David I. Levine

Eugene E. and Catherine M. Trefethen Chair in Business Administration
Haas Economic Analysis and Policy Group
Haas Organizational Behavior and Industrial Relations Group

Press release on study

Contrary to popular thinking among some diversity consultants, employing workers of many different races has little effect on average turnover in a retail workplace, although employees do quit more often if fewer colleagues are the same race, according to a recently published case study by Jonathan Leonard and David I. Levine -- two long-time affiliates of the Institute of Industrial Felations.

In one of few studies to explore how workplace demographics affect employee behavior, Leonard and Levine examined more than 70,000 employees at more than 800 workplaces of a national retailer. They outlined the results of their study in an article titled "The Effect of Diversity on Turnover: A Large Case Study" in the July issue of the journal Industrial and Labor Relations Review.

"The most important takeaway is diversity itself doesn't matter much in terms of turnover for most groups of workers," says Leonard, chairman of the Haas Economic Analysis and Policy Group. "It suggests that people are, at least in this sector, pretty tolerant."

Leonard and Levine's findings contradict one argument by some diversity consultants who claim that having a gender and racially diverse workforce reduces turnover. Leonard and Levine also failed to find support for another line of thinking that argues that diverse workplaces experience more friction and thus require special training.

"We were interested in seeing whether in fact there really was an empirical basis for a lot of advice that is pretty commonplace in the diversity consulting industry," explains Leonard, who holds the George Quist Chair in Business Ethics at the Haas School. "We discovered that, at least for the retail sector, diversity itself is not a big driver of turnover."

Leonard and Levine, director of the Center for Health Research at UC Berkeley, note that a workplace is most diverse when each racial group and gender has an equal share of employment.

Birds of a Feather

At the same time, Leonard and Levine did find support for the old proverb "birds of a feather flock together" when they studied another facet of diversity -- racial isolation. They defined racial isolation as being in a numerical minority in a workplace, whether it's white, black, Hispanic, or Asian. For instance, in a group composed five black employees and two white employees, the white employees would be more racially isolated than their black colleagues.

"The problem for managers is that each new hire raises isolation for some groups at the same time that it decreases isolation for others," Leonard and Levine noted.

Another discouraging finding was that all minority groups were more likely to quit a workplace in which a greater proportion of employees were white, suggesting that diversity is difficult to sustain. "Managers can benefit by helping employees thrive in a world of racial diversity - a prescription that is easier to state than to implement," the authors wrote.

Gender Diversity

One surprising finding was that women seemed to dislike gender diversity. Women were slightly more likely to quit when the gender breakdown of their workplace was closer to 50% female and 50% male, and less likely when their workplace was less diverse, with either mostly female or mostly male employees.

Leonard and Levine point out that their results best apply to the low-wage service sector, which is already characterized by high turnover. But as a chief entry point into the workplace, that sector is significant, Leonard notes. "The fact that there seems to be a lot of tolerance in that entry-level job is good news," he says.

Other results of the study include:

  • Racial isolation from potential customers - not just coworkers - also increased turnover, the authors found. Black and Hispanic employees in particular were less likely to quit in heavily black and Hispanic communities, respectively.
  • White employees left more often in situations where there were fewer whites. Although the sample was two-thirds white, almost a quarter of the workplaces had a nonwhite majority.
  • There was evidence that blacks and Hispanics preferred each other to white coworkers. Black exits were particularly rapid when more of their coworkers were white or Asian, while Hispanic colleagues did not increase black employees' exit rate.
  • For Hispanics, unlike other groups, turnover was lowest with a mixture of Hispanics and others. Hispanics left stores with many whites or Asians, but were not more likely to leave stores with black coworkers.

For a copy of the article "The Effect of Diversity on Turnover: A Large Case Study" in Industrial and Labor Relations Review, please contact the Haas School of Business.

David I. Levine
Professor, Haas School of Business
Chair, Center for Health Research