Hi Donald,
Thanks a lot for your suggestions. I'll study them in depth and let you (and the list) know if I have more questions.
I like keeping things simple, probably due to my non-statistical mind :) Therefore, if I find a reliable way to decide which FIR function is the "best" one for the purpose of my study, could it be possible to make a non-parametric randomise test of only that FIR, with a 2 paired samples design, and forget about the others?
Following this line, if I find that for subject 1 the best FIR function is #3, for subject 2 is #6... and so on, would it be acceptable to pick the beta values maps from those particular functions and put them in randomise?
Thank you again. I'll keep researching on your suggestions.
Regards,
Javier
|