Dear FSL group,
I've also looked through the archives and had the same question. I have two groups (controls and patients) and want to examine behavioral performance as a covariate of interest. Please assume that my covariate is demeaned according to the respective group. My model is below:
Group EV1(controls) EV2(patients) EV3(covariate controls) EV4(covariate patients)
1 1 0 -2 0
1 1 0 4 0
...
2 0 1 0 5
2 0 1 0 -2
...
Contrasts:
Controls-Patients: 1 -1 0 0 (show me the voxels where controls have greater activation than the patients taking into account my covariate)
Patients-Controls: -1 1 0 0 (show me the voxels where patients have greater activation than controls taking into account my covariate)
Covariate (controls): 0 0 1 0 (show me the voxels where controls have a positive correlation with the covariate)
Covariate (controls): 0 0 -1 0 (show me the voxels where controls have a negative correlation with the covariate)
Covariate (patients): 0 0 0 1 (show me the voxels where patients have a positive correlation with the covariate)
Covariate (patients): 0 0 0 -1 (show me the voxels where patients have a negative correlation with the covariate)
Could you please let me know if my model, contrasts, and my interpretation of the covariate in relation to my model is correct? Also, if I find activation differences between the groups for the first 2 contrasts, could I conclude that it is not due to differences in my covariate? Would it matter if there are differences in the covariate between the groups (I.e. the controls have significantly better behavioral performance than the patients)?
|