Dear all,

I am currently trying to figure out how I can use the Optimal Design Software (OD, link below) to do some kind of post-hoc power analysis. I am aware of the problems of post-hoc power, but nevertheless want to do the analysis.

With OD, one can create graphs displaying power on the y-axis and sample size on the x-axis. I have a repeated measures MLM design with two predictors: time (2 measurements) and group (intervention: yes/no). Time points (level 1) are nested within persons (level 2).

I computed my multilevel-model with R and the lme-function. Beside intercept there are no random effects in the model, only fixed ones (time, group, time*group).

Now I'd like to type in the appropriate values OD needs to compute the graph (for linear change and with unstandardized values).

These are:

Variability of level-1 residual ... That's the measurement error (sigma^2) and I get it through the VarCorr(model)-function

Variability of level-1 coefficient ... That's the between-person variability (tau_11) and I think there is a typo in the manual and it should read "variability of level-2 coefficient", because persons are level-2 here. I am not sure, how to get this variance value...

Treatment effect ... I thought, this would be the interesting time*group-Interaction effect (i.e., the regression coefficient b), but I'm not sure anymore...

Maybe one of you can give me a hint, what values I need to type in and where to find them?

I thank you all in advance!

Best regards,

Daniel Spitzenstätter

University of Innsbruck/Austria

Optimal Design(OD) - Manual and .exe-file:

http://hlmsoft.net/od/