 |
 |
Dear all, I am interested in assessing the differential relationship between serum testosterone levels and brain activation in 2 groups of subjects (a clinical group vs a control group). I have set up a flexible factorial with group as a factor and testosterone level as a covariate (technically there are 2 covariates, namely one for each group), giving me a a design matrix with 4 columns (the first two indicating the groups, and the second two indicating the two covariates). I then ran a contrast to examine the effect of the covariate in one group against the other group: [0 0 1 -1]. My main question is: how do I interpret the resulting clusters of activation? Does this indicate that the slope of the relationship between the covariate and the activation level is greater (more positive) in group 1 compared to group 2 in that particular cluster? Does anyone have a suggestion as to how to best plot the group differences in terms of the correlation between the covariate and the activation level? I have saved the clusters as ROIs, and have extracted the beta-values for the contrast of interest (which was taken to the second level in the flex factorial) for each individual subject in each of the two groups. And I have correlated those values against the individuals' testosterone levels, which I can plot in a scatter diagram. Does this make sense? I realize that this is a fairly basic question, but I could not find the answer on the mailing list, so hopefully someone can assist. Thanks, Dr. Ans Vercammen Postdoctoral Research Officer Neuroscience Research Australia Sydney