Hi Alfonso,

I am trying to run a regression in CONN where I am examining if some a priori chosen ICA resting state networks (e.g., "DMN") are associated with a composite executive function outcome variable (e.g., "EF"), accounting for a couple covariates (age at MRI, sex). So my regressions are EF, age at MRI, female, male, [1 0 0 0] for DMN, and same but in the negative direction.

While I demeaned, or "mean centered" if you like, "age at MRI" covariate in Stata,  I'm not sure about the appropriate way to demean the composite "EF" outcome variable for this CONN regression analysis.

This composite "EF" outcome variable is comprised of 5 individual variables, chosen from the results of a Stata factor analysis. Is it appropriate to demean these 5 individual variables separately, and then purely sum them together to become the composite demeaned "EF" outcome variable that is used in the CONN regression analysis (i.e., the 1 in [1 0 0 0])?