owever, it is difficult to find the exact distribution of the
likelihood ratio statistic,Λ. For large samples the statistic
2ln − Λ is approximately distributed as chi-squared with
k-1 degrees of freedom, where k is a number of the
population. For small samples the approximated chi-square
may be inaccurate. Alternatively, the Kruskal-Wallis nonparametric
test can be used. ANOVA can be applied if and
only if the nonnormal data are transformed to fit the
required assumption of it. In this paper, the power of the
ANOVA and the Kruskal-Wallis test for comparing the
several Weibull population means is investigated.