Research

Do Campaign Visits Increase Local Media Coverage? Evidence from the 2000 Election

Predoc Application Data Task

I study whether presidential campaign visits causally increase local newspaper coverage during the 2000 U.S. election. Because visit timing and location are strategically chosen, naïve comparisons confound visit effects with underlying political salience and time-varying shocks. I implement a dynamic staggered Difference-in-Differences design using two-way fixed effects with county and day fixed effects, estimating event-time leads and lags within a five-day window around first visits. I explicitly assess violations of parallel trends, no anticipation, and spillover assumptions, and discuss recent critiques of staggered TWFE estimators under treatment effect heterogeneity. Across candidate-specific and pooled specifications, I find no robust evidence that visits increase local coverage; unstable pre-trends and small treated samples limit causal interpretation.

Data: County × day newspaper mentions and campaign visit records (2000 election) · Identification: Dynamic staggered DiD (TWFE) with county and day fixed effects · Methods: Event-study specification; clustered standard errors; pooled and candidate-specific models

Paper Code