I am trying to fit a mediation model for my data where color of food (manipulated between subjects) predicts levels of consumption, but the relationship is mediated by the temporal dynamics of felt pleasure and guilt (10 time points, measured within-subjects). Specifically, I would like to see if the linear and quadratic trends in the emotions are different for the two color conditions, and whether these trends in turn predict consumption. I understand this is an example of a 2-1-2 model with mediator varying within-subjects but IV and DV are at the individual level. Could you give me some advice on how I should approach this?
It seems like this could simply be viewed as a single-level analysis. You have 2 groups and within each group you have growth data on emotions, that in turn predict a distal outcome of consumption. The growth modeling is done in a wide, single-level format. So it is a multiple-group growth model with a distal outcome. A couple of different UG examples can be drawn on to build this model.