Effect Size
Magnitude of an effect independent of sample size — complements the p-value, which only shows significance.
Also known as: effect size, Cohen's d, delta
Effect size quantifies how big an observed or expected effect is. It is the practical counterpart to the p-value: a tiny effect can be highly significant with huge n, a large effect can be non-significant with small n. Without effect size the picture stays incomplete.
- Cohen's d (mean comparisons): ; small 0.2 / medium 0.5 / large 0.8
- Pearson r (associations): small 0.1 / medium 0.3 / large 0.5 — see correlation
- η² / partial η² (ANOVA): share of explained variance
- Odds Ratio / Rate Ratio — see OR / RR
See also
Used in
In the Algorithm Lab
Sources
- Cohen, J. — Statistical Power Analysis for the Behavioral Sciences, 2nd Edition, Erlbaum
- Cumming, G. — The New Statistics: Why and How, Psychological Science, 25(1), 2014
- Wasserstein, R. L. & Lazar, N. A. — The ASA Statement on p-Values, 2016