Finite Population Quantile Estimators

Improved estimates of a survey population parameter can be obtained by using information about co-related auxiliary variable. There is a huge literature on such methods for estimating the mean. Here we explore some median-versions of these mean-based methods. Following an introduction in sampling theory and a brief overview on quantile regression, two new quantile based estimators are introduced and some of their properties are examined. A proof for consistency of the marginal quantile estimator and for Bahadur-Expansion validity of the conditional quantile estimator is included. Then, a look by means of the conditional double exponential likelihood model reveals the quantile base estimator connections to some of the already classical estimators. Some simulation results exploring the performance of the new estimators conclude the presentation.

Name thesis:degree_name

Ph.D.

Level thesis:degree_level

Dissertation

Discipline thesis:degree_discipline

Statistics

Grantor

University of Illinois at Urbana-Champaign

Year dc:date

2015

Author dc:creator

• Georgescu, Constantin

Contributors dc:contributor

• Stephen Portnoy

• Statistics

Language dc:language

eng

Identifier

(MiAaPQ)AAI3153300

OAI identifier oai:identifier

oai:www.ideals.illinois.edu:2142/87397