Confidence Limits for the Common Odds Ratio Based on the Asymptotic Distribution of the Mantel-Haenszel Estimator

Abstract

This paper proposes a new approximate confidence limit method for the common odds ratio based on the asymptotic distribution of the Mantel-Haenszel estimator. The proposed method provides easily computed limits, and is applicable to both sparse-data and large-strata cases. Monte Carlo experiments compare its performance with that of a method based on the asymptotic distribution of the log-Mantel-Haenszel odds ratio. Applications of the proposed method to the common risk and rate ratio cases are given.