Modeling Multiple Types of Outcomes Using SAS® PROC MCMC

• Activity Number: 414
• Type: Contributed
• Date/Time: Tuesday, August 5, 2014 : 2:00 PM to 3:50 PM
• Sponsor: Section on Bayesian Statistical Science
• Author(s): Sitaram Vangala*+ and Li-Jung Liang and Yih-Ing Hser
• Companies: University of California, Los Angeles and University of California, Los Angeles and University of California, Los Angeles Integrated Substance Abuse Programs
• Keywords: Bayesian model ; Zero-inflated model ; Longitudinal study ; Substance use ; Incarceration

Abstract

Many public health studies involve multiple response types. In this paper, we develop various joint models to account for correlations between repeated measurements (e.g., zero-inflated counts) and binary outcomes, leading to more precise estimates. We adopt a Bayesian approach, using SAS® PROC MCMC, a recently introduced Markov chain Monte Carlo sampler, to implement the proposed models. A long-term follow-up study of narcotics addicts is used to illustrate our approach.