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.

112 NORMATIVE THEORY OF PROBABILITY<br />

Constructing probability judgments<br />

According to the frequency theory <strong>and</strong> the logical theory, probability statements are<br />

reflections of objective facts in the world. According to the personal view, probability<br />

statements are judgments, <strong>and</strong> probability judgments need to be constructed by<br />

appropriate methods.<br />

Probability as willingness to bet<br />

What should we be trying to do when we make a probability judgment? Many theorists<br />

have argued that we should think about our personal willingness to bet that the<br />

proposition in question is true. Surely, the probability of a proposition should be one<br />

of the determinants of our willingness to act as though the proposition were true —<br />

to “bet on it” (see Chapter 10). In addition, the idea that probability can be defined<br />

in terms of a tendency to take action ties the concept of probability closely to observable<br />

behavior. However, the idea that probability judgments can be constructed by<br />

thinking about our willingness to bet has drawbacks.<br />

Ramsey (1931) argued that if we believe that the probability of an event is .5,<br />

this amounts to saying that we are equally willing to bet on the event’s happening<br />

as on it not happening. To take a more complicated case, suppose I value $3 three<br />

times as much as I value $1. Suppose you ask me the probability that it will rain<br />

tomorrow, <strong>and</strong> I say that I would be willing to bet, at three-to-one odds, that it will.<br />

This means that I would be willing to accept $1 if it does rain, on condition that I<br />

would pay $3 if it does not. Moreover, I would be just as willing to do the opposite,<br />

to pay $1 if it rains or accept $3 if it does not. According to the betting interpretation<br />

of probability, this means that my subjective probability of rain is .75. I would regard<br />

it as three times as likely to rain (.75) as not to rain (.25).<br />

Over a long series of making such bets, based on this same degree of belief,<br />

I would come out even, winning <strong>and</strong> losing the same amounts, if my judgments<br />

corresponded to the relative frequencies of the events. If I collect $1 every time<br />

the .75-probability event happens <strong>and</strong> pay $3 every time the .25-probability event<br />

happens, then, out of 100 events, I will collect about $75 for the more frequent event<br />

<strong>and</strong> pay out about $75 (25 times $3) for the less frequent event. This fact may help<br />

to make clear the use of the concept of betting in the measurement of probability.<br />

In sum, by this view, the way to construct personal probability judgments is to ask<br />

ourselves what odds we would consider fair. (We shall return to the connection<br />

between the personal view <strong>and</strong> relative frequencies.)<br />

It is not clear that thinking about bets is the best way to elicit probability judgments<br />

(Shafer, 1981). We think about probabilities, <strong>and</strong>, more generally, the strengths<br />

of our beliefs, because our probability judgments help us to make decisions. We often<br />

do not know what decision to make until we think about the strengths of our<br />

beliefs about the consequences of the various options before us. We cannot consult<br />

our decisions to determine what we believe, because we need to know what we<br />

believe in order to make our decisions.

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

Saved successfully!

Ooh no, something went wrong!