What it is
Base rate neglect is the systematic tendency to underweight or ignore background frequencies — the statistical rate at which events occur in the relevant population — when evaluating specific cases. When vivid, detailed, individuating information is available, people treat it as the primary input to their judgment and push the base rate into the background or ignore it entirely.
The base rate is the answer to the question: before you know anything specific about this case, how often does the outcome you're assessing actually occur? It is the starting point for any honest probability estimate.
The classic example
Kahneman and Tversky ran studies where subjects were told that a group of 100 people was composed of 70 engineers and 30 lawyers (or vice versa). Subjects were then given a personality description: "Jack is a 45-year-old man. He is generally conservative, careful, and ambitious. He shows no interest in political and social issues and spends most of his free time on his many hobbies, which include home carpentry, sailing, and mathematical puzzles."
When the description matched stereotypes of an engineer, subjects estimated Jack was highly likely to be an engineer — roughly equally regardless of whether the group was 70% engineers or 70% lawyers. The base rate (70 vs. 30) barely moved their estimates, even though it should have been the dominant factor.
The Linda problem is even sharper. Linda is described as a philosophy graduate, deeply concerned with social justice. Subjects consistently judge "Linda is a bank teller who is active in the feminist movement" as more probable than "Linda is a bank teller" — which is logically impossible, since the second is a subset of the first. The specific description overwhelms the base rate arithmetic entirely.
In markets and analysis
Venture capital is a domain where base rate neglect is endemic and expensive. The base rate for startup success is brutal: the vast majority of funded startups fail to return capital. Yet investors repeatedly engage with specific, compelling pitches and push the base rate out of mind.
A persuasive founder, a compelling demo, a large-sounding market — none of these change the base rate. They might justify shifting your probability estimate upward from the base rate, but that shift must be proportional to how strongly the specific evidence distinguishes this case from others. The evidence that a charismatic founder delivers a great pitch does not tell you much, because charismatic founders also drive most of the failed companies.
The same problem appears in credit analysis (ignoring the base rate of corporate defaults for companies in a given sector), in HR (ignoring the base rate of success for candidates from a particular school), and in trading (ignoring the base rate of profitable trades with a given pattern).
The Bayesian corrective
Bayesian reasoning structures the problem correctly. Start with the prior probability — the base rate. Then update based on the likelihood ratio: how much more likely is this specific evidence under the hypothesis you're considering versus under the alternative?
If the base rate for startup success is 10% and a founder's charisma is equally common among successful and unsuccessful founders (likelihood ratio ≈ 1), then the pitch doesn't change the 10% much. Only evidence that distinguishes winners from losers in the population should move the needle.
The practical discipline: before evaluating any specific case, write down the base rate explicitly. Then ask what specific evidence would cause a rational updating, and by how much. The typical finding is that base rates should anchor much more heavily than they do intuitively.
Where it shows up
In forecasting: political analysts who spend weeks reading detailed narratives about an election and ignore the historical base rate for incumbent parties. In medicine: doctors who ignore the base rate for rare diseases when ordering tests. In hiring: interviewers who ignore the base rate for job performance from a given profile in favor of a strong interview.
In each domain, the pattern is the same. Specific, narrative, emotionally engaging information feels more informative than it is. Background statistics feel less informative than they are. Closing that gap is the core of calibrated thinking.
One thing most people get wrong
Base rate neglect does not improve when people are experts. Kahneman's research showed that physicians, statisticians, and experienced investors all neglect base rates in proportion to the vividness of the specific case. Expertise helps with the technical content of the decision but does not automatically produce correct base rate weighting. The discipline must be deliberately applied — the cognitive instinct runs in the other direction.