Psychology

Anchoring Bias: Why the First Number Wins

Tversky and Kahneman (1974, Science) demonstrated that a randomly spun wheel stopping at 10 or 65 shifted participants' estimates of African countries in the UN by 20 percentage points. The first number encountered anchors all subsequent judgments: in salary negotiations, pricing, and project estimates.

6 min read
Quick Answer

What is the anchoring bias?

  • The tendency to rely too heavily on the first piece of numerical information encountered (the "anchor") when making estimates or decisions
  • Documented by Tversky and Kahneman (1974): a randomly spun wheel stopping at 10 or 65 shifted UN country estimates by 20 percentage points
  • The mechanism is insufficient adjustment: people start from the anchor and adjust, but stop too early
  • Persists even when participants are told the anchor is random and offered financial incentives for accuracy

The Wheel-of-Fortune Experiment

Tversky and Kahneman introduced the anchoring heuristic in their landmark 1974 paper “Judgment under Uncertainty: Heuristics and Biases,” published in Science (Vol. 185, No. 4157, pp. 1124–1131). The anchoring experiment is among the most cleanly designed in behavioral research.

45 vs 25

median estimates of the percentage of African countries in the United Nations: participants who saw the wheel stop at 65 gave a median estimate of 45%; those who saw it stop at 10 gave a median of 25%. A 20-point gap driven entirely by an arbitrary number.

Tversky & Kahneman (1974), Science, 185(4157)

Participants were shown a wheel of fortune rigged to stop at either 10 or 65. They were then asked to estimate the percentage of African countries that are members of the United Nations. The wheel spin was explicitly presented as random: participants were told the number was chance-generated and had no bearing on the question. Yet those who saw 65 gave median estimates of 45%, while those who saw 10 gave median estimates of 25%. The anchor shifted the final estimate by 20 percentage points despite being demonstrably irrelevant.

The mechanism Tversky and Kahneman identified is insufficient adjustment: people start from the anchor value and adjust in the direction they believe is correct, but they stop adjusting before they should. The anchor creates a starting point that biases the search for a plausible answer; the adjustment process terminates too early, leaving the final estimate too close to the anchor.

Why Anchors Work Even When You Know

Subsequent research has consistently shown that anchoring effects persist even when participants are explicitly warned about the bias, offered financial incentives for accuracy, or told that the anchor was generated randomly. Strack and Mussweiler (1997) demonstrated that anchors activate associated knowledge: after seeing a high anchor, information consistent with that anchor becomes more accessible, and the estimation process disproportionately draws on that information.

This makes anchoring unusual among cognitive biases: it cannot be eliminated by knowing about it. A negotiator who understands anchoring in principle is still subject to the same directional pull from the first number on the table. The structural response is not awareness but procedure: who presents first, what number appears in written materials, what baseline is established before discussion begins.

Professional Applications

Try alfred_

Try alfred_ free for 30 days

AI-powered leverage for people who bill for their time. Triage email, manage your calendar, and stay on top of everything.

Get started free

Frequently Asked Questions

Does anchoring work even when the anchor is obviously irrelevant?

Yes, and this is one of the most counterintuitive findings. In the Tversky and Kahneman experiment, the anchor was explicitly described as randomly generated by a wheel of fortune, yet it still produced a 20-point shift in estimates. Subsequent replications have tested anchors generated by social security numbers, birthday years, and other clearly unrelated sources, consistently finding that arbitrary numeric exposure shifts numerical judgments in the direction of the anchor. The anchor's irrelevance does not eliminate the effect; it merely reduces its magnitude slightly in some conditions.

How strong is anchoring in salary negotiations specifically?

Research consistently shows that the first offer in salary negotiations predicts final outcomes more strongly than almost any other single factor. Studies by Galinsky and Mussweiler found that first offers explained a substantial portion of variance in final negotiated outcomes. Counter-anchoring, which means aggressively naming a number far from the anchor in the other direction, reduces the effect, but even with a strong counter-anchor, the initial anchor still exerts measurable pull. The practical implication: in salary negotiations, the party who names a number first sets the range within which negotiation occurs.

Can you train yourself to be less susceptible to anchoring?

Training specifically focused on generating independent estimates before seeing any anchor, and on considering extreme counterarguments rather than simply adjusting away from the anchor, shows some effect in research settings. Galinsky and Mussweiler (2001) found that 'consider the opposite' instructions, where participants were asked to think about reasons why the anchor value was incorrect, reduced (though did not eliminate) anchoring effects. The key insight is that simply being told about anchoring produces little benefit; the corrective must be structural: generate your own estimate before seeing the anchor, or explicitly construct a counterargument before adjusting.