ABSTRACT

Perhaps the greatest strength of BUGS is its flexibility — the language can represent statistical models and data structures of arbitrary complexity. We now have the tools and skills to tackle a theoretically limitless range of models, knowing the basics of Bayesian analysis and MCMC computation (Chapters 1-5), the Bayesian view of the standard linear and generalised linear models pervasive in applied statistics (Chapters 6-7), how to assess and compare models (Chapter 8), and how BUGS can deal with the common complications of real data analysis (Chapter 9). Chapter 10 discussed hierarchical models, which are naturally suited to the graphical modelling principles of BUGS, and now we present an overview of many other specialised applications. Of course, the flexibility of BUGS comes at the cost of learning the many idiosyncrasies of the language and its practical limitations, and we hope to illustrate these issues in the examples here.