Dear Donald and other gPPI users,
I am using the toolbox with an event-related design (flanker task).
In this experiment, I have a non-uniform number of regressor of no interest (any incorrect and/or missed trial, only correct trials are of interest, all different across subjects). When I include all of them in the master file (P.Tasks={'0' 'cond1' 'cond2' 'cond3' 'cond4 'cond5' 'cond6' 'cond7' 'cond8' 'cond9' 'cond10' 'cond11' 'cond12}), I get sometimes, when there is only 1 event of no interest, a problem of co-linearity (the regressor, its PPI and the constant become grey in the design matrix).
I was therefore wondering, since all the regressors (of interest and of no interest) are included in my model, am I allowed to create PPI and interaction only for my regressors of interest (P.Tasks={'1' 'cond1' 'cond2' 'cond3' 'cond4'})? Let say for example I will have my 4 regressors of interest, 2 regressors of no interest and 4 PPI created for regressor of interest + interaction + movement parameters + constant. If not, what would be the correct way to conduct this analysis?
Thanks in advance!
Cheers,
Yann
|