Thinking and Deciding

THE MEASUREMENT OF UTILITY 331
When we decide how much money to spend on the prevention of blindness, for
example, it may not be correct to consider the fact that blind people will change their
goals to adapt to their handicap. We may do better to honor the goals of the people
who have not yet become blind, before adaptation. Arguably, these are the people
who matter. We should, however, make sure that, when they evaluate blindness, they
consider the pressure to change their goals as part of what blindness means. This
consideration can make blindness seem less bad, because of the adaptation of goals,
but worse because of the need to change the goals. This need may itself be undesired.
If the question is cure rather than prevention, then the people affected are those
with the condition. If their goals are rationally chosen, then we should honor them.
We may find ourselves placing a higher value on the prevention of a condition than
on its cure, because those with the condition have rationally adapted to it. This need
not be as crazy as it sounds. If, however, the adaptation is irrational, a sour-grapes
effect, then we should ignore it and treat both groups equally, those who do not have
the condition yet and those who have it.
Other methods involving matching and comparison
The time tradeoff and person tradeoff methods ask the subject to equate two situations
so that they match. The subject does this by changing the amount of time or
the number of people in one of the two situations. In the time tradeoff method, we
assume that utility per time is constant, so we assume that the total utility is the product
of the duration times the utility per unit time. Likewise, in the person tradeoff,
we multiply utility per person by the number of people in each situation.
We can do this in other ways, even when the units in the two situations are different,
so long as there are units. We can compare any two dimensions of situations
by matching the quantity of one of them that has has just as much utility as a given
quantity of the other. For example, we could compare two medical treatments of an
infection in terms of their cost and speed of recovery (in days). We could ask how
many days is equivalent to a cost of $50.
We do not need to assume that the utility of days or dollars is constant. We can
specify the numbers involved. Let us, for the moment, assume that the utility of
dollars is linear (within the range of interest) but the utility of days is not. Suppose,
for example, that you are sick and missing work. You can afford a day or two, but
after that things get increasingly difficult. So the utility difference between 2 and 3
days recovery time is greater than that between 1 and 2. The question then might be
to match the following two situations:
A. 2 days recover time, for $50.
B. 1 day recovery time, for $X.
If you said $70, we would infer that the difference between 2 days and 1 day has a
utility equal to that of $20. We could ask the same question using 3 days and 2 days,
and you might give a larger number of dollars, say $80. If we defined the utility of