Analysis shows hostile rhetoric reduces vote share in gubernatorial elections, suggesting underlying voter biases.
We analyze the relationship between hostility in campaign rhetoric (``incivility'' and character attacks) and electoral outcomes in US gubernatorial elections from 2010 to 2023. We use three LLMs to measure the degree of incivility and character attacks used by each candidate, and provide evidence of the validity of these measures. We find that both incivility and (especially) character attacks are robustly associated with lower vote share, controlling for factors found to be relevant to these elections in prior research and several new factors including LLM measures of ideological extremism and populism. The effects continue to be significant with controls for early and mid-campaign polls, indicating the negative relationship between hostile rhetoric and vote share does not just reflect underdog candidates strategically going negative early in campaigns.
No takes yet. Share an insight, caveat, or question.
Banks et al. (2025) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: