A didactic discussion of a latent variable modeling approach is presented that addresses frequent empirical concerns of social, behavioral, and educational researchers involved in longitudinal studies. The method is suitable when the purpose is to analyze repeated measure data along several interrelated dimensions and to explain some of the associated patterns of change in terms of other developmental patterns via regressions among random effects. The procedure utilizes the framework of latent variable modeling and in addition is readily applicable when not all subjects provide complete records on all assessments and variables. The approach is illustrated with data from a cognitive intervention study.