The base rate fallacy happens when people ignore how common something is before looking at new evidence. In medical testing, this can make a positive result seem more alarming than it really is. Even a highly accurate test can produce many false positives when the disease is rare.
Understanding base rates helps students interpret test results, screening programs, and risk claims more accurately.
Bayes' theorem gives a clear way to combine the base rate with test accuracy. A positive test result comes from two sources: true positives from people who really have the condition and false positives from people who do not. The probability of having the disease after a positive result depends on both groups, not just on the test's sensitivity.
This is why a test that is 99% accurate can still give a surprisingly low chance of disease when the base rate is very small.
Understanding Statistics: The Base Rate Fallacy
A useful way to reason about evidence is to imagine a real group of people instead of starting with a formula. This method is called using natural frequencies. It turns percentages into counts that can be checked.
First picture everyone who truly has the condition. Next picture everyone who does not. Then apply the test errors to each group separately.
The important comparison comes at the end. Among all people with a positive result, count how many are genuinely affected. This approach prevents the mind from treating every positive result as if it came from the same kind of person.
Consider a condition that affects one person in every thousand. Suppose one hundred thousand people take a test. About one hundred people in this group have the condition.
If the test detects ninety nine out of every hundred affected people, it gives about ninety nine correct positive results. The other ninety nine thousand nine hundred people are unaffected. If the test wrongly gives a positive result to one percent of unaffected people, it produces about nine hundred ninety nine false positives.
There are then about one thousand ninety eight positive results altogether. Only ninety nine belong to people with the condition. A positive result in this case means the person has roughly a one in eleven chance of having the condition, not a near certainty.
The common mistake comes from reversing a conditional statement. A test may be very likely to show positive when disease is present. That does not mean disease is very likely when the test shows positive.
These statements use the same words but describe different groups. One begins with people known to be ill. The other begins with people who received a result.
The second group includes people reached by false alarms. When a condition is uncommon, the healthy group is so large that even a small error rate can create many false alarms. The full group of positive results is the denominator that students must keep in view.
This reasoning matters beyond medicine. Schools may use plagiarism detectors, companies may flag fraudulent payments, and security systems may identify possible threats. In each case, a rare event can be hard to identify accurately because most flagged cases may still be ordinary cases.
The details can change when a test is used on a high risk group rather than the general public. A doctor may test people with symptoms, family history, or a known exposure. Their starting chance is higher, so the same test result carries different meaning.
A second independent test can improve confidence, but only if its errors are not caused by the same problem. Students should write down the population, the starting frequency, the correct detections, and the false alarms before drawing a conclusion.
Key Facts
- Base rate = the prior probability that an event is true before new evidence is considered.
- Sensitivity = P(positive test | disease), the chance a test correctly detects disease.
- Specificity = P(negative test | no disease), the chance a test correctly clears a healthy person.
- False positive rate = 1 - specificity.
- Bayes' theorem: P(disease | positive) = P(positive | disease)P(disease) / P(positive).
- For testing: P(disease | positive) = sensitivity x base rate / [sensitivity x base rate + false positive rate x (1 - base rate)].
Vocabulary
- Base rate
- The overall frequency or probability of a condition in the population before considering a specific test result.
- Prior probability
- The probability assigned to an event before using new evidence or data.
- Sensitivity
- The probability that a test gives a positive result when the person truly has the condition.
- Specificity
- The probability that a test gives a negative result when the person truly does not have the condition.
- Posterior probability
- The updated probability of an event after combining the prior probability with new evidence.
Common Mistakes to Avoid
- Confusing P(positive | disease) with P(disease | positive). These are different conditional probabilities, and Bayes' theorem is needed to convert one into the other.
- Ignoring the number of healthy people in the population. When a disease is rare, even a small false positive rate can create many false positives.
- Calling a 99% accurate test a 99% chance of disease after a positive result. Test accuracy alone does not determine the probability because the base rate also matters.
- Using percentages without converting them consistently. Mixing 1%, 0.01, and 1 in the same calculation can lead to errors by factors of 100.
Practice Questions
- 1 A disease affects 1% of a population. A test has 95% sensitivity and 90% specificity. If a person tests positive, what is P(disease | positive)?
- 2 In a group of 10,000 people, 2% have a condition. A test has 99% sensitivity and a 5% false positive rate. How many true positives and false positives do you expect, and what fraction of positive tests are true positives?
- 3 A rare disease screening test has very high sensitivity but only moderate specificity. Explain why most positive results might still come from healthy people.