A Bayesian Approach to Using Trajectories of Alcohol Use as Predictors of Young Adult Behavior

• Activity Number: 363
• Type: Contributed
• Date/Time: Wednesday, August 14, 2002 : 2:00 PM to 3:50 PM
• Sponsor: Biometrics Section*
• Author(s): Sujata Patil*+ and Trivellore Raghunathan and Jean Shope
• Affiliation(s): University of Michigan and University of Michigan and University of Michigan Transportation Research Institute
• Keywords: Bayesian ; Trajectories ; Longitudinal Data ; Hierarchical Models

Abstract

Longitudinal data are often used to study individual developmental growth curves or "trajectories." Some substantive analyses explore the predictive relationship between trajectories and a future event. Here, we are interested in studying trajectories of adolescent alcohol use as predictors of vehicle crashes incurred during young adulthood. Analysis methods for these types of substantive questions have proceeded in two stages. In stage 1, individual trajectories are estimated and summarized by two or more latent trajectory variables. For example, in polynomial models the coefficients for linear or quadratic terms are the latent trajectory variables. In stage 2, the latent trajectory variables are used as predictors of a future event. This traditional approach conditions on the estimated trajectories, ignoring the measurement error in these estimates and affecting inference. In this presentation, I describe a Bayesian approach used to account for all uncertainties. Results from the Bayesian method are compared with results from the traditional approach that are corrected to account for measurement error. Supported by NIAAA grant AA09026 and NIDA training grant T32DA07267.