For my dissertation I am looking at the temporal order of physiological and behavioural events before a voluntary action. I have 12 different repeated measures variables including: Time of intention (W), Readiness potential 1 (as measured by task A) Readiness potential 2 (task A), RPmax (task A), ERDalpha(task A), ERDbeta (task A), Time of probed intention (T), Readiness potential 1 (as measured by task B) Readiness potential 2 (task B), RPmax (task B), ERDalpha(task B), ERDbeta (task B).

And with these variables in order to establish the order of events I need to make many many comparisons (and use the appropriate bonferroni correction).

My issue is that some of the variables are normally distributed and some aren't - for example, time of intention (W) is normally distributed, but Readiness potential is not.

I know that if one of your varibles is not normallly distributed, you should use a non-parametric test ie Wilcoxons. But should I be using Wilcoxons for some of theses comparisons and then for the odd comparison jump back to a t-test? Or should I just use Wilcoxons for all my comparisons?

Thanks