Science has a self-correcting system built into it. The problem is that the self-correction rarely travels as far, or as fast, as the original claim. A 1998 paper published in The Lancet by Andrew Wakefield reached every corner of the English-speaking world within weeks. The retraction, 12 years later, barely registered by comparison. The myth it spawned is still active in 2026.
What follows is not a catalogue of innocent mistakes. Some of these studies were fraudulent. Some were methodologically reckless. Some were simply over-interpreted by researchers who let enthusiasm outrun their data, and then amplified by a media ecosystem that rewards a clean, counterintuitive finding far more than a careful, hedged correction. All eight have one thing in common: they lodged in the public mind so deeply that evidence alone has not been enough to dislodge them.
Discredited scientific studies shape whether children get vaccinated, whether prison reform policy gets taken seriously, whether job candidates spend two minutes in a bathroom stall doing superhero poses before interviews. The distance between “the study said” and “the study was retracted” is wider than most people realise.
1. Wakefield’s MMR-Autism Link: The Most Consequential Fraud in Modern Medicine
On February 28, 1998, a paper primarily authored by physician Andrew Wakefield and twelve coauthors was published in The Lancet, falsely claiming causative links between the measles, mumps, and rubella (MMR) vaccine and both colitis and autism. The paper involved only 12 children, had no control group, and relied on parental recall – scientific limitations that were clear even at the time of publication.
The study’s influence did not come from its science. It came from what happened after publication. When the original article was picked up by the general media, the findings were amplified by speeches and public appearances in which Wakefield recommended single vaccines rather than the combined MMR. Parents didn’t read the methodology section. They saw a doctor on television warning them.
The fraud involved data bias and manipulation and two undisclosed conflicts of interest, including a patent Wakefield held for a single measles vaccine and evidence that he had been paid by lawyers acting for parents in lawsuits against vaccine manufacturers. In 2010, the UK General Medical Council found that Wakefield had been dishonest in his research and had acted against patients’ best interests. The Lancet fully retracted the 1998 paper, stating that several elements of the manuscript were incorrect and that the journal had been “deceived.” The BMJ separately published an investigation concluding the data had been falsified. Three months later, Wakefield was struck off the UK medical register.
The paper’s publication and the media attention it received are widely regarded as a pivotal moment in mainstream scepticism about MMR vaccine safety. The methodological flaws were readily apparent to followers of medical research, and several larger and better-conducted studies definitively countered Wakefield’s claim. As Johns Hopkins Bloomberg noted in a March 2025 article, the paper was retracted and several large studies have since shown no association between vaccines and autism, but the idea persists among some groups. Wakefield’s subsequent media appearances ignited a worldwide vaccine scare that resulted in declining vaccination rates and outbreaks of measles and other vaccine-preventable diseases.
In 2026, pediatricians still spend appointment time refuting a paper that was retracted sixteen years ago and whose author lost his medical licence.
2. The Stanford Prison Experiment: When the Researcher Became the Story
The Stanford Prison Experiment, conducted in August 1971, was designed as a two-week simulation of a prison environment to examine how situational variables affect participants’ behaviour. Psychology professor Philip Zimbardo managed the research team. The study’s headline claim was dramatic and simple: ordinary people, when assigned the role of prison guard, would rapidly become cruel and abusive. It became one of the most cited studies in social psychology and was taught in undergraduate courses for half a century.
Recent critiques and replication attempts have highlighted methodological flaws, ethical concerns, and researcher bias, casting serious doubt on the original conclusions. The most damaging revelations came from audio recordings and documents showing that Zimbardo himself had coached the guards to act harshly, meaning the behaviour wasn’t spontaneous, but partially scripted.
A notable replication attempt, the BBC Prison Study, aimed to recreate the conditions of the Stanford experiment with more rigorous controls and found different outcomes. Following Zimbardo’s death in October 2024, researchers and scholars renewed calls for a formal reassessment of whether the original study’s conclusions could be considered scientifically valid given the documented coaching of participants.
The conclusion that ordinary people will commit atrocities when given authority and permission feels intuitively right. It matches a widely held sense of how power corrupts. That intuitive fit is precisely what keeps the study in circulation even after the methodology has been picked apart. A belief that confirms something we already suspect is far harder to dislodge than one that merely tells us something new.
3. The Milgram Obedience Experiments: Contested, Not Collapsed
In the early 1960s, Stanley Milgram, a social psychologist at Yale, conducted a series of experiments showing that a significant majority of individuals, when instructed by an authoritative figure, were willing to administer what they believed were severe electric shocks to another person. In the best-known variant of the study, roughly two-thirds of participants went all the way to the maximum shock level, an apparently lethal 450 volts.
The popular takeaway became a fixture of public psychology: people are terrifyingly obedient. The Holocaust was explicable. Atrocities happen because ordinary people follow orders.
The reality is more complicated. Given the high levels of stress Milgram imposed on participants, ethical concerns now make a direct replication untenable. Researchers who examined Milgram’s original notes found evidence that participants were pressured beyond the scripted instructions, raising questions about whether compliance was freely given or coerced by the experiment itself. Some researchers argue the experiment endures because it functions as a powerful parable rather than as a scientific finding.
It would be inaccurate to place Milgram in the same category as Wakefield. A 2025 study published in Scientific Reports found that levels of obedience in a new experimental context closely mirrored those Milgram observed in the 1960s, with neither location nor framing significantly influencing participants’ actions. The obedience effect, at some level, appears real. What is contested is the interpretation: the grand narrative about human nature that Milgram, and the textbooks that followed him, built around a limited set of experiments.
4. Power Posing: The TED Talk That Outlasted the Science
Following the 2010 publication of “Power posing: Brief nonverbal displays affect neuroendocrine levels and risk tolerance,” one of its authors, Amy Cuddy, achieved a degree of fame unusual among academic psychologists when she publicised its results in a popular TED talk and book. The fame turned sour when subsequent research failed to reproduce key results of the paper.
The original claim was attention-grabbing: standing in a “high-power” pose, feet wide, arms up, chest open, for two minutes before a job interview would measurably raise testosterone, lower cortisol, and make you more confident and willing to take risks. A critical replication study led by behavioral economist Eva Ranehill at the University of Gothenburg, using a sample five times larger than the original group, found no scientific evidence to support power posing, saying it did not affect hormones or risk tolerance. The replication was published in 2015 by the same journal that first published Cuddy’s original article.
After these findings were released, Dana Carney, a co-author of the original power posing paper, released a statement highlighting severe deficiencies in the paper, stating she did not believe “power pose” effects were real and that “the evidence against the existence of power poses is undeniable.”
The original claim covered two effects: increasing feelings of power and producing measurable hormonal changes. The hormonal half has never been substantiated, and there is insufficient replication to say that power posing does anything to physiology at all. Some evidence suggests people feel slightly more confident after expansive poses, which is a far more modest claim than the one that drove Cuddy’s TED talk to over 70 million views. That smaller, hedged version doesn’t make headlines, so the bolder original claim persists.
The TED talk remains live. The correction has been viewed by considerably fewer people.
5. The Marshmallow Test: Self-Control Versus Circumstance
Walter Mischel’s marshmallow test, developed at Stanford in the late 1960s and 1970s, told a story that people wanted to hear. Put a marshmallow in front of a four-year-old and tell them they can have two marshmallows if they wait fifteen minutes. The children who waited, the story went, grew up to have better SAT scores, lower body mass index, and stronger coping skills. Willpower at age four predicted success for life.
In 2018, a major replication study published in Psychological Science, led by Tyler W. Watts of Columbia University, followed a much more diverse group of 900 preschoolers into their teens. Controlling for differences such as household income and cognitive abilities, the researchers found only weak relationships to academic outcomes and no significant correlations to later behaviours such as anti-social tendencies.
Children from lower-income homes had more difficulty resisting treats than affluent children, and that difference reflected not weaker character but rational decision-making rooted in real-world scarcity. Scholars had known for decades that affluence and poverty shape the ability to delay gratification. The marshmallow test dressed that well-established pattern in a story about individual willpower and sold it as a discovery.
The test’s staying power in public consciousness comes from the same source as its flaw: it makes self-control feel like a character trait you either have or you don’t, a marker of potential identifiable in a preschooler. The 2018 replication found no childhood skills that function as reliable predictors of adult success. That finding doesn’t fit on a motivational poster, so the original version persists in parenting books, corporate training materials, and the pop psychology canon.
6. Sugar and Hyperactivity: The Study That Parents Refuse to Release
Ask any parent at a birthday party whether sugar makes children hyper, and the overwhelming majority will say yes. They have watched it happen. They are certain of it. The science has been equally certain, for decades, in the opposite direction.
A 1995 meta-analysis published in JAMA, examining 23 double-blind, controlled trials, found no evidence that sugar consumption causes hyperactivity or other behavioural changes in children, even in children diagnosed with ADHD or children described by parents as sensitive to sugar. When parents were told their children had consumed sugar, regardless of whether they actually had, they rated the children’s behaviour as more hyperactive. The effect was entirely in the observer, not in the child.
Confirmation bias is working in plain sight. Parents expect sugar to cause hyperactivity, so they see it. The child runs around at a birthday party, surrounded by friends, with zero normal routine, overstimulated by games and music and cake, and the sugar gets the blame. The belief has not shifted meaningfully in thirty years, despite being addressed by some of the most methodologically rigorous trials in pediatric behavioural research.
7. The Replication Crisis and the “Ego Depletion” Theory
One of the casualties of psychology’s broader reckoning with its own methodology is ego depletion, the theory proposed by Roy Baumeister in the late 1990s. The premise: willpower is a finite resource, like a muscle that fatigues. Use self-control in the morning and you have less of it in the afternoon. Resist the cookies and you’ll be worse at resisting the argument. The theory produced thousands of studies and became the intellectual backbone of an entire genre of productivity literature.
The replication crisis in psychology was initially spotlighted in 2015 when a team led by Brian Nosek published a paper in Science providing evidence of systemic reproducibility problems, finding that only 40% of 100 studies could have their findings reproduced. Ego depletion was among the high-profile casualties. A pre-registered replication effort coordinated across 23 international labs, published in Perspectives on Psychological Science, found no statistically meaningful ego depletion effect, with over 2,000 participants collectively producing a result indistinguishable from zero. The glucose theory, that willpower was literally fuelled by blood sugar, was tested and found equally unsupported.
The theory survives in management consulting, self-help books, and corporate wellness programs. Executives are still advised to schedule important decisions in the morning before their willpower runs out. The intervention recommendations persist. The foundational science has not.
8. Cold Fusion: The Energy Discovery That Wasn’t
In March 1989, electrochemists Martin Fleischmann and Stanley Pons announced at the University of Utah that they had achieved nuclear fusion at room temperature using a simple tabletop apparatus. The claim promised virtually unlimited cheap energy and triggered a worldwide wave of excitement and immediate replication attempts. Laboratories around the world scrambled to reproduce the results, and almost none could.
Unlike the psychology studies on this list, cold fusion was not a case of fraud or data manipulation in any proven sense. It was a case of scientists who moved too fast, announced results before adequate verification, and set off a chain reaction they could not control. The press conference came before the peer review. The subsequent failure to replicate, across dozens of independent labs, effectively ended cold fusion as a credible scientific program within eighteen months.
Cold fusion did not die. It retreated into what its proponents now call “condensed matter nuclear science” and has maintained a committed community of researchers and enthusiasts for more than three decades. Government agencies, including units within the US Department of Defense, have at various points funded small investigations into the phenomenon. The core claim, that anomalous heat can sometimes be produced in certain electrochemical systems, remains contested rather than outright dismissed by a small subset of legitimate scientists, though the original announcement’s claim of room-temperature nuclear fusion has never been replicated under controlled conditions.
A spectacular claim, once attached to the words “nuclear” and “unlimited energy,” can resist complete erasure from public consciousness no matter what the subsequent data shows.
Why Discredited Scientific Studies Are Almost Impossible to Kill
These eight studies did not survive because people are irrational, though cognitive biases play a role. They survived because of a structural feature of how science communicates with the public. The original findings arrived with credibility markers, a prestigious journal, a university affiliation, a TED stage, that felt authoritative. The corrections arrived through the same channels but carried a different social weight: they were about something being taken away, something that felt true being declared untrue. One factor is the illusory truth bias, the idea that hearing something repeatedly, even if you know it is wrong, makes it seem more true over time.
Motivated reasoning compounds the problem. Decades of research in experimental psychology show that providing detailed factual information is not an effective antidote to deeply held concerns. When asked in an experimental setting, individuals may repeat the corrected facts in the short term but easily revert without reinforcement. The Wakefield paper didn’t just give parents a theory about vaccines. It gave them a reason for their children’s suffering that felt like an answer, and answers that feel meaningful are not surrendered in response to a retraction notice in a medical journal.
Belief shifts slowly through sustained contact with accurate information across multiple trusted relationships. The pediatrician who takes the concern seriously rather than dismissing it. The science journalist who explains not just that a study was wrong but why it was wrong, and what the failure reveals about the conditions that produced it. The science curriculum that teaches how studies work before it teaches what studies found.
Retracted papers are also not reliably marked as such in the wild. Wakefield’s 1998 paper is still shared on social media. Cold fusion articles written in 1989 circulate as though they are current. The internet does not timestamp credibility. A decade-old refuted claim sits alongside a 2025 peer-reviewed systematic review with equal visual authority, and most readers have no reliable way to tell them apart. The Wakefield paper alone is among the most cited retracted papers in scientific history, according to Retraction Watch, which tracks ongoing citations of retracted work and notes that some papers continue accumulating references even after formal retraction.
Science communication is not primarily a problem of information. People who believe the MMR vaccine causes autism have often seen the counter-evidence. People who believe sugar makes children hyper have often been told otherwise. The belief persists because it is embedded in something larger than a data point: an identity, a community, a felt experience, a distrust of institutions that predates and outlasts any specific claim.
The Part That Doesn’t Have a Clean Fix
Retractions are built for scientists. The notification goes back through the same journals and databases that published the original finding, into a system that most people never directly access. By the time a study has become a parenting belief, a management principle, or a cultural shorthand, it has already left the laboratory far behind.
Some of these patterns go back further than any single study. The fear that vaccines cause developmental harm predates Wakefield. The intuition that willpower is finite predates Baumeister. The belief that sugar overstimulates children predates any published trial. What the studies did was hand people a credentialed frame for something they were already inclined to believe, and credentialed frames are extraordinarily difficult to remove once installed.
That isn’t a failure of public intelligence. It’s a failure of how science reaches the public, and how rarely the corrections that matter most are given the same resources, the same urgency, and the same storytelling as the original claims they’re trying to displace.
Disclaimer: This information is not intended to be a substitute for professional medical advice, diagnosis, or treatment and is for information only. Always seek the advice of your physician or another qualified health provider with any questions about your medical condition and/or current medication. Do not disregard professional medical advice or delay seeking advice or treatment because of something you have read here.
AI Disclaimer: This article was created with the assistance of AI tools and reviewed by a human editor.