222 Understanding Intuitive Decision Making
They told a second group that the individual had been chosen from a set of 30 engineers and 70
lawyers. Both groups were able to determine the correct probability that the person chosen would
be an engineer. The fi rst group estimated a 70% probability and the second group estimated a 30%
probability. Then the researchers told both groups that another person, Jack, was chosen at ran-
dom from the same set of 100 people. In addition, the researchers gave both groups the following
description of Jack:
Jack is a 45-year-old man. He is married and has four children. He is generally conservative,
careful, and ambitious. He shows no interest in political and social issues and spends most of
his free time on his many hobbies, which include home carpentry, sailing, and mathematical
puzzles.
This time, both groups estimated that there was a 90% probability that Jack was an engineer. If
the second group had used the base-rate information of 30%, as Bayes Theorem demands, their
probability estimate should have been much lower. Thus, the second group neglected the base-
rate information and based its estimate solely on the description that seemed representative, or was
stereotypical, of an engineer.
In a follow-up experiment, Kahneman and Tversky told one group that there was a 70% prob-
ability that a randomly chosen person, Dick, was an engineer and told another group that there was
a 30% probability that Dick was an engineer. But in this experiment they gave the two groups a
description that provided no diagnostic information about Dick’s profession:
Dick is a 30-year-old man. He is married with no children. A man of high ability and high
motivation, he promises to be quite successful in his fi eld. He is well liked by his colleagues.
According to Bayes’ theorem the two groups should not modify their original probability estimates
of 70% and 30% since the description was not informative. However, both groups now estimated
a 50% probability that Dick was an engineer. In this study both groups neglected the base-rate
information and based their estimates solely on the extent to which the description seemed rep-
resentative of an engineer. A number of other studies have since confi rmed that audiences tend
to ignore base rates when they receive less valid, but easier-to-comprehend, anecdotal evidence.^150
Audiences are susceptible to base-rate neglect in many different decision-making tasks,^151
especially those involving person perception.^152 For example, a large proportion of actual voting
behavior refl ects how much voters think the candidate “looks like” a competent leader.^153 Audi-
ences making staffi ng decisions are also susceptible to base-rate neglect. A study of performance
appraisal decisions fi nds that appraisers give much more weight to supervisors’ subjective verbal
descriptions of employees’ performance than to objective statistical data about the employees’ per-
formance.^154 In this study, supervisors’ subjective assessments accounted for 68% of the variance in
the performance appraisal ratings.
Investor behavior often refl ects base-rate neglect as well. For example, investors tend to predict
the future value of a company’s stock will go up after they read a description of the company that
“sounds good,” since a good stock price will then seem to be most representative.^155 But if they
read a description that makes the company sound mediocre, a mediocre stock value will appear
most representative, and investors will tend to predict the stock price will drop. In both these cases
investors are predicting future stock values solely on the basis of company descriptions without
questioning the reliability of the evidence or its statistical relevance to future profi t. Likewise,
when making health care decisions, patients do not think about numerical probabilities when esti-
mating risk, but instead focus on descriptive information regarding their physical symptoms and