The analysis of matched proportions

Abstract

SUMMARY The analysis of the difference between two treatments is discussed, the available data being the dichotomous responses to each of the two treatments by individuals matched in pairs. The model used in a Bayesian analysis is compared with the logistic model suggested by Cox (1958), and with a random effects model. A model which assumes the positive association of responses of matched individuals is considered. Finally, some of these methods are extended to the case when the order of application of the two treatments may matter.