Cognitive strategies in a complex judgment task: Analyses of concurrent verbal reports and judgments of cumulated risk over different exposure times

Subjects judged the cumulated risk of dying during a year for a set of persons or cases each characterized by the risks during three different periods of risk exposure during the year (e.g., 18 weeks with 2.3 deaths per thousand persons per year of exposure, 28 weeks with 2.5 deaths and 6 weeks with 16.0 deaths per thousand per year). Three main strategies were used for solving the problems: (1) addition of the risk levels, (2) computing the mean of the risk levels, and (3) the most frequently used anchoring and adjustment strategy most closely approximating the normatively correct risk over the year. The first two strategies do not use all relevant information because they neglect the importance of different exposure times. The anchoring and adjustment strategy used all information and was executed in a first stage in which an initial anchor or reference risk was selected followed by a second stage in which this anchor was adjusted according to the remaining information. The process for finding the first anchor was quite flexible, using different principles (e.g., extreme values or similarity) depending on the pattern of cues characterizing each case. Biases in relation to normative theory originated from subjects misunderstanding or simplifying the problem (Strategies (1) and (2)) or the characteristics of the anchoring and adjustment process which resulted in relative overestimation of short-time, high-risk cases. A few subjects, however, all of whom used sequential anchoring and adjustment, succeeded in giving quite accurate estimates.