[FieldTrip] one-sample cluster test

Pomper, Ulrich Ulrich.Pomper at charite.de
Tue Mar 27 15:00:32 CEST 2012


Dear list members,

this topic was discussed some time in 2008, but it seems to me no 
conclusion was found.

I would like to find source activity that differs significantly from 
zero. I could run a one sample t-test for every voxel and end up having 
a huge MC problem, or (preferably) use some kind of cluster based 
permutation statistics. However, the option for a one sample cluster 
test is not implemented in FT.

Would it be correct to test the data from my experimental condition 
against a set of data containing all zeros using a cluster based 
permutation test?

Any help is very much appreciated!
Best, Ulrich




More information about the fieldtrip mailing list