Information Technology Reference
In-Depth Information
Repeat this procedure with
powcolor
and
boxcolor
and
granules
and
box-
color
. Finally, while holding the Control key down, select
powcolor
,
gran-
ules
, and
boxcolor
and then click the
Factorial
pushbutton;
powcolor
∗
granules
boxcolor
will automatically appear in the
Effects
window.
Select
Fixed Effect Model Options
in the navigation panel as shown
in Figure 12.24. Check
Type 3
under
Hypothesis test type
. Then
select
Residual maximum likelihood
under
Estimation method
,and
Between and within subjects portions
under
Degrees of freedom
method
.
Select Random Effects
in the navigation panel as shown in Fig-
ure 12.25. When you select
Random effects
under the
Effects to use
menu, a box will appear at the far right of that menu. Click on that box
and the
Effects Builder - Random effects
window appears. Select
subid
and click the
Main
pushbutton;
subid
will automatically appear in the
Random effects
panel. Click the
OK
pushbutton.
Now select
Subject identifier
under the
Model subjects
menu; a
box will appear at the far right of that menu. Click on that box and
the
Effects Builder - Subject identifier
window appears. Select
subid
and click the
Main
pushbutton;
subid
will automatically appear in the
Subject identifier
panel. Select the
OK
pushbutton. This is shown in
Figure 12.26.
Select
Least Squares Post Hoc Tests
in the navigation panel. Click the
Add
pushbutton at the bottom of the
Effects to estimate
pane. The screen
shown in Figure 12.27 appears. Highlight each of the seven effects in turn
under the
Effects to use
menu and select
True
for each; this command
will cause the means for each of the effects to be output. Click the
Run
pushbutton.
∗
12.11 SAS OUTPUT FROM THE OMNIBUS ANALYSIS
The least squares means are shown in Figure 12.28. The results of the
omnibus analysis are shown in Figure 12.29. These
F
ratios are numerically
somewhat different from those produced by SPSS, although they lead us
to draw the same conclusions. As can be seen, all effects except the main
effect of
boxcolor
are statistically significant.
12.12 PERFORMING THE SIMPLE EFFECTS ANALYSIS IN SAS
To perform the simple effects analysis in
SAS Enterprise Guide
, configure
the analysis as shown above. Then in the
Least Squares Post Hoc Tests
screen shown in Figure 12.30, in the
Effects to use
frame, set the three-
way interaction to
True
and all others to
False
. Then in the
Comparisons
frame, set
Show p-values for differences
to
All pairwise differences
and
set
Adjustment method for comparison
to
Bonferroni
.Click
Run
to
perform the analysis.