Effect size calculation

[themostgod]

Hello
I have been asked to add the estimation of effect size for PALM outputs. I have been searching for similar threads for randomize or other permutation-based software packages, but there seems to be nothing. Given the non-parametric nature of the test, what would you recommend? Computing the effect size for individual clusters as [statistic]/sqrt (n)? Cohen's d seems to be far from optimal...
Thank you!
Pavel

[andersonwinkler]

Hi Pavel,
This is work in progress and it will take a while until it is fully implemented... you can nonetheless always take a test statistic and use the sample size to back convert to some equivalent effect size. R^2 is also generally a good measure.
All the best,
Anderson