Abstract

We identify an exogenous source of variation in exposure to campaign advertising in the 2000 presidential election, based on residence in battleground states. If exposure to campaign advertising makes a potential voter significantly more likely to vote, then we should see significantly greater turnout in battleground states. We do not. This result is robust to several specifications and evident in a natural experiment consisting of New Jersey residents. Conditional on existing campaign targeting strategies, campaigns do not affect the turnout decisions of the voters we study.