Hi,
I'm wondering about modeling scans for exclusion in a paradigm. I am
modeling scans to remove with a single regressor that is 0 for all scans bar
the one to be removed (when it has the value 1).
My question: should these regressors be entered first or last in the model?
I read in the HBF book that when SPM models blood flow, the parameter
estimates are calculated sequentially, which makes me think that 'missing
scan' regressors should be modeled first so that they remove all variance
associated with the to-be-removed scans, and the parameter estimates I am
interested in will use only the 'good' scans. Or are the parameter
estimates calculated simultaneously so that regressor order does not matter?
Apologies for such an elementary question, any advice (or direction to
relevant literature) appreciated!
|