Esprit critique
← All biases
bias · Probabilities

Base rate neglect

Base rate fallacy, Négligence du taux de base

Ignoring how common a phenomenon is to begin with (its prior probability) in favour of a more striking piece of specific information.

Mechanism

Specific information (a testimony, a test result, a description) feels more relevant than an abstract statistic. Yet the correct probability combines the two, according to Bayes’ formula.

In practice

Example

A disease affects 1 person in 1,000. A test always detects it, but also comes back positive for 5% of healthy people. Yet a positive result means only about a 2% chance of being ill: out of 1,000 people tested, about 1 true positive for 50 false positives.

Famous case

Casscells, Schoenberger and Graboys (1978) put a question to 60 doctors and students at Harvard Medical School: a disease with a prevalence of 1 in 1,000, a test with 5% false positives; what is the probability that a patient who tests positive is ill? Nearly half answer 95%. Only about 18% give the right answer, about 2%.

Defending yourself

How to protect yourself

Reason in natural frequencies (“out of 1,000 cases…”) rather than percentages. Always ask: what is the starting proportion?

Don’t confuse

Prosecutor’s fallacy

The prosecutor’s fallacy is a special case of base rate neglect: it inverts a conditional probability, taking P(evidence | innocent) for P(innocent | evidence).

Going further

Source

Kahneman and Tversky (1973); the taxi problem (Tversky and Kahneman, 1982). 85% of taxis are green and 15% blue. A witness who is 80% reliable says they saw a blue taxi. The typical answer is 80%; the correct answer is about 41%.

Practise

In the app, each entry offers ways to practise recognising the concept. No sign-up: your progress stays on your device.

Open the interactive entry