Information Technology Reference
In-Depth Information
compared to each of the remaining groups. You can perform this test in
either a two-tailed manner (direction of difference not hypothesized) or
in a manner that hypothesizes the control group to have either a higher or
lower mean than the other groups. This is the same general strategy that
is used by Simple contrasts; the directional contrast has about the same
amount of power as the Simple contrasts procedure.
7.15.2.4 Helmert Contrasts
Each group except for the last group is compared to the average of the
groups coded with higher values (the last group by definition has no
groups with value codes above it). In our SAT study time example, the
zero-month study group would be compared to the groups having studied
for two, four, six, and eight months; the two-month study group would
be compared to the groups having studied for two, four, and six months;
and so on. This is the mirror image of Difference contrasts.
7.15.2.5 Repeated Contrasts
Adjacent groups are compared to each other. In our SAT study time
example, the zero-month study group would be compared to the two-
month study group, the two-month study group would be compared to
the four-month study group, and so on.
7.15.2.6 Simple Contrasts
The reference group (first or last group) is the focus of the analysis. It
is separately compared to each the remaining groups. This is the same
strategy that is used by Dunnett contrasts with power that is comparable
to a directional Dunnett comparison.
7.16 PERFORMING SIMPLE CONTRASTS IN SPSS
We w i l l u s e Simple contrasts to illustrate how to work with preset contrasts;
the setup for the analysis and the output are very similar for the other
preset contrasts. For these preset contrasts, we use the General Linear
Model (GLM) procedure. Once you have opened your data file, follow this
navigation route from the menu bar at the top of the screen: Analyze
General Linear Model Univariate. That will bring you to the main
dialog window for the procedure as shown in Figure 7.12. You can see in
thefigurethatwehaveclickedover satscore to the Dependent Variable
panel and have clicked over group to the Fixed Factor(s) panel. This setup
is quite similar to what we have already described in the context of the
One-Way ANOVA procedure.
Clicking the Contrasts pushbutton brings us to the Contrasts dialog
window shown in Figure 7.13. When it opens, the independent variable is
Search WWH ::




Custom Search