Dear Peter,
Thank you so much for your useful remarks,
But I think in my example the transpose of the matrix symbol after bracket at the end of the line wasn’t clear, in my example I have two columns, the first column is about ones and the second column is about a covariate of no interest like handedness, so I have 18 subjects and two covariates in my design matrix.
In my real design matrix in addition to handedness, I have other covariates of no interest like age, etc.
my objective for using the design matrix is removing the effect of these covariates of no interest from inference in the group-level analysis (connectivity).
As I understand from some texts in GLM by demeaning (like mean-center the covariates) and define contrast can remove effect of covariates of no interest, as I see in PEB in SPM we can’t define contrast, on the other hand, if I mean-center the covariates, first column show average connectivity (commonalities and other columns show group differences) that the result of average group connectivity is similar to when I didn’t define the covariate of no interest in my design matrix ( when just we have a column of one in design matrix, that in PEB if we didn’t define any design matrix, consider this automatically), and now I want to know, is the average group connectivity the only way to achieve my goal or should I design the design matrix the other way to get the desirable results? Or is there a better way generally?
Best regards,
soroor
|