東華大學圖書館 |

語系: 繁體中文

說明(常見問題)

回圖書館首頁

手機版館藏查詢

登入

回首頁

切換: 標籤 | MARC模式 | ISBD

Selected Problems in Survey Sampling...

Stubblefield, Alexander Ross.

FindBook

Google Book

Amazon

博客來

Selected Problems in Survey Sampling Design and Survey Missing Data Analysis.

紀錄類型:	書目-電子資源 : Monograph/item
正題名/作者:	Selected Problems in Survey Sampling Design and Survey Missing Data Analysis./
作者:	Stubblefield, Alexander Ross.
出版者:	Ann Arbor : ProQuest Dissertations & Theses, : 2019,
面頁冊數:	65 p.
附註:	Source: Masters Abstracts International, Volume: 80-12.
Contained By:	Masters Abstracts International80-12.
標題:	Biostatistics. -
電子資源:	http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=13862045
ISBN:	9781392170885

Selected Problems in Survey Sampling Design and Survey Missing Data Analysis.
Stubblefield, Alexander Ross.

Selected Problems in Survey Sampling Design and Survey Missing Data Analysis. - Ann Arbor : ProQuest Dissertations & Theses, 2019 - 65 p.

Source: Masters Abstracts International, Volume: 80-12.

Thesis (M.S.)--The University of Oklahoma Health Sciences Center, 2019.

This item must not be sold to any third party vendors.

Sample surveys are becoming increasingly more common as they allow for the collection of data on a population using only a sample of units from that population. These sample surveys have cost and significant time savings and provide for the ability to estimate quantities of a population that would be otherwise impossible to estimate. Two core components of survey sampling are sample design and data analysis. Within sample design, many different methods of selecting the sample from the population exist, including methods to minimize variance whilst controlling cost and methods to oversample minority populations for more precise estimators. We propose a novel dual-frame oversampling approach with optimal allocation of sample sizes, which has great potential use in future dual-frame landline and telephone surveys and is found via real data application to be superior to a non-oversampled approach. For data analysis, missing data presents a large and frequent problem, as nonresponses cannot be ignored without risking substantial biases. Many methods for imputing missing values exist, including predictive mean matching imputation. Predictive mean matching imputation, however, relies upon the correct specification of the regression model used to predict the means. This is not robust and risks an incorrect model being used and biasing estimates. We propose a multiply robust predictive mean matching approach for imputation that offers one the ability to specify multiple regression models, increasing the chance that one of the models is correct. The imputation method is valid if one of the models is correctly specified. This method is demonstrated via simulation study to be superior to prior approaches when the models of the prior approaches are incorrectly specified, and to not significantly underperform prior approaches even when the prior approaches use a correct mode.

ISBN: 9781392170885Subjects--Topical Terms:

1002712
Biostatistics.

Selected Problems in Survey Sampling Design and Survey Missing Data Analysis.
LDR:02957nmm a2200325 4500 001 2208544
005 20191021073500.5
008 201008s2019 ||||||||||||||||| ||eng d
020 $a 9781392170885
035 $a (MiAaPQ)AAI13862045
035 $a (MiAaPQ)ouhsc:10186
035 $a AAI13862045
040 $a MiAaPQ $c MiAaPQ
100 1 $a Stubblefield, Alexander Ross. $3 3435580
245 1 0 $a Selected Problems in Survey Sampling Design and Survey Missing Data Analysis.
260 1 $a Ann Arbor : $b ProQuest Dissertations & Theses, $c 2019
300 $a 65 p.
500 $a Source: Masters Abstracts International, Volume: 80-12.
500 $a Publisher info.: Dissertation/Thesis.
500 $a Advisor: Chen, Sixia.
502 $a Thesis (M.S.)--The University of Oklahoma Health Sciences Center, 2019.
506 $a This item must not be sold to any third party vendors.
520 $a Sample surveys are becoming increasingly more common as they allow for the collection of data on a population using only a sample of units from that population. These sample surveys have cost and significant time savings and provide for the ability to estimate quantities of a population that would be otherwise impossible to estimate. Two core components of survey sampling are sample design and data analysis. Within sample design, many different methods of selecting the sample from the population exist, including methods to minimize variance whilst controlling cost and methods to oversample minority populations for more precise estimators. We propose a novel dual-frame oversampling approach with optimal allocation of sample sizes, which has great potential use in future dual-frame landline and telephone surveys and is found via real data application to be superior to a non-oversampled approach. For data analysis, missing data presents a large and frequent problem, as nonresponses cannot be ignored without risking substantial biases. Many methods for imputing missing values exist, including predictive mean matching imputation. Predictive mean matching imputation, however, relies upon the correct specification of the regression model used to predict the means. This is not robust and risks an incorrect model being used and biasing estimates. We propose a multiply robust predictive mean matching approach for imputation that offers one the ability to specify multiple regression models, increasing the chance that one of the models is correct. The imputation method is valid if one of the models is correctly specified. This method is demonstrated via simulation study to be superior to prior approaches when the models of the prior approaches are incorrectly specified, and to not significantly underperform prior approaches even when the prior approaches use a correct mode.
590 $a School code: 0361.
650 4 $a Biostatistics. $3 1002712
650 4 $a Statistics. $3 517247
690 $a 0308
690 $a 0463
710 2 $a The University of Oklahoma Health Sciences Center. $b Biostatistics. $3 3435581
773 0 $t Masters Abstracts International $g 80-12.
790 $a 0361
791 $a M.S.
792 $a 2019
793 $a English
856 4 0 $u http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=13862045