Subjective probability
Also known as: Bayesian probability, Credence
Probability as a personal degree of belief, not a long-run frequency.
What it means
The interpretation of probability as an individual's degree of belief in an uncertain proposition, rather than as the limiting frequency of an event in repeated trials. On this view a probability can be attached to one-off, unrepeatable events — an election, a diagnosis, a single project's success — by eliciting how strongly someone believes it, ideally in a way that is coherent and updates by Bayes' rule as evidence arrives. Coherence is enforced by the 'Dutch book' argument: anyone whose subjective probabilities violate the laws of probability can be offered a set of bets guaranteed to lose. Human-elicited subjective probabilities are workable but prone to overconfidence, anchoring, and partition dependence, so they require careful elicitation and calibration. The concept is foundational to Bayesian statistics, decision analysis, and the whole study of judgment under uncertainty, where the object being judged is precisely a personal probability.
Examples
A forecaster saying there is a '70% chance' this startup survives the year is expressing a subjective probability about a single, non-repeatable event.
A doctor saying there is a 'small chance, maybe one in twenty' that the lump is cancerous is stating a degree of belief about this one body, not a long-run frequency.
If someone says there is a 60% chance their team wins and also a 60% chance it loses, a bookmaker can build a set of bets that takes their money either way.
First described in Ramsey (1926); de Finetti (1937); Savage (1954).