Thinking and Deciding

318 UTILITY MEASUREMENT
we can meaningfully ask how we can minimize this error. We shall thus distinguish
here between judged utility and true utility.
Direct versus indirect judgments
The simplest methods for measuring utility are direct. In these methods, the subject
simply makes a judgment of the utility of one object as a proportion of the utility
of another, or adjusts the end of one interval so that it is equal to another. Note that
such matching responses need not involve hypothetical decisions. Another way to
do this, which we shall examine later, is to make rating responses to stimuli that vary
on several dimensions. For example, subjects can rate the badness of situations involving
combinations of health states (X) and monetary losses (Y). We can compare
the utility of X and Y by asking how much Y is needed to compensate for a given
loss of X, so as to make the average rating the same.
Indirect judgments are based on hypothetical decisions. They thus represent decision
utility (as opposed to predicted utility). For example, in the time-tradeoff
(TTO) method, the subject indicates how many years of future life with normal vision
is just as good as fifty years with blindness. The idea here is to assume that the
subject makes decisions based on utility and to infer the utility from the decision.
If the subject says forty years, and if we assume that utility adds up over time, then
we can conclude that the subject’s utility for blindness is .8, on a scale where death
is 0 and life with normal vision is 1. This is because .8 · 50 = 1 · 40. Or, in the
person-tradeoff (PTO) method, we can ask how many cases of blindness is just as
bad as ten deaths. If death is five times as bad as blindness, and if the judgment is
based on adding utility across people, then the answer should be fifty.
When subjects make indirect judgments, they do it in two ways. In one way,
they make a direct judgment and then infer their indirect judgment from that. For
example, when they are asked (in the person tradeoff method) how many people
losing a finger is equivalent to ten people losing an arm, they may ask themselves,
“How many times worse is it to lose an arm than a finger,” and then they multiply
this by ten. In the other way, they think about the people, not the badness of the
losses. Indirect judgments can be seen as a form of predicted utility, as distinct from
decision utility.
We may summarize the situation as follows:
True Utility ✏✏✏✏✏✏✶
Decision Utility
✻
❄
Judged Utility
Judgment
✏✏✏✏✏✏✶ Here, True Utility represents the actual outcome evaluated in terms of the subject’s
ultimate criteria. Decision Utility represents what is inferred from decisions. Judged