Sample sizeBMJ 1994; 308 doi: https://doi.org/10.1136/bmj.308.6939.1304 (Published 14 May 1994) Cite this as: BMJ 1994;308:1304
- I Hill-Smith
- Wigmore Lane Health Centre, Luton LU2 8BG.
EDITOR, - J Martin Bland and Douglas G Altman provide refreshing insight into a statistical trap for the unwary.1 They point out that in estimations of the correlation between measurements the sample size is usually the number of subjects, not the number of measurements. How bad the error is depends on whether the variability in observations on different subjects is much greater than the …