How do individuals in non-democracies organize collective action? Throughout history, it is common for those in power to structure society so as to make mass action highly unlikely; yet, from peasant revolts in the Middle Ages to urban arisings during the Arab Spring, individuals surmount these barriers. This dissertation argues that they surmount these barriers because mobilization occurs through interpersonal connections that connect individuals. These social networks transmit information necessary for protest mobilization, while state repression is most effective against institutions and prominent individuals.
This argument is tested in the context of the Arab Spring. In Chapter 1, 13.8 million geolocated messages and daily data on protests are gathered for the Middle East and North Africa from November 1st, 2010 through the end of 2011. These data reveal that protest mobilization correlates with coordination from individuals on the periphery of their country’s social network, not those who are prominent. Because the identification strategy for who is prominent in a network may average away impor- tant differences, Chapter 2 focuses on activists in Bahrain and Egypt. A supervised topic model on those activists’ messages still shows no effect for coordination; this finding reinforces Chapter 1 since these individuals are those most likely to affect protest mobilization. Qualitative evidence suggests that their effect occurs through generating common knowledge about mass dissatisfaction before protests occur.
Chapters 3 and 4 focus on methodological points. Chapter 3 introduces a technique for using Twitter data to analyze network structure as it evolves. Analyzing network structure over time requires data on the entire network, which is difficult for two reasons. First, Twitter imposes limits on acquiring data that make it effectively impossible to get complete network data. Second, those same restrictions make it effectively impossible to analyze that network as it changes. This chapter introduces a measure that can be collected rapidly enough from Twitter to obviate these issues. Chapter 4 is a general purpose introduction to Twitter. It explains what questions Twitter data can answer, how to acquire those data, and programming approaches.