東華大學圖書館 |

Language: English

Help

回圖書館首頁

手機版館藏查詢

Back

Switch To: Labeled | MARC Mode | ISBD

Design and efficient estimation in r...

Zhao, Yang.

Linked to FindBook

Google Book

Amazon

博客來

Design and efficient estimation in regression analysis with missing data in two-phase studies.

Record Type:	Electronic resources : Monograph/item
Title/Author:	Design and efficient estimation in regression analysis with missing data in two-phase studies./
Author:	Zhao, Yang.
Description:	118 p.
Notes:	Source: Dissertation Abstracts International, Volume: 66-06, Section: B, page: 3216.
Contained By:	Dissertation Abstracts International66-06B.
Subject:	Statistics. -
Online resource:	http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=NR02963
ISBN:	0494029633

Design and efficient estimation in regression analysis with missing data in two-phase studies.
Zhao, Yang.

Design and efficient estimation in regression analysis with missing data in two-phase studies. - 118 p.

Source: Dissertation Abstracts International, Volume: 66-06, Section: B, page: 3216.

Thesis (Ph.D.)--University of Waterloo (Canada), 2005.

Regression analysis that involves incomplete observations can utilize auxiliary information to provide more efficient estimates of regression parameters. In addition, many two-phase studies were designed to obtain efficient estimates of regression parameters while minimizing the costs of data collection.

ISBN: 0494029633Subjects--Topical Terms:

517247
Statistics.

Design and efficient estimation in regression analysis with missing data in two-phase studies.
LDR:03465nmm 2200277 4500 001 1817535
005 20060814143914.5
008 130610s2005 eng d
020 $a 0494029633
035 $a (UnM)AAINR02963
035 $a AAINR02963
040 $a UnM $c UnM
100 1 $a Zhao, Yang. $3 964626
245 1 0 $a Design and efficient estimation in regression analysis with missing data in two-phase studies.
300 $a 118 p.
500 $a Source: Dissertation Abstracts International, Volume: 66-06, Section: B, page: 3216.
502 $a Thesis (Ph.D.)--University of Waterloo (Canada), 2005.
520 $a Regression analysis that involves incomplete observations can utilize auxiliary information to provide more efficient estimates of regression parameters. In addition, many two-phase studies were designed to obtain efficient estimates of regression parameters while minimizing the costs of data collection.
520 $a In this thesis we consider efficient estimation methods for regression models with missing or mismeasured covariates or responses. We have developed maximum likelihood methods for estimating regression parameters from a semiparametric likelihood where the model for response given covariates is parametric and other nuisance distributions or conditional distributions are nonparametric. Surrogates or auxiliary variables in the nuisance distributions are used to increase the efficiency in parameter estimation. An EM algorithm for obtaining the estimates has been described. These methods work well even though the nuisance parameter is infinite dimensional. Implementation is easy for covariates or responses missing at random problems. Profile likelihood methods of variance and interval estimation have also been established. Extensive simulation studies are reported. The analyses of data from a population based case-control study of leprosy and data from a prevalence study of dementia illustrate these methods.
520 $a In regard to two-phase sampling we have also examined the information available from incomplete observations and auxiliary variables through the study of the asymptotic relative efficiency of the maximum likelihood estimators for some generally used parametric regression models. The results of asymptotic relative efficiency indicate (i) a two-phase study can be more efficient than a complete sample study; (ii) the optimal sizes of the two-phase samples can be estimated for both fixed budget and fixed precision problems; (iii) regression parameters can be consistently estimated based on phase I observations alone if the unobserved variable is binary; (iv) regression parameters are estimable when only the cases or the controls are observed at phase II for binary regression models with continuous covariate missing at random.
520 $a Finally, the optimal design problem for stratified simple random sampling at phase II is discussed for some fully parametric regression models. We considered studies using only complete observations and studies using both complete and incomplete observations separately. In a simulation study we compare the optimal design with the balanced design and the simple random sampling design. Some recommendations for two-phase sampling designs are provided.
590 $a School code: 1141.
650 4 $a Statistics. $3 517247
690 $a 0463
710 2 0 $a University of Waterloo (Canada). $3 1017669
773 0 $t Dissertation Abstracts International $g 66-06B.
790 $a 1141
791 $a Ph.D.
792 $a 2005
856 4 0 $u http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=NR02963