Comparisons of conditional logistic regression vs. a discriminant function approach in a case-control study where matching is performed Open Access

Li, Ruoxing (Spring 2020)

Permanent URL:


The logistic regression model has been widely used for estimating adjusted odds ratio associated with a binary outcome in case-control study. When matching is involved, conditional logistic regression is more commonly used to estimate the odds ratio corresponding to a continuous predictor as an alternative to standard unconditional model to decrease the bias caused by sparse data. In this thesis the discriminant function approach is suggested to generate closed-form estimators, especially under conditions involving few or small matched sets. The application of this approach, which given a multiple regression model form with the continuous predictor of interest on the outcome, includes fixed intercept effects for each matched set. It is demonstrated that the estimator based on discriminant function approach outperform the usual maximum likelihood estimator from logistic regression based on our simulation works and examples. The advantages have seen in reducing bias and width of CI for odds ratio, as well as generating reliable estimator under separation situations where logistic regression fails. Potential improvements for this study are also talked in the end of the article.

Table of Contents

Introduction 1

Methods 3

---Standard logistic regression 4

---Conditional logistic regression for matched case-control studies 5

---Discriminant function approach 7

Examples 10

Simulation Studies and Results 14

Discussion 19

Reference 22


About this Master's Thesis

Rights statement
  • Permission granted by the author to include this thesis or dissertation in this repository. All rights reserved by the author. Please contact the author for information regarding the reproduction and use of this thesis or dissertation.
Subfield / Discipline
  • English
Research Field
Committee Chair / Thesis Advisor
Committee Members
Last modified

Primary PDF

Supplemental Files