Start from the base rate

Before judging one case, ask how often cases like it turn out each way. Start from that, then adjust for what this case shows.

Daniel Kahneman and Amos Tversky showed in 1973 that people judge how likely something is by how well a case fits a picture, and largely ignore how common each possibility is. Given a description that sounds like an engineer, people said engineer whether they were told the group held 30% or 70% engineers.

The fix is to take the outside view first. Kahneman and Dan Lovallo (1993) describe it: find a reference class of similar cases, see how they turned out, and start there. Only then adjust for what you know about this case. Roger Buehler and colleagues (1994) found students took 55 days on average to finish their theses, against a best guess of 34; fewer than half finished even by their worst-case date. The outside view would have warned them.

Counting helps. Rather than percentages, imagine 100 or 1,000 cases and count how many end each way. The answer is often far from what the evidence alone suggests.

Why it matters

Vivid evidence about one case crowds out how common things are. People often neglect base rates even when they are told them, especially as percentages rather than counts, and so overrate rare explanations.

Away from the game

The planning fallacy: asked how long their own project will take, people picture it going to plan. Ask first how long similar projects took, the outside view, then adjust for what really differs about yours.