Abstract: Shifting Paradigms in Vaccine Discourse: A Comparative Analysis of Pre- and Post-COVID-19 Conversations on Twitter

◆ Fan Wang, University of Texas at Austin

The global health crisis precipitated by COVID-19 has dramatically altered the landscape of health communication, particularly in vaccine discourse. This study examines the evolution of vaccine-related conversations on Twitter, focusing on the transition from general vaccination discussions to the intensified debates surrounding COVID-19 vaccines. Our literature review encompasses research prior to the pandemic (Sak et al., 2016; Fu et al., 2016; Faasse et al., 2016) and extends into the COVID-19 era, providing a comprehensive understanding of the historical and current vaccine discourse dynamics.

Acknowledging the pre-existing anti-vaccination sentiment (Benoit & Mauldin, 2021), we critically analyze how these views have been amplified during the pandemic. The study employs social network analysis and linguistic inquiry to explore the interactions and linguistic patterns within pro- and anti-vaccine communities on Twitter. Two seed accounts were strategically chosen for their prominent roles and influence in these communities: @gavi, a pro-vaccine NGO (joined Twitter in December 2009 with 164.3K followers), and @ChildrensHD, an anti-vaccine NGO (joined in September 2016 with 132.6K followers). These accounts provided a focused lens through which to examine the vaccine debate.

The data collection spanned from 2015 to 2022, deliberately chosen to encapsulate the evolution of discourse both before and during the COVID-19 pandemic. This timeline allows for an analysis of the discourse shift and the impact of the pandemic on public opinion about vaccines. The dataset, comprising 5,646 tweets, revealed a disparity in the volume of tweets from the pro- and anti-vaccine communities, potentially reflective of the differing levels of online engagement and activity in these groups (Durmaz & Hengirmen, 2022).

Using the 'Clout' scale of Linguistic Inquiry and Word Count (LIWC-22) (Boyd et al., 2022), we evaluated the confidence levels expressed in the tweets. Our findings indicate an inverse relationship between word count and confidence, suggesting that tweets with fewer words exhibit higher confidence. This was substantiated by a regression analysis where word count emerged as a significant predictor of the Clout score.

Our analysis offers a detailed examination of the Clout scores over time, showcasing how vaccine-related discourse has transformed, especially in the wake of COVID-19. The study not only highlights the shifting paradigms in health communication on social media but also underscores the importance of understanding these trends for effective public health messaging during and beyond global health emergencies.