When is More Better? The Effects of Racial Composition on Voluntary Turnover

Abstract

This study examines the relationship between racial composition and individual, voluntary turnover for minorities (i.e., Asians, blacks, and Hispanics) in a large organization. We present a critical test for two sets of contrasting predictions. The first draws on similarity attraction, social contact, and social identity theories to suggest that working with racially similar others enhances the work environment in terms of perceived career opportunities, mentoring relationships, and network ties, all of which would increase the likelihood of remaining in an organization. The contrasting predictions draw on group competition and group threat theories and propose that working with racially similar others might increase competition for resources and generate a backlash effect against minorities that would induce their turnover. We suggest the paradox that these two approaches might be compatible if the effect of demographic composition is nonlinear. Our data analyses show that individuals' likelihood of turnover decreases as the proportion of employees in a job from one's own race increases. Furthermore, this relationship is nonlinear: Members of minority groups with very small representation benefited more from the increased presence of their own race than minorities who already had a substantial presence. This finding suggests a potential backlash effect at higher minority proportions. Results also show that turnover decreases as the proportion of employees from one's own race increases in the level above an employee's job. Overall, these findings suggest that working with others of the same race reduces the likelihood of minority exits. Interestingly, the proportion of other minorities in a job has a marginally significant, negative effect on employees' voluntary turnover. Thus, increasing racial diversity from one's own race and other minorities appears to strengthen minority workforce retention.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 108 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.