Page 72 -

3.4 Investigating more than one independent variable 57

Either a between-group design or a within-group design may be adopted in this
experiment. In a between-group design, each participant completes tasks under
only one of the six conditions. As a result, six groups of participants would be
required, one group for each condition. In a within-group design, each partici-
pant completes tasks under all six conditions. The advantages and disadvantages of
between-group design and within-group design that we discussed in Section 3.3.2
also apply to factorial designs. No matter which design is adopted, it is impor-
tant to counterbalance the orders and conditions in the experiment. In a between-
group design, the participants need to be randomly assigned to the conditions. In
a within-group design, the order in which the participant completes the six tasks
needs to be counterbalanced.

3.4.2 SPLIT-PLOT DESIGN
In experiments that study one independent variable, we can choose to implement
the study as a between-group design or a within-group design. In a factorial study,
we can also choose a split-plot design. A split-plot design has both between-group
components and within-group components. That is, one or more independent vari-
ables are investigated through a between-group approach and the other variables are
investigated through a within-group approach.
Table 3.3 illustrates an experiment that employs a split-plot design. The experi-
ment investigates two independent variables: age and the use of GPS. The variable
“age” has three levels: people who are 20–40 years old, people who are 41–60 years
old, and people who are older than 60. The second variable has two levels: driving
without GPS and driving with GPS assistance. Therefore, the total number of condi-
tions in this experiment is six.

Table 3.3 A Split-Plot Design
20–40 Years Old 41–60 Years Old Above 60

Driving without GPS 1 2 3
assistance
Driving with GPS 4 5 6
assistance

The impact of age is investigated through a between-group design since three
groups of participants from different age ranges are studied. The impact of the use
of GPS can be examined through a within-group approach. We can require each
participant to complete the same driving task both with and without the assistance of
the GPS. This gives us a typical split-plot design that involves both a between-group
component (age analysis is based on the columns) and a within-group component
(GPS use is analyzed by comparing condition 1 with condition 4, condition 2 with
condition 5, and condition 3 with condition 6).