Witnessing Political Protest on Civic Engagement∗

How does physically witnessing a protest in a democratic society affect citizens in authoritarian societies? Existing data are unable to answer the question by failures to capture witnesses, to construct meaningful comparison groups and to obtain information on pre-protest political behaviors. Using a quasi-experiment design, we report a first causal estimation on impact of 13 protests in Hong Kong from 2012-14 on witnesses from mainland China, We used geocoded posts from Chinese social networking site to collect a panel of Chinese users who visited Hong Kong but at different timing: treated users were physically close to one of the protests when it occurred; control users already left Hong Kong before protests and could not witness protests. With the control group and users’ ex-ante posting history, difference-in-differences methods are used to estimate changes of intensities and issues of discussion of civic and political problems. Treated users discussed more civic and political problems by 40.39% (95% confidence interval: 3.62%, 75.48%) after protests, relative to the change of control ∗Direct all correspondences to Han Zhang, Department of Sociology, Princeton University. Address: 125 Wallace Hall, Princeton University, Princeton, NJ, 08544. Email: [email protected]. Special thanks to Xing Xie and Nicholas Jing Yuan for the support of data collection and analysis. We thank Matthew Salganik, Paul DiMaggio, Miguel Centeno, Andreas Wimmer, Rory Truex, Rene Flores, Winston Chou, Clark Bernier, Kyle Chan for helpful comments on research design; Mark Beissinger, Ran Liu, Lynn White, for feedbacks. We would also like to thank participants in various seminars at Princeton: Digital Methods in Political Science Workshop, Political Methodology Seminar, as well as Contemporary China Colloquium.Draft version, please not spread without authors’ permission. †[email protected]