Francesca Gino i Katherine Coffman

AUTHORS Professor, Harvard ; Associate professor,

Business School ; Harvard Business Schooi
DIVERSITY

Unconscious Bias
Training That Works

Harvard Business Review

114 September-October 2021 TOR MARK MARRIS
 Only 10% of training programs gave attendees strategies for reducing bias. Imagine a
weight-loss program that told participants to step on the scale and left it at that.

But conventional UB training isn’t working, research implemented rigorous UB programs across a variety of
suggests. In a 2019 meta-analysis of more than 490 studies industries. In this article we’11 share what we’ve learned A More Effective Model
involving some 80,000 people, the psychologist Patrick about how they’re leveraging a more practical approach to Successful UB training gives people concrete tools for chang-
Forscher and his colleagues found that UB training did not UB training. (Disclosure: One of us—Gino—has conducted ing their behavior. It helps them better understand others’
change biased behavior. Other studies have revealed that antibias training at organizations as a consultant, but they experiences and feel more motivated to be inclusive.
the training can backfire: Sending the message that biases include none of the companies featured in this article.) Consider an approach that Patricia Devine of the Univer-
sity of Wisconsin and her colleagues have developed, called
Across the globe, are involuntary and widespread—beyond our control, in
other words—can make people feel they’re unavoidable
and lead to more discrimination, not less. In fact, in a 2006 The Flaws in Conventional Approaches
"prejudice habit-breaking.” Like conventional UB training,
it teaches what implicit bias is, how it’s measured, and how
public outcrv over racist incidents review of more than 700 companies, Alexandra Kalev, Frank
Dobbin, and Erin Kelly showed that after UB training, the
Traditional UB training falls short in a number of ways.
In a recent survey we did of more than 500 working adults
it harms women and people of color. After being educated,
participants take the Implicit Association Test, which
inthe workplaceand mounting likelihood that Black men and women would advance in
organizations often decreased. It’s no wonder that women
from a wide range of U.S. organizations, three findings
stood out. First, most organizations, worried about a back-
demonstrates that we all fall prey to unconscious bias to a
degree, and then get feedback on their personal level of bias.
evidence of rhe cos: of employees' and people of color continue to report high levels of unfair
treatment at work.
lash, make UB training voluntary. As a result it’s embraced
only by people who are already familiar with bias and
Next they’re taught how to overcome bias through a combi-
nation of strategies. These include calling out stereotyped
feelingexcluded, leaders are striving The most effective UB training does more than increase
awareness of bias and its impact. Itteaches attendees to
interested in reducing it. Second, 91% of the respondents
indicated that their firms don’t collect information on
views, gathering more individualized information about
people, reflecting on counterstereotypical examples, adopt-
fo make their companies more manage their biases, change their behavior, and track
their progress. It gives them information that contradicts
the metrics they claim to care about, such as the race and
gender of new hires and recipients of promotions and
ing the perspectives of others, and increasing interactions
with different kinds of people. After learning about each
diverse, equifable, and inclusive. stereotypes and allows them to connect with people whose
experiences are different from theirs. And it’s not just a
employee recognition awards. It’s hard to improve some-
thing you’re not even tracking. Third, 87% of the respon­
strategy, participants are asked to come up with examples
of how they could use it in their own lives. They’re taught

Unconscious bias Training has onetime education session; it entails a longer journey and
structural changes to policies and operations—like the
dents indicated that at their firms UB training doesn’t go
much past explaining the science behind bias and the costs
that the strategies reinforce one another and that the more
they’re practiced, the more effective they will be.

plaved a majorroie in their efforls. standardization of hiring processes, the elimination of self-
assessments from performance reviews, and the institution
of discrimination in organizations. In fact, only 10% of
training programs gave attendees strategies for reducing
This approach really works. In a longitudinal experi­
ment, Devine and her colleagues had 292 college students

UB Training seeksto raiseawareness of incentives for improving diversity. Rather than providing
UB training as a check-the-box exercise, companies make
bias. Imagine a weight-loss program that told participants
to step on the scale and left it at that. The idea that we can
participate in prejudice habit-breaking with a focus on race.
Two weeks later the attendees noticed bias in others more

ofrhe mental shorfcufsfhaflead a real, long-term commitment to it because they think it’s
worthy and important.
reduce our bias simply by being aware of it is the fatal flaw
in most UB training. In fact, most programs end exactly
than students who hadn’t participated did, and were also
more likely to label that bias as wrong. Two years later the
where they should start. researchers went back to a subset of the students and found
tosnapiudgrnenls-often based In a study we conducted at a pharmaceutical company,
the results of two surveys we did, and the work of other

Previous spread: Mark Harris; woman: oonal/G etty Images

scholars, we see the positive impact this style of UB training
on has. Not only do employees report heightened awareness of
bias, but they also show less bias and prejudice weeks after
falents or diameter. Ifs goal is to the training. They start fmding the workplace to be more
inclusive—somewhere that differences are cherished rather
IDEA IN BRIEF

THEPROBLEM THE CAUSE THESOLUTION

reduce bias ia attitudes and than tolerated. And women, people of color, and people with
disabilities report feeling a greater sense of belonging and
Conventional training
to combat unconscious
This trainingaims to raise employees’
awareness of biases based on race or
Companies must go beyond raising awareness
and teach people to manage biases and change

behaviorsatworkjrom hinngand respect for their contributions.

To unpack what drives these positive changes, we inter-
bias and make the
workplace more diverse,
gender. But by also sending the message
that such biases are involuntary and
behavior. Firms should also collect data on
diversity, employees’perceptions, and training

promofiondecisionsto interactions viewed dozens of leaders—including chief human resource

officers, learning and development executives, and diver­
equitable, and inclusive
isn’t working.
widespread, it can make people feel that
they’re unavoidable.
effectiveness; introducé behavioral “nudges";
and rethink policies.

with cusfome rs and col leagues. sity, equity, and inclusion officers—at companies that have

Harvard Business Review

Harvard Business Review
116 September-October 2021
September-October 2021
 > Training should look mostly at scenarios where leaders and employees subtly exclude
others or downplay their contributions-one of the most widespread forms of bias.

that those who had participated were still more likely to hiring. Over seven months in 2001, matched pairs of 23-year- skin color or make assumptions about how they’11 behave unconscious biases are prevalent, and help employees
speak out against bias than students who had not. old college students were asked to apply to 350 entry-level because of their appearance. address them. Working with human resources, she found
Using similar techniques, the same researchers trained job openings in Milwaukee that had been randomly Focus on the potential for growth. Accordingto the that performance reviews were biased: Women received less
STEM faculty at the University of Wisconsin to reduce selected from a newspaper’s Sunday classified section and respondents in our recent survey of nearly 1,300 work- helpful feedback than men did. Why? The forms included a
gender bias. Afterward, departmental hiring patterns a state-sponsored online job site. The applicants were two ing adults in the United States, learning that the brain is self-assessment, which managers read before filling out the
began to change. Over the next two years the proportion of white students (one with a fictional criminal record and one malleable and capable of positive change is the single most feedback forms. Women, consistent with research findings,
female faculty hired in departments that had undergone without) and two Black students with the same profiles as efïective component of antibias training. Participants need were less likely to promote themselves and were harder on
the training rosé from 32% to 47%; in departments that the white applicants. The difference in the responses the to be taught that while bias is normal, it’s not acceptable or themselves than men were. By eliminating self-evaluations,
hadn’t received the training, the hiring of women remained applicants got was striking: Thirty-four percent of white unavoidable. HR was able to reduce bias against women in managers’
flat. Faculty members in participating departments—both students without records and 17% of white students with To help its employees and managers grow, the Canadian feedback. Examples like this are used in the company’s
women and men—reported feeling more comfortable records received callbacks. Only 14% of Black students with­ energy company Suncor encourages them to write in a UB training to concretely show how bias can be effectively
bringing up family responsibilities and even that colleagues out records got callbacks, and a mere 5% of Black students “reflection notebook” about any instance in which they saw addressed through changes in behavior.
valued their research and scholarship more, an independent with records did. Yet in a follow-up survey the managers themselves acting with bias. They’re also asked to think Break stereotypes. Stereotypes we hold about ourselves
survey conducted months after the workshop found. indicated that they had no racial preference. Denial is about the life experiences that shaped their biases. and others profoundly influence our behavior. For instance,
The companies we’ve seen get good results from UB train­ widespread, but if people don’t admit bias exists, they can’t Microsoft’s UB training prompts participants to think one of us (Coffman) has found that lack of confidence in
ing take an approach similar to Devine’s. In addition, they address it. more deeply about the examples of bias depicted in its their own talent leads women in heids stereotypically
have both individuals and the organization track and reflect One of the issues is that traditional UB training tends videos. For instance, one shows a team discussion of who is considered “male,” like technology, to behave in ways that
on progress and identify where broader change is needed. to focus on extreme cases of abuse and harassment, giving most suited to lead a project. After watching it, participants can jeopardize their success, such as suggesting fewer ideas,
We observed this at the pharmaceutical company, where we employees another easy out: Td never do that,” they say and are asked to indicate which of various statements made particularly when working with men, and not applying for
conducted a three-month study with about 400 people. Six tune out. While it’s important to cover the extremes, training during the discussion are valid: (1) Technical understanding roles for which they are qualified. But encouraging trainees
weeks after UB training there, participants reported greater should look mostly at scenarios where leaders and employees is important when leading a technical project; (2) Cynthia’s to present examples that defy stereotypes can reduce bias.
feelings of inclusion, showed less bias and prejudice, and subtly exclude others or downplay their contributions—one young children will make it harder for her to be fully com- In one experiment by Nilanjana Dasgupta and Anthony
made greater commitments to organizational change than of the most widespread and insidious forms of bias. mitted to the project; (3) Ravi’s introverted nature makes Greenwald, students exposed to images of admired African
the employees who hadn’t participated. Moreover, after the Microsoft’s online UB training, which is also available him an unsuitable project leader; and (4) Gerry’s relaxed Americans showed a weaker preference for white individ­
company worked to eliminate bias from its performance publicly, includes videos depicting various everyday work- demeanor would not serve this project’s goals. The partici­ uals. In another study that Dasgupta and Shaki Asgari did,
reviews, employees feit that they seemed fairer, and data place scenarios. In one, the only woman on a team tries pants learn that only the first statement is valid. There is no female college students were less likely to view leadership
analyses confirmed that their perception was accurate. to add her views and is interrupted multiple times until evidence that having children, being introverted, or having and math as male domains after encountering female fac­
Now let’s examine in greater depth the elements of suc- another member flnally notices and asks her to speak. a relaxed demeanor negatively affects a project leader’s ulty members in those departments at their school.
cessful UB training and the complementary measures that Highlighting common forms of bias is also helpful effectiveness. Exposure to counterstereotypical information reduces
should be taken to reinforce its goals. when addressing more-blatant discriminatory behavior. Provide examples of how to change behavior. Each prejudice and results in more-positive interpersonal inter­
Starbucks took an approach similar to Microsoft’s when section of Microsoft’s training includes a best practice for actions. Leaders of Coming, the manufacturer of high-tech
designing a new antibias training in reaction to a highly overcoming bias, such as “examine your assumptions.” glass and ceramics products, provide it as part of a broader
Stress That “You Hold the Power” publicized 2018 incident in one of its Philadelphia stores. Participants are told, “The next time you catch yourself initiative to address unconscious bias. Intersections, a
UB training needs to help employees act on their aware- One day in April, two African American entrepreneurs making a judgment about someone’s background or working learning site on Corning’s internal community platform,
■ ness of bias. The idea is to empower them to change arrived at a Starbucks store for a meeting. They sat without preference, ask yourself, could this be an asset? This is a hosts Collective Voices, a podcast series that showcases
while cutting off their escape route—the inner voice that ordering, waiting for a local businessman to join them. The simple way of reframing your thinking about a person or employees and leaders across the business discussing diver­
says, Tm born like this, and there’s nothing I can do about store manager asked them to either place an order or leave, a situation.” The training also teaches employees that they sity, equity, and inclusion and sharing personal experiences.
itHere are some ways to do that. and then called 911 when they did not. The police arrived can counteract bias when hiring or assigning projects by In some episodes employees and leaders talk about times
Overcome denial. Even if we’re aware of our bias, we’re and arrested them. The chain’s leadership responded clearly identifying the requirements of a role before evaluat- they behaved in a biased manner—for example, by stereo-
often ignorant of its extent and its consequences. A large- by closing all 8,000 of its U.S. stores for half a day of UB ing potential candidates. typing a colleague because of his or her affiliation with a cer-
scale audit of human resource managers conducted by training focused on race. During it, leaders gave concrete At one public relations company, the chief diversity tain political or racial group. Their vulnerability helps others
Devah Pager and Lincoln Quillian, for instance, found clear examples of how bias can show up in stores, such as when officer works with various divisions to examine practices examine their own biases. Some of the stories have opened
evidence of discrimination against Black candidates in employees treat customers differently on the basis of their like customer interactions and hiring, identify where up discussions of counterstereotypical examples, such as

Harvard Business Review Harvard Business Review

September-October 2021 September-October 2021 119
 DIVERSITY

women who are thriving in traditionally male roles, showing education and support. Employees are given time off to go
that there are ways for everyone to succeed in the organi- through each module in groups of three to five people and
zation. Since the launch of the podcast, which has become discuss the questions it poses. Each session is 30 minutes
quite popular, internal surveys have revealed that employ­ for baristas and other nonmanagement employees, and
ees are more comfortable talking about their mistaken views 60 minutes for managers and above.
and flnd themselves relying less on stereotypes. Discussions can also occur virtually. On Corning’s Inter-
sections site, employees talk about issues related to inclu­
sion in an online forum, where a moderator is available to
Create Empathy answer questions and provide resources.
Research shows that we have less empathy for people
who seem different from us and are likely to treat them
Encourage Interactions Among
S
worse as a result. That’s why connecting with others through
empathy can improve our interactions across racial, gender, People from Different Groups
and other differences. Let’s look at some ways to nurture
empathy. These can be a powerful antidote to bias. Research
Offer opportunities to take the perspective of others. shows that white people who’ve had few interracial encoun-
We d'on’t put ourselves in someone else’s shoes naturally, ters often experience anxiety when interacting with Black
much research finds, but doing so can lead to greater people and try to avoid them altogether. But forming
interest in others’ welfare and more-positive relationships. relationships with members of other groups can widen our
In laboratory studies, instructing participants to take social networks, decrease our stress around people who are
another person’s point of view has been found to reduce bias different from us, and reduce our prejudices. Here are some
against stigmatized groups, such as African Americans, and effective methods for building them.
to suppress unconscious prejudices. In one study a group Expand inner circles. Training sessions themselves
of white Americans watched clips from the movie The Joy can help people get to know colleagues who are unlike

Mark Harris; braln; sdecoret/G etty Images; eye: m edlar/G etty Images; x-ray; oonal/G etty Images
Luck Club and were asked to put themselves in the place of them, even when the sessions are virtual. In our survey of
the Chinese American heroine, June. In comparison with a nearly 1,300 working adults, respondents said they bene-
control group, participants later showed less implicit prej- fited from the opportunities that training offered to interact
udice toward “outgroups” (people who were not like them). with diverse colleagues and to examine whom they had
During the UB training at Starbucks, attendees who were contact
is obviously most
withtied tooften. The success
how diverse of this is
a workforce kind of effort
in the first
shown videos in which employees from minority groups told place. The fact that it can cause discomfort should not
their stories were asked to adopt the perspective of those be a barrier. Research has shown that moderate discomfort
colleagues. is a critical catalyst for the introspection that can lead to
Hold small group discussions. These create opportu­ more-egalitarian behavior.
nities for people to learn about others’ views and expe- At Starbucks pulse surveys and interviews with employ­
riences. About every seven weeks, as part of an ongoing ees revealed that the discussions about the UB modules,
training called the Third Place development series, Star­ which were open to all employees, allowed colleagues with
bucks provides new guided UB learning and discussion different experiences and backgrounds to learn about one
modules, which are delivered on iPads in its stores for retail another and create new connections and made workers
employees and on the company’s intranet for the rest of the more empathetic.
workforce. The topics are crowdsourced from leaders of dif­ Nurture curiosity. The natural desire to acquire new
ferent departments, including human resources, employee knowledge and information can reduce prejudice and
development, and inclusion and diversity. But employees discriminatory behavior, research by Gino finds. (See "The
also can suggest areas where they feel they could use more Business Case for Curiosity,” September-October 2018.)

Harvard Business Review

120 September-October 2021
Harvard Business Review
September-October 2021
 People should reflect oo how they spend their time ai work and with whom.
When they’re handing out assignments, do their choices indicate bias?
DIVERSITY

Curiosity prompts us to get to know our colleagues better gain a reputation for always striving to be inclusive. Some time, and money. Here’s how organizations can provide candidates. In these ways organizations can ensure that
rather than make assumptions. UB training can encourage might buy into an organizational goal such as better serving such support. training lessons influence employees’ everyday behavior.
it by having people work together in diverse teams. At its diverse customers by creating a diverse workforce. Or they Build the foundations. First, organizations can collect Review and rethink policies. The leaders of Starbucks
UB training the multinational professional services firm EY may want to address UB simply because it’s the right thing data on the representation and dispersion of people from revised store policies that they believed led employees in
assigns colleagues from varying cultural backgrounds to to do. No matter what the goal, choosing a specific reason for different groups across the business, employees’ perceptions Philadelphia to call the police on the two Black men. The
teams and encourages them to ask one another questions, a commitment is a first step toward improvement. of inclusion, and where diversity-related process failures company issued clear guidelines stating that everyone was
find out what bias means to each person, and explore how to Choosing to commit is easier when participants see might be occurring (such as during hiring or performance welcome to spend time in its stores, with or without making
overcome it together. evidence of how unconscious bias affects their work. For reviews). That data will suggest which training topics might a purchase. And employees facing a challenging situation
Starbucks organizes regular sessions featuring outside instance, a large company that Gino worked with showed have the greatest impact on employees. Leaders can also were encouraged to move beyond their gut reaction by
speakers, from successful Black entrepreneurs to well- racial bias in performance evaluations. After HR directors establish a committee to oversee and report on progress consulting a checklist, considering the context, and seeking
regarded influencers who belong to other minority groups reviewed the data, they and the company’s senior leaders toward diversity, equity, and inclusion goals. These endeav- advice from others before taking action. Store managers
or have disabilities. The aim is to spark interest in people committed to elimiriating unfair practices. ors will need dependable funding. At Starbucks, which has were taught to ask, “Would I take this action with any cus­
whose experiences employees may not be familiar with. Find a mentor and solicit feedback. We often lack made a multiyear financial commitment to reducing bias, tomer in the same circumstances?”
Attendees are invited to practice asking curious questions. a clear sense of our own bias and how it affects others. One Molly Hill, the company’s vice president of learning, devel- As Starbucks recognized, UB training alone can’t stamp
Urge employees to track their interactions. UB train­ male leader we coached learned from a trusted colleague opment, and partner experience, says it speaks volumes out bias. Systemic changes are needed as well. Leaders
ing should encourage people to reflect on how they spend that in meetings he frequently interrupted people—primar- that her team doesn’t need to ask for a training budget every should revise long-standing practices that unfairly disad-
their time at work and with whom. When they’re handing ily women—and often attributed women’s ideas to their year—the money is allocated to the initiative by default. vantage certain groups, such as relying on unstructured
out assignments, do their choices indicate bias? Whom do male colleagues. Once his eyes had been opened, he began Measure the effectiveness of UB training. This is interviews or self-assessments. When managers at the
they gravitate toward in brainstorming sessions and sponta- paying closer attention to how he managed meetings. critical to improving the training over time. It involves, financial services firm realized there was a gender bias in
neous conversations? At UB training, participants might identify a mentor first, gathering data on engagement with the training itself. the way they assigned deals, they instituted a regular review
When Gino coached a group of leaders on unconscious who could observe their behavior for bias or advise them Microsoft does that with participant surveys and by study- to ensure that it stopped happening.
bias, she asked them to review their calendars to see whom on how to solicit feedback from others. Team members may ing what makes employees, teams, and units most likely to
they had met with in the previous month and whom they feel more comfortable providing input anonymously or consume UB training content. Starbucks similarly assesses leaders’ desire for their companies to be more diverse,
invited to meetings, and to think back about whom they may appoint one person to monitor meetings for signs of engagement with antibias materials through pulse surveys equitable, and inclusive has never seemed stronger. But
called on to speak during those meetings. The data was bias. Accepting that we’re biased isn’t easy, but learning of employees. conventional UB training programs aren’t delivering the
eye-opening: People of color were not invited to meetings from feedback is key to becoming more inclusive. Second, organizations must track the outcomes they’re changes they’re supposed to produce. By following our blue-
as often as white individuals, were called on less frequently Track improvement. It’s crucial to hold people account- trying to change. To promote improvements, companies like print, organizations can create programs that inspire people
in meetings they did attend, and met informally with their able by monitoring whether behaviors truly change over Microsoft and Coming publish demographic employment to more courageously examine and improve their behavior.
bosses less often. And when managers at a financial services time. When the leaders at the pharmaceutical company data in public reports each year. Starbucks tracks customer By replacing superficial, one-shot training with longer-term
firm analyzed investment opportunities and deals they’d tracked their practices, they saw differences in their promo­ engagement with employees in different stores, asking efforts that do a better job of helping people understand
passed on to colleagues, they found evidence of gender bias: tion patterns during the two years after their UB training. whether their efforts are improving customer experiences. their own unconscious biases and see how to overcome them
More men than women were offered those opportunities. Similarly, the managers at the financial services firm were Asking employees directly affected by bias to share their and measure their progress, leaders can turn their work-
more unbiased and fair in their assignment of deals after experiences before and after companywide UB training can places into environments where everyone truly feels a sense
they started tracking whom they passed their deals on to. also help leaders understand whether meaningful change ofbelongingandappreciation. ©
Encourage Good Practices and is occurring. One way to do this would be through surveys HBR Reprint R2105H
Continued Learning done just before and few months after the training.
If leaders want their organizations to become more Set a Broader Strafegy for Broader Impact Nudge people to make better decisions. After train­
When organizations make a broad commitment ing, organizations can establish what behavioral scientists FRANCESCAGINO is the Tandon Family Professor of Business
equitable and diverse, they need to help employees imple-
call “nudges”—measures that prompt people to engage in Administration at Harvard Business School and the best-selling
ment the lessons of UB training. Here are some measures to fostering diversity, equity, and inclusion, employ­
author of Rebel Talent. She studies how people can have more
they can encourage people to take. ees’ buy-in increases. In our survey of more than 500 new strategies. For example, before managers write per­
productive, creative, and fulfilling lives. Twitter: @francescagino
Commit to improvement. UB training should offer employees, participants reported taking UB training more formance reviews, they might be reminded to avoid giving KATHERINE COFFMAN is an associate professor of business
leaders and employees alike time to thoughtfully consider seriously when it was accompanied by resounding insti- feedback about employees’ personalities. Recruiters might administration at Harvard Business School. Her research focuses
their motivations for reducing bias. Some may want to tutional support demonstrated through thoughtfulness, be asked to reflect on key job requirements before discussing on how stereotypes affect beliefs and behavior.

Harvard Business Review

122
Harvard Business Review
September-October 2021 September-October 2021 123