Subjective and Objective Probability

Bayesian probability can express a rational agent's degree of belief or be constrained by formal rules that minimize the analyst's influence, and this topic contrasts the two stances.

Definition

Subjective probability is a coherent quantification of an individual's uncertainty, derived from rational preference axioms; objective probability, in the Bayesian context, refers to priors and procedures chosen by formal conventions intended to be minimally informative or invariant.

Scope

This topic covers the subjective (personalist) interpretation grounded in coherent preferences, the objective Bayesian program that seeks rule-based priors, and the decision-theoretic axioms that derive probability and utility from rational behavior.

Core questions

• How is subjective probability derived from axioms of rational preference?
• What distinguishes objective Bayesian methods from subjective ones?
• Why do coherence and the avoidance of Dutch books motivate the use of probability for beliefs?
• What roles do utility and decision theory play in defining Bayesian probability?

Key concepts

• subjective probability
• objective Bayesian
• coherence
• Dutch-book argument
• utility
• rational preference axioms

Key theories

Personalist (subjective) probability

Probability is defined as a coherent degree of belief; Savage's axioms show that a rational agent's preferences imply both a probability measure and a utility function.

Objective Bayesianism

Objective approaches seek priors determined by formal rules, such as invariance or maximum information content, so that conclusions depend as little as possible on personal choices.

Clinical relevance

The choice between subjective and objective stances shapes how priors are justified in regulated settings such as clinical trials, risk assessment, and policy analysis, where transparency about prior assumptions matters.

History

Ramsey and de Finetti developed subjective probability in the 1920s-1930s; Savage's 1954 axiomatization unified probability and utility. In parallel, Jeffreys pursued objective rule-based priors, seeding a long-running methodological dialogue.

Debates

Should priors be subjective or objective?

Subjectivists argue that all probability is personal and priors should encode honest belief, while objectivists seek conventional priors to make analyses reproducible and reduce the influence of the analyst.

Key figures

• Bruno de Finetti
• Leonard J. Savage
• Frank Ramsey
• Harold Jeffreys

Seminal works

• savage1954
• bernardo1994

Frequently asked questions

Does using a subjective prior make Bayesian analysis unscientific?

No. Subjective priors are stated explicitly and can be examined, varied, and checked through sensitivity analysis, which makes the assumptions transparent rather than hidden inside the choice of method.