When you're told the AI gender gap is about confidence
The Move
Cite the Bayesian meta-analysis that kills the confidence-gap framing, then pivot to the real driver: information and anchoring.
The Script
"I hear you, and I used to think that too. But Bandiera et al. ran a Bayesian meta-analysis of the overconfidence experiments and could not reject the hypothesis that the gender difference is zero. The real gap is information, not confidence. Roussille tracked 110,000 candidates and found that when you supply the anchor — the market number — the ask gap, bid gap, and offer gap all go to zero."
Your AI Prompt
Someone just told me the AI gender gap (or pay gap, or leadership gap) is about women's confidence. Draft a 3-sentence response that: (1) acknowledges the intuition without being dismissive, (2) cites the Bandiera et al. 2022 meta-analysis finding, (3) pivots to the information/anchoring mechanism from Roussille 2024. Tone: collegial and precise, not lecturing.
Why This Works
Bandiera, Parekh, Petrongolo & Rao (Economica 2022) conducted a Bayesian meta-analysis of 90 paired observations from 38 papers on overconfidence and could not reject the hypothesis that the gender difference is zero. Roussille (QJE 2024) tracked more than 110,000 candidates: when a platform prefilled expected salary with the market median, all gaps went to zero. Mazei et al. (Psychological Bulletin 2015) found that under low-ambiguity conditions, women's negotiation outcomes reverse in their favor. The intervention isn't encouragement — it's information.
Research Behind This Move
Men Are from Mars, and Women Too: A Bayesian Meta-Analysis of Overconfidence Experiments
Bandiera, Parekh, Petrongolo & Rao · Economica 89(S1) · 2022
The Role of the Ask Gap in Gender Pay Inequality
Roussille · Quarterly Journal of Economics 139(3), 1557–1610 · 2024
A Meta-Analysis on Gender Differences in Negotiation Outcomes and Their Moderators
Mazei, Huffmeier, Freund, Stuhlmacher, Bilke & Hertel · Psychological Bulletin 141(1), 85–104 · 2015