Seeing is Not Believing: A Nuanced View of Misinformation Warning Efficacy on Video-Sharing Social Media Platforms

Misinformation warnings have become the de facto solution for fighting fake news online. Our study brings attention to the challenge of developing effective misinformation warnings on short video-sharing platforms. We conducted semi-structured interviews with the think-aloud protocol to understand how users interact with and perceive misinformation warnings, specifically the interstitial and contextual warnings adopted by TikTok and Instagram Reels. We recruited 28 regular users of TikTok and Instagram Reels for this study. We contribute to the evolving scholarship on social media misinformation mitigation by casting light on nuanced participant interactions with and perceptions of misinformation warnings and how these interactions and perceptions influence the perceived accuracy of short video content. Our findings are threefold. First, the present study shows that specific contextual warnings do not always elicit behavioral adherence but can alert users to be vigilant about misinformation. Second, users' perceptions of interstitial and contextual warnings are influenced by the warning's explicitness and the risk level of the misinformation. Third, we identify the least and most effective/favored warning designs to help make accuracy judgments according to the participants. To this end, our findings have implications for improving the design of misinformation warnings on short video-sharing platforms.