02.03.2013 Views

Thinking and Deciding

Thinking and Deciding

Thinking and Deciding

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

THE MEASUREMENT OF UTILITY 317<br />

Many treatments are far more cost-effective than this, <strong>and</strong>, when resources are limited,<br />

we can improve matters on the whole by taking money from the less effective<br />

treatments <strong>and</strong> spending it on the more effective ones. Note, however, that this kind<br />

of argument is useful only when costs are limited <strong>and</strong> can be redirected in this way.<br />

This argument does not tell us how much a QALY is truly worth. From a decisiontheoretic<br />

point of view, that would depend on the alternatives available for spending<br />

the money, <strong>and</strong> the utility that those provide.<br />

We will begin the discussion of utility measurement with the kind of measures<br />

used in health. The basic problem is to elicit judgments about health-related quality<br />

of life, but we will speak of utility. The methods are useful elsewhere, outside of the<br />

health domain.<br />

The measurement of utility<br />

The problem of measuring utility has both a theoretical <strong>and</strong> a practical aspect. The<br />

theoretical problem is to state the conditions under which utility can be measured<br />

using each of the methods that have been devised. What would it mean for each<br />

method to be internally consistent? The practical problem is to state the conditions<br />

under which each method should be used. Which methods are easiest to use? Most<br />

internally consistent? Most sensitive to factors that should affect them <strong>and</strong> most<br />

insensitive to factors that should not affect them? How can we improve each method<br />

in these respects? Before dealing with the various methods for measuring utility, let<br />

us consider some general aspects of the problem, the nature of utility measurement,<br />

<strong>and</strong> the distinction between direct <strong>and</strong> indirect measures.<br />

Utility measurement as prediction<br />

When we try to measure utility in any way, we are, in essence, making a prediction<br />

about the utility of an outcome. We do not know everything relevant that would allow<br />

us to judge the outcome in the light of our goals. Even putting aside the uncertainty<br />

about which variants of an outcome will occur, we might be unaware of the extent to<br />

which factors will affect the achievement of our goals. I have discussed the difference<br />

between predicted utility <strong>and</strong> true utility in section 11.<br />

Utility measurement is not always prediction. Some of our goals do not concern<br />

experiences, or things that happen at a given point in time. (For example, we can<br />

want the house we build to last for a long time after our death, or want our great<br />

gr<strong>and</strong>children to have good lives.) It is still helpful to suppose that utility is real<br />

<strong>and</strong> that we are trying to judge it. We can be more accurate or less accurate in our<br />

judgments, just as we can predict our experiences accurately or inaccurately. The<br />

judgment from which we infer utility is not the same as the utility itself. All along,<br />

we have assumed that the question “which option is best” has a right answer (which<br />

might be that two are equally good). We have thus assumed that utility is real. We can<br />

try to judge what it is, <strong>and</strong> our judgments will be inaccurate to varying degrees, but

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!