02.03.2013 Views

Downloadable - About University

Downloadable - About University

Downloadable - About University

SHOW MORE
SHOW LESS

Create successful ePaper yourself

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

Consistency and coherence checks 285<br />

mean testing to see if an assessment obtained by different methods or<br />

approaches is firm). Our view is that when a probability distribution<br />

has to be assessed the probability method is usually the best approach to<br />

adopt because, as we argued earlier, it tends to overcome the propensity<br />

of decision makers to estimate distributions which have too narrow<br />

arange.<br />

Consistency and coherence checks<br />

Consistency checks are, of course, a crucial element of probability<br />

assessment. The use of different assessment methods will often reveal<br />

inconsistencies that can then be fed back to the decision maker. These<br />

inconsistencies should act as a stimulant to more intense thought which,<br />

hopefully, will result in greater insight and improved judgment. Indeed,<br />

the axioms of probability theory give no guidance as to which is the best method<br />

for the elicitation of subjective probability. Empirical research in the psychological<br />

laboratory has shown that sometimes the indirect methods<br />

are inconsistent with direct methods and sometimes they are not. Some<br />

investigators have demonstrated consistency between probability estimates<br />

inferred from wagers and direct estimates. 8 Others have shown<br />

that statistically naive subjects were inconsistent between direct and<br />

indirect assessment methods, whereas statisticians were not. 9 Generally,<br />

direct odds estimates, perhaps because they have no upper or lower<br />

limit, tend to be more extreme than direct probability estimates. If probability<br />

estimates derived by different methods for the same event are<br />

inconsistent, which method should be taken as the true index of degree<br />

of belief?<br />

One way to answer this question is to use a single method of assessing<br />

subjective probability that is most consistent with itself. In other words,<br />

there should be high agreement between the subjective probabilities,<br />

assessed at different times by a single assessor for the same event, given<br />

that the assessor’s knowledge of the event is unchanged. Unfortunately,<br />

there has been relatively little research on this important problem. One<br />

review evaluated the results of several studies using direct estimation<br />

methods. Test–retest correlations were all above 0.88 with the exception<br />

of one study using students assessing odds – here the reliability was<br />

0.66. It was concluded that most of the subjects in all experiments were<br />

very consistent when using a single assessment method.<br />

The implications of this research for decision analysis are not clear<br />

cut. The decision analyst should be aware that different assessment

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

Saved successfully!

Ooh no, something went wrong!