Source-linked AI summary
Gender differences in grant peer review: A meta-analysis
Lutz Bornmann, Ruediger Mutz, Hans-Dieter Daniel
TL;DR
The paper addresses conflicting conclusions about gender bias in grant peer review by synthesizing evidence from 21 studies. Its meta-analysis finds that men have statistically significantly greater odds of approval than women by about 7%, while the causes of this discrepancy remain unknown.
Problem
Narrative reviews had concluded that evidence for gender bias in grant awards was negligible, leaving contrary evidence from grant peer review to be assessed quantitatively.
Method
The paper conducts a meta-analysis of studies of grant peer review procedures and models gender effects across those procedures.
Results
About 7% greater odds of approval were estimated for men than women applying for grants.
Takeaways & Limitations
The findings support a robust average gender difference in grant peer review and motivate measures such as masking applicants’ gender to rule out bias.
Takeaways & Limitations
The cause of the discrepancy is unknown, and incorporating procedure characteristics into the meta-analysis produced models that did not converge.
Abstract
from arXiv · showhide
Narrative reviews of peer review research have concluded that there is negligible evidence of gender bias in the awarding of grants based on peer review. Here, we report the findings of a meta-analysis of 21 studies providing, to the contrary, evidence of robust gender differences in grant award procedures. Even though the estimates of the gender effect vary substantially from study to study, the model estimation shows that all in all, among grant applicants men have statistically significant greater odds of receiving grants than women by about 7%.
1.07. In other words, in grant peer review procedures, men have on average
This meta-analysis finds robust overall gender differences in grant peer review: men have about 7% greater odds of approval than women, although effects vary substantially across procedures.
- The study used Empirical Bayes estimates as effect sizes and 95% credible intervals to represent uncertainty across procedures.Negative mean log-odds ratios indicate preference for men, whereas positive values indicate preference for women.
- The estimated procedure-level effects ranged from 22.1% in favor of men to 22.9% in favor of women, although few procedures favored women.These values correspond to the reported log-odds extremes of -0.26 and 0.20.
- The estimated gender effects varied widely across individual peer review procedures, with statistically significant variability between studies.Individual mean log-odds ratios ranged from -0.26 to 0.20.
- Under an illustrative allocation of 100000 applications with equal numbers of men and women, the estimates imply 26000 approvals for men and 24000 for women.The assumed overall approval rate is 50%, producing a difference of 2000 approvals associated with applicant gender in this example.
- The cause of the observed discrepancy remains unknown, and adding procedure characteristics to the model produced nonconvergent or overly complex estimations.The authors discuss aggregation across fields and possible gender, institutional, or application-pool explanations but do not establish a cause.
- The authors propose masking applicants’ gender as one way to rule out intentional or unintentional gender bias, while noting that masking may not suit every submission type.Grant decisions also assess applicants’ track records, which complicates gender masking.