Probably the most used effect size measure at least in the psychology, MH and therapy realms. A “standardised effect size” describing the effect size of a difference in means between two independent groups: it’s that mean divided by the SD.
$$ d = \frac{mean}{SD} $$
Also used to describe the standardised mean difference in within subject, repeated, paired data. The devil is in the choice of standard deviation.
Details #
I think (I would) that there’s a good introductory summary in my Rblog post: Hedges’s g and Cohen’s d. Beyond that the main issues are about the choice of standard deviation for the denominator of the simple equation above. The usual choice, I would say the only choice for the between groups value is the “common” (sometimes called, a bit misleadingly, the “pooled” SD: it’s not the SD you get pooling the observations in both groups). See that blog post for more on that SD. Things are a bit more complicated for the repeated measures situation and I’ll give that its own Rblog post when I can.
Try also #
Effect size
Hedges’s g
Repeated measures
“Standardising”
Chapters #
We didn’t put it in the book but Chapter 8 and service comparisons would probably be where you might encounter effect sizes.
Online resources #
My Rblog post about Hedges’s g and Cohen’s d
Dates #
First created 21.i.24.
