multilevel modelling and Cohen's d

[fran]

Hi Statalisters,

I am running a regression analysis to compare two groups on my outcome y (while adjusting for some covariates) while taking into account clustering using xtmixed in Stata 11.

I would also like to calculate the Cohen's d value associated with this analysis. Normally I would calculate it as:

d = abs(mean_group1-mean_group2)/pooled sd

how can I obtain the pooled standard error? should I account for the clustering to estimate it?

Any help would be appreciated.

Thanks,
Fran

[Dave Airey]

Here are two interesting (open access) articles related to your question: http://www.ncbi.nlm.nih.gov/pmc/arti...g-03-00111.pdf and http://www.ncbi.nlm.nih.gov/pmc/arti...ihms489048.pdf

[fran]

Thank you - I will check them out!

[Fanny Leyton]

Hi Fran, I want to do the same, calculate the Cohen's d value in a multilevel regresion analisys using xtmixed.
Did you figure out how to do it?

Thanks
Fanny