A/B Test Autopsy·interactive demo
Pathology · 02

Minimum detectable effect

Required sample size scales with the inverse square of the relative effect. Halving the MDE quadruples the N. Under-powered tests do not decide. Running them longer rarely closes the gap.