there are different ways to conduct moderator analysis. If the modereator is categorial (e.g., sex), a multi-groups model in which you specify the model in each of the categories of the moderator is the ways to go. This allows testing the invariance of essential measurement parameters (e.g., loadings). If you allow me a self-citation:
If the moderator is semi-continous, there are the so-called "constained approaches" and "unconstrained appraoches": Recently, Little et al. proposed the "residual centering approach" which in my point of view, is the most straightforward approach.
In our paper on interaction modeling we discuss all approaches and give a tutorial:
Hi, Marsh et al (2012) provide a good overview over the unconstrained approach, which you can easily implement in AMOS.
Marsh, H. W., Wen, Z., Nagengast, B., & Hau, K. T. (2012). Structural equation models of latent interaction. In R. H. Hoyle (Ed.), Handbook of structural equation modeling (pp. 436–458). New York, NY: Guilford Press.
Iacobucci D, Saldanha N, and Deng X (2007). A Meditation on Mediation: Evidence That Structural Equations Models Perform Better Than Regressions. JOURNAL OF CONSUMER PSYCHOLOGY, 17(2), 139–153