Hi Alfonso, 

Thank you so much for this clarification. Would it then be accurate to say that using multivariate regression is better when looking for the unique contribution of each ROI that is within a known network? For example, if I was looking at the DMN and wanted to see the unique contribution of the PCC after controlling for the contribution of the MPFC, LLP, etc, then it would be better to employ multivariate methods? But if examining a number of different a priori ROIs that are not necessarily a part of a FC network (as in my study; ie., insula, hippocampus, putamen) then it would be better to use separate regression models for each ROI (i.e., bivariate)?

With regard to the way I set it up, yes I did employ the steps that you suggested by using dummy-coding and making C1-C4 variables all 0 for control subjects.

However after doing step c) in the second-level results tab select 'ASD', 'C1' ... 'C4' in the subject-effects list and enter a between-subjects contrast [0 1 0 0 0; 0 0 1 0 0; 0 0 0 1 0; 0 0 0 0 1]

... would I scroll through the source ROIs on the right separately to see how my predictor variables are associated with each ROI? Or together, since the results are only giving me connectivity associated with ANY of the four C1 - C4 variables?

And for future - if I did ever want to employ multivariate regression would I also set up the second-level results in the same way? ie [0,1,0,0,0; 0,0,1,0,0; 0,0,0,1,0; 0,0,0,0,1]

Thank you so so much, this has been so helpful!

Jenna