Thanks for the quick reply!
To verify, as there are quite a few possibilities: In case the assumptions of SPM hold, then which specific models for a within-subject model would be correct?
1) Flexible factorial with factor "subject"
2) Flexible factorial without factor "subject"
3) Full factorial with factors (independence no, variance equal) according to design of the experiment
4) Full factorial similar to 3) with additional factor "subject" (independence yes, variance unequal)
As far as I can see, Full factorial does not directly compare e.g. con image 1 of subject 1 with con image 2 of the same subject, so this approach should be wrong in any case?
|