Hi it's me again! Coming to you with a question about including covariates in a model.

So I'm running a mixed effect model in my longitudinal imaging data. I have 4 possible covariates to include: age, sex, ICV, and years of education.

I ran correlations between these covariates and my brain measures (separately at each time point). My mixed model will be set up with each region over 3 years.

My questions are: if a covariate is significantly correlated with one region but not another, do I have to include it in all models in order to maintain consistency in all my analyses? Also, let's say a covariate is significantly correlated with a baseline measure but not at years 1 and 3, should I still include it in the model?

Thanks!