Bayesian Variable Selection with Robit Model for Ordinal Data

• Activity Number: 130
• Type: Contributed
• Date/Time: Monday, August 4, 2014 : 8:30 AM to 10:20 AM
• Sponsor: Section on Bayesian Statistical Science
• Author(s): Chi-Kin Lam*+ and Guosheng Yin
• Companies: University of Hong Kong and University of Hong Kong
• Keywords: Bayesian Variable Selection ; Data Augmentation ; Hierarchical Mixture Prior Distribution ; Ordinal Data ; Robit Model

Abstract

Ordinal response data are commonly analyzed using logistic and probit regression models. However, these models are typically sensitive to outliers. A robust alternative, which is called the robit regression model, imposes Student's $t$ distribution to the model error. We study Bayesian variable selection with the robit regression model for ordinal data. In particular, we assign different hierarchical mixture prior distributions on the model parameters in conjunction with Bayesian data augmentation by introducing latent variables. Our proposed approaches are demonstrated by simulation studies and a real data example.