Bayesian nonparametric tests for multivariate locations

In this paper, we propose novel, fully Bayesian non-parametric tests for one-sample and two-sample multivariate location problems. We model the underlying distribution using a Dirichlet process prior, develop testing procedure based on posterior credible region spatial median functional of distribution. For problem, fail to reject null hypothesis if set contains value. form difference medians two samples equality zero. derive local asymptotic power under shrinking alternatives, also present simulation study compare finite-sample performance our procedures with existing parametric tests.