How this instrument works
The false positive paradox is the counterintuitive statistical result that a test can be highly accurate — high sensitivity, high specificity — and still be wrong more often than right when someone tests positive, purely because the condition it's testing for is rare. This calculator computes positive predictive value (PPV): given a positive result, what's the actual probability that the person really has the condition, using Bayes' theorem to combine the test's own accuracy with the condition's prevalence in the population.
The reason lies in simple arithmetic, not in the test being flawed. In a population of 10,000 people at 1% prevalence, only 100 truly have the condition; the other 9,900 don't. Even a 99%-specific test still misclassifies 1% of that huge negative group — 99 false positives — which is nearly as many false alarms as true positives (99, from 99% sensitivity on the 100 real cases). Positive results end up split roughly 50/50 between real cases and false alarms, even though the test itself is 99% accurate on both fronts.
This is also known as the base rate fallacy: the intuitive mistake of judging a positive result mainly by the test's accuracy while ignoring how rare the condition is to begin with. It's one of the most cited illustrations of Bayesian reasoning in statistics and clinical-epidemiology teaching, precisely because the arithmetic is simple but the result routinely surprises even trained readers.
- Enter the test's true positive rate into Test sensitivity (0-1) — the share of actual positive cases it correctly flags.
- Enter the test's true negative rate into Test specificity (0-1) — the share of actual negative cases it correctly clears.
- Enter how common the condition is in the population being tested into Disease prevalence (0-1, prior probability in the population).
- Read Positive predictive value (PPV) — the probability that someone who tested positive actually has the condition.
- Try lowering prevalence toward zero with sensitivity and specificity both held near 1 to see PPV drop sharply — the heart of the paradox.
- All three inputs must combine to allow at least one positive result; if the formula's denominator is zero the instrument reports PPV as undefined rather than force an answer.
Worked example — the classic 99%/99%/1% rare-disease case
Enter Test sensitivity = 0.99, Test specificity = 0.99, and Disease prevalence = 0.01. The numerator is 0.99 × 0.01 = 0.0099. The denominator adds that to the false-positive contribution, (1 − 0.99) × (1 − 0.01) = 0.01 × 0.99 = 0.0099, giving 0.0099 + 0.0099 = 0.0198. Positive predictive value reads exactly 0.5.
In other words: with a test that is 99% accurate in both directions, testing positive still only puts your true odds of having the rare condition at 50/50 — because in a population where only 1% truly have it, the 99% who don't still generate almost as many false alarms (0.99% of that huge group) as the true cases produce correct detections (0.99% of the tiny group). This is the textbook example used across statistics and epidemiology courses precisely because the surprise is so consistent.
Questions
How can a 99%-accurate test still be wrong half the time?
Because '99% accurate' describes the test's behaviour within each group (positives and negatives) separately, not the reliability of a positive result overall. When the condition is rare, the negative group is enormous compared with the positive group, so even a tiny 1% false-positive rate applied to that huge group produces almost as many false alarms as the small positive group produces correct detections. The worked example above shows this hitting exactly 50/50 at 1% prevalence with a 99%/99% test.
What's the difference between this calculator and the post-test-probability instrument?
They compute mathematically equivalent results by two different routes. This calculator plugs population prevalence directly into Bayes' theorem to get PPV. The post-test-probability calculator instead starts from a pre-test probability (which doesn't have to be population prevalence — it can be a clinician's estimate based on symptoms or risk factors) and updates it using the test's likelihood ratio. Set pre-test probability equal to prevalence and specificity equal in both, and the two instruments agree exactly, as this site's test vectors confirm.
Does a low PPV mean the test is bad?
Not necessarily — it means the test alone isn't decisive when the condition is rare, which is a property of the base rate, not a flaw in the test's engineering. Screening programs handle this by using a highly sensitive first test to catch nearly every real case, accepting a lower PPV, then confirming any positive result with a second, more specific test on that smaller, already-enriched group, where PPV rises sharply because prevalence within that group is now much higher.
What happens to PPV as prevalence increases?
PPV rises as prevalence rises, holding sensitivity and specificity fixed — because a larger share of the tested population genuinely has the condition, so a positive result is more likely to be a true positive rather than a false alarm from the (now relatively smaller) negative group. At 50% prevalence with a 90%/90% test, for example, PPV reaches 0.9, much closer to the test's own accuracy figures than the paradox case at 1% prevalence.
Is this calculator meant for real medical diagnosis?
No — it's an educational tool for understanding how prevalence distorts the reliability of a positive test result, a foundational idea in statistics and epidemiology teaching. Real diagnostic decisions depend on validated test characteristics, clinical context, and confirmatory testing, and should always involve a qualified clinician rather than a single probability figure from a calculator.