Survey Inflation: Many Conspiracy Believers Are Just Trolling, Study Finds

The Hidden Problem in Conspiracy Research

For years, public opinion surveys have painted a disturbing picture: large percentages of the public appear to believe in everything from alien cover-ups to secret global cabals. But a forthcoming study in a leading social science journal argues that these numbers may be dramatically overstated. The research team, drawing on multiple large-scale datasets, found that a sizable fraction of respondents who endorse the most bizarre conspiracies — such as claims that the Earth is flat or that lizards control world governments — are not expressing genuine conviction. Instead, they appear to be answering insincerely, often as a form of trolling, satire, or playful defiance toward the survey itself. The study's lead author, a psychologist at a major European university, described the phenomenon as "mischievous responding," a known but frequently ignored threat to the validity of self-report data.

The problem is not new. Survey researchers have long known that some respondents pay little attention, rush through questionnaires, or deliberately give absurd answers. What is new is the scale and sophistication of the problem in the context of conspiracy theories, where the content itself invites mockery and paradox. When a survey asks whether someone believes that the moon landing was faked and simultaneously that NASA faked the Earth's curvature, a respondent can easily answer "yes" to both — not because they hold both beliefs, but because they are making a joke. The researchers set out to detect exactly this kind of behavior by analyzing response patterns across dozens of items, looking for contradictions, impossibly fast completion times, and patterns that statistical models could flag as inconsistent with genuine belief.

How the Researchers Separated Signal from Noise

The study employed a two-pronged approach. First, the team embedded 'catch trials' — nonsensical statements designed to be logically impossible to believe with sincerity, such as "the Sun is made of bananas" — into otherwise standard conspiracy belief questionnaires. Respondents who agreed with these absurd items were flagged as potential insincere responders. Second, the researchers applied latent class analysis, a statistical technique that groups respondents based on their answer patterns, allowing the team to distinguish between a cluster of people who consistently endorse real conspiracy theories and a larger cluster who endorse everything, including the impossible. The results were striking. In one nationally representative sample, roughly 15% of respondents initially appeared to endorse a bizarre conspiracy item. After screening out insincere responders, that number dropped to under 4%.

The drop was even more pronounced for the most outlandish claims. For example, belief in a "hollow Earth" — a theory rejected by all credible geology — fell from an apparent 9% to less than 2% when only sincere respondents were considered. Belief in reptilian humanoids running the government, often cited as a meme rather than a genuine conviction, fell from 7% to roughly 1.5%. By contrast, more mainstream conspiracy theories, such as the belief that the assassination of John F. Kennedy involved a broader conspiracy, showed smaller declines, suggesting that those numbers are more reliable. The study's authors argue that this distinction is crucial: while insincere responding inflates nearly every conspiracy estimate, it has the largest effect on theories that are widely perceived as ridiculous or that have strong cult-movie or internet-meme associations.

The research also identified demographic correlates of insincere responding. Younger respondents, particularly those aged 18 to 29, were significantly more likely to be classified as mischievous responders. Men were more likely than women to engage in this behavior, and those who reported higher levels of internet use and social media engagement were disproportionately represented in the insincere cluster. This aligns with the observation that trolling and ironic posting have become common social practices online, especially in communities that mock survey research. The authors caution, however, that these demographic patterns do not mean that young men never hold conspiracy beliefs; rather, their survey responses are more likely to be contaminated by non-belief-driven factors.

The Academic Debate Over What Counts as Belief

The findings have already sparked heated debate among experts in political psychology and public opinion research. Some scholars argue that even "insincere" responses reflect something meaningful — namely, a cultural identity or a desire to signal opposition to mainstream institutions. A respondent who jokingly claims to believe in lizard people may be expressing distrust in a political system, even if they do not literally believe the claim. The study's authors acknowledge this point but push back. "There is a difference between a metaphor and a measurement," they write in the paper. "If we want to understand how many people actually believe that shape-shifting reptiles are running the Pentagon, we should not count jokes as evidence." They emphasize that their goal is not to dismiss the real problem of conspiracy theories, but to get a more accurate estimate of its true scale.

Other researchers have raised methodological questions, suggesting that the catch trials themselves may be too easy to detect. Some respondents might understand that agreeing with an absurd statement would mark them as insincere and therefore avoid it, even if they are otherwise trolling. Conversely, some genuinely confused or low-literacy respondents might agree with the absurd statements for reasons that are not malicious but also not indicative of conspiracy belief. The authors reply by noting that their latent class models do not rely solely on the catch trials; they look at the entire pattern of answers across dozens of items, and they validated the classification using follow-up interviews with a subset of respondents. In those interviews, most classified as insincere admitted that they had not taken the survey seriously.

This debate matters well beyond academic circles. Public agencies, health organizations, and social media platforms use survey-derived estimates to decide where to allocate resources for misinformation countermeasures. If the rate of genuine belief in, say, vaccine conspiracy theories is half of what surveys report, then a one-size-fits-all communications strategy may be built on a false foundation. The study's authors suggest that a smaller, more targeted group of sincere believers exists alongside a much larger group of people who are simply performing distrust. The latter group may still be worth engaging, but they require a very different intervention than the former. "You do not reason someone out of a position they do not actually hold," the lead author said in a press release. "Our data suggest that many apparent conspiracy believers are actually just messing with us."

What This Means for Polling, Politics, and Platforms

The implications are significant for the 2024 and 2028 election cycles. Political campaigns and media organizations frequently commission polls that ask about fringe beliefs, often to demonstrate the supposed insanity of an opposing candidate's followers. A poll that finds "30% of Americans believe the election was stolen by alien computers" may be cited widely, but if most of those respondents were joking, the statistic becomes a form of misinformation itself. The researchers urge polling organizations to adopt new quality-control measures, including multiple catch trials, completion-time thresholds, and post-survey debriefing questions. They also call for peer reviewers to require such checks before publishing conspiracy belief numbers in academic journals.

For social media platforms, the findings suggest that flagging or demoting content based on simplistic "belief" metrics may be less effective than assumed. A user who tweets "I believe the Earth is flat" as a joke is not the same as a user who sincerely argues for a flat Earth in a dedicated forum. Automated moderation systems that treat all such statements equally may over-police ironic speech while missing the smaller, more dedicated network of true believers. Platforms may need to focus their fact-checking resources on the latter group, which is probably more influential in spreading actual misinformation. At the same time, the study does not let sincere believers off the hook. Even if the true rate is 2% rather than 10%, that still represents millions of people in a large country — enough to fuel social movements, organize conferences, and influence local school boards.

The authors plan to make their screening method available as an open-source toolkit, so that other researchers can apply it to their own surveys. They are also planning a follow-up study that distinguishes between different types of insincerity: playful joking, politically motivated trolling, and "satirical endorsement" where a respondent uses the survey to mock the survey itself. Each of these may require different handling. "Surveys are conversations," the lead author said. "If a respondent treats the conversation as a joke, the joke tells us about the respondent's mood, not their worldview. Our job is to tell the difference."

The Road Ahead: Better Questions, Better Answers

The study is a reminder that survey research is as much an art as a science. The same questionnaire can produce radically different interpretations depending on the context, the sample, and the willingness of respondents to be honest. In the age of viral skepticism and internet prank culture, the temptation to answer ironically is greater than ever. Researchers cannot simply assume that a yes is a yes and a no is a no. The study's authors recommend that future surveys design items that are less susceptible to parody, perhaps by asking about concrete behaviors rather than abstract beliefs. For example, instead of asking "Do you believe in the Illuminati?" a survey might ask "Have you ever attended a meeting organized by a group that calls itself the Illuminati?" The former invites a joke; the latter invites a factual answer.

There is also an ethical dimension. If researchers inadvertently amplify the apparent prevalence of absurd beliefs, they may be contributing to the very panic they are studying. The paper's conclusion is measured but firm: "The prevalence of sincerely held conspiracy beliefs is lower than the public discourse suggests. But the prevalence of distrust is not. We should measure both accurately, and not confuse one with the other." As the study enters peer review and eventually publication, it is likely to be cited in court cases, policy briefs, and media commentary. Its central message — that a laugh is not a datapoint — is one that every polling official should take seriously.

A More Honest Picture of a Baffling Phenomenon

The takeaway for the public is simple: when you see a shocking statistic about how many people believe in something outlandish, it is worth asking whether the statistic was collected with care. Many such numbers circulate endlessly on social media, gaining credibility through repetition rather than evidence. Studies using rigorous screening techniques are still rare, but they point toward a more nuanced and sobering reality. Conspiracy theories are not going away, and the people who truly believe them are not vanishing. But the field around them is becoming clearer. The next time a survey reports that a third of the population believes the world is run by shape-shifting lizards, it may be wise to ask whether a third of the population is simply having fun at the researcher's expense. In the end, the study is an invitation to look past the meme and into the data, with the patience and skepticism that the subject demands.