Reliability vs. Diagnosticity in Hierarchical Inference

摘要

This study examined the performance of subjects in a cascaded inference task where two subjects worked together, one subject having diagnosticity information and the other having reliability information. This was compared to a condition in which a single subject received both types of information. Additionally, the effects of different 'experts' having the power to make the final decision in the two-person conditions was explored. Seventy-two subjects made inferences about the probability of success vs. failure of hypothetical job applicants presented in a personnel manager scenario. Subjects were paid bonuses according to their performance on the task. Contrary to hypotheses, there were no between conditions differences. Single subjects performed just as well as subjects working together. This study replicates previous work using single subjects in the general pattern of responses: subjects were somewhat radical in comparison to the normative model. (Author)