On April 12, 2018, a Philadelphia Starbucks manager called police on two African American men for doing nothing (not buying coffee while awaiting a friend). Was this but “an isolated incident”—as 48 percent of White and 10 percent of Black Americans presume? Or did the arrests reflect “a broader pattern” of bias? Starbucks’ CEO later apologized for the incident, said the manager was no longer with the company, and announced that on May 29, the company would close 8000 domestic stores to enable employee racial bias training.
The Starbucks fiasco drew national attention to implicit bias. It also illustrates what we social psychologists agree on, what we disagree about, and what can be usefully done:
Agreement: Bias exists. When one study sent out 5000 resumes in response to job ads, applicants with names such as Lakisha and Jamal received one callback for every 15 sent, while names such as Emily and Greg received one callback for every 10 sent. Similar racial disparities have been found in Airbnb inquiry responses, Uber and Lyft pickup requests, and descriptions of driver treatment during police traffic stops.
Agreement: Unconscious (implicit) biases underlie many racial disparities. Such biases are modestly revealed by the famed Implicit Association Test (IAT). Likely the Starbucks manager never consciously thought “those two men are Black rather than White, so I’ll call the police.”
Disagreement: How effective is the IAT at predicting everyday behavior? Its creators remind us it enables study of a real phenomenon, but was never intended to assess and compare individuals and predict their discrete behaviors.
Disagreement: How effective is implicit bias training? Skeptics argue that “blame and shame” diversity training can backfire, triggering anger and resistance. Or it may seem to normalize bias (“Everybody is biased”). Or it may lead to a temporary improvement in questionnaire responses, but without any lasting benefits. Even Anthony Greenwald and Brian Nosek, social psychologists and two of the IAT co-creators, echo some of these concerns. Greenwald notes that “implicit bias training . . . has not been shown to be effective, and it can even be counterproductive.” And Nosek warns that “diversity trainings are filled with good intentions and poor evidence.”
Greenwald and Nosek doubt the likely effectiveness of Starbucks’ planned training day. Nosek believes the company would be better advised to pilot and assess their intervention in a few stores and then scale it up.
But some research offers more hopeful results. As part of their research on automatic prejudice, Patricia Devine and her colleagues trained willing volunteers to replace biased with unbiased knee-jerk responses. Throughout the two-year study follow-up period, participants in their experimental intervention condition displayed reduced implicit prejudice. Another team of 24 researchers held a “research contest” that compared 17 interventions for reducing unintended prejudice among more than 17,000 individuals. Eight of the interventions proved effective. Some gave people experiences with vivid, positive examples of Black people who countered stereotypes.
Recently, Nosek and Devine have collaborated with Patrick Forscher and others on a meta-analysis (statistical summary) of 494 efforts to change implicit bias. Their conclusion meets in the middle: “Implicit bias can be changed, but the effects are often weak” and may not carry over to behavior.
So, what should we do? And what can we—and Starbucks and other well-intentioned organizations—do to counteract implicit bias?
First, let’s not despair. Reacting with knee-jerk presumptions or feelings is not unusual—it’s what we do with that awareness that matters. Do we let those feelings hijack our behavior? Or do we learn to monitor and correct our behavior in future situations? Neuroscience evidence shows that, for people who intend no prejudice, the brain can inhibit a momentary flash of unconscious bias in but half a second.
Second, we can aim toward an all-inclusive multiculturalism. As race-expert Charles Green explains, “ It’s about addressing unequal power distribution and creating opportunity for all. It is structural, not personal.”
Third, we can articulate clear policies—behavior norms—for how people (all people) should be treated in specific situations. Organizations can train employees to enact expected behaviors in various scenarios—dealing with customers in a coffee shop, with drivers at a traffic stop, with reservation inquiries at a rental unit. Happily, as people act without discrimination they come to think with less prejudice. Attitudes follow behavior.