Many reports in the medical literature are flawed through poor presentation of numerical results, or, more seriously, by the use of inadequate or inappropriate statistical methodology. This article expands on the more complex methodological issues. The main emphasis is on discussing key aspects of experimental design, and the idea of reporting results in terms of confidence intervals rather than P values.
|Number of pages||5|
|Journal||British Journal of Surgery|
|Publication status||Published - Jul 1991|