P-value (probability value)
The probability of obtaining an experimental result at least as extreme as the one observed, where a relationship does not exist between phenomena.
A Significant History of P-values
The widespread use of p-values in significance testing today can be largely attributed to the work of British statistician and biologist Sir Ronald Fisher. In Statistical Methods for Research Workers, published in 1925, Fisher famously proposed: "The value for which P=.05, or 1 in 20, is 1.96 or nearly 2; it is convenient to take this point as a limit in judging whether a deviation is to be considered significant or not.” Fisher never intended for P=0.05 to become a fixed rule and, 30 years later, he clarified his belief that significance levels should be set according to the circumstances of the experiment.
Fisher’s clarification was mostly ignored and P=0.05 has become the standard for determining statistical significance across the physical and social sciences. However, with ever-growing computing power and datasets available for experimentation and researchers failing to replicate the results of many studies, some have begun to question whether the arbitrary threshold is still fit for purpose. The behavioural economist Daniel J. Benjamin, for example, is calling for the threshold to be reduced to P=0.005. In some disciplines, researchers already adjust the p-value according to the circumstances: in the ATLAS experiment at CERN, which was devised to detect the Higgs particle, the p-value was extremely small. It corresponded to a 1 in 3,500,000 chance that the signal detected would appear by chance, if there was no Higgs particle.