 |
 |
Dear FSL Experts, I have two groups with resting state data (group A = 293 subjects, group B = 382 subjects) and I want to compare the DMN between them with dual regression. When constructing the study specific resting state networks in the ICA group concat analysis, I wonder if it is necessary to select an equal number of subjects from A and B to avoid selection bias (as in favouring one of the groups)? That is, using only 293x2=586 subjects instead of using all 675 subjects in the ICA step? In the dual regression step, I would of course still use all subjects. Many thanks! Best regards, Lars