Historically, social media engagement has been imagined as any interaction between organizations and stakeholders that occurs online (Taylor & Kent, 2014), which is often measured by analytical tools that count strategic communication activities such as directed communication (e.g., @), likes, favorites and follows on channels such as Twitter and Facebook. By defining engagement solely as an interpersonal interaction, this conceptualization fails to capitalize on the strengths of a social media channel like Twitter, which includes asymmetrical communication, network features, and the ability to reach mass audiences. An alternative and more theoretically oriented approach to analyzing social media engagement is to consider communication as dialogue. Dialogic communication is concerned with the co-creation of meaning and understanding between an organization and its publics, which ultimately serve to create and maintain relationships among multiple stakeholders. Within this context, dialogue is an outcome, rather than a set of procedures, and is the result of relationship building which occurs due to continuous communication (Kent & Taylor, 2002). While interaction is an important component of dialogue, dialogic communication encompasses far more than what can be captured by traditional analytical tools to include a focus on empathy, supportiveness, and commitment between an organization and its publics, with dialogue becoming the result of specific communication procedures and collaboration with the public (Kent & Taylor 2002). However, dialogic communication on social media has not been studied in the context of the NWS.
The NWS has begun to lay the groundwork for a continuum of communication for high-impact weather with a new program called Forecasting a Continuum of Environmental Threats (FACETs). This program seeks to replace product-centric methodologies with a modernized, science-based system, which will deliver a continuous stream of weather information extending from days to within minutes of an event (Executive Summary, Page 2). One key goal of FACETs is to facilitate “effective response” to NWS hazard messages, including all factors leading up to the receipt of the message (e.g., education, preparedness, situational awareness, understanding, response, and recovery). With these proposed changes and goals via the FACETs program, the NWS is laying the groundwork for ongoing engagement with the public, necessitating a dialogic, process-oriented communication strategy.
In this paper, we add to the scholarship on risk communication by examining Twitter use across the weather communication continuum in order to identify message strategies that enhance engagement during threat and non-threat periods. Specifically, we identify the messaging content that is shared over time with a specific focus on the contextual effects of threat and non-threat periods and develop a theoretically grounded model of social media engagement that takes into account these two distinct time periods
For this study, we selected 12 local NWS weather forecasting offices (representing communication in the NWS six regions) and investigate their communication strategies for three continuous months on Twitter (April-June, 2016). Threat and non-threat periods were identified by linking weather products that indicate the presence of an active threat (watch, warning, advisory) to the messages distributed by these 12 accounts. Threats represented during this three month period include flood, high surf, wind, tornado, fire weather, fog, severe thunderstorm, tropical storm, cyclone, winter weather, frost, freeze, and heat. Tweets per account during this period ranged from 69 (@NWSHonolulu) to 2213 (@NWSNorman) (total N = 10,784). Using content analysis methods, we develop a classification scheme informed by previous social media coding research including new media and engagement literatures (Lovejoy and Saxton, 2012; Waters & Williams, 2011) and risk communication (Sutton et al. 2015). We identify three primary message categories (action, information, and dialogic) and a set of subcategories for the two time periods of interest, showing the messaging strategies for preparedness and response (action), weather facts and forecasting (information), and community building and situational awareness (dialogic). We also code for message design features that indicate symmetrical communication (@user and requests for personal observations of weather and visible environmental impacts). We provide descriptive frequencies for each of the codes, and offer visualizations of social media engagement over the course of the threat/non-threat communication continuum.
From this analysis, we present a social media engagement model, which suggests the use of an ongoing communication strategy, which incorporates multiple messaging types, across the threat/non-threat time continuum. The findings from this research will help to inform messaging effectiveness to help increase organizational trustworthiness and credibility as part of an engagement strategy that is directly relevant to the next generation messaging paradigm represented by FACETs.