Researchers who followed the same people for five years found that belief in vaccine conspiracy theories splits into two distinct forms: the loud form is rare, while the quiet form is widespread.
The rare version involves claims most people recognize as fringe, such as vaccines containing microchips for tracking or mind control. Very few respondents endorsed it. The common version holds that governments and pharmaceutical companies systematically conceal unfavorable vaccine-related information, and it varied substantially across the sample.
The study, published in Scientific Reports on August 3, 2026, was led by Hitoshi Yamamoto of Rissho University with colleagues at Tsuda University, Tokyo City University, the National Institute of Advanced Industrial Science and Technology, and Meiji University. It analyzed five annual waves of panel survey data collected in Japan each April from 2020 through 2024.
Two Factors, and Only One of Them Is Common
The researchers measured conspiracy beliefs in the fifth wave using the established eight-item anti-vaccine conspiracy belief scale, then ran a factor analysis with maximum likelihood estimation and Promax rotation.
Two factors emerged. The team labeled the first institutional conspiracy, built from items such as the claim that the government is trying to cover up a link between vaccines and autism. The second, implausible conspiracy, captured claims about microchip-based tracking or mind-control experiments.
Both showed high internal consistency, with Cronbach's alpha of .90 and .88, respectively. The correlation between them was moderate at .543, indicating they were related but not redundant.
The distributions differed sharply. Institutional beliefs varied substantially across the sample. Implausible beliefs showed a pronounced floor effect and were endorsed by only a very small minority.
One item loaded moderately on both factors: the claim that immunizations allow governments to track and control people, which the authors noted combines a concern about governmental control with a more implausible tracking narrative.
Skepticism Explained More Than Media Diet Did
The analysis used latent growth modeling, which separates a person's starting level on a measure from how that measure changed over five years, then uses both to predict a later outcome.
Institutional conspiracy beliefs were associated with lower reliance on mass media, persistent skepticism toward mainstream media, increasing skepticism over the five-year period, and younger age.
The most consequential result is easy to skim past. The researchers estimated three models: one with media usage variables only, one with subjective media literacy and media skepticism only, and one with both. Media usage predicted conspiracy beliefs on its own. Once skepticism and subjective literacy were incorporated into the model, the effect of media usage disappeared.
That ordering matters for anyone designing health communication. It suggests the pattern is less about what information reached someone and more about the disposition they brought to whatever reached them. As Yamamoto summarized the research team's findings, "more common forms of conspiracy-related thinking are associated with persistent distrust of social institutions and the mainstream media."
The five-wave structure supports the interpretation that these are relatively stable orientations rather than reactions to pandemic-era news, because the researchers could observe starting levels at the beginning of the pandemic.
Why This Complicates Fact-Supply Strategies
Much of the public health response to vaccine misinformation assumes an information deficit. Correct the false claim, supply the accurate one, and belief should shift.
That model fits the implausible factor reasonably well, and the study found that almost nobody holds those beliefs. It fits the institutional factor poorly, because a person who believes agencies conceal unfavorable information will tend to process a correction issued by an agency as further evidence of the thing being described.
The authors draw the practical distinction themselves. Addressing institutional conspiracy beliefs may require rebuilding trust in mainstream media and political institutions, whereas responding to highly implausible narratives may require more targeted approaches directed at relatively small, potentially isolated groups.
There is a related warning about how researchers read online discourse. Extreme claims endorsed by a small minority may nevertheless attain disproportionate visibility on platforms that amplify striking content, which means computational studies of high-visibility posts may misrepresent what the public actually believes. This aligns with other longitudinal work on conspiracist worldviews, suggesting these orientations develop over long stretches of a life rather than in response to a news cycle.
What This Study Cannot Establish
The limitations are substantial, and the authors state them plainly.
The sample was recruited via an online crowdsourcing platform, so the findings may not fully generalize to populations with different media-use patterns or levels of digital engagement.
Attrition was heavy. The first wave included 2,000 participants, and the fifth included 600, with only prior-year participants invited back each year. Younger participants and women were more likely to drop out, though the researchers reported that attrition was not substantially associated with the main predictors. The mean age of the sample rose from about 46 to about 53 across the study, and the proportion of men rose from about 62 percent to about 72 percent.
Age effects should therefore be read cautiously. The authors note that in this design, age may reflect a combination of cohort-related differences in media environments and selective retention rather than a simple age gradient.
The media measures were broad additive indices that combined heterogeneous platforms, which may obscure meaningful differences between them. The design is observational, so it does not establish that skepticism causes conspiracy belief rather than the two sharing common roots.
Most importantly for American readers, this was conducted in Japan, a media environment the authors describe as comparatively consensus-oriented, with stronger normative expectations of political neutrality and less partisan fragmentation than many Western democracies. The authors present that as a strength, showing conspiracy beliefs can take hold even without a segmented partisan press. It also means the effect sizes should not be transferred to a US population.
The team is explicit that media skepticism alone does not explain vaccine conspiracy beliefs, and that trust in science, perceived ingroup norms, and national solidarity also shape them.
Key Questions Answered
What are the two types of vaccine conspiracy beliefs the study identified?
Institutional conspiracy beliefs hold that governments and pharmaceutical companies conceal unfavorable vaccine information, and implausible conspiracy beliefs involve claims such as microchip tracking or mind control.
Which one is more common?
Institutional beliefs varied widely across the sample. Implausible beliefs showed a pronounced floor effect and were endorsed by only a very small minority of respondents.
What predicted institutional conspiracy beliefs?
Lower reliance on mass media, persistent skepticism toward mainstream media, skepticism that grew across the five years, and younger age. Once skepticism and subjective media literacy were in the model, media usage no longer had an effect.
Does this show that media skepticism causes conspiracy beliefs?
No. The design is observational and associational. It cannot separate whether skepticism drives conspiracy belief or whether both arise from a shared underlying disposition.
How large was the study, and how many people remained in it?
It began with 2,000 participants in 2020 and ended with 600 in 2024. Only prior-year participants were invited back each year, and younger respondents and women were more likely to drop out.
Does this apply to the United States?
The associations may, but the magnitudes should not be transferred. The study was conducted in Japan, whose media environment the authors describe as less partisan and more consensus-oriented than many Western contexts.
What does this suggest for public health communication?
Correcting individual false claims may be ineffective against widespread beliefs, since a correction issued by a distrusted institution can be interpreted as confirmation. The authors suggest institutional trust-building for that group and targeted approaches for the small group holding fringe beliefs.
