|
Activity Number:
|
582
- Nonparametric Methods for Statistical Inference
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Wednesday, July 31, 2019 : 2:00 PM to 3:50 PM
|
|
Sponsor:
|
Section on Nonparametric Statistics
|
|
Abstract #301750
|
Presentation
|
|
Title:
|
Improved Exact Confidence Intervals for a Proportion Using Ranked-Set Sampling
|
|
Author(s):
|
Yimin Zhang* and Jesse Frey
|
|
Companies:
|
Villanova University and Villanova University
|
|
Keywords:
|
Clopper-Pearson;
Expected length;
Imperfect rankings;
Maximum likelihood estimation
|
|
Abstract:
|
We consider the problem of finding exact confidence intervals for a proportion using ranked-set sampling (RSS). The existing exact method consists of applying the method of Clopper and Pearson (1934) to the total number of successes. We improve on this method by using the method of Blaker (2000) and by replacing the total number of successes with the maximum likelihood estimator of the proportion. All of this work assumes perfect rankings, and the reduction in average expected interval length is substantial. The new intervals are also good in an absolute sense, as they come close to a new theoretical lower bound on the average expected length.
|
Authors who are presenting talks have a * after their name.
