Evaluating Human Judgments and Decision Aids

Abstract

One goal of applied decision research in accounting is to improve decision making (Libby and Lewis [1982]). The typical approach is to identify situations in which human judgments diverge from the output of a formal (normative) model and then recommend a decision aid to eliminate or reduce the divergence. The mere fact that human judgments diverge, however, does not mean that decision aids are always appropriate; and even if they were, the type of aid which might be employed is not obvious from such an observation. First, the opportunity cost of divergence from formal models must be sufficient to justify the cost of developing and implementing decision aids. Second, the reasons for divergence must be understood before the appropriate aids can be identified. The accounting literature is weak in both of these areas. With respect to opportunity costs, most researchers have ignored the problem of trying to estimate the benefits to be obtained from decision aids. Regarding causes of divergence, few accounting researchers have attempted to analyze the components of decision processes. This holds both for lens model studies and those involving the study of heuristics and biases. Each can be characterized as an input-output method in which causes of divergence are inferred from the relationships between the information provided to the decision maker (the inputs) and the judgments or responses of the decision maker (the outputs).