Hi all
is this the correct way to report the group networks observed in my population (after having obtained my dr_stage_2_icXX and dr_stage_2_subjYYY.. files) ??
- make a single column (of 1s) model
- define the 1 and -1 contrast to look for activation/deactivation
- perform a (e.g.) TFCE analysis
- report voxels > 0.95
looking at the script, it looks like randomize is performed independently for each component. isn't it?
so..
when i want to correlate networks strenght with behavioural variables, is it acceptable to perform one separate randomize for each component using a IC-specific mask.
e.g: the one obtained from the above procedure. a binary mask obtained as the sum of +1 and -1 contrasts.
this should reduce the number of voxels tested and be, in the meanwhile, formally correct
thanks in advance
regards
Alberto
|