Information Technology Reference
In-Depth Information
null hypothesis is true. As you can see, all of the groups are significantly
different from the reference group.
7.18 PERFORMING SIMPLE CONTRASTS IN SAS
Unlike SPSS,
SAS Enterprise Guide
does not have a large list of preset
contrasts, but it does have a
Dunnett
contrast on its menu system. We
perform this contrast in the same way that we did to perform a Tukey post
hoc test except that we will make somewhat different selections on the
Least Squares
tab. Thus, from the main menu select
Analyze
➜
ANOVA
➜
Linear Models.
On the
Task Roles
tab, specify
satscore
as the
Dependent
variable
and
group
as the
Classification variable.
On the
Model
tab,
select
group
and click the
Main
bar in the middle of the window to place
group
in the
Effects
panel. In the
Model Options
window click only the
Type III sum of squares to show.
Now select
Least Squares
on the
Post Hoc Tests
tab and click the
Add
pushbutton to display the
Options for means tests
. Here are the settings
for the drop-down menus.
On the
Class effects to use
drop-down menu now showing
False,
click
False
and the down arrow and select
True .
This will focus on
the differences of the means for the
group
variable.
On the
Comparisons
drop-down menu for
Show p-values for dif-
ferences
now showing
None,
click in the panel, then click the down
arrow, and select
Control using first level.
On the
Comparisons
drop-down menu for
Adjustment method
for comparisons
now showing
Default,
click in the panel, then click
the down arrow, and select
Dunnett
.
On the
Show confidence limits
drop-down menu now showing
False
,click
False
and the down arrow and select
True
.Thiswill
expose a second drop-down menu for you to specify the
Confidence
level (%)
. Retain the default of
95
.
Ignore the
Covariates
drop-down menu because there is no covariate
in the analysis.
When you have completed all of these specifications, your screen should
be configured as shown in Figure 7.18. Click
Run
to perform the analysis.
7.19 THE SIMPLE CONTRASTS OUTPUT FROM SAS
The output for the Dunnett comparisons is shown in Figure 7.19. With
the first group defined as the reference group, each of the other groups is
compared to that group. Each probability in the last column is based on
comparing that particular group to the reference group. Thus, in compar-
ing group 0 to group 2, the probability associated with that statistical test
is .0128. As can be seen, all comparisons yielded statistically significant
differences.