Effect size: A statistical basis for clinical practice

Abstract

Objective: Effect size (ES) is the statistical measure which quantifies the strength of a phenomenon and is commonly applied to observational and interventional studies. The aim of this review was to describe the conceptual basis of this measure, including its application, calculation and interpretation. Results: As well as being used to detect the magnitude of the difference between groups, to verify the strength of association between predictor and outcome variables, to calculate sample size and power, ES is also used in meta-analysis. ES formulas can be divided into these categories: I. Difference between groups, II. Strength of association, III. Risk estimation, and IV. Multivariate data. The d value was originally considered small (0.20 > d ≤ 0.49), medium (0.50 > d ≤ 0.79) or large (d ≥ 0.80); however, these cut-off limits are not consensual and could be contextualized according to a specific field of knowledge. In general, a larger score implies that a larger difference was detected. Conclusion: The ES report, in conjunction with the confidence interval and P value, aims to strengthen interpretation and prevent the misinterpretation of data, and thus leads to clinical decisions being based on scientific evidence studies.