Hi David, For the 2nd level, for each subject, use a design that has a single EV and the 2 timepoints, and use "Fixed Effects" (it will allow 2 inputs if you select FE first). All the best, Anderson On 24 February 2015 at 15:26, Dr. David Watson <[log in to unmask]> wrote: > Dear Mail_list > I have a study made up of two groups tested at two times (pre/post). The > FSL wiki suggests that to find effect of group, average the 2 measures > within-subject and run a 2 group mean comparison on these data in a 3 level > analysis approach. What is the best way to actually do this in practice? In > a second level model do you create an EV1 containing all the first scans > (pre) and then a second EV2 containing all second scans (post) and then use > a contrast 0.5 0.5? Followed by a simple two-group t-test model at level > three? > > Thanks in advance > > David >