SYNOPTIC ABSTRACT
Following a brief survey of existing tests for the homogeneity of I population variances, two new tests are proposed. One test uses subsampling; the other test uses jackknifing. Both tests employ the analysis of means (ANOM) and can be presented in a graphical form that is suitable for easy assessment of statistical as well as practical significance. Both tests are robust to nonnormality and can be used when fat-tailed distributions are suspected. Monte Carlo results are used to show that both ANOM type tests have power comparable to their analysis of variance (ANOVA) type counterparts.
Key Words and Phrases: