How Public Health Agencies Break through COVID-19 Conversations: A Strategic Network Approach to Public Engagement

ABSTRACT

In times of public health emergencies, health agencies need to engage and communicate with the public in real-time to share updates and accurate information. This is especially the case for the COVID-19 pandemic where public engagement can potentially save lives and flatten the curve. This paper considers how the use of interactive features and strategic network positions of health agencies on social media influenced their public engagement outcomes. Specifically, we analyzed 203 U.S. public health agencies’ Twitter activity and the public engagement they received by extracting data from a large-scale Twitter dataset collected from January 21st to May 31st, 2020. Results show that health agencies’ network position in addition to their two-way communication strategy greatly influenced the level of public engagement with their COVID-19 related content on Twitter. Findings highlight the benefits of strategic social media communication of public health agencies resides not only in how agencies use social media but also in their formation of network position to amplify their visibility. As official sources of health and risk information, public health agencies should coordinate their social media communication efforts to strategically position themselves in advantageous network positions to augment public engagement outcomes.

Notes

1. To ensure that it was not determined by the order of entering star network and community network position, we tried the other way around by adding community network variables first and then star network variables: Significant change in variance from Model 2 (interactivity) to community position by 4%, F(2, 194) = 10.634, p < .001 with significant predictive power of average neighbor in degree, b= 0.159, p< .001 and k-core, b= 0.193, p < .001. Subsequently, adding star position variables improved the model by 12%, F(2, 192) = 39.014, p < .001. Results were consistent such that greater model change was observed in the star network position than community network position with the predictors remained significant.