Effect size measures are a way to separate the strength of an effect. An effect might be the difference between the mean of people’s scores on a measure in a help-seeking group from the mean of scores on the same measure in a non-help-seeking group. Using effect sizes rather than the raw mean difference allows for comparison across different measures. For example, the effect size for that comparison for the CORE-OM could be compared with that for another measure. To some extent effect size measures can allow some comparison not just for exactly the same effect but also across different scenarios, for example an effect size measure can be used to compare the relative strength of the relationship between that score mean score difference on the CORE-OM against say the mean scores differences on say the OQ-45 across three categories of gender. There the use of an effect size has allowed comparison across different measures, across different predictor variables and across different numbers of groups in the predictors.
Cohen’s d is arguably the paradigmatic effect size measure in our field and developed for two group mean comparisons like that help-seeking versus non-help-seeking above.
The full range of effect size measures is huge and the field is pretty complicated. For once I’m not sure I recommend the wikipedia post about it. I’ll try to find something or do more myself.
Details #
This all sounds nice and the best way (I think!) to get a sense of the simple end of the details is to go to my Rblog post about Hedges’s g and Cohen’s d.
Try also #
My Rblog post about Hedges’s g and Cohen’s d
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 #
So far I’m only suggesting my Rblog post about Hedges’s g and Cohen’s d (getting repetitive!)
Dates #
First created 20.i.24.
