Jiansong - Replications are over obs (what your measures taken over, i.e. subjects) Correlated repeated measures as you say should be no (it would be yes if your t-test was between 2 conditions for a single group) HTH, Alexa On Mar 24 2005, Jiansong Xu wrote: > Dear Friends: > > I have 2 group subjects, one group was scanned for 8 sessions, the second > was scanned for 2 sessions. Now I want to do two sample t test. When ask > for: non-sphericity correction? Click Yes > > replications are over: obs or group (What Should I Click Here?) > > correlated repeated measures: no? > > Thanks > > Jiansong >