Key points are not available for this paper at this time.
Abstract This article studies the construction of Bayesian confidence interval for the difference of two proportions in the matched-pair design, and applies it to the equiva-lence or non inferiority test. Under the Dirichlet prior distribution, the exact posterior distribution of difference of two proportions is derived. The tail confidence interval and the highest posterior density (HPD) interval are studied, and their frequentist performance are investigated by simulation in terms of the mean coverage probability of interval. Our results suggest to use tail interval at Jeffreys prior for testing equivalence or non inferiority in matched-pair design. Keywords: Bayesian analysisConfidence intervalDirichlet priorEquivalence or non inferiority testMatched-pair designMathematics Subject Classification: Primary 62F15Secondary 62F25 Acknowledgments We are very grateful to an anonymous referee for many helpful comments and suggestions which make the current version available. This research was supported by National Natural Science Foundation of China (10761010).
Ши et al. (2008) studied this question.