Dear FSL Group,
I am interested in determining the combined effect of two EVs, and ideally, whether that effect is significantly different from the effect of either EV individually. Is there a way to do that?
If I selected an F test to cover those EVs as
F test1
EV1 0 1 0 X
EV2 0 0 1 X
I think that just tells me whether either variable is associated with fMRI signal rather than giving me a combined effect, right? I see in the archives that the F tests are non-directional, but if they cover 2 EVs, would significant results only appear in voxels in which both EVs have effects in the same direction (like giving me either the overlap of the two separate contrasts)? Alternatively, should I be adding the 2 EVs together to create a new EV in order to assess combined effects? If I did that I'd have to do it in a separate analysis from the original two EVs to avoid collinearity though, right? So I wouldn't be able to compare the combined EV to the original ones statistically?
Thank you!
Meredith
|