The “1-800 Critique” and the implications of descriptive models of human behavior

It is not always clear how leading theories of human behavior can be used to derive quantitative predictions. Often they are stated with enough qualifications about the applicable domain so that authors might be well advised to supply a 1-800 number for perplexed readers to call when they want to know how a particular theory applies to a novel problem. To demonstrate and address this problem we consider the task of combining theory-based predictions with new observations. We propose a two-stage estimation procedure to derive the initial theory-based prediction for a novel domain, and the best weighting of this initial prediction with accumulating observations. We show that the optimal weighting of the initial prediction and the new data can be summarized with a single number: the model's equivalent number of observations (ENO).