Contextual effects on the calibration of probabilistic judgments

Abstract

24 practicing auditors employed by public accounting firms participated in a study to determine the extent to which contextual factors affect the calibration of their subjective prior probability distributions (PPDs). Probabilistic responses to general-knowledge (almanac) questions were analyzed and compared to responses obtained in a previous study by the present 3rd author et al (see record 1983-07156-001 ), which used the same type of Ss and methods in a substantive audit judgment context. Results indicate that Ss' judgments in the general-knowledge task context were miscalibrated and significantly overconfident, whereas their judgments in the substantive audit task context were less miscalibrated and predominantly underconfident. Findings suggest that calibration research results may not be generalizable across applied judgment–decision contexts. Implications for practical applications of Bayesian decision models, improvement of preexperimental training methods, and the effects of incentives on miscalibration are discussed. (23 ref) (PsycInfo Database Record (c) 2025 APA, all rights reserved)