Despite millions of likes, comments and reposts, government microblogging may fail to achieve real communication impact during public events. This study aims to evaluate the effectiveness of information release by government microblogging and identify which public-response indicators drive success or failure across different events.
Drawing on 304,588 government microblog posts and 1,546,665 associated user responses from four major public events on Sina Weibo, the authors developed a multi-indicator evaluation system using topic modeling and qualitative analysis. A high-performing Robustly Optimized BERT Pretraining Approach (RoBERTa)-based classifier was then trained to automatically categorize user responses. To evaluate information release effectiveness, the super-efficiency slack-based measure data envelopment analysis model was used. Furthermore, key evaluation indicators functioning as sources of efficiency loss or drivers of success across different public events were identified using slack analysis and Cliff’s Delta.
Three key findings emerge. First, the evaluation indicators based on public responses include aligned appraisal of event, divergent appraisal of event, affirmation of government, criticism of government, affective reassurance, affective venting, prosocial behavior, organizational expected behavior, adverse behavior, liking behavior, favorable sharing behavior and unfavorable sharing behavior. Second, emergency responses led by government microblogging accounts with superior information release effectiveness exhibit a structural pattern of “function-situation” matching. Third, key evaluation indicators causing efficiency loss vary across public events while core drivers are highly consistent.
This study contributes to the existing knowledge of digital government communication and proposes practical approaches for governments, social media platforms and ordinary users during public events.
