Sustaining corrected beliefs in false news headlines over time: The roles of correction format and recognizing corrections.

Abstract

Exposure to false information on social media can erode the accuracy of beliefs and memory. These effects can be mitigated by issuing fact-checked corrections. Mounting evidence suggests that corrections improve belief accuracy the most when consumers recognize that a statement was earlier corrected. Two experiments examined this relationship further by testing whether correction formats vary in how they promote encoding and recognition of corrections as well as belief accuracy. To assess baseline beliefs, participants first rated the veracity of true and false headlines from the internet. Participants then read three types of corrections that all included veracity labels. Corrections showed only false headlines from the prior phase, only true headlines that contradicted false headlines from the prior phase, or false and true headlines, in that order. Finally, participants rerated the veracity of the original headlines and rated their recognition that headlines were corrected, both immediately after the correction phase and again after 1 week (Experiment 1) or month (Experiment 2). Corrections improved belief accuracy by lowering veracity ratings for false headlines, especially for corrections with false and true information. Highly confident recognition of corrected headlines was associated with durable improvements in rbelief accuracy, whereas less confident recognition was associated with less improvement and sometimes impairment. Corrections with false and true information led to the most high-confidence recognition of corrections. Collectively, these findings suggest that modifying correction formats to promote encoding and recognition of corrections can lead to sustained improvements and belief accuracy over time that counteract regression to false beliefs.