Hi Mahsa, None of these options is really appropriate for this hypothesis. Instead, please follow this example from the manual: http://fsl.fmrib.ox.ac.uk/fsl/fslwiki/GLM#Two_Groups_with_continuous_covariate_interaction All the best, Anderson On 21 October 2015 at 17:44, Mahsa <[log in to unmask]> wrote: > Dear FSL experts > > I'm interested to see if the correlation with a given score is different > among patients and controls. There are 2 options for GLM setup: > > 1) specify the two groups as 1 and 2 in the first column (the “Group” > column) and the score of interest as EV1. > > 2) specify the groups as separate EVs (EV1 and EV2) and the score as EV3. > > I should also mention that I'm only interested in difference between the > two groups in correlation with the score and not the difference in groups > mean values. > > I'd appreciate any comments on the above options. Would these two designs > return different results? > > Many thanks > Mahsa >