Hi SPM Experts!
I am running a second level fMRI analysis with 2 groups and 1 covariate. It's set up as a full factorial design in spm 12.
The design matrix has 3 columns.
Column 1 is group A, Column 2 is group B, and Column 3 is the covariate.
I am interested in testing if there are any regions where my Betas scale with the covariate, but in different directions for each group.
In other words, I expect a positive relationship for group A and a negative relationship for group B in the same region.
I am not sure of the correct way to set up the contrast for this test, or if I can even use contrasts to really test this.
I had thought that I could do the following F-contrast, thinking this indicates a positive relationship with Group and and negative with group B:
1 0 1
0 1 -1
And in fact, the region I expected did show a significant cluster. However when I tried it invert the direction of the effect (to confirm I was really measuring what I thought), I essentially got the same result suggesting this didn't actually tell me what I wanted.
This is the other contrast matrix I tried (I thought this would indicate a negative relationship for group A and positive relationship for group b):
1 0 -1
0 1 1
So my questions are:
Can I test what I want with the F or T contrasts at the second level?
If I can, what is the right way to set up the contrast?
If I can't, what is the right way to do this analysis?
Thank you!
Cheers,
Brandon
|