Modified tests for comparison of group means under heteroskedasticity and non-normality caused by outlier(s)

Year 2017, Volume: 46 Issue: 3, 493 - 510, 01.06.2017

Mustafa Cavus , Berna Yazici , Ahmet Sezer

Abstract

There are several approximate tests proposed such as Welch's F-test (W), the Parametric Bootstrap Test (PB) and Generalized F-test (GF) for comparing several group means under heteroskedasticity. These tests are powerful and have nominal type 1 error rates but they are not performing satisfactorily under non-normality caused by outlier(s). To handle this problem, we investigate tests that are powerful and provide nominal type 1 error rates by using robust estimators both for location and scale parameters. The performance of the modified tests are examined with Monte-Carlo simulation studies. Results of simulations clearly indicate that Generalized F-test modied with Huber's M-estimators achieves the nominal type 1 error rate and provide higher power than alternative methods.

Keywords

Outlier, Heteroskedasticity, Unbalanced, Huber's M-Estimators, Generalized F-test

References

• Barnett, V. and Lewis, T. Outlier in Statistical Data, Wiley and Sons, 1994.
• Ben, M.G. and Yohai, V.J. Robust analysis of variance for randomized block design, Com- munications in Statistics 216, 1779-1798, 1992.
• Braun, H.I. and McNeil, D.R. Testing in robust ANOVA, Communications in Statistics 102, 149-165, 1981.
• Chang, C.H., Pal, N., Lim, W.K., Lin, J.J. Comparing several population means: a para- metric bootstrap method and its comparison with usual ANOVA F-test as well as ANOM, Computational Statistics 25, 71-95, 2010.
• Chang, C.H., Lin, J.J., Pal, N. Testing the equality of several gamma means: a parametric bootstrap method with applications, Computational Statistics 76, 26-55, 2011.
• Cavus, M., Yazici, B., Sezer, A. Comparison of means under non-normality with outliers for unbalanced data, 8th International Conference of the ERCIM Working Group on Computing and Statistics Proceedings, London, ENGLAND, 2015.
• Cavus, M., Yazici, B., Sezer, A. Robust methods to compare means under violation of nor- mality with outliers, 12th German Probability and Statistics Days Proceedings, Bochum, GERMANY, 2016.
• Cavus, M., Yazici, B., Sezer, A. Modified tests for comparison of group means in the presence of outlier(s), 2th International Researchers-Statisticians and Young Statisticians Congress Proceedings, Ankara, TURKEY, 2016.
• Fan, W. and Hancock, G. Robust mean modelling: an alternative for hypothesis testing of independent means under variance heterogeneity and nonnormality, Journal of Educational and Behavioral Statistics 37, 137-156, 2012.
• Fisher, R.A. Statistical methods and scientific inference, Oliver and Boyd, 1925.
• Gamage, J. and Weerahandi, S. Size performance of some tests in one-way ANOVA, Com- munications in Statistics 273, 625-640, 1998.
• Gokpinar Yigit, E., Polat, E., Gokpinar, F., Gunay, S. A new computational approach for testing equality of inverse Gaussian means under heterogeneity, Hacettepe Journal of Mathematics and Statistics 42 5, 581-590, 2013.
• Huber, P.J. Robust estimation of a location parameter, The Annals of Mathematical Statis- tics 69, 73-101, 1964.
• Huber, P.J. and Ronchetti, E.M. Robust Statistics, John Wiley and Sons, New Jersey, 2009.
• Jian-Hong, S. and Jiang-Long, L. A new generalized p-value for testing equality of inverse Gaussian means under heterogeneity, Statistics and Probability Letters 82, 96-102, 2012.
• Karagoz, D. Modied Welch test statistic for ANOVA under Weibull distribution, Hacettepe Journal of Mathematics and Statistics 45 56, 2015.
• Karagoz, D. and Saracbasi, T. Robust Brown-Forsythe and robust modified Brown-Forsythe ANOVA tests under heteroscedasticity for contaminated Weibull distribution, Revista Colombiana de Estadistica 39, 17-32, 2009.
• Krishnamoorthy, K., Lu F., Mathew, T. A parametric bootstrap approach for ANOVA with unequal variances: Fixed and random models, Computational Statistics and Data Analysis 51, 5731-5742, 2007.
• Lee, H. and Fung K.Y. Robust procedures for multi-sample location problems with unequal group variances, Journal of Statistical Computation and Simulation 18, 125-143, 1983.
• Lee, S. and Ahn, C.H. Modified ANOVA for unequal variances, Communications in Statistics 32 4, 987-1004, 2003.
• Li, X. A generalized p-value approach for comparing the means of several log-normal populations, Statistics and Probability Letters 79, 1404-1408, 2009.
• Ozkip, E., Yazici, B., Sezer, A. Comparing the means for unbalanced several skewed data with unequal variances, 5th International Conference of the ERCIM Working Group on Computing and Statistics Proceedings , 2012a.
• Ozkip, E., Yazici, B., Sezer, A. Comparison of population means for skewed data with unequal variances, Joint Statistical Meetings 2012 Proceedings , 2524-2529, 2012b.
• Ozkip, E., Yazici, B., Sezer, A. Inference on the mean of unbalanced several lognormal population using generalized p-value and generalized condence intervals, 8th International Symposium of Statistics Proceedings , 2012c.
• Ozkip, E., Yazici, B., Sezer, A. Performance of some tests compares for equality of means in skewed distribution data, Joint Statistical Meetings Proceedings, Montreal, CANADA, 2013.
• Ozkip, E., Yazici, B., Sezer, A. A simulation study on tests for the Behrens-Fisher problem, Turkiye Klinikleri Biyoistatistik Dergisi,6 2, 59-66, 2014.
• Ozkip, E., Yazici, B., Sezer, A. Comparison of tests for the ANOVA with unequal variance, Joint Statistical Meetings Proceedings, Boston, USA, 2014.
• Reed, J.F. and Stark, D.B. Robust analysis of variance: a simulation study, Journal of Applied Statistics 22 1, 87-104, 1995.
• Rousseeuw, P. and Croux, C. Alternatives to the median absolute deviation, Journal of American Statistical Association 88 424, 1273-1283, 1993.
• Sadooghi-Alvandi, S.M., Jafari, A.A., Mardani-Ford, H.A. One-way ANOVA with unequal variances, Communications in Statistics 41, 4200-4221, 2012.
• Schrader, R.M. Robust analysis of variance, Communications in Statistics 6 9, 879-894, 1977.
• So, Y.C. and Sen, P.K. M-estimators based repeated signicance tests for one-way ANOVA with adaptation to multiple comparisons, Communications in Statistics 1 2, 101-119, 1982.
• Tan, W.Y. and Tabatabai, M.A. Some robust ANOVA procedures under heteroskedasticity and non-normality, Communications in Statistics 14 4, 1007-1026, 1985.
• Tiku, M. and Akkaya, A. Robust Estimation and Hypothesis Testing, New Age International, 2004.
• Tsui, K. and Weerahandi, S. Generalized p-values in signicance testing of hypotheses in the presence of nuisance parameters, Journal of the American Statistical Association 84, 602-607, 1989.
• Weerahandi, S. Testing variance components in mixed models with generalized p-values, Journal of the American Statistical Association 86, 151-153, 1991.
• Weerahandi, S. ANOVA under unequal error variances, Biometrics 51, 589-599, 1994.
• Weerahandi, S. Exact Statistical Methods for Data Analysis, Springer, 1995.
• Weerahandi, S. Generalized Inference in Repeated Measures: Exact Methods in MANOVA and Mixed Models, Wiley, 2004.
• Welch, B.L. On the comparison of several group means values, Biometrika 38, 330-336, 1951.
• Wilcox, R.R. Simulation results on solutions to the multivariate Behrens-Fisher problem via trimmed means, The Statistician 44 2, 213-225, 1995.
• Wilcox, R.R. Introduction to Robust Estimation and Hypothesis Testing, Elsevier, 2012.
• Wu, L.L. Robust M-estimation of location and regression, Sociological Methodology 15, 316-388, 1985.
• Yazici, B., Sezer, A., Ozkip, E. A simulation study to compare several population means under heteroskedasticity, Abstract Book of 5th International Conference of the ERCIM Working Group on Computing and Statistics, Oviedo, SPAIN, 2012.
• Yazici, B., Sezer, A., Ozkip, E. Generalized p-value test to compare several population means for unequal variances, Joint Statistical Meetings Proceedings, San Diego, USA, 2012.