3dttest++ for interaction with covariates


I am analyzing a between groups dataset (2 groups) using 3dttest++, and not sure if other function should be used rather than 3dttest++ for the following purpose. I also attached the code I used for 3dttest++ on the bottom. The datasets are seed-based resting-state functional connectivity (rsFC). Working Memory (WM) score was obtained from a behavioral task.

I would like to look at the 3-way interaction at the whole brain level of group x WM x age or the 2-way interaction of group x WM controlling age. Based on a previous post (link), I set up the analysis with setA being one group (e.g., control) and setB being the other group (e.g., patient), and put into my covariate file a column for age, a column for WM score, and an interaction term of age x WM score. I centered age and WM score for each group separately. Please let me know if this is the correct way, or are there better ways, to get the 3-way or 2-way interactions.

Also, what’s the regression model for the above 3dttest++ analysis? Is the following correct?

rsFC = a1(Age – mean_Age) + a2(WM – mean_WM) + a3(Age – mean_Age)*(WM – mean_WM) + a4(group) + a5(Age – mean_Age)*group + a6(WM – mean_WM)group + a7(Age – mean_Age)(WM – mean_WM)*group

Then, the a7 and a6 are the 3-way and 2-way interactions, respectively?


-prefix …/Results_reg/reg_WM
-center NONE
-covariates …/Results_reg/reg_covariates.txt
-setA HC
A0037 Corr_A0037_r_STG_Z+tlrc
A0046 Corr_A0046_r_STG_Z+tlrc

A1065 Corr_A1065_r_STG_Z+tlrc
A1082 Corr_A1082_r_STG_Z+tlrc
-setB SZ
A0017 Corr_A0017_r_STG_Z+tlrc
A0050 Corr_A0050_r_STG_Z+tlrc

A1161 Corr_A1161_r_STG_Z+tlrc
A1514 Corr_A1514_r_STG_Z+tlrc

XD, the most straightforward approach is to simply use the versatile program 3dMVM.

Thank Gang! Will use 3dMVM instead.