The aim is to provide an overview of Bayes modelling emphasising the principles of prior selection, identification and interpretation, and associated issues in statistical computing in a range of modelling areas.
It includes worked illustrations of Bayesian approaches to model selection, time series (ARMA and state space methods), latent variable problems, survival analysis, general linear mixed models, analysis of designed studies (case-control studies and clinical trials), and multi-level data analysis. Standard topics such as estimation from standard distributions, tests about the parameters of such distributions, power and sample size, and are also included (with Bayesian issues to the forefront).
The fully worked examples (including listings of programs, data and initial values) include a substantial range of applications and often reveal insights additional to those in the original source studies. The intention is to reveal the benefits of the Bayesian approach to substantive analysis, as well as illustrating the range of methods which can gain from an MCMC implementation. A range of application areas is drawn on, for example in the social and health sciences.