Fifty volunteers with sore throats were randomly allocated to a herbal spray or an identical-looking placebo spray (25 in each group). Neither the volunteers nor the nurse recording symptoms knew who had which. After 2 days, 18 of 25 in the herbal group and 11 of 25 in the placebo group reported feeling better.
Figure 1 shows the design of the experiment. Figure 2 shows the results of 1000 re-randomisations of the 50 outcomes (29 'better', 21 'not better') into two groups of 25, recording the difference in proportions (herbal − placebo) each time; 44 of the 1000 re-randomised differences were 0.28 or more.
| Group | Better | Not better | Total |
|---|---|---|---|
| Herbal | 18 | 7 | 25 |
| Placebo | 11 | 14 | 25 |
Flow diagram: 50 volunteers with sore throats are randomly allocated to a herbal spray group (25) or an identical placebo group (25); after 2 days a blinded nurse records whether each person feels better.
Histogram of 1000 re-randomised differences in the proportion feeling better (herbal − placebo), centred on 0, with the observed difference of 0.28 marked by a dashed line and the tail proportion 0.044 labelled.Facts you may need
(a)
Calculate the proportion feeling better in each group and the difference.
(b)
Explain why a placebo was used and why the nurse was blinded.
(c)
Explain why a further step is needed before concluding the spray works, and describe how the re-randomisation in Figure 2 was produced.
(d)
Use Figure 2 to interpret the result of the re-randomisation test and state what can be concluded about the herbal spray.
A herbal throat spray trial — Scholarship Statistics | Xametric
