LIKELIHOOD RATIO TEST BASED CONFIDENCE INTERVAL FOR THE RELATIVE RISK OF Two INDEPENDENT BINOMIAL SAMPLES

摘要

In this paper we study the range of the confidence interval of the relative risk of two independent binomial outcomes based on the likelihood ratio test (LRT). We also compare the coverage probability of this test with some other widely used methods. Our simulation studies show that the LRT based confidence interval generally has better and more stable coverage than other other methods for different sample sizes and different combinations of the success probabilities. This is a follow-up work of Feng, Wang and Tu (2010).