Hi Everyone,

I am running a panel model using population averaged model. I have different hypothese regarding the main effect and some moderating effect. for instance, the model 1 is only the main effect, model 2 includes the moderating effect number 1, model 3 includes moderating effect 2, and the overall model includes both of the moderating effects. Now, the question is that based on which model should I calculate the marginal effect of moderating factor? the overall model? or the models I used separately for each moderators?

Thanks a lot!