Sacco, J.M., Scheu, C.R., Ryan, A.M., & Schmitt, N. 2003. An investigation of race and sex similarity effects in interviews: a multilevel appraoch to relational demography. Journal of Applied Psychology, 88, 852-865.
There is a large body of literature supporting the notion that demographic similarity affects important outcomes at work. However, demographic similarity is often measured at the individual level even though it occurs between pairs of individuals or within a group.
Similarity as a construct in psychological research: similarity judgments changes as we age and gain experience, also, similarity judgments change as the context changes. Applied research includes P-O fit. For instance, Kristof’s (1996) review indicated that supplemental fit was related to job choice, work attitudes, and the intention to remain on the job. Another theory is Schbeider’s (1987) influential attraction-selection-attrition framework. The similarity-attraction paradigm is complemented by social identity (Tajfel & Turner, 1986) and self-categorization theory (Turner, 1987), which proposed that our self-concepts are in part formed by the groups to which we think we belong.
Research on demographic similarity and evaluation: several studies prove evidence that demographic similarity effects fade over time or as people get to know each other. Collectively, this research and theory suggests that the interpersonal interactions that naturally occur over time mitigate the effects of demographic dissimilarity.
Operationalizing the demographic similarity effects: 1) interaction in the ANOVA framework, which needs the equal number of raters and ratees. 2) perceived similarity 3) Euclidian distance (being criticized on a host of conceptual and methodological gorunds), 4) interaction (suffered from repeated measurements).
The authors’ approach: HLM.
In Chen’s seminar: notes from his PPT:
1. When we have nested data, and when non-independence occurs, it means we really have 2 variance components:
2. Within-group variance = Level 1 variance = Individual-level variance (termed σ2)
• σ2 indicates the sum of squared deviations of individual responses from their respective group means
3. Between-group variance = Level 2 variance = Group-level variance (termed t00)
• t00 indicates the sum of squared deviations of group means from the grand (overall sample) mean
4. Unfortunately, in OLS regression, the error term (i.e. the standard error) focuses on the overall variance
5. Level 1 and Level 2 variance components are collapsed into a single (overall) error term
6. When testing nested and cross-level relationships, the overall error term is inappropriate
7. Can either upwardly or downwardly bias the standard errors (and thus the significance tests)!
8. When we test Level 1 – Level 1 relationships (i.e., nested individual-level relationships):• The correct error term associated with the Level 1 predictor should focus on s2
• The actual error term uses [t00 + s2] – too large!
• This means that Type I errors in OLS regression will be artificially deflated (i.e., we’ll be less likely to detect significant effects, or have lower power)!
9. When we test Level 2 – Level 1 relationships (i.e., cross-level relationships):• The correct error term associated with the Level 1 predictor should focus on t00 + (s2/n)
• The actual error term uses [(t00 + s2)/n] – too small!
• This means that Type I errors in OLS regression will be artificially inflated (i.e., we’ll be more likely to detect significant effects, or have higher power)!
In the performance evaluation context, one rater gives ratings for multiple ratees, the between effect is the rater (interviewer effect), and the within effect is the ratee (e.g., the effects of applicant-level characteristics on ratings).
Method: 1) steps: from the null modelàintercept variance sig. or not? à predict the intercept à slope variance sig. or not? à predict the slope.
Centering: 1) grand mean; 2) group mean, 3) no centering. ß No correct centering approach, they can yield different results and different interpretations.
This research studied the effects of race and sex similarity on ratings in one-on-one highly structured college recruiting interviews (708 interviewers and 12,203 applicants for 7 different job families). A series of hierarchical linear models provided no evidence for similarity effects, although the commonly used D-score and analysis-of-variance– based interaction approaches conducted at the individual level of analysis yielded different results. The disparate results demonstrate the importance of attending to nested data structures and levels of analysis issues more broadly. Practically, the results suggest that organizations using carefully administered highly structured interviews may not need to be concerned about bias due to the mismatch between interviewer and applicant race or sex.
Some useful reference:Chatman, J.A., & Flynn, F.J. 2001. The influence of demographic heterogeneity on the emergence and consequences of cooperative norms in work teams.
Academy of
Management Journal, 44:956-974.Riordan, C.M. (2000). Relational demography within groups: Past developments, contradictions, and new directions. In G.R. Ferris (Ed.), Research in personnel and human resources management (Vol.19, pp.131-174).
New York:JAI Press.Riordan, C.M., & Shore, L.M. 1997. Demographic diversity and employee attitudes: an empirical examination of relational demography within work units. Journal of Applied Psychology, 82:342-358.
