Dear all,
I want to perform statistical analysis on my TBSS FA skeleton. I have two groups of subjects (CON & PAT) matched for age but not for years of education. I have the "all_FA_skeleton" with the first half of volumes CON and the second half PAT. Now I want to check if the diagnosis influences the FA and, if yes, I also want to know if such influence is age dependent or not .
With Glm GUI I created a matrix like this, let's say I have two subjects per group:
CON PAT age_CON age_PAT y_edu
1 0 a 0 e
1 0 b 0 f
0 1 0 c g
0 1 0 d h
My idea is that in this way I can model a design made by one factor (DIAGNOSIS) with two levels (CON & PAT) + 2 covariates (AGE & Y_EDU), the first (AGE) interacting with the factor DIAGNOSIS, while the second has no interactions.
To check if the diagnosis influences FA I prepared an F-test:
1 -1 0 0 0
If I'll find any significant results I will check if the difference between FA of CON & PAT depends upon age. To do that I'll use t-tests:
0 0 1 -1 0
0 0 -1 1 0
Is there any macroscopic error in the procedure described? Am I assuming only linear relations between FA and age?
Any help from the experts will be really appreciated.
Cheers,
Chiara
|