Dear fsl experts,
I have a question as to setting up the design matrix for VBM. I use the glm_gui to set up the design matrix.
My study is to look at the association between gray matter volume and a continuous clinical measure (covariate of interest) adjusting for other confounding variables.
In addition to the group variable built in the glm_gui, do I need to set up another group variable with all values "1" in the design matrix?
I understand in fMRI analysis I need that group variable to get mean signal activity but in VBM this would be irrelevant.
Do I still need this? and what's the differences between having this group variable and not?
What I am talking about is the results from the following two design matrix:
1)
group covariate of interest confounding(age)
1 15 61
1 20 55
1 5 70
. . .
. . .
. . .
vs.
2)
covariate of interest confounding(age)
15 61
20 55
5 70
. .
. .
. .
Thanks in advance.
Yi-Fang
|