Solution to question 11.2
I will use the 'magic number' method of calculating the sample size.
Difference worth finding = 1·5
Standard deviation = 3·0
α = 0·05 (5%)
Power = 90%
From table, magic number =< 10·5
We need:
2 x 3·0 x 3·0 x 10·5 ÷ 1·5 ÷ 1·5 = 84 patients per group
Or, 168 patients for the whole study, which we would probably round up to 170.
