Social context theory suggests that an important driver of political participation is the behavior of family, friends, co-workers and neighbors. How do social ties between individuals shape equilibrium behavior in larger populations? Despite theoretical inroads into this question, direct empirical tests remain scarce due to data limitations. We fill this gap using full social network data from 15 villages in rural Uganda, where village-level turnout is the outcome of interest. We find that levels of participation predicted by structural features of village networks are strongly associated with actual village-level turnout in low-salience local elections, and weakly associated in high-salience presidential elections. We also find that these features predict other forms of political participation, including attending village meetings and contributing to village projects. In addition to demonstrating that networks help explain political participation, we provide evidence that the mechanism of influence is that proposed by social context theory rather than alternative mechanisms like the presence of central brokers or the ability of networks to diffuse information.