When someone quotes a stat you know is wrong
The Move
Know the three claims most likely to surface in your meeting and the one-sentence correction for each. The person who debunks on purpose is impossible to ambush.
The Script
"That's actually a great example of why I check primary sources. The 'women apply only at 100% qualifications' stat traces to an unfootnoted line in a 2008 McKinsey article — there was never a study. The BIT ran a real experiment with 10,000+ job seekers and found the gap is 56% vs 52%. Small, real, and fixable."
Your AI Prompt
I heard someone cite '[paste the claim]' in a professional context. Trace this claim to its primary source. Report: (1) who published it, (2) sample size, (3) whether it is peer-reviewed, (4) whether it has replicated. If the claim is debunked or exaggerated, give me the one-sentence correction and the correct citation. Keep it factual, not snarky.
Why This Works
The SCORE project (Tyner et al., Nature 2026) tested nearly 4,000 social science papers: only 55% of positive claims replicated, and those that did were about 40% the size originally published. The 'women apply at 100%' stat traces to a 2008 McKinsey article with no underlying study — the BIT tested it for real and found 56% vs 52%. Kennedy, Kray & Lee (2024) showed 'women don't ask' is superseded: 54% of women vs 44% of men negotiated. Knowing these corrections is a strategic advantage.
Research Behind This Move
Investigating the Replicability of the Social and Behavioural Sciences
Tyner et al. (SCORE project, 292 authors) · Nature 652(8108) · April 2026
Now, Women Do Ask: A Call to Update Beliefs about the Gender Pay Gap
Kray, Kennedy & Lee · Academy of Management Discoveries 10(1), 11–37 · 2024
Men Are from Mars, and Women Too: A Bayesian Meta-Analysis of Overconfidence Experiments
Bandiera, Parekh, Petrongolo & Rao · Economica 89(S1) · 2022