Scott Alexander critiques a study claiming gender bias in GitHub, pointing out methodological flaws and media misrepresentation of its non-peer-reviewed findings.
Longer summary
Scott Alexander critiques a study about gender bias in GitHub pull request acceptance rates. He points out several issues with the study's methodology and interpretation, including the lack of peer review, ambiguous statistical significance, and potential confounding factors. He also criticizes media outlets for misrepresenting the study's findings, exaggerating its conclusions, and failing to mention its non-peer-reviewed status. Scott emphasizes that the study actually shows women's pull requests are accepted more often overall, and that the observed bias against women in one subgroup is small and possibly not statistically significant. He expresses concern about how such studies are used to promote a narrative of widespread sexism in tech.
Shorter summary