Exploring Managerial Personality across Europe with the ASA Model: Predicting Within- and Between-Organization and Country Variation

Importance and Key Contribution This study replicates and extends the landmark study by Schneider, Smith, Taylor, and Fleenor (1998) establishing the homogeneity hypothesis (HH) of personality. The extension explores the... [ view full abstract ]

Importance and Key Contribution
This study replicates and extends the landmark study by Schneider, Smith, Taylor, and Fleenor (1998) establishing the homogeneity hypothesis (HH) of personality. The extension explores the hypothesis within work organizations, between work organizations, and between European countries. The study sample consists of 2,839 managers from 143 organizations and 32 European countries. Multilevel analyses are conducted to investigate the trait homogeneity of organizational and country personality profiles. Consistent with previous literature that posits significant between-organization variance, the results of our study show some variance between-organizations - and partially between-countries. However more variance can be explained at the individual level of analysis. Evaluation of within group agreement using rwg and ADM scores revealed that personality profiles within-countries and within-organizations are quite diverse.

Theoretical Base
Human beings are fascinated by diversity though obsessed by homogeneity. Management literature acknowledges that organizations grow and survive when diversity within-organization is recognized and enhanced, but there are clear examples that shows organizations calling for diversity while striving for homogeneity. Schneider (1987) has partially addressed this issue with the Attraction-Selection-Attrition (ASA) model. Schneider’s insight has shaped the research and application agenda regarding diversity and homogeneity for over 25 years. Changing understandings of the field, new analytical tools and questions about underlying assumptions surrounding personality and culture mandate the retesting of the model in new contexts.

The ASA model and the threat of homogeneity
ASA is a model based on three compelling processes. The first process (attraction) predicts that people will choose a particular organization based on the congruence between their personal characteristics and the attributes of that potential work organization; the organization will then select people with the attributes the organization desires (e.g., congruence with the organization). Finally, the attrition process predicts that people will leave an organization if they do not fit.
Research studies have accumulated examining the ASA cycle. By far one of the dominant ways of conceptualizing homogeneity was through the use of congruence with values or organizational culture. Less consensus exists around the ASA framework when “personal characteristics” are specified in terms of individual needs or personality (Bretz, Ash, and Dreher, 1989; Turban and Keon, 1993; Denton, 1989; Slaughter et al., 2005).
The most comprehensive test of the ASA cycle is the Schneider et al., (1998). The authors tested the model by looking at the difference between within-organization variability compared to between-organization variability. Organizational membership had a multivariate effect on the four MBTI dimensions, and accounted for 24% of the variance in MBTI scores, suggesting that organizations are rather homogeneous as far as personality characteristics of their managers is concerned.

The ASA model plays a central role in the study of person-environment fit, OB and HRM (Edwards, Cable, Williamson, Lambert and Shipp, 2006). A question that remains is whether homogeneity operates primarily at the individual level (attraction), the organizational level (selection), or both, as well as a third variable, relevant in Europe: the influence of the cultural diversity between-countries on homogeneity.
The first contribution of this study is that the organization system is considered a relevant unit of analysis, and commensurate measures are used. The paper’s second contribution is to examine HH through the lens of multilevel analysis, which account for the hierarchical structures in the working population. A third contribution is to examine a cross-cultural and non-student sample.

Research Questions
Schneider et al. (1998), tested HH through MANOVA conducted on a sample of about 13000 managers from 142 organizations in the US. Europe’s different cultural and historical background is reflected in economic differences and in managerial styles and practices. Conclusions from USA samples cannot automatically be assumed to accurately describe European realities (cf. Guest & Zijlstra, 2012; Warr & Pierce, 2004). For that reason it is relevant to test whether the ASA model applies to the European context.
From a theoretical perspective the mechanisms of the Attraction-Selection-Attrition model primarily operates at the level of the organization. And since labor mobility in Europe is not as high as in the USA we cannot expect that a similar mechanism plays a role at the country level. Thus one can expect that there should be less variance within a country than between-countries. This would validate the underlying process of the ASA-model (i.e. divergent validity). Organizations are nested within-countries, and we have incorporated this in our multilevel analysis by treating country variance at the group level.
The above results in the following hypothesis:
H1a: The personalities of managers will vary between-organizations.
Following Bradley-Geist and Landis (2012) the ASA hypothesis suggests that besides differences between-organizations, there should be homogeneity within-organizations and tests that evaluate between group differences are not necessarily establishing within group agreement (George and James, 1993). To this effect we formulate an additional hypothesis:
H1b: The personalities of managers, as measured with the MBTI, within an organization are more alike than a random sample of managers.
Central to the ASA framework is the concept of goals (Schneider & Goldstein, 1995). Some organizations pursue business-related goals, others a service for the community at large. Thus:
H1c: Membership in private or public organizations has an effect on personality.
Studies on personality have long sought to compare frameworks across cultures or countries . Different modal managers’ personality patterns in different countries might be associated with the creation of different organizational contexts and outcomes (Yen, Krumwiede & Sheu, 2002). Consequently, we wanted to investigate the explanatory power of country, as country might shape people personality throughout the socialization process. Thus:
H2a: The personalities of managers in the same organization will vary between-countries.
Previous studies ignored nesting of organizations within-countries.
With the use of multilevel analysis we moved beyond simple models of variance decomposition through examining explanatory variables at multiple levels of a data hierarchy.

Method
Sample
The sample consisted of 2839 European managers from 143 organizations spanning 32 European countries. The maximum number of managers from a single organization was 215, the minimum 3. The median number of participants per organization was 19.74. The maximum number of managers from each country was 603, the minimum 1.
Personality Measures
Each respondent completed the Myers-Briggs Type Indicator (Myers et al., 1998, as part of a leadership development program. The MBTI contains four separate bi-polar indices that reflect individual preferences for exercising perception and judgment: sensing-intuition (SN), thinking-feeling (TF), extraversion-introversion (EI) and judging-perceiving (JP). Data were analyzed as continuous dimensions using parametric statistics, and the scores from the four indicators were standardized so that they ranged from -30 to +30.
In addition we collected demographic information to use as control variables.

Analysis Procedures
The four personality indicators were analyzed jointly by using a multivariate multilevel model. The model is defined as a random intercept non-nested or cross-classified model with three levels of analysis: individual-, organizational- and country-level (model specification in Appendix1). Analyses were conducted using MLwiN, (Rasbash, Steele, Browne and Prosser, 2004).

Findings and Implications
H1a investigated the personalities of managers within an organization and whether they were more alike than a random sample of managers. The intercepts of the individual organization lines differ for the four indicators. Comparing managers’ variability within-organization and the standard errors we can assert that there is variation across individuals within-organizations. We also detected a significant variability of individuals between-organizations when the scores for the four MBTI variables were analyzed. The four components had a similar variance at the individual level (within-organization) but TF and SN, respectively, had larger variance between-organizations, and compared with the other two indicators.
The size of the ICC1 for EI shows that about the 1.43% of the total variance in EI and 8% of TF may be attributed to differences between-organizations [(2.866/(2.866 + 197.234) and 9.780/(9.780 + 108.183)].
The test statistic for the Normal test is calculated for EI as 2.866/1.410 = 2.032 (>1.96); we can reject the null hypothesis that there is no variance across organizations. The same applies to the other three indicators. This confirms that managers’ personalities within an organization are not more alike than a random sample of managers. Work sector and gender did not affect personality.
As for H2a, the intercepts of the individual organization lines differ; this signifies the presence of a variation between-countries for the four personality dimensions. There is variation across individuals within-organizations, between-organizations and between-countries and this is statistically significant for EI and JP. SN and TF have a larger amount of variation at the between-organizations level but at the country level the variation is not significant. A significant heterogeneity for the SN at the organizational level is consistent with previous empirical research (cf. Bradley-Geist & Landis, 2012).
The intercepts of the country lines differ for the four indicators and were statistically significant. This means that there was variation across individuals within-organizations; when the country was considered in the hierarchy, all the indicators but SN were significant.
Interesting the relationship across the indicators at the country level. The null-hypothesis of no between-countries differences can be accepted for TF and SN and rejected for EI and JP.

Full reference list from the author.

homogeneity hypothesis, ASA MODEL multilevel analysis, managerial development, personality. [ view full abstract ]

homogeneity hypothesis,
ASA MODEL
multilevel analysis,
managerial development,
personality.