Covid 19: RAPID: Collaborative Research: Agency COVID-19 Risk Communication on Social Media: Characterizing Drivers of Message Retransmission and Engagement

Public health and emergency management agencies are on the front lines of informing and educating the public about the science of virus transmission and prevention. In response to a threat such as COVID-19, their mission requires them to communicate accurate and credible information to local populations using all means of information delivery. Currently, one of the most effective means of directly reaching the public is through the use of digital and social media. Social media enables delivery of timely and actionable risk information to the public while also supporting ongoing dialog, potentially increasing trust and reducing fear and erroneous rumors fueled by misunderstanding. This is increasingly important in the case of the coronavirus pandemic, which has now affected at least 40 states across the U.S. and requires nearly hourly public health alerts and warnings to at risk individuals. In our prior work on online communication in the recent Zika and Ebola outbreaks (Vos et al., 2018) we established that effective messaging on Twitter depended upon employing a combination of content, style, and structure features - but that the right mix seemed to depend upon properties of the disease event (including the uncertainty and ambiguity of the threat, the nature of the consequences involved, and the need for public information). COVID-19 poses a distinct risk profile, with a disruption potential to the American public not seen by any threat within decades. In this project, we propose to identify the key drivers of effective messaging in an emerging pandemic, and specifically to rapidly identify strategies for improving effectiveness in social media communication involving COVID-19 by public agencies. Our specific focus will be on the outcomes of message retransmission (an essential outcome for both high levels of message penetration and ensuring the multiple exposures needed for behavioral influence) and engagement (a critical indicator of attention and a driver of trust), both of which are measurable and established as core outcomes in prior studies of effective social media communication (Sutton et al., 2015). We will pursue this objective through the following core activities: • Collection of perishable social media data on COVID-19 messaging by public agencies, and public engagement with/retransmission of those messages. Here, we build on our established social media monitoring infrastructure, as well as experience with related events, to rapidly collect information from responding agencies posted to social media. We will prospectively collect all messages posted by a set of over 700 local, state, and national public agencies relating to COVID-19 on Twitter, as well as information on Follower relationships and additional meta-data. • Content coding of COVID-19 messages, to typologize information that is specific to the present event. We will employ a combination of expertise-based manual coding (drawing on our and others' prior work in the area) and automated coding, with use of machine learning techniques to generalize from manually coded samples as necessary. We have successfully used this strategy in prior events. • Characterization of messaging strategies used by public agencies in the evolving COVID19 response. Using a combination of manual content analysis and exploratory multivariate analysis of message sequence and features, we will identify the dominant de facto strategies used by agencies in their social media communications, and track their use over time. • Predictive analysis of message outcomes based on message context, content, style, and structure. We will use statistical models of retransmission rates and engagement outcomes to identify the predictors of effective messages, while controlling for contextual factors (e.g., account-specific effects, time of day, or background attention levels) that can also influence message success. We pioneered this approach within the social/media hazard space, and have successfully employed it in many previous studies. • Development of evidence-based guidance for effective social media messaging by public agencies in response to this and similar events. Using the above analyses, we will develop accessible guidelines for messaging strategies around COVID-19 and similar threats, in a practitioner-ready format. Our research strategy in all of the above points builds on our successful prior work in this area, both in response to emergent infectious disease threats (Vos et al., 2018) and in the context of anthropogenic (Sutton et al., 2016; Sutton et al., 2013; Sutton, Spiro, Fitzhugh, et al., 2014) and natural hazard events (Olson et al., 2019; Sutton et al., 2015; Sutton, Spiro, Johnson, et al., 2014).The Co-PIs for this project, Jeannette Sutton and Carter Butts, have been studying online information communication in the context of hazards and disasters since 2009. During this time, we have built a data infrastructure system that has collected well over 10 billion social media posts and tracked several thousand accounts over a multi-year period, supporting more than fifteen papers in peer reviewed publications, dozens of presentations at scientific meetings, and multiple practitioner-oriented documents. Importantly, this body of work has led to the identification of messaging characteristics that increase message diffusion and amplification of risk communication messages on social media under contexts of imminent threat and longer term emerging infectious disease outbreaks. The findings from our work have informed the messaging strategies of local, state, and Federal agencies including NOAA, NIST, DHS, and the CDC. Intellectual Merit: Risk communication messages on social media are real time traces of online in/formal communication shared under conditions of imminent and ongoing threat. Research on communication and messaging dynamics online provides insights into the social amplification of risk, via diffusion of information, and strategies to design effective messages. This project will test the risk communication on social media model in response to a global pandemic by analyzing official communication from public health and emergency management Twitter accounts. The findings from this work will lead to the further development and refinement of the social amplification of risk framework and the risk communication on social media model. Broader Impacts: This research will have immediate benefits to organizations and agencies tasked with communicating to at risk populations about emergent infectious disease. Our findings will inform the design and dissemination of risk communication messages and will be immediately applicable to public health and safety organizations in the context of COVID-19. Results will be shared via fact sheets, webinars, published papers, and presentations with academic and practitioner audiences.