Hi Anderson
Thank you very much for the help and clarifications. This is very helpful to me. I will use the design you proposed. Let me bother you with one more question about these permutations, if you have the time.
I don’t understand why we need the within permutations to test for example the group*factorA(2)*factorB(2) interaction. For example, I think a valid option would also be to build the desired contrasts in each subject individually (from 4 observations to one) and then just permute the subjects (like a two sample t-test). I guess this is then different from permuting the whole subjects (without within) in the original design with 4 observations per subject, but I don’t see why?
Best wishes,
Silvano
|