How do events in the real world affect sentiment and user behaviour on social media? In recent years, TikTok has emerged as a significant force in political campaigning, reshaping how candidates influence public opinion. As the 2024 US presidential elections approaches, researchers at SnT studied the impact of Joe Biden’s withdrawal from the race on the social media platform.
A turning point for sentiment and engagement on TikTok
The study, spearheaded by Yuwei Chuai and Gabriele Lenzini, encompassed an extensive review of 680,609 TikTok videos, accruing over 4 billion views, 23 million comments, 31 million shares, and 335 million likes from November 1, 2023, to October 6, 2024.
These videos included official political party content as well as content generated by individuals, where either the Republican or the Democratic candidate was mentioned.
The goal of the study was to understand the effect of Joe Biden’s withdrawal from the US elections 2024 on user sentiment on TikTok. Positive sentiment is based on the video description and voice-to-text transcript, these are included in the original video post. This sentiment can potentially affect the sentiments in comments to the post, which is worth to explore in the further.
Before Joe Biden withdrew from the race, videos that featured Republican candidate Donald Trump showed higher positive sentiment and lower negative sentiment compared to those that featured his Democratic opponent. But the scenario flipped dramatically post-Biden’s exit: “Positive sentiment in video messages featuring the Democratic candidate rose by 46.8% and negative sentiment fell by 52%” Chuai observed.
The increase in engagement was particularly notable. Videos mentioning either the Democratic candidate (Kamala Harris) or the Republican candidate (Donald Trump) consistently received more engagement, being 53.3% more likely to be shared and 77.4% more likely to be liked.
From Data Collection to viral campaign strategies
The researchers developed a rigorous methodology for collecting data from TikTok’s open API. They set up search queries to randomly select videos mentioning either the Republican or Democratic candidates and collected data multiple times. This approach helped minimise bias and ensure a representative sample of political discourse on TikTok, capturing a wide array of user interactions and reactions for over a year.
However, this study focuses solely on TikTok, whose API, or official data access interface, is open to scientists. The findings could differ across different online platforms, highlighting the need for more social media platforms to provide accessible data for scientific research.
“The current closure of certain APIs is a concern” says Valérie Schafer, a Professor at C²DH who specialises in the history of computing, telecommunications and data networks. Schafer led a recent research project on the history on online virality and noted that some analyses conducted on social media (and notably on Twitter) would no longer be possible.
She also highlights the need for qualitative studies, for instance on user creativity. Schafer refers to the reactions to the false claim made by Trump during ABC’s Presidential Debate, asserting that illegal immigrants had been eating domestic pets. This allegation triggered a strong wave of online reactions, memes, parodic videos and songs.
“This presidential campaign clearly reinforces the existing trend of fake news and going viral. There are some interesting, newer elements, like the strong participation of influencers or live streaming on platforms like Twitch. New technologies are also in the mix: audio and visual deepfakes and AI-generated content are clearly present in the campaign” develops Schafer.
There have been many twists and turns in recent weeks, the outcome of the US presidential election remains uncertain. Next Tuesday will paint a clearer story.
About SnT’s team
Yuwei Chuai is part of the Sociotechnical Cybersecurity Research Group (IRiSC) at SnT led by Gabriele Lenzini. IRiSC integrate methods from social sciences and legal compliance into computer science to conduct research on sociotechnical cybersecurity. This research project is supported by the Luxembourg National Research Fund (FNR) and Belgian National Fund for Scientific Research (FNRS), as part of the project REgulatory Solutions to MitigatE DISinformation (REMEDIS).
About C²DH’s team
Valérie Schafer is a Professor in Contemporary European History (EIH) and Head of this research area at the C²DH since 2020. Between 2021 and 2024, Valérie Schafer lead the FNR-supported research project “A history of online virality” (HIVI). It aims to reveal and historicise the heritage of online viral content.