|PROC MCMC Compared with Other SAS Procedures|
PROC MCMC is unlike most other SAS/STAT procedures in that the nature of the statistical inference is Bayesian. You specify prior distributions for the parameters with PRIOR statements and the likelihood function for the data with MODEL statements. The procedure derives inferences from simulation rather than through analytic or numerical methods. You should expect slightly different answers from each run for the same problem, unless the same random number seed is used. The model specification is similar to PROC NLIN, and PROC MCMC shares much of the syntax of PROC NLMIXED.
Note that you can also carry out a Bayesian analysis with the GENMOD, PHREG, and LIFEREG procedures for generalized linear models, accelerated life failure models, Cox regression models, and piecewise constant baseline hazard models (also known as piecewise exponential models). See Chapter 39, The GENMOD Procedure, Chapter 66, The PHREG Procedure, and Chapter 50, The LIFEREG Procedure.