Factor effects on the variance in a replicated two-way treatment structurein an agricultural system

We consider methods for drawing inference to factor effects on response var iability in factorial designs. Levene's test, the jackknife test, and Flign er and Killeen's test for heterogeneity of variance an generalized to facto rial designs. Application of these analyses to the 2 by 2 factorial design is examined in detail. A simulation study examines and compares power and r obustness properties of the tests. Based on these simulations, variants of both Fligner and Killeen's test and Levene's test are recommended. An examp le illustrates the usefulness of these tests in understanding data from a m ultifactor agricultural system that includes chile, nematodes, and yellow n utsedge.