Summary: A Dartmouth-led study finds that addressing the network of beliefs a person holds is often more effective at correcting false views—such as election fraud claims—than attempting to rebut a single falsehood in isolation. The research shows people maintain interlocking beliefs that reinforce one another, which can make factual correction difficult unless the supporting beliefs are also addressed.
Source: Dartmouth College
Understanding how beliefs form and why they resist contrary evidence is crucial in a polarized era, where disputes over topics like vaccines and climate change mirror the dynamics seen around election fraud claims.
A new Dartmouth-led study published in Nature Human Behaviour suggests that to counter false beliefs—like claims of widespread election fraud—it may be more effective to target the broader belief system that supports those claims rather than focusing solely on the specific falsehood. The research analyzed how people updated their beliefs about fraud during and after the 2020 U.S. presidential election.
“People do not hold isolated beliefs; they hold networks of related beliefs that influence each other,” says Rotem Botvinik-Nezer, lead author and postdoctoral researcher in Dartmouth’s Cognitive and Affective Neuroscience Lab. “That interdependence helps explain why simply presenting evidence against fraud often fails: many people have auxiliary beliefs—about who should have won or who benefits from fraud—that anchor the system.”
The research team, who have prior experience studying placebo effects and how perceptions shape outcomes, designed a large-scale survey to examine belief formation under real-world uncertainty. On November 4, 2020—while votes in several key states were still being counted—they surveyed more than 1,600 Americans about partisan preferences and beliefs related to hypothetical election outcomes.
Participants reported which candidate they preferred, how strongly they favored that candidate, and their estimates of how likely each candidate would have won the true vote absent fraud. They also rated how likely they thought fraud would change the reported outcome. Respondents were randomly shown one of two maps depicting different hypothetical winners in the remaining states and were then asked to reassess their beliefs about fraud and the true winner. A subset of participants completed a follow-up survey roughly three months later to report their enduring beliefs about the true vote and who, if anyone, benefited from fraud.
Results showed a clear partisan pattern: when a respondent’s preferred candidate appeared to lose, their belief in election fraud increased; when their candidate appeared to win, their belief in fraud decreased. The intensity of these “desirability effects” scaled with how strongly individuals preferred their candidate—stronger preference produced larger shifts in fraud belief.
To explain these shifts and predict them quantitatively, the researchers built a probability-based computational model grounded in Bayesian reasoning. The model represented a compact system of three interrelated beliefs: whether fraud had occurred, who would win the true vote, and who would benefit if fraud occurred. Importantly, the model did not include direct information about participants’ partisan desires; instead, it inferred how people updated beliefs from the structure of their prior beliefs.
Contrary to interpretations that attribute partisan asymmetries entirely to motivated irrationality, the Bayesian model accurately reproduced the observed patterns of belief updating and outperformed alternative models that assume purely biased or goal-directed reasoning. The key insight is causal attribution across competing explanations: when new evidence conflicts with someone’s belief that a particular candidate should have won, it can be rational—within their belief system—to infer fraud as an explanation for the discrepancy.
The study found that roughly one-third of respondents explained a hypothetical loss nearly entirely by fraud rather than by the true vote. That pattern illustrates how an auxiliary belief—such as who benefits from fraud—can short-circuit the usual link between evidence and belief about the true outcome.
“If fraud is treated as a plausible alternative explanation, it can take causal credit for an unexpected result, weakening the impact of straightforward evidence about who actually won,” says Tor Wager, Diana L. Taylor Distinguished Professor in Neuroscience and director of the Dartmouth Brain Imaging Center. “So changing a false belief often requires addressing the supporting beliefs that make the falsehood plausible.”
The study’s co-authors are Rotem Botvinik-Nezer, Tor Wager, and Matt Jones from the University of Colorado Boulder. Their findings highlight that correcting misinformation about election fraud and similar topics may demand multi-pronged approaches that simultaneously target key beliefs within a person’s broader belief system rather than relying on single-piece debunking.
About this psychology research news
Author: Amy Olson
Source: Dartmouth College
Contact: Amy Olson – Dartmouth College
Image: The image is in the public domain
Original Research: Open access. “A belief systems analysis of fraud beliefs following the 2020 US election” by Rotem Botvinik-Nezer et al., Nature Human Behaviour
Abstract
A belief systems analysis of fraud beliefs following the 2020 US election
Beliefs that the 2020 U.S. Presidential election was fraudulent persisted despite substantial contradictory evidence. This study asks why such beliefs are resistant to counter-evidence and whether that resistance reflects irrational, motivated reasoning or rational updating over a network of beliefs.
We surveyed 1,642 Americans during the vote count and tested how fraud beliefs changed under hypothetical outcomes. Participants’ beliefs in fraud rose when their preferred candidate appeared to lose and fell when their candidate appeared to win; both effects scaled with the strength of partisan preference, revealing partisan asymmetry or desirability effects.
A Bayesian model that represents a system of beliefs—about the true vote winner, the prevalence of fraud, and who would benefit from fraud—accurately accounted for these asymmetries and outperformed alternative models that assume purely motivated, irrational updating or that omit the full belief system. These findings suggest partisan asymmetries may reflect rational attributions across multiple causes of evidence, and that effective correction of false beliefs may require addressing multiple supporting beliefs at once rather than single-point debunking.