the ledger · draw · EP-037

Base Rate Explained: What a Positive Test Really Means

A test that's 95% accurate comes back positive — and your real chance is about two percent. That isn't a trick; it's the base rate, the single most protective idea in statistics, and this episode works the arithmetic by hand until you can do it yourself.

Airs Thursday 14 January 2027 on the channel. The video appears here the day it is published.

A test that's 95% accurate comes back positive — and your real chance is about two percent. That isn't a trick; it's the base rate, the single most protective idea in statistics, and this episode works the arithmetic by hand until you can do it yourself. You'll also learn the question to ask when something "doubles your risk" — doubles it from what? — and why you fear sharks when the cows are the ones keeping count.

The arithmetic here is a worked hypothetical, not a reading of anyone's own result — the numbers change with the condition and the test. A real test result belongs with the clinician who ordered it.

Drawn from: plain probability arithmetic, worked through in the episode (1-in-1000 prevalence, 95%-accurate test → roughly 50 false alarms per true positive → about a 2% chance); relative-versus-absolute-risk arithmetic (1 in 10,000 doubled is 2 in 10,000); CDC injury surveillance on cattle-caused deaths (a 2009 CDC report counted 108 cattle-involved deaths among US agricultural fatalities over 2003–2007 — about twenty-two a year) against the International Shark Attack File's US average of about one fatality a year.

General education, not advice. Sources are named on screen and in the description; where a guideline body is named, its current guidance governs. The disclaimer.

as spoken

The words, in order.

  1. A thousand stones. One is gold.

    Before dawn, a thousand stones lie on the riverbed, grey and quiet, and one of them is gold. Hold that picture. One in a thousand. Everything that follows depends on it.

  2. How common, before anything else.

    Today you draw from the well a tool with an unlovely name: the base rate. It means one thing only — how common something is before you know anything else about it.

  3. The story arrives. The count slips away.

    Your mind was not built to hold that number. It was built to be seized by the story in front of it — the test result, the headline, the shadow in the water. The count of the riverbed slips away the moment the story arrives.

  4. One in a thousand. A good test.

    So let us work one through, slowly, with real numbers. A condition touches one person in a thousand. There is a test for it, and the test is good — accurate ninety-five times in a hundred. You take it. It comes back positive.

  5. Instinct says ninety-five. Instinct is wrong.

    Your instinct says: ninety-five percent accurate, so I am ninety-five percent doomed. Your instinct is wrong, and not by a little. Watch.

  6. One true alarm in the field.

    Put a thousand people in a field. One of them truly has the condition, and the test, being good, will very likely find that one. So far, one true alarm.

  7. And about fifty false ones.

    But nine hundred ninety-nine of them are healthy, and a test that errs five times in a hundred will wrongly flag five of every hundred healthy people. Five out of every hundred, across nearly a thousand — that is about fifty false alarms.

  8. One in fifty. Not ninety-five.

    So the field holds roughly fifty-one people staring at a positive result, and only one of them is truly ill. Your positive test means about one chance in fifty. Around two percent. Not ninety-five.

  9. Even a good search turns up mostly fool's gold.

    The test did not fail. The riverbed did the arithmetic. When a thing is rare enough, even a good search turns up mostly fool's gold.

  10. Doubles it — from what?

    Now the second tool, and it guards the same gate. When someone tells you a thing doubles your risk, you must ask: doubles it from what?

  11. From one to two — in ten thousand.

    Suppose a misfortune strikes one person in ten thousand, and some habit doubles that. The headline says one hundred percent increase, and it is telling the truth. The truth is that your risk went from one in ten thousand to two in ten thousand.

  12. Absolute risk is your actual life.

    Nine thousand nine hundred ninety-nine out of that ten thousand are untouched before the habit, and nine thousand nine hundred ninety-eight after. The doubling of almost nothing is almost nothing. That is the difference between relative risk, which is a ratio, and absolute risk, which is your actual life.

  13. Headlines print the bigger number.

    Headlines live on relative risk for a simple reason: it is always the bigger number. "Doubles your risk" fills a page. "From one in ten thousand to two" would never be printed, and it is the honest sentence.

  14. Notice which number you were handed.

    Neither number is a lie. But only one of them was chosen to move you, and you should notice which one you were handed.

  15. Cows, twenty. Sharks, one.

    One more stone from the riverbed. In America, cows kill twenty or more people a year. Sharks kill about one. Yet no one has ever made a summer film about a pasture.

  16. Fear follows teeth, not the ledger.

    Your fear follows the vivid, not the common. It answers to teeth and dark water, not to the slow, dull, ordinary things that actually fill the ledger — the stairs, the road, the habit repeated daily for thirty years.

  17. The seller brings the shark.

    This is not a flaw to be ashamed of. It is old equipment doing its old job. But the world now sells to that equipment on purpose, and the seller always brings the shark, never the cow.

  18. Two questions before the door opens.

    So carry two questions with you from this well, and ask them in order, before you let any frightening number in the door.

  19. First: how common to begin with?

    First: how common is this to begin with? Find the riverbed. Find the one-in-how-many before the story started.

  20. Second: ratio, or actual people?

    Second: is that number relative or absolute? A ratio, or a count of actual people? If the answer is a ratio, go and find the count.

  21. Ask both questions. Then feel.

    Here is your bucket to carry home. The next headline that frightens you — this week, there will be one — stop before you feel anything, and ask it those two questions out loud. How common to begin with. Relative or absolute. Then decide what you feel.

read alongside it

From the shelf.

As an Amazon Associate I earn from qualifying purchases. The commission costs you nothing and bought none of these; the whole shelf is on the reading page, and how that works is on the about page.

  1. Naked Statistics — Charles Wheelan

  2. Thinking, Fast and Slow — Daniel Kahneman

keep drawing

More from The Ledger.

drink deep

Here is the map.

The Codex of Nine is a free twelve-page primer: the well beneath the tree, then one page for each domain — the question it answers, three truths already drawn, and one thing to do this week.

After that, one note per episode — the video and its sources — twice a week and nothing else. Unsubscribe in one click. Privacy.