Rumors, Truths, and Reality: Political Misinformation in the Modern Day

Key themes

• The problem is widespread. Roughly one-third of Americans accept the Obama-birther rumor; 33% believed the ACA had "death panels"; significant fractions believe various election-fraud and 9/11-truther claims. Over 70% of Americans endorse at least one false political claim, though very few believe many false claims simultaneously.
• Two sources of false belief. Conspiratorial thinking is a stable personality trait—some people are more naturally skeptical and prone to entertaining outlandish stories. However, partisanship is the stronger structural force: what Republicans and Democrats believe diverges sharply on partisan rumors. Obama-birthplace rumors show a 60-point Republican–Democrat gap; 2004 election-fraud rumors (Democrat-favoring) show a 60-point gap the other direction. Non-political rumors (Roswell aliens) show no partisan gap.
• The sticky "not-sure" category. Many people say they are uncertain ("not sure") rather than outright rejecting rumors. When asked why, they invoke "where there's smoke, there's fire"—repetition and ambient rumors create perceived validity even without explicit belief. This uncertainty is politically damaging because it undermines confidence in elected officials and institutions.
• Expert correction often fails. When fact-checkers or nonpartisan organizations debunk a rumor, the effect is minimal—equivalent to showing nothing. Nonpartisan corrections cancel out the rumor's negative effect but provide no net benefit.
• Credible messengers are rare. The most effective correction comes from an unlikely source—someone arguing against their partisan interest. When a Republican senator debunked death-panel claims, 68% of subjects rejected the rumor (vs. 43% in the rumor-only group and 58% in the control). This works because the source is credible: a Republican has no partisan motive to defend the ACA, so their denial is persuasive.
• Effects are temporary. Within two weeks, rumor-rejection rates drop by 7–10% even when the original correction was effective. The rumor-only condition's effect persists (and worsens), while corrections fade. Corrections do not "stick" the way rumors do.
• Media environment matters. Rumors spread faster and more visibly in the internet age, but the phenomenon is not new. Berinsky documents conspiracy beliefs going back to the founding era, McCarthyism in the 1950s, and World War I–era fears. What has changed is the ease of spread, not the underlying psychology.
• Partisan media ecosystem is imbalanced. Research from Harvard's Shorenstein Center shows an asymmetry in the distribution of misinformation: there is more high-volume misinformation on the right than the left (an empirical fact, not a normative claim). This creates a policy dilemma: interventions that treat both sides equally do not yield equal outcomes.
• Critical thinking can backfire. Teaching people to question everything is a double-edged sword—it can also prime conspiratorial thinking. There is no single educational solution.
• Novel approaches show promise. Work with David Rand on "accuracy nudges" (brief prompts reminding users they value accuracy before sharing) shows measurable effects on social-media sharing. Collaborations with tech companies (Google Jigsaw, Twitter, Facebook) are exploring how to scale these interventions. Cognitive-science partnerships at IDSS are investigating how to get people to deliberate more deeply before endorsing false claims.
• Many solutions, no magic bullet. Berinsky emphasizes there is no single fix. Effective responses require coordinated efforts across education, media design, fact-checking, and behavioral interventions.

Rumor landscape

Berinsky discusses several recurring political rumors:

• Obama birther myth (2009 onward). A persistent, partisan (Republican-skewed) conspiracy that persisted even after Obama released his long-form birth certificate. Supporters were skeptical of the evidence itself (claims about typeface, differences from twins' certificates).
• "Death panels" (2009–2010). A metaphor used to characterize end-of-life counseling provisions in the Affordable Care Act as government-mandated euthanasia. About 33% of respondents accepted this claim; 46% rejected it; 21% were unsure. It undermined ACA support and was extremely sticky.
• 2004 election fraud in Ohio (Kerry myth). Democrats speculated that electronic voting machines made by Diebold (a Republican donor) were rigged. Berinsky notes Democrats have conspiracy theories too, though the current media environment amplifies Republican-favoring rumors more.
• 9/11 inside-job conspiracy. About 15–20% of Democrats endorsed claims that the federal government assisted or allowed the attacks to enable Middle East wars.
• Illegal immigrant durable goods myth. A false claim that illegal immigrants are given rooms full of refrigerators, cars, and washing machines to take to their new lives in America. This rumor clusters with anti-immigration sentiment and partisan identity.

Berinsky's research design

Berinsky uses multi-wave survey experiments to test:

1. Baseline belief distributions. Eight-item scale combining partisan rumors and non-partisan rumors (Roswell, aliens) to separate conspiracy-proneness from partisan structure.
2. Correction experiments. Randomized trials comparing: (a) control (no information), (b) rumor only, (c) rumor + nonpartisan expert correction, (d) rumor + Republican correction, (e) rumor + Democratic correction. Outcomes measured immediately and two weeks later.
3. Follow-up justifications. Open-ended prompts asking why respondents believe what they do, revealing "where there's smoke, there's fire" reasoning.

Implications for practitioners

• Correction strategy matters more than content. Who says it is often more important than what is said.
• Identify unlikely messengers. Find credible figures who oppose partisan interests when debunking the opposing side's rumors.
• Plan for decay. Design systems for repeated reinforcement, not one-off corrections.
• Address the uncertain middle. Interventions should focus on people expressing doubt ("not sure"), not hard believers or hard skeptics.
• Scale requires partners. Tech companies, media organizations, and trusted community leaders must be part of the solution.
• Media design matters. Platforms can implement accuracy nudges and other behavioral designs to promote deliberation.

Notes

This talk represents the state of research on political misinformation circa 2021, capturing Berinsky's decade-long research program on rumors and corrections. The emphasis on the primacy of source credibility over message content is central to his framework and has become increasingly influential in the field. His candor about the absence of a simple solution is notable and contrasts with the "just add media literacy" or "just label false content" prescriptions common in policy circles. The research on temporary effects of corrections—showing that rumor persistence exceeds correction persistence—is particularly important for platform designers and policymakers who assume that one correction solves the problem.

The talk also surfaces the political complexity of addressing an asymmetrically distributed problem (more misinformation on one side of the political spectrum) without appearing to take partisan sides. Berinsky's empirical framing—"this is what the data shows"—is a useful model for researchers navigating these sensitive topics.

Berinsky's ongoing work at the intersection of cognitive science, political science, and computer science (the IDSS initiative) exemplifies a trend in the field toward interdisciplinary solutions rather than single-discipline approaches.