Wheel of Fortune African-country anchoring experiment

The Wheel of Fortune African-country anchoring experiment is a landmark demonstration of the anchoring effect: people’s numerical judgments can be pulled toward an arbitrary value even when that value contains no useful information. By making the anchor visibly absurd—a wheel that could stop only at 10 or 65—Daniel Kahneman and Amos Tversky exposed how readily judgment follows a number that reason should have discarded.

Design and Participants

Daniel Kahneman and Amos Tversky recruited students at the University of Oregon for an experiment built around a rigged wheel of fortune. The wheel was marked from 0 to 100, but its mechanism had been arranged so that it would stop only at 10 or 65. One researcher stood before a small group, spun the wheel, and asked the participants to record the number at which it stopped. The procedure therefore supplied two groups with different numerical starting points while giving neither group any relevant evidence about the question that followed.

The researchers then asked the students whether the percentage of African nations among United Nations members was larger or smaller than the number they had just written down. After that comparison question, each participant was asked for the best estimate of the actual percentage. The experiment thus separated the arbitrary number from the estimation task only formally: psychologically, the number had already entered the participants’ thinking.

The Numerical Shift

The results showed a striking divergence. Participants who had seen the wheel stop at 10 gave an average estimate of 25 percent, whereas those who had seen it stop at 65 gave an average estimate of 45 percent. The wheel had provided no information about the composition of the United Nations, and its restricted outcomes made its irrelevance especially plain. Yet the two groups’ judgments remained substantially closer to their different starting numbers.

The comparison question helped create the effect by requiring participants to consider the anchor before producing their own estimate. Once 10 or 65 had become part of the mental frame, the final judgment was not generated afresh. Instead, the estimate moved away from the anchor but did not move far enough to become independent of it. The difference between the groups therefore measured the influence of the number itself rather than a difference in knowledge about African nations.

Significance for Judgment

The experiment gave the anchoring effect its clearest absurdity. People had not merely been influenced by a plausible estimate, an expert forecast, or a relevant piece of evidence; they had been influenced by the outcome of a wheel that could not possibly answer the question. Kahneman and Tversky published the finding in their Science paper, where it became one of the best-known results they reported.

Anchoring was presented as a general feature of estimation. A person asked whether Gandhi was more than 114 years old at death will usually give a higher estimate than a person asked whether he was more than 35. A house can seem more valuable when its asking price is high than when the same house is listed at a low price, even when buyers intend to resist the influence. Any number introduced as a possible answer to an estimation problem can become a psychological starting point.

The experiment also connects numerical judgment with Bayesian discipline. In a well-structured inference, prior beliefs should be combined with the diagnostic value of evidence. An arbitrary wheel outcome has neither the status of a meaningful base rate nor the diagnosticity of relevant evidence, so a disciplined reasoner should ignore it. The students did not, illustrating how intuitive coherence can overwhelm the logical demand to exclude irrelevant information.

Competing Mechanisms and Wider Consequences

Daniel Kahneman and Amos Tversky initially disagreed about the psychological mechanism behind anchoring. One favored an interpretation based on deliberate adjustment; the other preferred a different account, and they never settled the disagreement during their collaboration. Later research showed that both interpretations captured genuine processes. One form of anchoring arises when System 2 begins from an initial value and adjusts away from it, with the adjustment remaining insufficient. Another arises through priming, an automatic manifestation of System 1 that makes the considered number influence subsequent thought.

The same logic appears in the assessment of subjective probability distributions. When experts select high and low percentile values for a quantity such as the Dow Jones average, they often begin with a best estimate and adjust upward or downward. If those adjustments are too small, the resulting confidence interval becomes too narrow. Empirical studies found that actual values fell outside such stated limits far more often than proper calibration would allow, often for about 30 percent of assessed problems rather than the expected two percent outside a 98 percent interval. The wheel experiment therefore stands not as an isolated curiosity but as a vivid model of a broader failure: judgments retain traces of starting points that should have no authority.

The experiment’s lasting importance lies in the mismatch between the objective emptiness of the anchor and the subjective persistence of its influence. It transformed an everyday susceptibility into an experimentally measurable phenomenon, showed that numerical judgment can be shifted without informative evidence, and helped establish anchoring as a central case in the study of human reasoning.