We present a Bayesian method for pairwise comparisons of hypotheses in a multiple regression with partition on the parameter space of a regression coefficient. The method is derived by introducing partial and multiple Bayes factors for the regression ...
We present a Bayesian method for pairwise comparisons of hypotheses in a multiple regression with partition on the parameter space of a regression coefficient. The method is derived by introducing partial and multiple Bayes factors for the regression coefficient having a uniform conditional prior. To show the performance of the method, a comparison has been done with other Bayes tests in terms of their respective powers. The comparison via a simulation study indicates that the suggested test enables us to deal simultaneously, without loss of power and any constraints of symmetry about prior distribution, with any number of alternative hypotheses.