Information Technology Reference
In-Depth Information
Our next computational steps would be to examine the main compar-
isons of our between-subjects variable (Factor
A
) and our within-subjects
variable (Factor
B
) and explore the nature of the statistically significant
A
B
interaction through simple effects analysis of Factor
A
at
b
k
or
Factor
B
at
a
j
. The procedures for the between-subjects main compar-
isons are straightforward and follow the same procedures and formulas
we reviewed in Chapter 8 on between-subjects factorial analyses of main
effects. The only difference is that the error term becomes
MS
S
/
A
.
The hand computations for the within-subjects main comparison and
the simple effects analyses require the computation of separate error
terms and are somewhat tedious to do and will not be covered here.
The interested reader is encouraged to review Keppel (1991, Chapter 5)
and Keppel and Wickens (2004, Chapter 19) for numerical examples.
Keppel et al. (1992, Chapter 12) offers a more “simplified” and com-
putationally more elegant approach using
MS
w
.
cell
(the average of the
variances of all treatment combinations) as the error term when con-
ducting simple effects analyses of
A
at
b
k
and simple comparisons of
A
at
b
k
. This latter approach is recommended only for students who are
just beginning to learn hand computational procedures of simple mixed
designs.
×
13.5 PERFORMING THE OMNIBUS ANALYSIS IN SPSS
13.5.1 STRUCTURING THE DATA FILE
The data file for our 3
2 simple mixed design example is shown in
Figure 13.4. The first column, as always, is used for our participant iden-
tification number; we have named this variable
subid
. The next column
holds our between-subjects variable of leadership style. Between-subjects
variables take arbitrary numeric codes; here we have coded authoritarian,
democratic, and laissez faire as 1, 2, and 3, respectively.
The last two columns in the data file represent our repeated measure
of project type. Under each level of this within-subjects variable we have
recorded the performance evaluation of management.
×
13.5.2 STRUCTURING THE DATA ANALYSIS
From the main SPSS menu, select
Analyze
➜
General Linear Model
➜
RepeatedMeasures
. You select
RepeatedMeasures
for the following rea-
son. If there are two or more cases in the study and at least one of the
independent variables is a within-subjects variable, then we must partition
the total variance of the dependent variable into a between-subjects por-
tion and a within-subjects portion. This partitioning is accomplished by
the
Repeated Measures
module of the
General Linear Model
procedure
in SPSS.
Selecting this path will open the dialog window shown in Figure 13.5.
As you will recall from Chapters 10, 11, and 12, this is the initial window
used by SPSS to have you name the within-subjects variable(s) in the
