Network, nexus, nonsense : infodemic studies on vaccine hesitancy in Taiwan

Abstract

Vaccine hesitancy is a growing, global public health problem. No more has this been more apparent than after the first pandemic of our information age. The World Health Organisation (WHO) recognised these trends, proposing a new “infodemiology” research agenda studying information transmission during outbreaks, integrating social media studies into public health research. This thesis uses forum data from PTT and social network analysis to study this growing problem in the context of COVID-19 in an understudied region: Taiwan.

The first chapter, “network”, describes how communities form among different vaccine stances, and how their engagement behaviour unfolds temporally. Using a multilayer network model where each layer represents a vaccine stance, I compare and contrast across three stances of pro-vaccination, vaccine hesitant, and anti-vaccination. I find that the anti-vaccination layer is more insular, and forms tighter clusters of communities. In addition, their engagement behaviour is more bursty compared to the other two layers.

The second chapter “nexus” measures the existence of “echo” chambers – with conceptual clarification on the term – across the three stances. Using the same multi-layer analysis, I analyse this on the layer level but also user level to understand the drivers, focusing on influential users. I find while there are interactions that promote chambering within layers, there are also those that span across layers, suggesting simultaneous mechanisms. Identifying which users are responsible for these mechanisms, and within which vaccine stance, is crucial for information management.

The third chapter “nonsense” assesses the diffusion of misinformation through measuring polarised engagement of either true or false information for influential users. Findings show that while the breadth of misinformation may be larger than true information, its speed is slower. Despite this, there are key users functioning as brokers that spread misinformation with their increased engagement with it. Identifying and reducing the connectivity these users have in the network may prevent cascading misinformation chains.

The last chapter maps the work of the three empirical chapters onto the WHO infodemiology research agenda, all the while discussing limitations of infodemic studies and generalisability of these studies.