1: With small samples, tests may not pick up violations. With huge samples, the tests may show a significant violation when in fact that actual size of the violation is very small. (E.g. the variances may be very nearly equal, but the test will be significant and equality rejected). This is a version of the high power problem. [return]