Thinking and Deciding

380 QUANTITATIVE JUDGMENT
Anchoring and underadjustment
We often make numerical judgments by anchoring on some number and then adjusting
for other things that we know. The anchoring and adjustment heuristic affects
quantitative estimates of all sorts. In general, people tend to underadjust. They do
this even when the anchor is totally irrelevant to the judgment. For example, Tversky
and Kahneman (1974) asked subjects to estimate certain proportions, such as the
percentage of African countries in the United Nations. “For each quantity, a number
between 0 and 100 was determined by spinning a wheel of fortune in the subject’s
presence. The subject was instructed to indicate first whether that number was higher
or lower than the value of the quantity, and then to estimate the value of the quantity
by moving upward or downward from the given number” (p. 1128).
The number the subject started with, determined solely by the spin of a wheel,
had a marked effect on the final estimate. Subjects who got high numbers from the
wheel gave higher estimates than those who got low numbers. The adjustment was
insufficient. This effect is found even when the subject needed only to attend to
the anchor (for example, by rewriting it); it does not require an initial judgment of
whether the anchor is too high or too low (Wilson, Houston, Etling, and Brekke,
1996). Of course, when the number is not arbitrary, the effect is even larger. One
practical consequence of this effect is in the domain of liability law. The plaintiff’s
request for damages in a lawsuit influences the jury’s award (Chapman and Bornstein,
1996), even though, in principle, the size of the damage award should depend
only on the facts presented. Apparently, the jury anchors on the request.
There is nothing wrong (in principle) with forming an estimate by starting with
one value and then adjusting it successively as each new piece of information comes
to mind. The mistake that subjects make is not adjusting enough. Underadjustment,
in turn, may result from a failure to look for counterevidence, a failure of actively
open-minded thinking. Chapman and Johnson (1999) asked subjects to think of features
of the target item that were different from the anchor, and this reduced the
underadjustment effect. When the subjects tried to think of features of the target that
were similar to the answer, the effect did not change. This phenomenon supports the
claim I made (in Chapter 3) that we tend to be biased in favor of our present beliefs,
even when these beliefs are induced in an experiment. We do not take sufficient
account of evidence against them. In this kind of experiment, subjects could manifest
this bias by searching their memories for evidence consistent with the initial
estimate, or anchor, even though they know that this estimate is completely arbitrary.
Simple heuristics for judgment
People use a variety of simple heuristics for judgment. These heuristics take less
time than a full MAUT analysis would take, by far. We have evidence for some of
these heuristics. Others have the status of models that fit the data, but possibly only
as approximations, just as linear models fit despite the fact that people probably do
not think like computers following linear models.