Thinking and Deciding

THE PSYCHOLOGY OF HYPOTHESIS TESTING 173
5 could support the hypothesis, if the answer were no, since the hypothesis predicts
a negative answer for this item. Certainly the subject expects her hypothesis to be
supported, but this is true no matter what kind of sequence she gives.
In sum, if subjects have a problem here, it is not in “trying to confirm.” They
may have no problem at all. Subjects may well be trying to test alternative hypotheses
(such as ascending even numbers less than 100). The hypotheses they have in
mind may seem unlikely to us, but nothing in the experiment tells the subject what
probabilities to assign to various alternative hypotheses.
What subjects do in these studies is to provide tests that are congruent with their
favored hypothesis. As noted, these studies have not shown that this violates any
normative model. But perhaps it results from the use of a heuristic that might lead
to such violations. We might call this a congruence heuristic: “To test a hypothesis,
think of a result that would be found if the hypothesis were true and then look for that
result (and do not worry about other hypotheses that might yield the same result).”
Baron, Beattie, and Hershey (1988) found that subjects did indeed seem to use
a “congruence heuristic” in which they favored tests or questions that gave a yes
answer if their favored hypothesis was true, even if this question was nonnormative.
Here is one of their test questions:
A patient has a .8 probability of having Chamber-of-Commerce disease
and a .2 probability of Elks disease. (He surely has one or the other.)
A tetherscopic examination yields a positive result in 90% of patients
with Chamber-of-Commerce disease and in 20% of patients without it
(including those with some other disease). An intraocular smear yields a
positive result in 90% of patients with Elks disease and in 10% of those
without it. If you could do only one of these tests, which would it be?
Why?
The “tetherscopic” examination yields a yes answer if the favored hypothesis (“Chamber-of-Commerce
disease”) is true. Many subjects chose the tetherscopic examination
for this reason, even though the “intraocular smear” is a better test in this case.
(To see why it is better, note that one could interpret a negative result on the intraocular
smear as a positive indication of Chamber-of-Commerce disease; looking
at it this way, the test has a 10% false-positive rate, instead of the 20% rate for the
tetherscopic examination.)
In another procedure, subjects were asked to imagine themselves in everyday situations,
in which they could ask yes-no questions to test a hypothesis. They were
asked to evaluate each question (with a numerical rating) as a test. They were later
asked to give their own probability judgments for the hypothesis and for a yes answer
to each question, first assuming the hypothesis to be true, and then assuming
it to be false. In the following typical example, a subject gave the highest rating to
the question that yielded a yes answer if the hypothesis was true, question 2, even
though question 1 would be more informative, as determined from the subject’s own
probabilities: