On computable estimates for accuracy of approximation for the Bartlett–Nanda–Pillai statistic

For the Bartlett–Nanda–Pillai statistic, we find computable estimates for accuracy of approximation, i.e., we describe explicitly the dependence on all parameters of the distributions that occur in the inequalities. For the other two classical statistics traditionally used in multivariate analysis o...

Full description

Saved in:
Bibliographic Details
Published inSiberian advances in mathematics Vol. 27; no. 3; pp. 153 - 159
Main Authors Lipatiev, A. A., Ulyanov, V. V.
Format Journal Article
LanguageEnglish
Published New York Allerton Press 01.07.2017
Springer Nature B.V
Subjects
Approximation
Estimates
Inequalities
Mathematical analysis
Mathematics
Mathematics and Statistics
Multivariate analysis
Variance analysis
accuracy of approximation
multivariate analysis of variance
Bartlett–Nanda–Pillai statistic
computable estimates
Online AccessGet full text

Cover

More Information
Summary:For the Bartlett–Nanda–Pillai statistic, we find computable estimates for accuracy of approximation, i.e., we describe explicitly the dependence on all parameters of the distributions that occur in the inequalities. For the other two classical statistics traditionally used in multivariate analysis of variance, i.e., the likelihood-ratio and Lawley–Hotelling statistics, similar computable estimates were found earlier.
ISSN:1055-1344
1934-8126
DOI:10.3103/S1055134417030014