Hi Eugene,
Thanks for your quick reply!
I hope you can help some more to clarify a few things for me:
Hope this is clear: I'm interested in both EV1 and EV2. For EV1 I want to know the averaged effect, specifically for EV2 i'm interested whether there is an additional effect of this stimvalue (the third column value in this case).
You are right, my contrast prob should be:
1 0
0 1
What I did not understand: both EV's are highly correlated in time (they consider the same stimulus presentation / duration and only differ in value). Is it therefore not essential to orthogonalize them? Or is it possible/preferred to continue without orthogonalizing?
Further, when is orthogonalizing and when is demeaning preferred, or is it the same (how do they deliver similar/diff results?)
What's your advice for including a first level covariate?
Thanks a lot!!
Anna
|