I have performed an experiment wherein the same people got to rate product 1 and product 2 on several items. These items together form latent constructs. I am modeling a simple model where I estimate the paths between 1 independent latent variable, a mediator (latent) and a dependent (observed) variable. I also drew a direct path from the IV to the DV. Now I want to compare this model between the two products. I performed a group-comparison and results looked fine. However, the same people rated both products. So these should be considered within-groups. How do I deal with that when wanting to compare these groups in AMOS? Are you allowed to do a normal group-comparison in AMOS or should you do this in a different way? Can anyone help?