Definition
Plain language
A statistical safeguard that raises the bar for calling a result real when you've run many comparisons, so you don't get fooled by chance.
As stated in the literature
A multiple-comparisons adjustment that divides the significance threshold by the number of tests performed, controlling the family-wise error rate at the cost of statistical power.
Why it matters: It matters because without guarding against many comparisons, pure chance will eventually hand you a false 'discovery' that wastes effort to chase.
For example, if you test twenty different diets for an effect and want to avoid a fluke, the Bonferroni correction makes you treat only much stronger results as real.
Heard on the show
“… After Bonferroni correction across sixty-two baseline comparisons — which is a fancy way of saying they were appropriately …”Episode 073 — When Three LLMs Talk to Each Other, Their Ideas Quietly Stop Moving