From Statistical Decision Theory to Practice: Some Problems with the Transition

While considerable progress has been made in the development of sophisticated statistical decision models, little attention has been paid to the problem of applying them in the "real world." This paper approaches one aspect of the necessary conversion process, namely the conceptualization and use of probabilistic data for decision making. Three points are demonstrated on the basis of clinical experimental evidence: (1) decision makers have difficulty conceiving of probability distributions, rather they are more comfortable with point estimates; (2) the concept of "risk of failure," the probability of failing to meet a perceived target, plays a major role in choice; and (3) the format in which probabilistic data are presented can affect choice behavior. A simple descriptive decision model is then developed, based on both the clinical evidence and two years of observation of a committee empowered to invest in new product development and product expansion activities.