A Simple Guide to Effect Size Measures.

ABSTRACT

Importance Effect size quantifies the magnitude of the difference or the strength of the association between variables. In clinical research it is important to calculate and report the effect size and the confidence interval (CI) because it is needed for sample size calculation, meaningful interpretation of results, and meta-analyses. Observations There are many different effect size measures that can be organized into 2 families or groups-d family and r family. The d family includes measures that quantify the differences between groups. The r family includes measures that quantify the strength of the association. Effect sizes that are presented in the same units as the characteristic being measured and compared are known as nonstandardized or simple effect sizes. The nonstandardized effect sizes have the advantage of being more informative, easier to interpret, and easier to evaluate in the light of clinical significance or practical relevance. Standardized effect sizes are unit-less and are helpful for combining and comparing effects of different outcome measures or across different studies (ie, meta-analysis). Conclusions and Relevance The choice of the correct effect size measure depends on the research question, study design, targeted audience, and the statistical assumptions being made. For a complete and meaningful interpretation of results from a clinical research study, the investigator should make clear the type of effect size being reported, its magnitude and direction, degree of uncertainty of the effect size estimate as presented by the CIs, and whether the results are compatible with a clinically meaningful effect.