JOURNAL OF AGRICULTURAL BIOLOGICAL AND ENVIRONMENTAL STATISTICS, cilt.4, sa.2, ss.166-184, 1999 (SCI-Expanded)
We consider methods for drawing inference to factor effects on response variability in factorial designs. Levene's test, the jackknife test, and Fligner and Killeen's test for heterogeneity of variance an generalized to factorial designs. Application of these analyses to the 2 by 2 factorial design is examined in detail. A simulation study examines and compares power and robustness properties of the tests. Based on these simulations, variants of both Fligner and Killeen's test and Levene's test are recommended. An example illustrates the usefulness of these tests in understanding data from a multifactor agricultural system that includes chile, nematodes, and yellow nutsedge.