Nonparametric Bayes Inference for Concordance in Bivariate Distributions
S.R. Dalai, Eswar G. Phadia · Communication in Statistics- Theory and Methods · 1983
In this article the Bayesian approach to solving various inferential problems related to dependent bivariate distributions is explored by using Ferguson's theory of Dirichlet processes. Analogs of Kendall's τ and the concordance coefficient are defined in Section 2 to deal with discrete data. The Bayesian estimator under the squared error loss turns out to be a slightly modified version of Kendall's . The analysis is carried out to include the empirical Bayes approach in Section 3. In Section 4 a Bayesian test for testing positive dependence is derived. Some small sample comparisons are carried out in Section 5. Finally a numerical illustrative example is given in Section 6.