In short: The base rate fallacy is ignoring how rare an event is when you read a test result. A test that is 99% accurate can still produce mostly false positives if the disease affects only 1% of people, because the few true positives are swamped by false positives drawn from the large healthy majority. Bayes' theorem forces the base rate back into the calculation.
The classic base rate fallacy probability example is the medical test, and JEE Advanced loves to dress it up as a "factory" or "signal detection" problem. The trap is intuitive: you hear "the test is 99% accurate" and conclude a positive result almost certainly means disease. That confuses P(positive | disease) with P(disease | positive) — and it ignores the base rate, the prior probability of the disease. Once you put the base rate back in with Bayes' theorem, a positive result is often far weaker evidence than it feels.
The reliable method
To dodge the base-rate trap, force the prior into your working:
- Write down the base rate. This is the prior P(disease) — how common the condition is before any test.
- List the two conditionals. Sensitivity P(+|disease) and the false-positive rate P(+|no disease).
- Find the total P(+) by total probability. P(+) = P(+|disease)·P(disease) + P(+|no disease)·P(no disease).
- Apply Bayes for the answer you want. P(disease|+) = P(+|disease)·P(disease) / P(+).
- Compare to the base rate. A positive should raise the probability above the prior — but for a rare disease it often stays surprisingly low.
A worked example
A disease affects 1% of a population. A screening test correctly flags 99% of people who have the disease (sensitivity), but also wrongly flags 5% of healthy people (false-positive rate). A randomly chosen person tests positive. Find the probability they actually have the disease.
Let D = "has the disease" and + = "tests positive". The base rate is the key input:
- P(D) = 0.01, so P(no D) = 0.99
- P(+|D) = 0.99
- P(+|no D) = 0.05
Total probability of a positive result:
P(+) = P(+|D)·P(D) + P(+|no D)·P(no D) P(+) = (0.99)(0.01) + (0.05)(0.99) = 0.0099 + 0.0495 = 0.0594
Now Bayes:
P(D|+) = P(+|D)·P(D) / P(+) = 0.0099 / 0.0594 = 1/6 ≈ 0.167
So even after a positive on a "99% accurate" test, the chance of actually having the disease is only about 16.7%. The method holds because the 0.0495 stream of false positives from the large healthy 99% dwarfs the 0.0099 stream of true positives from the tiny diseased 1% — the base rate dominates the denominator.
Common mistakes to avoid
- Reading "99% accurate" as P(disease|positive) = 0.99. Fix: that figure is P(positive|disease); the answer you want is the reverse conditional via Bayes.
- Dropping the base rate entirely. Fix: always multiply each conditional by its prior — the rarity of the disease is doing most of the work.
- Computing P(+) from the diseased group only. Fix: total probability sums positives from both the diseased and the healthy groups.
- Assuming a low false-positive rate is harmless. Fix: even 5% applied to a 99% healthy majority produces more positives than the disease itself.
- Confusing sensitivity with predictive value. Fix: sensitivity is P(+|disease); the predictive value P(disease|+) depends on the base rate too.
Frequently asked questions
Why is a positive test result often a false positive? When a disease is rare, the small group of true positives is outnumbered by false positives drawn from the much larger healthy group. So a positive result can more often come from a healthy person than a sick one.
What is the base rate fallacy in simple terms? It is judging a test result without accounting for how common the underlying condition is. Ignoring that prior probability makes a positive look far more conclusive than it is.
How does Bayes' theorem fix the base rate fallacy? Bayes multiplies each conditional by its prior, so the base rate enters the denominator through total probability. This pulls the posterior P(disease|positive) back down toward reality.
Does a more accurate test remove the problem? It helps, but if the disease stays rare, false positives from the large healthy population can still dominate. Lowering the false-positive rate matters as much as raising sensitivity.
Is the base rate fallacy the same as P(A|B) versus P(B|A)? They are closely linked. The base rate fallacy is what happens when you transpose those conditionals and forget the prior — Bayes corrects both at once.
Practise this
Find out if this trips you up — [take the free diagnostic](/diagnostic), then work through the [free Socratic lesson](/lesson/start) to make Bayesian thinking second nature.