Dear SPMers,
I was wondering if I could get some feedback on my analysis to make sure I'm on the right track. I have two groups of subjects, each with two scans. I also have behavioral scores for the subjects at each of the two timepoints. I'm running two separate analyses: 1) for scan 1 only, and 2) for scan2-scan1 vs behavioral score changes. For the scan1 analysis, I've entered the data into a full factorial with group as the between subject factor and behavior score as covariate with group interaction. My design matrix is ordered: group1 group2 behav1 behav2. I'm testing the following with T contrasts:
1. Brain areas where there are group by behavior interactions - contrast 0 0 1 -1
2. post-hoc test on interaction results (i.e. masked with interaction contrast at p<0.05)
a) main effect of group 1 -1 0 0& -1 1 0 0
b) main effect of behav 0 0 1 1 & 0 0 -1 -1
For the scan2-scan1 analysis, I first calculated the difference images and difference behavioral scores, and then used the same steps as above. Is there a more elegant way to do this?
Another question I have about this analysis is that the groups are unbalanced - 16 in group 1 and 6 in group 2. Is it still ok to mean center within group or should I do some sort of correction to address this?
Thanks so much!
Jennie
|