P-values aren't the end-all, be-all of good data analysis, write Jeffrey Leek and Roger Peng from the Johns Hopkins Bloomberg School of Public Health at Nature.
They note that the journal Basic and Applied Social Psychology in February basically banned the use of null hypothesis significance testing, arguing that the procedure is invalid.
Leek and Peng say that move won't affect the quality of published science as "there are many stages to the design and analysis of a successful study," the last of which is usually the calculation of an inferential statistic like a P-value. Decisions at other stages of the experimental process such as not adjusting for confounding or batch effects can have a greater effect on results. Further, "[a]rbitrary levels of statistical significance can be achieved by changing the ways in which data are cleaned, summarized, or modeled," they add.
Instead, Leek and Peng argue that researchers need to be better trained in data analysis and that there needs to be better study of how such analysis is actually carried out. "The ultimate goal is evidence-based data analysis," they say.