Hi Daniel,

Yes - this design looks ok.

Note that this design will model variance across the sessions, yet across-session variability is irrelevant to your hypothesis. In other words, you can test your hypothesis simply by acquiring one scan in each subject.

Therefore, an alternative design would be to average the four connectivity matrices across each subject and also average the four behavioral measures across subjects.

You could try both designs.

Andrew


Originally posted by Daniel Brennan: