Information Technology Reference
In-Depth Information
Keppel and Wickens (2004), and Kirk (1995), for more information on
this topic.
17.3 NESTED DESIGNS
The factorial designs used throughout this topic have purposefully been
conceptualized to include all possible combinations of the levels of the
independent variables. These designs are known as
crossed
or
completely
crossed factorial designs
and allow researchers to examine the interaction
among the independent variables or factors. Occasionally researchers are
unable or do not wish to employ a full factorial design. One instance of
a design that is not a full factorial is the
nested
or
hierarchical design
.Ina
nested design the levels of one factor do not occur at all levels of another
factor.
As an example of a nested design, consider the hypothetical crossed
and nested designs depicted in Figure 17.3. In the crossed design shown
in Figure 17.3A, Factor
A
has three levels. Assume that it represents type
of treatment (brief psychotherapy, cognitive-behavioral, psychoanalytic).
This variable is completely crossed with Factor
B
, which could be the
community mental health center location (Location 1, 2, 3, 4, 5, and 6)
at which the data were collected. Each level of treatment (
a
)appearsonce
with each level of treatment (
b
), producing (
a
)(
b
)
=
(3)(6)
=
18 treatment
combinations.
In the nested design shown in Figure 17.3B, Factor
B
(location) is
nested within Factor
A
(type of treatment), or
B
/
A
. Here, locations
b
1
A
Crossed
Factor A
a
3
a
1
a
2
Factor B
b
1
b
2
b
3
b
4
b
5
b
6
1
b
2
b
3
b
4
b
5
b
6
1
b
2
b
3
b
4
b
5
b
6
B
Nested
Factor A
a
2
a
3
a
1
Factor B
b
1
b
2
b
3
b
4
b
5
b
6
Figure 17.3
Example of a completely crossed design, A, and a nested design, B.
