ESCI Article
ESCI Article
ESCI Article
Findings from a pilot study to achieve a higher psychometric standard with the ECI
by Richard E. Boyatzis, PhD Professor in the Departments of Organizational Behavior and Psychology Case Western Reserve University Department of Human Resources, ESADE
Summary
The challenge
The Emotional Competence Inventory (ECI-2) and the ECI-U (University Version) show acceptable validity and reliability in a variety of studies; in fact, levels that are relatively rare for tests available to most practitioners and consultants. However, there is some criticism of our approach to conceptualizing and measuring EI in the professional research community. In some studies the competency scales do not appear valid as separate scales and the clusters do not differentiate themselves from each other. While this is of less concern to accredited users of the ECI, we decided to seek out a higher psychometric standard.
Our response
We started by re-conceptualizing the tests, both the ECI-2 and the ECI-U, as measures of social and emotional intelligence competencies. We then reviewed every item (360 0 survey question) and competency scale, applying factor analyses and revising them as necessary to ensure that they identified specific behaviors and were understandable and concise. This resulted in fewer competencies (12 instead of 18) and replaced the ECI-2 algorithm based on developmental levels with a measure based on consistency of behavior. The new instrument resulting from this work, the Emotional and Social Competency Inventory (ESCI), was piloted with a total of 116 participants and 1022 raters in the US and the UK.
The findings
The pilot study affirmed that the ESCI measures the behaviors that matter: those that contribute to effective performance. The psychometric standards achieved in the statistical analyses provided reassurance that the ESCI focuses on behaviors and the relationships between them that are observable, recognizable and distinct. The removal of developmental levels delivers a behavioral model which can be applied more satisfactorily to a wide range of work contexts, job roles and levels. The feedback package behind the ESCI will show participants how others experience their behavior in terms of the consistency with which they demonstrate emotional and social intelligence competencies. It will help participants to appreciate their strengths, to recognize how consistently they do certain things, and to identify what they can do more of to be even more effective.
The implications
The outcome of this work for the coaches and practitioners who are accredited to use the ECI is that there is now a choice of instruments. While we are more than pleased with the psychometric goal that the ESCI has attained, and satisfied that the revisions to the items and feedback package ensure even wider applicability, we realize that there are many reasons why practitioners may want to continue to use the ECI-2, or to use both in different circumstances.
www.haygroup.com
www.haygroup.com
Awareness competency clusters from the Relationship Management cluster (previously known as Social Skills). In particular, we wanted to differentiate those competencies that focus on the use of ones ability to understand and use emotions about oneself from the ability to understand and apply emotional understanding when dealing with others. The division between Emotional and Social Intelligence seemed to be the best way to address this. So the new tests are now called: The Emotional and Social Competency Inventory (ESCI) and The Emotional and Social Competency Inventory University Version (ESCI-U). However, there is some criticism of our approach to conceptualizing and measuring EI in the professional research community, and this has started to affect practitioners. While the detailed research programs and analyses using the ECI-2 and/or ECI-U continue to show reliability and validity, they also indicate instability: that is, the competency scales do not often appear valid as separate scales. In some of these studies the clusters do not differentiate themselves from each other. The major cause is that the scales and clusters are intercorrelated at a high level. This has not been an issue for coaches and practical users of the test, but it does raise questions in the academic community.
and a re-examination
In addition, weve examined every item (question) and competency scale against three criteria: Does it really differentiate from any of the other scales? Is it close to the related behavior that we observe in BEIs, videotapes and simulations? Can we drop this scale or item in order to achieve greater conciseness?
Some of the existing items and scales were good and some were less so (like many items in Initiative and Conflict Management). Some automatically collapsed into one larger scale in every data set with correlations of .85 to .95 (like Emotional Self-Awareness and Accurate Self-Assessment). So we decided which items to change and sought guidance in changing them by: going back to an assortment of actual BEIs, videotapes and simulations to reacquaint ourselves with the actual behavior that we, and others, saw. increasing the number of items per scale to 6 for the ESCI (5 plus 1 reverse-scored item in most scales) and 5 for the ESCI-U (because the ESCI-U is hand-scored, reverse-scored items will not be included in the ESCIU, the same as in the present ECI-U). We do not expand the length of the test, which is at a burdensome maximum already, because the total number of competency scales is reduced. revisiting the wording of the items for each scale to ensure considerable overlap in language and concepts. using statistical results from many analyses of doctoral students in their dissertations, as well as detailed exploratory and confirmatory factor analyses.
To increase the reliability 2 in terms of Exploratory and Confirmatory Factor Analysis and Item Response Theory, and in the hope of establishing better divergent validity3 of the scales and clusters, we needed to include more items per scale. Three data analyses were used in our work on scale redesign:
2 The extent to which the instrument yields consistent results with repeated use. 3
www.haygroup.com
The extent to which the instrument yields results that indicate different, distinguishable competencies and clusters of competencies.
1 2 3
The correlation matrix of the scales by Steve Wolff in the November, 2005 Technical Manual (n = 21,256); The Exploratory Factor Analysis by Boyatzis and Sala, 2004 (n ~ 6,500); The correlation matrix generated from the factor scores of the Confirmatory Factor Analysis scale correlations by Joan Manuel Battista, 2005 (n ~ 6,500).
To take into account the high intercorrelation among certain scales, and to help create more divergent validity among the scales and clusters (but keep the time burden of test completion the same as it is now) we reduced the number of competencies to 12. The ESCI competencies are now:
Emotional Self-Awareness: Recognizing ones emotions and their effects Emotional Self-Control: Keeping disruptive emotions and impulses in check Adaptability: Flexibility in handling change Achievement Orientation: Striving to improve or meeting a standard of excellence Positive Outlook: Persistence in pursuing goals despite obstacles and setbacks Empathy: Sensing others feelings and perspectives, and taking an active interest in their concerns Organizational Awareness: Reading a groups emotional currents and power relationships Coach and Mentor: Sensing others development needs and bolstering their abilities Inspirational Leadership: Inspiring and guiding individuals and groups Influence: Wielding effective tactics for persuasion Conflict Management: Negotiating and resolving disagreements Teamwork: Working with others toward shared goals. Creating group synergy in pursuing collective goals.
See Appendix 1 for an outline of the thinking behind the removal, merging or change of some of the ECI-2 competencies, and Appendix 2 for the revisions made to the ECI-U competencies. Removal of developmental levels
We dropped the competency developmental levels. This was a direct result of revising the wording of items for each scale to ensure overlap in language and concepts, and the consequence has been the ESCIs improved psychometric rigor. This does not affect the ESCI-U because the ECI-U doesnt use developmental levels. Competency levels can be helpful in capturing desired behaviors in a tailored competency model for a specific client. But when moving to generic models that should be valid across different roles and contexts, it becomes harder to apply them meaningfully for participants. Many accredited coaches also reported that their clients
www.haygroup.com
attempts to transform scores from the item responses to the levels often distracted from the true purpose of the coaching process. Why does the ECI pose this challenge? Those who find the developmental levels critical can continue to use the ECI-2.
The removal of developmental levels results in the removal of the algorithm that lies behind the ECI-2 and therefore the removal of a target level for each competency. Instead, competency scores are presented in terms of consistency of demonstration (from Never to Consistently) and summarized as an average score. In the future we are considering re-introducing tipping points (points on the scale that differentiate performance for each competency) once we have collected and analyzed a larger, representative sample with performance data. As a proxy for tipping points, norms for the ESCI will be established on the basis of frequency of demonstration of the competencies by the sample population participants will be able to compare their score against the 25th to 75th percentile range. Initially the sample population will consist of the pilot group, but it will grow with the use of the instrument. When sufficient data have been generated, norms will be re-calculated and re-applied.
Self-Awareness concerns knowing ones internal states, preferences, resources and intuitions . Self-Management refers to managing ones internal states, impulses and resources. Social Awareness refers to how people handle relationships and awareness of others feelings, needs and concerns. Relationship Management concerns the skill or adeptness at inducing desirable responses in others.
4 Although self-awareness serves as a foundation for the other clusters of competencies, ECI-2 scores often showed
lower scores for self-awareness than for the competencies that theoretically build upon it. This is because it is difficult for raters to assess a persons self-awareness. In the ESCI we attempted to correct this issue by re-wording the items to reflect behaviors that would be more visible
www.haygroup.com
Psychometric properties: Reliability The reliability of the scales in the ESCI remains comparable with the ECI-2. Table 1 shows the reliabilities for each competency and provides a comparison with the most similar ECI-2 competencies. These numbers are based on the pilot study, which contained 1022 raters.
Table 1: Cronbachs Alpha Reliability for ESCI Competencies
Cronbachs Alpha Reliability for ESCI Competencies Emotional Self-Awareness Achievement Orientation Adaptability Emotional Self-Control Positive Outlook Empathy Organizational Awareness Conflict Management Coach and Mentor Influence Inspirational Leadership Teamwork
5
Cronbachs Alpha Reliability5 for the most similar ECI-2 Competencies Emotional Self-Awareness Achievement Orientation Adaptability Emotional Self-Control Optimism Empathy Organizational Awareness Conflict Management Developing Others Influence Inspirational Leadership Teamwork & Collaboration
.83 .74 .76 .80 .76 .79 .76 .84 .83 .74 .79 .87
Total Others Ratings (N = 22,089), as referenced in the ECI Technical Manual November 2005.
Psychometric properties: Factor Analysis A principal axis Exploratory Factor Analysis with promax rotation 6 showed the factor analytic properties of the instrument to be outstanding. Table 2 shows a summary of the results of the factor analysis.
Table 2: Summary of Factor Analysis Results
Competency Emotional Self-Awareness Achievement Orientation Adaptability Emotional Self-Control Positive Outlook Empathy Organizational Awareness Conflict Management Coach and Mentor Influence Inspirational Leadership Teamwork Summary of Factor Analysis Results All questions loaded on expected factor 4 of 6 questions loaded on expected factor All questions loaded on expected factor All questions loaded on expected factor All questions loaded on expected factor All questions loaded on expected factor All questions loaded on expected factor All questions loaded on expected factor 5 of 6 questions loaded on expected factor 5 of 6 questions loaded on expected factor All questions loaded on expected factor All questions loaded on expected factor
Naming the new instrument ESCI recognizes the fact that it measures both emotional intelligence competencies (i.e. those in the self-awareness and self-management clusters) and social competencies (i.e. those in the social awareness cluster and the relationship management cluster). The pilot study has reaffirmed that the ESCI measures the behaviors that matter: those that contribute to emotionally and socially effective performance. The psychometric standards achieved in the statistical analyses provide reassurance that the ESCI focuses on behaviors and the relationship between them that are observable, recognizable and distinct.
The final items (a total of 72) were drawn from a larger number of piloted items (96), allowing us to select the most valid and reliable ones. In the final version of the ESCI: each competency scale has five items and most have an additional reverse-scored item. the items in each competency contain a key word or phrase to link all items in a scale. This is often related to the intent of the expressed competency or the central concept of the competency. This means that the instrument is easier to understand and complete, enhancing the credibility and usefulness of participants feedback data.
The removal of developmental levels (resulting from the re-wording of items) delivers a behavioral model which can be applied more satisfactorily to a wide range of work contexts, job roles and levels.
6
See G. Buchanan and M. Seligman (eds.), Explanatory Style (Hillsdale, NJ: Lawrence Erlbaum.)
www.haygroup.com
After completing the ESCI and seeking feedback from others, participants will receive an ESCI feedback report which, as with the ECI-2, provides the basis for a coaching conversation. The feedback package behind the ESCI will show participants how others experience their behavior in terms of the consistency with which they demonstrate emotional and social intelligence competencies. The format of the feedback report, and the removal of developmental levels, will free participants from the performance anxiety that can result from comparing their data to a set of target levels: levels which may or may not be relevant to their job roles. Instead, it encourages participants to appreciate their strengths, to recognize how consistently they do certain things, and to identify what they can do more of to be even more effective.
However, the ECI-2 will meet the needs of clients who: need to make comparisons between current feedback data and previous ECI feedback. want to compare composite data, across a group of participants, to previous ECI feedback for the same or different groups. find the ECI-2 developmental levels useful and applicable to their organization. need the credibility that comes from the large database behind the ECI-2, and are unwilling to wait for the ESCI database to grow sufficiently to meet their needs.
Because of the fundamental similarities between the ESCI and the ECI-2, minimal additional training will be required to equip practitioners and coaches with the knowledge required to deliver either instrument.
www.haygroup.com
www.haygroup.com
Dropping Change Catalyst This scale is highly correlated with Achievement Orientation. It aligns with it in the Factor Analyses. It seems to be a consequence of other competencies being used in a situation, or of a disposition to want to change things. Changing the name of Developing Others to Coach and Mentor The term Coach and Mentor more accurately reflects the essence of the Developing Others, when taking into account the changes made to the item wording for this competency. In addition, it seemed a better description of what effective leaders do.
www.haygroup.com
www.haygroup.com