Effect Sizes Over Significance: Reporting That Travels

Abstract

A p-value below a threshold says little about magnitude and nothing about practical relevance. We make the case for effect sizes with intervals as the default reporting unit, and work through examples where the two framings lead to opposite conclusions.