Critiques du petit p

Critiques du petit p

by Thierry Ancelle

Formal & Physical Sciences Literature & Writing DSRLiterary reference worksDSRCLiterary companions, book reviews and guides
🎙 Thierry Ancelle 👥 25K 📅 September 30, 2019 ⏱ 16 min 👁 5K 📄 expert opinion 🧭 2026-08-17
Available in: English (current) Français

Keywords

p-valuestatistical significancehypothesis testingpublication biasBayesian statistics

Summary

This video by Thierry Ancelle presents a critical examination of the p-value, a cornerstone of classical statistical hypothesis testing. It begins by recalling common misinterpretations of the p-value, then systematically outlines its limitations. The first issue is that the p-value does not measure the degree of evidence against the null hypothesis, as it depends on sample size. The second issue is the violation of the frequentist principle, as the p-value includes probabilities of unobserved outcomes. The video also discusses the impossibility of accepting the null hypothesis, leading to the null hypothesis paradox, and the resulting publication bias that favors significant results. A simulation demonstrates that a large proportion of published significant findings may be false, especially when the prior probability of the null hypothesis is high. The video concludes by highlighting the neglect of base rates, using the controversial example of hepatitis B vaccination and multiple sclerosis, and advocates for Bayesian methods as a more coherent alternative, referencing a Nature article calling for the abandonment of statistical significance.

168 words

Critical Evaluation

Value of the Information & Strength of the Argument

The video provides a valuable and well-argued critique of the p-value. It systematically addresses major criticisms, using clear examples and a simulation to illustrate the impact of publication bias and base rates. The argumentation is logical and builds a compelling case for the limitations of p-values. The video also offers a constructive alternative by mentioning Bayesian methods and the Bayes factor. However, it is an opinion piece and does not present original research, but it effectively synthesizes existing critiques.

Scientific Rigor, Source Quality, Title Accuracy

The video demonstrates scientific rigor by referencing a Nature article from March 2019 signed by 800 scientists calling for the abandonment of statistical significance. It also mentions the work of statisticians like Bezian on base rates. The sources are credible and relevant. The title accurately reflects the content, which is a critical review of the p-value. The video is well-structured and the arguments are presented clearly, though it does not provide a comprehensive list of references beyond the mentioned ones.

174 words

Title / Content Match

The title 'Critiques du petit p' accurately reflects the content, which is a critical review of the p-value.

Quality & Reliability

8/10

The video is a well-structured critique of p-values, presented by an expert in statistics and epidemiology. It covers major criticisms (violation of frequentist principle, non-acceptance of H0, null hypothesis paradox, publication bias, neglect of base rates) and references a Nature article calling for abandoning statistical significance. The content is accurate and aligns with current statistical debates, though it is an opinion piece rather than original research.

Key Moments

Markers derived by PSI from the transcript: the creator did not define chapters.

Cited Sources

Concurring Sources

  • The ASA Statement on p-Values — Supports the critique of p-values and recommends alternative approaches.
  • Nature article on statistical significance — The article mentioned in the video calling for the abandonment of statistical significance.

Dissenting Sources

  • Defending p-values — Some statisticians argue that p-values are useful when used correctly and that the criticisms are overstated. This video does not present counterarguments.

Contribution & Novelties

The video provides a comprehensive and accessible synthesis of the major criticisms of the p-value, making it a valuable educational resource. It goes beyond simple critique by offering a simulation that quantifies the impact of publication bias and base rates, and by suggesting Bayesian methods as a solution. The video is particularly useful for students and researchers who are new to these debates.

Pour aller plus loin :

  • The ASA Statement on p-Values — The American Statistical Association’s official statement on p-values, a key reference.
  • Bayes factor — Wikipedia article on Bayes factor, a Bayesian alternative to p-values.
  • Publication bias — Wikipedia article on publication bias, a central issue discussed in the video.

113 words

Radar Profile

The radar profile shows high scores in quantity and quality of information, and moderate to high technical level and reliability. This indicates a well-informed and credible presentation, though it is an opinion piece rather than original research.

Reliability 8/10