Lesson 8: Reducing prejudice through intergroup contact
Analyse before-and-after data from a jigsaw program
SkillsAnalysing cross-cultural dataReducing prejudice through contactResearch methods across cultures
Worked example
Illustrative data: ten students completed a prejudice scale (higher scores mean more prejudice) before and after eight weeks of jigsaw groups.
| Student | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 |
|---|---|---|---|---|---|---|---|---|---|---|
| Before | 34 | 29 | 38 | 31 | 27 | 36 | 33 | 30 | 35 | 28 |
| After | 30 | 27 | 33 | 30 | 24 | 31 | 31 | 29 | 30 | 27 |
(a) Identify the design and the appropriate t-test. (b) Calculate the mean of the differences (before − after) and the t statistic, to 2 decimal places. (c) With , state a conclusion. (d) Identify one limitation of the design and suggest an improvement.
Step 1: Identify the design and test
Each student was measured twice: repeated measures, so the paired t-test.
Step 2: Find each student's difference
Before − after: 4, 2, 5, 1, 3, 5, 2, 1, 5, 1.
Step 3: Mean and standard deviation of the differences
. With technology, .
Step 4: Standard error and t
; with .
Step 5: Conclude
, so reject the null hypothesis: prejudice scores fell significantly, by a mean of 2.90 points, after the program.
Step 6: Limitation and improvement
There is no control group, so something else over eight weeks (other lessons, maturation, getting used to the scale) could explain the fall. Add a control class taught without jigsaw groups, with classes randomly allocated, and compare the change in both.
A repeated measures design controls for differences between students, but not for events that happen to everyone over time.
Answer
(a) Repeated measures; paired t-test. (b) Mean difference 2.90; (). (c) : prejudice scores fell significantly after the program. (d) No control group, so other changes over the eight weeks could explain the fall; add a randomly allocated control class without jigsaw groups.
Try a similar one
A fresh question of the same kind, checked as you answer, with hints and a full solution.
Method
Match the test to the design (paired for repeated measures or matched participants).
Compute each difference, then , , and .
Compare p with 0.05 and state the conclusion in words.
Watch out for
Using the unpaired test: it gives here and would wrongly suggest no significant change.
Claiming the program caused the fall without a control group.