Improved Auxiliary Mixture Sampling for Hierarchical Models of Non-Gaussian Data.pdfVIP

Improved Auxiliary Mixture Sampling for Hierarchical Models of Non-Gaussian Data Sylvia Fru?hwirth-Schnattera, Rudolf Fru?hwirthb, Leonhard Heldc and H?avard Rued aDepartment of Applied Statistics and Econometrics Johannes Kepler Universita?t Linz, Austria bInstitute of High Energy Physics Austrian Academy of Sciences, Vienna, Austria cInstitute of Social and Preventive Medicine University of Zurich, Switzerland d Department of Mathematical Sciences Norwegian University of Science and Technology, Trondheim, Norway May 29, 2007 Abstract The article proposes an improved method of auxiliary mixture sampling for count data, binomial data and multinomial data. In constrast to previously proposed samplers the method uses a limited number of latent variables per observation, independent of the intensity of the underlying Poisson process in the case of count data, or of the number of experiments in the case of binomial and multinomial data. The smaller number of latent variables results in a more general error distribution, which is a negative log-Gamma distribution with arbitray integer shape parameter. The required approximations of these distributions by Gaussian mixtures have been computed. Overall, the improvement leads to a substantial increase in efficiency of auxiliary mixture sampling for highly structured models. The method is illustrated on two epidemiological case studies. Key words: Count data, Binomial data, Disease mapping, Gaussian mixture, Log-Gamma distribution, Multinomial data Submitted to Journal of Computational and Graphical Statistics 1 Introduction During the past years, auxiliary mixture sampling has turned out to be a useful tool for the Bayesian analysis of hierarchical and parameter-driven models of non-Gaussian data. The method has been used first by Shephard (1994) for stochastic volatility models and has been applied in this context by a couple of authors (Kim et al., 1998; Chib et al., 2002; Omori et al., 2004). Recently, auxiliary mixture s