Chapter 11
I. Definition of related samples
A.Repeated-measures (or within-subjects) -
B.Matched-subjects (matched-groups) -
(Matching overhead)
Table 1
Before Matching
| Group 1 | Group 2 | |||||
| Subject | X | MV (IQ) | Subject | X | MV (IQ) | |
| 1 | 22 | 100 | 6 | 16 | 101 | |
| 2 | 17 | 95 | 7 | 26 | 110 | |
| 3 | 31 | 110 | 8 | 28 | 105 | |
| 4 | 27 | 105 | 9 | 18 | 103 | |
| 5 | 24 | 102 | 10 | 14 | 100 |
Table 2
After Matching
| Group 1 | Group 2 | |||||
| Pair | Subject | X | Subject | X | D | |
| 1 | 3 | 31 | 7 | 26 | 5 | |
| 2 | 4 | 27 | 8 | 28 | -1 | |
| 3 | 5 | 24 | 9 | 18 | 6 | |
| 4 | 1 | 22 | 6 | 16 | 6 | |
| 5 | 2 | 17 | 10 | 14 | 3 |
C. Fundamental unit of analysis is the difference score (D
= X1 - X2)
(Table 11.3 overhead)
II. t-test for two related samples
A. Assumptions
1. Population distribution of difference scores is normal.
2.Observation are independent within each treatment
B.Formula for t-test for two related samples - same as one sample formula except with D scores
C. standard error for t-test is pooled from the
individual sample variances
D. Degrees of freedom df, n = number of pairs
df = n - 1
III. Steps in hypothesis test using related samples t
A. State hypotheses - now about relation of difference
scores means
Ho: ×D = 0
H1: ×D ‚ 0
(Fig. 11.1 overhead)
B. Set alpha
C. Match subjects, collect data, compute difference scores and perform statistical test
D. Make decision about Ho:
V. Design considerations with repeated measures
A.Advantages
B.Disadvantages
1.Carryover effects -
2.Progressive error -