ABSTRACT
A review of the structural equation modeling literature reveals a dearth of studies involving multi group comparisons of latent mean structures. This situation likely derives from the many complexities associated with the implementation of these analyses using earlier versions of the various SEM statistical packages. Indeed, it was this very issue that led to the writing of the first paper to explicitly model and demonstrate the necessary specification "tricks" required in testing for differences in mean structures using pre-LISREL 7 (Joreskog & Sorbom, 1988) versions of that program (see Byrne et al., 1989). However, over the past 2 to 3 years, recent versions of SEM programs have addressed these complexities and, as a result, the process is becoming more straightforward. As a consequence of this simplification trend, I fully expect more studies involving latent mean structures to be forthcoming; currently, they are still relatively far and few between. The AMOS program was designed to make mean and intercept modeling easy, and that it has done in spades. Believe me, testing for latent mean differences can't get much simpler than the AMOS approach to these analyses!
