Dear SPMers,
With a 2x2 design my design matrix looks like this:
A1 A2 B1 B2 A1B1 A1B2 A2B1 A2B2
I'm interested in the interaction contrast so I did 0 0 0 0 1 -1 -1 1 (or the opposite contrast).
If I understand correctly this will test voxels which show A1B1>A1B2 but A2B1<A2B2. But what if my hypothesis is that A1B1 > A1B2=A2B1=A2B2 (or A1B1 > A1B2 > A2B1=A2B2). I came up with the contrast 0 0 0 0 3 -1 -1 -1 but SPM says it's an invalid contrast.
Is it not possible to test an interaction such as the above?
Thank you in advance,
AJ
|