categorical covariate in 3dMVM

AFNI Message Board
1

Hi, I’m new to AFNI but have spent a lot of time reading through the threads and handouts and am still trying to get the hang of 3dMVM.

I have a quick question: I’m using 3dMVM to do a simple model with two between-subjects factors. There are no within-subjects factors. I can’t use 3dANOVAx because I have unequal sample sizes in each group. I want to assess for differences between two groups (‘disease’ and ‘no.disease’), controlling for a categorical covariate (study.site, which has 3 levels).

I’m scripting for it as follows:
3dMVM -prefix MVM_output -jobs 6 -bsVars group+study.site -SS_type 2 -num_glt 1 -gltLabel 1 disease_vs_no.disease -gltCode 1 ‘group : 1disease - 1no.disease’ -dataTable @datatable.nifti.txt

I’m getting nice results, but I’m not sure whether I have properly controlled for site effects by simply specifying it under -bsVars.

Thanks very much,
Jaya

2

Welcome to AFNI, Jaya!

I’m getting nice results, but I’m not sure whether I have properly controlled for site effects by simply specifying it under -bsVars.

Your model is fine except that you assumed that there would be no interaction between the two factors. So I would also try the following

-bsVars “group*study.site”

and see how the two models compare to each other.

3

Thanks so much for your reply.

I just have a quick confirmatory type question: I’m assuming that using num_glt/gltCode yields t values for output, and using num_glf/glfCode yields F ratios. However, it seems the t values are all positive - so I just wanted to confirm that the absolute value of the t is outputted by 3dMVM (using num_glt).

Thanks and sorry for this elementary question.
Jaya

4

Never mind! I just realized that when viewing nifti files with multiple sub-bricks in FSLview, it doesn’t show the other sub-bricks. Thus I wasn’t seeing the t-test sub-brick with positive and negative values.

Thank you,
Jaya