abstract: We investigate how college participation may differentially influence civic behaviors among individuals who were between 18 and 20 years old in 2012. Using data from the High School Longitudinal Study of 2009, we consider two direct measures of civic behavior, voter registration and volunteerism. We generate our estimates with propensity forests, a machine learning algorithm that can mitigate bias when using observational data and supports investigation of heterogeneous treatment effects. Overall, we find that college-goers are more likely to register to vote and volunteer, though, conditional on volunteering at all, likely to volunteer fewer hours. We find limited evidence of heterogeneous returns across various groups, suggesting that civic returns to higher education are broadly shared by those who attend.