Development of and
in Behavioral Decision Research
Carnegie Mellon University
In Ward Edwards's original formulation, behavioral decision research has three interrelated components: Normative analysis characterizes the choices that decision makers actually face. Descriptive analysis addresses how decision makers intuitively conceptualize those choices. Prescriptive analysis proposes ways to bridge the gap between the normative ideal and the descriptive reality (Edwards, 1954, 1961; von Winterfeldt & Edwards, 1986).
The complete treatment of any decision requires all three components. Without normative analysis, one cannot know what behavior is worth describing as decision relevant. Indeed, without it, the concept of bias, so central to behavioral decision research, is meaningless. Without prescriptive analyses, the field becomes a purely academic exercise, uninterested in bettering the human condition. It is also a somewhat cowardly exercise, unwilling to face the test of demonstrating that it knows enough to change behavior for the better (or, perhaps, to exploit it).
The prospect of an integrated epistemology contributed to the excitement of the field's early days. A once-legendary research program tried to do it all. Edwards and colleagues devised an experimental task, designed to capture the essential features of the practical task facing radar operators on the Cold War's DEW (Distant Early Warning) Line, in northern latitudes. That task included interpreting uncertain signals, in the context of varying prior probabilities (of attack), with explicit incentives for correct and incorrect summary judgments.