Exact Logistic Regression for a Matched Pairs Case-Control Sampling Design with Polytomous Exposure Variable.
by
Shyam Sundar Ganguly
Assistant Professor, Department of Epidemiology and Medical Statistics, College of Medicine, Sultan Qaboos University, P.O. Box 35, Muscat 123, Oman

Logistic regression methods have been found useful in the estimation of 'odds ratios' under matched pairs case-control design when the exposure variable of interest is binary or polytomous in nature. Usually, the analysis is performed using large sample approximation techniques. While carrying out the analysis with polytomous exposure variable, sometimes we encounter the situation where the number of discordant pairs in the resulting cells are small or the data structure sparse. In such situations, the asymptotic method of analysis is in question and hence an exact method of analysis may be more suitable. In this paper we discuss a method that uses the distribution of the conditional sufficient statistics of the logistic model parameters to perform exact inference in case of a pair-wise matched case-control data with more than two unordered exposure categories.We describe a computational method which can be used for obtaining the combinatorial coefficients involved in the distribution.

Date received: October 13, 2002

Copyright © 2002 by the author(s). The author(s) of this document and the organizers of the conference have granted their consent to include this abstract in Atlas Conferences Inc. Document # cais-46.