2
Associate Professor, Department of Statistics, Shiraz University
3
Professor, Department of Statistics, Shiraz University
Abstract
If the two-way ANOVA model assumes equal variances, the classical F-test is used to examine the effect of two factors on the response variable. However, in most practical problems, the assumption of equal variances does not hold. In recent years, various tests have been proposed to examine the effect of factors in the case of unequal variances. In this article, while introducing two tests, the approximate degrees of freedom and the parametric bootstrap, we evaluate the performance of these two tests in terms of power and type I error by studying simulation. The simulation results show that both tests perform well in controlling type I error and have a negligible difference in terms of test power. In terms of calculations, the parametric bootstrap method is based on simulation and is time-consuming; while the approximate degrees of freedom method is simpler in terms of practical use.
Boromandi, F., Kharati Kopaei, M. & Behboudian, J. (2016). Comparison of two tests: approximate degrees of freedom and parametric bootstrap in the heterogeneous unbalanced two-way ANOVA model. Andishe_ye Amari, 21(1), 49-55.
MLA
Boromandi, F., Kharati Kopaei, M., & Behboudian, J. "Comparison of two tests: approximate degrees of freedom and parametric bootstrap in the heterogeneous unbalanced two-way ANOVA model", Andishe_ye Amari, 21, 1, 2016, 49-55.
HARVARD
Boromandi F., Kharati Kopaei M., Behboudian J. (2016). 'Comparison of two tests: approximate degrees of freedom and parametric bootstrap in the heterogeneous unbalanced two-way ANOVA model', Andishe_ye Amari, 21(1), pp. 49-55.
CHICAGO
F. Boromandi, M. Kharati Kopaei & J. Behboudian, "Comparison of two tests: approximate degrees of freedom and parametric bootstrap in the heterogeneous unbalanced two-way ANOVA model," Andishe_ye Amari, 21 1 (2016): 49-55,
VANCOUVER
Boromandi F., Kharati Kopaei M., Behboudian J. Comparison of two tests: approximate degrees of freedom and parametric bootstrap in the heterogeneous unbalanced two-way ANOVA model. Andishe_ye Amari. 2016;21(1):49-55 (In Persian).
