I ran two group source localizations of MEG data: identical but one with GS and one with MSP.
I entered the nifti files to an ANCOVA.
The GS gave me nice results, in line with the literature.
The design matrix with MSP-localized files showed that the 4 regressors were not orthogonal, even before estimation. There were no significant voxels and the SPM was greyed out.
I don't understand why the design matrix showed correlations even before estimation?
If this is not an error, can I learn, from this difference between the models, something about my data?