Doctoral Dissertation Research: Information, Public Opinion, and Behavior

Regents Of The University Of Michigan - Ann Arbor, Ann Arbor MI

Investigators

Abstract

This project explores how political information manipulation in authoritarian regimes can affect the behaviors, attitudes, and opinion of individuals. This project is important to our understanding of how authoritarian regimes gain popular support and maintain control over populations using social media and big data. Planned survey experiments will measure the effect of political information manipulation tactics (censorship, astroturfing, and digital repression). As much of social interaction has moved into a digital space, governments have greater access to personal and social data. At the same time, the fields of artificial intelligence and machine learning have experienced unprecedented growth along with a similarly rapid growth in the power of microprocessors. These advances have given governments better tools to digest, process, and make informed decisions based on these massive stores of personal data. This project is a first step toward understanding how these common data-driven information manipulation tactics work and impact individuals in authoritarian governments, essential knowledge for scholars of modern politics and public opinion in a highly networked and digital age. In this project, the effects of experiencing political information manipulation on individual behaviors and opinion in authoritarian regimes are explored using both survey experiments and large-N correlational analyses. It is hypothesized that government information manipulation that is covert---when the identity of the government as information manipulator obscured---will increase the magnitude of individual-level effects in a direction favorable to the government. More specifically, when compared to more overt government information manipulation, covert information manipulation will be more persuasive, will more dramatically increase positive perceptions of the state, will more dramatically decrease one's sense of political efficacy, and will decrease one's willingness to express opinions more than overt manipulation. To test these hypotheses, a pseudo experiment will be run using a large database of syndicated articles identified using automated near-duplicate detection. Syndicated state media articles, appearing on different news websites with identical text but different comments, will be compared across a treatment group (articles with comments from astroturfers and ordinary users) and a control group (articles with comments from only ordinary users). Variables targeting each of the hypotheses above will be drawn from the text using statistical and natural language processing methods. Additionally, three survey experiments will be conducted to test, respectively, the effect of censorship, astroturfing, and digital repression in overt and covert forms. This work is critical to understand how modern authoritarian regimes leverage social media to guide public opinion, mobilize regime support, and demobilize opponents. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.