We tested the homogeneity of variance (HOV) assumption of ANOVA by
employing Levene's test, widely considered to be the standard HOV test. Available
procedures such as Levene’s test allow HOV testing for simple one-way models only.
HOV testing for more complex models is a subject of current research. Therefore, we
separately tested HOV across forecasting methods at each safety stock level (Table 29)
and HOV across safety stock levels for each forecasting method (Table 30). The HOV
assumption held in 40 of 63 cases at an a level of 0.05 and in 50 of 63 cases at an a level
of 0.01. The ANOVA test is relatively robust when the groups being compared are all of
approximately the same size, unless the group variances are extremely different or the
number of groups is large. Since the groups in our study were relatively few in number
and of the same size, we believe that our use of the HOV assumption in the present study
was not problematic. As noted by Box (1953), “To make the preliminary test on
variances is rather like putting to sea in a rowing boat to find out whether conditions are
sufficiently calm for an ocean liner to leave port!”