I am using STATA to perform linear regression. Here is the analysis I would like to perform:
I have weight measured at Time 1 and Time 2 (continuous variable). I have fatigue measured at Time 1 and Time 2 (continuous variable). I have race (binary) and I have the group participants were apart of (intervention or control).
I would like to assess the association between change in fatigue (IV) and change in weight (DV) while controlling for the group and interaction of race with fatigue. I do not want to use change scores, but rather simply control for weight and fatigue at baseline. So the model, without the race variable included, would look like this:
DV: weight Time 2 IV: Fatigue Time 2; Fatigue Time 1; Weight Time 1; Group
My question is in regards to how to interact race with fatigue. Should I interact race with both Fatigue at Time 2 and Time 1? Or only at Time 2?