Effect Size Calculator

Means, Standard Deviations, and Sample Sizes with Pretest

Note: This method adjusts Cohen’s d and Hedges’ g for baseline differences in the outcome measure. As such, it is a difference-in-differences effect size. This method requires the pretest and post-test means and standard deviations. A separate method exists for working with mean gain (change) scores. The dependent (paired) t-tests are used to determine the pretest/post-test correlation. These correlations do not affect the effect size but do affect the standard error and variance. If they are missing, a correlation of 0.50 is assumed. Alternatively, you can specify the pretest/post-test correlation directly.

There are three ways to compute the pooled within-groups standard deviation: (1) using only the pretest standard deviations, (2) using only the post-test standard deviations, and (3) using both the pretest and post-test standard deviations. The default is method 1.