Research Corner: Leadership Affects Tie Between Diversity, Voluntary Turnover

Common sense suggests that leadership should affect the relation between diversity and group performance, but common sense is frequently wrong and invariably imprecise. Fortunately, scientific research can be used to evaluate conclusions based on common sense. A nice illustration is provided in a recent paper by Lisa Nishii of Cornell University and David Mayer of the University of Michigan.

As Nishii and Mayer point out, past research has found inconsistent relations between group diversity and performance. They propose that this inconsistency is due in part to differences in leader behavior.

Specifically, when managers develop good relationships with all their subordinates, the negative effects of diversity are diminished, eliminated or reversed. Although those of us in the diversity arena tend to emphasize the positive effects of diversity, it is important to acknowledge that diversity also may have negative effects. Nishii and Mayer point out that diversity can stimulate in-group favoritism, relational conflict, miscommunication, power imbalances, and the like, all of which may motivate group members to leave. Thus, they explore the relation between diversity and voluntary turnover—and how this relation can be affected by leader behavior.

Leader behavior can be analyzed and described in many ways. Nishii and Mayer adopt the well-established leader-member exchange, or LMX, theory of leadership. Briefly, LMX theory posits that leader-employee relations vary in their quality. With some employees, leaders develop high-quality social exchange relationships. With other employees, the relationships are of lower quality and limited to economic exchanges. Nishii and Mayer argue that negative effects of diversity should be diminished when the average level of LMX is high (on average, the leader has high quality relations with subordinates), when LMX differentiation is low (the leader has similar relations with all subordinates), and especially when the average level of LMX is high and the differentiation is low (the leader has high quality relations with all subordinates).

They argue that these effects should be true both for demographic diversity and for a form of diversity more directly relevant to task performance—tenure diversity. Tenure, of course, is associated with experience, information, perspectives and other individual differences relevant to task performance.

To test these predictions, Nishii and Mayer studied a large supermarket chain. They obtained data from 348 departments, with five to 64 respondents per department. Their measure of demographic diversity combined differences across departmental employees in terms of race, gender and age. Their measure of tenure diversity was based on differences in the number of years the employees had worked in the department. They used a standard measure of LMX to assess the leader’s relations with the employees. Finally, their dependent variable was the voluntary monthly turnover rate, averaged over seven months.

For demographic diversity, the results fully supported the predictions. As demographic diversity increased, so did turnover when the average LMX was relatively low (the leader did not have especially high quality relations with employees), but turnover was unrelated to demographic diversity when the average LMX was high. Similarly, turnover increased along with demographic diversity when LMX variability was high (the leader treated different employees differently), but turnover was unrelated to demographic diversity when LMX variability was relatively low. Finally, as predicted, when the LMX average was high and differentiation was low (e.g., leader had high-quality relations with all employees), turnover was unrelated to demographic diversity. On the other hand, high demographic diversity was associated with much higher levels of turnover when both the LMX mean and differentiation were high (when the leader had high-quality relations with most employees but low-quality relations with a few employees).

The results were somewhat different for tenure diversity. Interestingly, higher tenure diversity was associated with lower turnover when the average LMX was high (on average, the leader had high-quality relations with employees) and when the variability of LMX was low (the leader had similar relations with all employees).

This research has important implications for diversity practitioners, for several reasons. Often, when diversity practitioners discuss the link between leadership and diversity, they mean leader commitment to diversity programs, not the specific personal behaviors required to change the environment. This research speaks specifically to the explicit need for all leaders at all levels to create inclusive environments for everyone, without marginalizing anyone.


— Dr. David A. Kravitz is a professor of management at George Mason University. Dr. Renée Yuengling is a workplace diversity consultant in Washington, D.C.