Hello Conn Community,
 
I am interested in testing competing explanations for a behavioral difference (i.e., is the deficit related by motor connectivity over attentional connectivity. I ran a group ICA and identified the relevant networks. I think that the proper way to test this is to set a contrast under the between measures contrast to examine the unique relationship to the behavioral deficit [0 1] / [1 0] and differences between the components themselves [1 -1]. Is this correct? 

Also if I wanted a sense of network cohesion I should look at variability or frequency? Or would it be preferable to extract ICA ROIs for a graph theory analyses? 

Thank you!

Kate