Temporary versus persistent influence in crisis communication: Analyzing Twitter communication during the early stages of COVID-19 in Japan.

Twitter plays an important role in crisis communication in many countries. It is clear that influential accounts play a bridging role connecting the public and the government in times of crisis. Previous research has measured user influence during crises using the PageRank algorithm. However, few studies have investigated the persistence of influence during lengthy crises. Research on the COVID crisis in Germany showed news organizations’ and journalists’ accounts to have been influential in all phases of the crisis, while government accounts were particularly important shortly before and after the lockdown (Shahi et al., 2021). If governments can identify accounts with long-term influence, then collaborating with those accounts’ owners could help to build a more stable and responsive communication system. In contrast, cases where accounts exert only temporary influence could be due to the content of specific posts; analyzing the content of such posts might suggest ways of avoiding misinformation. This study aims to identify the continuity of influence of different categories of Japanese Twitter accounts during the early stage of COVID-19 (February to May, 2020). The categories and features of influential accounts were identified by retweet network analysis and text mining. Influential accounts are classified into five categories: “Business and Civil Society”, “Government”, “Media”, “Experts”, and “General Individuals”. Compared with ordinary users who speak from personal perspectives, influential accounts conduct significant discussions on “social level”, relating to the Japanese government’s pandemic prevention policies and medical responses, which usually present strong negative emotions. To investigate the persistency of influence held by influential accounts, a regression analysis model was constructed by using the number of times the accounts were defined as influencers throughout each stage of crisis communication as the independent variable. The predictive analysis aims to contribute to an appropriate definition of influential accounts in future crisis communication. Since Japan was legally unable to enforce a lockdown, citizens were asked to adopt “self-restraint” measures. The government’s ambiguous and uncertain response policies provoked public discontent and undermined trust in government policies and announcements. This makes the Japanese case particularly suitable for identifying influential accounts that exerted a lasting influence over the direction of public opinion.