Exploring influences on user engagement in online platforms : three studies on user-generated content

Abstract

With the prevalence of online platforms, user-generated content (UGC) is developing rapidly, creating new jobs, markets and policy needs. UGC has evolved into a more dynamic form that demands users’ attention and continuous consumption. While existing research has predominantly focused on the influence of sentiment on online user engagement, there is less understanding of how dynamic sentiment patterns influence subsequent discussions. Additionally, despite abundant research on internal factors on engagement in online communities, the impact of external event on user engagement in online communities remains unclear. Moreover, as regulations tightened, there is an increasing need to understand legitimate strategies for companies and platforms to foster positive UGC. My dissertation presents three studies that address these questions, enhancing our understanding of UGC in the modern digital era. In the first study, I examine the effect of sentiment congruency on subsequent discussion by drawing on the priming theory. Utilizing a dynamic panel model, I empirically test my hypotheses using data from a popular online automobile forum in Asia. The empirical evidence demonstrates that higher sentiment congruency can motivate shorter response interval, more positive sentiment, and increased post volume in subsequent discussions. Additional exploration of contingent factors suggests that sentiment congruency effect is stronger in discussions that primarily comprise a higher proportion of inquiries and in relatively later discussion phases. This study highlights the importance of content and display sequence of user posts, providing valuable implications for platform designers aiming to boost trending topics in online discussions. In the second study, I investigate the impact of public negativity on engagement within online fan communities. Leveraging natural experiment design and weighted regression discontinuity in time model, I explore the public negativity effect using data from three online fan communities. The results suggest a decrease in comments and increase in likes when facing public negativity, suggesting reserved engagement within online fan communities. Moreover, I examine the common assumption of members’ homogenous responses by exploring the moderating influence of member types, demographic characteristics, and status characteristics. This study highlights the potential risk associated with making engagement in online fan communities visible to the general public, providing valuable insights for celebrities, influencers, entertainment companies, and platform designers. In the third study, I explore the emerging industry of digital serial publications, where publishers release creators’ content incrementally, and consumers make rating decisions with each new update. Using an analytical model, I investigate how publishers can use preview strategies to increase the equilibrium of user rating when the market reaches a static state. I find that previews and rating equilibrium follow a U-shape pattern, and the optimal preview strategy depends on rating value and market scale. Theoretically, these findings contribute to user-generated content (UGC) literature, particularly in terms of sentiment in UGC. They provide new insights into the dynamic sentiment patterns, influence of external events, and company strategies to enhance content generation. These insights deepen our understanding of the evolving landscape of UGC.