I appreciate this clear exposition of effect sizes - Stata's -esize- command is a very useful improvement! However, I would like to add a warning: Users should be aware of a Babylonian confusion of names and symbols (the Stata help to -esize- mentions this confusion, as well): Some renowned scholars denote the pooled but biased effect size measure (here: Cohen's d) as Hedges' g (which to my mind is unfortunate). As a consequence, when reporting the respective unbiased estimate calculated by -esize- one should not simply use the name "Hedges' g" but always add that it in fact denotes the unbiased effect size estimate. Similarly, when reporting "Cohen's d" one should add that it is the pooled (but biased) effect size estimate.
↧