In summary, the new F -max x test has a better control of Type I error than the other tests. When the
homogeneity assumption is violated, the power of the new Fmax test is increased with the block size or the degree of departure from the equal variances assumption. In most of the simulated situations, the
new Fmax-test has a higher power than other tests. When variances from multiple groups are distinct
(as in the last scenario in each set), Yitnosumarto’s test seems to have equal or higher power.