interaction, and speaks to the value of examining simple effects. The fact that an effect we
seek is significant does not necessarily mean that it is significant in the direction we desire.14.9 Two Within-Subjects Variables and One Between-Subjects Variable
The design we just considered can be seen as a straightforward extension of the case of one
between- and one within-subjects variable. All that we needed to add to the summary table
was another main effect and the corresponding interactions. However, when we examine a
design with two within-subjects main effects, the problem becomes slightly more compli-
cated because of the presence of additional error terms. To use a more generic notation, we
will label the independent variables as A, B, and C.
Suppose that as a variation on the previous study we continued to use different subjects
for the two levels of variable A (Gender), but we ran each subject under all combinations
of variables B(Condition) and C(Trials). This design can be diagrammed as
A 1 A 2
C 1 C 2 C 3 C 1 C 2 C 3
B 1 G 1 G 1 G 1 G 2 G 2 G 2
B 2 G 1 G 1 G 1 G 2 G 2 G 2
B 3 G 1 G 1 G 1 G 2 G 2 G 2488 Chapter 14 Repeated-Measures Designs
Table 14.9 Analysis of simple effects
(a) Between-subjects effects (Condition, Sex, and Condition 3 Sex) at Pretest
Source df SS MS F
Condition 1 403.225 403.225 1.45
Sex 1 2544.025 2544.025 9.13*
Condition 3 Sex 1 119.025 119.025 0.43
Error 36 10027.100 278.530
Total 39 13093.375(b) Within-subject effects (Sex, Time, Time 3 Sex) at BST
Source df SS MS F
Between subjects 19 7849.13
Sex 1 2173.61 2173.61 6.89*
Error (between) 18 5675.52 315.30Within subjects 60 3646.26
Time 3 338.94 112.98 1.88
T 3 S 3 54.54 18.18 0.30
Error (within) 54 3252.78 60.24
Total 79 11495.39
*p,.05