Dear FSL experts,
In my analysis I have an interaction term between two binary variables, coded as '0' and '1'. Normally, with non-fMRI data, I would just create an interaction term by multiplying these two variables together. However, the GLM wiki of FSL leads me to believe that this is different when constructing a .mat and .con file for Randomise. Is it truly wrong to do it 'the old-fashioned way', and if so, could you please explain why?
Another issue that I'm running in to with these binary variables is that the Randomise wiki suggests to use the -D option for demeaning, is this good to use even with '0' (reference group) and '1' coding?
Thank you very much for helping me. Regards, Johannes
|