Hello. I am attempting to analyze some fMRI data in SPM8 and am encountering some problems with design estimation at the individual analysis level. I have an events-within-blocks design where each subject undergoes two sessions with two conditions each (AB - B’A’). I think I’ve specified my model correctly (see specifed matrix jpeg attached), but when I estimate the model the correlations matrix shows the two conditions (DNT & VNT) to be collinear (see attached jpeg). I am having a hard time understanding how this could be the case and how to fix it. Did I do something wrong in my model specification? Any help would be much appreciated as I am feeling rather stuck. I have tried shortening the block lengths (to make the blocks more temporally discreet) in the off chance that hrf convolution is driving the colinearity, but that did not seem to help. I am happy to provide more information as needed, but did not want to overload this message with superfluous details.
Thanks so much,
Britt Carlson Emerton
Doctoral Candidate in Clinical Psychology, Neuropsychology
Suffolk University, Boston, MA
|