In general, if you don't find a significant interaction, then you don't need to model both factors. However, it depends on how you defined "significance". I would use a more liberal threshold to exclude the possibility of interaction. An alternative approach would be to run the two-sample t-test in each scanner separately and see if you get the same results. Best Regards, Donald McLaren, PhD On Sun, Nov 22, 2015 at 9:50 PM, Jess Reynolds <[log in to unmask]> wrote: > Dear SPM mailing list, > > I am undertaking a VBM study where data was collected on two different > scanners. The data set has clinical and control groups, with data from the > same number of participants from each of the two groups collected in each > of the scanners. When I run my grey matter group contrast analyses, is it > correct to run as a two sample t-test, with the scanner entered as a > covariate (scanner 1 entered as 0s and scanner 2 entered as 1s)? > > I have checked that there are no differences in the group contrasts by > running a full factorial design with two factors: 1) Group (2 levels), and > 2) Scanner model (2 levels), and there are no significant clusters for the > interaction between group and scanner. > > Thank you in advance for assistance. > Jess. >