Fundamentals of Predictive Analytics with JMP

by Ron Klimberg

On the other hand, if you reject H0, the variances can be assumed to be unequal, and the ANOVA output cannot be trusted.