SMALL-SAMPLE PROPERTIES OF ODDS RATIO ESTIMATORS UNDER MULTIPLE MATCHING IN CASE CONTROL STUDIES

Small-sample properties of old and new odds ratio estimators are studied with regard to bias and mean squared error using Monte Carlo experiments. Their performances in interval estimation are also compared. A new and simple estimator is found to perform better than almost any other estmator, including the conditional likelihood estimator, both for point and interval estimation for a wide range of sample sizes.