Introduction to Bayesian Statistics and WinBUGS

 

3 Day Workshop

 

Russell Millar

Department of Statistics

University of Auckland

 

Tentative Outline:

 

Day 1: Powerpoint presentation

  1. Introduction to Bayesian methodology and comparison with frequentist approach.

  2. Obtaining the posterior distribution using Bayes theorem. Simple example using normal data with quick demonstration in WinBUGS. Bayesian inference including point and interval estimation, hypothesis "testing".

  3. Prior distributions. Reference priors, vague priors, informative priors, hierarchical models, sensitivity to the prior.

  4. Implementation. Markov chain Monte Carlo for sampling from the posterior distribution, Metropolis-Hastings algorithm, Gibbs sampler, the dangers of MCMC.

 

Day 2: Introduction to WinBUGS

  • Morning: Introduction to WinBUGS: Model syntax, compiling the model, initial values, updating, monitoring.

  • Afternoon: Practice with real data.

    •  

    Day 3: WinBUGS continued and introduction to CODA

  • Morning: More WinBUGS (e.g., DoodleBUGS) and demonstration of CODA (Convergence Diagnostics Analysis) for checking the WinBUGS output.

  • Afternoon: Practice with real data.

    •