> a further question about the exclusive mask: if i use p = 0.5
uncorrected
> threshold in the exclusive mask, I would be restricting my analysis on
A
> in
> areas where the test on B were very unlikely to be rejected. will this
> make
> my arguement that the resultant areas indicate "A and not B" stronger?
>
> i can see this may reduce the power to find areas activated for A
becasue
> the exclusive mask will be much larger than when set at p = 0.05, but
> since
> my interest is in "A and not B" for this particular test, finding less
A
> related activation might be fine.
Well, yes, it will make your argument "A but not B" stronger, but you
will pay the price for it. Your exclusive mask will include only voxels
with p>0.5 for B (i.e., voxels with negative T-scores). So even if there
is no activation associated with B, you will be throwing out half of all
the voxels.
-Satoru
|