CROSSVALIDATION OF LATENT CLASS MODELS OF EARLY SUBSTANCE USE ONSET

Cudeck and Browne (1983) were among the first to discuss the advantage s of taking a crossvalidation approach to testing of covariance struct ure models. The purpose of this paper is to expand on Cudeck and Brown e's work in two directions. The first direction of expansion is into t esting of latent class models. The second direction of expansion invol ves using crossvalidation to examine differences between groups, where groups may be formed by gender, ethnicity, region, etc. In the presen t article crossvalidation is used to help select models of early subst ance use onset in a sample of young adolescents. The results suggest t hat the nature of the substance use onset process and the rate of move ment through the process are the same for males and females at seventh and eighth grade. However, the present study did find evidence for ge nder differences in substance use experience at the beginning of seven th grade, with males somewhat more advanced in the onset process. The results also suggest that double crossvalidation is greatly to be pref erred over single crossvalidation.