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

Simulation-based approaches to nonli...

Kim, Jonghyeon.

Linked to FindBook

Google Book

Amazon

博客來

Simulation-based approaches to nonlinear measurement error models.

Record Type:	Language materials, printed : Monograph/item
Title/Author:	Simulation-based approaches to nonlinear measurement error models./
Author:	Kim, Jonghyeon.
Description:	141 p.
Notes:	Adviser: Leon J. Gleser.
Contained By:	Dissertation Abstracts International61-08B.
Subject:	Biology, Biostatistics. -
Online resource:	http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=9985054
ISBN:	0599920572

Simulation-based approaches to nonlinear measurement error models.
Kim, Jonghyeon.

Simulation-based approaches to nonlinear measurement error models. - 141 p.

Adviser: Leon J. Gleser.

Thesis (Ph.D.)--University of Pittsburgh, 2000.

Two topics are discussed in this dissertation: <italic>SIMEX approaches to measurement errors in ROC studies</italic> and <italic>a Bayesian semiparametric approach to covariate measurement error models</italic>.

ISBN: 0599920572Subjects--Topical Terms:

1018416
Biology, Biostatistics.

Simulation-based approaches to nonlinear measurement error models.
LDR:03269nam 2200301 a 45 001 929299
005 20110427
008 110427s2000 eng d
020 $a 0599920572
035 $a (UnM)AAI9985054
035 $a AAI9985054
040 $a UnM $c UnM
100 1 $a Kim, Jonghyeon. $3 1252783
245 1 0 $a Simulation-based approaches to nonlinear measurement error models.
300 $a 141 p.
500 $a Adviser: Leon J. Gleser.
500 $a Source: Dissertation Abstracts International, Volume: 61-08, Section: B, page: 4237.
502 $a Thesis (Ph.D.)--University of Pittsburgh, 2000.
520 $a Two topics are discussed in this dissertation: <italic>SIMEX approaches to measurement errors in ROC studies</italic> and <italic>a Bayesian semiparametric approach to covariate measurement error models</italic>.
520 $a The first part of the dissertation investigates the estimation of the area under an ROC curve when test scores are observed with errors of measurement. A naive approach that ignores measurement errors generally yields inconsistent estimates. Finding the asymptotic bias of the naive estimator, Coffin and Sukhatme (1995, 1997) proposed bias-corrected estimators for parametric and nonparametric cases. However, the asymptotic distributions of the bias-corrected estimators have not been developed because of their complexity. We propose several alternative approaches, including the SIMEX procedure of Cook and Stefanski (1994). We also provide the asymptotic distributions of SIMEX estimators for statistical inferences. Small simulation studies suggest that SIMEX estimators perform reasonably well compared to bias-corrected estimators.
520 $a The second part of the dissertation considers a Bayesian perspective on general errors-in-variables regression models. Taking advantage of improved computational tools, likelihood approaches have recently received much attention. However, previously proposed approaches are fully parametric in the sense that the distribution of the unobserved covariate is parametrically modeled. Parametric modeling results in greater efficiency when the assumed model is valid, but inference may not be robust to departures from the assumed model. To remedy the lack of robustness of the parametric approach, we propose a Mixture of Dirichlet Process (MDP) model that can account for departures from standard parametric modeling of the unobserved covariates. A Bayesian framework is adopted with a two-stage random effects model of the observed surrogates and estimation of the model parameters is easily carried out by embedding data augmentation within Gibbs sampling. Compared to other proposed methods, our MDP approach can accommodate various types of error-free covariates and also apply when exact measurements are available on a subset or when a suitable external validation data is available. To numerically investigate the performance of our MDP approach, comparisons to other existing methods are made in real examples.
590 $a School code: 0178.
650 4 $a Biology, Biostatistics. $3 1018416
650 4 $a Statistics. $3 517247
690 $a 0308
690 $a 0463
710 2 0 $a University of Pittsburgh. $3 958527
773 0 $t Dissertation Abstracts International $g 61-08B.
790 $a 0178
790 1 0 $a Gleser, Leon J., $e advisor
791 $a Ph.D.
792 $a 2000
856 4 0 $u http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=9985054