Information Technology Reference
In-Depth Information
Linear Models
The GLM Procedure
Dependent Variable: satisfac satisfac
Sum of
Squares
Source
Model
Error
Corrected Total
DF
Mean Square
F Value
Pr
>
F
11
48
59
5360.000000
1688.000000
7048.000000
487.272727
35.166667
13.86
<.0001
These are the
probabilities
of the
F
ratio
occurring by
chance alone
for the main
effects and
interactions.
R-Square
Coeff Var
Root MSE
satisfac Mean
5.930149
26.00000
0.760499
22.80827
Source
kidshome
voter
politics
kidshome
∗
voter
kidshome
∗
politics
voter
∗
politics
kidsho
∗
voter
∗
politic
DF
Type III SS
Mean Square
F Value
Pr
>
F
326.666667
106.666667
836.250000
326.666667
530.416667
245.416667
687.916667
9.29
3.03
23.78
9.29
15.08
6.98
19.56
0.0037
0.0880
<
1
1
2
1
2
2
2
326.666667
106.666667
1672.500000
326.666667
1060.833333
490.833333
1375.833333
.0001
0.0037
<.0001
0.0022
<
.0001
Figure 9.25
ANOVA results.
The middle portion of Figure 9.25 presents
R-Square
; this is an alter-
native label for eta squared. However,
R-Square
is computed based on the
model with all seven effects combined (added) together. In the context
of ANOVA, we ordinarily wish to obtain the eta squared value for each
separate effect. To do this, we must perform the hand calculation, dividing
each sum of squares by the total sum of squares (
Corrected Total
)shown
in Figure 9.25. The coefficient of variation (the ratio of the standard devi-
ation of the sample as a whole to the mean of the sample as a whole), the
root mean square error (the square root of the mean square associated
with the error term), and the grand mean of the dependent variable are
also displayed in that middle table.
9.12 PERFORMING THE POST-ANOVA ANALYSES IN SAS
The omnibus analysis indicated that all but one of the effects were sta-
tistically significant. The significant three-way interaction supersedes all
of the two-way interactions and so we will not follow up on these. Fur-
thermore, we ordinarily would not bother with the main effects either
because we have obtained a statistically significant three-way interaction
in the omnibus analysis. However, to illustrate how we would approach
the statistically significant main effects, we will discuss them here.
