Understanding mental health-related discussion on social media and factors associated with supportive responses to mental health-related misinformation: a study of Sulli’s case on Weibo

Social media functions as a primary conduit for the aggregation and dissemination of information, while also providing a platform for the spread of mental health-related misinformation (MHRM). Although extensive research has examined health misinformation more broadly, studies focusing specifically on MHRM remain limited, particularly in contexts such as China where mental health resources are scarce, stigmatisation is prevalent, and the digital environment is complex. This study examines mental health-related discussions on Weibo and investigates factors associated with individuals’ support for MHRM. Drawing on the Health Belief Model, alongside risk and trust theories and dual-process frameworks, the research identifies key heuristic cues that shape mental health-related risk perception and trust in misinformation, and examines how these factors correlate with support for MHRM.

Data were harvested via automated crawlers from Weibo, targeting discussions surrounding depression and other mental health issues in the wake of a high-profile case. The dataset comprises textual content—including posts, comments, reposts, and replies—and diverse social media metrics. Instances of MHRM were identified and processed using deep learning-based natural language processing techniques to annotate semantic features such as civility, sentiment, and call-to-action functions. Subsequent statistical, social network, and computational thematic analysis were employed to understand online discursive practices and to elucidate the factors associated with support for MHRM.

The findings reveal that misinformation incorporating more visual elements or extended text is less likely to receive support, and that individuals’ heuristic evaluation of information quality plays a significant role in shaping both risk perception and trust. These results underscore the importance of pre-existing beliefs in the processing of mental health-related information and highlight vulnerabilities in heuristic evaluations. Methodologically, the study contributes through the application of computational social science techniques, while theoretically it offers a novel integrative model to guide future research on misinformation support in digital environments.