Information acquisition for model construction: An integrative, decision-theoretic perspective
Abstract
Most mathematical models of decision making assume that either complete and accurate input parameter information is freely and readily available or that one must resort to probabilistic modeling when dealing with uncertain information. Very often, the decision maker has the time and resources needed for acquiring additional and more accurate information prior to model building, solution, decision making and implementation. This is especially true in the context of planning and strategic decisions. Further, practicing decision makers favor easily understood and implemented 'satisficing' solution methods for the decision models constructed, as opposed to complicated optimization techniques. This is a key requirement for DSS design. For this purpose, we propose the construction and use of meta models that guide the decision model specification process and also suggest appropriate decision strategies. Such meta models assume that the decision maker can select one of several alternative information levels ranging from no information to complete information. Choosing a higher information level entails higher information acquisition costs but could improve decision quality. Essentially, the model suggests the appropriate tradeoff between information costs and decision quality. Examples of the construction and solution of such meta models in the context of static and dynamic decision scenarios are presented.
Degree
Ph.D.
Advisors
Whinston, Purdue University.
Subject Area
Information Systems|Management
Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server.