The p-value and the statistical significance debate
Ronald FisherJerzy NeymanEgon PearsonRonald WassersteinDaniel BenjaminValentin AmrheinDaniël Lakens
This page delivers no verdict; it presents claims and evidence with their sources. No theory in science is absolute. Method
⑂ Divergence of views. Views diverge on this question; each prong on the tree is one position.
For everyone
In scientific studies the threshold p < 0.05 is usually used to count a result as "statistically significant". The tradition rests largely on Ronald Fisher's book of 1925. After the replication crisis it became widely debated that p-values are misinterpreted and that tying conclusions to a single threshold leads to misleading results. The American Statistical Association issued a statement on the matter in 2016. Researchers disagree on the proposed remedies.
For specialists
The p-value is the probability of obtaining a result at least as extreme as the one observed if the null hypothesis and all the model's assumptions are true; it is not the probability that the null hypothesis is true. Fisher saw the p-value as a continuous measure of the strength of evidence, while Neyman and Pearson proposed making decisions with pre-set error rates. Today's practice is a hybrid of the two. Low statistical power, flexible analysis choices and the publication of significant results only raise the false positive rate.
Views on dating
No single date is given because the date is not documented with certainty; the views of different sources are listed.
| Origin of the 0.05 threshold | Fisher's book of 1925; Fisher proposed the value as a convenient reference point, not as a strict rule.[1] |
|---|
Claims
- P-values alone show neither the probability that a hypothesis is true nor the size of an effect; scientific conclusions should not rest only on passing a significance threshold.[2]
Positions
- Those proposing a lower threshold (Benjamin and 71 co-authors)
- The default significance threshold for new discoveries should be lowered from 0.05 to 0.005.[3]
- Those proposing to retire significance (Amrhein and more than 800 signatories)
- The concept of "statistical significance" should be retired altogether, and p-values reported as continuous values together with confidence intervals.[4]
- Those proposing to justify the threshold (Lakens and co-authors)
- Instead of one universal threshold, each researcher should justify in advance the threshold chosen for the conditions of their study.[5]
- Those proposing Bayesian methods
- The relative support for hypotheses should be measured directly with Bayes factors and posterior distributions.[3]
Connections
Sources
- Book (1925). Statistical Methods for Research Workers. Oliver and Boyd, Edinburgh
- Article (2016). The ASA's statement on p-values: context, process, and purpose. The American Statistician 70(2), 129–133DOI 10.1080/00031305.2016.1154108archive
- Article (2018). Redefine statistical significance. Nature Human Behaviour 2, 6–10DOI 10.1038/s41562-017-0189-zarchive
- Article (2019). Scientists rise up against statistical significance. Nature 567, 305–307DOI 10.1038/d41586-019-00857-9archive
- Article (2018). Justify your alpha. Nature Human Behaviour 2, 168–171DOI 10.1038/s41562-018-0311-xarchive
Search the literature
Look for scholarly work on this topic in other indexes:
Cite this page
Athena contributors. (2026). The p-value and the statistical significance debate. Athena: The tree of human knowledge. Retrieved November 1, 2026, from https://latifkedi.github.io/athena/en/d/p-degeri/
@misc{athena_p_degeri,
title = {The p-value and the statistical significance debate},
author = {{Athena contributors}},
year = {2026},
howpublished = {\url{https://latifkedi.github.io/athena/en/d/p-degeri/}},
note = {Athena: The tree of human knowledge. Accessed: 2026-11-01}
}Added: September 28, 2026 · Last changed: September 28, 2026 · Page history