Show simple item record

dc.contributor.advisorMaWhinney, Samantha
dc.contributor.authorWilson, Melissa Pierce
dc.contributor.committeememberMoore, Camille
dc.contributor.committeememberErlandson, Kristine M.
dc.date.accessioned2020-01-14T15:44:30Z
dc.date.available2020-07-06T15:44:31Z
dc.date.submitted2019
dc.descriptionIncludes bibliographical references.
dc.descriptionFall
dc.description.abstractWith the increasing availability of longitudinal cohort study data and associated biospec- imen libraries, retrospective testing of stored samples to address hypotheses beyond the original aims of the cohort study has become more common. Outcome-dependent sampling (ODS) methods have been developed as an alternative to random sampling, to identify more powerful subsets of cohort participants when testing all available samples is not feasible. In the setting that cost of desired testing is prohibitive or preservation of available samples is desired, available data can inform the choice of a subset of participants whose samples will be most informative. Existing methods often involve population average (PA) models with transitional terms and/or random effects. These models may perform well in the setting of data that is missing completely at random or missing at random, but little has been published on performance within the ODS framework in the setting of data missing not at random. As non-ignorable dropout can be an issue for longitudinal data collection and subsequent analysis, the assessment of bias in these settings is important. The goal of this thesis is to explore proposed models for binary outcomes and to evaluate these methods in the setting of non-ignorable dropout.
dc.identifierWilson_ucdenveramc_1639M_10692.pdf
dc.identifier.urihttps://hdl.handle.net/10968/4775
dc.languageEnglish
dc.publisherUniversity of Colorado at Denver, Anschutz Medical Campus. Health Sciences Library
dc.rightsCopyright of the original work is retained by the author.
dc.rights.accessEmbargo Expires: 07/06/2020
dc.subjectMissing data
dc.subjectOutcome-dependent sampling
dc.subjectBiobanks
dc.subject.meshCohort Studies
dc.titleEffects of missing data on outcome-dependent sampling methods for longitudinal cohorts with binary data
dc.typeThesis
dcterms.embargo.expires2020-07-06
thesis.degree.disciplineBiostatistics
thesis.degree.grantorUniversity of Colorado at Denver, Anschutz Medical Campus
thesis.degree.levelMasters
thesis.degree.nameMaster of Science (M.S.)


Files in this item

Thumbnail

This item appears in the following Collection(s)

Show simple item record