Effect size

An effect size is a quantitative measure of how large a difference or association is, stated independently of sample size. Common forms include the raw mean difference, Cohen's d, correlation coefficients, risk ratios, odds ratios and hazard ratios. The point is separation from statistical significance. A p-value addresses whether an effect is distinguishable from zero given the data. The effect size addresses how big it is. The two are independent. With 50,000 participants, a difference of 0.4 kg in body weight can reach p below 0.001 and mean nothing clinically. With 20 participants, a real and large effect can miss the 0.05 threshold entirely. Significance reflects precision, not importance. Effect sizes belong with a confidence interval, so the reader sees both the estimate and its uncertainty. Standardised forms carry Cohen's 0.2, 0.5 and 0.8 bands, which are conventions rather than clinical thresholds. The opposite error is treating any large effect size as meaningful. Small studies overestimate effects systematically, because only the larger estimates clear the significance filter and reach publication. In longevity and nutrition research, the first report of a supplement effect is frequently two to three times larger than the pooled estimate that later trials settle on.

Last reviewed: August 25, 2026

This definition is educational and is not medical advice, a diagnosis, or treatment. Talk to a doctor about any health decisions. Read our full medical disclaimer