> How does the search volume affect the severity of correction for
multiple
> comparisons in SnPM? In SPM the severity of the multiple comparison
> correction is proportional to the volume of search volume (to the
number
> of resels). What kind relation can be expected in SnPM?
Smaller the search volume, less severe the multiple comparison
correction becomes. Unfortunately there is no formula to describe the
relationship between the search volume and the multiple comparison
correction in SnPM, since it is a non-parametric method. I imagine it
wouldn't be very different from SPM, but probably it will depend on your
data.
> We know that the permutation distribution of the maximal statistic
deals
> with the multiple comparisons correction. Is it just that with a small
VOI
> the critical threshold of the permutation distribution of maximal
> statistic
> gets lower as compared to the whole brain analysis thus enhancing
> sensitivity?
Yes, you are correct. SnPM does this correction for the search volume
implicitly.
-Satoru
|