13.1 Nonrepeatable Events
It has been shown in Chapter 7 how you may assess your probabilities in many cases using classical ideas of equiprobable outcomes or, more often, by employing frequency concepts. Historically, these have been the most important methods of assessment and have led to the most valuable applications. However, there remain circumstances where neither of these ideas is relevant and resort has to be made to other methods of assessment; to other methods of measuring your uncertainty. For example, if you live in a democracy, the event that the political party you support will win the next election is uncertain, yet no equiprobable cases or frequency data exist. It is clearly unsound to argue that because over the past century your party has been successful only 22% of the time, your probability of success now is around 0.22, for elections are not usually judged exchangeable. No really sound and tested methods exist for events such as elections and as a result, this chapter is perhaps the most unsatisfactory in the book. What is really needed is a cooperative attack on the problem by statisticians and psychologists. Unfortunately, the statisticians have been so entranced by results using frequency, and the psychologists have concentrated on valuable descriptive results, that a thorough treatment of the normative view has not been forthcoming. What follows is, hopefully, not without value but falls short of a sound analysis of the ...