Advances and Applications in Statistics
Volume 47, Issue 1, Pages 65 - 89
(October 2015) http://dx.doi.org/10.17654/ADASOct2015_065_089 |
|
ANALYSIS OF THE NHANES III DATA SET USING RANKED SET AND JUDGMENT POST-STRATIFIED SAMPLES
Jeffrey Gory and Omer Ozturk
|
Abstract: Ranked set sampling (RSS) and judgment post-stratification (JPS) are two sampling procedures that have been shown to lead to more efficient estimators of population parameters than the more common simple random sampling (SRS) procedure. This paper provides a new approach for analysis of the Third National Health and Nutrition Examination Survey (NHANES III) data set using JPS and RSS sampling designs. An important aspect of these procedures is that one must assign judgment ranks to the units in the population based on auxiliary information. We explore a model-based approach to ranking and investigate how different ranking procedures impact RSS and JPS estimators of population means and proportions. Through a simulation study, we compare the RSS, JPS and SRS sampling procedures under various ranking models. We find that both RSS and JPS lead to more precise estimators than SRS. Further, JPS has practical advantages over RSS. Finally, we establish bootstrap inference for the rank-based sampling procedures. |
Keywords and phrases: rankingmodel,judgmentranking,coverageprobability,empty classes, bootstrap inference. |
|
Number of Downloads: 396 | Number of Views: 1264 |
|