Base Rates: The Question Your Brain Skips

A base rate is just how common a thing is before you know anything about your specific case. Out of 100 people, how many have the cough, fail the exam, get audited, win the raffle? It's the background frequency — the crowd you start in before any detail narrows you down. Plain, dull, and almost always the most useful number in the room.

Here's the trouble: your brain treats one vivid example as if it were a frequency. You read a story about a young marathon runner who collapsed, and suddenly running feels dangerous. But one story is one story. It tells you the thing is possible, not how often it happens. The base rate is the part nobody puts in the headline, because "this almost never happens to almost anyone" doesn't trend.

Why the test result lies

This is where the machine below earns its keep. A test can be 99% accurate and still hand you mostly false alarms — which sounds impossible until you watch it happen. The reason is the crowd. When a condition is rare, the few true cases get swamped by a tiny error rate spread across a huge healthy group. Drag the slider and watch the false alarms pile up.

Try it. Set the condition to something rare — say, 1 in 100 — and a sharp 99% test, then count the green flags that are actually wrong. In a grid of 100 people, one truly has it; the test catches that one, but also wrongly flags about one healthy person. So a positive result is a coin flip, not a verdict. Most people guess 99%. The real answer is closer to half. That gap is the whole lesson.

Estimating one with no data

Usually you won't have a clean statistic, and that's fine. You can sketch a base rate from a crowd you already picture. Ask: out of 100 people roughly like me, in this situation, how many would this be true for? Ten? One? One in a thousand? You're not chasing precision. You want the right order of magnitude — rare, occasional, or common — because that's what flips a conclusion.

A scary story raises the ceiling of what's possible. The base rate sets the floor of what's likely. Live on the floor.

So make it the first question, not the last. Before "is this happening to me," ask "how often does this happen to anyone." The specifics — your symptom, your bad quarter, your weird email — only matter once you know the crowd they sit in. Start with the boring number. The boring number is usually right.