南開科技大學圖書館 |

Language: 繁體中文

說明(常見問題)

南開科技大學

圖書館首頁

編目中圖書申請

登入

回首頁

切換: 標籤 | MARC模式 | ISBD

Assessing and correcting the effects...

Dube, Tina Juliet Thandeka.

Assessing and correcting the effects of measurement error among correlated covariates in a proportional hazards setting.

紀錄類型:	書目-電子資源 : 單行本
正題名/作者:	Assessing and correcting the effects of measurement error among correlated covariates in a proportional hazards setting./
作者:	Dube, Tina Juliet Thandeka.
面頁冊數:	131 p.
附註:	Source: Dissertation Abstracts International, Volume: 69-08, Section: B, page: 4509.
Contained By:	Dissertation Abstracts International69-08B.
標題:	Biology, Biostatistics. -
電子資源:	http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=3316439
ISBN:	9780549697381

Assessing and correcting the effects of measurement error among correlated covariates in a proportional hazards setting.
Dube, Tina Juliet Thandeka.

Assessing and correcting the effects of measurement error among correlated covariates in a proportional hazards setting. - 131 p.

Source: Dissertation Abstracts International, Volume: 69-08, Section: B, page: 4509.

Thesis (Ph.D.)--The University of Alabama at Birmingham, 2008.

Older drivers, when contrasted with their younger counterparts, tend to drive fewer miles annually, yet their motor vehicle crash rates adjusted for miles driven are higher than all groups except the youngest drivers. Identifying factors associated with the risk of automobile crashes among the elderly is vital. In many elder driver studies, error free covariates such as gender or age, and error prone covariates like annual mileage or cognitive performance, are used to study time to an at-fault motor vehicle crash. We performed a simulation study that mimics this situation, using both an error free covariate as well as a correlated mismeasured covariate with replicates, to explore the effects of measurement error while conducting proportional hazards modeling. The pair, correlation and measurement error are considered, since both affect parameter estimates. We used modified regression calibration and risk set calibration to account for and correct for the measurement error among correlated variables.

ISBN: 9780549697381Subjects--Topical Terms:

1000006828
Biology, Biostatistics.

Assessing and correcting the effects of measurement error among correlated covariates in a proportional hazards setting.
LDR:03428nmm 2200337 4500 001 1000005445
005 20101022142722.5
008 101022s2008 ||||||||||||||||| ||eng d
020 $a 9780549697381
035 $a (UMI)AAI3316439
035 $a AAI3316439
040 $a UMI $c UMI{me_controlnum}
100 1 $a Dube, Tina Juliet Thandeka. $3 1000006826
245 1 0 $a Assessing and correcting the effects of measurement error among correlated covariates in a proportional hazards setting.
300 $a 131 p.
500 $a Source: Dissertation Abstracts International, Volume: 69-08, Section: B, page: 4509.
500 $a Adviser: Leslie A. McClure.
502 $a Thesis (Ph.D.)--The University of Alabama at Birmingham, 2008.
520 $a Older drivers, when contrasted with their younger counterparts, tend to drive fewer miles annually, yet their motor vehicle crash rates adjusted for miles driven are higher than all groups except the youngest drivers. Identifying factors associated with the risk of automobile crashes among the elderly is vital. In many elder driver studies, error free covariates such as gender or age, and error prone covariates like annual mileage or cognitive performance, are used to study time to an at-fault motor vehicle crash. We performed a simulation study that mimics this situation, using both an error free covariate as well as a correlated mismeasured covariate with replicates, to explore the effects of measurement error while conducting proportional hazards modeling. The pair, correlation and measurement error are considered, since both affect parameter estimates. We used modified regression calibration and risk set calibration to account for and correct for the measurement error among correlated variables.
520 $a The focus of this research was consideration of approaches that correct for measurement error when using correlated covariates in Cox Regression modeling. Three methods, regression calibration, risk set calibration and maximum likelihood each adjusted for correlated covariates, were explored. More specifically, the former methods were used to simulate situations where two correlated covariates are used as prognostic factors to determine a relationship between time to an at-fault or fault unknown motor vehicle crash. After construction of data conforming to the correlation/measurement error structure, the data were then analyzed via the aforementioned methods. The resulting parameter estimates were then assessed to determine the ability of each method to correct for the effects of correlation and measurement error. The amount of both absolute and relative bias, along with coverage, mean square error and relative efficiency was examined at each correlation/measurement error design point. As a result of this simulation study, it was concluded that regression calibration corrected for correlation method tended to outperform risk set calibration corrected for correlation in the estimation of the true parameter estimates.
590 $a School code: 0005.
650 4 $a Biology, Biostatistics. $3 1000006828
650 4 $a Gerontology. $3 168436
690 $a 0308
690 $a 0351
710 2 $a The University of Alabama at Birmingham. $b Biostatistics. $3 1000006827
773 0 $t Dissertation Abstracts International $g 69-08B.
790 1 0 $a McClure, Leslie A., $e advisor
790 1 0 $a Ball, Karlene $e committee member
790 1 0 $a Katholi, Charles $e committee member
790 1 0 $a Roth, David $e committee member
790 1 0 $a Voeks, Jenifer $e committee member
790 $a 0005
791 $a Ph.D.
792 $a 2008
856 4 0 $u http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=3316439