by hand, based on a procedure developed by
Green and his associates (Green and McManus, 1995; Green et al., 1998).
 THE INDIVIDUAL IN THE STRATEGY PROCESS 207

In both studies participants were presented with a comprehensive list of

variables that the research team had identified on the basis of a concep-
tual analysis of the relevant case vignette. Using only those variables that
they had considered as they thought about the decision problem, partici-
pants were required to construct a network diagram, with each variable
represented as a node. With the proviso that their representations must
incorporate both choice alternatives, participants were free to choose as
many or few of the variables as they saw fit, but were advised that between
three and ten variables are ordinarily incorporated in a typical map.
Within the system employed in these studies, causal relations are repre-
sented as linkages between nodes, drawn as lines with an arrowhead
depicting the direction of causality and a number (ranging from +3 to −3)
depicting the nature and strength of the relationship. Perceived positive
causal relations are indicated by positive strength ratings (+3 representing
the strongest possible positive effect) and perceived negative causal rela-
tions with negative strength ratings (−3 denoting the strongest possible
negative effect).

Research Design

In both studies we employed a two (positive vs. negative problem frame)

by two (pre- vs post-choice mapping) between-participants experimental
design. In other words, participants completed one version only of the
research tasks, allocated on a random basis.

FINDINGS

It is convenient to consider our findings under two separate headings: the

analysis of choice behaviours and the analysis of the cognitive maps.
Although data collection for both studies has been fully completed, analy-
sis of the data pertaining to cognitive mapping is ongoing, so should be
regarded as tentative at this time.

Analysis of Choice Behaviours

Thus far, our attention has largely been confined to the testing of two
crucial hypotheses:

H1 There will be a significant difference in risk preferences under post-choice

mapping conditions. Specifically, significantly greater numbers of partici-
pants allocated to the positively framed condition will express a preference
for the safe-choice alternative (risk-averse behaviour) as opposed to the
208 THEORETIC AND METHODOLOGICAL ISSUES

risky alternative (risk-seeking behaviour). Conversely, significantly greater

numbers of participants allocated to the negatively framed condition will
express a preference for the risky-choice alternative (risk-seeking behav-
iour) as opposed to the safe alternative (risk-averse behaviour).
H2 Under pre-choice mapping conditions there will be no significant differ-
ences in risk preferences.

Our first hypothesis is derived directly from previous work on framing

(Bateman and Zeithaml, 1989a). Support for this hypothesis would show
that the framing bias is not restricted to simple laboratory problems and
is a potentially influential factor in more complex strategic situations.
H2 follows directly from our earlier discussion of Type I and Type II
processing strategies in the context of interventionist work. If cognitive
mapping techniques encourage actors to engage in elaborative thinking
consistent with our notion of Type II processing, experimental manipula-
tions designed to induce cognitive biases should no longer be effective
when participants are required to engage in a cognitive mapping task
prior to decision making. In the context of the present study, this means
that the framing bias predicted under H1 should no longer be evident
under pre-choice mapping conditions.
As reported in Hodgkinson et al. (1999b), the evidence accumulated
from both studies strongly supports each of our hypotheses. As expected,
Study 1 revealed a statistically significant association between choice
behaviour and framing condition for those participants who were
required to complete the cognitive mapping task after deciding which of
the two choice alternatives they preferred (χ 2 (df = 1; n = 44) = 7.333, p = .007).
This finding provides strong evidence of bias attributable to the framing
manipulation. In keeping with the predictions of prospect theory, the
number of participants favouring the risk-seeking alternative was higher
following exposure to the negatively framed version of the problem
(N = 20, 45.5%) than the positively framed version of the problem (N = 12,
27.3%). Conversely, the number of participants favouring the risk-averse
alternative was higher following exposure to the positively framed ver-
sion of the problem (N = 10, 22.7%) than the negatively framed version of
the problem (N = 2, 4.5%).
Turning to the results of the pre-choice mapping conditions, by con-
trast, we find no evidence of a framing bias; rather, equal proportions of
participants express a preference for the risk-seeking and risk-averse
alternatives (N = 8, 18.2% vs N = 14, 31.8%), irrespective of whether they
have been exposed to the positively or negatively framed version of the
problem (χ 2 (df = 1; n = 44) = 1.00, p = 1.00).
The findings of our follow-up study in the field tell a similar story. The
results indicated that in the post-choice mapping condition the average
amount of money allocated to the safe alternative was higher following
 THE INDIVIDUAL IN THE STRATEGY PROCESS 209

exposure to the positively framed version of the problem (mean = £6.07

million, mean rank = 16.50) in comparison to the negatively framed
version of the problem (mean = £4.31 million, mean rank = 11.31). As
expected, the results of a Mann–Whitney U test (corrected for ties) con-
firmed that this trend is significant (z = −1.81, p = .035, 1-tailed), once again
providing evidence of a framing bias.
As in Study 1, the data gathered under the pre-choice mapping condi-
tion did not reveal any evidence of bias attributable to the framing mani-
pulation. A Mann–Whitney U test (again corrected for ties) indicates that
the average amount of money allocated to the safe alternative was no
higher following exposure to the positively (mean = £5.67 million, mean
rank = 12.40) or negatively (mean = £6.30 million, mean rank = 13.90)
framed version of the problem (z = −0.51, p = .30, 1-tailed), thus providing
corroborating evidence in the field for the findings obtained in our earlier
laboratory study.
The fact that a significant association between choice behaviours and
problem frame was observed in both the laboratory and the field study
under the post-choice mapping conditions confirms that the framing bias
is not just restricted to simple laboratory problems and is a potentially
influential factor in more complex strategic decisions involving experi-
enced, senior managers. However, no significant association between
problem frame and choice behaviour was observed for those participants
who completed the cognitive mapping task prior to decision making. This
suggests that engaging in the cognitive mapping task prior to decision
making attenuated the framing effects observed under the pre-mapping
choice conditions. The results of these studies are consistent with a grow-
ing body of opinion that effortful thought can attenuate or eliminate the
biasing influence of decision frames (e.g. Maule, 1995; Sieck and Yates,
1997; Smith and Levin, 1996; Takemura, 1994).

Analysis of the Cause Maps

To date, our analysis of the participants’ cause maps has been limited to a
preliminary inspection of their structure and content on a case-by-case
basis; nevertheless, sufficient evidence has accumulated to illustrate the
considerable potential of our approach to mapping as a basis for investi-
gating the nature and significance of the mental representations under-
pinning actors’ choice behaviours. For ease of presentation, the examples
that follow are all taken from individuals whose maps were elicited under
the post-choice mapping conditions of our laboratory investigation (i.e. in
all cases the mapping exercise was completed once the individual con-
cerned had made his or her decision choice), but are typical of those
collected across all four conditions of both studies.
210 THEORETIC AND METHODOLOGICAL ISSUES

(10)
(5) Proven
Need to sustain/increase technology
−3 past profit levels
−3 (17)
(15) Proven success in +3
Greater competition domestic market
in domestic market +3
(4)
−3 +2 Inexperience
+3 (1) −3 overseas
(2)
RENEW MARKETING −3
COMMITMENT TO THE
+3 EFFORT IN THE OVERSEAS MARKET
DOMESTIC MARKET −1 −3
+2 + 2 (16)
+2 Continued decline
+3 +1 +3 in profit
(14) +3
Growing innovations in
(11)
alternative technologies (8) Development of
in domestic market Long-term future +3
(12) new business
Excellent reputation success of ES
in domestic market Paint Systems

−3

Figure 8.2 Example of a cognitive map elicited from a participant

exposed to the negatively framed version of the problem in Study 1,
whose decision choice (‘commitment to the overseas market’) is in
accordance with the predictions of prospect theory

Figures 8.2 and 8.3 present the cause maps of two of the individuals
who were exposed to the negatively framed version of the problem, while
sample maps elicited from participants exposed to the positively framed
version of the problem are presented in Figures 8.4 and 8.5. As might be
expected, given the nature of the decision task employed in this study, the
most striking feature of these maps is the distinctive pattern of activity
centred on the choice nodes.
The map presented in Figure 8.2 was elicited from an individual whose
decision choice was consistent with the predictions of prospect theory;
that is, this individual chose the risky alternative (‘commitment to the
overseas market’), and this is reflected by a markedly greater number of
in-degrees associated with the chosen alternative relative to the rejected
alternative (nine vs four). Furthermore, the vast majority of in-degrees
(seven out of nine) associated with the chosen alternative are positively
valanced.
Figure 8.3, by contrast, presents the map of an individual whose choice
behaviour is not consistent with the predictions of prospect theory.
Notwithstanding the fact that this individual was similarly exposed to the
negatively framed version of the problem, they chose the safer alternative
(‘renew marketing effort in the domestic market’). As can be seen, this
 THE INDIVIDUAL IN THE STRATEGY PROCESS 211

(4)
(3) Inexperience in
Technology and overseas market (14)
product less well Growing innovations in
(5) known overseas alternative technologies
Need to sustain or
in the domestic market
increase past profit +3 −2
levels +2 −1 +1
+1 +2 −1
+1
(8) +1
+3 (1) (2)
Long-term future
RENEW MARKETING COMMITMENT TO
success of ES
+3 EFFORT IN THE THE OVERSEAS MARKET
Paint Systems +2
DOMESTIC MARKET
+2 +3
+3 −3
−1 +1 +1
+2
(12)
(13)
Excellent (15)
Network of overseas
domestic Greater competition in
distribution agents
reputation domestic market
required

−2

Figure 8.3 Example of a cognitive map elicited from a participant

exposed to the negatively framed version of the problem in Study 1,
whose decision choice (‘renew marketing effort in the domestic market’)
runs contrary to the predictions of prospect theory

map exhibits a very different structure to that presented in Figure 8.2,

with high levels of activity centred on both choice nodes. Although the
overall number of in-degrees associated with the chosen and rejected
alternatives is almost identical (eight vs seven), in the former case all eight
of the in-degrees are positively valanced, whereas in the latter case five of
the seven in-degrees are negatively valanced.
A similar pattern of findings can be discerned in the maps we have
elicited from participants exposed to the positively framed version of the
problem. The map presented in Figure 8.4, for example, was elicited from
another individual whose choice behaviour conforms to the basic predic-
tions of prospect theory. As expected, this individual chose the risk-averse
alternative (‘renew marketing effort in the domestic market’) when pre-
sented with the positively framed version of the problem. As in the case
of the map presented in Figure 8.2, the bulk of activity in this individual’s
map is centred on the chosen alternative relative to the rejected alternative
(four in-degrees and one out-degree vs two in-degrees and one out-degree),
the majority of the in-degrees associated with the chosen alternative being
positively valanced (three out of four) while both of the in-degrees associ-
ated with the rejected alternative are negatively valanced.
Turning to the map presented in Figure 8.5, we find a similar pattern to
the map presented in Figure 8.3. Contrary to the predictions of prospect
theory, this individual chose the risk-seeking alternative (‘commitment to
the overseas market’) when a risk-averse response (‘renew marketing
212 THEORETIC AND METHODOLOGICAL ISSUES

(8)
Long-term future
(5)
success of +1
Need to sustain or
ES Paint systems
increase past profit levels

+2 +2

(1) (2)
RENEW MARKETING −3
Commitment to
EFFORT IN THE the overseas market 6
DOMESTIC MARKET +2
−2 −3
+3
(4)
Inexperience in
(17) overseas markets
Proven successs in
domestic market

Figure 8.4 Example of a cognitive map elicited from a participant

exposed to the positively framed version of the problem in Study 1,
whose decision choice (‘renew marketing effort in the domestic market’)
is in accordance with the predictions of prospect theory

effort in the domestic market’) was expected to be the favoured outcome.

As in the case of the map presented in Figure 8.3, the map of this indivi-
dual exhibits a relatively balanced pattern of activity across the two choice
nodes, with roughly equal numbers of in-degrees associated with the
chosen and rejected alternatives (eight vs six). Similarly, in this instance
we find that all eight of the in-degrees associated with the chosen alter-
native are positively valanced, whereas three of the six in-degrees associ-
ated with the rejected alternative are negatively valanced.

DISCUSSION

The cognitive perspective on strategic management is still in its infancy.

As we have seen, two major streams of research have influenced its develop-
ment thus far: the first has applied concepts from the heuristics and biases
perspective of the experimental cognitive psychology and behavioural
decision-making literatures, relying primarily on extrapolation from the
laboratory in conjunction with the use of observational methods in the
field; the second has utilized cognitive mapping techniques in an attempt
to study actors’ mental representations of strategic problems in a rela-
tively direct fashion. We began this chapter with the assertion that there
is much to be gained for the MOC field as a whole by bringing together
these hitherto largely disparate streams of research. The programme of
ongoing research reported in this chapter demonstrates the considerable
benefits to be attained from such a synthesis.
 THE INDIVIDUAL IN THE STRATEGY PROCESS 213

(12)
Excellent reputation
+2 in domestic market
(15)
Greater competition +2 (11)
(8)
in the domestic market Development of
Long-term future success
of ES Paint System new business
−1
+2 +2
−1 +3
(1)
RENEW MARKETING EFFORT (2)
IN THE DOMESTIC MARKET COMMITMENT TO THE
+2 + 2 OVERSEAS MARKET
(10) +2
−1 Proven technology
+3
+2 (3)
(16)
Product and technology
Continued decline
less well known overseas
in profit
+2
(17)
Proven success in
domestic market

Figure 8.5 Example of a cognitive map elicited from a participant

exposed to the positively framed version of the problem in Study 1,
whose decision choice (‘commitment to the overseas market’) runs
contrary to the predictions of prospect theory

Our studies confirm that the framing bias, previously identified in

relatively simple decision problems given to relatively inexperienced
individuals, is also likely to be a feature of the complex decisions taken by
experienced individuals in the context of strategy making in business
organizations. In addition, perhaps the most important finding of our
work to date from an interventionist perspective is that causal cognitive
mapping provides an effective means for de-biasing strategists’ judge-
ments arising from the effects of framing. Our results indicate that causal
mapping is as effective for experienced as for inexperienced decision
makers, thus providing strong support for the validity of this technique
as a basis for improving the overall quality of decision making.
Work is still underway to identify the precise mechanism(s) by which this
de-biasing occurs, and to this end we are in the process of developing
several new indices for comparing the participants’ cognitive maps in order
to test these hypotheses formally, with an appropriate degree of statistical
rigour. In the longer term, this should enable us to refine the tools and tech-
niques of causal mapping as a basis for intervening in the strategy process.

FUTURE DIRECTIONS

Our studies have demonstrated the considerable benefits to be gained by

utilizing concepts and methods from the field of behavioural decision
214 THEORETIC AND METHODOLOGICAL ISSUES

making in conjunction with cognitive mapping techniques. Additional

work is now required in order to explore further these synergies through
the investigation of other cognitive biases. One area worthy of immediate
investigation in this respect is the well-documented ‘escalation of commit-
ment’ phenomenon.
‘Escalation of commitment’, the tendency to commit further resources
to a failing course of action (Staw, 1981), has been closely linked not only
to the framing bias, but also to a variety of other cognitive biases, includ-
ing ‘illusion of control’ and ‘over-confidence in judgement’ (see, e.g., Huff
and Schwenk, 1990; Schwenk, 1986). Extrapolating from our studies of the
framing bias, causal cognitive mapping techniques may also prove useful
as a means for enriching understanding of the ‘escalation of commitment’
phenomenon.
On the basis of our framing studies, it is tempting to assume that the
greater enrichment of actors’ mental models that results from the applica-
tion of cognitive mapping techniques will invariably lead to better-quality
decision making and the avoidance of systematic cognitive biases, and
that, therefore, organizations should embrace decision-aiding techniques
that systematically encourage effortful thought. However, this need not
necessarily be the case. Less effortful thought associated with Type I
(heuristic) processing may, on some occasions, be highly functional for
navigating an information-rich yet highly uncertain world, enabling
strategists to reduce greatly the information-processing burden that
would otherwise ensue. In addition, in the context of escalation behav-
iour, Type II processing may serve merely to affirm an individual’s extant
worldview, rather than challenge it. The suggestion here is that mapping
may simply act as a vehicle for elaborating existing ways of thinking, thus
providing the basis for decision makers to lend increasing support to the
failing course of action, rather than abandon their original decision. On
the other hand, following the success of our studies of the framing bias, it
may prove possible, through recourse to cognitive mapping, to overcome
a range of biases that previous research has linked to the escalation phe-
nomenon, thereby reducing the tendency to escalate. In an effort to clar-
ify this pressing issue, the Leeds team is investigating these possibilities
directly in a series of additional studies.
At the heart of this issue is the question as to when heuristic (Type I)
and elaborative (Type II) processing strategies switch from being func-
tional to being dysfunctional in terms of their respective roles in strategic
decision making. While Type II processing may enable strategic decision
makers to identify the pitfalls associated with one course of action over
another, excessive amounts of Type II processing may result in ‘paralysis
by analysis’. Such an outcome is particularly likely under conditions of
high decisional stress, in which the individual decision maker stands to
suffer potentially serious losses, irrespective of which course of action
is selected (Janis and Mann, 1977). Under other conditions, as noted
above, elaborative processing may prove dysfunctional by affirming or
 THE INDIVIDUAL IN THE STRATEGY PROCESS 215

bolstering a decision maker’s inappropriate worldview, rather than

challenging it.
Using the experimental method, in conjunction with cognitive mapping
techniques, it should be possible to discover more about the underlying
mechanisms of functional and dysfunctional Type I and Type II process-
ing strategies and begin to understand their impact on the structure and
content of decision makers’ mental representations of strategic problems.
In turn, these insights should help refine the use of cognitive mapping
techniques as a tool for intervening in the strategy process by specifying
the conditions under which such procedures are more or less appropriate.

CONCLUDING REMARKS

In 1982, Charles Schwenk published a paper entitled ‘Why sacrifice rigor

for relevance? A proposal for combining laboratory and field research in
strategic management’. The purpose of this paper was readily encapsu-
lated in its title: a plea for strategy scholars to ensure that their empirical
studies meet the twin imperatives of scientific rigour and applied rele-
vance in what is essentially a practitioner-oriented field. Twenty years
later, the experimental method continues to be a much under-utilized
mode of investigation within strategic management. Since the publication
of Schwenk’s (1982) paper, surprisingly few strategy researchers have
sought to exploit the combined strengths of laboratory experimentation
and field studies: rigorous control, accompanied by a fundamental con-
cern to address directly the most pressing issues facing practitioners, with
high levels of realism. In our ongoing work at Leeds we are explicitly
attempting to achieve this balance, and the evidence presented in this
chapter provides sufficient data for the reader to judge the extent to which
we have been successful, thus far.
As noted at the outset of this chapter, there can be no doubt that the
development and application of cognitive mapping techniques has
enriched considerably the field of strategic management over recent
years. The sheer range of topics addressed by the contributors to the
present volume attests to the fact that the mapping community en masse is
engaging with issues of fundamental concern to those responsible for the
realities of making strategies happen in businesses and public sector
organizations. However, in the absence of clear scientific evidence con-
cerning the efficacy of these procedures, both as a basis for uncovering
individuals’ mental representations of strategic problems and as a means
for intervening in processes of strategy making, there is a danger that our
enthusiasm to develop theory, methods and research that are perceived as
‘useful’ by those engaged directly in the craft of strategizing may outstrip
our capacity to deliver knowledge that is sufficiently robust to withstand
the critical scrutiny of our less sympathetic colleagues, academics and
practitioners alike.
216 THEORETIC AND METHODOLOGICAL ISSUES

At present, there is virtually no high-quality scientific evidence to

support the use of cognitive mapping procedures as a basis for interven-
ing in the strategy process. Such evidence as has been reported has been
largely confined to descriptive accounts of apparent ‘success stories’. In the
absence of additional supporting data, gathered systematically under
controlled conditions, these accounts provide little more than anecdotal
evidence. Ultimately, if cognitive approaches are to win wider acceptance
within the strategy field, and indeed the other organizational sciences,
there is a pressing need for researchers to move beyond the uncritical
application of mapping procedures in field settings per se. In order to
demonstrate the real value-added contribution of mapping methods, it is
necessary to employ these methods in conjunction with other approaches
in such a way that it is possible to disentangle cause and effect. Only when
we have submitted our methods to the rigours of this form of analysis will
it be possible to discover the extent to which mapping techniques can truly
provide useful, additional sources of insight into organizational strategy.

NOTE

1. The work reported in this chapter was funded by the UK Economic and Social Research
Council (ESRC) under Phase II of its ‘Risk and Human Behaviour Programme’ (Grant
number L211 25 2042). The contribution of Nicola Bown, Keith Glaister and Alan Pearman
is gratefully acknowledged.

REFERENCES

Anderson, N., Herriot, P. and Hodgkinson, G.P. (2001) ‘The practitioner-research divide
in industrial, work and organizational (IWO) psychology: Where are we now and
where do we go from here?’ Journal of Occupational and Organizational Psychology,
74: 391–411.
Axelrod, R.M. (ed.) (1976) The Structure of Decision: Cognitive Maps of Political Elites. Princeton
NJ: Princeton University Press.
Barnes, J.H. (1984) ‘Cognitive biases and their impact on strategic planning’, Strategic
Management Journal, 5: 129–137.
Bateman, T.S. and Zeithaml, C.P. (1989a) ‘The psychological context of strategic decisions:
A model and convergent experimental findings’, Strategic Management Journal, 10: 59–74.
Bateman, T.S. and Zeithaml, C.P. (1989b) ‘The psychological context of strategic decisions:
A test of relevance to practitioners’, Strategic Management Journal, 10: 587–592.
Bazerman, M.H. (1984) ‘The relevance of Kahneman and Tversky’s concept of framing to
organizational behavior’, Journal of Management, 10: 333–343.
Bazerman, M.H. (1998) Judgment in Managerial Decision Making. 4th edn New York: Wiley.
Chaiken, S. (1980) ‘Heuristic versus systematic information processing and the use of source
versus message cues in persuasion’, Journal of Personality and Social Psychology, 39: 752–756.
Chaiken, S. (1987) ‘The heuristic model of persuasion’, in M. Zanna, J. Olsen and
C.P. Herman (eds), Social Influence: The Ontario Symposium, Volume 5. Hillsdale, NJ:
Erlbaum. pp. 3–39.
 THE INDIVIDUAL IN THE STRATEGY PROCESS 217

Daniels, K., de Chernatony, L. and Johnson, G. (1995) ‘Validating a method for mapping
managers’ mental models of competitive industry structures’, Human Relations, 48 (9):
975–991.
Das, T.K. and Teng, B.-S. (1999) ‘Cognitive biases and strategic decision processes’, Journal of
Management Studies, 36: 757–778.
Eden, C. (ed.) (1992) ‘On the nature of cognitive maps’, Journal of Management Studies,
(special issue) 29: 261–265.
Eden, C. and Spender, J.-C. (eds) (1998) Managerial and Organizational Cognition: Theory,
Methods and Research. London: Sage.
Fiske, S.T. and Taylor, S.E. (1991) Social Cognition. 2nd edn. New York: McGraw-Hill.
Forbes, D.P. and Milliken, F.J. (1999) ‘Cognition and corporate governance: Understanding
boards of directors as strategic decision-making groups’, Academy of Management Review,
24: 489–505.
Goodwin, P. and Wright, G. (1997) Decision Analysis for Management Judgment. 2nd edn.
Chichester: Wiley.
Green, D.W. and McManus, I.C. (1995) ‘Cognitive structural models: The perception of
risk and prevention in coronary heart disease’, British Journal of Psychology, 86:
321–336.
Green, D.W., McManus, I.C. and Derrick, B.J. (1998) ‘Cognitive structural models of
unemployment and employment’, British Journal of Social Psychology, 37: 415–438.
Hodgkinson, G.P. (1997a) ‘Cognitive inertia in a turbulent market: The case of UK residen-
tial estate agents’, Journal of Management Studies, 34: 921–945.
Hodgkinson, G.P. (1997b) ‘The cognitive analysis of competitive structures: A review and cri-
tique’, Human Relations, 50: 625–654.
Hodgkinson, G.P. and Bown, N.J. (1999) ‘The individual in the strategy process: A cognitive
model’, paper presented at the Annual Conference of the British Academy of
Management, Manchester, UK, September.
Hodgkinson, G.P. and Herriot, P. (in press) ‘The role of psychologists in enhancing organi-
zational effectiveness’, in I. Robertson, M. Callinan and D. Bartram (eds), Organizational
Effectiveness: The Role of Psychology. Chichester: Wiley.
Hodgkinson, G.P. and Johnson, G. (1994) ‘Exploring the mental models of competi-
tive strategists: The case for a processual approach’, Journal of Management Studies,
31: 525–551.
Hodgkinson, G.P. and Thomas, A.B. (eds) (1997) ‘Thinking in organizations’, Journal of
Management Studies, (special issue) 34: 845–952.
Hodgkinson, G.P., Padmore, J. and Tomes, A.E. (1991) ‘Mapping consumers’ cognitive struc-
tures: A comparison of similarity trees with multidimensional scaling and cluster analy-
sis’, European Journal of Marketing, 25 (7): 41–60.
Hodgkinson, G.P., Tomes, A.E. and Padmore, J. (1996) ‘Using consumers’ perceptions for the
cognitive analysis of corporate-level competitive structures’, Journal of Strategic Marketing,
4: 1–22.
Hodgkinson, G.P., Bown, N.J., Maule, A.J., Glaister, K.W. and Pearman, A.D. (1998) ‘Dual
information processing in strategic decision-making? A theoretical framework and some
empirical data’, paper presented at the 18th Annual International Conference of the
Strategic Management Society, Orlando, USA, November.
Hodgkinson, G.P., Bown, N.J., Maule, A.J., Glaister, K.W. and Pearman, A.D. (1999a)
‘Navigating an uncertain world: Strategic cognition and risk’, ESRC Risk and Human
Behaviour Newsletter, Issue 5. Swindon: Economic and Social Research Council.
Hodgkinson, G.P., Bown, N.J., Maule, A.J., Glaister, K.W. and Pearman, A.D. (1999b)
‘Breaking the frame: An analysis of strategic cognition and decision making under uncer-
tainty’, Strategic Management Journal, 20: 977–985.
Huff, A.S. (1990) ‘Mapping strategic thought’, in A.S. Huff (ed.), Mapping Strategic Thought.
Chichester: Wiley. pp. 11–49.
Huff, A.S. (2000) ‘Changes in organizational knowledge production’, Academy of Management
Review, 25: 288–293.
218 THEORETIC AND METHODOLOGICAL ISSUES

Huff, A.S. and Schwenk, C. (1990) ‘Bias and sensemaking in good times and bad’, in
A.S. Huff (ed.), Mapping Strategic Thought. Chichester: Wiley. pp. 89–108.
Janis, I. and Mann, L. (1977) Decision Making: A Psychological Analysis of Conflict, Choice and
Commitment. New York: Free Press.
Johnson, G. (1987) Strategic Change and the Management Process. Oxford: Blackwell.
Kahneman, D. and Tversky, A. (1979) ‘Prospect theory: An analysis of decision under risk’,
Econometrica, 47: 263–291.
Kahneman, D. and Tversky, A. (1984) ‘Choices, values and frames’, American Psychologist,
39: 341–350.
Kahneman, D., Slovic, P. and Tversky, A. (eds) (1982) Judgment Under Uncertainty: Heuristics
and Biases. Cambridge: Cambridge University Press.
Lant, T.K., Milliken, F.J. and Batra, B. (1992) ‘The role of managerial learning and interpreta-
tion in strategic persistence and reorientation: An empirical exploration’, Strategic
Management Journal, 13: 585–608.
Maule, A.J. (1995) ‘Framing elaborations and their effects on choice behavior: A comparison
across problem isomorphs and subjects with different levels of expertise’, in J.-P. Caverni,
M. Bar-Hillel, F.H. Barron and H. Jungermann (eds), Contributions to Decision Research 1.
Amsterdam: Elsevier. pp. 281–300.
Maule, A.J. and Hodgkinson, G.P. (in press). ‘Heuristics, biases and strategic decision
making’, The Psychologist.
Meindl, J.R., Stubbart, C. and Porac, J.F. (eds) (1996) Cognition within and between organiza-
tions. Thousand Oaks, CA: Sage.
Mintzberg, H. (1983) Power in and around organizations. Englewood Cliffs, NJ: Prentice Hall.
Pettigrew, A.M. (1973) The Politics of Organizational Decision Making. London: Tavistock.
Pettigrew, A.M. (1985) The Awakening Giant: Continuity and Change in Imperial Chemical
Industries. Oxford: Blackwell.
Petty, R.E. (1995) ‘Attitude change’, in A. Tesser (ed.), Advanced Social Psychology. Boston:
McGraw-Hill. pp. 194–255.
Petty, R.E. and Cacioppo, J.T. (1986) ‘The elaboration likelihood model of persuasion’, in
L. Berkowitz (ed.), Advances in Experimental Social Psychology, Volume 19. New York:
Academic Press.
Porac, J.F. and Thomas, H. (eds) (1989) ‘Managerial thinking in business environments’,
Journal of Management Studies, (special issue) 26: 323–438.
Porac, J.F., Thomas, H. and Baden-Fuller, C. (1989) ‘Competitive groups as cognitive
communities: The case of Scottish knitwear manufacturers’, Journal of Management Studies,
26: 397–416.
Reger, R.K. and Huff, A.S. (1993) ‘Strategic groups: A cognitive perspective’, Strategic
Management Journal, 14: 103–124.
Reger, R.K. and Palmer, T.B. (1996) ‘Managerial categorization of competitors: Using old
maps to navigate new environments’, Organization Science, 7: 22–39.
Schwenk, C.R. (1982) ‘Why sacrifice rigor for relevance? A proposal for combining labora-
tory and field research in strategic management’, Strategic Management Journal, 3: 213–225.
Schwenk, C.R. (1984) ‘Cognitive simplification processes in strategic decision making’,
Strategic Management Journal, 5: 111–128.
Schwenk, C.R. (1985) ‘Management illusions and biases: Their impact on strategic decisions’,
Long Range Planning, 18 (5): 74–80.
Schwenk, C.R. (1986) ‘Information, cognitive biases and commitment to a course of action’,
Academy of Management Review, 11: 298–310.
Schwenk, C.R. (1989) ‘Linking cognitive, organizational and political factors in explaining
strategic change’, Journal of Management Studies, 26: 177–187.
Schwenk, C.R. (1995) ‘Strategic decision making’, Journal of Management, 21: 471–493.
Sieck, W. and Yates, J.F. (1997) ‘Exposition effects on decision making: Choice and confidence
in choice’, Organizational Behavior and Human Decision Processes, 70: 207–219.
Simon, H.A. (1955) ‘A behavioral model of rational choice’, Quarterly Journal of Economics,
69: 99–118.
 THE INDIVIDUAL IN THE STRATEGY PROCESS 219

Simon, H.A. (1956) ‘Rational choice and the structure of the environment’, Psychological
Review, 63: 129–138.
Smith, S.M. and Levin, I.P. (1996) ‘Need for cognition and choice framing effects’, Journal of
Behavioral Decision Making, 9: 283–290.
Spender, J.-C. and Eden, C. (1998) ‘Introduction’, in C. Eden and J.-C. Spender (eds),
Managerial and Organizational Cognition: Theory, Methods and Research. London: Sage.
pp. 1–12.
Staw, B.M. (1981) ‘The escalation of commitment to a course of action’, Academy of
Management Review, 6: 577–87.
Takemura, K. (1994) ‘Influence of elaboration on the framing of decision’, Journal of
Psychology, 128: 33–39.
Tesser, A. (ed.) (1995) Advanced Social Psychology. Boston: McGraw-Hill.
Tranfield, D. and Starkey, K. (1998) ‘The nature, social organization and promotion of man-
agement research: Towards policy’, British Journal of Management, 9: 341–353.
Tversky, A. and Kahneman, D. (1981) ‘The framing of decisions and the psychology of choice’,
Science, 211: 453–458.
Walsh, J.P. and Fahay, L. (1986) ‘The role of negotiated belief structures in strategy making’,
Journal of Management, 12: 325–338.
9
FA C I L I TAT I N G G R O U P
COGNITIVE MAPPING
OF CORE COMPETENCIES

Phyllis Johnson and Gerry Johnson

ABSTRACT

This chapter gives an account of the theoretical, methodological and practical

issues that emerged during the completion of a corporate research project. We
describe how expert facilitation overcame issues encountered during the research
process and suggest guiding principles for facilitating the surfacing of knowledge
across large multinational organizations. The chapter concludes by discussing the
impact that inappropriate assumptions about the nature of cognition can have on
the outcomes of both corporate interventions and empirical work.

DESCRIPTION OF THE RESEARCH PROJECT

A similar approach and group cognitive mapping method to that described

by Ambrosini and Bowman (Chapter 1 in this volume) was employed
during a programme of consultancy intervention with a UK-based major
multinational organization. At the time of the consultancy, this organization
was engaged in the process of rethinking its corporate strategy and the role
of its corporate centre. This conglomerate had businesses in consumer,
commercial and industrial sectors throughout the world.
In the past, the corporate centre had seen its parenting role as one of
offering guiding strategic principles and direction but with limited direct
involvement. The top team of the corporation wished to explore several
things: (1) why the most successful of their businesses were successful;
(2) where their core competencies1 lay within their organization as a
whole; (3) to what extent there were commonalties across business units
 GROUP COGNITIVE MAPPING OF CORE COMPETENCIES 221

in this respect; (4) how they could build on these competencies in terms
of their acquisition/divestment strategy and restructuring; and (5) what
the new role of the corporate centre should be in the light of the above
analyses. The impetus for the project was a need to explain clearly their
strategy to their major stakeholders; in particular, could they argue they
were building on synergistic threads and common competencies through-
out the organization?
The top team selected 21 of their high-performing businesses from dif-
ferent countries to take part in a series of group core competency mapping
events to be held in several locations around the world. At each workshop
a maximum of four, but usually three, business units were represented.
Before attending the workshop, the management team from each unit
was asked to complete a piece of work that would help identify what cus-
tomers believed them to be especially good at and which set them apart
from their competitors. The teams brought these data with them to the
workshop.
During the workshop, the teams were asked to take each of the most
important elements of their success (as identified by their customers) and
rate themselves and their top three competitors against each one.2 The
units had typically provided a list of five or six elements. When the teams
had plotted their ratings on a wall chart, the elements in which they
surpassed their competitors were extracted and given the title primary
success factors. The primary success factors that commonly emerged at this
stage were typically broad statements such as ‘customer service’ or ‘tech-
nical support’.
The teams from each business unit were asked to take the three most
distinct (i.e. the factors in which they perceived the greatest distance
between themselves and their competitors) of their primary success
factors and spend several hours using group mapping to explore where
they came from, and how they were achieved, maintained and embedded
within organizational life. Each team was asked to map at least two pri-
mary success factors (PSFs) over one and a half days. The mapping was
based on a simple process of placing post-it notes on very large blank
sheets of paper pinned to the wall of the team’s workspace. The teams
were asked to follow a ‘flowering-out principle’ where they would begin
with their PSF and place a set of post-its around it that described how
they were able to deliver it. Each of these post-its was itself surrounded
by explanatory post-its, and so on. A sample map is shown in Figure 9.1.
It is important to note that when the project began, these teams were
not continually facilitated. It was intended that they should be left alone
with one of the authors circulating between the four teams to troubleshoot
and problem solve.
Once this mapping was complete, the teams were asked to distil what
appeared to be their core competencies (in some ways similar to tacit rou-
tines) from their rich and detailed maps. At the end of each session the
maps produced by each team were retained, along with a record of the
 Manual
bility assembly
Flexi
ign
Des Variety of Low plant
manufacture utilization
resource Stock
dth s
Brea Change level
t production runs
uc Stock
r od ge
P n availability
ra

Ability to Small
respond sub-contract
Success transport
GSBOS Flexibility
Despatch
Customer flexibility
Service returns
Custom &
Logistics practice

Figure 9.1 An extract of a core competence map

 GROUP COGNITIVE MAPPING OF CORE COMPETENCIES 223

teams’ own analysis. These were then used by the authors and senior
executives at corporate centre to look for evidence of organization-wide
core competencies.

ISSUES ENCOUNTERED

After the first mapping session, it became apparent that the protocol
needed to be changed to ensure the company could get access to the sort
of detail it required: that is, to identify patterns and routinized ways of
behaving at the level of the ‘everyday’ that may combine and intermingle
to produce a core organizational competence. It appeared that the map-
ping groups selected from each of the high-performing business units had
difficulty carrying out the mapping task alone and needed close and
expert facilitation. This became apparent when one of the authors, a facil-
itator also experienced in cognitive mapping, sat with one group to assist
them when they encountered problems. At the end of the first day, the
data provided by this group far surpassed that which the other two
groups had been able to achieve working on their own, with only occa-
sional support of the overall co-ordinator. On the second day, it was
decided to give as full facilitatory support as was possible. The end result
was that the output from this first session was sufficient to go forward for
further analysis by the corporate centre. However, the protocol was
changed for future sessions. In subsequent sessions all groups were
closely facilitated throughout the two-day mapping event. Additional
facilitators and cognitive mappers were proposed by corporate centre
(drawn from human resources and consultative psychology departments)
and trained for the purpose.
In the context of the ideas developed in this chapter, the fact the proto-
col was altered and corporate centre were happy with the results is not the
main issue. What is important (and is the theme of this chapter) is that the
groups that were not facilitated were not producing information that
could be described as a cognitive map of the primary success factors they
had decided to focus on, whereas the facilitated groups produced data
that could be described as a cognitive map and therefore help to predict
influences on behaviour in the organization. There were several reasons
(which we explain here) to question the validity of the approach we were
intending to take and, more broadly, to question the validity of cognitive
data produced from non-facilitated groups.
The cause of the difficulties encountered by unfacilitated groups was
found to be multi-factorial. These difficulties hindered the groups’ ability
to produce core competency maps that had real utility and more broadly
highlighted problems with group-level cognitive mapping in general.
These have been collapsed into and discussed under two headings:
(1) mapping issues and (2) group dynamic issues.
224 THEORETIC AND METHODOLOGICAL ISSUES

Mapping Issues

The first, and perhaps most disabling, obstacle that the unfacilitated
groups faced was their inability to move beyond a discussion of the obvi-
ous when trying to access and describe the nature of their complex and
causally ambiguous core competencies. For instance, a primary success
factor that emerged in many of the groups was ‘customer service’. When
the groups were asked to explore this and follow the flowering-out pro-
tocol described earlier, they would not produce the kind of map shown in
Figure 9.1. Instead they found themselves unable to move much beyond
the level one branches shown in Figure 9.2.
When we consider the cognitive task we were presenting to our infor-
mants, it is not surprising that they are disabled in this way. They were
being asked to access their own long-term memory and bring into their
working memory (Baddeley and Hitch, 1974), any information that is
retrieved or triggered via the parallel spread of activation (Anderson,
1983, 1990; McClelland and Rumelhart, 1986), across their neural network
of nodes initially triggered by the stimulus ‘why is your customer service
outstanding?’ There are several reasons why this cognitive task can, if left
unfacilitated, produce only limited results. First, cognitive theorists sug-
gest that information is processed in long-term memory using parallel
distributed processing (see McClelland and Rumelhart, 1986, or
Sternberg, 1999, for a review). During this process, a stimulus will excite
and stimulate other associated nodes that contain linked information and
in this way long-term memory is searched and information retrieved. This
process is not serial, that is, one node stimulated after another in
sequence. Rather, multiple nodes are excited and stimulated at the same
time and in turn stimulate others. Hence the term ‘parallel distributed
processing’, designed to evoke the image of a flowering out of activation
from a central stimulus.
One of the major benefits of this form of information processing is its
speed, with long-term memory being searched in milliseconds. Part of the
ability to produce such speedy responses resides in the notion of familia-
rity or deeply ingrained paths. If stimuli usually spread activation to and
from one another, then their propensity to continue to do so increases with
each subsequent activation. For example, if the stimulus knife usually trig-
gers the activation of fork, and vice versa, then when either is presented as
a stimulus it will trigger the other – they are a familiar association.
Returning to our core competency mapping groups, we can be reason-
ably certain that when managers are presented with the stimulus question
‘why is your customer service outstanding?’, the response provided with-
out facilitation will be that which is familiar to the group, that which has
been triggered before and is more or less obvious to them. The clear diffi-
culty that this presents in core competency mapping is that the concept of
the core competency is not at all familiar to the informants. To explore
what their core competencies might be requires them to move beyond the
 GROUP COGNITIVE MAPPING OF CORE COMPETENCIES 225

A B

Success

E D

Figure 9.2 Example of a cognitive map with only level one

obvious (the taken for granted) and search their own structured knowledge
of the way they do business in ways that access the unfamiliar and the
unusual connections that they have perhaps never made before. It is ask-
ing the informants to do the impossible, to retrieve, verbalize and map
something they do not even consciously know that they know themselves.
By facilitating group processes, informants can be helped in this process
of discovery. The facilitator’s role is to work as a catalytic part of the cog-
nitive mechanism of the assembled memories available in the group. That
is, the facilitators can help identify and raise to consciousness connections
between concepts verbalized in the group. These verbalizations may have
been made by two different people or have occurred at different times but
are none the less connected and represent shared (socially constructed)
cognition. The facilitator’s role here is to help make the connections.
Alternatively, the facilitator can focus on individuals and encourage them
to say more, to explain in detail, to give examples of what they are trying
to explain. In this way, other group members are offered more potential
triggers for their own memory search. The facilitator’s role here is in pro-
viding alternative stimuli. In this exercise, encouraging informants to tell
stories was especially useful in provoking the response ‘That makes me
remember the time that …’, so that another series of possibly important
pieces of information and connections could be produced.
Although clearly rooted in cognitive theories of information processing,
this form of facilitation of the group’s collective construction of their cog-
nitive map (and in fact the whole skill of facilitation) is highly reminiscent
of counselling psychology. Standard protocols and practices can be bor-
rowed from this branch of psychology to provide a mechanism that
226 THEORETIC AND METHODOLOGICAL ISSUES

enables the facilitators to act as the central processing unit of the group’s
collective cognition. That is, in order to be able to make lateral connections
between verbalizations within the group and push individuals to say and
discover more, facilitators actually take groups through a process of self-
exploration and understanding. This same process of exploring your own
ideas, values, attitudes, memories and beliefs in order to come to under-
stand yourself better is mirrored in several schools of counselling practice
(see Woolfe, 1996, for a review), but particularly in Egan’s model of person-
centred counselling. Egan (1994) suggests that the counsellor actively
listens to the client’s story and acts as a mirror reflecting back to the client
what he or she seems to be saying (i.e. picking up on connections, repeti-
tions, themes, etc., that may not be obvious to the client). In this way, and
over a period of time, counsellors and therapists help individuals under-
stand themselves and their psychological life a little better. We say more
about active listening and general facilitation skills later in this chapter.
The second mapping issue we encountered is also linked to the task of
helping each individual in a mapping group to move beyond the obvious
and question his or her taken-for-granted ideas. Earlier, we described how
this might be difficult for informants simply in terms of the cognitive
logistics of the task. However, the mapping task we asked individuals to
complete presented difficulties on another level. When people are asked
to question what they take for granted to be true, uncertainty can be
created. Where there is uncertainty, associated anxiety can often be found.
Hirschhorn (1988) discusses the relationship between uncertainty and
anxiety in the workplace and describes the various ways in which indi-
viduals can seek to avoid, minimize and transfer such anxiety. In terms of
our mapping groups, we found that one mechanism that workshop
members employed to reduce uncertainty (and therefore anxiety) was to
avoid questioning taken-for-granted assumptions and ways of doing
things within the organization and to become defensive or dismissive
when challenged. In some cases, we found that this uncertainty avoidance
was led and positively re-enforced by the most senior member of the
group. This in turn made it difficult for more junior members of the group
to intervene and challenge what was being mapped and re-enforced
anxiety about challenging in the group.
With status as an impartial outsider, a well-trained facilitator can iden-
tify and break such patterns of behaviour before they become established
within a group. In addition (and this leads to our third point below), it can
be helpful for the facilitator to undertake the physical task of mapping, so
as to increase his or her ability to intervene when power games are taking
place. In the scenario where a powerful member is taking editorial control
of the mapping, the facilitator can repeatedly check with the rest of the
group to make sure a map entry reflects the views of others. Overall,
though, the simple awareness that challenging the taken for granted is
likely to induce a degree of anxiety and reticence can enable a facilitator
to recognize avoidance behaviour and work to overcome it.
 GROUP COGNITIVE MAPPING OF CORE COMPETENCIES 227

The third mapping issue that we faced concerned, not the elicitation of
the information to be contained in the map, but rather the construction of
the map. It became apparent when observing the groups that group
members were unable simultaneously to verbalize and map constructs.
There was a need for a facilitator to take a faithful note of their group’s
verbalizations and, with the co-operation of the informants and at appro-
priately spaced intervals, map the constructs noted. The unfacilitated
groups found that the maps that they produced showed little detail, not
only because they had not explored such detail in discussion, but also
because they found themselves unable to remember connections they had
made in conversation when they later paused to map.
The final issue that we encountered with unfacilitated groups was con-
cerned neither with elicitation of constructs nor map construction, but
with semantics and interpretation. The groups who worked alone on the
first day of our project produced maps that contained large amounts of
abbreviation and company jargon that were incomprehensible to an out-
sider or indeed those at the corporate centre responsible for content
analysing the maps. To compound this further, as our project was spread
internationally, we also had to contend with cultural differences and the
semantic difficulties that translation produced. Having a facilitator pre-
sent helped the group to master these problems as all abbreviations,
jargon and translation issues were dealt with at the time of mapping.

Group Dynamics Issues

In addition to specific mapping issues that the unfacilitated groups faced,

we were also aware from the beginning of our project of the problems that
can be presented by a reliance on individuals working well in groups.
There were a number of issues that seemed to be endemic throughout the
whole project. We believe they present themselves as problems in most
forms of group-work, especially in group cognitive mapping.
In our project, we were asking teams from high-performing business
units to explain what enabled them to out-perform competitors. The fact
that the teams were all high performing and had been recognized as such
by their selection to take part in the project led to problems. One conse-
quence was that many of the teams we worked with were somewhat self-
congratulatory. In itself this was not a problem. That is, as they were
high-performing units, they understood why we needed to understand
the basis of their success. However, on occasions, self-congratulatory ten-
dencies led informants to conclude that the reason for their success was
their own personal skills as individuals and as a management team. This
may indeed have been a significant contributor and a key element of the
core competencies. However, given that we know core competencies are
by definition likely to be causally ambiguous and arising from multiple
activities, it is unlikely to be the whole story. We found that one of the
228 THEORETIC AND METHODOLOGICAL ISSUES

important roles of the facilitator was to push the teams beyond

self-congratulation to examine other possible reasons and activities to
which they could attribute their competitive advantage.
Some of the teams that attended the series of workshops also had
amongst them at least one senior executive from the business unit nomi-
nated to attend. This was useful as it gave insight into each business unit’s
activities from a top team perspective. However, on occasions we wit-
nessed the dynamic in the team as a whole suffering as a result of the
presence of this senior member. Whether the senior executives courted it
or not, more junior team members behaved in a deferent fashion towards
them. In terms of the mapping exercise, this manifested itself in a reti-
cence to challenge ideas and opinions offered by senior team members, in
addition to a general wariness about disclosing some working practices
that, although giving the business unit the edge over competitors, were
either contrary to or a very liberal interpretation of company policy.
We found that storytelling about breaking rules was often the source of
what we began to refer to as ‘gold dust’. This was extremely rich data
about how individuals within the company wanted and were empowered
to ‘go the extra mile’ to win and keep customers. Again, the role of the
facilitator was important. The facilitator needed to manage the dynamic
to ensure that all views were heard and hierarchical deference did not
take over the team. It was important continually to reiterate and check
with senior team members that the team-working environment was a
‘safe place’ to disclose the ‘gold dust’ data.
As the project developed we began to find that, just as unfamiliar sto-
ries were frequently useful and to be encouraged by the facilitator, famil-
iar stories had to be treated with some caution. The group mapping
exercise was an ideal venue for ‘axe grinding’ and ‘hobby horse’ riding.
Individuals, or the team as a whole, were aware that, via the research
team, they had a direct line back to corporate head office. If the infor-
mants were left unfacilitated, old familiar issues could be woven into the
maps and take on central importance. Again, by using a facilitator to
question and challenge the group’s output, it was more likely that maps
were genuine representations of what the informants believed their core
competencies to be.
The final issue in terms of group dynamics that we repeatedly encoun-
tered during the completion of the project was associated with the length
of time that the teams spent mapping. The teams were asked to map a
minimum of two primary success factors (PSFs), with most attempting to
map three. This took at least a day and a half and was a tiring and at times
frustrating experience for those involved. One of the most important roles
that the facilitators took on was to maintain momentum, keep the teams
focused and aware of their progress through the entire process.
We also found that for the facilitators themselves simultaneously
managing the mapping task and managing group dynamics for hours at a
time was tiring and stressful. To help them overcome ‘facilitator burn-out’,
 GROUP COGNITIVE MAPPING OF CORE COMPETENCIES 229

we made sure there was always a ‘floating facilitator’ available to work

with groups for short periods helping the full-time facilitator pick up
momentum and maintain overall direction. The entire facilitation team was
trained together and met at the start and end of each day’s mapping to
exchange experiences and support. This helped the communication
between the floating facilitator and the full-time facilitators flow more eas-
ily. We developed code-words to indicate to the floating facilitator such
messages as ‘Go away we’re doing fine and you’re disrupting the dynamic’,
‘Help out here, this guy’s being difficult’, ‘We’re getting stuck and I can’t
see how to help them; a different approach from you might help’.
There are a number of points to be made to conclude this section, which
has focused purely on the issues that we faced in our own project using
group-level core competency mapping. First, we are not suggesting that
other unfacilitated group mapping techniques are fundamentally flawed,
but rather that without facilitation, they can run the risk of failing to pro-
duce the data that they could be capable of producing. Second, we found
there were two distinct categories that the majority of the issues we faced
fell into: (1) mapping issues and (2) group dynamic issues. In our view,
the first of these presents the greatest threat to the validity of the data
obtained from the mapping technique if conducted without necessary
facilitation. The third point is concerned with facilitation itself.
Facilitation is a skill that requires training, development and practice. The
limitations of untrained facilitation are heightened when combined with
a requirement to manage a cognitive mapping exercise properly. In our
view, such a task should not be undertaken lightly. Individuals who are
especially well suited are those who have experience in cognitive map-
ping and are confident in dealing with groups and individuals facing
difficult situations. We selected individuals with backgrounds in psychol-
ogy and human resources, and although these worked well for us, they
are not necessarily the only choices.

THE PRINCIPLES OF FACILITATION

There are several purposes of facilitation (see Redman, 1996; Westley and
Waters, 1988), but in general they can be summarized as follows:

• to get the best information out of the informants;

• to help them through the entire process;
• to manage the group dynamics; and
• to help the informants make sense of the information that emerges.

The most important skills that we focused on in our own facilitation and
training were: (1) active listening; (2) early diagnosis of problems in the
group dynamic; and (3) intervention in restoring balance to the group
dynamic. We also stressed as part of our training that it was vital to
230 THEORETIC AND METHODOLOGICAL ISSUES

remember that the purpose of facilitation is not for the facilitator to offer
his or her own view or in any way direct the group towards a particular
line of argument, rationale or outcome.
Active listening is a skill developed by those working in many fields of
organizational and one-to-one interventions. It involves the listener con-
centrating completely on what the individual speaking is saying, how he
or she says it and what he or she is not saying. Active listening is not
about a dialogue or a conversation, where part of the task of listening is
to construct your own response internally. Rather, it is about total listen-
ing, so if asked, the listener could repeat or reflect back (word for word or
paraphrased) what the speaker has said and meant to communicate. It is
important to remember that active listening is a skill and not a naturally
occurring attribute; therefore it requires practice.
Some of the common syndromes that are encountered in group facilita-
tion are described by Westley and Waters, (1988) as ‘multi-headed beast’,
‘feuding factions’, ‘dominant member’ and ‘sleeping meeting’. Each of
these is briefly described below, along with suggested interventions. As
will be apparent from the descriptions, interventions can either be direct
(verbalize the problem to resolve it) or indirect (facilitator manages the
dynamics without the need to verbalize the problem). Often, the use of
direct or indirect intervention is a matter of which method appears appro-
priate to a group’s own style and the facilitator’s natural inclination (i.e.
confrontational or not).
The ‘multi-headed beast’ syndrome involves group members talking
past one another. Here, no one is listening to anyone else. It appears that
group members are working to their own agendas. An intervention that can
work well in this situation is to give the group the shared agenda that is
missing: that is, to reiterate the task, where the group is and where it needs
to get to. In addition, it is important with groups displaying this syndrome
to ensure they are not being premature in their discussion of answers, as it
is often this rush to answers that causes the lack of built consensus.
The ‘feuding factions’ syndrome is the manifestation of unproductive
conflict between two sub-groups (or even two individuals). It is charac-
terized by repeated cycles of arguments, perhaps with a different topic
but a familiar theme. The feud might reflect ‘old stuff’ brought to the
mapping session. It is important to recognize that the role of the facilita-
tor is not to act as organizational analyst and solve endemic conflict.
Rather it is to see through the successful completion of the mapping task.
Consequently, interventions should be made with short-term goals in
mind. One direct form of intervention is to encourage other group
members to stop the feuding with their own intervention, for example:
‘We’ve heard a lot about this from you Craig and Jeff, perhaps we could
hear what Monica and Dave have to say.’ Alternatively, the facilitator
could get the group to take a coffee break and take both the feuding
factions to one side and ask them to leave their differences aside for the
sake of the task.
 GROUP COGNITIVE MAPPING OF CORE COMPETENCIES 231

The ‘dominant member’ syndrome is self-explanatory. It can be directly

managed by raising the issue of his or her domination with the individ-
ual concerned. This can be done in the group or perhaps during a break.
An indirect intervention would also be to strongly re-enforce contribu-
tions from other group members. This can be achieved by: making
repeated eye contact with them; placing your body in-between the domi-
nant member and the rest of the group; taking any control tools (i.e. the
post-its and pens) away from the dominant member; and, finally, being
verbally supportive when other members offer their view.
The ‘sleeping meeting’ syndrome is characterized by long silences, low
energy and no new ideas circulating in the group. Fear of volatile issues,
general hostility, a lack of understanding of the task, and fatigue can all
cause this syndrome. Depending on the cause of the syndrome, the facil-
itator can: take a break; reiterate aims and objectives of the task; reiterate
the rules of mapping (i.e. that it is a safe place), or isolate the hostility and
agree to handle it elsewhere. If all fails and the group are still ‘sleeping’,
it can be best to move on to another task rather than sacrifice the whole of
the group’s time to one blockage.
Overall the most important message about facilitation is that its
purpose is not for the facilitator to offer his or her own view or in any
way direct the group towards a particular line of argument, rationale or
outcome. It is to help the group find their voice and tell their own story.

OVERALL PROBLEMS WITH COGNITIVE MAPPING

In the field of managerial and organizational cognition (MOC) research,

methodology has always been a problem and a matter of compromise.
This results from the subject matter itself: human thinking and its relation-
ship to and explanation of human behaviour, which is captivating and
wonderfully rich but persistently illusive. The search for managers’
knowledge (either singly or in groups) can yield a powerful source of
explanation to help us make sense of organizational reality. However, to
state that the way someone thinks about something (assuming one has
confidence in the evidence) is likely to affect how that person behaves in
relation to that thing is a disarmingly simple link to make. Full apprecia-
tion and understanding of this link is a task that has engaged psycholo-
gists for over one hundred and fifty years.
This development in psychological understanding is largely the result
of specialized and consistent effort. For instance, cognitive scientists focus
on the activity of cognition, information processing, storage, retrieval and
use. Although they do not deny it is an important endeavour, they them-
selves are not necessarily interested in behavioural outcomes. Likewise,
discursive psychologists (e.g. Edwards and Potter, 1992; Harré and
Stearns, 1995) do not deny the cognitive hardwiring of the human mind,
but rather choose to envisage thinking as a social activity facilitated by
232 THEORETIC AND METHODOLOGICAL ISSUES

discourse – what people choose to say and hear. Social cognition (see
Augoustinos and Walker, 1995; Nye and Bower, 1996, for reviews), born
out of the advances of others, is trying to bridge the gap between cogni-
tion and behaviour. Through such specialization and focus, the field of
cognitive psychology as a whole is continuing to move forward. It is
important to note that no matter what their focus, none of the researchers
listed above would claim to have developed a foolproof method that
maps cognition or to argue that from such a map they can consistently
and accurately predict behaviour.
Management research has witnessed its own cognitive revolution
(Walsh, 1995). The managerial and organizational cognition literature has
chosen to explore the relationship between cognition and behaviour in the
context of socially complex issues in a complex environment. Many
varied tools and techniques have been used to access the cognitions. These
mapping techniques have been designed to approximate the contents and
or structure of managers’ mental models in varying degrees of detail. No
matter what the intended form of map, it is crucial to recognize that maps
are not actual paper or electronic transformations of managers’ thinking.
In other words they are not models of cognition but approximated displays
of elements of managers thoughts at a specific point in time, noted in particular
ways in particular environments that determine their format and in many
instances their content. The key to the quality of the end product of cog-
nitive mapping lies in the recognition of this and of two other associated
issues: (1) the quality of the theory of cognition the researcher employs to
frame his or her predictions about what is occurring when an informant
attends to and interprets a stimulus presented to him or her by a
researcher; and (2) the elicitation technique the researcher uses to access
an informant’s interpretation of the initial stimulus and in effect retrieve
information from long-term memory. Both issues have been raised and
discussed extensively by Eden (1992).
Embedded within these issues are the assumptions that were intro-
duced in the second section of this chapter, that is, (1) the assumptions
that researchers make about cognition and, leading from this, (2) the
assumptions they then make about cognitive mapping and (3) the nature
of the end product their methodology achieves.
In a discussion of the state of the art in cognitive science, Sternberg
(1999) highlights the potency of the connectionist school of cognitive the-
ory. For connectionists such as Anderson (1983, 1985) and McClelland and
Rumelhart (1986), knowledge is implicit in neural connections between
constructs, and working memory exists in the presence of a stimulus that
triggers a specific spread of activation amongst constructs. So in connec-
tionist models of cognition, what is important is not how constructs
are stored, but how they are activated. Therefore the context in which a
memory is triggered is crucial.
However, within the field of MOC research, there has not been a con-
sistent effort to ensure that mapping methods are able to cope with the
 GROUP COGNITIVE MAPPING OF CORE COMPETENCIES 233

issue of context dependency. For example, there has been research

published in the past that adopts a context-independent method. In each
of these studies (Bougon, 1983; Markóczy, 1995; Porac et al., 1989; Reger,
1988), the parameters of context are ignored as researchers make the
assumption that their results will be valid independent of the context in
which their data are gathered. The cognitive assumptions that would
appear to underpin this work are questionable, as is the validity of attach-
ing the label cognition or cognitive to the output of this work.
Alternatively, there are a range of context-dependent methods (Allard-
Poesi, 1995; Daniels et al., 1995; Eden, 1992; Johnson et al., 1998; Tomicic
and Hellgren, 1999) in which researchers do little to predefine the contents
of the map, in effect offering the informant a metaphorical blank sheet to
work with. By allowing informants to define their own constructs and
therefore the context in which they verbalize their current thoughts and to
do so in an unstructured format (i.e. imposing no pre-defined hierarchy
on the constructs elicited), such methods take account of the context
dependency of cognition.
In terms of the second criterion upon which mapping efforts can be
judged (the source of data), there is also a range of current practice. From
what we know of the context dependency of cognition, it would be diffi-
cult to support maps that have been crafted from documented archival
data such as memos, company reports and minutes of meetings
(Examples of this form of archival mapping are Barr et al., 1992; Fahey
and Narayanan, 1989; Fiol, 1989.) Such maps may tell us a lot about what
a company wants the outside world to think it is thinking, but little about
what individual managers are actually thinking about issues that are
likely to predict future actions of the company and the individuals within
it. (Eden, 1992 and Golden, 1992, 1997 extensively criticize the use of
retrospection in management studies.)
Arguably, better mapping practice (in terms of the source of data used
to elicit maps) recognizes the transient nature of cognition by accessing
thoughts directly from informants. Following this basic principle pro-
duces maps that can lay claim to a better approximation of thinking than
can those extrapolated from documentation of various forms. (Techniques
that gather data directly from informants are: Allard-Poesi, 1995; Bougon,
1983; Daniels et al., 1995; Eden, 1992; Jenkins, 1998; Johnson et al., 1998;
Walsh et al., 1988.)
Direct elicitation methods also suffer from the effect of audience. In
other words, they are likely to be influenced by the informant’s need for
social desirability: that is, to say what they think the mapper wants to
hear. The issues of audience and social desirability, although endemic in
much social science research, can, to some extent, be managed (Huber and
Power, 1985). It is quite probable that their effects can never be fully
removed. None the less, safeguards such as assurances of anonymity, con-
fidentiality and non-judgemental empathy projected during interview
should be employed to reduce their effects.
234 THEORETIC AND METHODOLOGICAL ISSUES

In summary, two common pitfalls in cognitive mapping are to employ

a method that is context-independent, and to use secondary data allow-
ing the informant little or no expressive adequacy. Overall, the comment
that we would make is that mappers should try to ensure that the method
they choose is genuinely fit for the purpose they intend to put it to. The
notion of ‘fit for purpose’ in mapping ought to take into account: cogni-
tive theory; the context in which the mapping is taking place; the source
of data; the philosophy being adopted; and, finally, the execution of the
mapping itself to ensure the informant receives facilitation in the telling
of his or her story.

CONCLUSION

This chapter has considered both competence mapping and cognitive

mapping in the light of an exercise conducted within a multinational firm.
It has concluded that the richness and value of a competence map is, in no
small way, a function of the extent to which the process takes into account
the difficulties of servicing and exploring aspects of organizational life that
are typically taken for granted. Moreover, that in so doing, there are likely
obstacles related to the behaviour of the group of people undertaking the
task. We have argued that it needs to be recognized that effective mapping
of this sort is unlikely to happen without skilled facilitation, and we have
provided guidance on what such facilitation might need to entail.
We have also argued that there are related concerns here with regard to
cognitive mapping. Specifically, the context dependency of cognition
requires methods that take account of such context.
If we bring together these two summary observations on the challenges
of competence mapping and of cognitive mapping, we must conclude a
third. If researchers or consultants choose to employ cognitive mapping
techniques as a means of understanding the competencies of organiza-
tions, they need to recognize the fragility of what they are exploring and
take pains to triangulate their findings. For example, in the study we have
reported here, the mapping exercise took place using a similar form for
21 business units with a conscious search for both convergence and diver-
gence of findings. Whilst replication of mapping of this order may not be
required, assumptions of useful findings based on one or two mapping
exercises may be misguided.

NOTES

1. A core competency is defined as an activity, a messy accumulation of learning com-

prising both tacit and explicit knowledge (Hamel and Prahalad, 1990) that provides funda-
mental customer benefit yet is difficult to define.
2. The exercise corresponds to that described as the ‘customer matrix’ by Ambrosini and
Bowman (2002).
 GROUP COGNITIVE MAPPING OF CORE COMPETENCIES 235

REFERENCES

Allard-Poesi, F. (1995) ‘Representations and influence processes in groups: Towards a

socio-cognitive perspective of cognition in organizations’, paper presented at the
3rd International Workshop on Managerial and Organizational Cognition, Glasgow,
14–16 June.
Ambrosini, V. and Bowman, C. (2002) ‘Mapping successful organisational routines’, in
A.S. Huff and M. Jenkins (eds), Mapping Strategy in Knowledge. London: Sage.
Anderson, J.R. (1983) The Architecture of Cognition. Boston: Harvard University Press.
Anderson J.R. (1985) Cognitive Psychology and its Implications. 2nd edn. Melon
University.
Anderson, J.R. (1990) The Adaptive Character of Thought. Hillsdale, NJ: Lawrence Erlbaum
Associates.
Augoustinos, M. and Walker, I. (1995) Social Cognition: An Integrated Introduction. London:
Sage.
Baddeley, A.D. and Hitch, G.J. (1974) ‘Working memory’, in G.H. Bower (ed.), Advances in
Learning Motivation, Vol. 8. New York: Academic Press. pp. 117–90.
Barr, P.S., Stimpert, L.J. and Huff, A.S. (1992) ‘Cognitive changes, strategic action, and organi-
zational renewal’, Strategic Management Journal, 13: 15–36.
Bougon, M.G. (1983) ‘Uncovering cognitive maps: The Self-Q technique’, in G. Morgan (ed.),
Beyond Method: Strategies for Social Research. Beverly Hills, CA: Sage. pp. 173–188.
Daniels, K., de chernatony, L. and Johnson, G. (1995) ‘Validating a method for mapping
managers’ mental models of competitive industry structures’, Human Relations, 48 (9):
975–991.
Eden, C. (1992) ‘On the nature of cognitive maps’, Journal of Management Studies, 29 (3):
261–265.
Edwards, D. and Potter, J. (1992) Discursive Psychology. London: Sage.
Egan, G.E. (1994) The Skilled Helper. 5th edn. Monterey, CA: Brooks/Cole.
Fahey, L. and Narayanan, V.K. (1989) ‘Linking changes in revealed causal maps and envi-
ronmental change: An empirical study’, Journal of Management Studies, 26: 361–378.
Fiol, M. (1989) ‘A semiotic analysis of corporate language: Organizational boundaries and
joint venturing’, Administrative Science Quarterly, 34: 277–303.
Golden B. (1992) ‘The past is the past – or is it ? The use of retrospective accounts as indica-
tors of past strategy’, Academy of Management Journal, 35 (4): 848–860.
Golden, B. (1997) ‘Further remarks on retrospective accounts in organizational and strategic
management research’, Academy of Management Journal, 40 (5): 1243–1251.
Hamel, G. and Prahalad, C.K. (1990) ‘The core competence of the corporation’, Harvard
Business Review, 72 (1): 77–86.
Harré, R. and Stearns, P. (1995) Discursive Psychology in Practice. London: Sage.
Hirschhorn, L. (1988) The Workplace Within: Psychodynamics of Organizational Life. Cambridge,
MA: MIT Press.
Huber, G.P. and Power, D.J. (1985) ‘Retrospective reports of strategic level managers:
Guidelines for increasing their accuracy’, Strategic Management Journal, 6: 171–180.
Jenkins, M. (1998) ‘The theory and practice of comparing causal maps’, in C. Eden and
J.-C. Spender (eds), Managerial and Organizational Cognition: Theory, Methods and Research.
London: Sage. pp. 231–249.
Johnson, P., Daniels, K. and Asch, R. (1998) ‘Mental models of competition’, in C. Eden and
J.-C. Spender (eds), Managerial and Organizational Cognition: Theory, Methods and Research.
London: Sage. pp. 130–146.
Markóczy, L. (1995) ‘States and belief states’, International Journal of Human Resource
Management, 6: 249–270.
McClelland, J.L. and Rumelhart, D.E. (1986) Parallel Distributed Processing: Explorations in the
Micro-Structure of Cognition Vol. 2: Psychological and Biological Models. Cambridge, MA:
MIT Press.
236 THEORETIC AND METHODOLOGICAL ISSUES

Nye, J.L. and Bower, A.M. (1996) What’s Social About Social Cognition? Thousand Oaks, CA:
Sage.
Porac, J., Thomas, H. and Baden Fuller, C. (1989) ‘Competitive groups as cognitive commu-
nities: The case of Scottish knitwear manufacturers’, Journal of Management Studies, 26 (4):
397–416.
Redman, W. (1996) Facilitation Skills for Team Development. London: Kogan Page.
Reger, R.K. (1988) ‘Competitive positioning in the Chicago banking market: Mapping the
mind of the strategist’, unpublished PhD thesis, University of Illinois at Urbana
Champaign.
Sternberg, R.J. (1999) Cognitive Psychology. 2nd edn. Orlando, FL: Harcourt Brace and Co.
Tomicic, M. and Hellgren, B. (1999) ‘A framework for analysing cognitive homogeneity and
heterogeneity in management teams’, paper presented at British Academy of
Management Conference, Manchester.
Walsh, J. (1995) ‘Managerial and organizational cognition: Notes from a trip down memory
lane’, Organization Science, 6 (3): 280–321.
Walsh, J.P., Henderson, C.M. and Deighton, J. (1988) ‘Negotiated belief structures and deci-
sion performance: An empirical investigation’, Organizational Behaviour and Decision
Processes, 42: 194–216.
Westley, F. and Waters, J.A. (1988) ‘Group facilitation skills for managers’, Management
Education and Development, 19 (2): 134–143.
Woolfe, R. (1996) ‘The nature of counselling psychology’, in R. Woolfe and W. Dryden (eds),
Handbook of Counselling Psychology. London: Sage. pp. 3–20.
10
USING A
KNOWLEDGE-BASED
SYSTEM TO STUDY
S T R AT E G I C O P T I O N S

Dale W. Jasinski and Anne Sigismund Huff

ABSTRACT

Knowledge-based computer systems (KBSs) provide tools for storing, manipulating

and analysing texts. These tools enable researchers to identify key concepts,
define typical relations among concepts, and dynamically explore alternative
definitions of these relationships. The distinctive advantages of such a system for
qualitative research are two-fold. First, KBSs provide a way graphically to
explore different aspects of the subjects’ mental models in search of patterns or
distinctive content. Second, the KBS allows researchers to develop their own
mental models and document for others how these structures relate to the data.
This chapter illustrates the use of such a system in a longitudinal study of strate-
gic decision making from an options perspective.

One cannot ordinarily follow how a researcher got from 3600 pages of field notes
to the final conclusions, sprinkled with vivid quotes though they may be.
(Miles and Huberman, 1994: 16)

Renewed interest in case studies and other qualitative research designs in

the management sciences (Bartunek et al., 1993; Eisenhardt, 1989; Larsson,
1993; Van de Ven and Poole, 1990)1 has elevated a number of methodo-
logical issues. First, these projects generate large data sets from field
notes, interviews and documents taken directly from organizational
archives (memos, presentation material, minutes of meetings, etc.). The
ability to manage text data is extremely critical to the success of such
research and computers are an obvious asset, but software is just now
238 THEORETIC AND METHODOLOGICAL ISSUES

becoming commercially available and the capabilities of these packages

are not widely known. Second, the aim of this kind of research is typically
theory development. It is important that the researcher establish a theo-
retical perspective, but maintain the ability to recognize new issues sug-
gested by the data that do not fit that theoretic framework (Miles and
Huberman, 1994; Weitzman and Miles, 1995; Yin, 1993). It is not clear how
computer tools interact with this double need. Finally, researchers are
increasingly being asked to provide the consumer of their studies with a
better understanding of how they came to their conclusions. Computer
tools can increase confidence in systematic analysis, but the criteria that
will convince a knowledgeable audience bear further discussion.
This chapter demonstrates that knowledge-based computer systems
(KBSs)2 are a critical tool for conducting case studies and other qualitative
research by addressing each of these issues. KBSs provide organizational
researchers with the ability to: (a) store large quantities of information ver-
batim; (b) manipulate the data in different ways to gain insights that man-
ual techniques would find extremely cumbersome; (c) identify structures
that appear to capture the mental models used by subjects, at a level of
detail appropriate to the researcher’s subject and ontological orientation;
and (d) document the research constructs that result from this analysis.
In this chapter we illustrate these benefits by drawing on a 15-month
study of strategic decision making in a high-technology firm. The purpose
of the chapter is not to discuss strategic decision making in detail, but
rather to provide the reader with an example of a substantive research
design that uses a knowledge-based computer program to assist in the
analysis of textual data. Accordingly, the remainder of this chapter is organi-
zed as follows. First, we give a brief background into mental models
as a key feature of qualitative research. Next, an overview of KBSs is pro-
vided. The third section of the chapter introduces the research we con-
ducted using one such system, ATLAS/ti™. The next section of the chapter
outlines specific coding decisions, and illustrates the kind of analysis pos-
sible with this knowledge-based system. We conclude by discussing some
methodological issues, including the key contributions knowledge-based
systems can make to qualitative research.

MENTAL MODELS

Mental models are the mechanisms whereby humans are able to generate
descriptions of system purpose (why a system exists), ... form (what a system
looks like), .. . functioning (how a system operates), ... observed system states
(what a system is doing), and . . . predictions of future system states.
(Rouse and Morris, 1986: 351)

The complicated ‘how’, ‘what’ or ‘why’ questions of organization partici-

pants and researchers require a framework. For example, the research
 USING A KNOWLEDGE-BASED SYSTEM 239

reported in this chapter centres on the decisions made within a single

high-technology firm that was trying to develop and choose among
options for future product development. How do individual decision
makers actually evaluate their options for the future? And what perceived
organization and environmental factors influence the process of strategic
option evaluation? Mental models must be developed by people within
the company to answer such questions.
Mental models help construct a sequence of related events, processes or
structures that connect concepts. To be understandable, each concept
must have a relationship to some other concepts. The researcher inter-
ested in understanding what a concept means to an informant must thus
seek out these connections. Meaning can be captured, at least in part, by
developing a semantic network generated by an exhaustive search of all
concepts connected to the subject in question (Carlson, 1993). For exam-
ple, an organizational function like ‘marketing’ is a concept defined by the
people who perform it, the organizations where those people work, the
managers of those organizations, the activities that those persons and
organizations are responsible for, and so forth, until all relations, and
relations of relations, have been exhausted.
In the process of trying to understand such a framework, the researcher
generates his or her own mental model. In theory development from a qual-
itative perspective the desire is typically to have a close correspondence
between the informant’s mental model and the researcher’s mental model,
but the researcher’s model will also include concepts made salient by the
research itself. In our view even the most inductive study cannot help but
carry some marks of the researcher’s past experience with academic study
and with other organizations. Further, the researcher cannot be just a pas-
sive recording instrument; the experience of conducting the research will
(and should) generate their own new connections. Further, the outputs of a
research project typically will be more valuable if the researcher’s mental
models edit aspects of the models gathered from the field. The researcher
should also be expected to highlight certain concepts and relationships as
worth special attention from those outside the context of study.
As this brief discussion illustrates, the task of describing and docu-
menting a mental model is likely to require much more than a couple of
boxes with connecting arrows. Accordingly, researchers can benefit from
tools that assist them in handling complex relationships. To minimize the
problems of information processing and to provide an efficient means by
which to document the models attributed to and derived from the data,
this research chose to use a knowledge-based computer system.

KNOWLEDGE-BASED SYSTEMS

The use of a computerized knowledge-based system (KBS) provides

the tools to create an ‘audit trail’ of the often implicit assumptions and
240 THEORETIC AND METHODOLOGICAL ISSUES

decisions made during the course of a research project. Miles and

Huberman argue that there are three steps in qualitative data analysis:
data reduction, data display and conclusion drawing/verification. Data
reduction is the ‘process of selecting, focusing, simplifying, abstracting,
and transforming the data’ that is ‘occurring as the researcher decides
(often without full awareness) which conceptual framework, which cases,
which research questions, and which data collection approaches to
choose’ (1994: 10, emphasis added).
KBSs require that these decisions are made more explicit than they other-
wise might be. The particular brand of software we used, ATLAS/ti
(Release 1.1E, Scientific Software Development, 1996), has a hypertext
interface for creating and browsing networks of related concepts.
Through the use of this KBS a researcher is able to identify concepts in
the context of use, explore relationships among concepts, gather and
assimilate knowledge (as represented by the text data collected) into a
model, and browse the resulting knowledge base in search of patterns or
unusual relationships. Though these activities are influenced by subjec-
tive researcher decisions, which we discuss in further detail at the end of
the chapter, the KBS imposes the discipline of considering all data
collected and maintaining the structure used throughout the project.
Further, it efficiently displays this structure for consideration and critique
by others so that they can follow the thought process that leads to a
study’s conclusion. The software also saves all data in a manner that
allows re-analysis under alternative mental models.
Conceptually, hypertext is analogous to the creation of 3 × 5 notecards
containing observations that are then stored simultaneously in numerous
fileboxes for later retrieval. ATLAS/ti extends this filing property by
allowing the researcher to establish any number of relationships among
individual notecards and fileboxes. Moreover, the researcher can modify
the number of relationships to any one card at any time. Thus, ATLAS/ti
allows seamless revisions of relationships to account for new develop-
ments that arise during the research or new insights from analysis. All
concepts affected by such a revision are then automatically updated.
Searching the notecards or browsing through the model via graphical
interfaces uncovers relationships and patterns among events and frame-
works. The set of data and the disclosed hypertext structure clearly
demonstrate the logical support for theory extensions and conclusions.
The advantage of such software over programs that merely enable text
to be parsed, coded, and retrieved based on coding is that KBSs support
the researcher in building theory. Weitzman and Miles point out that,
‘computers don’t think, and they can’t understand the meaning of your
qualitative data’ (1995: 18). What KBSs can do is assist the researcher to:
(a) build connections between various codes that enable the development
of higher-order classifications and categories; (b) visually document the
conceptual structure that fits the data (using specific user-specified rela-
tionships such as ‘is a kind of’, ‘belongs to’); and (c) generate attending
 USING A KNOWLEDGE-BASED SYSTEM 241

propositions concerning that structure, and, depending upon the design,

even test those assertions (Weitzman and Miles, 1995). The KBS enables
the researcher to ask questions like these:

• To what extent is a specific concept grounded in the use of other

concepts?
• Thinking about a particular organizational outcome, what conse-
quences might an actor anticipate by applying a specific strategy?
• What other roles does concept X play beside those identified in my
current analysis?
• Show me all text passages that directly reference phenomena P or any
other codes that play a role (via ‘consequence’, ‘condition’ or ‘strategy’
relations) with respect to P.
• What other codes have been used to describe these parts of the text?
• Which text passages provide evidence (or support, or contradict) the
statement in text passage T?
• What are the dimensions of category C?3

These are exactly the kind of questions we explored in our research project.

STRATEGIC OPTIONS EXAMPLE

The substantive questions driving the research reported in this chapter

grew out of our desire to further develop ideas from option theory as a
means of dealing with uncertain futures affecting current firm decisions.
Originally developed in finance literature, option theory was gaining
currency among management researchers as a way of capturing volatile
and uncertain decision contexts such as joint venture creation and
research and development funding (Bowman and Hurry, 1993; Hurry,
1994; Kogut and Kulatilaka, 1994; Sanchez, 1993). The study (more fully
reported in Jasinski, 1996) examined actions and decisions made by the
entrepreneur/founder and 24 other experienced managers of a high-
technology venture over the course of 15 months as they tried to identify
the direction of future product development.

THEORETICAL FRAMEWORK AND

RESEARCH DESIGN

Options theory posits that the firm’s existing resources and capabilities can
be positioned to capture future opportunities by either the expansion or
the contraction of investments across environmental domains. In essence,
the firm’s resources comprise a bundle of possible options for future strate-
gic choice (Bowman and Hurry, 1993). The interaction between a firm’s
existing investment, new investments in its capabilities and knowledge,
242 THEORETIC AND METHODOLOGICAL ISSUES

and decision makers’ perceptions of future environmental opportunities

create the option. Creation only occurs when decisions made today pro-
vide the firm with preferential (in relation to its competitors) access to
future opportunities. In these situations it is almost always necessary to
make some (relatively small) investments in the present in order to delay
the decision to make a more strategic investment in the future. This char-
acteristic is a key link back to financial options theory. Another important
connection involves the link between the value of an option and when it is
‘called’. However, the date is specified for a financial option, and this is
very often a difficult decision when working with strategic options.
The key point is that strategic decisions that have the character of
options cannot be adequately analysed with most models of decision ma-
king and strategic planning. Option theory differs from definitions of
scenario building (Schoemaker, 1995), for example, by focusing on the
strategic advantage of delayed investment because the future cannot be
adequately envisioned. In fact, standard tools of planning analysis (such
as discounted cash flow) typically lead to different recommendations than
option theory would provide.
At the time of the study the available literature on strategic options
(Bowman and Hurry, 1993; Hurry, 1994; Kogut and Kulatilaka, 1994;
Sanchez, 1993, etc.) was theoretic. The aim of the case study was to inves-
tigate options thinking empirically in an organizational setting. The organi-
zation chosen was not using options vocabulary, per se, but was
following suggestions by Hamel and Prahalad (1994), as presented by an
outside consultant. Hamel and Prahalad make very similar prescriptions
to those put forth in the strategic options literature, emphasizing, for
example, the importance of discovering ‘white space’ ideas – opportunities
that reside between existing products – based on business definitions. We
wondered whether this options orientation could be maintained, given
distinctive differences from other planning ideas. If options thinking did
persist, we wondered what organizational and environmental factors
would influence the development of options over time.

Research Site

The research site, given the pseudonym DataCorp, is an entrepreneurial,

privately held, high-technology firm, formed to design, assemble and sell
computer data communication products. The firm resembles hundreds of
entrepreneurial firms in the United States. It was founded in 1982 by a
small group of scientists and engineers who recognized and responded to
a market need for improved communication links among terminals,
mainframes and mini-computers.
The early years of the company were marked by tremendous excite-
ment generated by technical innovations and the risks and opportunities
of launching the new company. The company obtained venture capital
 USING A KNOWLEDGE-BASED SYSTEM 243

and outside board members to support further development, and

introduced its first commercial data communications product line of
mainframe peripheral controllers in 1984. In 1985 the firm generated its
first profit and positive cash flow. Revenues grew to an all-time high in
1989, and employment peaked at nearly 350.
In late 1989, however, the company’s original equipment manufacturer
partner unexpectedly developed proprietary communication technology
and dramatically reduced its orders from Data Corp for the following
year. In response, the company redirected its distribution strategy,
reduced its sales force, and began development of a new product line that
would facilitate communication between networked computers and
mainframe computers. These network products were still under develop-
ment when the market for its mainframe communication products shrank
in 1992. By the end of 1992, sales were almost half of their all-time high in
1989. The company reduced its workforce again and restructured into
three groups, one of which was in charge of developing a new product
that would operate on IBM’s new high-speed fibre-optic access architec-
ture. By the end of 1993, development was progressing with this product,
but sales were not yet being made in any significant amount. Another lay-
off was deemed necessary, this one even larger than the prior one in 1992.
It was at this point that DataCorp’s top management team decided to
engage in a strategic planning process markedly different from any used
in the past. Up to now, the planning process had been restricted to the
members of the top management team. The purpose of this new process
was to provide DataCorp with an opportunity to gain input from more
members of the organization. The logic behind this increased participa-
tion was that it would lead to more ideas and the development of a new
technology platform that would enable DataCorp to break its dependence
upon the mainframe computer. Consequently, the 7-member top manage-
ment team was joined by 18 other members of the organization to form
the ‘Gang of 25’ responsible for identifying promising ‘product plat-
forms’. There was a specific injunction against thinking about specific
product ideas in favour of ‘white space ideas’ that very much had the
character of strategic options. An initial group of five possibilities was
reshaped to three, then two, with a final choice just as a buyout offer was
accepted (to the surprise of many in the company). The company’s
demonstrated capacity to think in terms of new platforms appears to have
been part of the appeal of the company to the buyer.

Using ATLAS/ti to Store and Display Qualitative Data

Although a variety of organization documents were collected and

analysed following this interesting planning process, the following dis-
cussion focuses on a core set of interviews conducted at four critical
points with each of the seven members of the top management team.
244 THEORETIC AND METHODOLOGICAL ISSUES

These 27 interviews account for 6,895 lines of text (128 pages). The way
in which this data set was established and manipulated will be descri-
bed in some detail before turning to the analysis of option thinking we
accomplished.

data entry and sorting into ‘text families’

In order to utilize the ATLAS/ti text analysis software, each interview
tape was transcribed using a word-processing program. ATLAS/ti
required that each document be saved in an ASCII format (with file exten-
sion .txt) with approximately 50 characters per line. This is done to facili-
tate the display of text in the software and eliminates the need for
horizontal scrolling while reviewing the text.
To allow unambiguous reference within and among documents, each
file was then reformatted with every line numbered (these files are given
a .num file extension). Line numbering is used by the software to identify
the chunks of coded text and can be used to move from one location in a
file to another. Thus, once the coding procedure is begun, no changes
should be made to the text files.
The various documents entered into ATLAS/ti are classified as primary
texts. The software enables primary texts to be grouped into ‘families’,
with the possibility that a single document can be a member of multiple
families. This allows searching to be restricted to a specific subset of the
data. For example, each transcript from an interview is a member of the
‘interview transcript’ family, while only those conducted during the first
time period are members of the family ‘time period 1’. It is thus possible
to search for discussion of a particular topic (say the ‘WAN’ strategic
option) during the early days of the study by using the primary text family
‘time period 1’ as a filter prior to beginning a search of the interviews.
All family divisions divide the text into major ‘chunks’ that the
researcher might want to search independently. Thus, for example, the
initial coding scheme also identified files in terms of which individual
was being interviewed, since we knew we were interested in the develop-
ment of individual mental models. The families established for the project
are shown in Table 10.1.

data coding process

Once text has been transcribed and identified by text family, coding can
begin. The first step in the coding process is to select a passage of text (a
‘quotation’ within the vernacular of ATLAS/ti) for coding. Sentences or
paragraphs thus became the basic unit of coding. In no instance in the
research was a quotation comprised of less than one sentence. The longest
amount of text coded as a single quotation was three paragraphs (24 lines).
Coding quotations is a three-step process involving delimiting text pas-
sages, assigning one or more codes, and possibly annotating that passage
 USING A KNOWLEDGE-BASED SYSTEM 245

TABLE 10.1 Primary text families

Primary text family Purpose
Time period 1 All interviews, field notes and company documentation
pertaining to the first data collection time period: 1 August
through 21 September 1994.
Time period 2 All interviews, field notes and company documentation
pertaining to the second data collection time period:
22 September through 15 October 1994.
Time period 3 All interviews, field notes and company documentation
pertaining to the third data collection time period:
16 October through 15 November 1994.
Time period 4 All interviews, field notes and company documentation
pertaining to the fourth data collection time period:
16 November through 28 January 1995.
Transcripts Only the interview transcript files.
Individual (one family Each individual’s set of interview transcripts.
for each of the repeated
interviewees)
Field notes All word-processed field notes.
Company documents All scanned company documents such as memos and
presentation material.

with a memo. For example, since the core interviews in the data set
included the same basic questions repeated in each of the four critical
time periods, the question asked was used as an initial code.4 Inspection
of the data suggested that each of the questions asked during the inter-
view tended to be associated with certain subjects. These became a
secondary code. The questions, and the codes used to classify responses
to the interview questions, are shown in Table 10.2.
The memo function, briefly illustrated in the following section of
the chapter, is a very useful capability of ATLAS/ti. Memos make it
possible to annotate text and codes, which encourages and facilitates
capturing researcher insight throughout data entry and analysis. In the
initial coding period these memos are a useful way to note tone, distrac-
tions or other aspects of the data collection process that are not clear from
the transcript alone. When later considering alternative ways of combin-
ing and viewing the data, the memo feature of the software becomes a
repository for theoretical insight. Memos also offer a helpful way to
begin ‘writing’ analysis while the researcher is in the data collection
phase of a project.
After identifying quotations within the interviews that related to
specific questions, it was necessary to assign the secondary code(s) found
at the right of Table 10.2. In ATLAS/ti coding can be done either by select-
ing from a menu of existing codes or by creating a new code. The codes
shown in Table 10.2 are an initial set of codes established prior to begin-
ning the coding process. These categories were quickly elaborated.
246 THEORETIC AND METHODOLOGICAL ISSUES

TABLE 10.2 Initial coding scheme: responses to interview questions

Question Codes
What happens in your study group meetings? Process:
Methodology
Personality
Structure
Study Group:
Membership
What do you think your role is in the meeting? Individual
Founders
How do you think it’s going? Process:
Evaluation
What factors are important when you are evaluating these Decision Criteria
scenarios?
What factors are important to the executive staff? Decision Group:
Agreement
Who is going to make the final decision? Decision Maker
Right now, if you had to allocate 100 points among each of the Point Allocation
scenarios, how many points would you give each? Why?
How do you think the other members of the decision group would Decision Group:
allocate their points? Agreement
Evaluate the criteria on the Scenario Evaluation Form in terms of Decision Criteria
its relative importance to you in assessing the overall strength of
the scenarios to be presented next week.
Are there any other criteria that you feel will be used in making the Evaluation Sheet:
final decision regarding these scenarios? Completeness
What do you feel are the top three strengths (opportunities) and Option
weaknesses (problems) for each of the remaining scenarios? Decision Criteria
For the scenarios that were not presented to the board of directors, Decision Criteria
what factors made them less attractive than the ones selected
for presentation?
For each of the two remaining options, please describe the key Milestones
events that will unfold over the next year for each of these
scenarios. In other words, if you were to sketch out a timeline, say
by quarters, what would be the milestones, and what contingencies
or alternatives might exist at each of these points? In identifying
the timeline, are there any key resources (people, alliances,
financial, etc.) that need to be expended to pursue the option?

coding elaborations
As all qualitative researchers are quick to point out, the coding process is
highly iterative. Though many researchers now think that it makes sense
to begin with some theoretic framework (Eisenhardt, 1989), rather than
attempt to abandon all preconceptions in an attempt to ground codes
totally in the data alone (Glasser and Strauss, 1967), these initial ideas
almost always prove insufficient. Additional coding categories were
added in this study and augmented with a very useful additional step to
organize those categories.
 USING A KNOWLEDGE-BASED SYSTEM 247

coding families
At the very beginning of the coding process reported here, it became clear
that additional codes were needed. First, the strategic options defined by
the company (which changed over the course of the research, see Jasinski,
1996) were set up under the coding family ‘option’. Coding families oper-
ate in a similar manner to primary text families in that they can be used
as a means to sort or filter the data.
It was also necessary to treat individuals as a coding family. It was not
necessary to assign each quotation to the individual who authored it,
since this was already being tracked by the system. Rather, a new scheme
was needed to capture references to other individuals. For example,
during the interview with DP on 9/21, DP responded to a question con-
cerning decision group membership by stating:

166 Initially I wasn’t sure why BW was on

167 there except that JM really wanted BW to be
168 on there. We wanted LR to be on there because
169 in the strategic planning process she’s the most
170 effective at implementing things that have to get
171 done. You’ve got JM and RE who are pretty
172 intuitive and they don’t sit down and really
173 implement some of these things. They won’t take
174 notes and send out minutes and do that kind of
175 stuff. So LR had to be a part of it. (11:20)

Each of the individuals referenced (BW, JM, LR and RE) was coded so that
his or her role in the evaluation process could be tracked. These individu-
als were assigned either to the ‘top management team’ coding family or to
the broader 25-member ‘decision group’ family, depending upon their posi-
tion within the organization. (In addition to coding by person, the passage
as a whole was coded ‘decision group: membership’ as it pertained to the
rationale behind adding two members ([BW and LR] to the decision group.)

quotation identification
In the process of coding, each quotation from the text was also referenced
parenthetically with a two-number code. The prefix indicates which pri-
mary text file the quotation is from while the suffix indicates the quota-
tion number within that text file. Thus the above cited quotation reference
is (11:20), indicating that it is from primary text file number 11 and the
particular quotation is number 20 in that file. As data sets become more
248 THEORETIC AND METHODOLOGICAL ISSUES

voluminous, and coding decisions more complicated, this ‘audit trail’

procedure becomes more and more important.

stabilizing the coding framework

As shown in Table 10.2, it was initially decided to use a two-part classifi-
cation scheme. The prefix code would be the more general category (ques-
tion, process, decision criteria, option) while the suffix would be the more
specific category for that particular code. Thus, within the general code of
‘process’, specific factors such as personality, structure and methodology
were established.
These initial codes were derived prior to the actual coding of the docu-
ments and were based upon a review of the strategy literature as well as
from prior research at this site. For instance, the strategic planning process
followed the prescripts of Hamel and Prahalad’s Competing for the Future
(1994), which stresses the importance of identifying the organization’s
core competencies, developing and implementing a strategic intent state-
ment, and finding ‘white space’ of market opportunities. Accordingly,
each of these three concepts was initially established as a sub-code within
the code ‘decision criteria’.
After establishing the initial codes, a sample comprising the first two
interviews with RE was coded. This enabled the first author to gain famili-
arity with the software and gain insight into how to structure the system
to best enable the software to assist in the data analysis. From this sample,
it was decided to expand the coding system. Accordingly, the number of
codes increased from an original 63 prior to the sample coding to 74 after
coding the two interview transcripts. After the other two RE interviews
were coded, the number of codes increased to 79 and a level of comfort
was reached with the general structure. After five sets of interviews (total
of 20 transcripts) were coded, the number of codes stabilized at 98.
The software continues to allow the development of additional coding
categories, but each addition requires reconsidering all prior coding deci-
sions. Thus, it is important to work with the coding scheme until it is
fairly stable. One way to check the categorization system is to review all
quotations currently given a particular code; this is particularly helpful
in deciding whether or not a category needs to be subdivided, given the
purposes of the research project.
Once codes are established, the number of quotations in a particular cat-
egory is also of theoretic interest, but it needs to be pointed out that while
programs typically provide this capability, care must be exercised in inter-
preting frequencies. The fact that a particular code has more or less ‘hits’
than another code does not in and of itself imply significance. Rather, the
researcher must examine the underlying data structure (in this case the
quotation and perhaps the interview) to determine the meaning of the fre-
quency and prepare an explanation. In this research, for example, some
respondents were much more verbal than others. The issues that were par-
ticularly salient to them are thus ‘over-represented’ in the frequency count.
 USING A KNOWLEDGE-BASED SYSTEM 249

TABLE 10.3 ATLAS/ti relations

Relation Symbol Attribute
A is associated with B == Symmetric
A is a part of B [] Transitive
A is a cause of B >> Transitive
A contradicts B <> Symmetric
A is a B is-a Transitive
A is a property of B *} Asymmetric

establishing data relationships

After the first 20 transcripts were coded with the ‘stable’ coding structure,
data analysis was initiated using the network editor function within
ATLAS/ti. The network editor enables the researcher graphically to
depict relationships between the concepts discovered through textual
analysis. It facilitates a major step in data analysis, and also requires addi-
tional coding decisions.
All elements of the ATLAS/ti database (codes, memos, quotations and
primary text files) can be nodes within a particular network view. As fur-
ther discussed in the last section of this chapter, the relationships between
these nodes are both a product of the initial structuring of the data and the
result of additional researcher input. For instance, the code ‘option’ and
the designation of specific option alternatives is a fairly straightforward
categorization from the text. Also rather straightforward is the subse-
quent relationship established between these codes, indicating that each
of the individual options being evaluated by the firm has a predefined
connection ‘is-a’ to the family code ‘option’.
Other relationships are possible. Table 10.3 summarizes the types of
relationships that are predefined within ATLAS/ti, where A represents
the source node (code, memo, quotation or text) and B represents the tar-
get node of the relationship. The symbol indicates how the relationship
will be shown when viewing the data within the graphical mode (net-
work editor). The attributes indicate how the relationship chosen will
impact searches and data displays. For example, the transitive nature of
causality means that all data coded with a causal relationship will be
treated in the following way: If A is a cause of B, and B is a cause of C, A
will automatically be treated as a cause of C. And if a network view is cre-
ated (a capability of ATLAS/ti described in the next section of this
chapter) that requires examining A, relationships with both B and C will
be shown as well.
With this capability to relate codes to other codes, a further nesting or
hierarchy of codes is established that can then be used to examine
relationships between specific elements (codes, quotations or memos)
within the database. Remember, the relationships shown in Table 10.3
only specify the possible connections between data elements. Each specific
250 THEORETIC AND METHODOLOGICAL ISSUES

occurrence of such a relationship is considered a link. In other words,

while there is only one ‘is-a’ relation defined, there could be several links
between data elements using the ‘is-a’ relation.
One immediately useful output from relationship coding is a more
structured view of coding decisions. For the data set used as an example
here, the initial application of relationships yielded the coding structure
shown in Figure 10.1. Note that the higher-order codes begin on the far
left of the figure (Decision Criteria, Decision Group, etc.) with their sub-
ordinate codes branching out to the right.
The parenthetical numeric reference after the code name refers to the
number of quotations linked to that particular code (prefix) and the num-
ber of codes linked to that code (suffix). For example, at the time this
graph was made, the code ‘Decision Criteria’ had no quotations but 35
codes linked to it. The code ‘Decision Maker’ had 19 quotations but no
links to other codes.

network display of data structure

Once a relationship structure is established, analysis using the network
editor can begin. To illustrate, after the first 20 transcripts were coded, a
preliminary analysis of the role of competition in discussions about new
product development was carried out. The code ‘DC:Competition’ was
selected and a network view opened that asked for all quotations and
their codes to be displayed. This initial view is shown in Figure 10.2.
Rich with information, this view enables the researcher to retrieve quo-
tations in context by sizing the network view window and ‘shift clicking’
to display the quotation summarized by any given label. The researcher,
then, is never far from the data, and can easily switch between a detailed
examination of a specific piece of the data and a more holistic view of the
entire database.

initial analysis and data simplification: the initiation

of the researcher’s mental models
Although ‘rich’, the direct graphic display of data rarely lends itself to
direct explanation. Thus, after reviewing the relations shown in Figure 10.2,
and re-examining the quotations associated with each node, a new network
view was created that summarized the primary links in the data.
This in fact is a critical step in the process of analysis – which repre-
sents the move from models that are as close as possible to the mental
models of respondents (given the objectives of the specific research pro-
ject) to a more abstract picture that represent’s the researcher’s mental
model. Such a refined view is shown in Figure 10.3, which offers a more
comprehensible picture of the way competition seemed to be affecting
DataCorp.
 DC: Ahas {3–1}
DC: Build Business Case {1–1}
DC: Comfort {7–1}
DC: Communication skills {1–1}
DC: Company size {3–1}
DC: Distribution Channel {16–3}
DC: Competition {24–5}~ Process: Experience {18–4} –
Role Company History {47–2}
DC: Core Competency {15–1}
DC: Cost {4–1}
DC: Customer Need {2–1}
DC: Customer Service {8–1}
DC: Differentiatable position {6–1}
DC: Distribution Channel {16–3}
DC: Fit with Intent Stmt {17–1}
DC: Gut Level {7–1}
Decision Criteria (DC) {0–35} DC: Headcount {1–1}
DC: IBM Relationship {9–1}
DC: Impact {3–1}
DC: Initial {2–1}
DC: Mainframe Migration {13–1}
DC: Management Capability {1–1}
DC: Market Size {23–1}
DC: Market Timing {20–1}
DC: Marketing Capabilities {6–1}
DC: Move Forward {5–1}
DC: Novel {22–2}
DC: Overlap {1–1}
DC: Partnering {24–2}~ – DC: Distribution Channel {16–3}
DC: Pay Back {2–1}
DC: Presentations {3–1}
DC: Public {2–1}
DC: Risk {13–1}
DC: Success {13–1}
DC: Technical Capability {36–1}
DC: Understand Environment {9–1}
DC: White Space {17–1}
Bruce Walsh {14–3}
Darrayl Cannon {15–3}
Decision Group: Agreement {17–1}
Decision Group: Membership {44–1}
Decision Group: Size {2–1}
Decision Group {0–11} Dee Perry {20–3}
Jack McDonnell {56–3}
Linda Reed {26–2}
Paul Lilly {15–3}
Rich Carlson {23–3}~
Rory Enright {31–3}
Decision Maker {19–0}
DC: Distribution Channel {16–3}
DC: Competition {24–5}~
Process: Experience {18–4} –
Role: Company History {47–2}
Entry: Location {0–3} DC: Novel {22–2}
Process: Experience {18–4} – Role: Company History {47–2}
DC: Distribution Channel {16–3}
Entry: Timing {0–1} – DC: Competition {24–5}~
Process: Experience {18–4} –
Evaluation Sheet: Completeness {12–0} Role: Company History {47–2}
Evaluation Sheet: Inter-related {8–0}
Financial Status {3–0}
Bruce Walsh {14–3}
Founders {0–3} Jim Frugere {1–2}
Will Behl {2–2}
Andy Lemmon {1–1}
Bob Wald {3–2}
Bruce Walsh {14–3}
Darrayl Cannon {15–3}
Dee Perry {20–3}
Eddie Renoux {1–1}
Jack McDonnell {56–3}
Jim Crawley {3–1}
Individuals {0–18}
Jim Farver {3–1}
Jim Frugere {1–2}
Linda Reed {26–2}
Paul Lilly {15–3}
Rich Carlson {23–3}~
Rory Enright {31–3}
Upendra {3–1}
Veralynn {1–1}~
Vision {2–1}
Will Behl {2–2}

Figure 10.1 Hierarchical view of codes

 Competition Decision
and Mkt Criteria (DC)
isa DC: Technical
isa
~DC: Competition isa Option: LAN Capability
Switch
isa DC: Novel isa
== 4:37
reason for isa ==

DC: Differentiatable 7:12 isa

position Role: Company isa = > isa DC: Market Size
History 8:2 isa reason for
isa isa
10:7 => =>
reason for
Option: Storage
13:3 1
isa DC: Risk
reason for isa ==
13:8 isa ~DC: Partnering
DC: Cost
14:16
DC: Move Forward =>
=>
17:6
DC: Core DC: Distribution
Option: Desktop 17:11 Channel
Competency
17:13 Option: LAN
Option: Data
Center 18:12

DC: Company Size 19:26

DC: White Space
DC: Customer Need DC: Understand 20:10
Environment
20:12 DC: Market Timing
Role: Board of
Directors
Figure 10.2 Initial network view of competition
 USING A KNOWLEDGE-BASED SYSTEM 253

Role: Company
History

=>

~DC: Partnering Process: Experience

=>
DC: Distribution
[]
Channel
=> Entry: Timing
=> =>
~DC: Competition

=>
Entry: Location
Operationalize
==
Competition
DC: Novel

Figure 10.3 Initial analysis of the role of competition

The figure uses a new concept ‘Process: Experience’ (shown in the

middle of the figure) as an organizing device. It also specifies details
about entry into new markets that are not available in Figure 10.2. These
details grew out of examining the various nodes in that figure.
The process of analysis was facilitated by the note-taking capacity of
ATLAS/ti. As already mentioned, this ability to write while viewing the
data is a very powerful feature of ATLAS/ti. The memo icon entitled
‘Operationalize Competition’ (shown at the left in Figure 10.3) allows the
researcher to retrieve the details about how the various respondents
described competitors (size, number, where competition would likely
occur, etc.). The ability to record impressions about the data constantly
while maintaining the integrity of the data themselves expedites analysis
and explanation.

Options Reasoning

We now move to an abbreviated discussion of the substantive analysis

being carried out with the data set just described, though it should
be emphasized that other text files were also of great importance in the
project itself.
One focus of theory-building endeavours involved an effort to use and
possibly expand the conceptual structure of strategic option analysis.
Although, as noted above, DataCorp did not directly use the language of
options theory; interviews and other data satisfied us that this company
was often, though not always, operating in an options framework. Tables
of data (illustrated by Table 10.4) that were constructed using key con-
cepts in the options framework facilitate this effort. In toto, these quota-
tions make it possible to outline several new aspects of strategic as
254 THEORETIC AND METHODOLOGICAL ISSUES

TABLE 10.4 Examples of option-striking considerations

Person Response Code
BW We can’t hedge our bets too long, but its important for External:
us to be ready to jump into it when we do make a Time
decision. The passage of time will be one trigger. We
need to be doing everything that is flexibly useful during
that time, such as building alliances, watching for emerging
markets, etc. We need to use the time to keep progressing.
It could be as short as six to nine months.

GF We can do both for some period of time especially in an External:

unofficial capacity. You never want to walk away from Partner
a new product idea too soon. Again, we need to come
up with a compelling value-add for the one option. We
can continue to look into both until we have to commit to
one product or until it becomes forced upon us to do so.

Its probably going to be an outside event. For example,

with the 9XXX product, I can imagine DEC saying we
need it yesterday and all of a sudden it will begin to
snowball and gain momentum and a slew of decisions
will be made. But we want to keep our options open as
long as we can before we start making those decisions.

WB Well, I haven’t made up my mind yet as to where the Internal:

industry is going even after doing a lot of reading and Visionary
research. Seems like we are playing a game of musical
chairs with the scenarios we have examined so far; we
just threw out three chairs and I’m not sure that the
remaining two are really worth keeping.

We need a visionary to say this is where I think the

market is going to be in five years and take a guess at a
product and do it.

opposed to financial options. Since this work does not draw directly on
the capabilities of the ATLAS/ti program, however, we will not discuss
it further.
A more detailed analysis of the reasoning behind continued investment
in (or termination of) each option the company considered was also being
carried out. This analysis extensively relied on the software’s capabilities,
and will therefore be described in some detail. For purposes of the analy-
sis, the primary data set examined consisted of the interview transcripts,
though once again company documents and notes from critical meetings
were important as well. The following discussion draws primarily on the
interviews. As shown in Table 10.5, these documents were coded into
723 quotations.
These data were analysed in the following way:

• As noted earlier, each member of the top management team of the

organization (N = 7) was interviewed periodically over the course of
the planning process, (four interviews per person, with the exception
 USING A KNOWLEDGE-BASED SYSTEM 255

TABLE 10.5 Document statistics

Primary Document No. of No. of pages No. of
text no. name lines quotations
3 RE1111 555 10 61
4 RE1012 534 10 71
5 RE124 405 8 29
6 RE921 385 7 43
7 PL921 173 3 20
8 PL1012 171 3 17
9 PL1111 96 2 16
10 PL124 77 1 9
11 DP921 404 7 41
12 DP1012 86 2 20
13 DP1111 168 3 22
14 DP119 161 3 18
15 RC921 192 4 14
16 RC1012 70 1 12
17 RC1111 214 4 22
18 RC118 370 7 24
19 JM118 385 7 32
20 JM1111 224 4 26
21 JM1012 189 4 19
22 JM103 49 1 5
23 DC921 294 5 25
24 DC1012 178 3 21
25 DC1111 190 4 29
26 DC119 118 2 20
27 LR1012 323 6 24
28 LR111 151 3 24
29 LR21 733 14 59
Total 6,895 128 723

of LR, who was interviewed three times). These interviews form a total
population of 27 ‘explanation opportunities’, which presumably were
guided by more or less explicit mental models of the planning process
and option opportunities.
• For each interview, text passages pertaining to the decision criteria used
in judging the options currently being investigated by the company
were coded by decision criteria mentioned by informants (i.e. competi-
tors, market size, risk). When these factors were elicited from comments
directly regarding the study group responsible for developing a partic-
ular option, the passage was also coded to that group (i.e. WAN, LAN,
Data Centre, etc.). Thus, a matrix could be established that showed the
evaluation criteria used in judging each of the firm’s options.
• Next, each transcript was read to determine if the individual articu-
lated an ‘ideal future state’ for the company. In other words, without
referencing a particular option, did individuals express what they
thought the ‘ideal’ option should be and/or what the consequences of
that option would be for DataCorp?
256 THEORETIC AND METHODOLOGICAL ISSUES

• If the individual expressed relationships among decision criteria or

other events that would or did occur, those statements were classified
as ‘causal elements’. Words like ‘so’, ‘since’, ‘because’, indicated a
causal relationship between the elements while other statements were
linked in a temporal order such as ‘We’ll start from the Data Centre
side and work out towards the LAN.’
• A matrix was developed for each individual that compared these three
items (decision criteria, ideal future state and causal elements) across
each of the four time periods.

At this point it was hoped that a mental model could be constructed for
each individual’s explanation(s) of the information presented by each of
the study groups. Drawing on previous research on jury’s, medical team’s
and other group’s decision processes (Pennington and Hastie, 1986, 1988),
it was expected that the individual would construct an explanation or
story about each possible action (select, reject, success, failure) that could
be undertaken with respect to a given option, and that judgement would
be based on the best match between the constructed explanation and the
individual’s favoured option.
Unfortunately, it was not possible to pursue this expectation about knowl-
edge generation for all individuals, for each option, for each time period,
because the amount of explicit reasoning about each option by each indi-
vidual varied. In addition, as noted earlier, there was a great deal of variance
between subjects in terms of the amount of narrative obtained from the
interview, as shown in Table 10.5. The variance is partly a function of the
interview schedule (some respondents were pressed for time so answers
were brief), the amount of information gathered as part of each interview
(a second researcher also gathered data during the interview), and the
introverted personality of some of the members of the executive staff.
However, a composite explanation could be and was constructed
for each DataCorp option. This composite documents that knowledge
development about the company’s options in the first two time periods
was quite general, while in the latter two time periods, comments were
directed toward each option currently being considered, although some
comments still pertained to the set of options.
As an example of the rich analysis possible within this framework, one
option – the LAN Switch – is now presented in some detail. The detail
illustrates the kind of data available to qualitative researchers. It also illus-
trates the ‘embedded’ nature of the data, which will be discussed later as
a critical issue for research interpretation and coding reliability.

LAN switch decision

The LAN Switch option group exhibited a great deal of energy and enthu-
siasm in the Decision Group meetings and at their presentations to the
 USING A KNOWLEDGE-BASED SYSTEM 257

planning group. They tried to capture the imagination of the company as

evidenced by this statement from their leader:

If DataCorp wants to be something bigger than they are today, they have to
diversify and enter the world market. The mainframe is not a good market.
Other possible ways to go would be to acquire companies in the mainframe
market. There is certainly a way to do that. But the ideas that they proposed
in the mainframe market … [aren’t] exciting opportunities in terms of poten-
tial growth. DataCorp could certainly acquire some other companies in the
mainframe business. Become big internationally. They could do it. But still
the growth is questionable. Say you buy out a $100 million company, now
you have to become $200. How are you going to get from a $200 million com-
pany to $500? (30:3)

Feeding off the energy of their leader, UG, the group worked quickly
to develop ideas and by the middle of December were much farther along
in terms of definition than either the Data Centre group (which tem-
porarily disbanded at this point) or the Storage group. Yet the very nature
of the market demanded that the company make a quick decision on the
development of this option. One month later (18 January 1995), UG
expressed concern that the decision to give the group the go-ahead was
taking too long:

The problem is the time. Almost two months have passed. And in this market
two months is a long time to wait. Especially when the market has already
been recognized in the marketplace as an exploding market. (30:4)

An analysis of the evaluation of the LAN Switch helps to explain why the
decision was delayed and points out some key process issues that
impacted the ultimate decision to drop this option. Our interpretation and
subsequent discussion was facilitated by initially considering the positive
and negative evaluations of the LAN option separately.

LAN Switch: positive evaluation

One of the most frequently mentioned decision criteria that reflected a
positive evaluation of the LAN Switch option was the size of the potential
market and the fact that DataCorp’s core competencies matched the skills
needed to develop a product line (10:7, 19:26, 20:11, 28:11). For example:

This is clearly a major market scenario and our skills line up pretty well with
that. I think it plays well to the existing networking and switching skills. (20:11)

In addition to market size, the LAN Switch option was also perceived as
providing leverage of the company’s history:
258 THEORETIC AND METHODOLOGICAL ISSUES

Because we know something about inter-networking, we’ve been in that

environment. We’re coming along in understanding the details of the playing
field. Who’s doing what with what architectures. (3:37)

It is important to note here that these skills and experiences were primar-
ily developed for the line of business known as ‘inter-operability’, which
had been sold off almost one and a half years prior to the research into this
option. That point was noted by the respondents and is elaborated in the
discussion of the negative evaluation.
In addition to being ‘interesting’ (18:22) and providing a good migra-
tion from mainframe dependency (28:11), the LAN Switch also was con-
sidered to have the potential to define a differentiated market position for
DataCorp (3:10, 13:8) that would create value for the company:

The network switch, as I listened to that one before I got on the team, I sat
there and said ‘Aren’t there a ton of people in that position? How are we
going to be different?’ What I like about it was if we could create a position,
that would probably be the best leverage of the value of this company. But
that’s a big if. (3:10)

If such a position could be located, it would also help in DataCorp’s

efforts to find a suitable partner:

I think that there’s some significant players. If we could partner with one – in
other words, if we have enough specific skill competence in the switching
area that they don’t have – then we might have a better chance. (13:8)

JM summed up the positive evaluation in the following critique of the

LAN Switch:

LAN inter-networking is quite interesting. I think the proposal is interesting

in terms of if you can break into that there’s a lot of growth. This is clearly a
major market scenario and our skills line up pretty well with that. I think it
plays well to the existing networking and switching skills. (20:11)

This quotation by a single individual is actually very close to the mental

model that we constructed for the group as a whole, as illustrated by the
decision criteria used in the positive evaluation for the LAN Switch. This
model is summarized in Figure 10.4.

LAN switch: negative evaluation

Although the LAN Switch emerged as the option that seemed to provide
the company with the ‘white space opportunity’ (Hamel and Prahalad,
1994) it sought, ultimately it was rejected. The negative evaluations of this
option (presented in detail in Jasinski, 1996) are theoretically interesting,
 USING A KNOWLEDGE-BASED SYSTEM 259

DC: Mainframe
<> DC: Novel
Migration

DC: Differentiatable
reason for DC: Success
position =>

== =>
~DC: Partnering ~DC: Competition
reason for
Positive
reason for Evaluation
== =>
Role: Company
History DC: Market Size =>
=>
reason for
Focus =>

DC: Technical DC: Core

reason for
Capability Competency

DC: Move Forward

Figure 10.4 LAN Switch decision: positive evaluation model

in part, because they do not mirror the positive evaluation. When placed
beside a similar analysis of decision making around other options, the
study provides a very rich picture of option analysis to present to the field
of strategic decision making. The ATLAS/ti database permits comparison
of the structure of arguments across options, which makes it possible to
think about the influence of option content. It is also possible to compare
the development of arguments over time, which has led us to think about
the transfer of logic between options, and partial ‘collapse’ of reasoning
for a period of time into arguments that are not sensitive to content.
Clearly the KBS used is just a means to these ends. But if the findings
are to be believed, the impact of these means has to be considered. We will
therefore turn to how we evaluate the reliability of this study, and to what
extent the use of KBSs affects establishing reliability.

METHODOLOGICAL ISSUES

The use of qualitative methods is channelled by the researcher’s stance

towards what can be known and how it can be learnt. These complexities
of ontology and epistemology are too great to be reviewed here (see
Blaikie, 1993) except to remark that qualitative methods can be used in
many different kinds of studies and the standards of validity the
researcher hopes to meet depend very much on initial assumptions, even
if they are not made explicit.
260 THEORETIC AND METHODOLOGICAL ISSUES

This study takes an interpretive approach, which means that we

assume all descriptions of the world (organization members’ and
researchers’) are affected by the describer’s place in time and space, his or
her prior experience, and specific aspects of the immediate context. This
epistemic starting point is less demanding than more positivist
approaches in terms of establishing validity, and yet we want to commu-
nicate with others who are less inclined to an interpretive approach.
Thinking about the extent to which we can meet the standards of unbi-
ased design, systematic data collection and logical analysis that grow out
of more positivist approaches facilitates such communication. We also
believe such questions can facilitate interpretive work, even though an
interpretive approach denies such standards can be met and their pre-
scriptions are not essential for this approach.
The issue is basically one of trustworthiness:

How do you persuade your audience, including most importantly yourself,

that the findings of your inquiry are worth taking account of? What is it that
makes the study believable. . .? . . .This is not simply a presentational matter. ...
There are fundamental issues about the inquiry itself [that must be
addressed]. (Robson, 1993: 66)

The qualitative researcher trying to communicate with a broad audi-

ence has to think about establishing validity (‘whether the findings are
“really” about what they appear to be about’ [Robson, 1993: 66]) and reli-
ability (whether others would come to similar conclusions, given the data
collected). Even though we believe that there are no ultimate arbitrators
of these questions, we wanted to communicate with the reader about the
match between our subject of interest and the data collected, and the non-
arbitrary coding procedures and analysis applied to that data. To some
extent these are design issues that precede the use of a KBS; for example,
it makes sense to note the amount of time spent in DataCorp before co-
ding began to help establish the validity of conclusions drawn. A KBS is
very helpful in further addressing validity and reliability. Though the fol-
lowing discussion reflects an interpretive approach to these issues, the
categories of concern have applicability to qualitative studies with other
starting points.

Researcher Interpretation

The researcher’s focus of attention affects the data gathered, as well as all
subsequent research decisions. More specifically, to ATLAS/ti, that means
that researcher interpretation affects the way blocks of text are demarcated,
coding structures are defined and carried out, relationships are estab-
lished, networks are analysed and models are configured. Other decisions
are always possible. In this case, an interest in dealing with complex and
 USING A KNOWLEDGE-BASED SYSTEM 261

confusing strategic decision-making environments led us to note and work

with certain aspects of DataCorp. A researcher observing the very same
meetings, and even asking the very same interview questions, might carry
out these tasks quite differently if he or she was interested in the planning
situation as a means of interpersonal control and influence.
The roots of researcher influences on attention and decision making are
deeply embedded in socially situated past experience and current inten-
tions. There is no way these influences can be eradicated. There are only
other perspectives. On the other hand, just any interpretation is not
acceptable. The reader of a study deserves to know that researchers were
careful not to let initial mental models overwhelm the decisions that were
made to carry out the study. Further, the reader deserves to know that we
considered some of the most plausible alternative choices that might be
made. There are several ways in which this might be done.

respondent verification
One of the most obvious checkpoints on validity is to interact with the
sources of data themselves. Some methodologists feel that the respon-
dent’s interpretations of a given event or concept should dominate the
researcher’s since they are the source of the data. We take the position that
the subject’s perspective must be noted, but that it is not the only accept-
able or valid interpretation. Every point of view has its ‘blind spots’ as
well as its unique insights: for example, people within a context can be
more focused on detail than on pattern.
Some writers on methodology also worry about the possibility that
interaction between researcher and subject around constructs of interest
will itself bias the data. Though always something to be aware of, we
were not particularly concerned about this possibility, given that our sub-
jects were intelligent adults with far more at stake thinking about the
future of DataCorp than our study. Part of the study design therefore
included discussion of interpretation to date with people in the organiza-
tion. Notably, about seven months into the study the ideas behind strate-
gic option theory were discussed with the top management team. Their
responses are an interesting part of the data collected.

intercoder reliability
A frequently used method of assuring study reliability is to involve more
than one individual in the coding process, comparing coding decisions to
assure that significant overlap between coders exists. At the beginning of
this study it was assumed that a second coder would be used in this way.
However, the technical nature of much of the transcripts precluded an
efficient means of accomplishing this. An attempt was made to create a
translation scheme whereby a non-technical person could code the data,
but it was abandoned after it became apparent that there were too many
262 THEORETIC AND METHODOLOGICAL ISSUES

terms that had to be defined. The following two quotations are indicative
of the problems encountered:

Because we know something about inter-networking, we’ve been in that

environment. We’re coming along in understanding the details of the playing
field. Who’s doing what with what architectures. (3:37)
On the campus backbone I think our strength there is that it’s building off
the switching technology that we had developed for the Conductor. (13:5)

In both cases, the respondents are referring to the LAN Switch option but
in neither instance do they use that term. In quotation 3:37 the reference
is to ‘inter-networking’ and in 13:5 it is to ‘campus backbone’. Based on
field notes of the interview and familiarity with the technology, the first
researcher was able to make proper attribution. But it became obvious
that technical knowledge and a significant amount of time spent at the
research site (almost three years) provided a level of understanding and
insight about the people and the planning process that could not be eas-
ily duplicated.

multilectics
A more successful attempt to establish study validity and reliability
involved the use of more than one research interpreter. The people inter-
viewed and observed in this study were in general intensely involved in
their jobs, at a major turning point for their company. They had sufficient
interest in the subject of the study to provide extensive access, for which
we are very grateful. Their overall concerns, however, were not ours.
They could not be expected to be involved in some of the details of analy-
sis that interested us, though keeping in touch with their response to that
analysis is always important. Use of a second interpreter, more closely
aligned with the subject of interest and the ultimate audience of a large
portion of the research, is an important additional element of the research
design.
The study reported here is the dissertation research of the first author.
As supervisor, the second author was significantly involved at the early
stages of defining the study. During much of the data collection period the
two authors met weekly. Though less involved with data coding, for rea-
sons already described, the second author was again more involved as the
findings of the study took shape. Furthermore, for a portion of the data
collection period a third researcher attended meetings and interviews to
carry out a second research project. She also shared notes and discussed
ways in which the data could be understood.
The philosophical importance of this kind of relationship can be
described as ‘multilectic’ (Huff, 1981), a word that is expanded from
Hegel’s ‘dialectic’. From a multilectic perspective the inevitability of each
individual’s unique point of view is a strength in scientific inquiry. In a
 USING A KNOWLEDGE-BASED SYSTEM 263

world that is taken to be ultimately undescribable from one point of view,

more than one theoretical observer provides ‘different ways of logically
understanding an issue. Each individual perspective is expected to be
internally consistent, and to have gained from prior use in other settings
some confirmation that it is a viable way of looking at (at least parts of)
the world’ (Huff, 1981: 90).
ATLAS/ti was especially helpful in generating dialogue from multiple
perspectives. Interaction with graphic views, and the first researcher’s
mental models described within the system, provided the grounds for
discussion. A basic precept of multilectics is that total synthesis cannot be
achieved, and richer insight is provided by maintaining alternative
perspectives. The flexibility of the program made it possible to maintain
alternative interpretation of the data for further development and analysis.

data triangulation
A discussion that advances some, though not all, of the multilectic argu-
ment falls under the heading of ‘triangulation’. Denzin (1978) distin-
guishes triangulation by data source (which can include persons, times,
places, etc.), method (observation, interview document), researcher
(investigator A and B) and theory. Miles and Huberman (1994) add data
type (qualitative text, recordings, quantitative).
Despite the focus on one data set in this article, we were in fact very
much influenced by the importance of data triangulation for establishing
study validity. The goal of data triangulation is to pick sources that have
different biases and different strengths, so they can complement each
other. ‘In effect, triangulation is a way to get to the finding in the first
place – by seeing or hearing multiple instances of it from different sources
by using different methods and by squaring the finding with others it
needs to be squared with’ (Miles and Huberman, 1994: 297). Multilectics
expects points of view to be more intrinsically incommensurate.
Following the precepts of data triangulation, we obtained information
from public records (via Lexis searches of newspapers and trade maga-
zines), as well as private company written material such as memoranda,
brochures and financial statements. Moreover, information was obtained
from company informants both in public settings (attendance at company
meetings) and in confidence during private interviews. Using ATLAS/ti, it
was possible to compare comments made during interviews to the same
individual’s statements during group meetings and perhaps even to subse-
quent written documentation in company and/or public documents.
While these procedures are in line with classic, pre-computer, principles
of qualitative research design, the use of ATLAS/ti allowed much more
extensive data gathering. More important, it offered multiple views and
slices of the data. For example, data relationships could be viewed graphi-
cally, in text format or in numerical frequency distributions. Patterns
could be gleaned from one perspective and then pursued from another.
264 THEORETIC AND METHODOLOGICAL ISSUES

This was especially true in constructing the decision models for each of
the option decisions. Often the initial pass would be through graphical
examination of the data codes and quotations and would then proceed to
a review of the text passages in context of the transcript. To gain one sense
of the weight of a particular decision factor, the frequency of occurrence
of that criterion could be reviewed.
In short, ATLAS/ti provided us with what any computer support
system should be expected to provide: the capacity to more adequately
account for and deal with a complex data set. Ultimately, however, the
capacity of a KBS goes beyond helping to establish the trustworthiness of
a study, which is the subject of our conclusion.

CONCLUSION: DISTINCTIVE CONTRIBUTIONS

OF KBSs TO QUALITATIVE RESEARCH

The basic things KBSs accomplish merely take over jobs the qualitative
researcher has always done. These systems collect and code information,
and can carry out systematic searches of the database (e.g. for counter-
examples). Of course KBSs allow the single individual to accomplish
these things much more quickly. Given the important trade-off between
time and money, this in practice means that KBSs make much larger pro-
jects possible.
KBSs begin to make a more distinctive contribution to qualitative
research as the researcher starts to manipulate the data collected. Time is
the barrier that hinders the completion of qualitative research projects in
the collection, coding and systematic search phases. As the KBS aides data
manipulation, however, time is no longer the focus. KBSs are of greater
value at this stage of the research process because they make it possible to
consider different parts of the data simultaneously, thereby promoting dis-
covery of patterns that otherwise might be obscured by the weight of evi-
dence. At issue is the fact that the human brain is not very adept at keeping
many bits of information in mind at once (Miller, 1956), but is uniquely
suited to discovering patterns.5 The KBS thus extends a human strength.
Of equal importance for qualitative research, KBSs allow more alterna-
tives to be considered before a pattern is selected as meriting further
analysis. However skilled the individual researcher might be in establish-
ing such connections, KBSs again extend that reach. A system like
ATLAS/ti allows the researcher to carry out an infinite number of ‘what
if’ analyses. The potential contribution to qualitative research is as great
as the spreadsheet’s contribution to financial analysis.
KBSs are even more important as a tool to facilitate the development of
the researcher’s own mental models. Here the issue is that KBSs offer a
quick connect between researcher-defined variables and structures and
variables and structures more closely linked to the data. This possibility
puts the central characteristic of KBSs (that they can incorporate more
 USING A KNOWLEDGE-BASED SYSTEM 265

data, and iterate more alternatives than any human mind) in a new light.
The researcher skilled in interpretation and theory building in the past
typically had to move away from the data for a time in order to pursue
analysis. Data sets are distracting; they easily become clay around the feet
of the theorist. The KBS allows the needed distance, because it can ‘hide’
connections to the data. But the data are always only a click away.
The advantages of this capacity are enormous, and researchers are only
just beginning to discover how to use it. The potential is matched by that
derived from a second critical aspect of KBSs: their capacity to explore the
consequences of researcher structures. The search for counter-examples
has long been established as the hallmark of good qualitative research.
KBSs allow the introduction of a new – related but more important – hall-
mark. Given the capabilities of this new research tool, we propose that the
careful qualitative researcher should provide an audit trail back to the
data of the consequences of theoretically elevating specific concepts and
relationships. The KBS thus not only allows, but also demands a closer
connection between theory and data. As consumers of qualitative
research, we should now ask to see the larger picture; we can ask about
other concepts that are attached to those being proposed as central, those
that in the past have been left ‘on the cutting-room floor’.
We are just beginning to establish such an audit trail for the study sum-
marized in this chapter, but we can see the potential of the system we used.
For example, a key concept in the study involves timing. ‘When to call’ is a
central issue for strategic option theory; it was also a central issue for
DataCorp as it tried to find new sources of competitive advantage and was
ultimately approached with a buyout offer. ATLAS/ti allows us to explore
the ways timing is defined, and summarize simple links (of the kind
reviewed in the LAN option) between timing and other decision criteria. We
believe that the data can suggest more complicated connections. Starting at
the other end of the process, we also think that as we develop more compli-
cated theoretic ideas on our own, ATLAS/ti will allow us to test them more
adequately by returning to the complexity of the available evidence.
Having expounded on the virtues of KBSs, it is important to remember
in closing that qualitative analysis continues to be an art. Qualitative stud-
ies have always suffered from the problem of ‘too many trees to see the
forest’. The KBS proffers yet more, and new kinds of trees. The need for
artistry is thus greater than it ever has been. But KBSs are providing new
raw material that promises to elevate the status of qualitative studies as a
source of social science insight. Certainly these systems make qualitative
analysis more exciting than it ever has been.

NOTES

1. In addition to these published reports, the 1996 Academy of Management Meetings

featured a special session in the business policy and strategy section devoted entirely to
analytic case-based research.
266 THEORETIC AND METHODOLOGICAL ISSUES

2. While this chapter uses the term ‘knowledge-based system’ to refer to a generic type of
software, Weitzman and Miles (1995) use the terms ‘code-based theory-builders’ and ‘con-
ceptual network builders’. Commercially available programs include AQUAD, ATLAS/ti,
HyperRESEARCH and NUDIST.
3. Examples drawn from an electronic mail discussion with the Thomas Muir, developer
of ATLAS/ti.
4. Methodologically, the use of the same questions for each time period facilitated com-
parison of responses among individuals at a particular time period but also between time
periods for the same individual. Of course, this is just one aspect of the total data set, but a
particularly important part of the data being used as an example.
5. This point is frequently made by the director of the National Supercomputer
Applications Center at the University of Illinois in Champaign-Urbana.

REFERENCES

Bartunek, J., Bobko, P. and Venkatraman, N. (1993) ‘Toward innovation and diversity in
management research methods’, Academy of Management Journal, 36: 1362–1373.
Blaikie, N. (1993) Approaches to Social Enquiry. Cambridge: Polity Press.
Bowman, E. and Hurry, D. (1993) ‘Strategy through the options lens: An integrated view of
resource investments and the incremental choice process’, Academy of Management Review,
18 (4): 760–782.
Carlson, D. (1993) ‘Hypervision: A tool for semistructured data analysis’, unpublished work-
ing paper, University of Colorado at Boulder.
Denzin, N. (1988) Sociological Methods: A Source Book, 2nd edn. New York: McGraw-Hill.
Eisenhardt, K. (1989) ‘Building theories from case study research’, Academy of Management
Review, 14 (4): 532–550.
Glasser, B. and Strauss, A. (1967) The Discovery of Grounded Theory. Chicago: Aldine.
Hamel, G. and Prahalad, C.K. (1994) Competing for the Future. Boston: Harvard Business
School Press.
Huff, A. (1981) ‘Multilectic methods of inquiry’, Human Systems Management, 2: 83–94.
Hurry, D. (1994) ‘Shadow options and global exploration strategies’, Advances in Strategic
Management, 10A: 229–248.
Jasinski, D. (1996) ‘Using strategic option theory to provide sensible problem solving for
venture growth’, unpublished doctoral thesis, University of Colorado at Boulder.
Kogut, B. and Kulatilaka, N. (1994) ‘Options thinking and platform investments: Investing
in opportunity’, California Management Review, Winter: 52–71.
Larsson, R. (1993) ‘Case survey methodology: Quantitative analysis of patterns across case
studies’, Academy of Management Journal, 36: 1515–1546.
Miles, M. and Huberman, A. (1994) Qualitative Data Analysis. Thousand Oaks, CA: Sage.
Miller, G.A. (1956) ‘The magical number seven, plus or minus two’, Psychological Review,
63: 81–97.
Pennington, N. and Hastie, R. (1986) ‘Evidence evaluation in complex decision making’,
Journal of Personality and Social Psychology, 51: 242–258.
Pennington, N. and Hastie, R. (1988) ‘Explanation-based decision making: The effects of
memory structure on judgment’, Journal of Experimental Psychology: Learning, Memory and
Cognition, 14: 521–533.
Robson, C. (1993) Real World Research. Oxford: Blackwell.
Rouse, W. and Morris, N. (1986) ‘On looking into the black box: Prospects and limits in the
search for mental models’, Psychological Bulletin, 100 (3): 349–363.
Sanchez, R. (1993) ‘Strategic flexibility, firm organization, and managerial work in dynamic
markets: A strategic options perspective’, Advances in Strategic Management, 9: 251–292.
Schoemaker, P. (1995) ‘Scenario planning: A tool for strategic thinking’, Sloan Management
Review, 36 (2): 25–40.
 USING A KNOWLEDGE-BASED SYSTEM 267

Van de Ven, A. and Poole, M. (1990) ‘Methods for studying innovation development in the
Minnesota innovation research program’, Organization Science, 1 (3): 313–335.
Walsh, J. (1995) ‘Managerial and organizational cognition: Notes from a trip down memory
lane’, Organization Science, 6 (30): 280–321.
Weitzman, E. and Miles, M. (1995), Computer Programs for Qualitative Data Analysis.
Thousand Oaks, CA: Sage.
Yin, R. (1993) ‘Applications of case study research’, Applied Social Research Methods Series,
Vol. 34. Newbury Park, CA: Sage.
A N N O TAT E D
BIBLIOGRAPHY

Heidi M. Neck and Nardine Collier

Ackermann, F. and Belton, V. (1994) ‘Managing corporate knowledge

experiences with SODA and V.I.S.A.’, British Journal of Management,
5 (3): 163–176.

Two management science approaches are introduced that have been used for
the effective management of knowledge, toward the development of strategy,
planning and decision making in a wide variety of organizational settings. Both
approaches make use of computer technology to provide a flexible and inter-
active environment that incorporates a visual representation of the problem. The
SODA methodology is based on a multidisciplinary theoretical framework for
‘action research.’ Kelly’s (1955) Personal Construct Theory, from which the cog-
nitive mapping technique originated, is at the root of the framework. V.I.S.A. is
a decision support system that has been developed to assist in multiple-criteria
analysis. Underlying the approach is a conceptually simple, theoretically well-
founded and well-validated model – an additive-weighted value function. Case
studies illustrate the two approaches.

Andersen, C.R. and Paine, F.T. (1975) ‘Managerial perceptions and

strategic behavior’, Academy of Management Journal, 18 (4): 811.

The process by which the organization and the environment interact has
not been firmly established. Certainly, the strategy formulation process is a
crucial part of that interaction. Contributions to understanding organization-
environment interaction from the policy and organization areas are integrated
through a perceptual model of strategy formulation. Key inputs in the model are
managerial perceptions of environmental certainty and uncertainty and low and
high need for change. The resulting four quadrants present different kinds of
strategy formulation problems.
 ANNOTATED BIBLIOGRAPHY 269

Andersen, P.H. and Strandskov, J. (1998) ‘International market selection: A

cognitive mapping perspective’, Journal of Global Marketing, 11 (3): 65–84.

In the international market selection literature, selecting new markets is often

largely understood as an information-processing problem. The notion is that
managers require extensive market information to reduce decision-making
complexity. It is argued, however, that managerial cognition is more central to
international market selection processes, since recognition and evaluation of
strategic stimuli strongly affect the way this process is approached and exe-
cuted. Market environments are not unambiguous realities, but abstractions that
are given meaning through processes of selection, identification and screening.
To select among international markets, managers impose mental maps to
acknowledge markets proposed for mapping managerial decision making,
based on the so-called ‘means–end theory’.

Barnes, J.H., Jr (1984) ‘Cognitive biases and their impact on strategic

planning’, Strategic Management Journal, 5 (2): 129–138.

In strategic planning, human judgement is needed to interpret the facts and

determine their relevance for the future. Certain biases are inherent in any deci-
sion making as judgemental rules, or heuristics, are employed to simplify diffi-
cult mental tasks. Such heuristics as availability, hindsight, misunderstanding the
sampling process, judgements of correlation and causality, and representative-
ness can lead to large and persistent biases with serious implications.
Overconfidence results when people are insensitive to the tenuousness of the
assumptions upon which their judgements are based. While nearly everyone has
a desire for certainty, often some level of risk is inevitable. Experiments show that
a person’s cognitive limitations lead to simplifying the process of integrating
information when making even the most important decisions. Bias can be dealt
with by forcing planners to perform sensitivity analysis on their estimates.

Barr, P.S. and Huff, A.S. (1997) ‘Seeing isn’t believing: Understanding
diversity in the timing of strategic response’, Journal of Management
Studies, 34 (3): 337–370.

There is general consensus in the strategy literature that successful firms alter stra-
tegy to address changes in their environments and enact more favourable condi-
tions. Studies suggest that this adjustment is not always made in a timely manner.
A new study of six pharmaceutical firms suggests that multiple concepts associated
with environmental changes must be directly linked to organizational performance
270 ANNOTATED BIBLIOGRAPHY

before new strategies are initiated. The results emphasize the importance of stress
as a precursor to strategic response and have implications for the way firms
conceptualize ‘response’ when referring to significant changes in strategy.

Barr, P.S., Stimpert, J.L. and Huff, A.S. (1992) ‘Cognitive change, strategic
action, and organizational renewal’, Strategic Management Journal, 13
(Special issue: summer): 15–36.

Organizational renewal requires that a firm’s top managers make timely adjust-
ments in their mental models following significant changes in the environment.
The US railroad industry was selected as the subject of a study of whether
similar organizations in similar contexts differ in their ability to recognize
significant changes in their environments. The investigation focused on the
mental models of the top managers of a matched pair of firms. The need for
careful matching led to the selection of two Midwestern railroads, the C&NW
and the Rock Island. Analysis of longitudinal data from these railroads suggests
that renewal hinges not so much on noticing new conditions, but on being able
to link environmental change to corporate strategy and to modify that linkage
over time. The leaders of the C&NW not only recognize changes in their
environment, they also gradually change their mental models of how organiza-
tional performance is affected by the changed environment.

Bartunek, J.M. (1984) ‘Changing interpretive schemes and organizational

restructuring: The example of a religious order’, Administrative Science
Quarterly, 29 (3): 355–372.

A case study of a religious order is presented in an attempt to describe:

(1) processes through which interpretive schemes undergo second order change
(2) the relationship between second order change in interpretive schemes and
restructuring and (3) the role of the environment and of the organization’s
leadership in these processes. During the course of the study, the religious
order’s shared interpretive schemes, particularly its understanding of its mission,
were being substantially changed in terms of structure. It is suggested that major
changes in interpretive schemes occur through dialectical processes in which
old and new ways of understanding interact, causing a synthesis. The process of
change in interpretive schemes is in a reciprocal relationship with structural
changes. This relationship is mediated by the actions of organizational members
and their emotional responses to change. The way organization leadership
initiates or responds to alternate interpretive schemes restricts the type of change
in comprehension that can occur.
 ANNOTATED BIBLIOGRAPHY 271

Billman, B. and Courtney, J.F. (1993) ‘Automated discovery in managerial

problem formulation: Formation of causal hypotheses for cognitive
mapping’, Decision Sciences, 24 (1): 23–41.

Development of knowledge acquisition techniques known as automated dis-

covery systems has occurred in deep and narrow domains of knowledge.
Automated discovery is the generation of new knowledge by a computer
system on its own, without the help of another knowledge source. This study
describes research and validation of an automated discovery system for a
wide and shallow domain – business management. The system continues
recent advances in expert systems research that have enhanced cognitive
mapping, a problem formulation tool. The system perceives the behaviour of
distal variables in the environment through probabilistic cues-to-causality,
and generates previously unknown hypotheses by aggregating the probabili-
ties into a single criterion of causal relatedness. The system is validated
against the source code of a simulated managerial environment, and causal
relationships posited by decision makers experienced in the play of the
gaming simulator.

Bougon, M.G. (1992) ‘Congregate cognitive maps: A unified dynamic

theory of organization and strategy’, Journal of Management Studies,
29 (3): 369–389.

Independently, two dynamic approaches to organization and strategy have

emerged in fields traditionally confined to static methods. One approach uses
the cybernetic properties of collective cognitive maps to create a dynamic
theory of organization and social system change. The other approach uses the
hierarchic properties of collective cognitive maps to create a dynamic theory of
strategy. A dynamic cognitive approach makes organization theory and strategy
theory inseparable. The approach distinguishes between aggregate and congre-
gate collective cognitive maps. In this unified theory, the hierarchic and cyber-
netic aspects of collective cognitive maps combine with the cryptic aspect of
concepts and connections present in maps to further explicate the association
between organization and strategy. Cryptic labels are very different from equi-
vocal labels. Crypticality is a social phenomenon. The congregate cognitive
map is the social reality; there is no other underlying or deeper social reality
to discover.

Brown, S.M. (1992) ‘Cognitive mapping and repertory grids for qualita-
tive survey research: Some comparative observations’, Journal of
Management Studies, 29 (3): 287–307.
272 ANNOTATED BIBLIOGRAPHY

A study of business strategy was carried out in 86 organizations in the crop

protection industry. A multi-operational approach was used to enable validation
of data by triangulation. This provided an opportunity to conduct a comparative
evaluation of interactional investigative methods in a relatively controlled man-
ner. Some conclusions of the analysis are: (1) Cognitive mapping satisfies many
more process criteria than repertory grids. (2) The data that cognitive maps pro-
duce are less comparable than the data from repertory grids. (3) Splitting maps
produced under the constraints introduced for the purposes of the project
(actual versus ideal states) allows the respondent to focus on specific aspects of
the ideal state one at a time and in contrast to the actual. (4) Cognitive maps-
as-programs possibly do not represent the cognitive structure of the client’s own
problem conception. (5) Skeleton entries to cognitive maps (COPE) can be
preferable to full-string entry for certain purposes, such as the analysis of large
survey data.

Calori, R., Johnson, G. and Sarnin, P. (1994) ‘CEOs’ cognitive maps and
the scope of the organization’, Strategic Management Journal, 15 (6):
437–457.

A study considers chief executive officers (CEO) as cognizers charged with inte-
grating views in the top management team; a role that should require high
cognitive complexity, especially in diversified multinational corporations. A
methodology for studying top managers’ cognitive complexity is described and
then applied to a sample of 26 CEOs. The CEOs’ cognitive maps of the structure
and of the dynamics of their industry are analysed in terms of their degree of
complexity, in relation to the breadth of the business portfolio of the firm, its
geographic scope and the links the firms has with foreign parents. The results of
this exploratory test generally confirm the principle of requisite cognitive com-
plexity and reveal a new set of more precise hypotheses linking particular
dimensions of the scope of the firm with particular dimensions of the CEOs’
cognitive complexity.

Cossette, P. and Audet, M. (1992) ‘Mapping of an idiosyncratic schema’,

Journal of Management Studies, 29 (3): 325–347.

Among researchers interested in organizations and management, some of those

who use cognitive mapping share a common objective – to improve organiza-
tional action. Some intervene directly at the level of the organization while
others prefer to achieve this indirectly, by working at the individual level. Those
who work at the individual level rely on the emancipatory properties of a
cognitive map, which facilitates reflectiveness. The cognitive map and its
construction are characterized by the notions of: (1) natural logic, (2) schemati-
zation, (3) contextuality, (4) representation, (5) knowledge and (6) schema.
 ANNOTATED BIBLIOGRAPHY 273

A cognitive map is built from in-depth interviews. The idiosyncratic map of the
owner-manager of a small business contains 57 concepts, treated as variables,
and 87 links. By studying each variable and its link with other variables, it
becomes possible to consider many specific aspects of the owner-manager’s
vision of the firm. More than half the variables in the cognitive map are involved
in at least one loop homogeneity. Specific propositions with regard to such
research are drawn from the findings.

Daniels, K., Johnson, G. and de Chernatony, L. (1994) ‘Differences in man-

agerial cognitions of competition’, British Journal of Management, 5
(Special issue): 21–29.

It has been assumed that strategic decision making and implementation are
both achieved through managers’ sharing homogeneous cognitions of compe-
tition (e.g. Porac and Thomas, 1990). The assumption of homogeneity of
cognitions of competition was tested using a sample of 24 managers from the
offshore pumps industry. A variety of cognitive mapping techniques were used
and maps were compared using a self-rating methodology. The results indi-
cated that managers’ mental models of competition are diverse, rather than
homogeneous, but that this diversity increases as company and functional
boundaries are crossed.

Daniels, K., de Chernatony, L. and Johnson, G. (1995) ‘Validating a

method for mapping managers’ mental models of competitive industry
structures’, Human Relations, 48 (9): 975–991.

Cognitive mapping techniques refer to methods used to elicit the structure and
content of people’s mental models. There is growing interest in applying these
techniques to the study of managers’ mental models of strategic management
issues. A new method is suggested for mapping managers’ mental models of
competitive industry structures that is based upon recent developments in the
cognitive psychology of the categorization of concepts. The method is evaluated
with respect to its psychometric properties against the well-established, but
potentially cumbersome, repertory grid technique. The method is demonstrated
to have good validity by comparing it with the repertory grid technique.

Day, G.S. and Nedungadi, P. (1994) ‘Managerial representations of com-

petitive advantage’, Journal of Marketing, 58 (2): 31–44.

Managers use mental models of markets to simplify and impose order on complex
and ambiguous competitive environments and isolate points of competitive
274 ANNOTATED BIBLIOGRAPHY

advantage or deficiency. In a study of senior managers of 190 businesses, four

different types of mental models or representations of competitive advantage
were found, varying in the emphasis placed on customer or management
judgements about where and how competitors differ. These representations
were influenced equally by pressure points in the environment and choice of
strategy. The type of representation was also strongly associated with con-
strained patterns of information search and usage, raising the possibility that the
necessary simplifications and narrowing of perspective may come at the cost of
myopia and insensitivity to challenges from unexpected directions. There was
also a strong association between the completeness of the managerial repre-
sentation and relative financial performance, which supports related studies on
the profitability of a market orientation.

Diffenbach, J. (1982) ‘Influence diagrams for complex strategic issues’,

Strategic Management Journal, 3 (2): 133–146.

Influence diagramming is a qualitative approach to mapping the interrelationships

of complex strategic issues. Because of its qualitative nature, it is a simple desk-
top tool, requiring no computer software or particular expertise. The influence of
social, economic, technological and environmental factors can be assessed and
then used in conjunction with quantitative analytical techniques. The design and
interpretation of influence diagrams is illustrated and applied to a policy assess-
ment study of peak-load electricity pricing. Influence diagrams can vary in com-
plexity and detail, and can be refined and revised as needed. Initially, the
influence diagram can provide insight into relationships requiring further exami-
nation. It can uncover counter-intuitive relationships. It can be used effectively for
communication and can reveal differences in decision makers’ assessments.

Duhaime, I.M. and Schwenk, C.R. (1985) ‘Conjectures on cognitive sim-

plification in acquisition and divestment decision making’, Academy of
Management Review, 10 (2): 287–296.

Acquisition and divestment decisions usually involve complexity, ambiguity

and lack of structure. It is suggested that business decision makers may employ
some systematic and predictable simplification processes in defining such
poorly structured problems. It is further argued that four major biases, selected
from the organizational behaviour and cognitive psychology literatures, may
describe and explain the sorts of decision-making errors found in cases of ill-
considered acquisitions followed by unsuccessful turnaround efforts and even-
tually by divestment. The four biases are: (1) reasoning by analogy, (2) illusion
of control, (3) escalating commitment and (4) single-outcome calculation.
Examples from field research and the business press support the argument, and
implications for theory and research are discussed.
 ANNOTATED BIBLIOGRAPHY 275

Dunn, W.N. and Ginsberg, A. (1986) ‘A sociocognitive network approach

to organizational analysis’, Human Relations, 39 (11): 955–976.

A sociocognitive network methodology for investigating organizational dynamics

is presented. The methodology allows one to uncover and quantify differences in
the content of organizational reference frames. The resulting indexes of cognitive
content may be merged with standard sociometric data, creating a network matrix
that measures the sociocognitive connectedness of an organization’s participants.
When compared to other methodologies for investigating human consciousness,
the sociocognitive network approach has several advantages. The grid technique
is flexible, efficient, systematic and easily reproducible. Theoretically, it alleviates
the establishment of subjective meanings of organizational processes to partici-
pants before an attempt is made to measure them. Finally, the grid technique
permits one to monitor diverse organizational learning processes.

Dutton, J.E. (1993) ‘Interpretations on automatic: A different view of

strategic issue diagnosis’, Journal of Management Studies, 30 (3): 339–358.

Models of strategic decision making and environmental scanning typically

assume that decision makers diagnose issues actively, using conscious and
intentional effort to identify and to interpret potentially significant events,
developments and trends. When decision makers confront an issue that is novel
in content, about which persons have strong evaluations, where time pressure
and information load are intense, or decision makers exist in an organization
that has routinized means for dealing with issues, norms for consistency are
prevalent, or the organization has been successful, design and process inter-
ventions may be necessary to engage active strategic issue diagnosis.

Dutton, J.E. and Jackson, S.E. (1987) ‘Categorizing strategic issues: Links
to organizational action’, Academy of Management, 12 (1): 76–91.

A general conceptual framework is presented that is concerned with how the

meanings attached to strategic issues by decision makers are translated into
organizational responses. The model integrates an interpretive view of organi-
zational decision making with cognitive categorization theory, providing a
framework for understanding why organizations in the same industry respond
differently to the same environmental events and trends. It is believed that
labelling an issue as either a threat or an opportunity affects both subsequent
information processing and the motivations of key decision makers. It also is
argued that decision makers’ cognitions and motivations systematically affect
the processing of issues and the types of organizational actions taken in
response to them. The model helps researchers better understand the problem-
sensing and diagnosis process in decision making.
276 ANNOTATED BIBLIOGRAPHY

Dutton, J.E., Fahey, L. and Narayanan, V.K. (1983) ‘Toward understanding

strategic issue diagnosis’, Strategic Management Journal, 4 (4): 307–324.

A framework is presented to describe the strategic issue diagnosis (SID) process.

Strategic issues are poorly defined, emerging developments. Their diagnosis
requires interpretation, judgement and questioning traditional assumptions
about the business environment. The outcomes of SID serve as the basis for
future strategies. SID inputs include the concepts and beliefs of individual deci-
sion makers and their political interests, as well as the particular characteristics
of the issue facing them. Analysis of strategic issues will be characterized by
successive revisions of individuals’ judgements, and negotiations among indi-
vidual decision makers to reach a consensus on their assumptions about the
issue, its causes and effects, its expected outcomes, and the language and labels
chosen to characterize it. It is shown how the strategy formulation models of the
Boston Consulting Group’s growth-share matrix and the Profit Impact of Market
Strategies (PIMS) can be used as tools to aid decision makers in SID.

Dutton, J.E., Walton, E.J. and Abrahamson, E. (1989) ‘Important dimen-

sions of strategic issues: Separating the wheat from the chaff’, Journal of
Management Studies, 26 (4): 379–397.

Decision makers must use dimensions to sort the wheat from the chaff in the
field of potential strategic issues. The dimensions implied by 3 literatures and
dimensions generated by an empirical study are compared. A search of the 3
literatures identified 26 dimensions that differentiate strategic issues. Repertory
grid techniques were used to identify the range of attributes used to define the
‘meaning space’ for strategic issues in a sample of strategic decision makers at
the Port Authority of New York and New Jersey. Dimensions in the literature
overlap those in the empirical study; 13 dimensions accounted for 51 per cent
of the frequency with which all attributes were mentioned over all respondents.
However, these 13 dimensions exclude the one cited most often in the litera-
ture – immediacy. Also, marked differences in the relative importance of classes
of dimensions emerged in the sample relative to the literature. Issue content was
about 3 times as prevalent in the sample, action was about the same in both,
analytic characteristics were almost twice as prevalent in the literature, and
source did not emerge in the sample at all.

Eden, C. (1990) ‘Strategic thinking with computers’, Long Range Planning,

23 (6): 35–43.

Strategic options development and analysis (SODA) has been used extensively
by project teams working on strategic issues. A typical program involves a
SODA facilitator in capturing the views, knowledge and expertise surrounding
 ANNOTATED BIBLIOGRAPHY 277

the issue. It is important to include interviews with those who could sabotage
the outcome or those who are involved in implementation. Using cognitive
mapping as the core technique in work on major issues produces high levels of
ownership of the model constructed and of the agreed actions. In a SODA pro-
ject, it is imperative that the ownership of an action portfolio be increased by
the use of cognitive maps during the early stages of the project. As the project
progresses, the maps are aggregated into a single group map, which will not
belong to any member of the group. It becomes a device for facilitating negoti-
ation, synergy and creativity. Promoting a series of focused strategy workshops
can ensure coherence in strategy and create logical incrementalism.

Eden, C., Ackermann, F. and Cropper, S. (1992) ‘The analysis of cause

maps’, Journal of Management Studies, 29 (3): 309–324.

Cause maps are used to depict and explore the cognitive structures of members
of organizations. They are coded following many different conventions. A cause
map has several structural properties – the property of hierarchy and linkage. At
one extreme a map can comprise several clusters of nodes and links that are
each disconnected from one another. At the other extreme, a map can be highly
interconnected. One important analysis of emerging features relates to the
detection of clusters, where a cluster may be more or less separable from other
parts of the map. Within the context of analysis of directed graphs the existence
or not of feedback loops will be of interest. Nodes with a single link to other
parts of the map can be deleted to strip the map of detail. The effect of this
process of stripping out detail is to collapse the map to include only those nodes
with a domain score of three or more, thus retaining those nodes that sit at
branch points. As the analyses are completed, some nodes take on a different
status from others.

Fahey, L. and Narayanan, V.K. (1989) ‘Linking changes in revealed causal

maps and environmental change: An empirical study’, Journal of
Management Studies, 26 (4): 361–378.

The cognitive maps of the dominant coalition of one firm over time are explored.
Two key methodological issues warrant attention: (1) the process of deriving the
anticipated environmental elements in the causal maps; and (2) the derivation of
the causal maps themselves. The revealed causal maps of television receiver
maker Zenith over the period 1960–79 were traced. The twenty-year period was
grouped into five eras, based mainly on industry sales evolution. The structure of
the raw and reconstructed revealed maps indicated that decision makers were
cognizant of the complexity of the environment. The content of the maps changed
considerably from period to period. However, little interconnectedness between
elements of the macroenvironment and the industry was present in the maps. This
278 ANNOTATED BIBLIOGRAPHY

may reflect a difficulty on the part of decision makers to construct a complex and
integrated view of the environment. Results of the hypothesis-testing stage suggest
that the revealed causal maps evolve over time and sometimes in tune with the
environment.

Fiol, C.M. and Huff, A.S. (1992) ‘Maps for managers: Where are we? Where
do we go from here?’, Journal of Management Studies, 29 (3): 267–285.

Cognitive maps are graphic representations that locate people in relation to

their information environments. Cognitive maps are of potential interest to
managers because they are a means of displaying graphically the firm’s current
strategic position and because they hold the promise of identifying alternative
routes to improving that position. Broad strategic concerns of managers require
a portfolio of different kinds of cognitive maps. Cognitive maps encompass not
only specific choice points, but also information about the context surround-
ing the points. Maps can be products designed to remain relatively stable over
time or tools that people expect to modify. Current knowledge of cognitive
maps allows the identification of submaps of different aspects of cognitive
processes. These include: (1) identity submaps, (2) categorization submaps
and (3) causal and argument submaps. Taken together, these three provide a
tool kit for managers who must make sense of ambiguous and changing
environmental stimuli.

Fournier, V. (1996) ‘Cognitive maps in the analysis of personal change

during work role transition’, British Journal of Management, 7 (1): 87–105.

The way in which personal construct psychology and repertory grids can be
used to produce cognitive maps is illustrated, and psychological adjustment
during the transition from college to employment is analysed. Cognitive maps
are based on the principal component analysis of repertory grid data and
provide a visual representation of the way in which the subjects see them-
selves in relation to other people and in relation to their construct. The cogni-
tive maps produced by two graduates shortly after entry in employment and
six months later are analysed and compared in order to identify trends of per-
sonal change. The case studies show that the analysis of the cognitive maps
provides some valuable insights into the way in which graduates change their
construction of themselves and of their work environment during the period
of transition.

Gavetti, G. and Levinthal, D. (2000) ‘Looking forward and looking

backward: Cognitive and experiential search’, Administrative Science
Quarterly, 45 (1): 113–137.
 ANNOTATED BIBLIOGRAPHY 279

This paper examines the role and interrelationship between search processes that
are forward-looking, based on actors’ cognitive map of action–outcome linkages,
and those that are backward-looking, or experience-based. Results show that,
although crude, these representations still act as a powerful guide to initial search
efforts and usefully constrain the direction of subsequent experiential search.
Changing a cognitive representation itself can act as an important mode of adap-
tation, effectively resulting in the sequential allocation of attention to different
facets of the environment. This virtue of shifting cognitive representation, however,
may be offset by the loss of tacit knowledge associated with the prior cognition.

Ginsberg, A. (1989) ‘Construing the business portfolio: A cognitive model

of diversification’, Journal of Management Studies, 26 (4): 417–438.

Researchers investigating the relationship between diversification and perfor-

mance have been generally inattentive to the dominant general management
logics that corporate-level managers use to understand and manage strategy
diversity. In response to this, the use of the repertory grid is proposed. The grid
is a methodology adapted from the work of Kelly (1955) and other personal
construct theorists. As an approach to operationalizing the mental maps that
direct the management of strategic diversity, the business repertory grid appears
to have several important advantages. It is particularly useful in assessing the
ways in which top managers construe the corporate portfolio and manage
strategic diversity. Grid technique is flexible, efficient and systematic. It is also
highly reproducible, thus allowing researchers to validate or challenge the
results of previous studies. A main goal of the application of the repertory grid
to the study of strategic diversity is to measure the way in which top managers
perceive relationships among businesses in their firms.

Ginsberg, A. (1990b) ‘Connecting diversification to performance: A

sociocognitive approach’, Academy of Management Review, 15 (3): 514–536.

A socio-cognitive model of diversification is developed that bases the role of top

managers’ belief systems in the cognitive and behavioural attributes of corpo-
rate strategy. In contrast to more traditional approaches that emphasize the
economies generated by diversity and relatedness, this approach emphasizes
the creativity and flexibility associated with corporate-level decision-making
processes. The socio-cognitive perspective of strategy proposed makes three
important contributions: (1) It allows the examination of the ways in which
managers’ cognitive structures reflect the systemic properties of strategic posi-
tion. (2) It allows the examination of the ways in which shared cognitive struc-
tures reflect the behavioural dynamics of strategic decision-making processes.
(3) It allows the development of specific predictions and prescriptions regarding
the influence of top management teams’ socio-cognitive capacities on
processes of diversification and their economic outcomes.
280 ANNOTATED BIBLIOGRAPHY

Ginsberg, A. (1994) ‘Minding the competition: From mapping to mastery’,

Strategic Management Journal, 15 (Special issue): 153–174.

Cognitive approaches to strategy have examined the subjective nature of busi-

ness environments and competitive situations, but have failed to show how
managerial mental models lead to superior economic performance. In contrast,
resource-based views of strategy acknowledge the importance of managerial
skills in creating economic rents, but have not examined the processes through
which managerial cognitions lead to sustained competitive advantage. To
address this deficit, an analysis develops a socio-cognitive capability approach
that integrates cognitive and economic theories. The approach: (1) identifies
socio-cognitive foundations of differentiation and cost; (2) examines how these
foundations emerge from the process of strategy development; (3) explains how
group capabilities influence this process; and (4) shows how human and
organizational resources give rise to group capabilities.

Gripsrud, G. and Gronhaug, K. (1985) ‘Structure and strategy in grocery

retailing: A sociometric approach’, Journal of Industrial Economics, 33 (3):
339–348.

The objective market structure of a conventionally defined geographical market

area is contrasted with the relevant structure as it appears to individual retailers.
The impact of perceived competitive structure on the retailers’ choice of strat-
egy is examined. The interdependence in the strategy formulation of close com-
petitors is emphasized. Three hypotheses treating retail competition are
advanced. Consideration is given only to those grocery stores defined at the
four-digit level by SIC-data in one conventional market area – a small
Norwegian town encompassing 51 stores. Data are obtained from personal
interviews with store managers, and a complete sociometric mapping is made.
Results support the hypotheses and indicate that imitation or differentiation by
close competitors is related to whether intervening competitors of less impor-
tance do or do not exist. However, the findings indicate that an adequate assess-
ment of the subject is difficult using only objective market structure data.

Gronhaug, K. and Falkenberg, J.S. (1989) ‘Exploring strategy perceptions

in changing environments’, Journal of Management Studies, 26 (4):
349–360.

How a firm perceives its own strategy and that of its competitors is explored
using the forest products industry to represent the research context. Multiple
data sources include semi-structured interviews with top management, annual
reports, 10–K forms, articles from business periodicals and questionnaire data.
Great discrepancies in self-evaluations and competitors’ evaluations of the
 ANNOTATED BIBLIOGRAPHY 281

firms’ strategies were seen. The reported differences between actors and
observers indicate that perceptions are biased and that firms only partly under-
stand their environments. Also, it is observed that competitors’ perceptions of
each other are not necessarily mutual. This indicates that firms select competi-
tors based on their reality construction and that only a fraction of the potential
objectively defined competitors may be included in the firm’s evoked set of
competitors. This may be interpreted as firms selecting competitors as a basis for
strategy formulation and strategic actions.

Gupta, N. and Shaw, J.D. (1998) ‘Let the evidence speak: Financial incen-
tives are effective!’, Compensation and Benefits Review, 30 (2): 26–32.

It is an inevitable fact that most of us are motivated by money – some more so

than others. Behaviours that are rewarded are repeated, behaviours that are
punished are eliminated. When certain behaviours are followed by money, they
are more likely to be repeated. This means that employees will do the things for
which they are required; it also means that they will ignore the things for which
they are not rewarded. What management supports with money is given greater
weight than what management supports simply with words. In short, money
motivates by rewarding certain behaviours; it also motivates by showing people
what is valued in the organization – it provides a cognitive map of the path
people must take to succeed. There are good ways and bad ways to use finan-
cial incentives. Financial incentive programmes should: (1) be tied to valued
behaviours; (2) have good measurement systems; (3) use financial incentives to
supplement other rewards; and (4) have realistic goals. On the other hand,
financial incentive programmes should not: (1) be idealistic and unrealistic;
(2) equate rewards and punishments; or (3) violate employee expectations.

Hall, R.I. (1999) ‘A study of policy formation in complex organizations:

Emulating group decision-making with a simple artificial intelligence
and a system model of corporate operations’, Journal of Business
Research, 45 (2): 157–171.

A process theoretic framework using a brain–body analogy is used to concep-

tualize the interactions between the corporate and the policy-making systems of
an organization. The policy-making system is modelled as an artificial intelli-
gence containing behavioural rules that mimic the socio-cognitive and socio-
political logic used by policy-making groups in constructing a simple shared
cause–effect cognitive map from which to select policy options. The study pro-
vides the paradoxical finding that a better group cognitive map containing
causal relations more closely mirroring the real-world system will likely engen-
der more internal conflict. The study also provides insight into the process by
which key feedback loops of influences become overlooked in the building of
282 ANNOTATED BIBLIOGRAPHY

a group cause map. Implications for management training and practice and
future research are drawn from the analysis.

Hawes, J.M. and Crittenden, W. (1984) ‘A taxonomy of competitive retail-

ing strategies’, Strategic Management Journal, 5 (3): 275–288.

Top managers from a national sample of grocery store chains were surveyed
about adoption of generic products and the target market, marketing mix and
retailing mix strategies used by their firms in retailing generics. An 18-item
Likert-type scale was provided for answers. Using a multivariate clustering pro-
cedure, respondents were grouped according to their use of various strategies,
leading to the development of a taxonomy of four distinct competitive patterns,
including: (1) non-adopters, (2) aggressive initiators, (3) conservative reactors
and (4) submissive defenders. Aggressive initiators were early adoptors and
active promoters of generic products, expecting them to be a source of sales
growth from new customers. Of the three adopting groups, aggressive initiators
featured the largest selection and lowest prices for generic products.
Conservative reactors adopted generic products in response to competitor adop-
tion. Submissive defenders adopted generics at a low level, in fear of losing sales,
and devoted little effort to promotion. The aggressive initiator strategy was found
to yield superior performance.

Henderson, G.R. (1998) ‘Brand diagnostics: Mapping branding effects

using consumer associative networks’, European Journal of Operational
Research, 111 (2): 306–327.

Understanding consumer perceptions and associations is an important first step

to understanding brand preferences and choices. The ways in which cognitive
theorists would posit network representations of consumer brand associations
are discussed. Several empirical examples of consumer associative networks,
based on data from a variety of data collection techniques, are relied on in order
to demonstrate the tools available to the brand manager using network analytic
techniques. In addition to being grounded in theory, networks are shown to be
quite important to mapping an extensive array of branding effects, including:
(1) branded features, (2) driver brands, (3) complements, (4) co-branding,
(5) cannibalization, (6) brand parity, (7) brand dilution, (8) brand confusion,
(9) counter-brands and (10) segmentation.

Hodgkinson, G.P. and Johnson, G. (1994) ‘Exploring the mental models of

competitive strategists: The case for a processual approach’, Journal of
Management Studies, 31 (4): 525–552.
 ANNOTATED BIBLIOGRAPHY 283

Recently there have been a number of studies published that seek to further the
understanding of the competitive structures of markets. These studies have used
aggregate perceptual data in an attempt to uncover industry-level mental
models of business environments. It is argued that such studies are predicated
on the assumption that there are high levels of consensus within and among
organizations in a given industry concerning the bases of competition and the
positioning of particular organizations. A recent study revealed considerable
variation in terms of the nature of the cognitive categories elicited from the
participants and the overall complexity of their taxonomies relating to competi-
tive structures, both within and among the organizations.

Hodgkinson, G.P., Bown, N.J., Maule, A.J., Glaister, K.W. and Pearman, A.D.
(1999) ‘Breaking the frame: An analysis of strategic cognition and deci-
sion making under uncertainty’, Strategic Management Journal, 20 (10):
977–985.

The findings of two experimental investigations are reported into the efficacy
of a causal cognitive mapping procedure as a means for overcoming cogni-
tive biases arising from the framing of strategic decision problems. In Study 1,
participants in the post-choice mapping conditions succumbed to the fram-
ing bias, whereas those in the pre-choice mapping conditions did not. Study
2 replicated and extended these findings in a field setting. Taken together,
the findings of these studies indicate that the framing bias is likely to be an
important factor in strategic decision making, and suggest that cognitive
mapping provides an effective means of limiting the damage accruing from
this bias.

Hoffman, R.R., Shadbolt, N.R., Burton, A.M. and Klein, G. (1995) ‘Eliciting
knowledge from experts: A methodological analysis’, Organizational
Behavior and Human Decision Processes, 62 (2): 129–159.

The psychological study of expertise has a rich background and has recently gained
impetus in part because of the advent of expert systems and related technologies
for preserving knowledge. In the study of expertise, whether in context of applica-
tions or the context of psychological research, knowledge elicitation is a crucial
step. Research in a number of traditions – judgement and decision making, human
factors, cognitive science and expert systems – has utilized a variety of knowledge
elicitation methods. Given the diversity of disciplines, topics, paradigms and goals,
it is difficult to make the literature cohere around a methodological theme. The
types and subtypes of techniques are illustrated and a discussion of research cul-
minated that has empirically evaluated and compared techniques.
284 ANNOTATED BIBLIOGRAPHY

Ingoo, H. and Sangjae, L. (2000) ‘Fuzzy cognitive map for the design of
EDI controls’, Information and Management, 37 (1): 37–50.

Electronic Data Interchange (EDI) control design is ill structured and demands
consideration of the complex causal relationships among various components
of the controls, which may be broadly classified into formal, informal and auto-
mated types. Each of these can, in turn, be categorized as internal or external.
However, it is difficult even for EDI experts to predict the causal effects of one
control on another. In order to aid the design of EDI controls, the application of
a fuzzy cognitive map, EDIFCM (EDI-Control Design using a Fuzzy Cognitive
Map), was developed. Structural equation modelling was used to identify rele-
vant relationships among the components and indicate their direction and
strength. A standardized causal coefficient from structural equation modelling
was then used to create a fuzzy cognitive map, through which the state or
movement of one control component was shown to have an influence on the
state or movement of others. Thus, EDI auditors were able to enhance their
understanding of the causal relationship of controls and effectively design them.

James, L.R., Joyce, W.F., Slocum, J.W., Jr. and Glick, W.H. (1988)
‘Organizations do not cognize’, Academy of Management Review, 13 (1):
129–132.

Glick’s (1985) discussions of organizational climate (OC) contain some incon-

sistencies. He first defines OC as a non-psychological organizational variable
that is different from a shared assignment of psychological meaning. Then he
describes OC as a ‘shared meaning’ or a ‘collective attitude’. Glick also focused
attention on the need to obtain accurate descriptions of organizational charac-
teristics in order to assess OC. However, if OC is defined in terms of shared
meanings, then many of Glick’s recommendations need rethinking. In a
response, Glick contends that the disagreement between his views and the
views of James et al. rests on the differences between two conflicting definitions
of organizational climate. While Glick included shared meaning as a compo-
nent of organizational climate, this does not imply that organizational climate
is psychological. The individualist and macroclimate researchers disagree and
should work together to clarify definitions and resolve differences.

Jenkins, M. and Johnson, G. (1997) ‘Entrepreneurial intentions and out-

comes: A comparative causal mapping study’, Journal of Management
Studies, 34 (6): 895–920.

A study uses a causal map methodology to consider the contrasts between

entrepreneurial intentions and outcomes. In evaluating a series of propositions
drawn from the extant literature, the study finds that the elicited causal maps are
 ANNOTATED BIBLIOGRAPHY 285

consistent with contrasts in entrepreneurial intentions, but not outcomes. The

study develops a series of output propositions suggesting that entrepreneurial
outcomes are associated with causal maps that connect the internal operations
of the business with the external environment. This implies that entrepreneurial
success may be the result of intuitive systems thinking in which connections are
made between the environment and the internal operations of the business.

Jones, S. and Sims, D. (1985) ‘Mapping as an aid to creativity’, Journal of

Management Development, 4 (1): 47–60.

Three techniques for eliciting creative problem-solving ideas include: (1) brain-
storming, in which problem solvers generate as many ideas as possible about a
problem without evaluating any one idea; (2) force-field analysis, in which
problem solvers list the possible forces that will contribute to and inhibit solu-
tion of a problem; and (3) lateral thinking, in which problem solvers apply ideas
from diverse areas to an unrelated problem situation. While these techniques
are effective in generating creative ideas, translating creative ideas into practi-
cal problem solutions has been difficult. Cognitive mapping is proposed as a
means for managing the creative problem-solving process. Through cognitive
mapping, problem solvers’ explanatory and predictive beliefs about a problem
and its solution are graphically modelled so that members of a problem-solving
group are better able to understand one another’s positions and underlying
assumptions. Therefore, members can draw upon their own and others’ per-
sonal wisdom in creatively solving the problem. Examples of the practical busi-
ness implementation of the technique are presented.

Kiesler, S. and Sproull, L. (1982) ‘Managerial response to changing envi-

ronments: Perspectives on problem sensing from social cognition’,
Administrative Science Quarterly, 27 (4): 548–571.

Managerial problem sensing is characterized as a necessary precondition for

managerial activity directed toward organizational adaptation and is composed
of noticing, interpreting and incorporating stimuli. The constituent social cogni-
tion processes that make some types of problem-sensing behaviour, including
errors, relatively likely to occur are reviewed. Four simple ideas emerge as the
conventional wisdom of problem sensing. To improve problem sensing, indi-
viduals or organizations should: (1) learn from experience; (2) plan; (3) increase
the speed with which managers receive information; and (4) increase the range
of information managers receive. For social perception, five theories dominate:
(1) the augmentation principle, (2) the discounting principle, (3) illusory corre-
lation, (4) illusory causation and (5) automatic scanning. The social cognition
processes have organizational implications for such issues as crises, chance
events and breakpoints.
286 ANNOTATED BIBLIOGRAPHY

Klein, J.H. and Cooper, D.F. (1982) ‘Cognitive maps of decision-makers in

a complex game’, Journal of the Operational Research Society, 33 (1): 63–71.

Much attention is currently being directed towards the problems of strategic

decision making and decision making in conflict and crisis. An interdisciplinary
stance involving psychological research, political science and international
relations would be helpful in assessing the contributions of the individual deci-
sion maker to the model. This review illustrates the use of a cognitive mapping
technique to examine the behaviour and perceptions of individual decision
makers, a cognitive map being a representation of the subjective decision-
making environment of an individual. Seven military officers each played two
scenarios in a research war game. Analysis of their communications revealed
that the individual players were remarkably consistent over the two scenarios
but that their perceptions of their common decision-making environment differed
significantly. The differences noted focused on: (1) the size and complexity of
their cognitive maps; (2) detailed interpretation of the maps; (3) players’ confi-
dence and anticipation of the future; and (4) the way in which the maps were
altered as time progressed.

Kwahk, K.-Y. and Kim, Y.-G. (1995) ‘Supporting business process redesign
using cognitive maps’, Decision Support Systems, 25 (2): 155–178.

Turbulent changes and competitive pressures have forced organizations to con-

stantly change. Business process redesign has been widely adopted as an
organizational change method in the 1990s. Although BPR projects provide the
possibility of dramatic performance improvement, many organizations have
encountered serious problems due to the lack of commitment to such projects
and the difficulty of systematic targeting of critical processes. By identifying the
cause–effect relationships within an organization, an attempt is made to address
these issues. A cognitive map-based method, called two-phase cognitive mod-
elling (TCM), is proposed to help organizational members identify potential
organizational conflicts, capture core business activities and suggest ways to
support the necessary organizational change. To apply the method in the real-
world context, a prototype modelling tool, called two-phase cognitive model-
ling facility (TCMF), is developed. Working procedures of the TCM method and
TCMF features are illustrated with their application to the real BPR project of a
dairy company.

Langfield-Smith, K. (1992) ‘Exploring the need for a shared cognitive

map’, Journal of Management Studies, 29 (3): 349–368.

An experiment was undertaken to elicit the shared perceptions of six individu-

als employed in a Fire Protection Department regarding the important aspects
 ANNOTATED BIBLIOGRAPHY 287

of the job of a fire protection officer. The experiment consisted of two stages –
individual cognitive mapping sessions and group workshops. Shared percep-
tions were modelled in the form of a causal cognitive map. Participants were
divided into subgroups of two to three people, and each subgroup was asked to
cluster elements and to categorize the elements as agreed, disputed or dis-
carded. Participants met as one group to finalize the list of agreed elements. The
workshop terminated because of the lack of agreement on language and its
meanings. A new model that explained the relationships between individuals’
and shared belief systems and collective cognitions proposed that individuals
operating as a decision-making group do not necessarily need to share an
extensive system of shared beliefs.

Langfield-Smith, K. and Wirth, A. (1992) ‘Measuring differences between

cognitive maps’, Journal of the Operational Research Society, 43 (12):
1135–1150.

A causal cognitive map is a directed network representation of an individual’s

beliefs concerning a particular domain at a point of time. The nodes and the
arcs joining them indicate causal beliefs. There have been few attempts to
develop quantitative measures for such maps. The measures could be used to
compare the maps of different individuals and also to track the changes in the
beliefs of a single individual over time. They would assist in providing a more
objective basis for qualitative analysis. Current cognitive mapping research is
reviewed, and some measures are proposed for computing the difference
between two maps illustrating this work with a managerial example.

Larsen, E. and Lomi, A. (1999) ‘System dynamics and the “New

Technology” for organizational decisions: From mapping and simula-
tion to learning and understanding’, European Management Journal,
17 (2): 117–119.

Managers are surrounded by systems of all kinds: economic, cognitive, social

and organizational. These various systems are generally characterized by a
number of attributes that make them difficult to predict and control.
Computer simulation is a powerful tool to help managers deal with the
systematic properties of their decision environments. System dynamics offers
a comprehensive approach to the integration of organizational learning, com-
puter modelling and the creation of test-bed for evolving alternative solutions
and strategies.

Laukkanen, M. (1994) ‘Comparative cause mapping of organizational

cognitions’, Organization Science, 5 (3): 322–343.
288 ANNOTATED BIBLIOGRAPHY

A method for comparatively studying real-life managerial thinking – defined as

the respective manager’s beliefs about key phenomena and their efficacy links
in their strategic and operative situation – is described. A study case comparing
the cognitive structures of managers in two interrelated industries in terms of
their concept bases and causal beliefs is presented. It is shown that patterns of
industry-typical core causal thinking, manifestations of a dominant logic or
recipe, can be located, operationalized and comparatively analysed with this
method. The contents of management thinking are typically products of complex
long-term mechanisms.

Louis, M.R. and Sutton, R.I. (1991) ‘Switching cognitive gears: From
habits of mind to active thinking’, Human Relations, 44 (1): 55–74.

A perspective on the switch from automatic to active thinking and the condi-
tions that provoke it is discussed. Effective thinking may be related to an indi-
vidual’s ability to sense when a change in thinking style is needed.

Madu, C.N. and Jacob, R.A. (1991) ‘Multiple perspectives and cognitive
mapping to technology transfer decisions’, Futures, 23 (9): 978–997.

Multiple perspectives are used to analyse the technology transfer process. The
use of multiple perspectives avoids the problems encountered in using purely
technical or rational models, such as management science and operational
research techniques used in solving socio-technical problems. Emphasis is
placed on technology transfer to less developed countries where previous
rational techniques have been unsuccessful. Two case studies are used: one
focusing on the development of agricultural versus oil technology, and the other
focusing on the transfer of production technology. The case studies illustrate the
way in which the use of multiple perspectives allows the consideration of the
technical, organizational and personal perspectives in the technology transfer
decision-making process, thus enhancing and facilitating the process.

Marchant, T. (1999) ‘Cognitive maps and fuzzy implications’, European

Journal of Operational Research, 114 (3): 626–637.

A cognitive map is a collection of nodes linked by some arcs. Up to this point,

there is unanimity in the literature about the previous definition. However, if
one looks closer at the meaning of the nodes and links, one can expect that
there are crucial differences between the various authors. These differences are
 ANNOTATED BIBLIOGRAPHY 289

not always explicit. In spite of this, it seems that many authors perform on the
maps the same kind of analysis (strongly inspired by Axelrod, 1976), even if
these analyses are not consistent with their conception of a cognitive map. That
is why it is important to define the kind of map used clearly and formally. A
formal definition of a cognitive map relying on the concept of fuzzy implication
is proposed. In the framework, a node is a logical proposition and a link is an
implication. Starting from the definition, some properties of this kind of maps
and some analysis techniques are shown.

Morecroft, J.D.W. (1992) ‘Executive knowledge, models and learning’,

European Journal of Operational Research, 59 (1): 9–27.

Modelling and simulation have come of age, extending their influence beyond
the mind and desktop of the analyst into the boardroom and the mental
models of managers. Increasingly, models are seen to have a role as instruments
to support strategic thinking, group discussion and learning in management
teams. In this respect, they are quite similar to qualitative problem-structuring
approaches used by strategy advisers and process consultants. Models are
described in terms of three attributes that support different cognitive and group
processes in management teams. Models can be viewed as maps that capture
and activate knowledge. They can also be viewed as frameworks that filter and
organize knowledge. Finally, they can be viewed as microworlds for experi-
mentation, co-operation and learning. An explanation is given of how the
modelling process fits into conventional management team meetings. The value
chain methodology and system dynamics are then constrasted in order to illus-
trate the variety of group and cognitive support provided by different maps and
frameworks.

Porac, J.F. and Thomas, H. (1990a) ‘Taxonomic mental models in competi-

tor definition’, Academy of Management Review, 15 (2): 224–241.

A cognitive approach to the problem of competitor definition is outlined. Before

competitive strategies can be formulated, decision makers must have an image
of who their rivals are and on what dimensions they will compete. Given the
diverse range of organizational forms and decision makers’ limited capacity to
process complex interorganizational cues, the task of defining the competition is
both important and problematic. This dilemma leads to the use of simplified
mental models to define rivals. To the extent that the two can be separated con-
ceptually, the focus is primarily on the structure of these models, rather than on
the process of competitive identification. By internalizing a mental classification
of organizational forms, the strategist can simplify the interorganizational envi-
ronment by collapsing individual organizations into category types. Five propo-
sitions that clarify how this simplification process works in practice are provided.
290 ANNOTATED BIBLIOGRAPHY

Porac, J.F. and Thomas, H. (1994) ‘Cognitive categorization and subjective

rivalry among retailers in a small city’, Journal of Applied Psychology, 79
(1): 54–66.

Results are reported of two studies conducted to measure the cognitive struc-
tures underlying perceived competitive relationships among retailing firms in a
small city. Drawing from recent research on cognitive categorization, the theo-
retical importance of studying subjective rivalry is first discussed, and then an
explanation of how categorization processes influence perceived competitive
boundaries among firms is offered. The results of Study 1 suggest that cognitive
categories of firms are perceived to be largely independent sets of organiza-
tions. The results of Study 2 suggest that middle-level categories represent a
psychological inflection point differentiating rivals from non-rivals. The impli-
cations of these data for studying how managers make sense of competitive
structures are discussed.

Porac, J.F., Thomas, H. and Baden-Fuller, C. (1989) ‘Competitive groups as

cognitive communities: The case of Scottish knitwear manufacturers’,
Journal of Management Studies, 26 (4): 397–416.

The Scottish knitwear industry was chosen to study the influence of shared
beliefs because of its small size and long-standing traditions. Extensive semi-
structured interviews were conducted with top managers from approximately
35 per cent of these companies over a six-month period. It is suggested that the
Scottish knitwear sector exists as it does today because the mental models and
strategic choices of key decision makers intertwine to create a stable set of
transactions in the marketplace. Core identity and causal beliefs are identified
that allow managers to define competitive boundaries and make sense of inter-
actions within these boundaries. It is suggested that such beliefs are reinforced
by a mutual enactment process in which the technical choices of firms con-
strain the flow of information back to decision makers, thereby limiting their
vision of the marketplace to that which has already been determined by exist-
ing beliefs.

Reger, R.K. and Huff, A.S. (1993) ‘Strategic groups: A cognitive perspec-
tive’, Strategic Management Journal, 14 (2): 103–124.

The strategic group concept provides an attractive middle ground between firm
and industry for both theory development and empirical analysis. To date, this
concept has been defined by researchers in terms of secondary accounting and
financial data, and a number of critics have questioned the validity of this work.
Research shows that industry participants share perceptions about strategic
commonalities among firms, and that participants cluster competitors in subtle
 ANNOTATED BIBLIOGRAPHY 291

ways not reflected in extant academic research on strategic groups. Decision

makers’ perceptions and cognitions are phenomena that can be expected to
influence industry evolution. They are of research interest as an additional
source of data on firm commonalities that helps address concerns about previ-
ous strategic group research.

Reger, R.K. and Palmer, T.B. (1996) ‘Managerial categorization of com-

petitors: Using old maps to navigate new environments’, Organization
Science, 7 (1): 22–39.

The paper examines differences between automatic and controlled processing by

executives in an increasingly dynamic industry. The results suggest that cognitive
inertia affects judgements in both modes, but the effect is stronger with automatic
processing. The longitudinal results indicate that change creates diversity of
thought across managers in the same environment. Managers at competing firms
are therefore apt to view competition quite differently in turbulent environments.

Reger, R.K., Gustafson, L.T., Demarie, S.M. and Mullane, J.V. (1994)
‘Reframing the organization: Why implementing total quality is easier
said than done’, Academy of Management Review, 19 (3): 565–584.

A cognitive theory is presented of why planned organizational change efforts,

such as total quality initiatives, often fail. The theory suggests that employees
resist total quality because their beliefs about the organization’s identity con-
strain understanding and create cognitive opposition to radical change. A
dynamic model is proposed in which successful implementation of fundamen-
tal organizational transformation is partly dependent on management’s ability to
reframe the change over time. Implementation may best be accomplished
through a series of middle-range changes that are large enough to overcome
cognitive inertia and relieve organizational stress, but not so large that members
believe the proposed change is unobtainable or undesirable.

Reynolds, T.J. and Gutman, J. (1988) ‘Laddering theory method, analysis,

and interpretation’, Journal of Advertising Research, 28 (1): 11–32.

Laddering is an in-depth, one-on-one interviewing technique that is used to

develop an understanding of how consumers translate product attributes into
meaningful associations regarding the self. The underlying theory behind the tech-
nique is the Means–End Theory (Gutman, 1982). Several specific interviewing
devices used to initiate the laddering process are described, including: (1) triadic
sorting, (2) eliciting preference-consumption differences and (3) determining
292 ANNOTATED BIBLIOGRAPHY

consumption differences by occasion. The value of the occasional context is

stressed. The analysis of laddering data is discussed, and emphasis is placed on
the crucial difference between this technique and other, more traditional, quali-
tative ones. With the laddering technique, the primary output is quantitative and
in the form of a hierarchical value map (HVM). The construction of the HVM is
reviewed in detail. Applications also are discussed, with consideration of the key
research problems, including perceptual segmentation, product/brand assessment
and assessment and development of advertising strategy.

Roos, L.L. Jr and Hall, R.I. (1980) ‘Influence diagrams and organizational
power’, Administrative Science Quarterly, 25 (1): 57–71.

The use of influence diagrams to help understand political processes within

organizations is examined and illustrated through a case study of a new
extended care facility connected to a hospital. The high-cost unit was operat-
ing well below capacity, so an influence diagram was drawn to help under-
stand why the director was unwilling to increase the occupancy rate to
budgeted standards and how such action was related to his overall strategy. A
particular object of this study was to discover the influences and strategies
underlying the director’s policies and postures in dealing with the unit’s envi-
ronment. Influence diagrams helped to generate an overall view of the organi-
zational problems being considered. The influence diagram suggests ways in
which innovation might be facilitated. Proper positioning within the organiza-
tion, having the support of top management and establishing external support
help the innovator.

Russell, R.D. (1999) ‘Developing a process model of intrapreneurial sys-

tems: A cognitive mapping approach’, Entrepreneurship: Theory &
Practice, 23 (3): 65–84.

This paper uses a cognitive mapping approach to build a model of corporate

entrepreneurship from an organizational perspectives. This approach analyses
corporate entrepreneuring from a systems viewpoint and facilitates an under-
standing of the process through which entrepreneurial firms generate innovation.
Additionally, the application of Maruyama’s (1963) concept of ‘morphogenetic’
systems provides a possible explanation for the non-linear changes that are
characteristic of intrapreneurship.

Salancik, G.R. and Meindl, J.R. (1984) ‘Corporate attributions as strategic

illusions of management control’, Administrative Science Quarterly,
29 (2): 238–254.
 ANNOTATED BIBLIOGRAPHY 293

This study analyses the reasons given by 18 chief executive officers to explain
their firms’ performance in annual stockholder reports over the 1961–78 period,
comparing firms with stable and unstable performance. It is contended that the
managements of unstable firms, lacking real control over organizational out-
puts, strategically manipulate causal attributions to manage impressions of their
control. These managements claim liability for both positive and negative per-
formances more than do the managements of firms with stable performance,
and, contrary to psychological theories, seem unwilling to attribute poor per-
formance to uncontrollable environmental events. When environmental effects
are cited, they are associated with announcements of executive changes, thus
implying to readers that management is prepared to do battle with its volatile
environment. These attributional strategies resulted in better future performance
for the firms using them.

Schwenk, C.R. (1984) ‘Cognitive simplification processes in strategic deci-

sion making’, Strategic Management Journal, 5 (2): 111–128.

Generating ideas about the ways decision makers actually deal with complexity,
ambiguity and uncertainty, involves looking at cognitive simplification processes
in strategic decision making. The focus on simplification processes for which
laboratory and field support exists should increase the chance of identifying
cognitive processes that really do affect organizational decisions, rather than
processes produced only by the artificiality of the laboratory context or the politi-
cal processes in organizations. A review is provided of selected cognitive sim-
plification processes and the possible effects of these biases on strategic decision
making, even though these biases are not always harmful in organizations.
Future research in this area should concentrate on field settings to provide insight
into the effects of these processes. Laboratory research should use tasks more
representative of the ill-structured problems actually encountered in business.

Schwenk, C.R. (1985) ‘The use of participant recollection in the modelling

of organizational decision processes’, Academy of Management Review,
10 (3): 496–504.

Two basic differences exist between models of organizational decision

processes founded on participant recollection and those founded on other data
sources, such as analysis of meeting transcripts, archival data and field obser-
vation. Specific decision-process models illustrate these differences. Possible
reasons for the differences among models developed from different data sources
are outlined, and implications for decision-process research are discussed.
Approaches must be formulated for combining data from various sources so that
they challenge and improve models developed from a single data source rather
294 ANNOTATED BIBLIOGRAPHY

than merely confirming and extending them. Suggestions are offered to help
minimize the biasing effects of primary reliance on a single data source and to
reconcile some of the differences between decision process models based on
different data sources.

Schwenk, C.R. (1988) ‘The cognitive perspective on strategic decision

making’, Journal of Management Studies, 25 (1): 41–55.

Strategic management researchers are focusing increased attention on the cog-

nitions of key decision makers. This is an important development because:
(1) the study of cognition may improve understanding of industry and competi-
tive strategy and the ways environmental factors affect strategic decisions; and
(2) sufficient attention has not been paid to the study of cognition in the past.
Cognitive research can give insights into the ways decision makers with limited
cognitive capacities comprehend and solve very complex strategic problems
and into the types of errors they make. Recent research has explored such topics
as: (1) heuristics and biases, (2) assumptions and cognitive maps, (3) analogy
and metaphor, and (4) the development of strategic schemas. These separate
streams of research need to be integrated in order for future research to provide
a complete understanding of strategic problem solving.

Sims, D.B.P. and Doyle, J.R. (1995) ‘Cognitive sculpting as a means of

working with managers’ metaphors’, Omega, 23 (2): 117–124.

A new technique is presented for helping managers to talk through and develop
their views of difficult and complex issues, which are given expression by
arranging a collection of objects, some of them symbolically rich, in an arrange-
ment or ‘sculpture’. At the same time, the managers describe and develop the
meanings being given to, and the relationships among, the objects. The tech-
nique is in the tradition of elicitation techniques, such as cognitive mapping, in
that it encourages a person or a group to dialogue with a physical representa-
tion of their ideas. Meanings are not merely described but sometimes actively
constructed or negotiated. Theoretically, the technique draws on recent work in
cognitive psychology and linguistics on metaphors. The use of this technique in
a variety of different circumstances is described and evaluated. It is argued that,
more than two-dimensional techniques, cognitive sculpting offers the requisite
variety to capture and communicate the richness and metaphoric complexity of
managers’ views of their world.

Snaw, J.B. (1990c) ‘A cognitive categorization model for the study of inter-
cultural management’, Academy of Management Review, 15 (4): 625–646.
 ANNOTATED BIBLIOGRAPHY 295

A cognitive categorization model of intercultural management is developed that

focuses on interaction between an expatriate manager and a host country sub-
ordinate. The model’s premise is that difficulties occur between managers and
subordinates from different cultures because of basic differences in how the
individuals collect, process, store and use information about one another’s
behaviour. Some of the effects that culture may have on the content and struc-
ture of schemas are outlined. In addition, consideration is given to the extent to
which automatic versus controlled information processing occurs. The use of
the model to examine convergence of cognitive structures through intercultural
dynamics is also discussed. The model focuses on the daily interaction between
a manager and a subordinate. It does not attempt to explain culture-based
differences in other important interpersonal or group relationships.

Stubbart, C.I. (1989) ‘Managerial cognition: A missing link in strategic

management research’, Journal of Management Studies, 26 (4): 325–347.

The linkages between cognitive science and strategic management research

are explored. Schendel and Hofer (1979) implicitly assumed a cognitive basis
for much of the strategy-making process, but they did little to systematize a
cognitive approach. Strategy scholars have recently focused their efforts upon
describing managers’ semantic networks, or cognitive maps (what managers
know about their competitive situations). Semantic networks have several use-
ful properties, including flexibility, activation, reasoning and computation.
These networks also have limitations, such as: (1) there is no evidence that
humans reason according to matrix algebra; (2) cognitive maps cannot easily
represent temporal relations; and (3) cognitive maps used in strategy are often
excessively abstract. Today, some scholars regard cognitive simplification and
non-optimal heuristic inferences as a serious threat to managerial decision
making. However, strategic management stands to benefit greatly from an infu-
sion of relevant empirical findings, stronger theory and added support for
strategic planning.

Swan, J. (1995) ‘Exploring knowledge and cognitions in decisions about

technological innovation: Mapping managerial cognitions’, Human
Relations, 48 (11): 1241–1270.

Success and failures with technological innovation can, in some cases, be

attributed to the decision-making processes in the user organization. There has
been a growing interest among strategy researchers in managerial cognitions.
The nature and importance of knowledge bases and cognitions for decisions
about technological innovation are described and a suggestion is made about
how knowledge and cognitions can be explored using process research.
Cognitive mapping methodologies are useful to researchers, particularly in
296 ANNOTATED BIBLIOGRAPHY

conjunction with process research, and potentially useful to practitioners, for

understanding managerial cognitions and for anticipating problems that may
arise as a result of these cognitions.

Swan, J. (1997) ‘Using cognitive mapping in management research:

Decisions about technical innovation’, British Journal of Management, 8
(2): 183–198.

The paper highlights the importance of cognitions in decisions about techno-

logical innovation. However, cognitive processes have been under-emphasized
in empirical work on technological innovation and part of the problem may
have been lack of availability of research tools and techniques with which to
explore cognitions. Cognitive mapping methodologies are reviewed in terms of
their potential to fill this gap in the research into technological innovation. The
paper discusses these methodologies, evaluates their limitations and argues that
a distinction should be made between cognitive maps and the output of map-
ping techniques. It is concluded that cognitive mapping may provide a useful
addition to existing management research tools provided researchers are clear
about what is revealed by the particular methodology used.

Swan, J.A. and Newell, S. (1994) ‘Managers’ beliefs about factors affecting
the adoption of technological innovation: A study using cognitive
maps’, Journal of Managerial Psychology, 9 (2): 3–11.

Managers’ cognitions or belief systems play an important role in the decision-

making process that leads to the adoption of innovations, but research in this
area has been neglected. A cognitive mapping methodology is used to reveal
managers’ beliefs about the causes of and effects of a particular type of techno-
logical innovation. These managers’ beliefs are compared with suggestions
made in the academic literature about the factors that influence a firm’s level of
innovation. Factors that, according to the academic literature, increase the like-
lihood of innovation were not believed to be important direct causes by these
managers. These managers saw involvement in professional associations as a
very important causal factor in innovation in production and inventory control.
Other factors seen to be direct causes were the ratio of professional and tech-
nical staff to others in the firm, the promotion activities of vendors, and the
competitors’ levels of technology.

Thomas, J.B., Clark, S.M. and Gioia, D.A. (1993) ‘Strategic sensemaking
and organizational performance: Linkag’, Academy of Management
Journal, 36 (2): 239–270.
 ANNOTATED BIBLIOGRAPHY 297

The question of how key cognitive processes (scanning, interpreting and

responding) of top managers may be linked to organizational performance is
examined. Scanning involves searching the environment to identify information
that is pertinent to the organization. Interpretation involves comprehending the
information and deciding which strategic issues to address. A response, or
action, is a change in organizational practices. Using path analyses on data
from 156 hospitals, the direct and indirect effects among these sensemaking
processes and performance outcomes were tested and a model of their rela-
tionships developed. Top managers’ attention to high levels of information
during scanning was found to be related to their interpretation of strategic issues
as positive and as implying potential gains. There was no link found between
positive-gain interpretations of strategic issues and subsequent action. However,
the interpretation of a strategic issue as controllable had a positive effect on
product-service change.

Vislosky, D.M. and Fischbeck, P.S. (2000) ‘A mental model approach

applied to R&D decision-making’, International Journal of Technology
Management, 19 (3–5): 453–471.

A mental model approach is developed to map systematically an organization’s

R&D project-selection decision process from the cognitive beliefs of its key
decision-makers. Respondents revealed decision-criteria maps are judged against
a normative project-selection model to reveal systematic departures from rational
choice. The methodology is demonstrated within an industrial R&D organization
of a large production company. The results highlight the value of the mental
model approach applied to R&D decision making. Application of the methodo-
logy may prove useful to managers and policy makers by providing insightful
information about an industry’s R&D selection process.

Walsh, J. (1988) ‘Selectivity and selective perception’, Academy of

Management Journal, 31 (4): 31–35.

Dearborn and Simon’s (1958) work on selectivity and selective perception was
conceptually replicated and extended using data from 121 midcareer managers
enrolled in a two-year, part-time executive master’s degree programme at a
large university. The early work was replicated and extended through measures
of managers’ entire work histories, their belief structures and three indexes of
information processing in an ill-structured decision situation. Results indicated
that, contrary to Dearborn and Simon, few managers saw their organizational
worlds along narrow functional criteria. Three-quarters of the group either had
no firmly held dominant conception of success or had strong conceptions of
success. Little evidence was found of parochial information processing.
298 ANNOTATED BIBLIOGRAPHY

Managers with a marketing belief structure identified more external management

problems and asked for more external management information than did managers
in human relations and generalist groups.

Walton, E.J. (1986) ‘Managers’ prototypes of financial firms’, Journal of

Management Studies, 23 (6): 679–699.

The applicability of meaning giving as a categorization process of matching

stimuli against prototypes to organizations was explored by focusing on pro-
totypes as a central component of the perspective. A phenomenological
analysis of 22 prominent institutions in New York City focused on the mean-
ings that managers attach to organizations or to contents of categories and
prototypes. A four-part structured interview was used to elicit managers’
descriptions of their own firms and nine others. The responses were analysed
linguistically, and multidimensional scaling techniques were used to uncover
hidden structures in the data. The results of the study supported the premise
that people endow organizations with meaning by a prototype-matching cat-
egorization. The results suggest that prototypes affect structure of information
about organizations. Managers apparently used the labels provided to list pro-
totypical attributes.

Wang, S. (1996) ‘A dynamic perspective of differences between cognitive

maps’, Journal of the Operational Research Society, 47 (4): 538–549.

To complement current methods used in measuring the differences between

cognitive maps, the application of neural network tools is suggested. The neural
network approach aids decision makers in the identification of dynamic aspects
of target systems that are represented by the cognitive maps to be compared. A
real-data application of this method shows that the proposed technique can be
used to enhance the understanding of a decision environment.

Warren, K. (1995) ‘Exploring competitive futures using cognitive map-

ping’, Long Range Planning, 28 (5): 10–21.

Uncertainty over the direction and scale of change in future industry conditions
is one of the most intractable problems of strategic management. Anticipating
the future can be hampered by misunderstanding of the driving forces at work,
their relative significance and the ability of the firm to affect them. UK take-
home drinks retailing is examined and it is shown how diverse forces for change
may be combined in mental maps to build credible scenarios to inform a firm’s
strategy. An example shows the dangers of accepting conventional wisdom,
 ANNOTATED BIBLIOGRAPHY 299

the importance of including in scenarios the responses of participating firms,

and the potential to discover promising strategies in apparently unattractive
competitive conditions.

ACKNOWLEDGEMENT

Abstracts are reprinted with permission from ABI/INFORM. Copyright

1999 by UMI Company. All rights reserved. No further reproduction or
redistribution without prior permission from UMI.

REFERENCES

Axelrod, R.M. (1976) Structure of Decision: The Cognitive Maps of Political Elites. New Jersey:
Princeton University Press.
Dearborn, D.C. and Simon, H.A. (1958) ‘Selective perception: A note on the department
identifications of executives’, Sociometry, 21: 140–144.
Glick, W.H. (1985) ‘Conceptualizing and measuring organizational and psychological cli-
mate. Pitfalls in Multilevel Research’, Academy of Management Review, 10 (3): 601–616.
Gutman, J. (1982) ‘A means-end chain model based on consumer categorization processes’,
Journal of Marketing, 46 (2): 60–72.
Kelly, G. (1955) Theory of Personal Constructs. New York: Norman.
Maruyama, M. (1963) ‘The second cybernetics: Deviation-amplifying mutual causal
process’. American Scientists, 51: 164–179.
Porac, J.F. and Thomas, H. (1990) ‘Taxonomic mental models in competitor definition’,
Academy of Management Review, 15 (2): 224–240.
Schendel, D.E. and Hofer, C.W. (1979) Strategic Management: A New View of Business Policy and
Planning. Boston: Little Brown.
INDEX

Active listening, 230 Competitive advantage/organizational

Anonymity, 175 success, 8, 13, 19–21, 31, 37–8,
Artificial intelligence, 89, 90, 92–4, 98, 50–1, 113, 133, 139, 163, 192, 221,
100, 107–08 224, 227
Aspirations and goals, 173–5, 183, 185–6, Characteristics, 35–6
188–9, 191, 193 Vulnerability, 36–7
Ascanelli, 152 Consensus, 59, 91
ATLAS/ti, 118–19, 238, 240, 243–54, Content analysis, 54, 117–19
249, 263–5 Corporate strategy, 220–1
Analysis, 254–9, 263 Cosworth, 149
Coding, 244–8, 251 Coughlan, 152
Attribution, 78 Critical incident, 26
Cluster analysis, 119, 121–8, 198
Barnard, 152, 155 Culture, 146
Brabham, 155
Byrne, 147 Decision Explorer/ Group Explorer, 9, 29,
33, 175–6, 182, 185
Card sort, 197 Decision making, 9, 92, 191, 198–9, 203,
Case studies, 237–8, 265 209, 215
Causal mapping, 19, 22, 89–90, 105–06, 113, Strategic, 196–7, 201, 203, 213,
118, 174–5, 204 237–9, 253
Benefits, 34–5, 105–06, 109, 204 Type I and II, 199–200, 214–15
Complexity, 91, 94, 107 Dominant logic, 112, 115, 118, 128–9,
Group, 22–3, 89–90, 92, 94, 106, 108, 131–3
220–1, 225–8
Problems, 35, 223–8 Ecclestone, 161
Process, 24–5, 27–8, 38, 95–7, 206–07, Entrepreneurs, 113–18, 128–9, 132–3
221–3, 228 ESRC, 216
Clark, 161
Cognitive bias, 196–7, 200–1, 203, 213 Facilitation of mapping session, 220,
Cognitive mapping, 1–2, 21, 55–6, 60, 223, 225–9
91, 94, 113–15, 117, 118, 174–5, Principles, 229–31
197–8, 213–16 Federation Internationale de l’Automobile
Benefits, 175–6, 181, 187–8, 191–3 (FIA), 158, 161, 165
Group, 176–8 Feedback loop, 89–90, 94, 102–04, 106–08
Problems, 231–4 Ferrari, 159–160
Techniques, 9, 114, 118, 175–6, 178, Firm turnover, 149, 163
182–4, 200 Forti, 152
Cognitive models, 199 Frame, 52
Cognitive sculpting, 8, 63, 72–84 Framing, 200–204, 209, 211, 213
Communities of practice, 1
Competencies (organizational), 112, 114, Geography (role), 137–8, 141, 145
185–8, 190, 193, 220–1, 223, Gossip and rumour, 157, 163
227, 234 Govan Initiative, 185
 INDEX 301

Head, 151 Learning, 48–51, 55, 107

Health Authority, 182 Lola, 152
Hemisphere (brain), 79–80, 84 Lotus, 155
HyperCard/HyperTalk, 92
Managerial and organizational cognition,
Ignorance, 33 197–8, 212, 231–3
Industrial Organization Economics, 37 Managers’ role, 37
Industry cluster, 138–9, 141, 144 Maps
Innovation, 157–8, 163 Career trajectories, 147–52
Interviews, 24–5, 32, 53–4, 95, 105, 145, Characteristics/benefits, 2–6, 14
175–6, 182, 244–5, 254–5 Strategy, 175–6, 179, 183, 186–92
Interpretive view, 46–9, 51, 60, 78, 260 Strategic knowledge, 139
Methodological issues, 260–3 March, 149, 151, 153, 155
McLaren, 158
JOURNEY-making approach, 173–4, Mental models/mental maps, 197
183, 185 Customers, 46, 55–6
Entrepreneurs, 113–16, 128, 132
Kauffman Foundation, 116, 118, 133 Individuals, 199, 214–15, 237–8
Kelly’s personal construct theory, 174, Managers, 47–8, 51, 55–56, 107, 112–14,
176, 180, 194 119, 128
Knowledge Researchers, 9, 239, 250
Based software (see also ATLAS/ti; Mercedes-Benz, 161
Decision and Group Explorer), Metaphors, 19, 26–8, 63–72, 77, 83–4
237–41, 259–60, 264–6 Types, 64–69
Cluster, 9 Ubiquity, 66–7, 70
Community, 140, 145–6, 149, 155–6, Motor sport
159, 162–3 Industry, 137, 141, 144–6, 155, 158, 163
Dissemination, 137, 140, 144, 146–7, Valley, 137–8, 140–5, 149, 155–6,
155–6, 162, 165 159–63
Economy, 137–8, 140–1, 144, 163 Multilectic perspective, 262–3
Embodiment, 146, 155, 160, 174 Murray, 158
Entrepreneurial, 112–13
Explicit, 1, 14, 32 Narratives, 112, 116–19, 124, 128, 130,
Generation/production, 1, 14, 137, 133, 256
140, 144, 162, 165, 173 Negotiation, 175, 179, 183, 185
Individual, 71, 83, 163
Informal, 1 Oval Mapping Technique (OMT), 175,
Managerial, 1, 9 178–80, 185
Organizational, 20, 63–4, 173 Observation, 157
Role, 140, 144 Option theory, 241–2, 253
Scientific, 158–60
Spatial organization, 138–40, 143, Pacific, 152
162–3 Panlogos Consulting and University of
Strategic, 81, 83–4, 139 Witten-Herdecke, 46, 50
Surfacing (see also causal, cognitive Perceptions
mapping and cognitive sculpting, Customers, 52, 59
knowledge-based software), Managers, 59, 242
173–174, 193 Planning, 242–243
Tacit/implicit, 1, 8, 14, 19–20, 26–7, Policy making, 89–92, 94, 105, 108
32–3, 174–5, 232 Postlethwaite, 151, 159
Prospect theory, 201–03, 211
Labour
Market, 149, 155, 163 Qualitative research (see also knowledge-
Network, 149, 153, 155 based software), 237–8, 240, 265
Turnover, 46, 149, 163 Quantitative research, 49
302 INDEX

Repertory grid, 197 Strategy Delivery Support System, 193

Researchers (see also mental models) Strategies, 9, 55–6, 59, 128–9, 185, 188,
Interpretation, 9, 260–1 190–1, 193
Role, 32–3 Growth, 112–15, 118–23, 126–8, 130, 132–3
Resource-based view of the firm, 1, 8, 19, Strategic management, 1, 19, 36–7, 46, 199,
37, 114–15, 128, 131, 133 204, 212
Reynard Racing, 151–2 Techniques, 47, 49, 50, 60 (see also
Rinland, 150 SENSOR, Decision Explorer,
Rigour and relevance (in research), 215–16 Cognitive Sculpting and mapping,
Routines (organizational), 1, 8, 20, 30, 38 Causal mapping, JOURNEY)
Paradox, 37 Shell, 181
Tacit, 19–39, 221 System Dynamics model, 94, 106
Unmanaged, 36
Rules (behavioural), 92–3, 104 Technology, 157–61
Theories
Schema Espoused, 133
Image, 70, 72 In use, 133, 176
Scottish Natural Heritage, 189 Thomas, 151
Self-Q, 24–5, 28, 32 Trust, 32, 34
Senna, 161 Tyrell, 152
SENSOR, 46, 51–3, 55, 56
Sense making, 52, 174 UCINET IV, 123
Simtek, 152, 155 Untraded interdependencies, 141, 143
Social mechanisms, 1
Storytelling, 8, 19, 24–6, 28, 32, 52–3, Values (corporate), 112
55–6, 228
Strategy development process, 173–5, Wright, 158
191, 193