Decision Spaces

In many decisionmaking situations a fundamental question to be answered is, "At what point do I stop looking for additional alternatives and make a decision based on those I have seen so far?" While not all decisionmaking situations pose this problem, there is a class of decisions where one must worry about when to stop looking and begin deciding. This occurs when the nunmber of alternatives available for potential consideration is virtually unlimited and the number currently on hand for evaluation and decisionmaking is comparatively small. Moreover, little is known about how the alternatives on hand stack up against those "out there somewhere" still waiting to be discovered. The central issue is the lack of a context for figuring out when to stop looking and begin deciding, and the central idea is to be able to provide such a context. Two technologies exist that can do just that when combined: multiattribute utility measurement and Monte Carlo simulation. Such a context, called a decision space, can help answer questions about whether or not costly, time-consuming searches for better alternatives are likely to pay off. This paper discusses decision spaces in general, gives one operational example of how to produce them, and then applies the example to illustrate their use.