Statistics Glossary → Cohen's d (effect size)
Cohen's d (effect size)concept
An effect size that expresses the difference between two means in standard deviation units.
The p value tells whether a difference can be explained by chance, but not how large it is; with very large samples even trivial differences can be significant. Cohen's d makes the difference independent of the measurement unit: d = (mean₁ − mean₂) / pooled standard deviation.
By Cohen's (1988) rough benchmarks, d ≈ 0.2 is a small, 0.5 a medium and 0.8 a large effect, although what counts as important depends on the field. In small samples d is slightly inflated; the corrected version is called Hedges' g. The sign only shows the direction of the difference.
An everyday example
Suppose people on a new diet programme lost 3 kg on average and those not on it lost 1 kg. If weight change varies by about 4 kg from person to person, the 2 kg difference is half a standard deviation (d = 0.5, a medium effect); if weight change varies by only 1 kg, the same 2 kg difference is two standard deviations (d = 2, a very large effect).
In MerQur
In the effect size section of the t-tests, together with the Hedges' g, Glass's Δ and CLES options, with a normal-approximation or noncentral-t confidence interval.