Albert and Chib (1995)'s latent variable representation of the Bayesian probit regression model for categorical outcomes is widely recognized to facilitate model fitting. This representation has also been used in various settings to incorporate residual dependence into regression models with discrete outcomes. In this talk, we further extend this latent variable strategy to specify models for multicategory spatially-dependent outcomes. In particular, we discuss parameter identifiability issues in the latent mean specification and introduce covariance structures for describing the cross spatial/category residual dependence. We also consider data augmentation MCMC strategies for improving the efficiency of model fitting algorithms. Finally, we illustrate the proposed modeling framework through an analysis of land-cover/land-use observations taken over mainland Southeast Asia.

Keywords: Spatial statistics; Categorical data analysis; Bayesian modeling; MCMC

Biography: Catherine Calder is Associate Professor of Statistics in the Department of Statistics at the Ohio State University. Her research interests include spatial and spatial-temporal statistics, Bayesian modeling and computation, and applications in the environmental, social, and health sciences.