Analysis of power functions of multiple comparisons tests
Analysis of power functions of multiple comparisons tests
The power functions of the Studentized range test and the F-test for comparing the means of normal populations in the one-way fixed effects analysis of variance model is investigated. The least favourable configurations of population means under certain range restrictions of the population means are given. These results are important in the calculation of the sample sizes required to guarantee power level under the restrictions on the range of the population means. The optimal test procedures which maximise the least favourable power under the range restrictions of the population means are found for the known variance case. Then the relations between these optimal tests and the range type tests, sum of square tests are pointed out. The problem of comparing several treatment means with a control mean is also considered. Finally, these results are applied to the binary data, by using the arcsin-root transformation and the asymptotic distribution, for both comparison between k Bernoulli responses and comparison between k Bernoulli responses and a control Bernoulli response.
Liu, Wei
b64150aa-d935-4209-804d-24c1b97e024a
July 1990
Liu, Wei
b64150aa-d935-4209-804d-24c1b97e024a
Hayter, Anthony J.
0ac7de8f-0995-4766-afc0-c0c95313e94d
Liu, Wei
(1990)
Analysis of power functions of multiple comparisons tests.
University of Bath, Doctoral Thesis, 128pp.
Record type:
Thesis
(Doctoral)
Abstract
The power functions of the Studentized range test and the F-test for comparing the means of normal populations in the one-way fixed effects analysis of variance model is investigated. The least favourable configurations of population means under certain range restrictions of the population means are given. These results are important in the calculation of the sample sizes required to guarantee power level under the restrictions on the range of the population means. The optimal test procedures which maximise the least favourable power under the range restrictions of the population means are found for the known variance case. Then the relations between these optimal tests and the range type tests, sum of square tests are pointed out. The problem of comparing several treatment means with a control mean is also considered. Finally, these results are applied to the binary data, by using the arcsin-root transformation and the asymptotic distribution, for both comparison between k Bernoulli responses and comparison between k Bernoulli responses and a control Bernoulli response.
This record has no associated files available for download.
More information
Published date: July 1990
Identifiers
Local EPrints ID: 457940
URI: http://eprints.soton.ac.uk/id/eprint/457940
PURE UUID: 9052f6ca-4d96-4231-83e3-e4d042cc6bc6
Catalogue record
Date deposited: 23 Jun 2022 16:49
Last modified: 20 Feb 2024 02:33
Export record
Contributors
Thesis advisor:
Anthony J. Hayter
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