uncommon sense in offering training to faculty and students, and even mandate it (29% of all schools require faculty to undergo training), is par- Why Doesn’t Diversity ticularly surprising given that the research on Training Work? the poor performance of training comes out of academia. Imagine university health cen- The Challenge for Industry ters continuing to prescribe vitamin C for the common cold. and Academia Corporate antibias training was stimu- lated by the civil rights movement of the Frank Dobbin and Alexandra Kalev 1950s and 1960s and legal reforms that movement brought about. Federal agen- cies took the lead, and by the end of 1971, tarbucks’ decision to put 175,000 work- the Social Security Administration had put Sers through diversity training on May 29, 50,000 staffers through racial bias training. in the wake of the widely publicized arrest By 1976, 60 percent of big companies of- of two black men in a Philadelphia store, fered equal-opportunity training. In the put diversity training back in the news. But 1980s, as Reagan tried to tear down affir- corporations and universities have been do- mative action regulations and appointed ing diversity training for decades. Nearly Clarence Thomas to run the Equal Employ- all Fortune 500 companies do training, and ment Opportunity Commission, trainers be- two-thirds of colleges and universities have gan to make a business case for what they training for faculty according to our 2016 called “diversity training.” They argued that survey of 670 schools. Most also put fresh- women and minorities would soon be the men through some sort of diversity session backbone of the workforce and that em- as part of orientation. Yet hundreds of studies ployers needed to figure out how to better dating back to the 1930s suggest that antibias incorporate them. By 2005, 65 percent of training does not reduce bias, alter behavior large firms offered diversity training. Con- or change the workplace. sultants have heralded training as essential We have been speaking to employers for increasing diversity, corporate counsel about this research for more than a decade, have advised that it is vital for fending off with the message that diversity training is likely the most expensive, and least effec- tive, diversity program around. But they per- sist, worried about the optics of getting rid of Yet hundreds of studies dating back training, concerned about litigation, unwill- to the 1930s suggest that anti-bias ing to take more difficult but consequential training doesn’t reduce bias, alter steps or simply in the thrall of glossy training materials and their purveyors. That colleges behavior, or change the workplace. and universities in the United States persist

48 anthropology Volume 10 • Number 2 • September 2018

Anthropology Now, 10:48–55, 2018 • Copyright © Taylor & Francis Group, LLC ISSN: 1942-8200 print / 1949-2901 online • https://doi.org/10.1080/19428200.2018.1493182 lawsuits and plaintiffs have asked for it in Most of these studies look at interven- most discrimination settlements.1 tions that mirror corporate and university Yet two-thirds of human resources spe- training in intensity and duration. One im- cialists report that diversity training does not portant study by Patricia Devine and col- have positive effects, and several field studies leagues suggests that a more extensive cur- have found no effect of diversity training on riculum, based in strategies proven effective women’s or minorities’ careers or on mana- in the lab, can reduce measured bias.5 That gerial diversity.2 These findings are not - sur 12-week intervention, which took the form prising. There is ample evidence that training of a college course and included a control alone does not change attitudes or behavior, group, worked best for people who were or not by much and not for long. In their re- concerned about discrimination and who view of 985 studies of antibias interventions, did the exercises — best when preaching Paluck and Green found little evidence that to the converted. We do not see employers training reduces bias. In their review of 31 jumping on this costly bandwagon. Con- organizational studies using pretest/posttest sider Starbucks, which closed 8,000 stores assessments or a control group, Kulik and for half a day to train 175,000 workers, at Roberson identified 27 that documented im- an estimated cost of $12 million in lost busi- proved knowledge of, or attitudes toward, ness alone. Starbucks hires 100,000 new diversity, but most found small, short-term workers each year, and to match the Devine improvements on one or two of the items intervention they would need a dozen half- measured. In their review of 39 similar stud- day sessions, every year, for more than half ies, Bezrukova, Joshi and Jehn identified only the workforce. Unlikely they would go that five that examined long-term effects on bias, far, even if the logistics of scaling a class- two showing positive effects, two negative, room intervention to 100,000 people could and one no effect.3 be worked out. A number of recent studies of antibias Despite the poor showing of antibias train- training used the implicit association test ing in academic studies, it remains the go-to (IAT) before and after to assess whether un- solution for corporate executives and univer- conscious bias can be affected by training. sity administrators facing public relations cri- A meta-analysis of 426 studies found weak ses, campus intolerance and slow progress on immediate effects on unconscious bias and diversifying the executive and faculty ranks. weaker effects on explicit bias. A side-by- Why is diversity training not more effective? side test of 17 interventions to reduce white If we can answer that question, perhaps we bias toward blacks found that eight reduced can fix it. Five different lines of research sug- unconscious bias, but in a follow-up exam- gest why it may fail. ining eight implicit bias interventions and First, short-term educational interventions one sham, all nine worked, suggesting that in general do not change people. This should subjects may have learned how to game come as no surprise to anthropologists. De- the bias test.4 Effects dissipated within a cades of research on workplace training of all few days. sorts suggests that by itself, training does not

Frank Dobbin and Alexandra Kalev The Trouble with Diversity Training 49 do much. Take workplace safety and health found the message of multiculturalism, training which, it stands to reason, employ- which is common in training, makes whites ees have an interest in paying attention to. feel excluded and reduces their support for Alone, it does little to change attitudes or be- diversity, relative to the message of color- havior. If you cannot train workers to attach blindness, which is rare these days. Whites the straps on their hard hats, it may be well- generally feel they will not be treated fairly nigh impossible to get them to give up biases in workplaces with prodiversity messages.10 that they have acquired over a lifetime of me- Perhaps this is why trainers frequently report dia exposure and real-world experience. hostility and resistance, and trainees often Second, some have argued that antib- leave “confused, angry, or with more ani- ias training activates stereotypes. Field and mosity toward” other groups.11 The trouble is, laboratory studies find that asking people to when African-Americans work with whites suppress stereotypes tends to reinforce them who take a color-blind stance (rather than a — making them more cognitively accessible multicultural stance), it alienates them, re- to people.6 Try not thinking about elephants. ducing their psychological engagement at Diversity training typically encourages peo- work and quite possibly reducing their likeli- ple to recognize and fight the stereotypes hood of staying on.12 So perhaps trainers can- they hold, and this may simply be counter- not win with a message of either multicultur- productive. alism or color-blindness. Third, recent research suggests that train- Fifth, we know from a large body of or- ing inspires unrealistic confidence in anti- ganizational research that people react discrimination programs, making employees negatively to efforts to control them. Job- complacent about their own biases. In the autonomy research finds that people resist lab, Castilla and Benard found that when external controls on their thoughts and be- experimenters described subjects’ employ- havior and perform poorly in their jobs when ers as nondiscriminatory, subjects did not they lack autonomy. Self-determination re- censor their own gender biases.7 Employees search shows that when organizations frame who go through diversity training may not, motivation for pursuing a goal as originating subsequently, take responsibility for avoid- internally, commitment rises, but when they ing discrimination. Kaiser and colleagues frame motivation as originating externally, re- found that when subjects are told that their bellion increases. Legault, Gutsell and Inzli- employers have prodiversity measures such cht found this to be true in the case of anti- as training, they presume that the workplace bias training. Kidder and colleagues showed is free of bias and react harshly to claims of that when diversity programs are introduced discrimination.8 More generally, in experi- with an external rationale — avoiding law- ments, the presence of workplace diversity suit — participants were more resistant than programs seems to blind employees to hard when they were introduced with an organi- evidence of discrimination.9 zational rationale — management needs. In Fourth, others find that training leaves experiments, whites resented external pres- whites feeling left out. Plaut and colleagues sure to control prejudice against blacks, and

50 anthropology Volume 10 • Number 2 • September 2018 when experimenters asked people to reduce backlash? Plaut and colleagues found that bias, they responded by increasing bias un- when multicultural curriculum was framed less they saw the desire to control prejudice as inclusive of the majority culture, ma- as voluntary.13 Thus Robin Ely and David jority group members responded better.16 Thomas found that a discrimination/fairness Perhaps the curriculum should emphasize framing of diversity efforts, which evokes le- multiculturalism but stress that the majority gal motives, is less effective than an integra- culture is an important part of that multi- tion/learning framing that evokes business culturalism. motives.14 Fourth, can we prevent trainees from feel- What is a university administrator or cor- ing that training is an effort to control their porate executive to do? Some researchers thoughts and actions, and from rebelling suggest remedies. On the one hand, they against the message? Legault and colleagues have addressed problematic features of train- found that by manipulating the framing of ing. On the other, they address evidence that training, trainers can influence whether train- training tends not to change workplaces un- ees see it as externally imposed or voluntarily less it is part of a broader effort, involving chosen.17 We expect that two common fea- multiple components. tures of diversity training — mandatory par- First, can we prevent antibias training ticipation and legal curriculum — will make from reinforcing stereotypes, rather than participants feel that an external power is suppressing them? Devine and colleagues trying to control their behavior. By mandat- ask their trainees to practice behaviors that ing participation, employers send the mes- increase contact with members of other sage that employees need to change, and the groups, and empathy for other groups — employer will require it. By emphasizing the these behavioral changes appear to be part law, employers send the message that exter- of the secret to avoiding the reinforcement nal government mandates are behind train- of stereotypes. Second, can we prevent ing. These features may lead employees to training from making managers complacent think that commitment to diversity is being because they believe that the organization coerced.18 has handled the problem of discrimination? One possibility would be to introduce the “moral licensing” literature as part of train- ing.15 It suggests that when people do some- We expect that two common thing good (e.g., attend training) they are features of diversity training likely to feel licensed to do something bad — mandatory participation and afterward (e.g., discriminate in hiring). This legal curriculum — will make might equip trainees to look out for the ef- participants feel that an external fect in their own behavior. power is trying to control their Third, can we prevent antibias train- behavior. ing about multiculturalism from making whites and men feel excluded and eliciting

Frank Dobbin and Alexandra Kalev The Trouble with Diversity Training 51 Our surveys show that 80% of corpora- tions with diversity training make it manda- The key to improving the effects tory, and 43% of colleges and universities of training is to make it part of a with training for faculty make it mandatory. wider program of change. 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 not only implicit biases, but structural dis- regulations and procedures to comply with crimination. The trick is to couple diversity them — the legal case for diversity — as do training with the right complementary mea- about 40 percent of university trainings. Per- sures. Our research shows that companies haps employers should cut the legal content most often couple it with the wrong comple- and make training voluntary, or give employ- mentary measures.21 The antidiscrimination ees a choice of different types of diversity measures that work best are those that en- training. gage decision makers in solving the prob- This begs a bigger question: if employers lem themselves. could design a diversity course that reduced We find that special college recruitment bias, would it reduce workplace discrimina- programs to identify women and minorities tion? There is reason to believe that it would — sending existing corporate managers out not. A recent meta-analysis suggests that to find new recruits — increase managerial change in unconscious bias does not lead diversity markedly. So do formal mentor- to change in discrimination. Discrimination ing programs, which pair existing managers may result from habits of mind and behav- with people a couple of rungs below them, ior, or organizational practices, that are not in different departments, who seek mentor- rooted in unconscious bias alone.20 This rein- ing and sponsorship. So do diversity task forces the view that employers cannot expect forces that bring together higher-ups in dif- training to change the workplace without ferent departments to look at the data on hir- making other changes. ing, retention, pay and promotion; identify The key to improving the effects of train- problems; brainstorm for solutions and bring ing is to make it part of a wider program of those back to their departments. So do man- change. That is what studies of workplace agement training programs that use existing training in other domains, such as health and managers to train aspiring managers. All of safety, have proven. In isolation, diversity these programs put existing higher-ups in training does not appear to be effective, and touch with people from different race/ethnic/ in many corporations, colleges and univer- gender groups who hope to move up. All of sities, training was for many years the only them help existing managers to understand diversity program in place. But large corpo- the contours of the problem. And all of them rations and big universities are developing seem to turn existing managers into champi- multipronged diversity initiatives that tackle ons of diversity.

52 anthropology Volume 10 • Number 2 • September 2018 By contrast, popular human resources Notes policies thought to reduce discrimination and promote diversity by controlling mana- 1. Frank Dobbin, Inventing Equal Opportunity gerial bias seem to backfire.22 Companies (Princeton: Princeton University Press, 2009). that establish formal hiring and promotion 2. Sara Rynes and Benson Rosen, “A Field criteria — through job tests and perfor- Survey of Factors Affecting the Adoption and Per- mance rating systems — to limit managerial ceived Success of Diversity Training,” Personnel Psychology 48, no. 2 (1995); J. Edward Kellough discrimination see reductions in managerial and Katherine C. Naff, “Responding to a Wake-up diversity. Formal civil rights grievance pro- Call: An Examination of Federal Agency Diversity cedures, which give employees a means to Management Programs,” Administration & Society pursue complaints of discrimination, also 36 (2004); Alexandra Kalev, Frank Dobbin and backfire because managers find them threat- Erin Kelly, “Best Practices or Best Guesses? Assess- ening. Our statistical analyses show that di- ing the Efficacy of Corporate Affirmative Action versity training can improve the effects of and Diversity Policies,” American Sociological Re- certain diversity programs, but employers view 71, no. 4 (2006); Frank Dobbin, Alexandra have to complement training with the right Kalev and Erin Kelly, “Diversity Management in programs — those that engage rather than Corporate America,” Contexts 6 (2007). alienate managers. 3. Elizabeth L. Paluck and Donald P. Green, Starbucks got mixed press coverage for “Prejudice Reduction: What Works? A Critical its mass diversity training event, with some Look at Evidence from the Field and the Labora- tory,” Annual Review of Psychology 60 (2009); experts, such as University of Virginia psy- Carol T. Kulik and Loriann Roberson, “Common chologist Brian Nosek, expressing skepti- Goals and Golden Opportunities: Evaluations of cism that that particular quick fix would Diversity Education in Academic and Organiza- 23 fix anything. But Starbucks says that this tional Settings,” Academy of Management Learn- is the first volley in what they expect to be ing and Education 7 (2008); Katerina Bezrukova, a long game. To their credit, Starbucks has Aparna Joshi and Karen A. Jehn, “Can We Teach tried to address racial bias before, with its Diversity? A Review of Diversity Trainings in Edu- 2015 campaign encouraging baristas to cational and Organizational Settings,” (2008), write “Race Together” on customers’ coffee Working Paper, Psychology Department, Santa cups, as a conversation starter. Starbucks Clara University. pulled the plug on that campaign after a 4. Patrick S. Forscher et al., “A Meta-Analy- couple of shots of media criticism and a sis of Procedures to Change Implicit Measures,” PsyArXiv (2018); Calvin K. Lai et al., “Reducing dollop of ridicule. Starbucks faces much Implicit Racial Preferences: II. Intervention Effec- the same challenge that university adminis- tiveness across Time,” Journal of Experimental Psy- trators face: what to do in an age in which chology 145, no. 8 (2016): 1001–16. diversity in executive and faculty ranks has 5. Patricia G. Devine, Patrick S. Forscher, An- been at a standstill for decades? Social sci- thony J. Austin and William T.L. Cox, “Long-Term ence research now gives us a pretty good Reduction in Implicit Race Bias: A Prejudice idea of what does not work and what re- Habit-Breaking Intervention,” Journal of Experi- mains promising. mental Social Psychology 48, no. 6 (2012).

Frank Dobbin and Alexandra Kalev The Trouble with Diversity Training 53 6. Mary Lou Egan and Marc Bendick, Jr., sity Training,” Journal of Organizational Behavior “Combining Multicultural Management and Di- 28 (2007); Rohini Anand and Mary-Frances Win- versity into One Course on Cultural Competence,” ters, “A Retrospective View of Corporate Diversity Academy of Management Learning & Education 7 Training from 1964 to the Present,” Academy of (2008); Adam D. Galinsky and Gordon B. Mos- Management Learning & Education 7 (2008). kowitz, “Perspective Taking: Decreasing Stereo- 12. Victoria C. Plaut, Kecia M. Thomas and type Expression, Stereotype Accessibility, and Matt J. Goren, “Is Multiculturalism or Color Blind- in-Group Favoritism,” Journal of Personality and ness Better for Minorities?” Psychological Science Social Psychology 78 (2000); C. Neil Macrae, Ga- 20, no. 4 (2009). len V. Bodenhausen, Alan B. Milne and Jolanda 13. Lisa Legault, Jennifer N. Gutsell and Mi- Jetten, “Out of Mind but Back in Sight: Stereotypes chael Inzlicht, “Ironic Effects of Antiprejudice on the Rebound,” Journal of Personality and So- Messages: How Motivational Interventions Can cial Psychology 67 (1994); Carol T. Kulik, Elissa L. Reduce (but Also Increase) Prejudice,” Psycho- Perry and Anne C. Bourhis, “Ironic Evaluation Pro- logical Science 22 (2011); Deborah L. Kidder, cesses: Effects of Thought Suppression on Evalua- Melenie J. Lankau, Donna Chrobot-Mason, Kelly tions of Older Job Applicants,” Journal of Organi- A. Mollica and Raymond A. Friedman, “Backlash zational Behavior 24 (2000). toward Diversity Initiatives: Examining the Im- 7. Emilio J. Castilla and Stephen Benard, “The pact of Diversity Program Justification, Personal Paradox of Meritocracy in Organizations,” Admin- and Group Outcomes,” International Journal of istrative Science Quarterly 55, no. 4 (2010). Conflict Management 15, no. 1 (2004); Patricia 8. Cheryl R. Kaiser, Brenda Major, Ines Ju- G. Devine et al., “The Regulation of Explicit and rcevic, Tessa L. Dover, Laura M. Brady and Jenessa Implicit Race Bias: The Role of Motivations to Re- R. Shapiro, “Presumed Fair: Ironic Effects of - Or spond without Prejudice,” Journal of Personality ganizational Diversity Structures,” Journal of Per- and Social Psychology 82 (2002). sonality and Social Psychology 104, no. 3 (2013). 14. Robin J. Ely and David A. Thomas, “Cul- 9. Laura M. Brady, Cheryl R. Kaiser, Brenda tural Diversity at Work: The Effects of Diversity Major and Teri A. Kirby, “It’s Fair for Us: Diversity Perspectives on Work Group Processes and Out- Structures Cause Women to Legitimize Discrimi- comes,” Administative Science Quarterly 46, no. nation,” Journal of Experimental Social Psychology 2 (2001). 57 (2015). 15. Eric Luis Uhlmann and Geoffrey L. Cohen, 10. Victoria C. Plaut, Flannery G. Garnett, ‘“I Think It, Therefore It’s True’: Effects of Self-Per- Laura E. Buffardi and Jeffrey Sanchez-Burks, ceived Objectivity on Hiring Discrimination,” Or- ‘“What About Me?’ Perceptions of Exclusion and ganizational Behavior and Human Decision Pro- Whites’ Reactions to Multiculturalism,” Journal cesses 104, no. 2 (2007); Benoit Monin and Dale of Personality and Social Psychology 101, no. 2 T. Miller, “Moral Credentials and the Expression of (2011); Tessa L. Dover, Brenda Major and Cheryl Prejudice,” Journal of Personality and Social Psy- R. Kaiser, “Members of High-Status Groups Are chology 81, no. 1 (2001). Threatened by Pro-Diversity Organizational Mes- 16. Plaut, ‘“What About Me?’” sages,” Journal of Experimental Social Psychology 17. Lisa Legault, Isabelle Green-Demers, Pro- 62 (2016). tius Grant and Joyce Chung, “On the Self-Regula- 11. Carol T. Kulik, Molly B. Pepper, Loriann tion of Implicit and Explicit Prejudice: A Self-De- Roberson and Sharon K. Parker, “The Rich Get termination Perspective,” Personality and Social Richer: Predicting Participation in Voluntary Diver- Psychology Bulletin 33 (2007).

54 anthropology Volume 10 • Number 2 • September 2018 18. Legault et al., “On the Self-Regulation of 22. Dobbin, Schrage and Kalev, “Rage against Implicit and Explicit Prejudice”; Yogesh Malhotra, the Iron Cage.” Dennis F. Galletta and Laurie J. Kirsch, “How En- 23. Rachel Abrams, Tiffany Hsu and John Eli- dogenous Motivations Influence User Intentions: gon, “Starbucks’s Tall Order: Tackle Systemic Rac- Beyond the Dichotomy of Extrinsic and Intrinsic ism in 4 Hours,” New York Times, May 29, 2018. User Motivations,” Journal of Management Infor- mation Systems 25, no. 1 (2008). 19. Kulik et al., “The Rich Get Richer.” Frank Dobbin is professor of sociology at Harvard. 20. Forscher et al., “A Meta-Analysis of Proce- dures”; John F. Dovidio, “Reducing Prejudice May Not Be Effective for Reducing Discrimination: Al- Alexandra Kalev is associate professor of sociol- ternative Social Psychological Approaches” (pre- ogy and anthropology at Tel Aviv University. sentation, “What Works to Reduce Discrimina- tion” Conference, Radcliffe Institute for Advanced Study at Harvard University, Cambridge, MA, April 20, 2018). 21. Frank Dobbin, Daniel Schrage and Alex- andra Kalev, “Rage against the Iron Cage: The Var- ied Effects of Bureaucratic Personnel Reforms on Diversity,” American Sociological Review 80, no. 5 (2015): 1014–44; Frank Dobbin and Alexandra Kalev, “Why Diversity Programs Fail,” Harvard Business Review 94, no. 7 (2016).

Frank Dobbin and Alexandra Kalev The Trouble with Diversity Training 55