When we began this research in March 2020, conspiracy theories were not what we had in mind.
COVID-19 was rapidly spreading around the world, and Japan was entering an unusual period of voluntary restrictions. Unlike some countries that imposed legally enforced lockdowns, Japan largely relied on people voluntarily refraining from going out, gathering, and engaging in other activities. As researchers interested in cooperation and social dilemmas, we immediately saw this situation through the lens of a classic problem: individuals were being asked to restrict their own activities for the benefit of society as a whole.
Why would people cooperate under such circumstances? And why would some people cooperate more than others?
With these questions in mind, we hurriedly designed a questionnaire in March 2020. Our first survey was conducted on April 3, just as voluntary restraint was intensifying in Japan. We recruited 2,000 participants and asked about their attitudes and behaviours, including how they obtained information from mass media and social media.
At the time, we did not know that this would become a five-year project.
A changing pandemic brought new questions
As the pandemic continued, the social issues surrounding COVID-19 changed. Vaccines became available, but misinformation and opposition to vaccination also became increasingly visible. News reports described harassment of healthcare workers, anti-vaccine demonstrations, and extraordinary claims circulating online, including stories about microchips being implanted through vaccines.
These developments raised a new question for us. During a public health crisis that required collective cooperation, why were some people engaging in behaviours that appeared to run directly against this collective effort?
We therefore became interested in conspiracy beliefs surrounding vaccination and incorporated them into our continuing panel study. We surveyed the same panel annually from 2020 to 2024. Of the original 2,000 participants, 600 participated in the fifth wave in 2024.
This gave us an opportunity that we had not anticipated when we designed the first questionnaire.
When researchers study conspiracy beliefs using a cross-sectional survey, they can observe what people believe and what their current attitudes and media habits are. But it is much harder to know what came before. If someone who believes a conspiracy theory also distrusts mainstream media, did that distrust precede the conspiracy belief, or did it emerge alongside it?
Our panel data allowed us to approach this question differently. We had already measured participants' media use, subjective media literacy, and scepticism toward mainstream media repeatedly over several years. We did not have to ask people in 2024 to remember what they had thought about the media years earlier. We had measured those attitudes at the time.
Not all conspiracy beliefs looked alike
One of the most striking findings came from examining the conspiracy beliefs themselves.
The eight questions we used separated into two factors. The first involved distrust of established institutions, including beliefs that governments, pharmaceutical companies, scientists, or academics were concealing information about vaccines. We called these institutional conspiracy beliefs.
The second involved more extreme claims, such as vaccines containing tiny devices for tracking people or being used in mind-control experiments. We called these implausible conspiracy beliefs.
The distinction mattered because the two types of belief had very different distributions. Institutional conspiracy beliefs varied considerably across participants. By contrast, the implausible claims were rejected by the great majority and endorsed by only a small minority.
This made us reconsider what we mean when we talk about "conspiracy beliefs." The most spectacular claims may attract enormous attention, particularly online, but their visibility should not automatically be interpreted as evidence that large numbers of people believe them.
Media use was not the whole story
Our five-wave data also allowed us to distinguish relatively stable attitudes from changes occurring during the pandemic.
For institutional conspiracy beliefs, lower use of mass media was initially associated with stronger conspiracy beliefs. But a more important pattern emerged when we considered media use together with people's attitudes toward media.
People who had already been more sceptical of mainstream media, and those whose scepticism increased over the five-year period, tended to show stronger institutional conspiracy beliefs in 2024. Once these attitudes were taken into account, the association with media use itself disappeared.
This suggests that the issue may not simply be which media people use. A more fundamental factor may be how people relate to established sources of information and how much they trust them.
We should be cautious about causal interpretation. Our study does not show that media scepticism by itself causes conspiracy beliefs, and many other psychological and social factors are likely to be involved. Nevertheless, the longitudinal data show that institutional conspiracy beliefs are associated with orientations toward mainstream information sources that persist and change over time.
The implausible conspiracy beliefs told a different story. Because so few respondents endorsed claims involving microchips or mind control, our statistical models had limited explanatory power. These results should therefore be interpreted cautiously.
The loudest voices may not represent the majority
During the pandemic, extreme conspiracy claims were highly visible on social media. Such content is also relatively easy for researchers to observe: large numbers of public posts can be collected and analysed computationally. But highly visible online content does not necessarily tell us how widespread a belief is in the population.
Our results raise a simple but important possibility: the most visible conspiracy theories may not be the most widespread ones.
Claims involving microchips and mind control may attract attention precisely because they are extraordinary. Yet in our sample, such beliefs were rare. The more widespread phenomenon was less sensational: persistent scepticism toward mainstream media and established institutions.
For public communication, this distinction matters. Responding to a small number of extreme claims may require very different strategies from addressing a broader erosion of trust in institutions and information sources.
A study we did not originally plan
Looking back, perhaps the most interesting aspect of this project is that we could not have designed it in 2020 knowing where it would end.
We began with a question about cooperation. In March 2020, we saw voluntary restrictions in Japan as a real-world social dilemma and wanted to understand why people would restrict their own activities for the collective good. As the pandemic evolved, new social problems emerged, and our research questions evolved with them.
Five years later, the data collected for that original purpose gave us an unexpected window into the attitudes associated with vaccine conspiracy beliefs.
Longitudinal research requires patience, and sometimes its value becomes clear only later. Social crises change, public concerns change, and researchers' questions change as well. Following the same people through those changes can reveal patterns that would have been impossible to anticipate at the beginning.
Our experience with this study reminded us that sometimes the most valuable part of a longitudinal project is not only answering the question with which it began, but also being able to ask questions that did not yet exist when the first data were collected.