Individual Submission Summary
Share...

Direct link:

Assessing the Cost-effectiveness of Fact-checking Interventions

Thursday, November 5, 3:30 to 5:00pm, Property: Boston Marriott Copley Place, Floor: 4th Floor, Room: Nantucket

Abstract

Online misinformation has become a central public policy problem as it erodes trust, distorts democratic decision-making, and imposes broad social costs. Previous literature has pointed to the need of paternalistic intervention, yet few studies compare the cost-effectiveness of different policy design choices. Should platforms invest in resource-intensive fact-checking systems, or can lighter-touch interventions meaningfully improve citizens’ judgments? Evidence on low-cost, scalable intervention can help policymakers identify strategies that improve information quality without relying solely on expensive moderation systems or unevenly available fact-checking infrastructure. This paper evaluates whether a low-cost “accuracy nudge” can improve people’s ability to distinguish true from false news, and how its effects compare with a built-in fact-checking tool.
We study this question using a pre-registered 2×3 online experiment conducted on Prolific. Participants evaluated 18 news items drawn from Snopes, spanning political and nonpolitical topics and balanced by partisan slant. The experiment randomized subjects across three information environments: no fact-checking option, access to a quota of on-demand fact-checks, and an encouragement treatment that prompted subjects to verify claims using their own judgment and trusted external sources. Each environment was crossed with a monetary incentive condition that either offered a flat payment or accuracy-contingent bonuses. In addition to whether respondents correctly classified each item as true or false, we also measure response time as a proxy for attention and deliberation.
The results show that the encouragement nudge improves performance relative to the baseline, increasing mean accuracy by roughly 12%, though this gain is smaller in magnitude than the 34% improvement produced by direct fact-checking. The nudge also increases deliberation time by about 13–20%, suggesting that it works in part by encouraging greater attention to the task. Within the encouragement condition, subjects classified as “nudged” based on longer response times are substantially more accurate than non-nudged subjects, and regression estimates show a positive relationship between attention and correctness in the nudge treatments.
Our findings identify a scalable intervention that improves news judgment without requiring platforms to pre-screen content or provide costly labels for every claim. While direct fact-checking remains more effective in the short run, accuracy nudges may offer a cost-effective complement that broadens reach, encourages user-side deliberation, and helps reduce the spread of misinformation at scale.

Author