3dMVM_ Covariates

AFNI Message Board
1

I am trying to run 3dMVM with

Between-subject factor : group (G1, G2, G3, and G4), scanner (scan1 and scan2),
Quantitative Covariates : gender, age, and accuracy

I am using the contrast obtained from SPM instead of the beta images. I am trying to get the average effect similar to what we get from SPM.
In SPM when we look at the average effect F stat-map, it seems like the covariates are given zero weight. Is 3dMVM also doing the similar things or covariates is treated in different manner? I want to control for covariates, should I give any weight in the 3dMVM code?

Also when I see the results obtained by 3dMVM ‘gltCode’, on the AFNI window, it shows like ‘average_effect Z’. Does ‘Z’ mean the z-transform image?

Thanks
Jung

2

Jung,

Why is ‘gender’ considered a quantitative variable?

I am using the contrast obtained from SPM instead of the beta images

Contrast between what and what?

Is 3dMVM also doing the similar things or covariates is treated in different manner? I want to control for
covariates, should I give any weight in the 3dMVM code?

If you center those quantitative variables properly, they would be controlled at their respective center values. As for testing an effect associated with a covariate, read the 3dMVM help and check out those examples in the help.

Does ‘Z’ mean the z-transform image?

That’s Z-statistic (under Gaussian distribution).

3

Thanks Gang.

If I make the gender as categorical variable, it is going to be a factor. right?
We usually have been regressing out the gender variable as covariates.

The contrast means the contrast between two conditions. In my case for GoNogo task, the contrast is the same as Condition : 1Nogo -1Go’

Jung