Coherence and correspondence criteria for rationality: experts' estimation of risks of sexually transmitted infections

The aim of this study is to examine both coherence and correspondence criteria for rationality in experts' judgments of risk. We investigated biases in risk estimation for sexually transmitted infections (STIs) predicted by fuzzy-trace theory, i.e., that specific errors would occur despite experts' knowledge of correct responses. One hundred twenty professionals with specific knowledge of STI risks in adolescents were administered a survey questionnaire to test predictions concerning: knowledge deficits (producing underestimation of risks); gist-based representation of risk categories (producing overestimation of condom effectiveness); retrieval failure for risk knowledge (producing lower risk estimates); and processing interference in combining risk estimates (producing biases in post-test diagnosis of infection). Retrieval was manipulated by asking estimation questions that “unpacked” the STI category into infection types or did not specify infection types. Other questions differentiated processing biases from knowledge deficits or retrieval failure by directly providing requisite knowledge. Experts' knowledge of STI transmission and infection risks was verified empirically. Nevertheless, under predictable conditions, they misestimated risk, overestimated the effectiveness of condoms, and also suffered from processing biases. When questions provided better retrieval supports (unpacked format), risk estimates improved. Biases were linked to gist representations, retrieval failures, and processing errors, as opposed to knowledge about STIs. Results support fuzzy-trace theory's dual-process assumptions that different types of errors are dissociated from one another, and separate failures of coherence and correspondence among the same sample of experts. Copyright © 2005 John Wiley & Sons, Ltd.