Bayesian Models for Binary Responses with Markov Dependence

• Activity Number: 520
• Type: Contributed
• Date/Time: Wednesday, August 6, 2014 : 10:30 AM to 12:20 PM
• Sponsor: Survey Research Methods Section
• Author(s): Michelle Marcovitz*+ and John W. Seaman
• Companies: Baylor University and Baylor University
• Keywords: Bayesian sample size determination ; Markov dependence ; prior distribution ; Bernoulli random variable

Abstract

Survey methodology for sensitive questions is of great interest to psychologists, social scientists, and statisticians. In this poster session, we consider a questionnaire that asks three binary sensitive questions of each individual participating in the survey. We regard the responses as dependent Bernoulli random variables and estimate the probabilities of "yes" responses to the three questions using Bayesian methods. Our Bayesian models complement the likelihood methods introduced by Klotz (1973) and Bonney (1987). We formulate a Bayesian logistic regression for these dependent dichotomous responses and consider Bayesian sample size determination.