Valuing Infor Mation in Decision Making Under Uncertainty

Abstract

A methodology is outlined for objectively determining the potential value of infor Mation in a decision theoretic context in the absence of any prior infor Mation. Accordingly, two or more decision situations can be valued to determine where scarce infor Mation resources are best deployed. It is shown how minimum and maximum EVPI can be determined for a given payoff matrix as a game against nature. The infor Mation gains (or losses) in dollar or percentage terms can also be determined as the decision maker moves through various infor Mation stages such as null, vague and proper prior distributions. Differences between this methodology and an entropy approach to infor Mation valuation are illustrated.