An Algorithm for Testing the Robustness Properties of Two-Way Nonorthogonal Analysis of Variance

Abstract
A simulation algorithm for examining the robustness properties of two-way nonorthogonal analysis of variance (ANOVA) is described. A nonorthogonal ANOVA occurs when unequal numbers of subjects appear in the cells of a factorial design. The algorithm can be used to study the effect of the violation of the assumptions of homogeneity of variance or normality on the Type I error rates. The algorithm offers a researcher the opportunity to provide additional evidence as to the validity of the results for an analysis based on real life data when robustness of the tests might be in doubt.