Hi Anderson,
thanks for the quick response.
So i Have 40 subjects overall (20 each Group)
36 fMRI nuisance variables (for our purpose i think each of them should get two EVs, so 72EVs)
2 cortisol data (this would sum up to 4 EVs)
3 Rating Scales (6 EVs)
Age and Gender (one EV for each of them should be enough, because they are highly correlated)
and 2 EVs of interest.
If I understood you correctly, I need more EVs than Subjects. In the best case 10-20 subjects more than EVs.
That means I cannot put everything in one matrix. Does the same go with testing for interactions? And how would you recommend to proceed with the matrices? (would the 5 different design matrices, as explained in my previous message, be a good idea in my case with 86 EVs and 40 Subjects?)
I didn't quite understand what you meant by
"If you happen {...}, although making the model more parsimonious by reducing redundancies is often a good idea."
Thanks a lot. You are of great help to me!
Berel
|