Information Technology Reference
In-Depth Information
Car/No
Drinks
Car/One
Drink
Car/Three
Drinks
SUV/No
Drinks
SUV/One
Drink
SUV/Three
Drinks
Subid
1
1
1
2
2
0
0
4
6
7
9
10
7
6
2
5
6
6
4
4
2
4
7
5
6
5
3
2
5
1
3
6
1
1
4
5
5
3
2
6
1
1
Mean
1.00
4.00
7.50
1.33
4.50
5.00
Figure 11.1
Raw data for numerical example.
11.2 A NUMERICAL EXAMPLE OF A TWO-WAY
WITHIN-SUBJECTS DESIGN
The simplified data that will serve as our example use two time-
independent variables as shown in Figure 11.1. We are studying driving
errors that are made in emergency handling situations on a standardized
route. Participants are six students who are given extensive experience
with the route so that it is familiar to them. The dependent variable is the
number of errors made by the students during a loop around the route.
There are two independent variables in the study. For each jaunt the
students will be tested after they have consumed either zero, one, or
three alcoholic beverages. They will drive a sport sedan (car) on one type
of driving occasion and a family-sized, sport utility vehicle (SUV) on
another type of occasion. Because this is a within-subjects design, each
student will be tested under all of the experimental conditions although
the order will be randomized for each student to balance the influence of
any carry-over effects on the conditions. That is, each student will drive
the car after consuming no alcoholic drinks, also after consuming one
alcoholic drink, and also after consuming three alcoholic drinks; he or
she will also drive the SUV after consuming no alcoholic drinks, also after
consuming one alcoholic drink, and also after consuming three alcoholic
drinks.
11.3 PARTITIONING THE VARIANCE INTO ITS SOURCES
The two summaries of the ANOVA, one for the between-subjects variance
and another for the within-subjects variance, are shown in Table 11.1. We
will discuss the
between-subjects effects
first.
11.3.1 BETWEEN-SUBJECTS EFFECTS
The upper portion of Table 11.1 shows the between-subjects effects; it
summarizes the between-subjects source of variance. Some students gen-
erally make fewer driving errors than other students. This represents the
individual differences among the six participants; this source of variance
is known as
error
(it is actually
between-subjects error variance
) rather than
between-subjects variance. With six participants there are 5
df
.No
F
ratio
is ordinarily computed for individual differences.
