How do autocrats maintain power? In many cases, autocrats lack the support of a majority of the population. This problem is particularly stark in electoral autocracies, where autocrats must generate millions of favorable votes in order to stay in power. Coercion, or the forcible exclusion of some segments of the population from power, is one tool that many autocrats use to solve this problem. However, creating coercive institutions is also dangerous for autocrats, as the same forces that can be used to coerce citizens can also be used to depose or demand resources from the autocrat himself. In the first paper, I contend that autocracies can use the psychological effects of fear to coerce citizens at a lower cost and at lower personal risk. This psychological theory of autocratic coercion has two core implications that I test. First, I use a lab-in-the-field experiment to show that the emotion of fear reduces participation in pro-democracy action, and that this may work through its effects on perceptions of risk and risk attitudes. Second, I show using correlational evidence that propensity to feel fear predicts variation in participation in dissent. In the second paper, I examine how poverty conditions the way that citizens respond to the threat of coercion. I argue that poverty may make coercion more effective in reducing citizen dissent both by making citizens more prone to fear, and by increasing their physical vulnerability to violence. I test this prediction at the individual and constituency level using data on public opinion and voting in Zimbabwe, drawing on random variation in recent exposure to violence and poverty. The third paper tests whether emotions can also be used by activists to increase dissent among citizens with anti-regime preferences. I partnered with an opposition party that ran an experimental test of angry against enthusiastic campaign messages using video and images sent out via mobile phone chat groups. Analysis of the transcripts of these groups shows that the anger appeals generated significantly more pro-opposition participation in the groups. There is some evidence that anger was most effective in constituencies that had experienced violence in the past.