A recent study found that women early-career scientists had higher rejection rates than their male counterparts when applying for Discovery Grants from the Natural Sciences and Engineering Research Council of Canada. The study also found that, among successful applicants, women obtained slightly less funding than men.
The study’s authors hypothesized that these discrepancies in funding outcomes might be due to differences in the language used by men and women in their grant summaries, but no noticeable differences were detected. The study, by University of British Columbia doctoral student Mackenzie Urquhart-Cronish and UBC zoology professor Sarah Otto, was published in September in the open-access journal FACETS.
The two authors analyzed the text of research summaries from nearly 2,000 Discovery Grant applications from 2016 to see whether they could detect gendered language via indicators like analytic thinking and emotional tone. “There is some evidence that this occurs in general language use, although not necessarily in scientific discourse,” said Ms. Urquhart-Cronish. Gendered language differences might trigger implicit bias reactions, added Dr. Otto, “or alter the outcomes of grant applications in a way that we were unaware,” explaining that once identified, possible sources of bias can be considered by writers and reviewers.
The researchers also obtained data for the number of applications applied for versus those funded between 2012 and 2018, and the funding level granted. If not indicated, the presumed gender was assigned using a name association algorithm, with researchers acknowledging possible error in mismatching each researcher’s self-identified gender, including nonbinary genders. Their analysis of funding decision and amount also examined career stage and scientific discipline.
Granting success rates for 2016 applicants self-identifying their gender were 67.5 percent for men and 64.6 percent for women. The proportion of female awardees varied by discipline, ranging from roughly 10 percent in the physical sciences, engineering and mathematics to about 30 percent in life and Earth sciences. Average grant size was $33,155 per year, with men at a similar career stage and discipline receiving $1,756 more.
Career stage, inferred from available data, was the most important predictor of award value. Rejection rates were higher for women (40 percent) than men (33 percent) in the early career phase. “Early-career gender disparity is a potentially important obstacle impeding the research progress of female scientists and should be investigated,” the study authors write.
Rebecca Goldin, professor of mathematical sciences at George Mason University and director of STATS, a collaboration between the American Statistical Association and Sense About Science USA, says, “It’s interesting work because they documented differences.” However, she cautions, “Even if the evaluators know the gender, that does not, in and of itself, mean that there is bias in the evaluation of proposals,” noting that the study did not quantify the quality of the applications themselves.
Dr. Goldin also cautions that their language analysis may be of “limited value” because the scientific summaries – the publicly accessible part of the grant application – may be stylistically different than the body of the application, a consideration echoed by Lesley Shannon, a professor of engineering science at Simon Fraser University who holds the NSERC Chair for Women in Science and Engineering for BC and Yukon. “What would be good is if they could actually do this [analysis] on the proposals themselves,” she says. Drs. Goldin and Shannon were not involved in the study.
The authors conclude that the root causes of the gender differences detected remain unknown. Nevertheless, they say the findings underline a need to “reduce obstacles facing the progress of underrepresented groups as they proceed through tenure-track career stages.”
In a prepared statement, NSERC senior communications advisor Martin Leroux says the agency “recognizes the challenges faced by women in STEM, particularly early career researchers.” In recent years, NSERC has introduced bias mitigation measures, including the collection of more accurate self-identification data from applicants starting in 2018. “The granting agencies are working together to address broader cultural and systemic issues in academia and create an inclusive research enterprise that goes beyond gender equity to recognize diversity in the broadest sense,” says Mr. Leroux.
Ms. Urquhart-Cronish and Dr. Otto say they shared their results with NSERC and plan to meet with the agency to review their findings and discuss opportunities for a more comprehensive study.