Auxiliary mixture sampling for count data

The advent of MCMC methods led to a burst of papers discussing the employment of Bayesian methods for modelling complex structures in the social and economic sciences. We mention here in particular the area of marketing science, where many new and exciting modeling technique are nowadays applied to problems, that could not be dealt with previously, see for instance the book by Rossi, Allenby and McCulloch (2005). Sylvia Frühwirth-Schnatter, the head of the IFAS, has been working with Bayesian statistics since her 1988 PhD thesis on dynamic Bayesian models with applications to hydrological short-term forecasting, and published a couple of papers on MCMC estimation of complex statistical models, like mixture models, or state space models. Since Sylvia moved to the IFAS in February 2003, Bayesian statistics and Markov chain Monte Carlo methods became a major focus of statistical research at the IFAS. Recently, important progress has been made concerning Bayesian inference for discrete-valued data. The MCMC group at the IFAS developed new Gibbs sampling schemes for logit type models as well as for count data model, that need draws from standard densities, only, and avoid constructing proposal densities within a Metropolis-Hastings scheme.