Statistical Methods for Psychology

not pool other error terms that apply to other simple effects. In other words, it will be based

solely on the Control group. We can test the Interval simple effects quite easily by running

separate repeated-measures analyses of variance for each of the groups. For example, we

can run a one-way repeated-measures analysis on Intervals for the Control group, as dis-

cussed in Section 14.4. We can then turn around and perform similar analyses on Intervals

for the Same and Different groups separately. These results are shown in Table 14.6. In

each case the Interval differences are significant, even after we correct the degrees of free-

dom using or , whichever is appropriate.

If you look at the within-subject analyses in Table 14.6, you will see that the average

is (2685.669 1 3477.571 1 1871.026)/3 5 2678.089, which is

from the overall analysis found on page 473. Here these denominators for the Fratios are

noticeably different from what they would have been had we used the pooled term, which

is the traditional approach. You can also verify with a little work that the terms

for each analysis are the same as those that we would compute if we followed the usual

procedures for obtaining simple effects mean squares.

For the between-subjects simple effects (e.g., Groups at Interval 1) the procedure is

more complicated. Although we could follow the within-subject example and perform sep-

arate analyses at each Interval, we would lose considerable degrees of freedom unnecessarily.

Here it is usually legitimate to pool error terms, and it is generally wise to do so.

MSInterval

MSerror MSI 3 Ss w/in groups

́N ~ ́

480 Chapter 14 Repeated-Measures Designs

Table 14.6 Calculation of within-subjects simple effects for data from King (1986)

(a) Interval at Control

Source df SS MS F

Between subjects 7 134,615.58

Interval 5 76,447.25 15,289.45 5.69*

Error 35 93,998.42 2685.67

Total 47 305,061.25

*p,.05; 5 .404; 5 .570

(b) Interval at Same

Source df SS MS F

Between subjects 7 175,600.15

Interval 5 193,090.85 38,618.17 11.10*

Error 35 121,714.98 3477.57

Total 47 490,405.98

*p,.05; 5 .578; 5 1.00

(c) Interval at Different

Source df SS MS F

Between subjects 7 74,506.33

Interval 5 211,018.42 42,203.68 22.56*

Error 35 65,485.92 1871.03

Total 47 351,010.67

*p,.05; ́N 5 .598; ~ ́ 5 1.00

́N ~ ́

́N ́~