Key points are not available for this paper at this time.
Several power functions were computed to determine the influence of violations of assumptions on the analysis of variance. For each analysis two groups of 10 scores each were randomly selected from a normally distributed population of 10,000 cases. In all cases each of the several points on the power curves was determined by the percentage of rejected null hypotheses in 5,000 analyses with alpha set at .05. The CDC 1604 computer was used.
Young et al. (Mon,) studied this question.