Double reading reduces miss errors in low prevalence search.

Abstract

Low prevalence studies show that people miss a large proportion of targets if they appear rarely. This finding has implications for real-world tasks, such as mammography, where it is important to detect infrequently appearing cancers. We examined whether having people search in pairs in a "double reading" procedure reduces miss errors in low prevalence search compared with when participants search the displays alone. In Experiment 1 pairs of participants searched for a mass in a laboratory mammogram task. Participants either searched the same display together (in the same room) or searched the displays independently (in separate rooms). Experiment 2 further manipulated the reading order so that paired participants either read the mammograms in the same or different orders. The results showed that, although there was no effect of reading order, double reading led to a substantial reduction in miss errors compared with single reading conditions. Furthermore, the reason for the double reading improvement differed across reading environments: When participants read the displays in a shared environment (i.e., in the same room) the improvement occurred due to an increase in sensitivity; however, when participants read the display in different rooms the improvement occurred due to a change in response bias. (PsycInfo Database Record (c) 2021 APA, all rights reserved).