The study finds that candidates whose research closely matches that of evaluation committee members are significantly more likely to be selected, regardless of publication quality. Using AI techniques on over 40,000 abstracts, the authors identify a “self-similarity bias” that rewards research aligned with evaluators’ own scientific interests. Because committees are still predominantly male, men are more likely than women to share strong research affinities with at least one committee member. This dynamic disadvantages women and ultimately limits diversity and innovation within the discipline.

Read the full article