Empirical Likelihood Confidence Intervals for ROC Curves with Missing Data

• Activity Number: 172
• Type: Contributed
• Date/Time: Monday, August 1, 2011 : 10:30 AM to 12:20 PM
• Sponsor: Section on Nonparametric Statistics
• Author(s): Yichuan Zhao*+ and Yueheng An
• Companies: Georgia State University and Georgia State University
• Keywords: Confidence interval ; Missing data ; ROC curve ; Smoothed empirical likelihood

Abstract

The receiver operating characteristic (ROC) curve is widely utilized to evaluate the diagnostic performance of a test, in other words, the accuracy of a test to discriminate normal cases from diseased case. In biomedical studies, we often meet with missing data. In this situation, the regular inference procedures cannot be applied directly. In this paper, a random hot deck imputation is used to obtain a "complete sample". After that, empirical likelihood (EL) confidence intervals are constructed for the ROC curves. The empirical log-likelihood ratio statistics is derived whose asymptotic distribution is a weighted chi-square distribution. The results of simulation study show that the EL confidence intervals perform well in terms of the coverage probability and average length for various sample sizes and respondent rates.