ABSTRACT
As noted in the previous two sections, when a one-way ANOV A is statistically
significant, a researcher cannot be sure which specific differences are significant.
Thus, for the example in Section 48, there are three differences between pairs of means:
1. Group A versus Group B (6.000-5.500 = 0.500)
2. Group A versus Group C (6.000-2.333 = 3.667)
3. Group B versus Group C (5.500-2.333 = 3.167)
A number of tests have been suggested for determining the significance of the
differences between pairs of means after obtaining an overall significant result us-
ing ANOVA.1 In Section 49, Tukey's HSD test was illustrated. An alternative test,
known as Scheffi's test, is considered in this section. Scheffe' s test is more con-
servative than Tukey's. That is, Scheffe's test is less likely to lead to rejection of
the null hypothesis than is Tukey's.2