Negativity drives engagement on political TikTok

A new study published in Computers in Human Behavior suggests that political videos on TikTok that criticize opposing political parties and use emotionally charged or uncivil language tend to generate higher levels of engagement. These patterns appear across both liberal-leaning and conservative-leaning accounts, though with some notable differences in how each group responds.

Short-form video platforms like TikTok have become key venues for political messaging, particularly among younger users. Traditional news outlets, political influencers, and grassroots organizations increasingly use the platform to share perspectives, raise awareness, and mobilize supporters.

Although many studies have examined political engagement on platforms such as Facebook and Twitter, there has been limited research on what drives user responses to political content on TikTok. Prior research has often relied on small samples or focused on specific political actors. The current study aimed to address these gaps by examining a much larger dataset and exploring how content features and account characteristics shape audience behavior on TikTok.

The researchers also sought to differentiate between types of engagement. They distinguished between internal actions, such as liking and commenting, and external actions like sharing, which signal a willingness to spread content beyond one’s personal viewing.

“I’ve been studying TikTok as a news and civic space since 2022. Over time, I noticed how news influencers and news organizations were using the platform to share information and shape public conversations. That really sparked my curiosity,” said study author Zicheng Cheng, an assistant professor in the School of Journalism at the University of Arizona.

Free daily newsletter

“I am interested in digging into these questions further: what kind of news or news-like content is being created on TikTok, and how do people respond or engage with it? Overall, my research explores how audiences interact with political and news content on TikTok and what those interactions mean for society and democracy. I approach these questions using a variety of methods, from computational content analysis and interviews to experiments.”

In the current study, the researchers analyzed more than 578,000 TikTok videos posted between 2019 and 2023 by nearly 10,000 elite accounts (defined as those with more than 10,000 followers). These accounts included a wide range of political actors, from news organizations and politicians to individual influencers and activist groups. The researchers used machine learning, topic modeling, and large language models to analyze the text, audio, and visual components of each video. They also assessed sentiment, civility, and content type.

One of the most consistent findings was that videos that criticized an opposing political party received the most engagement. This includes higher rates of likes, comments, and shares. Across the dataset, roughly one-third of videos featured such content, and it was the strongest predictor of engagement for both liberal and conservative accounts.

Videos that praised one’s own party also tended to increase likes and comments, although they had little effect on sharing. In contrast, videos that criticized one’s own political group sparked more conversation through comments and were more likely to be shared, but were not particularly well-liked. Notably, there were virtually no videos in the dataset that praised opposing parties, suggesting this type of cross-partisan messaging is extremely rare on TikTok.

The researchers also examined emotional features of the videos. Content with negative sentiment was more engaging overall, especially when it came to comments and shares. This pattern aligns with existing psychological research suggesting that people are more drawn to negative information. However, there were some differences across political lines. While liberals generally responded strongly to negative sentiment, conservatives were somewhat more likely to like content with a positive tone.

Videos that included uncivil language (such as personal attacks or insults) were also more engaging than those that used polite or neutral language. This trend held across both liberal and conservative audiences. According to the study, civility was one of the strongest predictors of commenting and sharing behavior. The researchers suggest this may reflect how confrontational or emotionally charged content captures attention and provokes stronger responses.

“We found that negativity drives engagement on political TikTok,” Cheng told PsyPost. “Videos that criticize the other political side, also known as out-party critique, consistently attract the most likes, comments, and shares. TikTok videos using uncivil language also tend to perform better, regardless of political leaning.”

Another feature that encouraged interaction was when creators explicitly asked viewers to like, comment, or share the video. These direct prompts were relatively common and consistently predicted higher engagement.

In terms of who posted the videos, individual influencers and activist groups tended to generate more likes and comments than legacy media outlets or politicians. Influencers likely benefit from the perception of authenticity and personal connection, which fits well with TikTok’s informal, creator-driven culture. Activist groups also saw relatively strong performance, particularly in terms of comments and shares.

Political figures, by contrast, generally saw lower engagement—except when it came to shares. Their videos were shared more often than other categories, suggesting they may serve as reference points or signals for supporters to spread messages to their broader networks.

“Interestingly, news outlets, whether legacy media or digital-native media, struggle to gain the same level of engagement as influencer accounts,” Cheng said. “Political videos from individual influencers tend to grab the audience’s attention much more than those from news outlets.”

The study also analyzed specific political issues. Topics like racial justice, student debt, child welfare, and the January 6 Capitol attack were among the most engaging overall. However, some topics showed distinct ideological patterns. Liberal accounts saw more engagement on racial justice and COVID-19 transparency, while conservative accounts engaged more with abortion rights and climate policy.

“We found that videos addressing specific political issues tend to generate higher engagement than those that didn’t address any concrete topics,” Cheng explained. “This suggests that substantive political content can still capture attention on a platform often known for entertainment.”

The researchers also found that conservative users were less likely to share videos about divisive topics, even when they liked or commented on them.

“Another interesting pattern came from conservative accounts,” Cheng told PsyPost. “We noticed that while conservatives frequently like and comment on videos about abortion and racial justice, they are much less likely to share them outside of TikTok. It’s possible that sharing represents a more public endorsement, and some users may hesitate to broadcast their views on such sensitive topics to a broader social network.”

Finally, the study found that liberal audiences tend to be more responsive overall to content features than conservative audiences. This includes stronger reactions to emotional tone, issue framing, and content with calls to action or justification.

“Overall, our findings suggest that political conversations on TikTok are often driven by emotional and confrontational content, which has important implications for how people experience politics and information online,” Cheng said.

While the study provides new insight into political engagement on TikTok, the authors note some limitations. Engagement metrics such as likes, comments, and shares do not necessarily indicate agreement or approval. Users may interact for various reasons, including irony or opposition, and these motives cannot be easily inferred from quantitative data alone.

“Our engagement metrics (number of likes, comments, and shares) mainly capture surface-level interactions,” Cheng noted. “That means, these numbers don’t necessarily fully reveal people’s deeper attitudes or beliefs. Future research could build on this by examining the content of comments more closely. For example, whether users are showing support, disagreement, or offering more nuanced perspectives.”

Despite these limitations, the findings offer a detailed and large-scale look at political communication on TikTok. They suggest that emotionally charged, partisan content continues to attract the most attention, raising questions about the platform’s role in shaping political discourse. The Political Content Engagement Model introduced in this study provides a framework for future research and a potential guide for communicators navigating the evolving landscape of digital politics.

“We plan to continue studying how people express and engage with politics on TikTok, particularly after the 2024 U.S. presidential election,” Cheng explained. “Future studies will take a closer look at the content within these news or news-like videos. One interesting direction is to examine whether political information shared on the platform is factual or contains misinformation or disinformation, and whether audiences are able to distinguish false information from factual content on TikTok.”

The study, “Political Content Engagement Model: A large-scale analysis of TikTok political video content features and audience engagement,” was authored by Zicheng Cheng and Yanlin Li.