Short-lived persuasion from campaign rallies: Evidence from the 2016 U.S. presidential election

This paper estimates the impact of presidential campaign rallies on candidate support and policy attitudes in the 2016 U.S. presidential election. Using CCES microdata matched to rally dates and locations, I implement an event-study design that exploits within-county variation in interview timing around rallies. Rallies generate short-lived shifts that fade within 1 week. Clinton rallies consistently increase support for Clinton immediately after these events take place. Trump rallies show geographic heterogeneity—support falls in urban counties and rises in suburban counties. Policy attitudes respond weakly overall: I find little systematic evidence that Trump rallies shift issue positions, while Clinton rallies generate a modest leftward shift in a broad ideology index in suburban counties.