Anchoring and Adjustment Heuristic

Definition

Life requires people to estimate uncertain quantities. How long will it take to complete a term paper? How high will mortgage rates be in five years? What is the probability of a soldier dying in a military intervention overseas? There are many ways to try to answer such questions. One of the most common is to start with a value that seems to be in the right ballpark and then adjust it until a satisfactory estimate is obtained. “My last paper took a week to write, but this one is more demanding so maybe two weeks is a good guess.” “Mortgage rates are low by historic levels, so perhaps they'll be a couple of points higher in five years.” “The fatality rate in the last war was 1.5%, but our enemies are catching up technologically; maybe 4% is a more likely figure in the next conflict.”

Estimates such as these are based on what psychologists call the anchoring and adjustment heuristic. You start with an initial anchor value and then adjust until an acceptable answer is found. The choice of the term anchor for the starting value speaks to one of the most interesting features of this procedure: People typically fail to adjust sufficiently. That is, the initial value exerts some “drag” on the final estimate, systematically biasing the result.

Amos Tversky and Daniel Kahneman, who brought the anchoring and adjustment heuristic to psychologists' attention, provided a clear demonstration of the insufficiency of adjustment. They spun a “wheel of fortune” and asked participants if certain quantities were higher or lower than the number on which the wheel landed. The participants were then asked to estimate the precise value of the quantity in question. For example, some participants were asked whether the percentage of African countries in the United Nations is higher or lower than 10%. Their subsequent average estimate of the actual percentage was 25%. Other participants were initially asked whether the percentage of African countries in the United Nations is higher or lower than 65%. Their average subsequent estimate was 45%. Thus, the initial anchor value, even when its arbitrary nature was quite apparent, had a pronounced effect on final judgments.

In another telling demonstration, Tversky and Kahneman asked people to tell them within five seconds the product of either 1 × 2 × 3 × 4 × 5 × 6 × 7 × 8 or 8 × 7 × 6 × 5 × 4 × 3 × 2 × 1. Because the allotted time was too short to permit an exact calculation, respondents had to estimate. The first group did so by extrapolating from a relatively low number (“one times two is two, two times three is six… so it's probably about…”). The second group started from a larger number (“eight times seven is fifty-six… so…”). Because the two groups of respondents started with different anchor values, they came up with predictably different estimates. The average estimate of the first group was 512, whereas the average estimate of the second group was 2,250. If initial anchor values did not bias final estimates, the average estimates of the two groups would have been the same. Clearly, they were not. Note that the actual answer is 40,320, which shows even more powerfully that both groups adjusted insufficiently.

...