The use of sample survey data for estimation of the tails of distribution functions /
An examination of continuous study variables of a population may be provided by the associated distribution function and quartiles. Due to the complex design features employed in large-scale population surveys, analyses of distribution function and quartiles for the population often use design-based...
| Main Author: | |
|---|---|
| Format: | Thesis Book |
| Language: | English |
| Published: |
[Place of publication not identified] :
[publisher not identified] ;
1999.
|
| Subjects: | |
| Online Access: | http://proxy.library.tamu.edu/login?url=http://proquest.umi.com/pqdweb?did=731681401&sid=1&Fmt=2&clientId=2945&RQT=309&VName=PQD |
| Summary: | An examination of continuous study variables of a population may be provided by the associated distribution function and quartiles. Due to the complex design features employed in large-scale population surveys, analyses of distribution function and quartiles for the population often use design-based approaches. Customary design-based estimation procedures lead to consistent point estimators and valid large-sample confidence intervals. However, customary design-based methods may not provide satisfactory performance in the statistical inference for the tail quartiles due to limitations on effective sample sizes. We develop alternative model-based quantile estimators by fitting a parametric model to the tail of an underlying distribution. Extending ideas from the quantile plot literature for simple random samples, we propose a line-fitting procedure for estimating quartiles from the tail of a normal quantile plot. Least squares methods are used to fit a line through the tail plot of which in turn leads to estimators of the associated mean, variance and quartiles. We also consider use of three types of weighting matrices for the least squares fit. One is a design-based covariance matrix estimator of a weighted sample quantile vector. The other two are alternative model-based quantile covariance matrix estimators. The related simple test statistics of the goodness-of-fit and its asymptotic distribution are also investigated. The proposed methods are applied to medical examination data from the U.S. Third National Health and Nutrition Examination Survey (NHANES III). |
|---|---|
| Item Description: | Vita. "Major Subject: Statistics". |
| Physical Description: | xiv, 126 leaves : illustrations ; 28 cm. Issued also on microfiche from University Microfilm Inc. |
| Bibliography: | Includes bibliographical references (leaves 121-125). |