Robust One-Way ANOVA Tests under Heteroscedasticity and Nonnormality

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) , parametric bootstrap test proposed by Krishnamoorthy et. al (2007) ; and generalized F test proposed by Weerahandi (1995) ;, respectively. These tests are based on the modified maximum likelihood (MML) estimators proposed by Tiku(1967, 1968) , .