Ohio State is in the process of revising websites and program materials to accurately reflect compliance with the law. While this work occurs, language referencing protected class status or other activities prohibited by Ohio Senate Bill 1 may still appear in some places. However, all programs and activities are being administered in compliance with federal and state law.

Do Americans Really Support Political Violence? New IDEA Research Says No.

November 4, 2025

Do Americans Really Support Political Violence? New IDEA Research Says No.

wize_600per_400_pixel_news_titles_everywhere.jpeg
Image created by AI.

 

As voters go to the polls today, a new study led by IDEA’s Ryan Kennedy challenges the idea that many Americans support political violence. His latest survey used AI to probe for deeper responses than traditional polls, found that far fewer people endorse political violence than recent polls have shown.

Recent polls have raised concerns about rising support for political violence. For example, a POLITICO article yesterday reported that 24% of people said there are “some instances where violence is justified,” and most (55%) expect political violence to rise in the next five years. 

 

But Dr. Ryan's research suggests that these results may be misleading. His AI-driven interview system revealed that most “yes” answers stemmed from misunderstandings of the questions, rather than genuine support for violence. Here are the key findings from the study, which he also discussed in The Conversation.

 

  • Misinterpretation inflates the appearance of support for political violence. The majority of “yes” responses were rooted in ambiguous phrasing; once clarification was provided, responses exhibited greater polarization, suggesting the AI interaction acted as an intervening stimulus that shaped opinion. Respondents associated “force” with legal pressure, the military as a police force, or historical struggles abroad. “The surprisingly high level of support in earlier polls appears to be an artifact of question design, not a true reflection of Americans’ willingness to commit political violence,” Kennedy says. “Our AI‑driven approach shows that once respondents clarify what they meant, the genuine support drops dramatically.”

 

  • Online survey noise is also a likely factor in many polls. Dr. Kennedy notes that “bogus respondents” and “online disinhibition” can inflate support for extreme positions in web-based panels.

 

  • Broad public condemnation in other studies also challenges these polls. When asked directly about contemporary U.S. violence, only a small fraction (around 3 %) expressed support, confirming that most Americans denounce the recent spate of attacks.

 

How the AI‑Driven Survey Was Conducted

  1. Cognitive interviewing with AI: Respondents first answered the same yes-or-no questions used in earlier polls, such as “Do you think it is ever justified for citizens to resort to violence to achieve political goals?

     

  2. Follow-up probes: The AI interviewer then asked additional questions to gain a deeper understanding of each person’s reasoning, adjusting the questions based on their responses. This helped reveal that terms like “use of force” were often understood as legal action, removing a president, or non-violent military involvement.

     

  3. Automated coding:  An AI algorithm categorized the open‑ended explanations, separating genuine support for violent tactics from broader, non‑violent interpretations.

     

Why This Matters for Election Day

  • Reducing fear-mongering: Overstating public support for violence can encourage extreme talk and make things worse. Accurate information helps keep voters calm and prevents a vicious cycle.
  • Focusing on democratic participation: When officials understand what people really think, they can focus on protecting voting sites instead of preparing for unlikely large-scale unrest.

Learn more