Lies to Friends, Truths to Strangers: Preference Falsification 2.0 in China

• In Event: Information and Misinformation in Autocracy: Evidence from China

Abstract

Against predictions, authoritarian regimes survived the appearance of social media and other new information and communication technologies. The dominant paradigm for the social media in authoritarian regimes focuses on how the state manipulates the flow of information. As recent scholarship shows, online strategies of dictators include censorship, selective responsiveness, contamination, and mobilization. In this paradigm, the state is a key player: simply put, it delivers the punishments and rewards for political talk.

This paper shifts the paradigm by initially moving the state into the background to focus on how people deter one another from expressing political discontent. As a theoretical point of departure, I link a simple fact and an established proposition. First, as a simple fact, the social media allow individuals to express their political opinions anonymously to an audience of “strangers,” those to whom they are unconnected. Second, as the theory of preference falsification posits, individuals may lie about political preferences, especially for fear of reputational sanctions by “friends,” those to whom they are connected. Linking the two, I theorize that, in political talk, individuals lie to “friends” online (and, as originally posited, offline) and tell the truth to “strangers” online to minimize losses from reputational sanction by others in society.

I theorize that this strategic expression in online and offline political talk has important implications for an authoritarian state: it allows the state to collect true information at low risk of collective action against it. On the one hand, as findings on the social media echo chamber show, truthfully expressed preferences to “strangers” have greater difficulty finding an audience and are more likely to reveal polarization; on the other hand, as formal models of political communication under dictatorship show, polarization on the social media discourages collective action against the state. In sum, not only can the authoritarian state use truthfully expressed preferences among “strangers” on the social media to inform its policies, but also strategic political expression disorganizes collective action efforts.

I test the theory and its implications with an original large dataset of social media political talk I collected from zhihu.com, one of the largest social media sites in China, which contains high-quality political deliberation. I apply text-as-data methods including word embedding, named-entity recognition, and sentiment analysis to measure political attitudes in over 2 million pieces of political expression contributed by 1.8 million zhihu.com users. I then use methods of matching and weighting for high-dimensional data to perform causal identification. Along the way, I propose a new workflow of text data analysis with low computational cost, high replicability, and high interpretability for social scientists.

My theory provides a micro-foundation for information control in authoritarian regimes in the social media era. Distinctively, it implies that the dictator’s classic trade-off between information scarcity and collective action threat can be alleviated by strategic preference expression in society. It thereby provides a distinctive answer to the broader question of how authoritarian regimes have survived and thrived since the appearance of social media.

Author

• Haohan Chen, Duke University