The Integrated Contingency Model: Range-sensitive decision models

Weighted averaging models commonly are used to aid policy makers in their deliberations. A number of studies have submitted such models to range sensitivity tests which assess the models' validities. The findings of these studies have questioned the use of these models and the policies that have been adopted based on their application.;In this dissertation, a theoretical framework, the Integrated Contingency Model, was developed to explain when models would be range-sensitive. The framework posits that passage of such a test requires people's cognitive processes be cue level dependent and stable. Furthermore, an averaging rule must be descriptively valid. These properties are a function of cognitive effort and of task demand, both of which are contingent on the decision context. The decision context encompasses the decision environment, the characteristics of the individual and of the task, and the presentation of the task.;It was hypothesized that the range sensitivity of judgment models is contingent upon the decision context. To test this hypothesis, two experiments were conducted. In both, the range sensitivity of weighted averaging judgment models was assessed. Experiment 1 examined the range sensitivity of this model when the task required people to express preferences and when the decision context was expected to facilitate range sensitivity. Two different range sensitivity tests were used, and the effect of response scale was tested. Experiment 2 examined the range sensitivity of a weighted averaging model when the task required people to make predictions and when the decision context was expected to inhibit range sensitivity. Within the second experiment, two factors--response scale and cue definition--were tested.;The results of both experiments were consistent with the ICM. The models in Experiment 1 generally passed the range sensitivity tests, and the results were not affected by the choice of response scale. The models in Experiment 2 generally did not pass the range sensitivity tests, and the degree of range sensitivity was contingent upon the experimental factors. These results suggest that range sensitivity is contingent upon the decision context.