To P or Not to P: That Is the Question

In 1993 Eugene Harris, a well-known biostatistician at the NIH, wrote an editorial for Clinical Chemistry regarding the use of P values vs confidence intervals in the evaluation of scientific data (1). This has long been a contentious topic in the scientific world and one with which each new generation of scientists has had to grapple on the basis of arguments advanced for the use of various statistical approaches. Harris recommended use of the statistical technique that best met the objectives of the experiment, but this presumes at least some knowledge of the underlying statistical techniques.An excellent and instructive article recently appeared in Nature that explores this subject further, giving insights into the ramifications of incorrect interpretation of statistics (2). This article, written by Regina Nuzzo, is highly recommended to our readership because it provides a very clear and interesting presentation. The article presents a history of the competing models of legendary statisticians, the model of Ronald Fisher, who initially proposed …