3dLME vs 3dMVM

AFNI Message Board
1

Hello,

I would like some help deciding which multivariate analysis program to use for my repeated measures experiment: 3dLME vs 3dMVM.

My independent variable is Time. It has 4 levels and it is a repeated measure. I also have 3 nuisance behavioral measures that I would like to covary out of the analysis of the effect of Time. These three measures also change with the passage of Time (e.g., memory accuracy, memory confidence, and response time).

Thus, I would like to look at the effect of Time on brain activity with concomitant changes in those 3 nuisance measures covaried out.

Would it also improve the model to include inter-subject differences in the three covariates? That is, should I include a component in the model that codes for differences in memory accuracy, memory confidence, and response time across participants?

Thank you,

Christine

2

Christine,

My independent variable is Time. It has 4 levels and it is a repeated measure. I also have 3 nuisance
behavioral measures that I would like to covary out of the analysis of the effect of Time. These three
measures also change with the passage of Time (e.g., memory accuracy, memory confidence, and
response time).

With within-subject covariates, 3dLME is definitely the solution for you.

I would like to look at the effect of Time on brain activity with concomitant changes in those 3 nuisance
measures covaried out.

Be careful about centering if each of those covariates has different means across the 4 levels of Time: https://afni.nimh.nih.gov/pub/dist/doc/htmldoc/STATISTICS/center.html

Would it also improve the model to include inter-subject differences in the three covariates? That is,
should I include a component in the model that codes for differences in memory accuracy, memory
confidence, and response time across participants?

I don’t see any rationale for doing so if you’ve already incorporated those 3 covariates in the LME model.