Why Diversity Management Backfires (And How Firms Can Make it Work)

by Frank Dobbin and Alexandra Kalev

Corporate America has spent billions on diversity programs since the late 1960s, with little to show for it recently; the growth of white women and African-Americans in management stalled about two decades ago. We analyze data from more than 800 firms, over 30 years, to figure out why. Most firms try to control managerial bias through programs like diversity training and grievance procedures. This approach backfires, leading to reductions in diversity. Some firms try to engage managers in promoting diversity through programs like mentoring and special college recruitment. This approach works, leading to increases in diversity. The problem is that too many firms try to control managers and too few try to engage them.  

We have been developing an evidence-based approach to workplace equity programs by using statistical models to look at how different corporate reforms affect the numbers of white, African-American, Latino, and Asian-American men and women in management.1 Which reforms backfire, and which actually help? 

First, the bad news. Management’s knee-jerk response to the Civil Rights Act of 1964 was to try to prevent managers from acting on their biases by limiting their discretion. It is true that managerial bias is a key cause of inequality. Plenty of studies document the prevalence, and effects, of bias. But it turns out that not all approaches to addressing bias are effective. To stop discrimination, executives implemented quantitative metrics for structuring hiring and promotion decisions. Managers were asked to choose the highest scorers on paper-and-pencil job tests and annual performance ratings. This was supposed to stop managers from hiring their white, male cronies, and make them hire the best candidates. To hold managers accountable for discrimination, executives put in grievance procedures that would give employees a way to contest biased promotion, pay, and discharge decisions. Job tests, performance ratings, and grievance systems are supposed to constrain the discretion of middle managers, but they are often counterproductive. We find that job tests and grievance procedures lead to significant decreases in each of seven underrepresented groups in management: white women, black men and women, Latino men and women, and Asian-American men and women. Performance evaluations harmed only white women, but they didn’t help any of the other groups.

How could such well-intentioned, commonsensical, reforms backfire? Studies from sociologists and psychologists suggest that managers and workers rebel against controlling workplace rules and regulations.2 Indeed we find evidence that managers rebel against job tests, performance evaluations, and grievance procedures in one way or another. In interviews, human resources executives tell us that hiring managers, most of whom are white and male, often offer jobs to their cronies without making them take “mandatory” job tests, but make everyone else take the tests. Laboratory and field studies show that managers inflate performance ratings for white men, harming others’ chances of promotion. Managers often retaliate against people who bring grievances, causing them and their friends to look for new jobs. Retaliation is no minor issue. Of the 100,000 discrimination complaints brought to the Equal Employment Opportunity Commission every year, nearly 40% include charges of retaliation for the initial complaint.  

When reforms that backfire are popular among firms, they can do a lot of harm. About 44 percent of medium and large firms have job tests and more than 50 percent have civil rights grievance procedures. Over 90 percent have performance evaluation systems, although their popularity may begin to wane as they come under fire for being ineffective. There may be a role for performance review systems in firms, but as currently implemented, they aren’t helping to achieve equity.

Diversity training is another good-intentions-bad-outcomes reform. It has been heralded by diversity consultants and called a nuisance by managers. It may be more than just a nuisance. Like job tests, performance ratings, and grievance procedures, training is designed to control managerial bias. Companies such as Texaco, Coca-Cola, and CBS spend millions of dollars a year to train managers, usually in face-to-face workshops, with lectures, videos, and role-playing. These sessions are supposed to make people aware of bias and move them to control their biases. Forty percent of all medium and large firms offer training, as do virtually all Fortune-500 companies. Yet business-school studies of workplace training in safety and other fields show little effect, because training is infrequent and disconnected from day-to-day workplace routines.3 Our research suggests that diversity training can also incite managerial rebellion. The typical diversity program leads to reductions in white women, African-Americans, and Asian-Americans in management. One reason may be that, as some psychological studies show, rather than quashing bias, diversity training can activate it.

The good news is that programs that engage managers in promoting diversity, rather than trying to control them, have broad positive effects on actual workforce diversity. Studies from sociologists and psychologists suggest that when managers and workers are engaged in pursuing a goal they become committed to achieving it. Cognitive dissonance theory provides an explanation; when people engage in a behavior, their attitudes migrate toward support of that behavior so as to resolve cognitive dissonance. Ask someone to write an essay (the behavior) that conflicts with their beliefs (the attitude) and their beliefs will begin to change. Ask the most rambunctious kid in class to be in charge of discipline and order and it improves. When firms ask managers to try to solve the problem of equity, they see increases in diversity.

Firms engage managers in several different ways. They send managers on special recruitment visits to historically black colleges, women’s colleges, and professional events for women and minorities. Special recruitment programs encourage managers to get out there and find recruits. Many firms use management training programs to help women and minorities move up. These programs use existing managers to train a diverse pool of future managers. Our research shows that these things work. Special recruitment efforts substantively increase the share of women and minorities in management. Management training helps white women in particular, as do special efforts to get managers to nominate female trainees.

Mentoring programs similarly engage managers in solving the problem of retaining and promoting women and minorities. We see significant increases in women from all races and ethnic groups, and increases in the three groups of minority men in firms that initiate mentoring programs. Managers we interviewed told us that as mentors, they became champions for the protégés assigned to them, helping protégés to move up. Mentoring seems to be good for the mentors as well; a study following mentoring program participants at Sun Microsystems found that it helped the careers of protégés, but did even more to help the careers of mentors.4

One of the most effective things a firm can do to open opportunity is to appoint a diversity taskforce. Like special recruitment programs and mentoring programs, these engage managers. The typical taskforce brings together heads of different units for monthly meetings to look at data, discuss problems in recruitment and retention, and brainstorm for solutions. “We don’t have a single female engineer, what are we going to do about that?” They find local engineering schools that enroll female students and send recruiters. “We haven’t been able to hire Latinos in packing.” They set up referral programs so that the few Latinos they have can spread the word that the company is hiring. “We haven’t been able to keep our female managers.” They do exit interviews to find out whether caregivers would stay on if offered flexible hours. Chief executives usually encourage managers from different units to join the taskforce. In interviews with taskforce members, some told us they wanted to solve the problem and move on, but that by looking at the PeopleSoft data every month, they saw that the problem wasn’t going away. Being on a taskforce seemed to turn the average manager into a champion for diversity. And when the taskforce initiated a new recruitment, training or retention program, those taskforce members made sure the program was carried out in their own departments. Our statistical analyses show that diversity taskforces lead to increases in all seven of the underrepresented groups we examine.

The successful programs make managers part of the solution, rather than labeling them as an impediment to diversity. Our analyses point to two other ways in which firms can promote diversity. First, firms that put someone in charge of overseeing diversity—a full-time diversity manager—see bigger improvements than others. A diversity manager can have effects beyond the reforms she institutes, by asking middle managers to explain their hiring and promotion decisions and checking in with mentors and recruiters. Psychological experiments show that when people believe they may have to explain themselves to someone else, they view their own decisions from the perspective of an outsider and scrutinize those decisions for hints of bias. We find that the mere presence of a diversity manager makes most of the reforms work better, and erases some of the negative effects. Performance evaluations, for instance, do not backfire and have some positive effects when a diversity manager is in place.

Second, women from all races and ethnic groups gain ground in firms that put in work-family programs. Work-family reforms appear to operate as much by signaling that the firm supports workers who are juggling both work and family demands as by offering tangible benefits. Thus, childcare referral services are nearly as effective as subsidized on-site child care, despite the fact that referral services don’t provide a financial benefit and cost the employer little. Flextime and parental leave programs also make a difference.

So employers might consider ditching reforms designed to control managerial prejudice that have proved ineffective, or worse. They might monitor the implementation of other programs, such as performance evaluations, through full-time diversity managers. They might engage managers in solving the problem of equity by putting them on taskforces, sending them to colleges to recruit women and minorities, asking them to train new managers, and making them mentors to aspiring managers. And they might put in new work-life programs—even low-cost programs—to signal that they welcome workers juggling work and family demands. Companies that do those things see change.     


1. Frank Dobbin, Alexandra Kalev, and Erin Kelly, “Diversity Management in Corporate America,” Contexts 6.4 (2007): 21–28; Alexandra Kalev, Frank Dobbin, and Erin Kelly, “Best Practices or Best Guesses? Assessing the Efficacy of Corporate Affirmative Action and Diversity Policies,” American Sociological Review 71.4 (2006): 589–617; and Frank Dobbin, Daniel Schrage, and Alexandra Kalev, “Resisting the Iron Cage: The Diverse Effects of Bureaucratic Reforms to Promote Diversity,” Working paper, Department of Sociology, Harvard University.

2. Cheryl R. Kaiser, Brenda Major, Ines Jurcevic, Tessa L. Dover, Laura M. Brady, and Jenessa R. Shapiro, “Presumed Fair: Ironic Effects of Organizational Diversity Structures,” Journal of Personality and Social Psychology 104 (2013): 504–519; Timothy A. Judge, Carl J. Thoresen, Joyce E. Bono, and Gregory K. Patton, “The Job Satisfaction-Job Performance Relationship: A Qualitative and Quantitative Review,” Psychological Bulletin 127.3 (2001): 376–407; and Randy Hodson, “Workplace Behaviors: Good Soldiers, Smooth Operators, and Saboteurs,” Work and Occupations 18.3 (1991): 271–290.

 3. Donald L. Kirkpatrick and James D. Kirkpatrick, Evaluating Training Programs: The Four Levels (Berrett-Koehler, 2006).

 4. Capital Analytics, “Case Study: Sun Microsystems University Mentoring Program,” 2012, http://www.symbiontperformance.com/sitebuildercontent/sitebuilderfiles/sunmentoring-case-study-2012.pdf.