This thesis develops a model that estimates the number of participants in a protest event. My central assumption suggests that those engaged in the same collective action should have similar trajectories during the event. Individual trajectories of those who are away should not express the same patterns in trajectories. I developed a Convolutional Neural Network (CNN) model with two input layers. The first layer contains data on the trajectories of easily identified subgroups of protesters and easily identifies subgroups of nonprotesters. The second layer indicates if a specific part of each trajectory is not observed directly but, instead, is imputed. The resulting classifier identifies citizens who share the same trajectory during the protest with an accuracy of 94 percent.

The proposed algorithm significantly outperforms these alternatives: Logistic Regression, Random Forest, and Artificial Neural Network. The advantage of CNN in this setting comes from the fact that this architecture applies feature-recognition filters to all segments of the trajectory and thus is robust to potential shifts among various trajectories. These shifts exist because protesters do not appear at the same time or in the same place. The resulting estimates are validated using estimates of protest events from the national press.

This dissertation explores the strategies that modern authoritarian leaders use to survive in office. Unlike many 20th century dictators, today's autocrats must operate in the "spotlight'' — new media and information technology enable the political opposition and the public to observe their actions. This greater observability limits the effectiveness of government repression, sometimes forcing the authorities to shift to other tools of political control. I study two of these alternative tools: the staging of pro-government rallies to create an image of invincibility and the recruitment of armies of paid supporters to shape the narrative on the Internet and disrupt online conversation.

To explore these strategies, I focus on the case of Vladimir Putin's regime in Russia. I argue that, faced with a wave of anti-government protests, an autocrat such as Putin can discourage further demonstrations by organizing pro-government rallies that — perhaps surprisingly — convey credible information to regime opponents about the dictator's popularity. Moreover, this discouragement effect will be stronger — under certain conditions — if the autocrat allows some media freedom. I test this theory using data I collected on which Russian cities had access to broadcasts of the independent radio station, "Echo of Moscow.'' Combining matching techniques with a difference-in-differences design, I compare protest dynamics in the cities that received broadcasts and in those that did not.

To better understand the second strategy, I explore the behavior and impact of several hundred "trolls'' — paid supporters of the regime who are allegedly employed to leave pro-government comments on social media platforms. Using probabilistic topic modeling, I develop a method to estimate the causal effect of troll interventions in online discussions. I find that trolls are able to successfully divert online discussions from politically charged topics, but are ineffective in promoting a pro-government agenda. In a separate chapter, I develop a methodology for the study of such Internet actors. Specifically, I devise a set of classification models to detect paid "political commentators.''