Spring Seminar Series - February 27, 2003
University of Minnesota
School of Statistics
College of Liberal Arts
Weighted Empirical Likelihood
Jian-Jian Ren
Department of Mathematics
University of Central Florida
Thursday, February 27, 2003
4:00 PM, 115
Ford Hall
Minneapolis, East Bank Campus
Social at 3:30 PM, 300
Ford Hall
Abstract
Weighted empirical likelihood is a newly developed likelihood method which
can be used to construct tests and confidence intervals for various types
of incomplete data, such as right censored data, doubly censored data and
interval censored data. In this talk, we discuss the confidence intervals
for survival probabilities based on weighted empirical likelihood. The
theoretical coverage accuracy of the confidence intervals is established
in a unified form for the censored data aforementioned. Due to technical
difficulties, work on interval estimates for survival probabilities with
interval censored data Case 2 does not as yet exist in the literature.
Here a solution for this problem is provided. Some simulation results
will be presented with comparison between the weighted empirical likelihood
method and alternative methods, and an example of interval censored Case
2 data set from AIDS research will be discussed.