Description
Robust One-Way ANOVA Tests under Heteroscedasticity and Nonnormality.
Description
Robust tests (RW, RPB and RGF) are provided for testing the equality of several long-tailed symmetric (LTS) means when the variances are unknown and arbitrary. RW, RPB and RGF tests are robust versions of Welch's F test proposed by Welch (1951) <doi:10.2307/2332579>, parametric bootstrap test proposed by Krishnamoorthy et. al (2007) <doi:10.1016/j.csda.2006.09.039>; and generalized F test proposed by Weerahandi (1995) <doi:10.2307/2532947>;, respectively. These tests are based on the modified maximum likelihood (MML) estimators proposed by Tiku(1967, 1968) <doi:10.2307/2333859>, <doi:10.1080/01621459.1968.11009228>.