How to Estimate Public Support for Political Violence and Why It Matters: Impact of Sampling, Engagement Checks, and Question Phrasing

Abstract What is the true level of support for political violence in the United States? The answer to this question is extremely important to understanding the level of threat to American democracy, as well as how best to respond. Critics have argued that previous works by political scientists have drastically overstated the level of support for political violence, by 600 percent or more, due to disengaged respondents and poor question wording in surveys. Using probability panels, we conduct survey experiments to understand the impact of survey sampling methodology, respondent engagement, and question wording on the reported support for political violence in the United States. We find that disengagement is far less prevalent in probability panels, and that the effects of disengagement and question wording are less pronounced than in opt-in panels used by critics. Accordingly, policymakers and the general public have valid concerns about the magnitude of public support for violence.