An Investigation of Judgmental Reliability

点击翻译按钮获取中文摘要

An Investigation of Judgmental Reliability The proposal deals with the effects of the reliability of judgment (conceptualized as consisting of two parts--reliability of information acquisition and reliability of information processing) on judgment accuracy. The PI proposes a study of experts (radiologists reading mammograms) and a lab study that is a loose analog of radiographic diagnosis. The conceptual framework is an expanded version of Brunswik's lens model. The review of the literature shows that the topic of reliability has been neglected in the study of expert judgment, although it is known that there are individual differences in judgment reliability. The first study proposed would have 10 radiologists read a set of 100 mammograms twice. They would also judge a set of features of each mammogram twice to provide evidence about reliability of information acquisition. The second study is an experimental analog in which subjects are given visual cues or visual and numerical cues (such as age, etc.) which are probabilistically related to an outcome. Baseline and feedback trials will be given. Subjects will judge the outcome in some blocks of trials and will judge the cues in other blocks of trials. The criterion will be linearly related to the cues in all cases.