Row-column and cross-over designs for comparing dual with single treatments
Row-column and cross-over designs for comparing dual with single treatments
The study concerns the design of experiments whose purpose is to compare the joint effects of two drugs A and B, at n and m levels respectively, with the effect of each individual drug. The treatment containing both drugs at the zero level is excluded from the experiment on ethical grounds. The aim of this investigation is to identify efficient row-column and cross-over designs for estimating the dual versus single treatment contrasts, in the sense of having small total variance for the contrast estimators.
Initially, attention is restricted to row-column designs, that is designs for two orthogonal blocking factors. The criterion used for design selection is the A-criterion. Lower bounds on the total variance of the estimators of the contrasts of interest are developed and used to assess the performance of row-column designs obtained by amalgamating two single blocking factor designs which are part-balanced for the dual versus single contrasts. A subclass of part-balanced row-column designs, with the property that treatments are orthogonal to row blocks, is identified. A method of finding efficient designs for estimating any specific given set of treatment contrasts is also described.
In the latter part of this thesis, designs are identified for cross-over experiments where each subject is given a sequence of two or more distinct treatments. For such studies, it is necessary to consider the possibility that the effect of a particular treatment may persist beyond the period of application. Designs are considered for the cases where treatment effects persist for one and for two periods and simple additive models including first- and second-order carryover effects are assumed.(DX178951)
University of Southampton
Yates, Paula Kim
66ad7ade-7bc3-44c5-b012-b2a9fa73216a
1993
Yates, Paula Kim
66ad7ade-7bc3-44c5-b012-b2a9fa73216a
Yates, Paula Kim
(1993)
Row-column and cross-over designs for comparing dual with single treatments.
University of Southampton, Doctoral Thesis.
Record type:
Thesis
(Doctoral)
Abstract
The study concerns the design of experiments whose purpose is to compare the joint effects of two drugs A and B, at n and m levels respectively, with the effect of each individual drug. The treatment containing both drugs at the zero level is excluded from the experiment on ethical grounds. The aim of this investigation is to identify efficient row-column and cross-over designs for estimating the dual versus single treatment contrasts, in the sense of having small total variance for the contrast estimators.
Initially, attention is restricted to row-column designs, that is designs for two orthogonal blocking factors. The criterion used for design selection is the A-criterion. Lower bounds on the total variance of the estimators of the contrasts of interest are developed and used to assess the performance of row-column designs obtained by amalgamating two single blocking factor designs which are part-balanced for the dual versus single contrasts. A subclass of part-balanced row-column designs, with the property that treatments are orthogonal to row blocks, is identified. A method of finding efficient designs for estimating any specific given set of treatment contrasts is also described.
In the latter part of this thesis, designs are identified for cross-over experiments where each subject is given a sequence of two or more distinct treatments. For such studies, it is necessary to consider the possibility that the effect of a particular treatment may persist beyond the period of application. Designs are considered for the cases where treatment effects persist for one and for two periods and simple additive models including first- and second-order carryover effects are assumed.(DX178951)
Text
410682.pdf
- Version of Record
More information
Published date: 1993
Identifiers
Local EPrints ID: 462376
URI: http://eprints.soton.ac.uk/id/eprint/462376
PURE UUID: af91f808-2bd3-4288-8640-0a5da471aa68
Catalogue record
Date deposited: 04 Jul 2022 19:06
Last modified: 16 Mar 2024 18:55
Export record
Contributors
Author:
Paula Kim Yates
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