Enhancing a geographic regression discontinuity design through matching to estimate the effect of ballot initiatives on voter turnout

type="main" xml:id="rssa12056-abs-0001"> Ballot initiatives allow the public to vote directly on public policy. The literature in political science has attempted to document whether the presence of an initiative can increase voter turnout. We study this question for an initiative that appeared on the ballot in 2008 in Milwaukee, Wisconsin, using a natural experiment based on geography. This form of natural experiment exploits variation in geography where units in one geographic area receive a treatment whereas units in another area do not. When assignment to treatment via geographic location creates as-if random variation in treatment assignment, adjustment for baseline covariates is unnecessary. In many applications, however, some adjustment for baseline covariates may be necessary. As such, analysts may wish to combine identification strategies—using both spatial proximity and covariates. We propose a matching framework to incorporate information about both geographic proximity and observed covariates flexibly which allows us to minimize spatial distance while preserving balance on observed covariates. This framework is also applicable to regression discontinuity designs that are not based on geography. We find that the initiative on the ballot in Milwaukee does not appear to have increased turnout.