New York real-estate agent anchoring experiment

The New York real-estate agent anchoring experiment showed that an irrelevant asking price could substantially alter the judgments of experienced professionals evaluating the same house. Agents who believed they were independently assessing market value were nevertheless pulled toward an artificially high or low figure, revealing anchoring as a robust feature of judgment rather than a laboratory curiosity or a weakness confined to inexperienced students.

Experimental Design

The experiment used a house that was actually on the market. Real-estate agents visited the property and examined a comprehensive booklet of information about it, including an asking price. The researchers divided the agents into two groups: one saw an asking price substantially higher than the house’s listed price, while the other saw an asking price substantially lower. The manipulated figure was therefore embedded in an otherwise realistic professional evaluation, where agents had access to the physical property and detailed written information rather than having to make a judgment from a bare number.

Each agent was asked to estimate a reasonable buying price for the house. The agents were also asked to state the lowest price at which they would agree to sell the property if they owned it. These two judgments required them to translate the available evidence into monetary assessments, making the experiment a test of how a numerical starting point enters practical valuation. The asking price was not supposed to determine either answer: it was simply one item in the information environment, and the agents regarded their professional knowledge as sufficient to discount it.

Measured Anchoring Effect

The results showed a pronounced difference between the groups exposed to the high and low asking prices. Agents who saw the higher figure gave higher valuations, while those who saw the lower figure gave lower valuations. The effect was quantified through an anchoring index, calculated as the proportion of the difference between the two groups’ estimates relative to the difference between the two anchors. In the real-estate experiment, the anchoring effect was 41 percent, meaning that a substantial portion of the manipulated price difference carried over into the agents’ judgments.

The finding was especially striking because the agents did not experience the anchor as an obvious command. They did not simply repeat the asking price. Instead, they adjusted from it while believing that their own knowledge, inspection of the house, and professional standards had produced the final estimate. The experiment thus captured the characteristic form of anchoring: an initial number changes the answer without announcing itself as the cause.

A separate demonstration at the San Francisco Exploratorium made the same mechanism easier to measure. Participants first answered whether the tallest redwood was more or less than either 1,200 feet or 180 feet, then estimated its height. The high-anchor group produced a mean estimate of 844 feet, whereas the low-anchor group estimated 282 feet. Although the anchors differed by 1,020 feet, the estimates differed by 562 feet, yielding an anchoring index of 55 percent. The real-estate result, at 41 percent, showed that the effect remained powerful in a consequential professional setting.

Expertise and Denial

The agents were asked which factors had affected their judgments. Remarkably, the asking price did not appear among the factors they named. They took pride in their ability to ignore it and insisted that the listing price had not influenced their responses. Their answers created a sharp contrast between subjective confidence and measured behavior: the experimenters could detect the influence statistically even though the individual professionals could not detect it introspectively.

The comparison with business-school students made the contrast sharper. The students, who had no real-estate experience, produced an anchoring index of 48 percent, almost identical to the agents’ 41 percent. The important difference between the groups was not susceptibility but awareness. The students conceded that the anchor had influenced them; the professionals denied that it had. Expertise therefore did not eliminate the bias, and professional confidence may have made the bias harder to recognize.

This combination of influence and denial illustrates a broader problem in intuitive judgment. People often know the reasons they believe they used, yet those reasons do not necessarily include the forces that actually shaped the judgment. The agents’ inspection of the house and their knowledge of real-estate practice were genuine sources of information, but the experimental manipulation showed that those resources did not fully protect them from an irrelevant numerical starting point.

Mechanism and Significance

Anchoring is presented as an automatic operation of System 1. When a relevant value is available, numerical prediction commonly begins from that anchor and proceeds through adjustment. The adjustment may appear deliberate, but it is usually insufficient, leaving the initial figure embedded in the final estimate. The asking price thus functioned not as a reasoned valuation but as a starting point that quietly organized the agents’ subsequent thinking.

The experiment belongs to a larger research program on heuristics and biases. That program examined how people assign probabilities, forecast the future, assess hypotheses, and estimate frequencies. Its central claim was that intuitive shortcuts are economical and often effective, yet can produce systematic and predictable errors. Anchoring was especially important because it could be measured directly and because its effects appeared outside the laboratory, in decisions involving money and professional expertise.

The result also challenged the expectation that specialized knowledge automatically produces rational independence. Real-estate agents had the relevant experience, saw the property, and studied extensive information, but their estimates still moved with the manipulated asking price. The experiment did not show that expertise is useless; it showed that expertise does not make judgment transparent to the judge or guarantee freedom from cognitive influence. Its enduring significance lies in that uncomfortable gap between what professionals believe guides their decisions and what controlled comparison reveals actually guides them.