If you’re a big, powerful, name-brand company in 2017, the public is going to judge you based on both the diversity of your workforce and how welcoming a place your company is for members of minority groups. This is a good thing — it’s only recently that society has finally gotten a bit more serious about diversity and inclusion, and many employees suffer when the companies they work at make little or no effort to be diverse and inclusive.
The problem is that this pressure leads some companies to take big, flashy, headline-grabbing diversity moves that don’t actually accomplish anything. For a really good example, check out Beth Winegarner’s interesting, well-reported story on Google’s recent diversity travails, in Fast Company. The short version is that “[a]fter spending two years and $265 million on the effort, Google’s employee population was only 2% black in 2016, the same percentage as it was in 2014.” Google simply hasn’t found any approach to increasing diversity that has worked, especially when it comes to hiring more African-American employees.
The whole story is worth reading, but three key lessons jump out not only for tech companies, but for any companies trying to improve on this front:
1. Stop falling for the overblown claims of implicit-bias research. Implicit bias, or deep-seated biases that affect our behavior beyond our conscious awareness, is real and is almost certainly one factor (among many) when it comes to the lack of diversity in tech and other fields. A black person applying to be an engineer at Google likely faces obstacles a white or an Asian applicant doesn’t, in part because they might not fit certain deep-seated ideas of what an engineer is “supposed” to look like. Studies dealing with résumés and housing have shown over and over again that people of color face disadvantages in hiring and renting simply by dint of their race — people who are black, or who have names that “sound” black on a résumé, get interviewed and offered apartments less frequently than white people, even when all the characteristics that should matter are held constant. Given that fewer and fewer Americans would explicitly state an unwillingness to interview or rent to black people, implicit bias must be one of the mechanisms behind these unfortunate disparities.
So implicit bias does matter. But the question of whether it matters is different from the question of whether researchers have found a way to accurately measure it, and to reduce it in a way that will lead to less biased behavior. They haven’t, and they haven’t. As I wrote in a long recent article, the main tool used to measure implicit bias at the individual level, the implicit association test, is flawed almost to the point of uselessness (at least when it comes to IATs dealing with race and ethnicity), and, according to the best meta-analysis we have, there’s basically no evidence interventions geared at reducing implicit bias (as measured by the IAT or other tools) have a meaningful effect on outcomes.
All of which is to say, it’s unfortunate that the IAT has been super hot for a while — it’s seen as the way for a company to show it’s serious about diversity. And so it has been for Google. Winegarner writes that Erica Joy Baker, an engineer who left Google in 2015 “largely because she was fed up with the company’s failure to address a culture of casual racism,” “says it’s not clear to her whether Google’s newer initiatives are rooted in a real desire to change, or are a part of a public-relations effort, which she said was evident in 2014, when the company released its unconscious-bias training information through re:Work.”
If you follow the link, it takes you to a website where Google brags that “To date, more than 30,000 Googlers (over half of the company) have participated in the 60-90 minute [implicit bias] workshop, making it the largest voluntary learning program at Google.” Sure enough, the workshop in question, which you can watch a video of here, leans heavily on the IAT. That is a lot of money and time spent on a test that hasn’t been credibly linked to biased behavior in the first place.
2. Understand the broader limitations of the research on diversity trainings. Because companies want to avoid lawsuits, have a genuine desire to foster more welcoming environments to employees who are members of minority groups, or both, there is a large and growing cottage industry of diversity-training programs. Unfortunately, the best evidence we have suggests that these programs don’t work. For example, writing in Contexts, the American Sociological Association’s journal, in 2007, the researchers Frank Dobbin, Alexandra Kalev, and Erin Kelly reported the results of their examination of the diversity practices of 829 companies. “Three of the four most popular programs — diversity training, evaluations, and network programs — have no positive effects in the average workplace,” they explained. “The two least popular initiatives, mentoring and diversity managers, were among the most effective.”
Even a decade later there’s still simply a dearth of solid research on these programs — definitely not enough for companies to adopt the biggest, broadest, most sexy-looking diversity programs (unless their goal is simply to avoid lawsuits rather than improve their workplace). In an email, Betsy Levy Paluck, a social psychologist at Princeton University who studies intergroup relations, told me, “I’m on record saying that we haven’t studied diversity programs well enough, period.”
3. If you have resources, maybe focus more of them on “upstream” diversity spending. None of the above is a good reason for companies to give up on diversity initiatives, of course. For one thing, that just isn’t an option in a country as diverse and socially aware as ours; and for another, as the Contexts article notes, there is evidence some programs, such as mentorship ones, can positively affect the workplace.
That said, if a company has a lot of money to throw around, as Google does, and is concerned with actually changing the world, it might make the most sense to attack some of the problems that contribute to nondiverse workplaces at or near the root. Google does some of this. For example, Winegarner writes that the search giant is “a hefty supporter of Black Girls Code, an organization that exposes girls of color to computer science and technology, and last year it set up shop in Oakland’s predominantly black Fruitvale neighborhood in an effort to provide technology training to locals. A second site is slated to open in Harlem, New York, this year.”
These programs need to be judged like any other, of course, but many of them undeniably provide underprivileged young people with access to resources they wouldn’t have otherwise. While they are a drop in the bucket compared to the country’s huge disparities in education and access to STEM-training resources, they do make a difference — they likely produce, in the long run, more Google-hireable people people from underrepresented groups. This is a good thing, whether or not it leads immediately to tangible results for Google or other philanthropically minded corporations.