The full Bayesian significance test for symmetry in contingency tables

摘要

Tests for symmetry in contingency tables constitute a broad and important subarea in Statistics, and several methods have been devised for this problem. In this paper we propose the Full Bayesian Significance Test (FBST) for problems of symmetry and point-symmetry in contingency tables. FBST is an intuitive Bayesian approach which avoids to assign positive probabilities to zero measure sets when testing sharp hypotheses. Numerical experiments comparing FBST performance to power-divergence statistics suggest that FBST is a good alternative for problems concerning tests for symmetry in contingency tables.