See below

On Sat, Sep 21, 2013 at 10:05 AM, Qin Xu <[log in to unmask]> wrote:

Hi,
Thanks for your instant reply！
What’s the difference between demeaning within group and across all subjects? There were situations that demean within group in elsewhere while test the same model as mine, egg. https://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=ind1005&L=FSL&P=R18717&1=FSL&9=A&I=-3&J=on&X=3A86614E58392038A4&Y=qinxu.china%40gmail.com&d=No+Match%3BMatch%3BMatches&z=4

If you mean center your continuous covariate within group, the group difference contrast will no longer be adjusted for your continuous covariate. In other words, let's say your two groups have slightly different ages and you want to adjust your group difference for age. If you mean center age within group, you will not get an adjusted effect.

I didn't express my second question clearly. Yeah, I want to test the overall linear relationship between FA and behavior. But i want to make sure if i could add an EV with 1s besides the EV of behavior scores? The contrast was [0,1] if I add an additional EV considering the groups ?

If you only want to look at the linear relationship between the continuous covariate without adjusting for the 2 groups, you'd use this model

http://fsl.fmrib.ox.ac.uk/fsl/fslwiki/GLM#Single-Group_Average_with_Additional_Covariate

If you have two groups, you'd use the other model I suggested earlier.

Cheers,

Jeanette

Thanks.
Best,
Qin