Decision Analysis: General Description

Decision trees are a way of taking complex issues, breaking them into manageable units of decisions and their consequences, and then recombining those units into a single set of outcomes which can be used to inform one's choices for dealing with the issue. Usually, the results of the analysis are presented as a "stem-and-leaf" display (a collection of branching lines, beginning at a single point and ending with a large number of points).

Decision trees (like all decision analysis) forces the analyst to think carefully about:

The true nature of the issue (e.g., what are the decisions which must be made, what are the choices within each decision, what are the potential consequences of each choice....)
The role of chance (not all determinants of an issue are within the control of the decision-maker; the consequences of some decisions may be unknown--or even unknowable--or may depend on the choices of other decision makers--which may be unknown before one makes one's own choice)
The character of the interaction of decisions and chance events (upon analysis, it may become apparent that some decisions are unnecessary or that others could be greatly improved by more information--i.e., lessening the role of chance). (cf. Stokey & Zeckhauser, pp. 204-205)

Clarifying these issues of choice and chance can help decision makers make the best choice. But even then, the choice is never dictated by the analysis because the analysis almost always includes estimates of outcomes which are uncertain, and the decision maker must take into account his/her/their preferences for various outcomes. Under different circumstances, one might be more willing to accept greater risk; under other circumstances risk must be avoided.

609