P values are still usefulBMJ 1994; 309 doi: https://doi.org/10.1136/bmj.309.6966.1439a (Published 26 November 1994) Cite this as: BMJ 1994;309:1439
- E Cobo,
- M J Campbell
- Statistician, Departament d'Estadistica, Universitat Politecnica de Catalunya, 08010 Barcelona, Spain.
- Reader in medical statistics, University of Southampton, Southampton SO16 6YD.
EDITOR,—While we agree with Paul Brennan and Peter Croft that the greatest emphasis in reporting observational studies should be on questions of confounding and bias, we believe that their suggestion that the term “statistically significant” should be abandoned is akin to banning a useful implement such as a kitchen knife on …