WORKSHOP IN STATISTICAL MIXTURES AND LATENT-STRUCTURE MODELLING

International Centre for Mathematical Sciences, Edinburgh,   March 28 - March 30,  2000

Murray Aitkin, Department of Statistics, University of Newcastle, UK  and Education Statistics Services Center, Washington DC, USA
`Likelihood and Bayesian analysis of mixtures'


   The talk compares likelihood and Bayesian analyses of finite mixture distributions, and expresses reservations about the latter. In particular, the role of prior assumptions in the full Monte Carlo Markov Chain Bayes analysis is obscure, yet these assumptions clearly play a major role in the conclusions. These issues are illustrated with a detailed discussion of the well-known galaxy data. It is conjectured that differences in conclusions are due at least in part to the use of the unequal-variance model and the prior distribution of the component variances.