Three-Way Between-Subjects Design - Analysis of Variance Designs - page 220

variable is a subject identification code (subid) . The next three variables
are the independent variables in the analysis; each has been assigned
arbitrary codes to represent the categories. These variables are kidshome
(children present in the home coded as 1, children not in the home coded
as 2), voter (having voted in the last election coded as 1, not voting in the
last election coded as 2), and politics (liberal, moderate, and conservative
coded as 1, 2, and 3, respectively). The last variable in the data file is
our dependent variable, named satisfac , the score on the survey assessing
satisfaction with the public school system.

9.6.2 STRUCTURING THE ANALYSIS
From the main menu select Analyze ➜ General LinearModel ➜ Univari-
ate to reach the main dialog window shown in Figure 9.5. Click satisfac
over to the Dependent Variable panel and in this order click kidshome ,
voter , and politics over to the Fixed Factor(s) panel. We will first run
the omnibus analysis to determine which effects are significant. Based on
those results we will then determine which post-ANOVA analyses we will
perform.
The order in which we bring the variables into the Fixed Factor(s)
panel is important when we have three or more independent variables in
the analysis. Recall our discussion in Chapter 8 about the way in which
the observed means are displayed in the Descriptive Statistics output
table. We saw that earlier specified variables change more slowly than later
specified variables. Having structured the presentation of the example in

Figure 9.5 The main GLMUnivariate window.