Title: Shifting attention to accuracy reduces misinformation sharing: Evidence from computational modeling and field experiments

Abstract: In recent years, academics and practitioners have sought solutions to reduce the sharing of misinformation online. One promising intervention involves prompting people to think about accuracy. Although effective in experimental settings, it remains unclear how this accuracy prompt intervention works and how it can be scaled up and delivered on social media platforms. Here we address these two questions. We apply computational modeling (drift-diffusion models) to a large corpus of experimental data to test competing cognitive mechanisms. We find that accuracy prompts do not cause people to deliberate more but instead shift people’s attention to accuracy while they deliberate. That is, the intervention changes what people are thinking about rather than how much they are thinking. We then delivered the intervention at scale on Twitter using targeted digital advertisements encouraging users who regularly share misinformation to think about accuracy before tweeting (e.g., "think about accuracy before you share"). Meta-analytic evidence from three field experiments suggests that the ad campaigns successfully reduced the amount of misinformation shared on Twitter. Together, these findings suggest that shifting attention to accuracy is a promising approach for reducing the spread of misinformation online.

Zoom link: https://mit.zoom.us/j/8796050369