
Award won:
- The 2024 Kaid-Sanders Best Article of the Year Award, Political Communication Division. ICA.
Name(s) & affiliation:
- Sang Jung Kim (Assistant Professor, School of Journalism and Mass Communication, University of Iowa),
- Isabel Villanueva (Doctoral Candidate, Life Sciences Communication, University of Wisconsin-Madison),
- Kaiping Chen (Associate Professor, Life Sciences Communication, University of Wisconsin-Madison)
Project title:
- Going beyond affective polarization: How emotions and identities are used in anti-vaccination TikTok videos
Co-authors (if any):
- Isabel Villanueva (Doctoral Candidate, Life Sciences Communication, University of Wisconsin-Madison), Kaiping Chen (Assistant Professor, Life Sciences Communication, University of Wisconsin-Madison)
Publication reference, link (APA 7th):
- Kim, S. J., Villanueva, I. I., & Chen, K. (2024). Going beyond affective polarization: How emotions and identities are used in anti-vaccination TikTok videos. Political Communication, 41(4), 588-607.
Tell us something about you/your team and how and why you decided to focus on this research
- Our team shares a strong interest in investigating how the multi-modal delivery of political communication shapes audience engagement. This interest was sparked by the widespread misinformation about COVID-19 vaccination on social media, a central issue during the pandemic. These messages, often characterized by their multi-modal and emotional nature, frequently emphasized particular sociopolitical identities, which appeared to amplify their impact on the public. Despite this, there has been a lack of sufficient empirical investigations into these dynamics. Recognizing this gap, we were motivated to focus our research on understanding the mechanisms of such communication and its broader implications for political communication. Notably, our team highlights how emotions influence engagement differently depending on the distinct modalities of delivery and emphasizes the role of various social identities beyond partisanship, such as relational identities that highlight roles like being a mother.
In 280 characters or less, summarize the main takeaway of your project.
- Our study expands the emotions-as-frames model, affective intelligence theory, and social identity theory by revealing how audiences engage with emotional appeals and sociopolitical identity narratives delivered through multi-modal TikTok anti-vaccination videos.
What made this project a “polcomm project”?
- Emotions and social identities are foundational concepts in political communication, grounded in a rich historical tradition. However, the interplay between the multi-modal nature of social media messages and the emotions and social identities embedded in political communication remains underexplored. Our paper positions itself as a novel ‘polcomm project’ by addressing this gap and advancing foundational research in the field.
What, if anything, would you do differently, if you were to start this project again? (What was the most challenging part of this project? …& how did you overcome those challenges?)
- The most challenging aspect of this project was identifying distinct emotions across the various modalities in TikTok anti-vaccination videos. In this paper, we utilized an off-the-shelf library based on the traditional machine learning model and a dictionary-based approach to identify emotional appeals. Moving forward, our team is enhancing this process by leveraging foundation models—large-scale, pre-trained models that provide a versatile framework, enabling them to be adapted for a wide range of tasks across different domains—to improve the accuracy and depth of emotion detection. We are particularly excited to explore how foundation models can advance the identification of emotional appeals in multi-modal messages.
What other research do you currently see being done in this field and what would you like to see more of in the future?
- An exciting and rapidly growing area of research utilizes various computer-assisted methods, including computer vision models, to explore theoretical concepts in political communication by examining large-scale, multi-modalmessages, such as emotional expressions. Furthermore, as the social media environment matures, scholars are increasingly focusing on its multi-platform nature and how political communication flows from one platform to another. In the future, we hope to see more research efforts integrating multi-modal and multi-platform analyses, examining how different types of media—such as text, images, and videos—interact across platforms to shape political narratives and influence public opinion.
- For instance, our recent study (Chen et al., 2024) investigates the portrayal of gender stereotypes on YouTube and TikTok within controversial science discourses. Our analysis reveals that while both platforms associate distinct multi-modal features with men and women, they diverge significantly in thematic emphasis: YouTube’s thumbnails frequently feature climate activists or women depicted in harmony with nature, while TikTok’s thumbnails often highlight women in Vlog-style selfies with feminine gestures. These findings underscore the nuanced role that platform-specific affordances and audience preferences play in shaping gendered narratives.
- In light of these insights, we urge political communication scholars to prioritize understanding how multi-modal and multi-platform messages interact to construct and propagate political communication phenomena. Adopting this comparative and holistic approach is critical for uncovering the complexities of digital political discourse and its implications for public opinion and engagement.
What’s next? (Follow-up projects? Completely new direction?)
- Building on our team’s focus on advancing computational methods to analyze multi-modal messages and uncover political communication phenomena, we are currently developing a sociotechnical framework for evaluating computer vision models and introducing new approaches to assess biases in the classification of gender and emotions within images in these models, including foundation models (Sha et al., 2024). Simultaneously, as mentioned earlier, we are continuing our efforts to understand how political communication reveals differences in multi-modal messages across various social media platforms, integrating multi-modal and multi-platform approaches to advance political communication scholarship. Looking ahead, we aim to collaborate across disciplines to develop metrics that enhance the transparency and fairness of computer vision models. By doing so, we hope to bridge the gap between computational advancements and their real-world applications in understanding and improving political communication across diverse platforms.