The University of Southampton
University of Southampton Institutional Repository

Optimal designs in the presence of missing responses

Optimal designs in the presence of missing responses
Optimal designs in the presence of missing responses
Design of experiments is an approach that could minimise the costs of conducting experiments by maximising the information that could be obtained from the study prior to the implementation. It is crucial that the experimenters consider the impact of the presence of missing observations on the statistical power of the study and the precision of the inferences. This research incorporates the features of some missing data analysis approaches into the experimental design framework for finding a design that is robust to the presence of missing responses.

We propose optimal design framework for the linear regression models and the linear mixed models respectively. Assuming that missing responses are generated by a monotone missing at random (MAR) mechanism, we consider the features of complete case analysis and a multiple imputation respectively in the design framework for the linear regression models, and of available case analysis in the cohort design framework for the linear mixed models.

The optimal design framework for the linear regression models with complete case analysis is a generalisation to the work that is proposed by Imhof et al. (2002). Besides that, we believe we are the first who consider a multiple imputation approach at the design stage of an experiment. Moreover, having accounted for the presence of dropouts, we introduce two types of design regime in the cohort design framework for the linear mixed models.

Throughout this project we show that using the optimal designs that assume completely observed responses or the naive designs may not be the best option. There are statistical gains in accounting for the features of missing data analysis approaches at the design stage of an experiment, especially for the study that involves a small sample size and high costs. This novel research provides a new tool to tackle the presence of missing data in future experimental studies.
Lee, Kim May
8111b847-1f07-460f-88d7-e5a24d798e3a
Lee, Kim May
8111b847-1f07-460f-88d7-e5a24d798e3a
Biedermann, Stefanie
fe3027d2-13c3-4d9a-bfef-bcc7c6415039
Mitra, Robin
2b944cd7-5be8-4dd1-ab44-f8ada9a33405

(2016) Optimal designs in the presence of missing responses. University of Southampton, School of Mathematics, Doctoral Thesis, 168pp.

Record type: Thesis (Doctoral)

Abstract

Design of experiments is an approach that could minimise the costs of conducting experiments by maximising the information that could be obtained from the study prior to the implementation. It is crucial that the experimenters consider the impact of the presence of missing observations on the statistical power of the study and the precision of the inferences. This research incorporates the features of some missing data analysis approaches into the experimental design framework for finding a design that is robust to the presence of missing responses.

We propose optimal design framework for the linear regression models and the linear mixed models respectively. Assuming that missing responses are generated by a monotone missing at random (MAR) mechanism, we consider the features of complete case analysis and a multiple imputation respectively in the design framework for the linear regression models, and of available case analysis in the cohort design framework for the linear mixed models.

The optimal design framework for the linear regression models with complete case analysis is a generalisation to the work that is proposed by Imhof et al. (2002). Besides that, we believe we are the first who consider a multiple imputation approach at the design stage of an experiment. Moreover, having accounted for the presence of dropouts, we introduce two types of design regime in the cohort design framework for the linear mixed models.

Throughout this project we show that using the optimal designs that assume completely observed responses or the naive designs may not be the best option. There are statistical gains in accounting for the features of missing data analysis approaches at the design stage of an experiment, especially for the study that involves a small sample size and high costs. This novel research provides a new tool to tackle the presence of missing data in future experimental studies.

PDF
kim_thesis_r2.pdf - Other
Download (2MB)

More information

Published date: May 2016
Organisations: University of Southampton, Mathematical Sciences

Identifiers

Local EPrints ID: 397648
URI: http://eprints.soton.ac.uk/id/eprint/397648
PURE UUID: 8f888647-e279-469b-af63-9b0a162f7a91
ORCID for Stefanie Biedermann: ORCID iD orcid.org/0000-0001-8900-8268

Catalogue record

Date deposited: 06 Jul 2016 13:42
Last modified: 27 Jul 2019 00:34

Export record

Contributors

Author: Kim May Lee
Thesis advisor: Stefanie Biedermann ORCID iD
Thesis advisor: Robin Mitra

University divisions

Download statistics

Downloads from ePrints over the past year. Other digital versions may also be available to download e.g. from the publisher's website.

View more statistics

Atom RSS 1.0 RSS 2.0

Contact ePrints Soton: eprints@soton.ac.uk

ePrints Soton supports OAI 2.0 with a base URL of http://eprints.soton.ac.uk/cgi/oai2

This repository has been built using EPrints software, developed at the University of Southampton, but available to everyone to use.

We use cookies to ensure that we give you the best experience on our website. If you continue without changing your settings, we will assume that you are happy to receive cookies on the University of Southampton website.

×