Hi Helmut,

so, from your point of view, I should only look a the FWE corrected value, am I right?
Of course, since it is an independent ROI (so it does not correspond a specific cluster observed at the whole brain level), I can not double check if that is really really the case, but it makes sense that it is cutting off some parts of the cluster and thus change the p-value observed.

I would be interested in hearing other people's opinions about that.


On 9 June 2015 at 15:32, H. Nebl <[log in to unmask]> wrote:
Dear everyone,

I think the uncorrected cluster p values always refer to properties of the whole brain analysis. If your VOI cuts off some parts of a cluster from the whole-brain analysis then the uncorrected cluster p value in the results table is altered, but this new uncorrected cluster p value is identical to the uncorrected cluster p value of another cluster with the same extent in the original whole-brain analysis.

In contrast, FWE-corrected cluster p values are adjusted depending on size of the VOI and properties of the VOI (sphere, box, or image, they result in different calculations).

Thus, corrected cluster p values can indeed become smaller than uncorrected ones.