 |
 |
Hi Alex, I'm afraid this extra information you are providing now changes everything again... You don't really have 4 groups, just 2 groups. The T1 and T2 refer to within-subject effects. What exactly would you like to test? Are you interested in changes between timepoints and whether these changes vary according to group (interaction) or are you interested on average group differences between patients and controls? Design and contrasts would differ substantially depending on your answer. All the best, Anderson On 14 February 2018 at 11:45, SUBSCRIBE FSL Anonymous <[log in to unmask] columbia.edu> wrote: > Hi Anderson, > > Thanks again for your advice with our analysis. There is one more wrinkle, > which I did not mention earlier. > > To set this up at little, we scanned patients and controls at time 1 and > time 2 – roughly 3 months apart – (so we have 4 groups: Patient T1, Patient > T2, Control T1, Control T2). There are some subjects who participated at > one time point only or time point 2 only, and some who participated both at > time 1 and time 2 (in the same study). Clearly, the variance between the > same subject at T1 and T2 would be less than two different subjects. > However, T1 and T2 scans are always in different analysis groups. > > How would I take care of this? Would it be through sign flipping, or would > it be another approach, since the repeat subjects always fall into > different groups? Please let me know if I am not being clear, and thank you > again SO MUCH for your help! > > Sincerely, > Alex > >