Base Rate Fallacy

The base rate fallacy occurs when someone evaluates the probability of something by focusing on specific, vivid information while neglecting the base rate — the general frequency of that thing in the overall population. Because individuating details feel more relevant and memorable than dry statistics, people often overestimate how likely an unusual explanation is.

This matters because accurate reasoning requires combining both pieces of evidence: how well the case fits a description and how common that category is to begin with. Ignoring base rates leads to serious errors in medical testing, criminal profiling, risk assessment, and everyday judgments about people.

Example of Base Rate Fallacy

A disease affects 1 in 10,000 people, and a test for it is 99% accurate. Jamal tests positive and concludes he almost certainly has the disease — but because the disease is so rare, the vast majority of positive results are false positives, and his actual chance of being sick is under 1%.

Note
This fallacy is closely related to the representativeness heuristic and the prosecutor's fallacy. It is a reasoning error about probability rather than a flaw in formal logical structure, which is why it is classified as informal.

In more detail

Base Rate Fallacy is a logical fallacy and reasoning error in which a person ignores or underweights general statistical information about how common something is in a population (the “base rate”) and relies instead on specific, vivid, or seemingly diagnostic details about an individual case. Because the overall prevalence of a condition strongly constrains how likely any particular positive result or matching description really is, discarding the base rate can produce conclusions that are confidently stated and badly wrong. The error matters enormously in medicine, criminal justice, security screening, and everyday judgments about people, where it can turn a rare possibility into an apparent certainty.

The fallacy works by substituting representativeness for probability. When a description fits a stereotype or a test result matches a label, the mind treats the fit as if it were the answer, forgetting to ask how many people were in the pool to begin with. Consider a disease that affects 1 in 1,000 people and a test that is 99 percent accurate. Out of 10,000 people, about 10 have the disease and will nearly all test positive, but roughly 100 healthy people will also test positive by error. A positive result therefore means only about a 1-in-11 chance of actually having the disease — not 99 percent. Vivid individual information feels compelling, while abstract background rates feel dry and irrelevant, so intuition quietly drops the number that matters most.

The effect was documented in the 1970s by psychologists Daniel Kahneman and Amos Tversky, whose research on judgment under uncertainty showed that people asked to guess whether a described person was, say, a lawyer or an engineer largely ignored the stated proportions of lawyers and engineers in the sample and judged by how well the description matched a stereotype. A closely related real-world version appears in forensic science as the prosecutor's fallacy: in the English case of R v. Sally Clark, the probability of two infants in one family dying of natural causes was presented to the jury as if it were the probability of the mother’s innocence, while the comparably rare base rate of double infant murder went unaddressed. Clark was convicted in 1999 and her conviction was quashed in 2003, with the Royal Statistical Society publicly criticizing the statistical reasoning used at trial.

To recognize the fallacy, listen for arguments that leap from “this test is very accurate” or “this profile fits perfectly” to “therefore this person almost certainly has the trait,” without any mention of how common the trait is. The countermeasure is to think in natural frequencies: imagine a concrete population of 10,000 and count how many true positives and false positives it would contain. Always ask what the prior probability was before the evidence arrived, and remember that when a condition is rare, even a highly accurate test produces mostly false alarms. Treating evidence as an update to a base rate, rather than a replacement for it, is the heart of sound probabilistic reasoning.