Information Technology Reference
In-Depth Information
Kids in Home
Liberal
Moderate
Conservative
36
14
10
40
21
18
Vote Yes
32
7
2
39
20
15
33
36
8
14
5
10
35
32
15
40
39
21
Vote No
30
25
9
39
38
22
31
35
26
32
8
15
Satisfaction with
Public Schools
Range 0 - 40
Kids Not in Home
Liberal
Moderate
Conservative
33
35
20
40
40
27
Vote Yes
26
30
13
40
39
26
26
33
31
35
14
20
28
18
36
35
25
40
Vote No
21
11
32
36
22
40
20
28
14
18
32
36
Figure 9.1
Design and data for our three-way between-subjects example.
present in the home, although we could just as legitimately have broken
the data apart by one of the other independent variables. Once we select
the variable to split apart, it becomes a bit more difficult to conceptually
compare the levels of this variable because each of its levels is in separate
tables.
Based on how we schematically presented the design in Figure 9.1, we
can say that this design is a 2
2, although the order of specifying
the independent variables is arbitrary. The number of separate cells in the
design - the number of unique combinations of levels of the independent
variables - is known by multiplying the number of levels of each variable.
Thus,therearetwelvecellsinthedesign(2
×
3
×
12), each repre-
senting a unique combination of the three independent variables. This
tier of analysis - working with the twelve cell means - is the essence of the
design and represents the
highest order
effect (i.e., the effect with the most
detail).
×
3
×
2
=
9.2 PARTITIONING THE VARIANCE INTO ITS SOURCES
9.2.1 THE SUMMARY TABLE
As was true of the all the designs that we have been and will be discussing,
the total variance of the dependent variable is partitioned into its sep-
arate sources. These sources are the effects that are associated with the
independent variables and the error variance.
