We've all heard about the study suggesting that job applicants with black-sounding names should go to the back of the bus. The fact that white-sounding candidates named Emily Walsh were 50 percent more likely to get a callback than those named Lakisha Washington has become an oft-cited stat — and a political talking point
But a major new study from the University of Missouri casts doubt on those conclusions. Its markedly different findings suggest either that the oft-cited 2004 study may be due for an update — or that the researchers behind it perhaps inadvertently measured more than just race.
In fact, the new study shows that the response rates for male vs. female applicants, as well as black vs. Hispanic vs. white applicants, are statistically indistinguishable, says co-author Cory Koedel, an associate professor of economics and public policy.
Koedel says he believes it is the first major study to revisit the question raised in the landmark 2004 research. And if anything, a comparison shows, it's even more comprehensive.
The 2004 study that found such a marked racial divide, by Marianne Bertrand and Sendhil Mullainathan
, sent 5,000 fake resumes in response to help-wanted ads in newspapers in Chicago and Boston. The authors of the new study, Koedel and Rajeev Darolia, both professors at Mizzou's Harry S Truman School of Public Affairs, sent in 9,000, in seven different cities.
Koedel and Darolia surveyed how those fictitious applicants were treated in six broad occupational categories, including customer service, medical billing and information technology. The applications were randomly assigned names that are highly correlated with gender and race/ethnicity based on U.S. Census records — Washington and Jefferson for blacks; Hernandez and Garcia for Latinos; Anderson and Thompson for whites.
Across the board, 11.4 percent of resumes got a response from employers, the study found. And while white applicants — and white women in particular — got the most responses, those results were not statistically distinguishable.
That is also true when the professors tracked explicit requests for an interview by the employer, according to their report — a more clear sign of interest in the candidate. There, too, the responses were similar, Koedel says.
So what's changed since the 2004 study?
For one thing, the 2004 researchers, Bertrand and Mullainathan, gave their fake applicants distinctly African-American-sounding first names — Ebony, Jamal and Aisha. The more recent Mizzou study did not. The reason: "researchers have indicated concern that these names could be interpreted by employers as being associated with relatively low socioeconomic status," Koedel and Darolia write. They wanted to remove that as a variable. (The 2004 study, unlike the Mizzou one, didn't include Latino surnames at all.)
But while the Mizzou professors removed that variable, they acknowledge, they may have created another one. Census data shows that 90 percent of individuals named Washington are black, and 75 percent of people named Jefferson are as well, but is that widely known? By not using ethnic-sounding first names, Koedel and Darolia write, "a tradeoff is that the surnames in our experiment may not be as strong indicators of race as the distinctly African-American sounding names in" the previous study. Still, they note, the error rate would have to be nearly 60 percent to get similar results as in the 2004 study if confusion were entirely to blame.
There is another factor at play, however, and that's the passage of time. Have hiring managers in the U.S. come a long way in the last twelve years?
"I think that's absolutely a possibility," Koedel says. "Our study can't say why this happened, but that is one possibility."
After all, back in 2004, job applicants were still answering help-wanted ads they saw in the newspaper. Who's to say that the twelve intervening years haven't made them more receptive to candidates with a different racial or ethnic background?
It's a hopeful thought, and Koedel, for one, sees plenty of possibilities for future research to explore.
"I think in terms of this line of inquiry, we're at the end of the road for us right now," he says. "But there are lots of smart people who can pick this up and ask further questions in interesting ways."
We welcome tips and feedback. Email the author at firstname.lastname@example.org