Information Technology Reference
In-Depth Information
Figure 5.18
Options window in
One-Way ANOVA
.
Test of Homogeneity of Variances
The equal variances assumption is
violated; the probability (Sig.) is < .05.
GAFscore
Levene
Statistic
df1
df2
Sig.
5.108
2
12
.025
Figure 5.19
The Levene's test in
One-way ANOVA
.
analysis to be performed. The SPSS Explore procedure can also be used to
conduct the Levene and Brown-Forsythe tests, and will be demonstrated
in Chapter 6.
Figures 5.19 shows the results of the Levene test. The statistically signif-
icant outcome indicates that the assumption of homogeneity has not been
met. Instead of evaluating the mean differences with the usual Fisher
F
test,
we must instead use a test such as the Brown-Forsythe test, designed to
accommodate group differences in variance. As can be seen in Figure 5.20,
Robust Tests of Equality of Means
The means are significantly
different after correcting for the
unequal variances across the levels
of the independent variable.
GAFscore
Statistic
a
df1
df2
Sig.
Welch
Brown-Forsythe
24.104
2
5.519
.002
16.939
2
8.036
.001
a.
Asymptotically F distributed.
Figure 5.20
The Welch and the Brown-Forsythe tests.
