Analysis of Matched Data Using Logistic Regression

Abstract

Our discussion of matching begins with a general description of the matching procedure and the basic features of matching. We then discuss how to use stratification to carry out a matched analysis. Our primary focus is on case-control studies. We then introduce the logistic model for matched data and describe the corresponding odds ratio formula. We illustrate the use of logistic regression with an application that involves matching as well as control variables not involved in matching.