Investigating which corrective modalities generate greater positive user engagement on social media

Introduction Health-related misinformation correction on social media has become an important issue in digital health communication. However, the factors associated with user engagement with corrective information remain insufficiently understood. Based on the elaboration likelihood model (ELM), this study examines how information features, as central-route cues, and information source features, as peripheral-route cues, are associated with user engagement with health-related corrective posts on social media. Methods Using Sina Weibo as the research context, this study analyzed health-related corrective posts published or reposted by official debunking accounts. A data-driven analytical framework was adopted to examine the relationship between potential influencing factors and user engagement behavior. Correlation analysis, independent-samples nonparametric tests, and multiple linear regression models were used to test statistical associations. Machine learning algorithms were further introduced to assess variable importance structures and potential nonlinear patterns within the predictive framework. Results The findings show that information features are statistically associated with user engagement, but their effects vary substantially after controlling for contextual variables. By contrast, information source features, especially source influence, show a more consistent association with user engagement across different analytical frameworks. In the predictive analyses, the feature-importance structures across models are highly consistent, with source influence consistently occupying the dominant position. Discussion Overall, user engagement with health-related corrective information on social media is more strongly related to information source features than to textual language expression features, which appear to be more context dependent. This study extends the applicability of ELM to corrective information contexts on social media and highlights the important role of peripheral-route cues in high-noise information environments. The findings provide empirical insights for corrective information dissemination strategies, suggesting that enhancing the visibility and dissemination capacity of information publishers may be more practically significant than optimizing message content alone.