Reading a conspiracy theory does more than just introduce a suspicious narrative; it also inflates a person’s estimate of how many other people believe the claim is true. According to new research published in the European Journal of Social Psychology, this exaggerated sense of public agreement may help explain why people come to accept conspiracy theories after encountering them. In one of the experiments, showing people concrete statistical data about how few others actually endorsed these claims reduced their own belief in them.
Conspiracy theories are explanations for events that assume multiple actors are secretly working together to achieve a harmful or unlawful goal. When people feel uncertain about how to interpret events, they often look to others for cues on what to think or do. Psychologists refer to these cues as descriptive norms, which are perceptions of how common a particular belief or behavior is within a group. Relying on descriptive norms is a form of social proof, where people assume that if many others hold a view, that view is likely valid.
Although conspiracy theories frequently draw massive attention online, actual believers usually represent a minority of the broader public. Still, those who endorse these narratives often assume their views are widely shared. For example, a 2021 study of British participants found that people consistently overestimated how much their peers believed in conspiracy theories, and that thinking peers endorsed these ideas strongly predicted a person’s own belief. Similarly, a 2023 study in Turkey indicated that people who believed local conspiracy theories overestimated how many ordinary citizens and experts agreed with them.
Correcting these misperceptions can alter how people think. In a 2021 experiment with British parents, showing participants that most other parents rejected anti-vaccine conspiracy theories reduced their own endorsement of those claims. Building on this foundation, the new research was led by Jan-Willem van Prooijen, head of the Social Psychology section at Vrije Universiteit Amsterdam, senior researcher at the Netherlands Institute for the Study of Crime and Law Enforcement, and endowed professor of radicalization, extremism, and conspiracy thinking at Maastricht University.
He and his colleagues wanted to test whether simply encountering a conspiracy theory inflates perceived descriptive norms, and whether providing hard data on public skepticism can reduce belief in a wide variety of general conspiracy theories.
“People often encounter conspiracy theories, for instance online,” van Prooijen told PsyPost. “One problem is that extreme voices that spread conspiracy theories (e.g., online influencers) sometimes get excessive exposure. That might raise the impression that these conspiracy theories are more widely believed than is actually the case. We wanted to examine these dynamics in our research: Do conspiracy beliefs depend on how normative people think they are, and can we use these insights to develop an intervention?”
To investigate these questions, the researchers conducted six online studies totaling 2,906 participants from the United States and the United Kingdom. All six studies were preregistered, meaning the researchers publicly recorded their hypotheses and analysis plans before collecting data.
In the first pair of studies, the researchers examined the relationship between personal conspiracy beliefs and perceived norms using 300 American participants in Study 1a and 299 British participants in Study 1b. Participants rated how plausible they found 10 popular conspiracy claims, such as the idea that the 9/11 terrorist attacks were an inside job by the U.S. government, on a scale from 0 to 100, and then used the same scale to estimate the average plausibility rating given by the other participants in the study.
In both countries, the actual average belief in these 10 theories was relatively low, sitting at 33.09 out of 100 in the American sample and 34.16 in the British sample. However, participants estimated the group average to be higher, guessing 39.85 in the United States and 42.30 in the United Kingdom. Those who strongly endorsed the conspiracy theories themselves were much more likely to think other participants believed them too, with strong correlations of 0.65 in the U.S. sample and 0.71 in the U.K. sample, and their estimates of the group average were less accurate than the estimates of skeptics.
The researchers also tested whether conspiracy believers saw themselves as standing out from the crowd, given that past research has linked conspiracy belief to a desire to feel unique. By comparing each person’s own rating to their estimate of the group rating, the team found that stronger believers saw the gap between their own views and the group’s as no larger than skeptics did.
These initial findings align with research covered by PsyPost in 2025, which found that individuals who are overconfident in their thinking abilities tend to hold stronger conspiracy beliefs and vastly overestimate how many others agree with them.
Because the first two studies only looked at correlations, the researchers next tested whether reading a conspiracy theory directly causes people to think the idea is popular. They recruited 505 American participants for Study 2a and 501 British participants for Study 2b. In each study, half of the participants were randomly assigned to read an anonymous blog post claiming that pharmaceutical companies secretly suppress medical cures to boost their profits, while the other half read a blog post arguing that pharmaceutical companies do not hide cures.
After reading and summarizing the text, participants rated their own belief in four statements about pharmaceutical conspiracies on a 0 to 100 scale, and estimated how the other participants in the study rated them. In the American sample, reading the conspiratorial blog post raised personal belief to an average of 61.78, compared to 49.46 among those who read the non-conspiratorial text. It also pushed estimates of other people’s belief up to 62.64, compared to 51.00 in the control group.
The British sample in Study 2b showed the same pattern. Personal belief averaged 59.78 after reading the conspiracy post versus 41.99 in the control group, while estimates of others’ beliefs rose to 60.28 compared to 46.41 in the control condition. Statistical mediation analyses were consistent with the idea that the increase in personal belief ran through this heightened sense that others believed the claim. But the reverse pattern, in which people’s newly strengthened beliefs shaped their guesses about others, also fit the data, so these analyses cannot establish which came first.
“One reason why people believe conspiracy theories is because they assume many others believe them as well,” van Prooijen explained. “Conspiracy beliefs thus spread via social influence processes. Understanding these processes can help design effective interventions to reduce conspiracy beliefs.”
To test whether changing perceived norms can directly alter conspiracy beliefs, the researchers manipulated the norm information given to participants in Studies 3 and 4. In Study 3, 501 American participants were presented with six popular conspiracy theories, such as the claim that the government uses subliminal messages on television. Half were told that “especially many” previous participants found the statements plausible, while the other half were told that “only a few” found them plausible.
This vague wording did not produce a statistically significant difference between the two groups. Participants told that many others believed the claims scored an average of 44.48 out of 100 on personal belief, compared to 41.65 among those told that few others believed them. The authors noted that words like “many” and “few” are open to personal interpretation; for instance, learning that even 10 percent of the public believes an extreme claim might still register as a large number to some people. Because this study did not measure whether participants’ sense of the norm actually shifted, the authors acknowledge another possibility: that people do not take their cues from norm information at all, and that the link seen in earlier studies reflects people projecting their own beliefs onto others.
In Study 4, the team tested whether providing specific, factual statistics would be more effective. They recruited 800 American participants and showed them eight conspiracy theories from the first study. Half of the participants saw a bar chart for each theory displaying the actual distribution of ratings from Study 1a, along with the median score and a note explaining that most participants did not find the statement plausible at all. The other half saw the conspiracy statements without any statistical data.
Providing concrete data shifted how participants viewed both the public consensus and the theories themselves. Those who saw the charts estimated the previous group’s average belief at 20.95 out of 100, compared to 38.34 among those who received no data. In turn, personal endorsement of the conspiracy theories dropped to an average of 24.06 in the group that saw the factual charts, compared to 33.70 in the control group.
“We have done two studies testing the intervention, but it only worked if we gave people factual statistical information of how few people believed a conspiracy theory,” van Prooijen noted. “If we provided more abstract information, only stating that ‘few people’ believe the theory without providing concrete numbers, the intervention did not work. It remains speculative why that is the case, but it is nevertheless informative for future interventions based on this research.”
“The practical significance is that these insights quite directly suggest an intervention to reduce conspiracy beliefs,” van Prooijen said. “We have tested the effects of exposing people to factual statistical information of how widespread belief in a particular conspiracy theory is. Such information often clarifies to people that a conspiracy theory actually is not widely endorsed at all, which decreases the likelihood that people will believe the theory themselves.”
As with all research, there are some caveats to consider when interpreting these results. Participants in brief online experiments may partly respond the way they think researchers expect, which the authors note could mean these effects are smaller in everyday life, and the “others” in these studies were fellow online participants rather than the broader population. In addition, social media algorithms and online communities can surround users with like-minded believers, creating environments where conspiracy theories genuinely are the local norm.
“It is important to keep in mind that in everyday life, these processes are likely to be more complicated,” van Prooijen said. “For instance, we examined the effects of exposing people to conspiracy theories on their perceptions of how normative belief in these theories is. But in real life, people to some extent also choose what information they wish to be exposed to. People may for instance actively search the Internet looking for conspiracy theories, and be motivated to think belief in these theories is widely shared.”
The authors also point out that factual norm interventions could backfire if applied to the rare conspiracy theories that actually are endorsed by a majority of the public, such as theories surrounding the assassination of John F. Kennedy. Future studies could examine how these social processes operate in real-world online networks and in more collectivistic cultures, where pressure to conform to group norms may be even stronger.
“An important caveat is that the intervention that we tested might only work in communities where a particular conspiracy theory is indeed not so widespread,” van Prooijen added. “Some conspiracy theories ARE widespread, for instance in the general population (e.g., about the Kennedy assassination), or in specific groups (e.g., online anti-vaccination groups). These are complications that need to be investigated further, and are important to take into account before putting these insights into practice.”
“We were excited to see that our findings almost naturally suggested an intervention, which indeed had the anticipated effects,” he continued. “At the same time, the intervention was successful in only one experiment. We need more testing to assess the effectiveness of this approach. One possible goal is to examine these issues in a longitudinal design: What is the long-term effectiveness of an intervention based on information about descriptive norms?”
The study, “Everyone Knows It’s a Hoax: The Role of Perceived Descriptive Norms in Conspiracy Beliefs,” was authored by Jan-Willem van Prooijen, Paul K. Bergmann, Vladimira Čavojová, and Jakub Šrol.