Correlation, regression, and repeated dataBMJ 1994; 308 doi: https://doi.org/10.1136/bmj.308.6942.1510a (Published 04 June 1994) Cite this as: BMJ 1994;308:1510
- R Persaud
- nstitute of Psychiatry, London SE5 8AF
- Department of Public Health Sciences, St George's Hospital Medical School, London SW17 0RE Medical Statistics Laboratory, Imperial Cancer Research Fund, London WC2A 3PX.
EDITOR, - J Martin Bland and Douglas G Altman's note on correlation, regression, and repeated measures draws attention to the problem of calculating a correlation coefficient based on repeated observations of the same subject.1 But they fall prey themselves to the common misjudgment of placing too much importance on the significance of the correlation coefficient.
The basic problem is the way in which significance is calculated as the correlation coefficient is strongly affected by the number of cases for which there are pairs of data. If, for example, you have about 500 cases the correlation coefficient needs only to be 0.088 to be significant at …