Surprising Data on Hiring Bias
New research finds companies may be committing less racial bias in hiring than before, but some experts are skeptical about the results.
By Jack Robinson
In a famous experiment more than a decade ago, economists found strong evidence of racial bias in companies' responses to fake resumes. New research now suggests the problem may be easing.
Many experts are skeptical, though. They say unconscious bias is as big a problem as ever, hurting workforce diversity. And it's increasingly on the minds of top human resource executives.
"It's a very big issue, even globally," says Margaret Regan, president and CEO of The FutureWork Institute, a Brooklyn, N.Y.-based workforce-consulting firm. In workshops for employers that her company provides around the world, the topic of unconscious bias is "probably our most requested session," she says.
The issue gained attention in 2004 with a study by a pair of economists. Marianne Bertrand at University of Chicago and Sendhil Mullainathan, now at Harvard University, answered more than 1,300 job ads with about 5,000 fake resumes.
While qualifications listed on those resumes were the same, the names were different. Some resumes carried names such as "Brad Kelly," suggesting they were from white applicants. Others -- with names such as "Darnell Jackson" -- signaled the applicants were black.
The results were startling: Employers were 50 percent more likely to call candidates from resumes with white names. Applicants with black-sounding names needed eight years more experience to get the same call-back rate, researchers concluded.
A more recent experiment using similar methods has reached a different conclusion. Researchers at the University of Missouri and elsewhere sent out nearly 9,000 fake resumes but found little evidence of racial bias in company responses.
But is bias really receding? The new experiment was different in several ways, most notably in the names put on fake resumes. To suggest an applicant was black, researchers used two last names that are especially common in African-American families: Jefferson and Washington. But they paired them with first names that are not: Chloe and Ryan.
Cory Koedel, an associate professor of economics and public policy at the University of Missouri, says he and other study authors took this approach to limit the influence of socioeconomic status. Many first names common only among African Americans indicate family income and education as well as race, he and other researchers say.
This change in the experiment introduced "a little bit of ambiguity" into the results, Koedel says. With only last names as an indicator of race, were companies simply missing the signal that would have allowed them to discriminate?
Some experts say that's exactly what happened, and they doubt the study shows hiring bias is receding.
One is John M. Nunley, an associate professor of economics at the University of Wisconsin-La Crosse who has worked in this area. He recognizes the challenge of separating race and socioeconomic status.
"I get that concern," Nunley says. "But I also suspect that [employers] are not picking up race" without first names that distinguish between black and white applicants.
Nunley says his own 2014 study, which otherwise confirmed the original 2004 research, did find the gap between call-back rates for white and black resumes had declined to 14 percent from 50 percent. So he's cautiously hopeful that researchers are seeing changing behaviors by hiring managers.
"It appears that -- maybe -- things have improved," he says.
Nunley notes that his study found bias largely in hiring for customer-service jobs. This, along with other research, suggests some managers may be anticipating what customers will think about a potential employee. This effect was especially pronounced in cities where the population is largely white, he says.
Driven in part by research results such as these, unconscious bias in hiring has become a hot topic in recruitment circles. Many companies have launched training programs to help hiring managers counter unconscious bias.
Among many common recommendations, consultants suggest that recruiters use structured interviews, asking the same questions of each candidate in the same order. This can reduce the influence of extraneous influences, such as a school, mentor or company in common between candidate and interviewer.
Several new ventures also aim to help HR departments by providing new tools to reduce the opportunity for bias in hiring.
Textio, a Seattle-based start-up, for example, offers software for analyzing job ads that -- among other things -- identifies phrases that will send unconscious signals to applicants. That can help hiring managers craft job postings that encourage a diverse applicant pool.
Another new firm aimed at engineers, interviewing.io, lets job hunters interview with recruiters anonymously -- no names or video, just voice. Only if both sides are interested do things move to the next stage.
GapJumpers, also meant for technical workers, takes it a step further by allowing applicants to participate in a "blind audition" in which they can show their skills before revealing name or resume details that might get them screened out.
Addressing unconscious bias at any one point in the application process may not solve the problem, however. Unconscious bias can creep in anywhere, says Joelle Emerson, CEO and founder of the diversity consulting firm Paradigm.
"We've found that bigger disparities often emerge after the resume-review stage of a hiring process, when identity becomes more salient," Emerson says. And it can reflect employer prejudices about not only race, but colleges or previous employers.
"Essentially, bias can come into play throughout the hiring process," she says. "Even if you solve for bias at one stage, you can expect it will still affect the others."
Send questions or comments about this story to firstname.lastname@example.org.