Dr. Webster and Dr. Winkler,
Thank you for your responses. I tried running the analysis using only my variable of interest (i.e., only 1 EV), and some results were significant at .10 but not .05, which is a bit more of what I would have expected. Therefore, I am starting to think there is something going on with my model.
Dr. Winkler, with regard to your suggestion (i.e., adding a column of 1s), would the contrast setup remain the same? That is, I would now have the following, where age, sex, and education have all been demeaned:
EV1 SDMT Age Sex Edu
1 1 15 1 1.5
1 2 12 -1 2
1 3 -15 -1 -1.5
1 4 -12 1 -2 .... etc.
"Contrast & F-test":
Title EV1 SDMT Age Sex Edu
C1 Group mean 0 1 0 0 0
(If I demean the variable of interest (SDMT), do I still need to include a column of 1s?)
Then run the following:
% randomise_parallel –i all_FA_skeletonised –o tbss –m mean_FA_skeleton_mask –d comp1.mat –t comp1.con –n 5000 --T2
Thank you so much,
Cristina
|