Cognitive bias
Last reviewed:
Part of:Thinking clearly about risk →
All fallacies and biasesCognitive biases list
Base-rate neglect grabs a vivid likelihood and ignores how common the thing was to begin with. A test can be quite accurate and still be wrong most of the time it raises an alarm, if the condition is rare.
Base-rate neglect is filed here as a bias. If you cannot point to that move in the remark in front of you, you do not yet have this bias. Disliking the conclusion is not enough, and a true premise does not repair a broken step.
This page adds no new definition and no statistic. The three scenes below only move the library’s example into a meeting, a news sentence, and a social feed. None of them is a reported case, and none of them carries a number.
The library’s example is the anchor, quoted so the later scenes cannot drift: “A test is 95% accurate and flags a rare condition that affects 1 in 1,000. Most of the alarms will still be false, because the 5% error falls on a much larger healthy group.” The three that follow change the room and keep the step.
Picture that example in a meeting, aimed at someone who has to answer before the next item. If the room cannot give the reply on the card, it has not answered the point. It has used Base-rate neglect. No survey of offices is claimed. The test is whether the step matches the card.
The same step can sit in a speech or a headline, where the pause is shorter and the audience is larger. Imagine the stored example said into a microphone and left unpacked. This is not a claim about a particular broadcast or a particular paper. The correction is still the reply on the card, not a different theory of the press.
Imagine the example as a caption. Agreement is not evidence the step was valid, and outrage is not evidence a different bias was in play. Match the step to the definition before you name it.
Ask how common the thing is before the new clue, then how the clue changes that. A percentage about the test is not yet a percentage about the person.
“That is Base-rate neglect” is not yet an answer. Name the step, then ask the question the card tells you to ask. If the remark survives, drop the label. If it fails, you still owe a reason for what you think instead. The reply does not hand you a new conclusion.
Base-rate neglect is easy to shout in place of a neighbour. The nearest card people mix this up with is Confirmation bias. Use the comparison linked under this section when you are unsure which name fits. The comparison quotes the cards. It does not invent a third definition.
Write down the sentence you think is the mistake. If it matches one definition, use that name. If it matches neither, do not force a label. If it matches both, answer the two steps separately.
A 95% accurate test flags a condition that affects 1 in 1,000. Why might most alarms still be false?
The options and the explanation open in the quiz. Scoring stays on this device. There is no pass mark on this card.
Base-rate neglect grabs a vivid likelihood and ignores how common the thing was to begin with. A test can be quite accurate and still be wrong most of the time it raises an alarm, if the condition is rare.
A test is 95% accurate and flags a rare condition that affects 1 in 1,000. Most of the alarms will still be false, because the 5% error falls on a much larger healthy group.
Ask how common the thing is before the new clue, then how the clue changes that. A percentage about the test is not yet a percentage about the person.