Statistics Glossary → Effect size
Effect sizeconcept
A measure of how large a difference or relationship is, independent of sample size.
A p value only says whether an effect can be explained by chance; with very large samples even trivially small differences become significant. An effect size measures how big the difference or relationship actually is, and it makes comparisons across studies, meta-analysis and power analysis possible.
The main families are standardized mean differences (Cohen's d, Hedges' g, Glass's Δ), proportions of explained variance (η², ω², ε², R²), association measures (r, φ, Cramér's V) and ratio measures (odds ratio, risk ratio). Cohen's 'small, medium, large' benchmarks (e.g. 0.2 / 0.5 / 0.8 for d) are only a rough guide; whether an effect matters depends on the field. Effect sizes should be reported with their confidence intervals.
An everyday example
Two diet programmes are compared in a study of 10,000 people, and the difference in average weight loss is 0.3 kg. Because the sample is huge, p < 0.001, but with a standard deviation of 6 kg the effect is d = 0.05, which is negligible in practice. The effect size makes this distinction visible.
In MerQur
Reported in the t-tests, ANOVA, ANCOVA, MANOVA, nonparametric tests (Mann-Whitney, Wilcoxon, Kruskal-Wallis, sign test), chi-square tests and meta-analysis; the Permutation Test reports Cohen's d and Hedges' g in every run. The Effect Size Calculator also computes them from summary statistics through its 'Analysis type' option.
Analyses where this concept appears
One-Sample t-TestIndependent-Samples t-TestPaired t-TestOne-Way ANOVATwo-Way ANOVAMANOVAANCOVAPermutation TestMann-Whitney UWilcoxon Signed-RankKruskal-WallisBinomial TestSign TestChi-Square IndependenceChi-Square Goodness of FitEffect Size CalculatorMeta-AnalysisBayesian t-Test (JZS Bayes Factor)Mediation Analysis
Related concepts
Cohen's d (effect size)Eta squared (η²) and partial eta squared (η²ₚ)Cramér's V and Cohen's wOdds ratio (OR)