Understanding and Optimal Tightening of the KWW Joint Confidence Region for a Ranking

Written by:
RRS2024-01

Abstract

Klein, Wright, and Wieczorek (2020) presents a simple novel measure of uncertainty for an estimated ranking of K populations using a joint confidence region which tends to reveal more uncertainty for the middle of the estimated ranking than for the extremes. We explore and propose a framework which permits some control over the amount of uncertainty in various portions of the estimated ranking with an optimal allocation of sample among the K populations. A research animated data visual exists using U.S. Census Bureau data.

Page Last Revised - February 12, 2024