Hi Zheng, Please see a mock-up of a design here: https://dl.dropboxusercontent.com/u/ 2785709/outbox/mailinglist/design_zheng.ods One is for within-subject effects (differences between timepoints, further compared between groups). The other is for between-subject effects (average of timepoints, compared among groups). Hope this helps. All the best, Anderson On 8 December 2016 at 16:49, Zheng Liu <[log in to unmask]> wrote: > Hi Anderson, > > > > This project is to investigate the brain maturation effected by the degree > of the dietary. Group 1 is “control”, group 2 is “high-level” and group3 is > “low-level”. All subjects are scanned at two timepoints. I am trying to > find out the significant voxel-based defferences, including between- and > in-group. > > > > Thanks, > > Zheng > > > > *From:* FSL - FMRIB's Software Library [mailto:[log in to unmask]] *On > Behalf Of *Anderson M. Winkler > *Sent:* Thursday, December 08, 2016 2:34 AM > *To:* [log in to unmask] > *Subject:* Re: [FSL] randomise repeat ANOVA > > > > Hi Zheng, > > > > What is the hypothesis that you'd like to test? Thanks. > > > > All the best, > > > > Anderson > > > > > > On 7 December 2016 at 20:43, Zheng Liu <[log in to unmask]> wrote: > > Hi All, > > > > I have an ANOVA analysis problem using randomise. > > I have 21 subjects data and each subject was measured twiced. These 21 > subjects are divided into three groups, i.e group1=8 subjects, group2=6 > subjects, group3=7 subjects. > > Would anyone know how to design the matrix? My input data is a 4D matrix, > i.e. volume_3d*42, where the first 21 volume_3d are measured at the first > time and the order is accoring to the group1-3. The second 21 volume_3d are > measured at the second time and the order is the same. > > > > Thanks very much. > > > > Zheng > > > > >