Solution to question 11.5
I will use the 'magic number' method of calculating the sample size.
Standard deviation = 5·0
Difference worth finding = 4·0
α = 0·01 (1%)
Power = 80%
From table, magic number = 11·7
We need:
2 x 5·0 x 5·0 x 11·7 ÷ 4·0 ÷ 4·0 = 37 patients per group
(Remember to round up to the next whole number.) So, we need 74 patients for the whole study. We might phrase the sample size calculation for the research proposal as so:
Pilot studies indicated that measurements were likely to have a standard deviation of 5·0; we wish to detect minimum difference of 4·0 units. A sample size calculation indicates that a minimum of 37 patients per group would be required to give 80% power at a 1% significance level. To allow for drop-outs we propose to recruit 40 patients to each group (80 in total).
