Skip to main content

Paper No. 05-15

Download 05-15

MA Juarez

Normal Correlation: An Objective Bayesian Approach

Abstract: In this paper we give a decision-theoretic oriented, objective Bayesian answer to the problems of point estimating and sharp hypothesis testing about the correlation coefficient of a bivariate Normal population. Under this view both problems are deemed closely related and thus a coherent answer is developed. Comparisons with frequentist results are given and an alternative interpretation of the maximum likelihood estimator is found.

Keywords: Bayesian reference criterion, intrinsic divergence, intrinsic estimator, logarithmic discrepancy, reference prior.