Spring 2006, Vol. 4, No.

Qualitative Newsletter of the

Methods American Political Science Association

Organized Section on Qualitative Methods

Letter from the Editor

Contents
John Gerring
Symposium: Ethnography Meets Rational Boston University
Choice: David Laitin, For Example jgerring@bu.edu
Introduction In the Fall of 2002, APSA created its 37th Organized Sec-
Ted Hopf ..............................................................................2 tion, devoted to the study, development, and dissemination
Theory, Data, and Formalization: The Unusual Case of of qualitative methods. Since that time, I have served as the
David Laitin editor of this newsletter. My job, as I saw it, was to bring to
Yoshiko M. Herrera ..........................................................2 the attention of our members the most interesting, innovative,
Mechanisms vs. Outcomes and (it follows) contentious issues in the field of political
Kanchan Chandra ..............................................................6 methodology, regardless of whether they might be catego-
Understanding Rules and Institutions: Possibilities and rized conventionally as ‘qualitative’ or ‘quantitative.’ (Issues
Limits of Game Theory of import solely to quantitative work have been deferred to
David M. Woodruff ........................................................13 the Political Methodology section–no need to duplicate ef-
Ethnography and Rational Choice in David Laitin: fort.) With that caveat, the mission of the newsletter was in-
From Equality to Subordination to Absence terpreted broadly to include all methodological issues of rel-
Ted Hopf ..........................................................................17 evance to the study of politics. Symposia have ranged from
Recognizing the Tradeoffs We Make broad philosophy-of-science issues to narrower debates about
Ashutosh Varshney ........................................................20 technique. For the most part, these topics have been chosen
Ethnography and/or Rational Choice: in response to ideas from our members and as extensions of
A Response from David Laitin APSA panels and roundtables. Usually, the management of a
David Laitin .......................................................................26 symposium was delegated to the person taking the initiative
to organize a discussion on that topic.
Symposium: Alexander L. George and Andrew As editor I took a laissez-faire approach to the newsletter,
Bennett’s Case Studies and Theory Development asking authors to follow only a few stylistic and substantive
in the Social Sciences guidelines: contributions should be short, accessible to a
broad readership, written with some flair, and encompassing a
Introduction range of viewpoints on the chosen subject. The aim, while
Jack S. Levy ......................................................................34 retaining some of the intellectual rigor associated with more
Notes From a Generalist traditional academic journals, was to give writers scope to
Daniel W. Drezner ...........................................................34 opine–that is, to use the first-person pronoun and to adopt a
A Major Milestone with One Major Limitation more discursive manner than would be usual in a more formal
John S. Odell ...................................................................37 academic venue. In this manner, I hoped to reproduce the
Well Worth the Wait lively and candid views exchanged with each other in emails
Jack S. Levy ......................................................................40 and over cups of joe. “What do you really (in your heart of
Case Studies and the Philosophy of Science hearts) think of X?” This is the sort of conversation that I
David Dessler ..................................................................43 wanted to foster.
Advancing the Dialogue on Qualitative Methods During the last three years, the newsletter has covered a
Andrew Bennett ..............................................................45 lot of ground. In Spring 2003, we ran a symposium on teach-
ing qualitative methods, which featured a comprehensive re-
Book Notes ........................................................................48 view of textbooks and discussions of courses and various
Announcements ...............................................................51 approaches to the subject. In Fall 2003, our symposium ad-
dressed the knotty issue of “interpretivism,” with contribu-
tions from several scholars, including Clifford Geertz. In Spring
Qualitative Methods, Spring 2006

Concluding Remarks
Recognizing the Tradeoffs We Make
In this essay, I merely wanted to describe Laitin’s work in
terms of ethnography and rational choice, not evaluate it. The
Ashutosh Varshney1
differences are clear across the three pieces. It is also obvious
that each piece produces different kinds of knowledge, espe- University of Michigan, Ann Arbor
cially with respect to the expected generalizability of the find- varshney@umich.edu
ings. In HC, a rationalist theory of identity has been married to David Laitin defies a famous binary classification of schol-
a particular set of empirical circumstances. In IF, a more elabo- ars between hedgehogs and foxes. The late Isaiah Berlin’s
rate rationalist theory of identity has been tested in a much work, following Tolstoy’s, gave this distinction considerable
larger, though still bounded, domain. In EIC, a universalist currency in the social sciences. The hedgehog knows one
theory of identity has been tested nowhere, but has been dem- thing very well; and the fox knows quite a few things, if not
onstrated valid within a set of ultra-constraining assumptions. each in great detail.2 Hedgehogs work on one given topic/
Is the obvious true? Is ethnography the enemy of gener- theme/theory for an entire lifetime, adopting a cumulative re-
alization? Perhaps in practice, but not in principle. Wedeen search program, attempting to resolve one puzzle at a time, as
(2002) has recently written about the possibility of collecting they advance. Think of Arend Lijphart’s lifelong pursuit of the
intersubjective data based on phenomena such as identity, so idea of consociational democracy.
that conceptualization of variables need not be derived exclu- Foxes move from one big topic/theme/theory to another,
sively from a priori theories, but rather can remain true to the each topic keeping them engaged for a few years but not more,
ways in which concepts are understood in context. This would showing enormous intellectual breadth in the process. Con-
provide more reliable and valid data for those with statistical sider Samuel Huntington in political science, and Amartya Sen
inclinations, for those who wish to specify survey and focus in political economy. Huntington has provoked new debates
group instruments, and for those who wish to construct mod- in three fields of our profession: comparative politics, Ameri-
els with grounding in some reality. can politics, and international politics. Sen ranges from ratio-
The objective should be to return to Laitin’s original in- nality on one hand to famines and poverty, inequality, choice
sight. Theories of political action, of identity, of mobilization of techniques in planning, and, increasingly, identity on the
and identification require accounts of preferences that are not other.
merely assigned, but theorized and empirically uncovered. And Laitin has worked almost entirely on ethnic politics, rarely
preferences themselves are not just the oral statements or writ- if ever on development, economic reforms, democracy and auth-
ten testimonies of subjects, but are embodied in their mundane oritarianism, party politics, etc., let alone in other subfields of
daily practices. Ethnography, in this sense, is necessary for the political science discipline. Yet three things separate his
rational choice to produce creditable knowledge claims of any work from a classic hedgehog strategy. His substantive ques-
kind. tions have varied, even if the subfield has not; he has moved
from country to country in search of answers; and what is
Notes most pertinent to this symposium, his methodological commit-
1
Reminiscent of the efforts made to do so in Robert H. Bates, ments have radically changed over time.
Avner Greif, Margaret Levi, Jean-Laurent Rosenthal, and Barry R. Three of Laitin’s books deal with language politics. In
Weingast, Analytic Narratives (Princeton: Princeton University Press Politics, Language and Thought: The Somali Experience
1998). (1977), Laitin probed the political and social consequences of
2
This is reminiscent of Achen and Snidal’s recommendation that maintaining a neocolonial language like English, as opposed
qualitative case studies are best suited as the raw material appendages
to using a vernacular like Somali, as an official language. In
of rational choice models. Christopher H. Achen and Duncan Snidal,
“Rational Deterrence Theory and Comparative Case Studies,” World
Language Repertoire and State Construction in Africa (1992),
Politics 41:2 (January 1989), 167-69. he explained how very few African states went for linguistic
rationalization in the classical European sense of having only
References one language, but many others went for two other linguistic
strategies: a 2-language outcome, and what he came to call a
Fearon, James D. and David D. Laitin 1996. “Explaining Interethnic 3+1 solution, a formula he found in India and has applied to
Cooperation,” American Political Science Review 90:4 (Decem- other countries as well. In Identity in Formation: The Rus-
ber), 715-35.
sian-Speaking Populations in the Near Abroad (1998)–IF
Laitin, David D. 1986. Hegemony and Culture. Politics and Religious
Change among the Yoruba. Chicago: University of Chicago Press.
hereafter—the central issue is how to explain the emergence
Laitin, David D. 1998. Identity in Formation. The Russian-Speaking of a “conglomerate identity,” based primarily on linguistic
Populations in the Near Abroad. Ithaca: Cornell University Press. adaptation, among the Russian-speaking populations of Es-
Wedeen, Lisa. 2002. “Conceptualizing Culture: Possibilities for Po- tonia, Latvia, Ukraine, and Kazakhstan after the breakup of
litical Science,” American Political Science Review 96:4 (Decem- the Soviet Union.
ber), 713-28. His work on identity politics is, of course, not entirely
driven by language issues. In his second book, Hegemony
and Culture: Politics and Religious Change among the

20
 Qualitative Methods, Spring 2006

Yorubas (1986)–HC hereafter—Laitin asked why Yorubaland’s always entail a tradeoff. Each method in the social sciences
religious life was split between Muslims and Christians, but can handle some puzzles better, leaving others unresolved. It
Muslim-Christian differences were not the principal cleavage is not as helpful to say that ethnography is better than ratio-
in Yoruba politics. Finally, in two co-written articles with James nal choice, as to determine what their respective strengths
Fearon (Fearon and Laitin 1995, 2003), he has examined the are, and what they can handle best. Creative imagination can
consequences of ethnic diversity for peace and violence. In allow us to blunt the edges of the tradeoff, but the tradeoffs
the first joint article, Fearon and Laitin probed the conditions do not altogether disappear.
under which ethnic diversity would actually lead to peace, not In what follows, I will discuss this idea concerning each
violence; and in the second article, they asked whether ethnic of the methods Laitin has deployed: ethnography, surveys,
diversity was indeed a crucial determinant of civil wars, or and formal modeling. I will primarily use his work as an illus-
whether other factors were more significant. tration of tradeoffs, though in the process I will also discuss
The range of these questions makes Laitin a formidable other work. All three methods have their unique mix of
intellectual force, indeed a central figure, in the subfield of strengths and weaknesses, and we need to decipher what
ethnic politics. One can no longer write about language poli- kinds of questions can be best analyzed by each. It is both a
tics, identity formation, or ethnic peace and violence without question about the strategy and substance of research. Laitin
engaging his arguments. Moreover, his frequent forays into has not always been conscious of this point, nor has he con-
new empirical terrains add greatly to his output. His case ma- sistently followed it. Since he subscribes to the notion of
terials have come from Somalia, Nigeria, India, Sri Lanka, cumulation in social science, his recent critiques on purely
Catalonia, the Baltic Republics, Central Asia, and Ukraine. As methodological grounds seem quite puzzling and paradoxical.
my own research has become multi-country, it is now clear to Basically, the form his critique has taken and his commitment
me that developing intimacy with new political and cultural to the idea of cumulation are not logically consistent.
materials, a prerequisite for thoughtful work, is not easy. Con-
Shifts of Evidence, Shifts of Method
sequently, I have developed a strong admiration for those
who step beyond the existing zones of familiarity and develop Theoretical shifts, especially in light of changing evi-
ideas on that basis. Laitin has repeatedly allowed his intellec- dence, are quite common in scholarly life. Robert Dahl became
tual curiosity to migrate to newer lands, also sometimes lin- skeptical about the pluralist nature of American democracy,
guistically retooling himself. Many have worked on multiple once the tight hold of business over American politics be-
countries; very few have learned new languages. Laitin may came clear to him in the 1970s (Dahl 1982). And as the revolu-
have never left the subfield of ethnic politics, but his intellec- tions overthrowing Communism squarely questioned his as-
tual journeys within have a fox-like quality. sumptions about human behavior, Jon Elster developed seri-
Laitin’s substantive achievements are not the principal ous self-doubt about rational choice theories in the early 1990s
issues for this symposium. Rather, our focus is on his meth- (Elster 2000).
odological moves. The symposium seeks to assess the value These are examples of evidence-based theoretical shifts.
of Laitin’s methodological voyage from his early work based Are Laitin’s shifts evidence-driven, or method-driven? In his
primarily on ethnography, whether in Somalia or in Ile Ife in thoughtful essay for this symposium, Ted Hopf (2005) sug-
Yorubaland, Nigeria, to his work over the last decade and a gests that the reasons are methodological.
half, in which a rational choice stance has played a major role. But are methodological shifts entirely uncommon? In one
But before I proceed further, one should note that eth- of the famous interpretations of Marx’s overall body of work,
nography and rational choice are not the only methodological Louis Althusser argued that there was an “epistemological
alternatives which should be discussed here. Rational choice break” in Marx after his early work—before Das Kapital was
methodology, which does tend to rely heavily on formal logic written (Althusser 1969). According to Althusser, “early Marx”
and a priori assumptions, as opposed to ethnography which moved from the pseudo-scientific methods, when he gave too
is more empirically driven, is only one of the elements in Laitin’s much emphasis to human consciousness, to science later when
transformation. Some of his more recent work is heavily sta- he made it unambiguously clear that the structure of produc-
tistical, and we must draw a distinction between formal and tion determined the relations of production and, therefore,
statistical reasoning. If in “Explaining Interethnic Coopera- human consciousness. Epistemology is about the ways of
tion” (1995; EIC hereafter), assumptions, formal reasoning, generating knowledge. Willy nilly, it becomes inescapably
and equations abound, and only illustrations from the real methodological.
world are given but no systematic empirical evidence, in In short, method-based theoretical turns have precedence
“Ethnicity, Insurgency and Civil War” (2003; EICW hereaf- in intellectual history. Like Marx in his later works, Laitin to-
ter), there are only a few basic assumptions made and re- day tends to start with some universalist, apriori assump-
course to formal logic is minimal. Instead, existing theories of tions. Laitin, of course, does not leave it there, and unlike so
civil war are tested against a large statistical dataset. This many rational choice scholars, he does field work as well. But
kind of work is not ethnographic but it is still empirical, to be given that his survey questions are based on apriori theory,
differentiated from the rational choice tilt of the EIC. says Hopf, his empirical testing has become partial:
So how should we judge Laitin’s methodological trans-
I could not find a single case where ethnographic data
formation? My central argument in this essay is that methods

21
Qualitative Methods, Spring 2006

were advanced as evidence in contradiction to the sur- tics and in quite intense forms–open-ended, collecting narra-
vey data, or still less, as evidence to interpret the survey tives about them and postcoding them (once narratives have
data that were gathered. The intersubjective world of post- been collected) is perhaps one of the best ways to go. This
Soviet subjects was accorded far less evidentiary value survey strategy is different from following a theoretically de-
than the answers to survey questionnaires, questions termined finite-answers form and, therefore, a precodable for-
which were developed in light of a priori theories of the mat, as is typical of standard surveys. I am currently experi-
researcher, not from the ethnographic materials he gath- menting with such survey designs in my own research in four
ered (Hopf 2006, 18). countries. The upcoming results will show how far the rede-
signed survey technique works. Basically, those who survey
This way of generating knowledge, Hopf continues, is
do not collect narratives, and those who collect narratives do
“directly opposed” to the celebrated opening lines of Hege-
not survey, but there is no theoretical reason to see them as
mony and Culture (HC):
irreconcilable methodological adversaries. They can be sub-
When we try to interpret politics in Africa (or anywhere, stantially combined, blunting the edges of the tradeoff.4
of course!) in terms of our own structures of preference
Methods and Explanations
and categories of action, we learn less about either Afri-
cans or ourselves than we do by recognizing that our This said, another side of Hopf’s methodological critique
political understanding is not universal, but is contin- remains. Following his point about how method is deployed
gent on our sociological and historical experiences.” (HC, in HC as opposed to IF, one could also see the basic change
ix). in Laitin’s position on what structural transformations do to
human choices. In IF, Laitin argues that after a structural trans-
Hopf is insightfully identifying the difference between
formation brought about by the fall of the Soviet Union, the
surveys and ethnographies here. Though scholars select their
Russian-speakers in the Near Abroad calculated whether lin-
ethnographic sites for theoretical reasons, surveys are more
guistic assimilation was in their interest or not. In HC, Laitin
theory laden than ethnography. Survey questions are theo-
had said something dramatically different. The Yoruba did not
retically framed: only some questions are asked, not all pos-
calculate, when faced with the clear possibility of structural
sible questions. Ethnography facilitates a much more open-
transformation in their political arena. The fascinating ques-
ended “soaking and poking” and, as Hopf puts it, “it lets the
tion for Laitin’s inquiry during his Ile Ife field work became the
subject speak.”
following: why would the Yoruba still stick to a tribal (ances-
Hopf is right about this, but it is also worth asking whether
tral city) identity rather than a religious (Muslim vs. Christian)
ethnography has some limitations and surveys some advan-
one, even though a Civil War in Nigeria during 1967-70 at-
tages. Ethnography clearly allows us to deepen, but surveys
tempted to redefine Nigeria into a Muslim North and Christian
make broadening possible. Deepening and broadening as cat-
South, and again, when in the late 1970s, a debate on whether
egories of empirical observation generate trade-offs. Ethnog-
there should be a Federal Sharia Court of Appeal sought to do
raphy makes accuracy about a case or two possible in a way
the same? In their religious life, the Yoruba acted as Muslims
that surveys can not match; but surveys allow a broader range
and Christians, but they remained politically committed to
of observation, covering many more cases than ethnography
their tribal identity, refusing to react religiously to the cata-
can possibly do. I find Laitin’s belief in Identity in Formation
clysmic political events. Why?
(IF) that ethnography alone would not take him forward in the
Laitin explains why rational choice is unable to help him
Near Abroad well founded.
answer this question:
Though HC and IF seem to be asking the same broad
question–namely, what explains the choice of certain identi- Rational choice theorists...cannot tell us if ultimately but-
ties as opposed to others–the scale of observation is clearly ter is better than guns; it can tell us that at a certain point
different. In HC, Ile Ife was studied in depth and an assump- the production of a small number of guns will cost us a
tion was made that it was a microcosm of the entire Yorubaland. whole lot of butter, and at that point it is probably irratio-
In IF, unlike HC, four countries were observed. Surveys inevi- nal to produce more guns. Within a political structure,
tably had to be given greater weight than ethnography. Hopf individuals constantly make marginal decisions. [Ratio-
seems to suggest that Laitin should have done in Narva, his nal choice] theories can give us a grasp on how indi-
base in Estonia, what he did in Ile Ife, but seeing all of Estonia vidual political actors are likely to make choices within
through the prism of Narva, let alone three other societies, that structure.
was not the purpose of Laitin’s research. Nor might it have [Rational choice] theory cannot, however, handle
been a sensible methodological strategy.3 long-term and non-marginal decisions. When market
This does, however, lead to an important question: can structures are themselves threatened, and people must
surveys be designed in such a way that they pick up some of decide whether to work within the new structure or hold
the strengths of ethnography? A fuller discussion of this issue on to the old–without an opportunity for a marginal deci-
will lead us too far in a cognate area. It will suffice to note that sion–microeconomic theory is not applicable...Structural
making the survey questions about ambiguities, anxieties, fear transformations–changing the basic cleavage structure
and hopes—emotions that so often accompany identity poli- of a society–are not amenable to the tools of micro-eco-

22
 Qualitative Methods, Spring 2006

nomic theory... (HC, 148-9, parenthesis and emphasis would involve great uncertainty” (HC, 149). In IF, Russian-
added). speakers calculate even in times of great structural transfor-
mation–namely, the collapse of the Soviet Union.
Identity choice was not a marginal, but a structural deci-
The remarkable difference between the two methods is
sion. Rational choice arguments, therefore, were inapplicable.5
thus clear. In IF, Laitin starts with an apriori theory of indi-
What would apply instead?
vidual choices, as opposed to viewing choices as embedded
….Gramsci provides the solution….The model of he in a structural context, as was the case in HC. But is it neces-
gemonic control can help explain the reification of the sarily a problem? Is something lost in the process? Hopf (2006)
“tribe” in African politics–why that cleavage became the is sure about the great loss, as discussed below. I would like
dominant metaphor for political action and why it per- to answer the question in two ways, one of which goes in the
sisted... (HC, 150). direction of Hopf’s observation, but the other does not.
Hopf argues that if Laitin had allowed himself to be an
To explore how hegemony was created, Laitin then goes
ethnographer a la HC, he would have found that the absence
into history, fixing his gaze on the British colonial period,
of Russian ethnonational mobilization in the Near Abroad was
starting in the late 19th century:
consistent with a variable “omitted” from the surveys: Soviet
Claims based on religious identity were expunged from identity. Decades of history had made Soviet identity a lived
the political arena by British administrators...British everyday reality for Russians in the Near Abroad: “In under-
administration shied away from the promotion of Chris- standing themselves as multinational, as Soviet, the ethno-
tianity… British administrators…feared the revolution- national axis of identification was simply not available, or sa-
ary implications of religious fanaticism (HC, 154). lient, to millions of Russians living abroad. Therefore, they
did not understand themselves in opposition to Kazakhs,
Finally, Laitin sketches the impact of this decades-long
Ukrainians and Estonians, in precisely that way, just as the
principle of British rule on the Yoruba:
Yoruba did not mobilize along Christian and Muslim axes of
The idea that ancestral city represents ‘blood’ while reli- identification” (Hopf 2006, 18).
gion represents ‘choice’ is so deeply embedded into Notice the role structural context plays here in the exer-
commonsense thinking that experience and data demon- cise of individual choices. Some choices are simply not part of
strating otherwise fail to disabuse Yoruba people of this the institutionalized commonsense of politics because of how
‘truth’ (HC, 159). history played itself out. Hopf’s central methodological in-
sight is that an a priori theory led Laitin to formulate his
This is fascinating puzzle-solving. In many parts of the
survey questions in a way that ruled out this explanatory
world, religion is often not seen as a matter of choice, even
possibility, and an ethnographic soaking and poaking would
though it is in principle. Religion is more often seen as an
have made it transparent.
unchangeable reality inherited from forefathers. Moreover, in
If true, this is a very big conclusion, for it not only changes
other parts of the British Empire, the colonial authorities chose
how we explain the absence of Russian ethnonational mobili-
religion as a ruling strategy, for example in Northern Nigeria,
zation in the Near Abroad–as a result of each Russian calcu-
but in Yorubaland, they chose a different strategy, leaving a
lating how other Russians will behave, as Laitin does, or as a
quite different legacy. The distinctiveness of the institution-
result of a historically produced choice pattern, as Hopf pro-
alized commonsense of Yoruba politics, Laitin argues, is thus
poses–but it also shows that an important potential substan-
linked to the contingencies of colonial rule.
tial explanation is eliminated by a method relying on apriori
Let us now ask how the impact of a structural transforma-
assumptions.
tion on identities is handled in the Near Abroad. The unravel-
In the end, the area experts will have to judge the veracity
ing of the Soviet system is in many ways conceptually analo-
of either claim. What those of us doing surveys in different
gous to the Biafran Civil War. It ended a system as it existed,
parts of the world can do is ask whether questions about a
without making it clear what would replace it instead. In HC,
possible Soviet identity were included in Laitin’s question-
“the politicization of communal identities cannot be fully un-
naire–especially in an open-ended form which allows one to
derstood by examining the logic of individual choices” (103).
watch against excessive theoretical determination of survey
In IF, whether or not Russian speakers assimilate is based on
questions. The way Laitin’s survey questions are reported in
a strategic interaction with other Russian speakers, concep-
the appendix of IF does not make it clear whether he did ask
tualized as a matter of individual choice in a tipping model. In
questions about the possibility of a Soviet identity of Rus-
HC, calculations about identity, if made at all, were thought to
sians in the Near Abroad, and in what form.
take place in normal times, not in times of structural transfor-
If Hopf is right, he tellingly shows us the consequences
mation, for the latter was marked by a radical uncertainty about
of a method driven by a priori assumptions, but I also wish to
the future, making a cost-benefit calculus hard to practice.
argue that the same method has also generated some big ideas.
“The level of costs and benefits of different forms of political
EIC by Fearon and Laitin is another example of an argument
identification among the Yoruba is not at issue. For a Yoruba
based on apriori and universalist assumptions. It proposes
to reformulate his political identity on the basis of his religion
“in-group policing,” or “self-policing,” as a societal mecha-
nism of peace between diverse ethnic groups. The idea is de-

23
 Qualitative Methods, Spring 2006

ductively laid out, and Hopf is right that the theory is not ciated with Anderson (1983) and Gellner (1983), both of whom
empirically tested.6 attribute the emergence of nationalism to the rise of moder-
But the fact remains that it is a big and novel idea in the nity, and claim that nationalism was impossible before the
field, and it has a huge empirical potential. The existing theo- modern age, though their mechanisms are somewhat different
ries were either primordial (ancient hatreds), instrumental (po- (printing press and capitalism for Anderson, and industrial-
litical entrepreneurs mobilizing ethnicity for self-serving ends), ization for Gellner).
epochal (arrival of modernity), or institutional (consociational For a statistical testing of the modernist argument, Fearon
or liberal democracies; voting systems, etc).7 Using a simple and Laitin needed variables that could measure modernity, or
insight that ethnic groups can monitor their own group mem- proxy for it. Higher levels of per capita income became their
bers much more easily than those of a different ethnic group, proxy for modernity (Fearon and Laitin 2003, 78), and they
Fearon and Laitin turned it into a serious theoretical proposi- find that lower levels of per capita income increased the odds
tion, elaborated with game theory. of civil wars, not the other way round (83). Anderson and Gell-
I am empirically testing this theoretical idea in my current ner, they concluded, were wrong.
project in 18 cities across Indonesia, Malaysia, Sri Lanka, Ni- Were they? There are two conceptual problems in the
geria, and India, and the chances that some, if not all, of my way Fearon and Laitin formulated the test, partially inescap-
materials will bear it out are very high. My previous argu- able due to the requirements of regression analysis. First,
ments based on six cities in India had proposed a different both Anderson and Gellner made epochal arguments–argu-
societal mechanism of peace–interethnic civic engagement, ments that focused on a transformation of human conscious-
especially in organizations (Varshney 2002). As my research ness as it existed in the Middle Ages, once modernity arrived.
moves further, we will perhaps find out the conditions under Higher or lower income of countries today–or since 1945–is
which in-group policing works as a mechanism, and condi- quite beside the point. Pre-modern times may have had lower
tions under which interethnic engagement does. In short, even per capita incomes than modern times, but Anderson and
though the theory that I will develop will not be universal, it is Gellner also talk about print capitalism and industrialization.
the universalist assumptions and a deductive mode of theo- Their arguments are historically specific. The only way to
rizing that produced the idea of in-group policing. Fearon and test their arguments is to explore whether before the birth of
Laitin are certainly taking the world of knowledge forward. the printing press and/or industry, national consciousness
How does this discussion relate to my central argument? existed.9 Second, the arguments of Gellner and Anderson are
The same method that produced in-group policing as an idea about national identities, not nationalist civil wars. Having a
perhaps managed to rule out, if Hopf is right, the possibility national identity does not necessarily imply a hunger for war.
of a Soviet identity for the Russians in the Near Abroad. And Identities and wars are conceptually separable.
the method that identified the role of colonialism in producing Thus, regression analysis is a good way to test some
institutionalized commonsense in Yoruba politics cannot eas- theories, not all. Contemporary primary exports are easily quan-
ily tell us why despite British attempts at creating or freezing tifiable at a large-n level, but how does one quantify the extent
an institutionalized Hindu-Muslim divide through electoral of printing press penetration in a large number of cases in the
rules in India, South Indians managed to escape Hindu-Mus- 17th, 18th, and 19th centuries, and their consequences for hu-
lim cleavages, instead getting intra-Hindu caste divisions as man consciousness? And if a large-n dataset cannot be cre-
the master narrative of politics (Varshney 2002). Recourse to ated, how can one run regressions? In the absence of large
colonial practices resolves an intriguing puzzle about Yoruba datasets for the 17th through 19th centuries, one will have to
politics and generates a possible idea–contingencies of colo- discover a small number of critical cases that show the rise of
nial rule–for portability, without clinching it for all postcolonial national consciousness before the birth of the printing press
analytic sites. Historians had begun to zero in on this idea and/or industry. The debate between Gellner and Kedourie
elsewhere, but political scientists on the whole had not.8 goes precisely in that direction (Kedourie 1993, 136-144).
I hope it is now clear why we should not damn methods
Arguments and Statistical Testing
as intrinsically superior or inferior, neither ethnography, nor
Let us now turn to statistical methods such as regression surveys, nor for that matter deductive work, whether concep-
analysis, and examine the idea of tradeoffs. In EICW, Fearon tualized formally (as in game theory) or informally (as in the
and Laitin advance the argument about why civil wars occur writings, let us say, of a Rawls or an Elster). We should simply
in a very important way. A widely discussed recent theory, recognize the potential and limits of each method, and we
also based on statistical testing, had proposed that the odds should see whether the method proposed is suitable for the
of civil war were strikingly correlated with primary commodity problem at hand.10
exports (Collier and Hoeffler 2001). Primary commodities are This is true even in the natural sciences. Einstein famously
“lootable” commodities, and it is “greed” about these re- argued several decades back that physics and meteorology
sources that drive an insurgency, not “grievances” about eth- will neither have the same methods, nor the same degree of
nic discrimination, argued Collier and Hoeffler. Fearon and predictiveness. Physics typically studies a few variables in
Laitin disprove this argument conclusively. interaction, allowing parsimony and predictive accuracy.
But they run into some trouble when they test another Meteorology has so many variables that having more power-
argument—the so-called modernist view of nationalism asso- ful computers, which he saw coming, would only allow us to

24
 Qualitative Methods, Spring 2006

predict whether a broad area would get hurricanes in the week nores, first of all, the value of his own IF, where he studies
or so ahead, not predict that months in advance and if closer identity formation only in those parts of the former Soviet
to time, not predict whether a specific town or village would Union where conflict was absent or low in the immediate post-
be hit and with what intensity. Einstein’s reasoning was clear: Soviet phase: Kazakhstan, Estonia, Latvia, and Ukraine. He
it is the number of variables affecting the path and intensity of does not study the Chechen region, which had a lot of con-
a hurricane (or a snow storm) and their very complicated inter- flict and could have had very different identity outcomes.
action that was at issue here, not our computing powers. Me- Moreover, some of the most instructive social science
teorological problems cannot be reduced to a few variables, as work in recent decades selects even more on the dependent
in physics. Likewise, some problems in politics may well allow variable than Laitin does in IF. Sen’s Poverty and Famines
the parsimony of physics, but others may be more like meteo- (1981) and Bates’ Markets and States in Tropical Africa (1981)
rology, requiring very different kinds of conceptualization and are the best examples. Both are widely viewed as classics of
measurement. the development field, and justly so. Sen’s theory of famines
I should add that my argument about methodological was based on five famines; there were no half-famines or non-
choices entailing trade-offs is consistent with some new work famines in his research design. And as Rogowski pointed out
on methods, both of the quantitative and qualitative sort (Brady long ago (1995), Bates only studied agricultural stagnation in
and Collier 2004). It also underlines the value of methodologi- Sub-Saharan Africa, not cases of agricultural success.12
cal pluralism in the social sciences. Methodological pluralism It may be true that in most cases, it will be hard to clinch
is defensible not because anything goes, but because differ- an argument we want to make if we have no variation on the
ent methods will do different things well.11 dependent variable. But that is not the only way to contribute
to knowledge. Both Sen and Bates did two notable things.
Further Implications
First, they thoroughly undermined an existing conventional
For Laitin, this argument has some further implications. wisdom: food availability decline as a cause of famine, and
His movement from ethnography and case studies to sur- Africa’s cultural taste for leisure over work, producing back-
veys, formal reasoning, and statistical testing should allow ward bending supply curves instead of upward sloping ones,
him to deal with puzzles of a large variety. At the same time, his leading to agricultural stagnation.13 Second, they put a new
denunciation of other people’s work in his more recent schol- idea on the table for others to work with: entitlement failures
arly phase is puzzling and hugely paradoxical. Consider two as a cause of famine, as in Sen’s argument, and self-seeking
examples, one about case studies, another about selection on behavior of urban politicians, buying the rural rich through
the dependent variable. subsidies, and running policies that hurt the countryside as a
Laitin finds case studies unacceptable unless the study whole, as in Bates.14
of a single country is “transformed into a high-n research Research designs that select on the dependent variable
design, thereby increasing the scientific leverage” (Laitin 2003, can often do both of these, and that is reason enough to see
180). My argument with this reasoning is not that turning a them as contributions. Clinching theories in an ideal fashion
country study into a high-n design is wrong. Rather, I have is one way to contribute; undermining existing popular theo-
problems with Laitin’s insistence that that is the only way to ries and presenting elements of a new are another way. Inter-
save case studies, or by extension, ethnographies, which tend estingly, in IF, there is a point where Laitin says something
to study a village or a town. similar (Laitin 1998, 325), but he nonetheless attacks such stud-
Paradoxically, Laitin’s argument today amounts to de- ies elsewhere for they do not contribute to cumulation (Laitin
nouncing his old scholarly self, so evocatively in evidence in 2003, 179).
HC. More generally, Laitin’s insistence ignores the fact that
Conclusion
case studies can contribute to cumulation by producing in-
triguing ideas, even when the n is equal to one. This is true of Laitin has made remarkable contributions to our knowl-
critical cases, which even the more statistical view of King, edge, becoming a central figure in the subfield of ethnic poli-
Keohane, and Verba (1994) accepts as valid. To recall, critical tics. The kind of methodological evolution he has undergone
cases are those that, given theory, you would least expect to is also uncommon in the profession. For both of these reasons,
have outcomes that they do. With extensive low incomes and substantive and methodological, his scholarly output inspires
widespread illiteracy, India should not have been democratic, admiration. The admiration would be infinitely greater if he
but it is. With little sanctity of private contracts, virtually no could view methodological choices as consisting of tradeoffs
restrictions on the powers of the state, and a highly underde- and could thereby view work emanating from methods not cur-
veloped capital market, China should not have been an eco- rently favored by him as also contributing to the cumulation of
nomic dynamo for two and a half decades, but it has been. knowledge, for which the intellectual case is quite clear. It will
India and China thus become critical cases for the theories of also save him from self-inflicted paradoxes and contradictions.
democracy and economic growth. Theoretically unselfcon-
scious case studies are a problem, not country studies that Notes
are not transformed into a high-n design. 1
I would like to thank Anna Grzymala-Busse, Ira Katznelson,
Laitin’s critique of studies that select on the dependent David Laitin and Daniel Posner for some penetrating comments on an
variable is also oddly self-defeating (Laitin 2003, 179). It ig- earlier version of the argument presented here.

25
Qualitative Methods, Spring 2006
2
For a fuller development of this distinction, see Varshney (2003b). ods: Newsletter of the American Political Science Organization
See also Isaiah Berlin, 1979. Organized Section on Qualitative Methods 4:1 (Spring), 17-20.
3
Whether Laitin should have engaged in four ethnographies, select- Horowitz, Donald. 2001. The Deadly Ethnic Riot. Berkeley and Los
ing a central site in each country, is an important issue, and worth Angeles: University of California Press.
thinking about. But studying all four societies from the microcosm of Kedourie, Elie. 1993 [1961]. Nationalism. Oxford: Blackwell Pub-
Narva would not have been methodologically valid. lishers.
4
For some early thoughts on these lines, see Varshney, 2002, 19- King, Gary, Robert O. Keohane, and Sidney Verba. 1994. Designing
20. Social Inquiry. Princeton: Princeton University Press.
5
Whether this critique of rational choice theory is right is a differ- Laitin, David. 1977. Politics, Language and Thought: The Somali
ent point altogether. On what kind of rationality might apply to non- Experience. Chicago: University of Chicago Press.
marginal decisions like identity choices, see Sen (1982) and Varshney Laitin, David. 1986. Hegemony and Culture: Politics and Change
(2003a). Among the Yoruba. Chicago: University of Chicago Press.
6
Only examples of an approving sort are listed by Fearon and Laitin, David, 1998. Identity in Formation: The Russian Speaking
Laitin. See the empirical critique of Horowitz (2001, 475-6). Population in the Near Abroad. Ithaca: Cornell University Press.
7
A detailed elaboration of these traditions can be found in Varshney Laitin, David. 2003. “The Perestroikan Challenge to Social Science.”
(2002), Chapter 2. Politics and Society 31:1 (March), 163-184.
8
With the exception of Benedict Anderson (1983), especially in Lipton, Michael. 1977. Why Poor People Stay Poor: Urban Bias in
his account of “Creole Pioneers.” World Development. Cambridge: Harvard University Press.
9
Alternatively, can better proxies be developed for epochal argu- Rogowski, Ronald. 1995. “The Role of Theory and Anomaly in So-
ments? One should, of course, remain open to such possibilities. cial Scientific Inference.” American Political Science Review 89:2
10
In one of his recent essays, Laitin (2003) appears to have partly (June), 467-70.
moved in this direction. He argues for a tripartite method: formal Schultz, Theodore. 1980. Distortion of Agricultural Incentives. Bloom-
reasoning, statistical testing, and narratives. But it is unclear whether ington: Indiana University Press.
the ideal set forth has ever been realized, or can be. Moreover, consid- Sen, Amartya. 1983. Poverty and Famines. Oxford: Clarendon Press.
erable paradoxes in that position also remain, as discussed later. Sen, Amartya. 1982. “Rational Fools.” In Choice, Welfare and Mea-
11
On this matter, also see Laitin (2003) for a different view. surement (Cambridge: MIT Press).
12
To be fair to Bates, he does mention Kenya and Ivory Coast as Varshney, Ashutosh. 1995. Democracy, Development and the
cases of relative agricultural success, but that account comes at the Countryside. New York: Cambridge University Press.
end and is very brief. Basically, variation in outcomes is not the Varshney, Ashutosh. 2002. Ethnic Conflict and Civic Life: Hindus
centerpiece of the argument. and Muslims in India. New Haven: Yale University Press.
13
See Varshney, 1995, Ch. 2, for how popular the theories of Varshney, Ashutosh. 2003a. “Nationalism, Ethnic Conflict and Ra-
backward bending supply curves were. tionality.” Perspectives on Politics 1:1 (March), 85-99.
14
This idea did have a prior lineage in Lipton (1977) and Schultz Varshney, Ashutosh. 2003b. “Varshney and Bates; Two Views on
(1980), but Bates provided the most convincing links between poli- Seeing Like a State.” APSA-CP (Summer), 1492.
tics and economic outcomes. Lipton and Schultz assumed that the
urban bias of the political structure produced anti-agricultural out-
comes. Bates showed exactly the links worked. Ethnography and/or Rational Choice:
References A Response from David Laitin
Althusser, Louis. 1969. For Marx. London: Allen Lane.
Anderson, Benedict. 1983. Imagined Communities. London: Verso. David Laitin
Bates, Robert. 1981. Markets and States in Tropical Africa. Berkeley: Stanford University
University of California Press. dlaitin@stanford.edu
Berlin, Isaiah. 1979. Russian Thinkers. Henry Hardy, ed. New York:
Penguin.
As Ted Hopf presented his paper at the symposium held
Collier, Paul and Anke Hoeffler. 2001. “Greed and Grievance in Civil
War.” World Bank, Typescript. https://econ.worldbank.org/ at the 2005 annual APSA meeting, provocatively titled “Being
programs/library. David Laitin,” I felt as if I were in a chute on the 7½th floor of the
Dahl, Robert. 1982. Dilemmas of a Pluralist Democracy. New Ha- Marriott Wardman Park, ready to be discharged onto the New
ven: Yale University Press. Jersey turnpike. But I survived, enough so to offer the follow-
Elster, Jan. 2000. “Rational Choice History: A Case of Excessive ing remarks.
Ambition.” American Political Science Review 94:3 (September), The key substantive theme raised by the papers in the
685-95. symposium is the relationship of ethnography and a theory of
Fearon, James and David Laitin. 1995. “Explaining Interethnic Coop- purposive action. In the 1950s, the eminent anthropologist
eration.” American Political Science Review 90:4 (December), 715-
Frederic Barth encountered the work of John von Neumann
35.
Fearon, James and David Laitin. 2003. “Ethnicity, Insurgency and and Oskar Morgenstern, and immediately saw the deep impli-
Civil War.” American Political Science Review 97:1 (February), 75- cations of their game theory for anthropology. He then wrote a
90. game theoretic essay (Barth 1959) analyzing chieftaincy poli-
Gellner, Ernest. 1983. Nations and Nationalism. Ithaca: Cornell Uni- tics among the Pathans. This was one of the few lead balloons
versity Press. that Barth let fly in his distinguished career, and the anthropo-
Hopf, Ted. 2006. “Ethnography and Rational Choice in David Laitin: logical field has steered clear of game theory ever since. But
From Equality to Subordination to Absence.” Qualitative Meth- the scholarly relationship between game theory and ethnogra-

26