Degree Type

Dissertation

Date of Award

2018

Degree Name

Doctor of Philosophy

Department

Statistics

Major

Statistics

First Advisor

Jae-kwang Kim

Second Advisor

Zhengyuan Zhu

Abstract

Stratification is a common tool in survey sampling to reduce variance of estimates. Currently in the literature, there is very little work done in multivariate stratification - that is to stratify by multiple variables. We wish to stratify using multiple variables as a survey produces estimates for multiple variables. We focus on modernizing an agricultural survey that uses an area frame conducted by the National Agricultural Statistics Service. We first examine improving the stratification by automating the process then we extend to the multivariate case. Finally, we discuss model-based stratification for minimizing anticipated variance of estimates.

Copyright Owner

Stephanie Ann Zimmer

Language

en

File Format

application/pdf

File Size

95 pages

Share

COinS