Bayesian statistics may inform public policy better than significant odds ratiosBMJ 1997; 314 doi: https://doi.org/10.1136/bmj.314.7088.1202a (Published 19 April 1997) Cite this as: BMJ 1997;314:1202
- David Braunholtz, Medical statisticiana,
- Richard Lilford, Professor of health services researchb
- a Medical Statistics, Leeds University, Leeds LS2 9LN
- b University of Birmingham, Birmingham B16 9PA
Editor-We agree with many of the points raised in the correspondence1 commenting on our article2 but not necessarily the conclusions. For example, Paul Brennan states that insufficient attention is generally paid to the possibility of bias and confounding when assessing epidemiological data, and we agree with D R Cox and V T Farewell that sensitivity analysis is thus important.1 However, we see advantages in comparing the effects of a variety of assumed distributions for bias rather than a variety of fixed biases.
Cox and Farewell point …