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

Analysis of clustered longitudinal c...

Gao, Dexiang.

FindBook

Google Book

Amazon

博客來

Analysis of clustered longitudinal count data.

紀錄類型:	書目-語言資料,印刷品 : Monograph/item
正題名/作者:	Analysis of clustered longitudinal count data./
作者:	Gao, Dexiang.
面頁冊數:	110 p.
附註:	Source: Dissertation Abstracts International, Volume: 68-11, Section: B, page: 7049.
Contained By:	Dissertation Abstracts International68-11B.
標題:	Biology, Biostatistics. -
電子資源:	http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=3289830
ISBN:	9780549328933

Analysis of clustered longitudinal count data.
Gao, Dexiang.

Analysis of clustered longitudinal count data. - 110 p.

Source: Dissertation Abstracts International, Volume: 68-11, Section: B, page: 7049.

Thesis (Ph.D.)--University of Colorado Health Sciences Center, 2007.

Matched cohort study designs appear often in medical research. These designs match nt exposed study subjects to nc unexposed subjects on one or more characteristics. The purpose of matching is to control measured and unmeasured confounding effects and to increase estimation precision. A group of matched subjects is often referred to as a cluster. The interest of this dissertation is in the within-cluster exposure effect for count outcomes. Two types of data are studied, one is clustered cross-sectional count data in which there is one measurement for each subject, and the other is clustered longitudinal count data, where several measures are made on each subject. The motivation for this thesis project was a study conducted in young children in Indonesia to examine the association of incidence of Acute Lower Respiratory Tract Infection (ALRI) in children with characteristics of the mother. The outcome variable was the number of ALRIs during the follow-up period. A cross-sectional dataset and a repeated measures dataset were collected. The time during the year when the child was born could be an important factor for ALRI. To adjust for this effect, children were grouped into 24 birth cohorts of one half-month each according to their birth dates.

ISBN: 9780549328933Subjects--Topical Terms:

1018416
Biology, Biostatistics.

Analysis of clustered longitudinal count data.
LDR:04177nam 2200289 4500 001 1394041
005 20110419112704.5
008 130515s2007 ||||||||||||||||| ||eng d
020 $a 9780549328933
035 $a (UMI)AAI3289830
035 $a AAI3289830
040 $a UMI $c UMI
100 1 $a Gao, Dexiang. $3 1672625
245 1 0 $a Analysis of clustered longitudinal count data.
300 $a 110 p.
500 $a Source: Dissertation Abstracts International, Volume: 68-11, Section: B, page: 7049.
500 $a Adviser: Gary K. Grunwald.
502 $a Thesis (Ph.D.)--University of Colorado Health Sciences Center, 2007.
520 $a Matched cohort study designs appear often in medical research. These designs match nt exposed study subjects to nc unexposed subjects on one or more characteristics. The purpose of matching is to control measured and unmeasured confounding effects and to increase estimation precision. A group of matched subjects is often referred to as a cluster. The interest of this dissertation is in the within-cluster exposure effect for count outcomes. Two types of data are studied, one is clustered cross-sectional count data in which there is one measurement for each subject, and the other is clustered longitudinal count data, where several measures are made on each subject. The motivation for this thesis project was a study conducted in young children in Indonesia to examine the association of incidence of Acute Lower Respiratory Tract Infection (ALRI) in children with characteristics of the mother. The outcome variable was the number of ALRIs during the follow-up period. A cross-sectional dataset and a repeated measures dataset were collected. The time during the year when the child was born could be an important factor for ALRI. To adjust for this effect, children were grouped into 24 birth cohorts of one half-month each according to their birth dates.
520 $a For the cross-sectional clustered count data, we studied six statistical methods including an independent Poisson model, fixed cluster effects Poisson model, conditional likelihood Poisson estimation, generalized estimation equations, random cluster effects Poisson models and random cluster effects Poisson models with separate within- and between-cluster effects. We demonstrate that the ratio of the number of exposed to unexposed subjects in each cluster plays a key role in determining the behavior of various analytic approaches. When this ratio is constant across clusters, simple theoretical expressions are available for the estimate of log risk ratio and its variance, and several common analytic methods give equivalent results. Simulations were used to examine the performance of the six methods when the exposure ratios vary across clusters. Our results indicate advantages for random effects models. Random effects Poisson models maintained a valid Type I error rate. A random cluster effects Poisson model that separates the between- and within-cluster exposure effects also allows testing of the assumption of equality of the between- and within-cluster exposure effects. These results will be useful to investigators and statisticians not only in analyzing studies with within-cluster covariates and count outcomes but also useful in designing such studies.
520 $a For the longitudinal clustered count data, we studied the case where the correlation for repeated measures is due to subject heterogeneity. We demonstrated that the Markov Chain Monte Carlo (MCMC) sampling approach performed well for three-level hierarchical Poisson models. The estimates are not biased and its type I error rates are close to 0.55. In contrast, quasi-likelihood (QL) implemented in the SAS GLIMMIX procedure did not do well. The estimates are more variable than those from MCMC. The SE( b&d4; ) of QL underestimated the true variability of b&d4; and the type I error of QL is too large for this method to be recommended.
590 $a School code: 0831.
650 4 $a Biology, Biostatistics. $3 1018416
690 $a 0308
710 2 $a University of Colorado Health Sciences Center. $3 1023815
773 0 $t Dissertation Abstracts International $g 68-11B.
790 1 0 $a Grunwald, Gary K., $e advisor
790 $a 0831
791 $a Ph.D.
792 $a 2007
856 4 0 $u http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=3289830