Statistical decision theory attempts to provide a formal approach to decision making and statistical inference by asking that explicit numerical criteria be given and that the merits of a statistical procedure or decision be assessed by these criteria. Since its introduction in the 1940's, decision theory has served as a unifying force for apparently diverse branches of statistics. Its unifying nature is continually being rediscovered; for instance, a large body of recent work on function estimation has been found to be intimately related to classical minimax theory. Decision theory has also provided the basis for much of the foundational development in statistics. It also has a very practical side, and is increasingly being used in the modeling and analysis of real decision problems in business, economics, medicine, law, reliability and quality control, risk assessment, and numerous other fields. This project will support the Fifth Purdue International Symposium on Statistical Decision Theory and Related Topics to be held from June 14-20, 1992 at Purdue University. The broad topics chosen for this symposium include classical subjects within or related to decision theory, such as admissibility/minimaxity/invariance, sequential analysis, hierarchical and empirical Bayes analysis, multiple decision theory, optimal designs, and model selection. Newer topics will be included such as decision-theoretic function estimation, information/complexity, semiparametric models and adaptive methods, neural nets/influence diagrams/graphical structures, multiple prior/utility problems, computational Bayesian developments, and decision-theoretic aspects of noninformative priors, as well as a wide variety of applied problems in decision theory. Previous symposia in this series on decision theory and related topics were held in 1961, 1970, 1976, 1981, and 1986.
