Dear all,
I am running a TBSS analysis and got a little confused from what I read on this list about covariates and demeaning.
I have FA images from a group of healthy subjects, along with their age, sex and reaction times.
I want to run a TBSS analysis on a behavioural measure (BM) with age and sex as covariates of no interest.
1° Is this correct to have a design matrix such as
Group EV1 (BM) EV2 (Age) EV3 (sex)
1 10 31 1
1 12 28 2
1 8 30 1
1 20 22 1
with a design constrast such as
EV1 EV2 EV3
1 0 0
-1 0 0
(for the two contrasts modeling BM with positive and negative relationship to FA)
2° Is it sufficient enough to run randomise on these data with the -D option ?
Thanks,
Josselin Houenou
|