Power Analysis for 3 (between subjects) x 3 (within subjects) ANCOVA - How to

#1
Hi Stats Community,
I'm so pleased to have discovered this resource. I'm having a lot of trouble with my power analysis for my proposal. I'll be doing a 3 x 3 mixed ANCOVA with two covariates. I've used G power but I don't think it takes the repeated measures factor into account, which I think should boost my power. Therefore, I'm not sure if it's appropriate. Is there an alternative to G Power (some similar program) or a book or article that someone could recommend? This small but essential part of my proposal is dragging me backwards in my schedule so I would really appriciate any help at all.
Thanks!!
 

Dason

Ambassador to the humans
#2
You could always simulate lots of data sets under the "alternative" hypothesis and then run the analysis of interest. The proportion of times you reject the null is an estimate of the true power.
 
#3
A far as I remember from a couple of power calculations for repeated-measures experiments, I can recall that the power for repeated measures [using G power in particular] needs also to know and enter the correlation between the groups with repeated (or matched) data. So once you choose the right test in Gpower, it would ask you the correlation between groups which have repeated / matched values. You have to compute the correlations from the raw data beforehand (and feed it to the Gpower program) to become able to analyze the power. Otherwise, it won't work.