ABSTRACT
A large number of problems in statistics involves comparisons of different populations or the same population subjected to different treatments. In classical statistics, these problems are framed in the rubric of hypothesis testing and estimation of parameters. The conventional Bayesian approach is similar but differs in calculating either posterior odds of various parametric hypotheses or posterior distribution of parametric functions. The following is typical: Let two independent random samples X (N 1) and Y (N 2) be drawn from populations having possible different means µ and η say. A standard practice is to make a comparison of µ and η or in particular a posterior probability region interval on µ – η.
