The Anchoring Effect

Definition

Anchoring occurs "when people consider a particular value for an unknown quantity before estimating that quantity" and their final estimate stays implausibly close to that starting value. Kahneman and Tversky found the effect held even when the anchor was transparently random and useless as information — "the list of anchoring effects is endless." Later research resolved a long-running disagreement between the two of them by showing there are actually two distinct mechanisms: anchoring as deliberate (but insufficient) adjustment, a System 2 process, and anchoring as an automatic priming effect, a System 1 process.

In the Book

The chapter's central experiment: Kahneman and Tversky rigged a wheel of fortune to stop only at 10 or 65, spun it in front of University of Oregon students, and asked them first whether the percentage of African nations in the UN was larger or smaller than the number they'd just seen, then to give their best estimate. Average estimates were 25% for those who saw 10 and 45% for those who saw 65 — a huge, absurd effect from a number everyone knew was meaningless. He also gives the asking-price-on-a-house example (the same house seems more valuable with a high listing price) and the age-of-Gandhi-at-death example. For the "anchoring as adjustment" mechanism, he cites Eldar Shafir, Tom Gilovich, and Robyn Le Boeuf's line-drawing experiments (people undershoot when adjusting up from a low starting point, overshoot adjusting down from a high one) and applies it to real behavior like driving too fast after leaving a highway.

Why It Matters

Anchoring shows that a judgment can be moved by information the judger consciously knows is irrelevant — knowing about the bias does not immunize you against it, because it partly operates automatically (via priming) rather than through reasoning that could be corrected. This is directly actionable wherever a number is presented before an estimate is requested: negotiators open with extreme first offers, retailers list a high "was" price next to a sale price, and estimators should generate their own reasoning before being shown someone else's figure, precisely because "any number you are asked to consider as a possible solution to an estimation problem will induce an anchoring effect."