Product & Analytics / Experimentation

Statistical Significance (Experimentation) ~ (p-value; stat-sig; "significant at 95%")

A rule for deciding whether an A/B test result is real signal rather than noise.

The fork, why two teams get different numbers

Three methodologies give different verdicts on the same data.

In the Metric LibraryGet the Metric Library

Reference: Neyman-Pearson / NHST convention; sequential-testing literature (Wald SPRT · alpha-spending / Lan-DeMets); Bayesian A/B-testing convention