Actually, the wiki page is not quite right. For a simple interaction between age and weight, you would need to create a new regressor = age*weight, then set the contrast to [0 0 0 0 0 0 1]. If you want an interaction between group, age, and weight, then you would need two new regressors, age*weight for group1 and age*weight for group2, then the contrast will be [0 0 0 0 0 0 1 -1]. I would suggest demeaning both age and weight prior to computing the product.


On 8/27/2019 10:12 AM, Zhi Li wrote:

        External Email - Use Caution        

Dear Freesurfer experts,

May I ask how to understand the contrast of interaction between covariates in the example https://surfer.nmr.mgh.harvard.edu/fswiki/Fsgdf2G2V? The interaction of group x age x weight is:

[0 0 1 -1 0 0
 0 0 0 0 1 -1]

Why the interaction between group and weight should be put in the second row (what does each row mean in the F-contrast matrix)?


If there are three levels in the factor one (three groups), I guess the interaction of group x age x weight should be:

[0 0 0 1 -1 0 0 0 0
 0 0 0 0 1 -1 0 0 0
 0 0 0 0 0 0 1 -1 0
 0 0 0 0 0 0 0 1 -1] 

Is this correct? If not how should I set this contrast?

In addition, if I would like to know the interaction between age and weight without group, how should I set the contrast then?

Looking forward to your kind suggestions.

Thanks and best wishes,

Lizhi



_______________________________________________
Freesurfer mailing list
Freesurfer@nmr.mgh.harvard.edu
https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer