02.03.2013 Views

Thinking and Deciding

Thinking and Deciding

Thinking and Deciding

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

MAUT AS A TYPE OF DECISION ANALYSIS 353<br />

how such an analysis would work. You might try a holistic estimate first, in which<br />

you simply give your opinion. Then use your own best guesses for the effect of<br />

different speed limits on the various dimensions, <strong>and</strong> give your own utilities for the<br />

dimensions.<br />

MAUT can be modified considerably for the analysis of real decisions. For example,<br />

suppose you have one usable eye, <strong>and</strong> it has a cataract that your doctor must<br />

eventually remove. You must decide whether to have the cataract removed now or<br />

wait a month. Your vision is getting slowly worse. A successful operation will restore<br />

your vision fully, but a failure will leave you blind. The probability of failure<br />

is something your doctor can tell you. What you need is an estimate of the relative<br />

utility — for the next month — of the three possible outcomes: restored vision,<br />

blindness, <strong>and</strong> your present vision (assuming that it does not change over the course<br />

of the month). How can you think about this? One way is to think of the various activities<br />

that you can do given some outcome but not another. The activities should be<br />

separate <strong>and</strong> independent. For example, you cannot count both “reading” <strong>and</strong> “doing<br />

your job,” if your job involves reading. (You can count “reading for pleasure.”) You<br />

can assign utilities to these activities by finding groups of them that are equivalent.<br />

Suppose that there are ten activities, all with the same utility. Nine are activities<br />

you can do now but would not be able to do if you were blind, but there is one<br />

you cannot do now but would be able to do if your vision were restored. If we<br />

assign a utility of 100 to restored vision <strong>and</strong> 0 to blindness, the utility of your present<br />

condition is 90. You should have the operation now if the probability of success is<br />

greater than .90, since its expected utility would then be at least .90 · 100 + .10 · 0,or<br />

90. Of course, after doing an analysis like this, you could decide that the activities<br />

were not independent after all, because the loss of the ability to engage in one of them<br />

would be much worse if it were the only one left than if there were other alternatives.<br />

Nonetheless, an analysis of this sort could help you to think about your decision in<br />

a new way. If you repeated this process month after month, eventually you would<br />

reach a point at which the activities you would gain back from success would be<br />

worth the risk of failure.<br />

Table 14.2 shows another example of a multiattribute analysis. The question is<br />

to choose a birth control method. This too is a repeated decision. Let us assume it is<br />

made once a year, so the decision concerns the effect of using one method or another<br />

for a year. The particular analysis into attributes is one of many ways to do this. A<br />

unique feature is the separation of AIDS risk from risk of other sexually transmitted<br />

diseases. Some attributes are omitted, such as cost, religious issues, reversibility,<br />

<strong>and</strong> health effects other than sexually transmitted diseases. Many methods are not<br />

included, such as sterilization, implants, the “ryhthm method” <strong>and</strong> combinations of<br />

methods.<br />

In order to assign weights to the attributes, we need to know their ranges. Here<br />

is how they are defined, in order from left to right:<br />

HIV prevention. Range is 0 to 1 in 1,000 chance of infection per year. (These figures<br />

are hypothetical.)<br />

STD prevention. Sexually transmitted diseases other than HIV infection include

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

Saved successfully!

Ooh no, something went wrong!