Thinking and Deciding

HEURISTICS AND BIASES IN PROBABILITY 147
Tom W.
In perhaps the clearest example of the representativeness heuristic, Kahneman and
Tversky (1973) asked one group of subjects to estimate the proportion of first-year
graduate students in the United States in nine different fields (business administration,
computer science, engineering, humanities and education, law, library science,
medicine, physical and life sciences, and social science and social work). Subjects
judged computer science to be the second smallest (7% of all graduate students) of
the nine fields, with social science and social work the second largest (17%). (These
judgments were realistic at the time the study was done, around 1970.) These proportions
constitute the prior probabilities, which are often called the base rates, before
any additional evidence about a particular graduate student is provided.
Another group of subjects was given the following personality sketch of a firstyear
graduate student (1973, p. 238):
Tom W. is of high intelligence, although lacking in true creativity. He
has a need for order and clarity, and for neat and tidy systems in which
every detail finds its appropriate place. His writing is rather dull and mechanical,
occasionally enlivened by somewhat corny puns and by flashes
of imagination of the sci-fi type. He has a strong drive for competence.
He seems to have little feel and little sympathy for other people and
does not enjoy interacting with others. Self-centered, he nonetheless
has a deep moral sense.
The subjects were asked to judge how similar Tom W. was to the typical graduate
student in each of the nine fields. The subjects considered Tom to be most similar
to computer-science students and least similar to social-science and social work students.
These similarity ratings correspond to the representativeness of the sketch of
each of the categories.
A third group was told that the sketch of Tom had been written by a psychologist,
on the basis of projective tests, when Tom was a senior in high school. This group of
subjects was asked to rank the nine different fields according to the probabilities of
Tom’s being in them. The rankings matched almost perfectly the rankings given by
the similarity group. Tom was considered to be most likely to be studying computer
science and least likely to be studying social science or social work. The probability
ratings were not related at all to the prior probabilities. When subjects are asked to
make a probability judgment, they apparently base it entirely on their beliefs about
similarity, or representativeness, and not at all on their beliefs about prior probability.
(It can be assumed that this group of subjects would have given the same prior probability
and similarity ratings as the other groups; all groups were drawn from the same
population of respondents to an advertisement in a college newspaper.) This neglect
of prior probability leads to systematic errors in ranking the fields in probability, that
is, in saying which of two fields was more likely. The error is therefore not just a
matter of not assigning the right numbers to beliefs.