Why Doesn't Diversity Training Work? The Challenge for Industry and Academia

Why Doesn't Diversity Training Work? The Challenge for Industry and Academia. Frank Dobbin & Alexandra Kalev.  Anthropology Now, Volume 10, 2018 - Issue 2. https://doi.org/10.1080/19428200.2018.1493182

Starbucks’ decision to put 175,000 workers through diversity training on May 29, in the wake of the widely publicized arrest of two black men in a Philadelphia store, put diversity training back in the news. But corporations and universities have been doing diversity training for decades. Nearly all Fortune 500 companies do training, and two-thirds of colleges and universities have training for faculty according to our 2016 survey of 670 schools. Most also put freshmen through some sort of diversity session as part of orientation. Yet hundreds of studies dating back to the 1930s suggest that antibias training does not reduce bias, alter behavior or change the workplace.

We have been speaking to employers about this research for more than a decade, with the message that diversity training is likely the most expensive, and least effective, diversity program around. But they persist, worried about the optics of getting rid of training, concerned about litigation, unwilling to take more difficult but consequential steps or simply in the thrall of glossy training materials and their purveyors. That colleges and universities in the United States persist in offering training to faculty and students, and even mandate it (29% of all schools require faculty to undergo training), is particularly surprising given that the research on the poor performance of training comes out of academia. Imagine university health centers continuing to prescribe vitamin C for the common cold.

Corporate antibias training was stimulated by the civil rights movement of the 1950s and 1960s and legal reforms that movement brought about. Federal agencies took the lead, and by the end of 1971, the Social Security Administration had put 50,000 staffers through racial bias training.  By 1976, 60 percent of big companies offered equal-opportunity training. In the 1980s, as Reagan tried to tear down affirmative action regulations and appointed Clarence Thomas to run the Equal Employment Opportunity Commission, trainers began to make a business case for what they called “diversity training.” They argued that women and minorities would soon be the backbone of the workforce and that employers needed to figure out how to better incorporate them. By 2005, 65 percent of large firms offered diversity training. Consultants have heralded training as essential for increasing diversity, corporate counsel have advised that it is vital for fending off lawsuits and plaintiffs have asked for it in most discrimination settlements.1

Yet two-thirds of human resources specialists report that diversity training does not have positive effects, and several field studies have found no effect of diversity training on women’s or minorities’ careers or on managerial diversity.2 These findings are not surprising.  There is ample evidence that training alone does not change attitudes or behavior, or not by much and not for long. In their review of 985 studies of antibias interventions, Paluck and Green found little evidence that training reduces bias. In their review of 31 organizational studies using pretest/posttest assessments or a control group, Kulik and Roberson identified 27 that documented improved knowledge of, or attitudes toward, diversity, but most found small, short-term improvements on one or two of the items measured. In their review of 39 similar studies, Bezrukova, Joshi and Jehn identified only five that examined long-term effects on bias, two showing positive effects, two negative, and one no effect.3

A number of recent studies of antibias training used the implicit association test (IAT) before and after to assess whether unconscious bias can be affected by training.  A meta-analysis of 426 studies found weak immediate effects on unconscious bias and weaker effects on explicit bias. A side-byside test of 17 interventions to reduce white bias toward blacks found that eight reduced unconscious bias, but in a follow-up examining eight implicit bias interventions and one sham, all nine worked, suggesting that subjects may have learned how to game the bias test.4 Effects dissipated within a few days.

Most of these studies look at interventions that mirror corporate and university training in intensity and duration. One important study by Patricia Devine and colleagues suggests that a more extensive curriculum, based in strategies proven effective in the lab, can reduce measured bias.5 That 12-week intervention, which took the form of a college course and included a control group, worked best for people who were concerned about discrimination and who did the exercises — best when preaching to the converted. We do not see employers jumping on this costly bandwagon. Consider Starbucks, which closed 8,000 stores for half a day to train 175,000 workers, at an estimated cost of $12 million in lost business alone. Starbucks hires 100,000 new workers each year, and to match the Devine intervention they would need a dozen halfday sessions, every year, for more than half the workforce. Unlikely they would go that far, even if the logistics of scaling a classroom intervention to 100,000 people could be worked out.

Despite the poor showing of antibias training in academic studies, it remains the go-to solution for corporate executives and university administrators facing public relations crises, campus intolerance and slow progress on diversifying the executive and faculty ranks.  Why is diversity training not more effective? If we can answer that question, perhaps we can fix it. Five different lines of research suggest why it may fail.

First, short-term educational interventions in general do not change people. This should come as no surprise to anthropologists. Decades of research on workplace training of all sorts suggests that by itself, training does not do much. Take workplace safety and health training which, it stands to reason, employees have an interest in paying attention to.  Alone, it does little to change attitudes or behavior.  If you cannot train workers to attach the straps on their hard hats, it may be wellnigh impossible to get them to give up biases that they have acquired over a lifetime of media exposure and real-world experience.  Second, some have argued that antibias training activates stereotypes. Field and laboratory studies find that asking people to suppress stereotypes tends to reinforce them — making them more cognitively accessible to people.6 Try not thinking about elephants.

Diversity training typically encourages people to recognize and fight the stereotypes they hold, and this may simply be counterproductive.  Third, recent research suggests that training inspires unrealistic confidence in antidiscrimination programs, making employees complacent about their own biases. In the lab, Castilla and Benard found that when experimenters described subjects’ employers as nondiscriminatory, subjects did not censor their own gender biases.7 Employees who go through diversity training may not, subsequently, take responsibility for avoiding discrimination. Kaiser and colleagues found that when subjects are told that their employers have prodiversity measures such as training, they presume that the workplace is free of bias and react harshly to claims of discrimination.8 More generally, in experiments, the presence of workplace diversity programs seems to blind employees to hard evidence of discrimination.9

Fourth, others find that training leaves whites feeling left out. Plaut and colleagues found the message of multiculturalism, which is common in training, makes whites feel excluded and reduces their support for diversity, relative to the message of colorblindness, which is rare these days. Whites generally feel they will not be treated fairly in workplaces with prodiversity messages.10

Perhaps this is why trainers frequently report hostility and resistance, and trainees often leave “confused, angry, or with more animosity toward” other groups.11 The trouble is, when African-Americans work with whites who take a color-blind stance (rather than a multicultural stance), it alienates them, reducing their psychological engagement at work and quite possibly reducing their likelihood of staying on.12 So perhaps trainers cannot win with a message of either multiculturalism or color-blindness.

Fifth, we know from a large body of organizational research that people react negatively to efforts to control them. Jobautonomy research finds that people resist external controls on their thoughts and behavior and perform poorly in their jobs when they lack autonomy. Self-determination research shows that when organizations frame motivation for pursuing a goal as originating internally, commitment rises, but when they frame motivation as originating externally, rebellion increases. Legault, Gutsell and Inzlicht found this to be true in the case of antibias training. Kidder and colleagues showed that when diversity programs are introduced with an external rationale — avoiding lawsuit — participants were more resistant than when they were introduced with an organizational rationale — management needs. In experiments, whites resented external pressure to control prejudice against blacks, and when experimenters asked people to reduce bias, they responded by increasing bias unless they saw the desire to control prejudice as voluntary.13 Thus Robin Ely and David Thomas found that a discrimination/fairness framing of diversity efforts, which evokes legal motives, is less effective than an integration/ learning framing that evokes business motives.14

What is a university administrator or corporate executive to do? Some researchers suggest remedies. On the one hand, they have addressed problematic features of training.  On the other, they address evidence that training tends not to change workplaces unless it is part of a broader effort, involving multiple components.

First, can we prevent antibias training from reinforcing stereotypes, rather than suppressing them? Devine and colleagues ask their trainees to practice behaviors that increase contact with members of other groups, and empathy for other groups — these behavioral changes appear to be part of the secret to avoiding the reinforcement of stereotypes. Second, can we prevent training from making managers complacent because they believe that the organization has handled the problem of discrimination?

One possibility would be to introduce the “moral licensing” literature as part of training.  15 It suggests that when people do something good (e.g., attend training) they are likely to feel licensed to do something bad afterward (e.g., discriminate in hiring). This might equip trainees to look out for the effect in their own behavior.

Third, can we prevent antibias training about multiculturalism from making whites and men feel excluded and eliciting backlash? Plaut and colleagues found that when multicultural curriculum was framed as inclusive of the majority culture, majority group members responded better.16 Perhaps the curriculum should emphasize multiculturalism but stress that the majority culture is an important part of that multiculturalism.

Fourth, can we prevent trainees from feeling that training is an effort to control their thoughts and actions, and from rebelling against the message? Legault and colleagues found that by manipulating the framing of training, trainers can influence whether trainees see it as externally imposed or voluntarily chosen.17 We expect that two common features of diversity training — mandatory participation and legal curriculum — will make participants feel that an external power is trying to control their behavior. By mandating participation, employers send the message that employees need to change, and the employer will require it. By emphasizing the law, employers send the message that external government mandates are behind training.  These features may lead employees to think that commitment to diversity is being coerced.18

We expect that two common features of diversity training — mandatory participation and legal curriculum — will make participants feel that an external power is trying to control their behavior.

Our surveys show that 80% of corporations with diversity training make it mandatory, and 43% of colleges and universities with training for faculty make it mandatory.  Employers mandate training in the belief that people hostile to the message will not attend voluntarily, but if we are right, forcing them to come will do more harm than good.19 About 75 percent of company trainings cover regulations and procedures to comply with them — the legal case for diversity — as do about 40 percent of university trainings. Perhaps employers should cut the legal content and make training voluntary, or give employees a choice of different types of diversity training.

This begs a bigger question: if employers could design a diversity course that reduced bias, would it reduce workplace discrimination? There is reason to believe that it would not. A recent meta-analysis suggests that change in unconscious bias does not lead to change in discrimination. Discrimination may result from habits of mind and behavior, or organizational practices, that are not rooted in unconscious bias alone.20 This reinforces the view that employers cannot expect training to change the workplace without making other changes.

The key to improving the effects of training is to make it part of a wider program of change. That is what studies of workplace training in other domains, such as health and safety, have proven. In isolation, diversity training does not appear to be effective, and in many corporations, colleges and universities, training was for many years the only diversity program in place. But large corporations and big universities are developing multipronged diversity initiatives that tackle not only implicit biases, but structural discrimination.  The trick is to couple diversity training with the right complementary measures.

Our research shows that companies most often couple it with the wrong complementary measures.21 The antidiscrimination measures that work best are those that engage decision makers in solving the problem themselves.

We find that special college recruitment programs to identify women and minorities — sending existing corporate managers out to find new recruits — increase managerial diversity markedly. So do formal mentoring programs, which pair existing managers with people a couple of rungs below them, in different departments, who seek mentoring and sponsorship. So do diversity task forces that bring together higher-ups in different departments to look at the data on hiring, retention, pay and promotion; identify problems; brainstorm for solutions and bring those back to their departments. So do management training programs that use existing managers to train aspiring managers. All of these programs put existing higher-ups in touch with people from different race/ethnic/ gender groups who hope to move up. All of them help existing managers to understand the contours of the problem. And all of them seem to turn existing managers into champions of diversity.

The key to improving the effects of training is to make it part of a wider program of change. By contrast, popular human resources policies thought to reduce discrimination and promote diversity by controlling managerial bias seem to backfire.22 Companies that establish formal hiring and promotion criteria — through job tests and performance rating systems — to limit managerial discrimination see reductions in managerial diversity. Formal civil rights grievance procedures, which give employees a means to pursue complaints of discrimination, also backfire because managers find them threatening.  Our statistical analyses show that diversity training can improve the effects of certain diversity programs, but employers have to complement training with the right programs — those that engage rather than alienate managers.

Starbucks got mixed press coverage for its mass diversity training event, with some experts, such as University of Virginia psychologist Brian Nosek, expressing skepticism that that particular quick fix would fix anything.23 But Starbucks says that this is the first volley in what they expect to be a long game. To their credit, Starbucks has tried to address racial bias before, with its 2015 campaign encouraging baristas to write “Race Together” on customers’ coffee cups, as a conversation starter. Starbucks pulled the plug on that campaign after a couple of shots of media criticism and a dollop of ridicule. Starbucks faces much the same challenge that university administrators face: what to do in an age in which diversity in executive and faculty ranks has been at a standstill for decades? Social science research now gives us a pretty good idea of what does not work and what remains promising.