Is peer review at Science sexist?

Last week, researchers published a dataset of peer review outcomes for 110,303 scientific papers submitted between 2015 and 2020 to the prestigious journal Science and its open-access cousin Science Advances. The data tracks each manuscript from submission through editorial review, peer review, and eventual publication or rejection.

Getting published in glossy journals like Science can make or break a researcher's career, and most existing data on peer review is from less selective journals. So it's notable that the team released an anonymized version of the dataset for everyone to analyze.

The initial analysis presented with the dataset found some interesting patterns. One revealing result was the role of editors at elite journals. Because Science and Science Advances get so many submissions, both journals evaluate papers in two stages: peer review, in which outside experts review the work, and editorial review, in which editors (and boards of experts that advise editors) decide which papers to send for peer review and which ones to accept based on peer review comments. Only 17% of articles submitted to Science even made it to traditional peer review. But articles that made it to peer review were accepted over one-third of the time.

The team also looked at how topic, prestige, and geography affected acceptance. Papers in Topic 7 (includes viruses, RNA, immune response, and related terms) were accepted 9.6% of the time, but papers in Topic 3 (includes society, politics, gender, and related terms) were accepted just 1.7% of the time. That's a nearly 6-fold difference. Manuscripts with a corresponding author at an institution in the most prestigious quintile had an 11.6% acceptance rate; that rate was just 3.4% for papers in the two least prestigious quintiles. Papers by large teams were accepted at higher rates than those by small teams, and US and Canada-based corresponding authors were more than 3 times as likely to be accepted than those by China-based corresponding authors.

There was also small but statistically significant disadvantage in publication outcomes for studies with a woman as corresponding author. Papers with a man as corresponding author were accepted 6.4% of the time, as opposed to 5.3% for papers by their female colleagues (P < 0.001).

Some researchers on Bluesky saw the paper as proof of misogyny — not complex systemic factors, but plain old sexism — in the peer review system. "The paper confirms the hostility many of us feel is baked into the editorial & peer review community (sitting behind a veil of anonymity) that prefer to see papers led by women thrown out," read one post by a biophysicist. "New scientific paper by male authors laments continued gender-bias in reviewing of scientific papers led by female authors. Appreciate the solidarity, though I wonder if the reviews were 'easier' for it being a male crew," reads another by an anthropologist. Another post by a cognitive scientist called Science sexist.

Is it that simple? Are papers really being accepted or rejected just because they have a woman as corresponding author? If that's the case, it really, really sucks. Not just because it'd be sexist, but because it'd be very hard to do anything about directly. It's hard to change people's biases. But might there be factors like article topic or author location or institutional prestige that contribute to the gender gap — and which might be easier to address?

Since the anonymized data is free to access, I thought it'd be fun to see if I could find out.