Past empirical research on judgmental heuristics and biases has focused on questions of classification, with relatively little attention given to the development of general theoretical principles. It is the latter, however, that ultimately will lead to conclusions of greater generality and usefulness. A selected review of the literature on representativeness indicates some of the effects for which any complete theory must account and some of the limitations in many of the current experimental designs. The manner in which people use probabilistic information depends, in part, in (i) information specificity, (ii) information salience, (iii) individual differences, and (iv) problem wording. Theories are needed to elucidate these phenomena, and experimental designs are required that pay more attention to them.
The CAUSE Research Group is supported in part by a member initiative grant from the American Statistical Association’s Section on Statistics and Data Science Education