Abstract
We examine how race affects the employment status of subordinates following a job change by their immediate supervisors. We test whether racial homophily between a subordinate and a supervisor affects the odds of being let go. We also consider whether a racial match between an incoming head coach and assistant affects whether assistants retain their assistant coaching position. Data for these analyses come from a unique data set that explores what happens to 704 NCAA Division I college basketball assistant coaches after the head coach leaves the school. Logistic regression analyses confirm the benefit of working for a white head coach as this decreases the likelihood of being let go, compared to more positive outcomes such as following the coach to a new school, being internally promoted or retained after the head coach's departure. Furthermore, racial homophily with incoming head coaches insulates subordinates from having to search for new employment by increasing the likelihood of assistants being retained.
ACKNOWLEDGMENTS
Contributions of the authors are equal; authors are listed in alphabetical order. We thank Kayla Clarin and Dinur Blum for their research assistance.
NOTES
Notes
1 Given that directors of basketball operations typically make less than assistant coaches at the same school, this is technically a demotion. However, sensitivity analyses reveal that the results are substantively the same when we exclude these six cases; thus, we include them here under the status quo label.
2 Black or white coaches compose 99.5 percent of our sample. The remaining .5 percent were either Cuban, Puerto Rican, or Mexican, and were coded as nonwhite. While it is not our intention to ignore nonblack minorities, we follow CitationCunningham and Sagas (2005) and discuss our results in terms of a black–white dichotomy. Omitting these individuals did not affect results.
3 This analysis, therefore, compares each of the three employability resilience outcomes to being let go. We employed logistic regression instead of multinomial logistic regression because multinomial logistic regression violated the independence of irrelevant alternatives assumption and because logic pointed to different variables mattering for different outcomes. Consequently, our analyses preclude comparisons of each of the possible outcomes to one another.