 |
 |
Hi there, I am a bit confused on how to add another confound to an interaction model. So I have 2 groups and I want to check whether resting state connectivity as identified with seed analyses is affected by age (my interaction). In addition, I have a confound variable, ie several scans had a different acquisition matrix (45 instead of 47 slices). So far I have 5 EVs: 1. control, 2. patient, 3. acquisition, 4. age control, 5. age patient. Thus I have only one column for acquisition (not split into groups) and I indexed the scans with 45 slices with 1, and the majority having 47 slices with 0. a) So is it correct to just add a single column with nuisance variable to this kind of model b) is the indexing correct, or does it need to be orthotogonalised / demeaned as we may also be interested in group effects Thanks very much in advance for any input. Cheers, Torsten