Time use : a case for compositional data analysis /
Files
(Published version)
Date
2018
Authors
Dumuid, Dorothea,
Editors
Advisors
Journal Title
Journal ISSN
Volume Title
Type:
thesis
Citation
Statement of Responsibility
Conference Name
Abstract
The first chapter of this program of research explains how the issue of perfect multi-collinearity can be overcome using a specialised branch of statistics called compositional data analysis. This approach conceptualises daily time use as a composition made up of mutually exclusive and exhaustive time-use domains, rather than absolute durations of time spent in individual time-use domains. The composition is expressed in relative terms, as a set of log ratios, to enable its inclusion in traditional statistical models. The subsequent chapters describe how compositional data analysis can be applied to answer common research questions within time-use epidemiology, and provide examples from a variety of datasets. The thesis is presented in a hybrid format – the chapters contain traditional text to describe the methods, and publications to provide an example of their application.
School/Discipline
University of South Australia. School of Health Sciences.
School of Health Sciences
School of Health Sciences
Dissertation Note
Thesis (PhD(Physiotherapy))--University of South Australia, 2018.
Provenance
Copyright 2018 Dorothea Dumuid.
Description
1 ethesis (xxvi, 350 pages) :
illustrations (some colour)
Includes bibliographical references.
illustrations (some colour)
Includes bibliographical references.
Access Status
506 0#$fstar $2Unrestricted online access