Return to search

Estimation of survival of left truncated and right censored data under increasing hazard

When subjects are recruited through a cross-sectional survey they have already experienced the initiation of the event of interest, say the onset of a disease. This method of recruitment results in the fact that subjects with longer duration of the disease have a higher chance of being selected. It follows that censoring in such a case is not non-informative. The application of standard techniques for right-censored data thus introduces a bias to the analysis; this is referred to as length-bias. This paper examines the case where the subjects are assumed to enter the study at a uniform rate, allowing for the analysis in a more efficient unconditional manner. In particular, a new method for unconditional analysis is developed based on the framework of a conditional estimator. This new method is then applied to the several data sets and compared with the conditional technique of Tsai [23].

Identiferoai:union.ndltd.org:LACETR/oai:collectionscanada.gc.ca:QMM.100210
Date January 2007
CreatorsShinohara, Russell.
PublisherMcGill University
Source SetsLibrary and Archives Canada ETDs Repository / Centre d'archives des thèses électroniques de Bibliothèque et Archives Canada
LanguageEnglish
Detected LanguageEnglish
TypeElectronic Thesis or Dissertation
Formatapplication/pdf
CoverageMaster of Science (Department of Mathematics and Statistics.)
Rights© Russell Shinohara, 2007
Relationalephsysno: 002666607, proquestno: AAIMR38435, Theses scanned by UMI/ProQuest.

Page generated in 0.0019 seconds