3dMVM and covariates

Hello there,

I have a question about 3dMVM and covariates

My 3dMVM analysis has two within subjects factors - load (3 levels )and chromophore (2 levels). These are data from infants and children of different ages, but each infant/child has a value for each level of a factor. We think that our effects be it main effect of load and/or main effect of chromophore and/or interaction between load and chromophore might be driven by Age. Hence , we would like to use Age (in days) as a covariate. In this case, do I need to center Age with qVarsCenters) ? It would seem unnecessary as an average of the ages for each load and/or chromphore would be the same.

Does this seem accurate?

Any advice is appreciated.

Thank you.


each infant/child has a value for each level of a factor

Does age vary across the levels of load or chromophore, or both?

Each kid has a value for each of the three levels of load and each of the two levels of chromophore. So age wouldn’t vary between load 1, load 2 or load 3. Similarly, age wouldn’t vary between chromophore 1 and chromophore 2.

As an e.g…:

Age(days) Load Chromophore Inputfile
1000 1 1 xyz1.nii
1000 1 2 xyz2.nii
1000 2 1 xyz3.nii
1000 2 2 xyz4.nii
1000 3 1 xyz5.nii
1000 3 2 xyz6.nii
2000 1 1 xyz7.nii
2000 1 2 xyz8.nii
2000 2 1 xyz9.nii
2000 2 2 xyz10.nii
2000 3 1 xyz11.nii
2000 3 2 xyz12.nii

Does this answer your question?


Since age for each subject does not vary across the levels of both factors, you can let 3dMVM to automatically perform the default centering around the overall mean. In other words, you do’t have to do anything regarding the option -qVarsCenters.

Great. Thanks!!