How Misinformation Wins: What a Decade of Research Reveals About Fake News, Psychology, and Digital Influence
The 2016 U.S. presidential election transformed “fake news” from a novelty into one of the most urgent public problems of the digital age. Early scholarship by economists Hunt Allcott and Matthew Gentzkow, published in the Journal of Economic Perspectives, set the baseline: fake news stories were widely consumed, with pro-Trump misinformation enjoying substantially more engagement than pro-Clinton content, even as the average American recalled encountering only one to three such articles during the campaign. That seemingly modest exposure mattered because fake news was highly concentrated among specific audiences, and its effects were reinforced by the architecture of social media. Twitter-based research by Grinberg and colleagues found that nearly all fake news consumption during the 2016 election was concentrated in a tiny fraction of users, with 0.1 percent of accounts responsible for about 80 percent of shares. Similarly, Guess, Nyhan, and Reifler showed that visits to untrustworthy websites were heavily skewed toward older, conservative-identifying users, even after controlling for ideology. The viral character of misinformation was underlined by Vosoughi, Roy, and Aral’s landmark Science study, which analyzed millions of rumor cascades on Twitter and found that falsehoods spread farther, faster, and more widely than the truth, particularly in political content. The initial picture, then, was one of a decentralized media environment in which a small number of highly active, highly partisan users could generate enormous misinformation flows.
Scholars quickly recognized that “fake news” was too narrow a concept for the broader landscape of deception, manipulation, and simply careless talk. Vraga and Bode offered useful distinctions: misinformation is false but not necessarily intended to deceive, while disinformation is deliberately engineered to mislead. Guess and Lyons framed the problem as part of a broader toolkit of online propaganda, and Frankfurt’s classic essay On Bullshit was revived by researchers to describe content that is not even trying to be true but instead seeks to impress or persuade. MacKenzie and Bhatt connected these ideas epistemologically, arguing that fake news undermines the very standards by which knowledge is produced and trusted. Marwick and Lewis showed how fringe actors exploit journalists’ reliance on social media to manufacture morally panicked stories, and Starbird described disinformation as a hybrid system of bots, trolls, and unwitting human amplifiers. The same period produced a wave of studies on “dark platforms, from far-right networks on Telegram to conspiracy communities on 8kun and Gab, demonstrating that misinformation rarely stays on mainstream platforms; it is cultivated on the margins and then laundered into visibility. Garrett argued that focusing on echo chambers is a distraction, since disinformation campaigns deliberately inject false content into mainstream communities; the problem is not merely that audiences are fragmented but that malicious actors manipulate those fragmented publics. This broader definitional turn allowed researchers to include not only fabricated news articles but also unsubstantiated voter-fraud claims, COVID-19 conspiracy theories, and partisan rumors.
A major strand of the literature asks a deceptively simple question: why do people believe, share, and share false information? Pennycook and Rand discovered that susceptibility to fake news is strongly predicted not by partisanship per se, but by analytic thinking: individuals who score low on cognitive reflection and are more receptive to meaninglessness—what they call “bullshit receptivity”—are more likely to accept misinformation and to overestimate what they know. Their work suggests that “falling for fake news” reflects a general weakness in critical thinking, not simply a motivated desire to believe flattering lies. Yet personality and motivational factors also matter at least as much. Buchanan and Kempley found that cognitive-perceptual schizotypy and psychopathy predict sharing false political information, even after controlling for partisanship, while Lawson and Kakkar found that conscientiousness interacts with ideology to predict fake news sharing. Arceneaux and colleagues developed the concept of the “need for chaos, a desire to destroy existing social and political arrangements for its own sake, and showed that some individuals share hostile political rumors because they enjoy chaos rather than because they believe them. Armaly and Enders added perceived victimhood, finding that Americans who feel aggrieved and marginalized are more receptive to conspiracy narratives. The Green Paranoid Thought Scales and measures of intolerance of uncertainty have been imported into political communication to explain why ambiguous information feels threatening and therefore credible to some people. This psychological literature complicates any simple “both sides” story: misinformation susceptibility is partly cognitive, partly dispositional, and partly driven by deep emotional needs for status, meaning, and certainty.
Equally important, researchers have discovered that sharing false information is frequently a social act rather than an epistemic failure. Schaffner and Luks showed that some survey respondents report beliefs they know are false—such as crowd sizes or approval ratings—because truth-telling conflicts with identity expression. Serota and Levine found that lying is not normally distributed: a few prolific liars account for most falsehoods, which maps neatly onto Hughes’s finding that a small group of hyper-political users produce the majority of political tweets. Petersen, Osmundsen, and Arceneaux demonstrated that the need for chaos predicts sharing hostile political rumors even when people do not believe them, because the point is to troll, polarize, or disrupt. Bor and Petersen, in a cross-national test, argued that online political hostility arises from a mismatch between ancient coalitional instincts and modern anonymous social media, making hostility itself rewarding regardless of content. Lopez and Hillygus documented “survey trolling,” where participants deliberately answer survey questions mischievously, and Metzger and colleagues cataloged “many shades of sharing” misinformation, from accidentally sharing low-quality content to consciously weaponizing it. Celse and Chang found that observing political leaders lie increases the willingness of ordinary citizens to lie, a “politicians lie, so do I” effect, and Bullshitting Frequency Scales developed by Littrell and colleagues showed that frequent bullshitters are more receptive to various types of misleading information. In short, sharing misinformation is not reducible to being duped; it is often a strategic and expressive performance of identity, outrage, or just entertainment.
none of this happens outside platform and elite power. Bail and colleagues assessed Russian Internet Research Agency tactics and found that sophisticated influence operations can shape political attitudes and behavior even when their direct effects are modest. Elite actors matter as much as Russian bots: Lasser and colleagues showed that political elites share low-quality news sources at high rates, and Mosleh and Rand measured exposure to misinformation from political elites on Twitter, finding that falsehoods spread by prominent figures can dwarf organic misinformation flows. Sanderson and colleagues demonstrated that when Twitter flagged then-President Trump’s election misinformation, the flagged tweets continued to spread widely on and off the platform, raising hard questions about whether social media companies can ever effectively intervene. Berlinski and colleagues found that exposure to unsubstantiated claims of voter fraud depresses confidence in elections, even among people who do not accept the claim, and Edelson and colleagues tied conspiratorial thinking to belief in election fraud. DeVerna and colleagues found ideological asymmetries in the failure to correct misinformation, while Ecker and colleagues found no evidence that partisan worldview necessarily blocks correction, creating a complicated picture for fact-checking. Kim and colleagues showed that self-selection and exposure to incivility fuel online comment toxicity, while Vosoughi and colleagues’ finding about falsehoods speed is partly explained by novelty: false news is fresher and more surprising, making it more shareable. Garrett and Bond, however, found that conservatives are more susceptible to political misperceptions, and Garrett, Long, and Jeong identified affective polarization—deep dislike of one’s partisan opponents—as a major mediator. Trust in message source and risk propensity also predict sharing on Facebook, according to Buchanan and Benson, and Uscinski and colleagues connected conspiracy theory beliefs to vaccine hesitancy, while Moran and colleagues examined networked moral panic through #SaveTheChildren, demonstrating that misinformation often spreads through well-intentioned campaigns.
Can misinformation be contained? The emerging evidence is cautiously hopeful about targeted psychological interventions but also increasingly recognizes limits. Pennycook and colleagues showed in a Nature study that shifting people’s attention to the concept of accuracy, even with a simple prompt, reduced misinformation sharing by about half. Van der Linden and colleagues developed psychological inoculation, or “prebunking,” which exposes people to weakened versions of misleading arguments and builds resistance; Roozenbeek and colleagues adapted this into a scalable online game that improved misinformation resilience on social media. Durand and colleagues showed that graph literacy—a person’s ability to understand visual data—is a powerful and often overlooked predictor of health communication effectiveness, and Pennycook and Rand argue that most people can identify fake news when their attention is directed to accuracy. But the broader countermeasures literature, reviewed by Courchesne, Ilhardt, and Shapiro, finds surprisingly limited evidence for many platform interventions such as labels, downranking, or content removal; Vincent and colleagues found Facebook’s downranking of repeat offenders has measurable but modest effects. Corrections of political misinformation are often effective, as Ecker and colleagues concluded, but DeVerna and colleagues show that in a polarized environment corrections themselves can become tribal symbols. Sunstein’s Liars: Falsehoods and Free Speech in an Age of Deception warns that the law is ill-equipped to distinguish harmful lies from protected speech, while Arendt’s classic essays on lying in politics remind us that the problem is less about individual falsehoods and more about the destruction of a shared reality. There is no silver bullet, and the strongest consensus from this literature is that misinformation is a structural problem at the intersection of psychology, politics, platform design, and institutional trust. Any sustainable answer must therefore be equally structural, combining media literacy, better journalism, reformed algorithms, accountability for elites, and a renewed investment in the idea that factual accuracy is a public good rather than a partisan preference.