Geographic natural experiments with interference: The effect of all-mail voting on turnout in Colorado

We analyze a geographic natural experiment during the 2010 Colorado primary election in the USA, when counties in the state of Colorado had the option to have an all-mail election or retain traditional in-person voting on Election Day. The town of Basalt, in the southwestern part of the state, is split in half by two counties that chose different modes of voting. Our research design compares these two counties to understand whether turnout levels were altered by all-mail elections. Our analysis considers the possibility that social interactions may lead to spillover effects-a situation in which one unit's outcome may be affected by the treatment received by other units. In our application, treated and control voters lived in very close proximity and spillovers are possible. Using the potential outcomes framework, we consider different estimands under the assumption that interference occurs only when treated individuals are in close geographic proximity to a sufficiently high number of control individuals. Under our assumptions, our empirical analysis suggests that all-mail voting decreased turnout in Colorado, and shows no evidence of spatial interference between voters.