In the first article [1] of the statistics bites series it was stated that ideally, exact P values should be reported, for example P = 0.039, rather than inequalities such as 'P < 0.05' or 'P > 0.05'. Information presented this way enables clinicians to make their own interpretations [1] . It was also recommended that confidence intervals be reported in addition to P values, as they are more informative [1, 2] . There are guidelines produced by journals on how many things in an article should

more »

... written, P values are no exception. The International Committee of Medical Journal Editors (ICMJE) [3] advises that they should not be solely relied upon and the American Medical Association (AMA) manual of style [4] states that when P values are reported they should be written to two decimal places, regardless of whether they are significant. Sterne, [5] suggested that making it more precise helped the researcher make the appropriate decision. However, Prayle [6] reported the following P values; P = 2.6 × 10 −9 , P = 4.4 × 10 −11 , P = 2.2 × 10 −16 . The authors [6] have chosen to report exact P values, which will allow readers to interpret how important they are. Clearly, these are not written in the usual manner, normal form, but they are reported in a method known as standard form or scientific notation [7] . This leads to two problems, firstly, how should these numbers be interpreted, and secondly why are P values being reported this way? Standard form is a method used to report either very large or very small numbers easily [7] . A number is said to be in scientific or standard form when it is written