Dear SPM people,
I have a problem with my second level design: regressors of the model are not orthogonal.
Interestingly the design matrix on the first level is almost orthogonal. But if I put the regressors on the second level they become dependent. I did flexible factorial design and cells for subjects within each regressor are also correlated if I explore the design orthogonality. This results in high overall correlations between regressors in the model.
Does anyone have an idea, how can it be, that model, which is orthogonal on the first level becomes not orthogonal on the second level?
Thanks in advance!
Katharina
|