liveBook · Manning

Prior probability

As explained in Timeless Algorithms, The Foundational Ideas

A prior distribution represents uncertainty about an unknown quantity before incorporating observed data.

A positive diagnostic test is evidence of disease, but its meaning also depends on how likely the disease was before the test. That starting probability is the prior probability. More generally, a prior distribution represents uncertainty about an unknown quantity before new data are incorporated; it can reflect existing knowledge, experience, or theory. section 2.2.2 section 2.2.1 section 3.2.4

For a hypothesis H and observed data D, Bayes’ theorem gives P(H | D) = P(D | H) × P(H) / P(D). The prior P(H) describes the hypothesis before the new data; the likelihood P(D | H) measures how well it explains those data. The evidence P(D) accounts for the data across competing hypotheses and normalizes the result. The posterior P(H | D) is the revised probability. section 2.2.1 section 2.2

A rare disease and a positive test

Suppose disease prevalence is 0.001, test sensitivity is 0.99, and the false-positive rate is 0.01. The prior probability of no disease is 0.999. A positive result can arise either from disease or from a false positive, so P(positive) = 0.99 × 0.001 + 0.01 × 0.999 = 0.01098. The posterior probability of disease is 0.00099 / 0.01098, approximately 9%. The result raises the probability substantially, but it is not a definitive diagnosis. section 2.2.2 section 10.2.1

The prior matters because evidence is interpreted against what was plausible beforehand. In diagnosis, prevalence, seasonal trends, and clinical experience can inform initial beliefs; symptoms and tests can then revise them. Priors also introduce assumptions that can affect estimates. Fisher objected to assigning distributions to fixed but unknown parameters, arguing that inference should instead rest on observed data and the assumed probability model. section 2.1 section 3.2.4

How Timeless Algorithms, The Foundational Ideas covers it

How deep it goes
Defines the prior within Bayes’ theorem and works through a numerical update, while noting an objection to prior distributions.
Examples
Illustrates the prior through disease prevalence and a positive test for a rare condition.
What you can do after
Evaluate/understand how a low prior probability affects the interpretation of a positive diagnostic test.
How it connects
Connects prior probability to likelihood, posterior probability, marginal probability, and maximum likelihood estimation.

Also explained in

Compare how 12 books explain prior probability →

Related concepts

This explainer was generated by Manning's knowledge graph from the text of Timeless Algorithms, The Foundational Ideas. The section links point to the parts of the book it draws on.