As political conversations shift online, Chloe Wittenberg PhD ’23 is learning how the information Americans consume shapes their attitudes and beliefs. An MIT postdoc in political science who recently earned her doctorate at the Institute, Wittenberg is interested in comparing the persuasive powers of video-based political content to text-based content. As Americans increasingly turn to social media platforms like YouTube and TikTok for their news, her work takes on greater urgency.
“The American political system is founded on the notion of an informed, engaged public, so we should strive to understand both what citizens think and how they come to these conclusions,” says Wittenberg.
Through original surveys and experiments, Wittenberg explores what kinds of political content Americans find arresting — and convincing. “As information systems continue to change and evolve, it is increasingly important for political scientists to take stock of the impacts of different forms of political content on public opinion,” she says.
Data tell the story
Wittenberg came to MIT in 2017, she says, to train in “rigorous and cutting-edge methods” that would help her trace the role of the media in shaping political views. She found her research home in the MIT Political Experiments Research Lab, directed by Adam J. Berinsky, professor of political science.
“Political beliefs and opinions are the product of complex, social processes,” she says. “Survey and experimental methods can help us dig deeper into these processes to extract new and actionable insights.”
In the face of rising misinformation and political polarization, Wittenberg wanted to know “where political beliefs come from in the first place.” She found social science answers that pointed to “features of the content itself, such as its topic, slant, or framing” plausible, but incomplete. This prompted her to consider not just what media individuals were consuming — but also how they were consuming this information.
As she began pursuing these questions, Wittenberg became interested in a growing phenomenon: deepfakes, synthetic videos that blur the line between truth and fiction. A popular consensus was rapidly emerging that deepfakes “posed a new, unprecedented threat, because the technology could be uniquely capable of shaping people’s attitudes and beliefs,” she says. But that assumed that video itself was inherently powerful, capable of greater persuasion than other forms of information, such as text. Was it really? Wittenberg had found her dissertation topic.
Immersive versus persuasive
Through a series of projects, Wittenberg began testing widespread assumptions about video’s singular sway as a channel for political persuasion. In conducting this work, she closely collaborated with Berinsky, as well as colleagues outside of political science such as David G. Rand, the Erwin H. Schell Professor and Professor of Management Science and Brain and Cognitive Sciences, and Ben M. Tappin, a research affiliate at Rand’s Human Cooperation Lab and an Early Career Fellow at the Leverhulme Trust.
Her first project, done in partnership with Berinsky, Rand, and Tappin, investigated the effects of presenting political information in the form of either the original video or a text-based transcript of the video. Across two complementary survey experiments involving more than 7,500 Americans, Wittenberg and her co-authors examined responses to nearly 75 professionally produced political advertisements and online news videos, spanning such topics as climate change, gun control, income inequality, and the Covid-19 pandemic. Their analysis revealed that respondents perceived videos as much more credible than text, but were only slightly more likely to change their attitudes and behavior in response to these videos.
“Just because video seems more credible than text doesn’t mean it’s more persuasive,” says Wittenberg. “Seeing may be believing, but it does not necessarily change people’s minds.”
Her next stop was determining whether the changes in attitude effected by both video and text persisted over time. In a two-part experiment with over 2,000 Americans, Wittenberg found that political video and text messaging functioned equally well in shifting respondents’ policy opinions — both immediately and several days later. “Although the effects of video and text both eroded slightly over time, there were virtually no differences between these two types of media in either part of the study,” she says. “This upends the assumption that video makes people more receptive to political information in both the short and long term.”
Drawing a bead on belief
Wittenberg has long been curious about how people come to hold strong attitudes about political issues. During her undergraduate studies at Swarthmore College, coursework in political science and psychology, she says, “opened my eyes to a whole new way of understanding political belief systems.”
While at Swarthmore, she participated in an interdisciplinary program devoted to understanding the roots of peace and conflict through a political, psychological, social, and cultural lens. Her thesis looked at how people form opinions about conflicts abroad and examined the role of media in shaping those opinions. To do so, she combined laboratory experiments with in-depth text analysis to help her probe the connections between media narratives and public opinion.
“Through this work, I realized that I really enjoyed the process of puzzling through complex theoretical questions,” she says. “It also reinforced to me the importance of multidisciplinary approaches to social science research, which I have continued to use during my time at MIT.”
After three years working as a research analyst at the Center for Effective Philanthropy in Cambridge, Massachusetts, where she honed her survey design and analysis skills, Wittenberg decided to return to an academic setting “to advance her knowledge of best practices in public opinion research.”
The persuasive advantage
In the final phase of her dissertation, which she defended last summer, Wittenberg shifted focus. Rather than weighing the power of video versus text in altering people’s political attitudes, she investigated the possibility “that video’s persuasive advantage” might lie in its unrivaled capacity to capture and hold attention. In initial studies, she finds that individuals are indeed captivated by and highly likely to engage with political videos — but continue to find conventional political media, such as news articles, compelling.
Wittenberg, who continues her work as a postdoc in the Department of Political Science, believes that a nuanced understanding of public responses to different forms of political messaging could help inform debates about optimal strategies for political persuasion. “My research has clear, practical implications for campaigns and advocacy groups seeking to sway public support on political topics,” she says. However, she believes her work can also provide more general insights about the role of social media in American politics. “There are a lot of different directions to go with this research, which is exciting.”