Hi All,
I'm setting up a PPI analysis where I'm interested in looking at several psycho-
physiological interactions. Is it best to set up a model for each PPI of interest
(so, all task EVs, 1 ROI time series, 1 PPI) or to include all PPIs in one model
(all task EVs, 1 ROI time series, all PPIs of interest). The second option leads
to a lot of multicolinearity, so I'm leaning towards the first; just thought I'd
ask the advice of all of the PPI experts out there.
Also, is it valid to do contrasts between two PPIs derived from different
GLMs? In this case, I'd run a first level FEAT analysis for each PPI indidually
and do contrasts at the second level.
Thanks,
Liz Selgrade
|