The Influence of Biased Local Newspaper Endorsements

This paper investigates the direct effects of political endorsements on election results. I analyze county-level vote shares in statewide elections for Governor and U.S. Senate, where voters across the state participate in the same election but are "treated" with different local newspaper exposure. I find that credible endorsements have a small but statistically significant effect on vote shares of about 1.3 percentage points. I also demonstrate that county-level newspaper bias and vote shares are positively correlated, and that candidate quality can explain about three-fourths of the raw correlation between newspaper endorsements and electoral performance. In aggregate, endorsement effects only change election outcomes in a small subset of very close elections, but in those cases endorsements almost always help the higher quality candidate win. The results suggest that local newspapers can help improve political outcomes and that some voters may rely on newspaper endorsements to inform their perceptions of candidate quality.