Publication history explains roughly a quarter of the gap in success rates between black and white researchers who apply for the U.S. National Institutes of Health’s (NIH’s) bread-and-butter R01 research grant, according to a study published today in PLOS ONE. The finding sheds light on a tangible reason behind a known funding gap, but it doesn’t eliminate the possibility of bias in the NIH review process.
In a 2011 Science paper, a team of researchers led by Donna Ginther, a professor of economics at the University of Kansas in Lawrence, reported that black applicants’ chances of winning an R01 grant were 13 percentage points lower than those of white applicants. After accounting for confounding factors such as the applicants’ research institution and prior NIH funding, the gap narrowed to 10 percentage points. The new study reports that the funding gap between black and white researchers narrows further, to 7 percentage points, when the team accounts for additional information about publication history contained in applicants’ biographical sketches—brief CVs frequently referred to as biosketches that are submitted as part of NIH applications.
“The first paper from this team set in motion [a] reverberation that went around the country,” says Stephen Thomas, the director of the Maryland Center for Health Equity at the University of Maryland in College Park, who works to help increase diversity in the biomedical workforce and wasn’t involved in either study. The new publication is “a very important paper” because it uses a more sophisticated analysis to identify explanations for the funding gap, but it “has not removed the prospect of actual bias on the part of reviewers,” he says.
In the PLOS ONE paper, Ginther’s team extracted new information about applicants’ publication records by looking at their biosketches. The 2011 paper, which examined 83,188 R01 grant applications submitted to NIH between 1999 and 2006, also looked at the applicants’ publication history. But the data were gleaned from publicly available databases and the name-matching algorithm that the team used to link applications with publications sometimes yielded errors, Ginther says.
Scouring the biosketches for the new information was more labor-intensive, so the researchers focused their latest analysis on a subset of the applications analyzed in the 2011 study: 598 applications from black scientists and 599 applications from white scientists that were submitted to NIH between 2002 and 2006. (The researchers also looked at applications from Asian and Hispanic scientists, but the findings were comparable to their 2011 study, which reported that funding gaps for those groups could be entirely explained by confounding factors.)
In the new sample group, grants were awarded to 27% of white applicants and 14% of black applicants. That 13 percentage point difference mirrors what was found in the 2011 study. After accounting for the variables that were found to be important in the 2011 study and the publications listed on biosketches—evaluated by looking at journal impact factors, author placement, and citation rates—the black-white funding gap dropped by 6 percentage points, double the drop that was documented in the first study. The team also looked at training history—traineeships, fellowships, postdoc institutions—but that biosketch information did not appreciably reduce the gap. “It’s not a slam dunk,” Ginther says. “We’d like to be able to explain the entire gap. … And we don’t fully explain it, but we’re closing in on an explanation.”
The biosketches cited an average of 23 publications—more than would be the case on current applications because NIH changed biosketch guidelines in 2014. Ginther’s team found that the papers that black applicants listed in their biosketches tended to have lower journal impact factors than those of white applicants. On average, black applicants also reported fewer publications and their papers were cited less often—differences that were more apparent for papers that the applicants published when they were principal investigators, as compared with when they were Ph.D. students or postdocs. “As careers progress, the gap widens,” Ginther says.
The new study shines a brighter light on the importance of helping junior researchers develop solid publication records—although there are likely various complex reasons underlying the disparities, such as less access to resources, networks, and support. Moreover, Thomas says, “many racial and ethnic minorities are interested in conducting research that really addresses the needs of the populations they’re concerned with.” Those types of studies may be undervalued by high impact journals, which tend to focus more on basic research, he says. Black researchers may also spend more time doing service work, which could affect their publication productivity. “We call that the black tax,” Thomas says.
In the new analyses, roughly half of the black-white funding gap isn’t explained by any variable in the dataset, which means that the results don’t rule out the possibility that implicit bias creeps into NIH’s peer review process, Ginther says. “If I was reading this when I was going through the pipeline, I’d be very discouraged,” says Esteban Burchard, a professor of medicine and biopharmaceutical sciences at the University of California, San Francisco, who has written about the need for a diverse biomedical workforce.
Burchard also points out that the data are old. “I would love to see a repeat of this with current data,” he says. “Basing [analyses] on data from 2003, 2006—that’s a long time ago.”
NIH hasn’t repeated Ginther’s analysis in its exact form, but the agency has explored data from the past 3 years. “We’re seeing higher application rates [from black researchers] and a slight narrowing of the gap,” says Hannah Valantine, the chief officer for scientific workforce diversity at NIH. The gap still exists, “but it’s less than it certainly was in the Ginther era.”
Valantine notes that she and her NIH colleagues are “busy actively trying to address this issue.” They’re currently in the middle of a study in which they are re-reviewing R01 applications submitted by black and white researchers—but now stripping them of any identifying information. Valantine anticipates releasing those results in late 2019. “Hopefully that will enlighten us as to whether or not there is some degree of bias,” she says.