Screen a hundred thousand people for something rare. The test barely errs — yet the pile of positive results fills, overwhelmingly, with the healthy. Drag the base rate below and watch belief re-weigh itself: it is rarity, not inaccuracy, doing the damage.
One hundred thousand people, screened once
Populationyour chance of actually being sick
01 / Why the alarm lies
Sensitivity and specificity describe the test. They say nothing about you until the base rate speaks. When the condition is rare, the vast healthy majority — multiplied by even a one-percent error — produces more false positives than there are true cases to find.
02 / The prior does the work
Bayes' rule folds the base rate p into the verdict. Shift it and the posterior swings from near-certainty to near-nothing without the test changing at all. Evidence updates belief; it does not replace it.
03 / Read the area, not the accuracy
Trust the diagram over the percentage. The blue sliver of the flagged bar is every person the screen was right about; the red is the crowd it frightened for nothing. Rarity sets that ratio — drag the base rate up and watch the blue reclaim the bar.