Hi Zhou, Yes, this is fine, but note that you need to run randomise twice, one for A, another for B, unless none of the subjects that did A did also B (i.e., unless there is no overlap between A and B). All the best, Anderson On 16 April 2014 14:43, Zhou Deng <[log in to unmask]> wrote: > Dear FSL experts, > > Our team has a big sample of about 300 people, some people did no > finish inventory A, some did not finish > > inventory B, we meant to run TBSS analysis (before statistic analysis) of > the whole sample (300), then deselect subjects > > according to a specific inventory. > > Could I do this by specifying design.mat by assigning the sample > into two groups( Group1 who had finished the > > inventory, Group2 who hadn't), and only use Group1 for randomise analysis? > > For example, if i have 5 subjects, subject 5 did not finished > inventory A, could I specify design.mat, design.con like this: > > design.mat: > 1 0 x1 0 > 1 0 x2 0 > 1 0 x3 0 > 1 0 x4 0 > 0 1 0 x5 > PS: x means the mean-centered score of inventory A > (across all 5 subjects). > design.con > 0 0 1 0 > 0 0 -1 0 > > Then, randomise -i AllSkeleton_FA -o tbss -m Mean_FA_mask -d > design.mat -t design.con -n 5000 --T2 -D > > Is the above listed steps ok? thanks in advance. > > > > All the best, > Zhou > > > > >