Lesson 9: Brain research, data and recovery from injury
Analyse a treatment trial with a paired t-test
SkillsAnalysing brain research dataParkinson's and Alzheimer's disease
Worked example
Ten people with Parkinson's disease were scored on a motor symptom scale (higher = more severe) before and after four weeks of a new medicine (hypothetical data).
| Participant | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 |
|---|---|---|---|---|---|---|---|---|---|---|
| Before | 38 | 42 | 35 | 47 | 40 | 33 | 45 | 39 | 44 | 37 |
| After | 31 | 36 | 33 | 38 | 35 | 30 | 40 | 35 | 37 | 34 |
(a) Calculate the mean and standard deviation of the scores before and after treatment, to 2 decimal places. (b) Identify the appropriate -test and justify your choice. (c) A paired -test gives , , . Draw a conclusion about the hypothesis that the medicine changes motor symptom scores.
Step 1: Means
Before: . After: .
Step 2: Standard deviations
With the calculator's sample SD (): before , after .
Use the sample standard deviation , not : the participants are a sample.
Step 3: Choose the test
Each participant was measured in both conditions (repeated measures), so the scores are paired: use a paired -test, which tests each person's change.
Step 4: Use p
, which is less than 0.05, so reject the null hypothesis: the fall in mean score from 40.00 to 34.90 is statistically significant.
Step 5: Limit the conclusion
There was no placebo control group, so the improvement cannot be attributed to the medicine alone (expectation, practice or time could contribute), and 10 participants limit generalisation.
Answer
(a) Before: mean 40.00, SD 4.50. After: mean 34.90, SD 3.07. (b) Paired -test, because the same participants were measured twice (repeated measures). (c) , so the reduction in motor symptom scores is significant; but without a placebo control the medicine cannot be shown to be the cause.
Try a similar one
A fresh question of the same kind, checked as you answer, with hints and a full solution.
Method
Means and sample SDs to 2 decimal places.
Repeated measures or matched → paired; independent groups → unpaired.
Compare with 0.05; state the decision, then the limitation.
Watch out for
Using the population SD.
Concluding the medicine "works" without mentioning the missing control group.