Alcohol abuse has become a major public health concern in the United States. In view of the enormous costs involved in addressing this problem, it is imperative that researchers have access to appropriate statistical tools to analyze their data. In terms of statistical methodology, there has been a growing interest in recent years in fitting models to data collected from longitudinal surveys using complex sample designs. This interest reflects the needs of policy makers and researchers for in-depth studies of social processes over time. Many of the longitudinal studies on alcohol abuse are characterized by the fact that each individual longitudinal record is short, while the number of records is large. Consequently, software for structural equation modeling (SEM) and multilevel modeling is used to analyze these longitudinal data sets. We propose to adapt the existing LISREL program to provide researchers with correct estimates, standard errors, and fit measures under complex sampling schemes in the case of SEM for continuous variables. We further propose to conduct a feasibility study with the objective of extending this methodology to handle complex sampling schemes in the case of SEM for a mixture of continuous and ordinal variables and in the case of multilevel modeling.
