Anchoring Bias: The First Number You Hear Hijacks Your Judgment

The Wheel of Fortune

Tversky and Kahneman introduced anchoring in 1974 as part of their influential paper on heuristics and biases. In a demonstration that became a classic, they spun a wheel of fortune in front of subjects that was rigged to stop at either 10 or 65. Participants then estimated what percentage of African countries were members of the United Nations. Those who saw the wheel land on 65 gave estimates that were systematically higher than those who saw it land on 10 — even though everyone could see the wheel was random and clearly had no logical relationship to UN membership.

The Gandhi experiment extended this finding to the domain of factual estimation with obviously absurd anchors. Asking if Gandhi died before or after age 9 pulled estimates lower than asking about age 140, even though both anchors were transparently impossible. The irrationality of the anchor did not protect against its influence. The mind, having activated the number, used it as a starting point regardless.

Why Anchors Work

Anchoring operates through at least two mechanisms. The first is an adjustment process: given a starting number, people adjust from it toward a plausible value, but typically fail to adjust enough. The adjustment stops when the person reaches a value that feels "good enough," which tends to be closer to the anchor than a full adjustment would require. The second mechanism is confirmatory hypothesis testing: the anchor activates a hypothesis and the person searches for evidence consistent with it, generating a biased body of supporting evidence that pulls the estimate toward the anchor.

Both mechanisms operate unconsciously and cannot be overridden by knowing about them. In one study, participants were told explicitly that an anchor was generated randomly and had no relevance to their estimate. They acknowledged this verbally. Their estimates were still anchored. Instructions to "try not to be influenced" reduce anchoring effects modestly; they do not eliminate them.

Real-World Consequences

Anchoring effects are substantial in high-stakes domains. In salary negotiations, whoever names a number first anchors the subsequent discussion — research shows that final settlements are closer to the first offer made than to any objective value of the position. In legal sentencing, studies have shown that requesting a longer prison term leads to longer sentences, even from experienced judges who are instructed to set the request aside.

Real estate agents and car salespeople use anchoring deliberately: the sticker price or initial asking price sets a reference point that makes subsequent concessions feel like gains, regardless of the gap between the anchor and actual value. Retailers use similar tactics with "original price" tags. The anchor does not have to be reasonable to work — it merely has to exist and be encountered first.