Joan C. Williams
1,181,611 views • 14:43

In 2018, two Black men went to a Starbucks to wait for a business associate. But when they asked to use the bathroom, the manager ordered them to leave. They refused. He called the police, and the video went viral. Amidst an avalanche of bad publicity, Starbucks closed all stores across the country for four hours of diversity training. And so, baristas were handed workbooks with prompts like, "What makes me me and you you?" and, "Understanding our bias: from color-blind to color brave." This made newspapers across the country, and arguably, that was the goal. "Look, everyone! We're solving our diversity problem!" The assumption, though, was that you could address structural racism with an earnest conversation about our feelings. My take: give me a break.

To address structural racism, you need to change structures. So in the aftermath of George Floyd's death, my sense is that many companies are feeling pressure to actually deliver on their diversity goals, but they haven't a clue what to do. And that's because we spent probably close to a billion dollars on diversity. But the basic tools of the diversity industrial complex, they just don't work. A one-shot bias training — it doesn't work for a really simple reason: doing anything once won't change a company's culture. And the other basic tools — things like an employee resource group or a women's initiative — they're fine, if the problem is with the women and the people of color. But it's not. If a company faces challenges surrounding diversity, typically, it's because subtle and not-so-subtle forms of bias are constantly being transmitted through their basic business systems — through hiring, through performance evaluations, through access to opportunities. So we need to stop trying to fix the women and the people of color. We need to fix the business systems.

And if you think about it, this makes sense, because if a company was facing challenges with sales, it wouldn't respond by holding a series of sincere conversations about how much we all value sales and put on programming for "National Celebrate Sales Month" and expect sales to improve. But that's a lot of what we're doing in the diversity context. If we really want to tackle diversity effectively, we need to use the same tools businesses use to tackle any business problem — evidence and metrics. And, you know, I suspect this will come as a relief to a lot of CEOs who feel far more comfortable using those tools than they do with trying to lead a deep conversation about the inner workings of social inequality.

The first step is for us to understand what bias looks like on the ground. And I and my team at WorkLife Law, we have been studying how bias plays out in everyday workplace interactions for well over a decade. And what we find is that the same patterns of bias, the same five patterns, they emerge over and over again. So here's what the evidence looks like.

The first pattern we call "prove it again." Some groups have to prove themselves more than others. This is triggered by lots of different things. It's triggered by race and gender, age, disability, LGBTQ status, even social class. So one study, for example, looked at callbacks offered to white men with identical qualifications but different hobbies. One résumé listed things like sailing and polo, and the other résumé listed things like counseling first-generation college students and country music. And, if you can believe it, Mr. Polo — he got 12 times the number of callbacks as Mr. Country Music. Too often when we talk about privilege, we forget about class.

The second pattern is called "the tightrope," and it reflects the fact that a certain in-group of white men just need to be authoritative and ambitious in order to succeed. But women walk a tightrope, where they may be seen as abrasive if they're authoritative but unqualified if they're not. And people of color who behave assertively often are written off as angry if they're Black, even hotheaded if they're Latinx and sometimes as untrustworthy if they're Asian American.

The next pattern we call the "tug-of-war," and it reflects the fact that sometimes bias against a group fuels conflict within the group. So, for example, if there's room for only one woman or person of color, it's entirely predictable: women are going to be supercompetitive with other women, and people of color, competitive with other people of color.

The fourth pattern of bias is actually the strongest form of gender bias, called "the maternal wall." And it reflects assumptions that mothers aren't committed, probably shouldn't be and aren't competent — think "pregnancy brain." So mothers often find they have to prove themselves yet again when they return from maternity leave. And if they do, they may be seen as bad mothers and so as bad people and disliked.

The final pattern consists of racial stereotypes. So, Asian Americans again and again report that they're seen as a great match for technical skills, but lacking in leadership potential. And our studies show that Black professionals, again and again, report really high levels of isolation and often startling forms of disrespect. And an Asian American professional may be seen as too emotional in a discussion where, you know what, a white man behaving exactly the same way would be seen as having a career-enhancing passion for the business.

And so what we find is that white women report four patterns of bias. Men of color also report four. Women of color report all five in very substantial proportions. And among women of color, Black women report the most bias as a group. But the bottom line, really, is that the experience of white men as a group differs from that of every other group. If a white man is a first-generation professional or LGBTQ, he may encounter bias. But but most aren't.

These biases can have really serious negative effects. There's a ton of research. But here's a story that really says it all. We were working with one company, and we spoke to a woman engineer who had found a mistake in one of the calculations of a male colleague, and she pointed it out. When she pointed it out, she was violating an unwritten rule. The good woman is seen as modest, self-effacing and nice, not a mission-driven expert. That's why male experts in meetings exert more influence. But you know what? Female experts, they actually exert less influence than female nonexperts do. And so when this engineer pointed out the mistake in calculation, she told us, the response of her department was so massively negative that, she said, "Now I'm just smiling a lot and bringing in cupcakes." This company, by allowing gender bias to go unchecked, was literally jeopardizing their mission.

So what's the solution? The solution is to use bias interrupters, new tools my team has developed that are evidence-based and metrics-driven. And I've just told you about a lot of the evidence. Metrics are also superimportant because they help you pinpoint where things are going wrong. So if a company has challenges with hiring, they should be keeping track of who is in the original pool of candidates and who survives résumé review and who gets called to interview and who survives the interview. And the reason that's important is because the fix, if you have a nondiverse original pool, is totally different than the fix if no woman ever survives the interview because every woman is either too witchy or too meek.

Metrics are also superimportant for another reason: to establish baselines and measure progress. If you use evidence and metrics, what we have found is that small tweaks can have really big effects. So we've worked with one company, for example, who asked us to look at their performance evaluations. And when we did, we found that only 9.5 percent of the people of color had leadership mentioned in their performance evaluations. That was 70 points lower than white women. And that was superimportant because, as you can imagine, mentions of leadership predicted advancement. And so we worked with them to do two simple things. First, we redesigned the performance evaluations form. And second, we help them develop a simple one-hour workshop that, among other things, projected actual comments from the prior year's performance evaluations, and asked people a simple question: Which of the five patterns of bias does this represent, or is it no bias? Just doing that, we found in year two, 100 percent of the people of color had leadership mentioned in their performance evaluations.

At this company, white women, they had a different problem. Almost 20 percent had comments in their performance evaluations that they didn't really want to make partner — this was a partnership. And we suspected the women hadn't actually said that. It was just assumptions. And so in that one-hour workshop, we told people, "Hey, don't say this unless you've actually had a conversation, and someone has told you they don't want to make partner." In year two, only one woman got that comment — one woman in the entire company.

And so what we find is that we have helped over 100 companies actually make progress towards their diversity goals. And there's growing evidence that these bias interrupters work. And the best thing about them is that they help every single group. So in this company I've been talking about, in year two, people of color got wildly more constructive feedback — it was like a 30-percent jump. But white women, they got more constructive feedback, too, and so did white men. If you design your systems based on evidence, it's going to help every single group.

So the bottom line, if you think about it, your systems and your culture, they reflect the people you've already hired. So if you want to replicate that workforce into the future, definitely keep on doing exactly what you're doing. But if you don't, if you actually want to make progress on diversity, equity and inclusion — what we call DEI — my message to CEOs is reassuring: you already know what to do. Use standard business tools, start from the evidence, gather metrics to establish baselines and measure progress and keep at it until you achieve your goals. That's the new DEI playbook. And it works.

Thank you.