Extant research shows that individuals who discuss politics and current events with their peers also participate more actively in civil society. However, this correlation is not sufficient evidence of causation due to a number of analytical biases. To address this problem, data were collected through a panel study conducted on students who were randomly assigned to dormitories during their first year of college. In addition, the data were preprocessed before analysis with a matching procedure. These data show that discussing politics and current events caused these students to participate in civic activities during their first year of college. A follow-up study conducted on the same population during their fourth year of college shows that the positive effect of civic talk on civic participation still exists despite the passage of three years. Further analysis shows that the boost in civic participation initially after engaging in civic talk is the mechanism by which the effect of civic talk lasts into the future. (Contains 5 tables, 1 figure and 11 notes.)