Effect Size and Statistical Power in Psychological Research

Abstract

The probability associated with an inferential statistic, such as the t-test, is a function of the sample size employed in the study; it gives no indication of the size of effect, such as the magnitude of the difference between two conditions. This article explains how measures of effect size fulfil this role, and emphasises the importance of reporting effect sizes to help evaluate the results of a study. In addition, it shows how statistical power analysis should be conducted during the design stage of a study in order to choose an appropriate sample size. Problems which result from failure to take power and effect size into account are described, including preliminary data analysis, tests of assumptions, and model testing; possible solutions to these problems are also given. A range of effect sizes is provided for different types of analysis, and means for conducting power analysis are described