The MCMC Procedure

Getting Started: MCMC Procedure

There are three examples in this Getting Started section: a simple linear regression, the Behrens-Fisher estimation problem, and a random-effects model. The regression model is chosen for its simplicity; the Behrens-Fisher problem illustrates some advantages of the Bayesian approach; and the random-effects model is one of the most prevalently used models.

Keep in mind that PARMS statements declare the parameters in the model, PRIOR statements declare the prior distributions, MODEL statements declare the likelihood for the data, and RANDOM statements declare the random effects. In most cases, you do not need to supply initial values. PROC MCMC advises you if it is unable to generate starting values for the Markov chain.