[Skip to Content]
[Skip to Content Landing]
Citations 0
April 2010

Effect Size Estimation as an Essential Component of Statistical Analysis

Author Affiliations

Author Affiliations: Center for Clinical Trials, Johns Hopkins Bloomberg School of Public Health, Baltimore, Maryland (Dr Vedula); and Center for Statistics in Medicine, University of Oxford, Wolfson College Annexe, Oxford, England (Dr Altman).

Arch Surg. 2010;145(4):401-402. doi:10.1001/archsurg.2010.33

The term effect size has 2 related but different meanings. In its generic form, effect size refers to any estimate of the difference in effect between 2 treatments. In its more specific sense, effect size refers to a standardized measure of effect, in which the estimated difference in a continuous measure is expressed as a multiple of the within-group standard deviation (sometimes termed Cohen d). A recent editorial published in Archives describes methods to calculate the standardized measure but almost wholly ignores the generic use of the term.1 Furthermore, the authors argue that only standardized effects are comparable across studies. This is incorrect; indeed most meta-analyses conducted in health research are based on what they call raw effect sizes.

First Page Preview View Large
First page PDF preview
First page PDF preview