Test Accuracy
A test can be very accurate and still be wrong most of the time it says "positive". Whether it is depends on how common the disease is. Below, 1,000 people take the same test.
The situation illustrative
Share of the people tested who really have the disease.
Of the sick, the share the test catches.
Of the healthy, the share the test correctly clears.
1,000 people tested
Positive predictive value as the disease gets more common same test
Sensitivity
Out of everyone who has the disease, how many test positive. High sensitivity means few missed cases, which matters most when missing one is dangerous.
Specificity
Out of everyone who is healthy, how many test negative. Even 95% specificity means 5 false alarms for every 100 healthy people, and in a big screening programme most people are healthy.
The base rate
When a disease is rare, false alarms from the many healthy people can outnumber the true cases. That is why doctors confirm a positive screening result with a second test.
The scenarios are illustrative numbers, not figures for any real test. The idea of showing results as counts of people ("natural frequencies") comes from research by Gerd Gigerenzer on how doctors and patients read test results. Educational only; not medical advice.