Leonhard Foeger / Reuters

Decades' worth of social-psychology research has demonstrated that humans classify each other by race and gender and respond instinctively based on stereotype and social norms. In the workplace, these unconscious biases mean white men are more likely to be hired, promoted, and paid well compared with equally deserving women or racial minorities.

One famous experiment found that resumes that appeared to be from white people received 50 percent more interviews than identically qualified applications from black candidates. This effect shows up even when the decision maker is from the less-powerful group. A more recent study showed that men and women alike were twice as likely to hire a man for a math-intensive job, even when the candidates’ math skills were identical. Awareness of implicit bias has grown thanks to academic initiatives such as Project Implicit and grassroots blogs like the Microaggressions Project and I, Too, Am Harvard, which is now a YouTube meme, as well as media coverage and advocacy by business leaders such as Facebook's Sheryl Sandberg.

A key difference between this sort of unconscious bias and more deliberate discrimination is that those perpetuating it often wish they could stop it, if only they knew how. Existing efforts—such as diversity training programs and exhorting women to "lean in"—may actually backfire; for example, women who negotiate for raises may do themselves more harm than good. “There are all kinds of really well-meaning people in corporate America who are sincerely distressed that they have spent many millions of dollars trying to retain and advance women and minorities and it ain’t working. They are baffled about what to do,” says Joan C. Williams, a law professor at the University of California, Hastings.

Here’s the good news: For those who want to make their workplaces more fair, there are real, concrete steps available to get there. Williams outlines three steps that companies should take: First, conduct internal research to identify areas of possible bias; second, identify key metrics for tracking the results of interventions; and finally, make a change that will curb the effects of these subconscious prejudices on an ongoing basis—Williams calls these interventions "bias interrupters." Once those steps are taken, managers can assess the effect of their interventions, tweak, and repeat as needed.

What is a "bias interrupter"? A bias interrupter can be as simple as rewriting help-wanted advertisements to remove traditionally masculine words or redesigning the process by which people receive promotions, as Google recently did. The company had found that women were being promoted less often than men in part because of a system that required employees to nominate themselves for promotions, something that runs counter to norms of traditional feminine conduct. Men routinely nominated themselves at higher rates than women did.

To interrupt this, the technology firm launched a campaign to change the culture around self-nomination. Google started systematically asking all employees who met the criteria for promotion to nominate themselves and directed managers to follow suit within their workgroups. Meanwhile, female senior leaders spoke at meetings and within the women’s employee resource group about the importance of self-promotion, signaling that this behavior was expected and desirable for women as well as men. As a result, the gender difference on self-nomination dissipated.

“We all have unconscious bias because of our exposure to images in the media and our families of origin, but if we’re aware of it we can work together to combat that,” said Judith Williams, Google’s global diversity and talent programs manager. “It’s so important to understand that the diversity conversation is not that in a few months you’re going to see a huge change. We’ve made progress. We’re working for more. This is a long-term game; we’re in it for the long term.”

One phenomenon holding women back is what Williams terms "office housework," necessary but undervalued tasks, such as administrative chores, scheduling, meeting planning, taking notes, filling coffee cups, mentoring, serving on minor committees, and spearheading low-visibility projects. In a study of gender bias in science and tech, Williams found Latina scientists most likely to report being expected to do this sort of work, which helps an organization but doesn’t propel an individual’s career forward. To interrupt this bias, an employer could survey employees to develop metrics that measure who is doing these tasks, and then either dole out the “housework” randomly or assign it to an administrative employee. It’s important, Williams argues, to evaluate and ratchet up an intervention if the gender and racial inequalities persist.

Another common pattern of bias is known as the “prove-it-again” pattern, in which social psychologists find women and minority employees have to continually prove their skills and value, as compared with white men. This often comes across as women being promoted only based on their results, while men are promoted on their potential. To disrupt this bias, which falls most heavily on black women, organizations could pre-commit to a list of requirements for a promotion, and then require a detailed explanation for any waiving of those conditions, Williams suggests.

These sorts of interventions will be easier to design for certain challenges than for others. For example, when female law-firm partners are bullied out of their share of the revenue for clients they brought to the firm, culture-wide change is needed. And some companies are just not going to be on board with even trying: Research that finds organizations that aver a belief in meritocracy—common in the technology sector—are actually more likely to be discriminatory.

The dwindling numbers of minorities and women as you cast your eye up the ranks of any business or professional field lend credence to the argument that existing diversity efforts just aren’t working. With compelling research that diverse work groups produce better business outcomes, the profit motive should sway corporate America, even if the moral one falls short.

We want to hear what you think about this article. Submit a letter to the editor or write to letters@theatlantic.com.