Dear Hernando,
> We want to analyze a PET study comparing two groups: 4 patients in
> group 1 and 16 in group 2. For each patient, there are two different
> conditions (Awake and Sleep) and one covariate. Furthermore, there is
> only one scan for each subject/condition. We would like to test for the
> differences between sleep and awake between the two groups. I have a
> few questions:
>
> (1.) If there was no covariate included, it seems that the contrast
> vector should be [1 -1 -1 1]. How does one adjust for the covariate?
> (2.) Is there a test for a covariate effect in SPM96?
Simply entering the covariate into the design matrix adjusts the
parameter estimates and ensuing T values, The contrast should look
like [1 -1 -1 1 0] to assess the sleep/wake x group interaction and [0
0 0 0 1] to look at the main effects of covariate over all subjects.
I hope this helps - Karl
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|