Show description

Detail is also good. , linear regression models of experts, some applications of Bayesianism as a descriptive theory) treat the subject as a black box and simply claim that the input–output relations are pretty much what they would be if she actually went through the steps prescribed by the models (taking a weighted average of several predictor variables, multiplying prior odds by a likelihood ratio). By contrast, process models aim to trace the internal cognitive processes as a person engages in a stretch of reasoning.

These include exploiting people’s susceptibility to anchoring effects by staking out extreme positions, focusing on vivid and memorable sorts of dangers (even if their risk is much less than that of more pallid dangers), and inflating the likelihood of some risk by ignoring base rates. But it is doubtful that successful opinion makers needed to wait for behavioral decision theory to learn about tactics like these. Extrapolations and Applications 43 Notes 1 Bayes’s theorem provides a simple and appealing normative model for updating probabilities.

The use of base rates)? Or when experimenters investigate a wider range of tasks, types of information, formats for judgments, and the like, will the effect dissolve into a fog of ever higher-order interactions (specific information increases sensitivity to base rates, unless it is difficult to weave it into a causal scenario around it, but even then it increases sensitivity if it is also the case that . , unless . )? Some generalizations may end up being relatively unqualified. But it is also possible that further investigation will show that some phenomena must be hedged with so many qualifications that it will be almost impossible to predict with any confidence when they would occur outside the lab or, in the worst case, to determine in retrospect that they had occurred.