Performance of tests for homogeneity of variances in completely randomized designs

Authors

Keywords:

Heteroscedasticit, Type I error, Power

Abstract

By 1920, Fisher proposed the analysis of variance, which aims to decompose the total
variation in sources of variation known. For validity of the results of the analysis of variance,
this depends on some conditions are met presupposed. One reason to ignore checking of assump-
tions is the difficulty of finding adequate tests for such purpose. The assumption of homogeneity
of variances is the most important assumption of the analysis of variance. Violation of any
other assumption may result in heterogeneity of experimental error, and this further reinforces
the need for their study. Thus, the objectives of this research were to study the performance
and implement the control of type I error and power of 15 tests for homogenenidade variances
using Monte Carlo simulation, in varied settings of treatments and replicates. In normal and
randomized design proposals based on the likelihood showed the best results followed the proposal
presented by Bayesian Samiuddin. Variations Levene’s test had modest results in situations of
low reps which also happened to Cochran.

Author Biography

Denismar Alves Nogueira, Universidade Federal de Alfenas

Instituto de Ciências Exatas, unidade de Estatística

References

ARGAÇ, D. Testing for homogeneity in a general one-way classication with fixed effects: power simulations and comparative study. Computational Statistics and Data Analysis, 2002.

BARTLETT, M. S. Properties of sufficiency and statistical tests. Proceedings of the Royal Statistical Society - Serie A, v,60, p.268-282, 1937.

BOOS, D. D.; BROWNIE, C. Bootstrap methods for testing homogeneity of variances. Technometrics, v.31, n.1, p.69-82, 1989.

BOX, G. E. P.; ANDERSEN, S. L. Permutation theory in the derivation of robust criteria and the study of departures from assumptions. Journal of the Royal Statistical Society, Series B, v.17, n.1, p.1-26, 1955.

BOX, G. E. P. Some theorems on quadratic forms applied in the study of analysis of variance problems, I. Effect of inequality of variance in the one-way classification. Ann. Math. Statist., v.25, p.290-302, 1954.

BROWN, M. B.; FORSYTHE, A. B. Robust tests for equality of variances. Journal of the American Statistical Association, v.69, n.346, p.364-367, 1974.

BROWN, M. B.; FORSYTHE, A. B. The use of weighted contrasts in analysis of models with heterogeneity of variance. Proceedings of the Business and Economics Statistics Section American Statistical Association, p.347-352, 1983.

DIXON, W. J.; MASSEY, F. J. Introduction to statistical analysis. McGraw-Hill Book, New York, n.3, p.308-309 , 1969.

FERREIRA, D. F. Estatística básica, Editora UFLA, Lavras - MG, p.664, 2005.

FERREIRA, D. F.; DEMÉTRIO, C. G. B.; MANLY, B. F. J.; MACHADO, A. DE A.; VENCOVSKY, R. Statistical models in agriculture: biometrical methods for evaluating phenotypic stability in plant breeding. Cerne, v.12, n.4, p.373-388, 2006.

GOMEZ, K. A.; GOMEZ, A. A. Statistical procedures for agricultural research. John Wiley, n.2, p.680, 1984.

HARTUNG, J.; ARGAÇ, D.; MAKAMBI, K. H., Small sample properties of tests on homogeneity in one-way ANOVA and meta-analysis. Statist. Papers , n.43, p.197-235, 2002.

JAMES, G. S. The comparison of several groups of observations when the ratios of population variances are unknown. Biometrika, n.38, p.324-329, 1951.

JOHNSON, R. A.; WICHERN, D. W. Applied multivariate statistical

analysis. Prentice Hall: New Jersey, p.816, 1998.

KEYES, T. K.; LEVY, M. S. Analysis of levene's test under design imbalance. Journal of Educational and Behavioral Statistics, v.22, p.227-236, 1997.

LAYARD, M. N. J. Robust large-sample tests for homogeneity of variances. Journal of the American Statistical Association, v.68, n.341, p.195-198, 1973.

LEEMIS, L.; TRIVEDI, K. S. A comparison of approximate interval estimators of the Bernoulli parameter. The American Statistician, v.50, p.63-68. 1996.

LEVENE, H. Robust tests for equality of variances. Contribution to Probability and Statistics. Stanford, CA: Stanford University Press, p.278-292, 1960.

MANLY, B. F. J. Randomization, bootstrap and Monte Carlo methods in biology. University of Otago, New Zealand, p. 356, 1997.

MEHROTRA, D. V. Improving the Brown Forsythe solution to the generalized Behrens Fisher problem. Comm. Statist. Simulation Comput., v.26, p.1139 1145, 1997.

MILLER, R. G., Jr. Jackknifing variances. Annals of Mathematical Statistics, v.39, n.2, p.567-582, 1968.

O'BRIEN, R. G. A robust technique for testing heterogeneity of variance effects in factorial design. Psychometrika, v.43, n.3, p.327-342, 1978.

O'NEILL, M. E.; MATHEWS, K. L. A weighted least squares approach to levene's test of homogeneity of variance. Australian e New Zealand Journal Statistical, v.42, n.1, p.81-100, 2000.

O'NEILL, M. E.; MATHEWS, K. L. Levene tests of homogeneity of variance for general block and treatment designs. Biometrics, v.51, p.216-224, 2002.

R DEVELOPMENT CORE TEAM. R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. 2012. ISBN 3-900051-07-0, Disponível: http://www.R-project.org/.

RUBIN A. S. Experimentação em genética. Editora UFLA, Lavras - MG, n.2, p.303, 2005.

SAMIUDDIN, M. Bayesian test of homogeneity of variance. Journal of the American Statistical Association, v.71, n.354, p.515-517, 1976.

WELCH, B. L. On the comparison of several mean values: an alternative approach. Biometrika, n.38, p.330-336, 1951.

WILSON, E. B.; HILFERTY, M. M. The distribution of chi-square. Proceedingdof the National Academy of Science, v.17, p.684-688, 1931.

Published

31-12-2013

How to Cite

Nogueira, D. A., & Pereira, G. M. (2013). Performance of tests for homogeneity of variances in completely randomized designs. Sigmae, 2(1), 7–22. Retrieved from https://publicacoes.unifal-mg.edu.br/revistas/index.php/sigmae/article/view/141

Issue

Section

Probability and Statistics