Information Technology Reference
In-Depth Information
The GLM Procedure
Least Squares Means
Adjustment for Multiple Comparisons: Tukey
LSMEAN
Number
satscore
LSMEAN
group
0
2
4
6
8
1
2
3
4
5
412.857143
474.285714
552.857143
614.285714
622.857143
Least Squares Means for effect group
Pr > t for H0: LSMean(i) = LSMean(j)
Dependent Variable: satscore
The difference between
group 1 (no study time)
and group 2 (two months
of study time) is
statistically significantly
different.
i/j
1
2
3
4
5
1
2
3
4
5
0.0276
<.0001
<.0001
<.0001
0.0276
0.0030
<.0001
0.0276
<.0001
<.0001
<.0001
<
0.0030
<
0.0093
0.9918
.0001
0.0276
0.0093
.0001
0.9918
<.0001
The difference between
group 4 (6 months of
study time) and group 5
(8 months of study
time) is not statistically
significantly different.
Figure 7.11
SAS output for the Tukey post hoc test.
We briefly consider these characteristics here; we will then describe the
contrasts based on these characteristics in Sections 7.15.1 to 7.16. SAS has
nothing analogous to these preset contrasts in point-and-click mode.
7.15.1.1 Dialog Window Containing the Contrast
Five of the six contrasts are available through the Contrasts dialog window.
This window is not available in the One-Way ANOVA procedure we have
been using thus far. Rather, we will access Contrasts through the General
Linear Model (GLM) procedure. One of these contrasts, the Dunnett
procedure, is available in the Post Hoc dialog window which can be
accessed through either the One-Way ANOVA or the GLM procedure.
7.15.1.2 Pairwise/Nonpairwise
Pairwise comparisons deal with means of separate conditions. In the non-
pairwise comparisons for these contrasts, one of the two means being
compared is a linear combination of means from multiple groups.
Search WWH ::




Custom Search