Work in Progress

• Learning About Bias: An Experiment on News Consumption in Russia (With Georgiy Syunyaev)

  Most media that people consume exhibit certain political biases or slants. However, many citizens either do not understand or underestimate the slant of the media they consume. This study proposes a new experimental design to investigate whether making media slant more evident affects how citizens perceive news coverage, what news sources they consume, and how they form their political beliefs. Our experimental intervention fielded online among Russian citizens exposes respondents to news coverage by two major state and independent television channels and, at the same time, makes respondents more attentive to the slant of news reporting. Our panel design allows us to examine the subsequent impact of the intervention on respondents' perceptions of media, willingness to consume particular media outlets, political knowledge, and support for the government and its policies. We find that respondents who were exposed to news coverage in this way adjusted their beliefs about the slant and bias of state and independent media, reported greater awareness of news stories published by state and independent media, and revised their political views, expressing more concern about the government and its policies. In addition, respondents who were exposed to the coverage from an independent television station adjusted their news consumption, more often choosing to consume news from this independent outlet.

• Quantifying Narrative Diffusion Across Languages (With Hannah Waight, Solomon Messing, Margaret E. Roberts, Jonathan Nagler, Jason Greenfield, Megan Brown, Kevin Aslett, and Joshua A. Tucker)

  How can one trace the spread of information, ideas, and narratives across the world using text data? Social scientists have long sought to answer this question, which requires identifying pairs of documents that contain statements with the same underlying meaning about the same subject. Past approaches that rely on n-gram matching or topic modeling to date have yielded only a loose approximation to this ideal. We propose a method to track the global diffusion of information: first applying a highly scalable method called locality sensitive hashing (LSH) to cross-language embedded representations of text based on a large-language model (LLM) to generate a relatively small number of candidate pairs, then fine-tuning an instruct-trained LLM to identify the actual pairs of sentences that contain the same idea. It is extremely difficult to create a gold-standard labeled data set to evaluate performance for this pairwise problem--we do so by creating data set of thousands of benchmark sentence pairs that contain iterations of equivalent and different statements about the same and different topics. Our method has far higher recall than verbatim text reuse methods and is more precise than topic modeling. This approach can be applied to the study of propaganda, misinformation, diffusion of innovations. In this paper, we apply the approach to show how U.S. media sources reuse information from Russian state media in the context of the 2022 Russian invasion of Ukraine, for example, accusations that Ukraine is developing bioweapons.

• Combating Information Avoidance in Wartime Russia: A Multidisciplinary Field Experiment (With Julia A. Minson, Aaron Erlich, Jordan Gans-Morse, Christopher Higgins, and others)

  We report the results of two large-scale field experiments examining the effectiveness of email-based communication in the context of Russian information manipulation following its full-scale invasion of Ukraine. We contacted academics across the social sciences and asked them to produce messages to persuade Russian residents to watch a truthful video about the invasion. In partnership with the non-profit Mail2Ru, we randomly assigned eleven expert messages and two control treatments to approximately 260,000 email addresses. Only one message led to significantly greater engagement with the video relative to the control treatments, with 9 messages performing in line with the controls and one message underperforming them. We then successfully replicated the effectiveness of our top-performing intervention in a second experiment. Our work highlights the challenges associated with effective communication in the face of disinformation and government surveillance and point to the urgent need for experimental research in this area.

• Trust Building Across Identity Groups: An Experiment in Kazakhstan (With Yoshiko Herrera and Andrew Kydd)

  Mistrust is a common cause of conflict between individuals belonging to different identity groups. When can such mistrust be overcome? We study this question using an experiment based on a trust game between members of different social identity groups. In particular, we study the effect of hearing about positive interactions across group lines on the willingness of individuals to take a chance on cooperating with outgroup members. We field the experiment in Kazakhstan, focusing on relations between Kazakhs and Russians.

• The Effects of Social Media Censorship on Polarization (With Yoshiko Herrera, Mingcong Pan, and Yiming Wang)

  Political polarization is a pressing problem around the world and social media have come under scrutiny for their possible role in facilitating it. In democracies, some companies or governments restrict online expression or block access to certain platforms. In autocracies, digital restrictions are used widely to stem oppositional activity, often under the pretense of fighting extremism. But do digital restrictions reduce polarization, or do they exacerbate it? Existing work suggests both possible outcomes. We examine the effects of digital restrictions via a series of online survey experiments in the U.S., China, and other countries, focusing on how a threat of social media restrictions affects citizens' social identities, political polarization, and online behavior.

• Is Trust in Media Decreasing? Evidence from the World Values Survey

  View at SSRN

  Scholars and observers are increasingly concerned that confidence in mainstream media is decreasing, which could undermine the democratic political process and make citizens more vulnerable to manipulation. Various surveys have documented dwindling confidence in media in the U.S. and other democracies, but so far this research has not established whether this decline is a global trend. I suggest a more robust approach using data from the World Values Survey to establish worldwide trends in attitudes. The resulting analysis shows there has been some decline in confidence in media since the early 1990s, but there is no evidence of a decline among stable democracies. Confidence in media has been decreasing instead in democratizing countries and other states that have undergone substantial political changes.