Toward evidence-based medical statistics. 2: The Bayes factor

Area of Science:

• Statistics
• Scientific Inference

Background:

• Bayesian inference is often perceived as subjective by medical researchers.
• Despite this, Bayesian methods possess a data-driven core crucial for evidence evaluation.

Purpose of the Study:

• To highlight the objective, data-based core of Bayesian inference.
• To present the Bayes factor as a superior alternative to P values for measuring evidential strength.

Main Methods:

• Focus on the Bayes factor, a core component of Bayesian methodology.
• Utilize the minimum Bayes factor as an objective measure of evidence.

Main Results:

• Bayes factors provide a theoretically sound interpretation for inference and decision-making.
• P values are shown to significantly overstate evidence against the null hypothesis.
• Bayes factors clearly distinguish experimental evidence from inferential conclusions.

Conclusions:

• The Bayes factor serves as a reliable, objective measure of evidential strength.
• Bayesian methods, particularly Bayes factors, offer a robust framework for scientific reasoning.
• Integrating prior knowledge with experimental data is essential for robust inferential conclusions.