We have multichannel EEG and EMG data of stroke patients and I am aware that they may have different phase values, and therefore if you dichotomize them, for instance according to their lesion side, you might report a false negative due to potential out of phase coherences of different patients. 

 Is there a feasible method to compare cortico-muscular coherence (CMC) between two groups of stroke patients or it MUST be done individually?