Structuring the decision problem

Summary The basic structure of a decision problem entails alternatives, uncertainties, consequences of alternatives and uncertainties, as well as the objectives and preferences of the decision maker. Important decisions on the relevant alternatives have to be made in advance. They refer to such questions as: Should further alternatives be considered or should a choice be made from the existing ones? Should the number of existing alternatives be reduced by merging similar alternatives or increased by splitting existing alternatives into several variants? Should the options be designed as one-stage or multi-stage alternatives? Other important preceding decisions relate to the modeling of uncertainties. Can the future be predicted sufficiently well to neglect uncertainties in general? If not, what are the relevant uncertainties that influence the outcomes of the decision problem? In how much detail or how general should the universe of possible states be modeled? Uncertainty is described by states or events to which probabilities are allocated. Probabilities have to obey certain rules: joint probabilities and conditional probabilities are relevant to combining uncertain events. These combinations can be visualized by the use of event trees or cause trees. If an alternative is chosen and uncertainty resolved, a certain consequence will be obtained in a deterministic fashion. It might be necessary to formulate an effect model that determines the consequences. When modeling the preferences, a preliminary decision has to be made on whether a single objective or several objectives should be considered. The relevant objectives have to be identified. In the case of uncertain expectations, risk attitude has to be considered, and in case of consequences that occur at different points in time, it might be reasonable to model time preference. Usually, it is not possible to model the different components of the decision problem independently of one another. The components influence each other. The decision maker moves back and forth between the set of alternatives, the uncertainty structure and his model of preferences until he finishes modeling the decision problem and derives the optimal decision. Graphical forms of representation such as the influence diagram, the decision matrix, and the decision tree are very useful tools. They force the decision maker to clarify his conceptions and support him in communicating with and explaining the decision fundamentals to other people involved in the decision process.