Abstract
After the Fukushima nuclear power plant accident, there were various conflicting opinions regarding radiation risks, and especially on social media, there was a marked division between groups that disseminated information based on scientific facts and those that disseminated emotional rather than factual statements. It is necessary to consider how scientists can effectively disseminate correct information to the public, focusing on the use of social media, which is becoming more and more important. We have compiled a set of recommendations for science and risk communication in the new era, based on our research results on the analysis of Twitter big data concerning radiation, corresponding network simulations of information dissemination and direct interviews to influencers, as well as on our experience in Web-based information transmission after the accident. Our studies have shown that experts must send out scientific information promptly. Immediate reaction is most important in fact-checking circulating information and denying wrong messages before they diffuse quickly in the society. There is no time to wait for experts’ consensus, and anyway unified opinions are not trusted as the public suspects them to be biased. The scientific community must be prepared for crises and build cooperation networks among scientists.
INTRODUCTION
The Great East Japan Earthquake that occurred on 11 March 2011 and the subsequent accident at the Fukushima Daiichi Nuclear Power Plant have thrown Japanese society into confusion. In particular, television, weekly magazines, and quickly published books claimed the danger of radiation which made the public anxious. In spite of the government’s efforts to calm them down by preaching safety based on observed doses, many people did not take the message as it was. The social atmosphere was such that scientists who preached safety were strongly criticised for being ‘government scientists’ who spoke on behalf of the government.
Press conferences by the government’s Nuclear and Industrial Safety Agency and TEPCO, the company responsible for the accident, were broadcast daily on television, but their explanations were not to the point, frustrating the public. The government and administrative response was slow and confusing, with additional evacuation orders issued in some areas after they were once declared safe. Rather than dispelling the fear of radiation among the Japanese people, who still remember the atomic bombing, the accident has resulted in a growing distrust of the government and scientists, who may be hiding the truth in an attempt to make the effects of the accident seem smaller. Then, at the end of April, a professor at the University of Tokyo, an expert on radiation protection, held a press conference with tears announcing his resignation as a special advisor to the Cabinet, a shocking turn of events that has decisively discredited the government and made risk communication a complete failure.
Conflict between public opinions was severe especially in social media, which was developing rapidly at the time, and the public was completely divided into two groups: one that preached safety and the other that claimed the dangers of radiation. We have collected radiation-related posts on Twitter (now X) after the nuclear accident and analysed the big data to draw lessons learnt for scientists to effectively disseminate scientific information to the public.
STUDIES ON TWITTER-BASED COMMUNICATION
Twitter data analysis
Twitter data were purchased through a Japanese agency in several different datasets with different time periods and acquisition keywords. The main dataset of our study contains all 24 million tweets and retweets about radiation in the 6 months following the nuclear accident. These posts were grouped by text analysis using a Doc2Vec programme, and the network structure showing retweet relationships was visualised. As shown in Fig. 1, there was a clear division between Group A and C, which made their posts based on scientific facts, and Group B, which was more critical and emotional in their expression, claiming the dangers of radiation (Tsubokura et al., 2018). A characteristic of Twitter and other social media is that a small number of influencers exclusively receive a large number of retweets, and as a result, the influencers’ opinions dominate. In fact, in the dataset examined, half of all tweets were not original tweets but retweets of others, and 40% of the retweets were of content posted by influencers in the top 200, resulting in widespread dissemination of those contents. The influencers in Group B were mostly laypeople with no expertise in radiation or nuclear energy, but they retweeted each other and were closely united. On the other hand, the influencers in Group A were mostly scientists and other experts, while Group C had many media representatives, and they sent out tweets individually and showed little coordination with each other.
Texts of tweet contents collected with basic keywords related to radiation in the 6 months following the nuclear power plant accident were analysed, and user accounts were classified into three groups. This figure, which visualises the network structure showing retweet relationships, shows that influencers in Group B, who often showed emotional expressions, were closely connected with each other, while influencers in Group A and C (many of whom were scientists), who disseminated scientific information, had sparse mutual links.
To build a model of information diffusion in the Twitter space, we made simulations using the retweet network extracted from the data. Since influencers are important, we chose the top nine influencers from each side, i.e. Group A and C and Group B, and tested a simplified model in which information was initially dispatched only from these influencers. Information first reaches direct supporters who directly retweet the influencers’ posts, and in the next step, it is passed on to secondary helpers, as shown schematically in Fig. 2. In subsequent steps, other people receive information from those connected to them in the network, but when they receive opinions from both sides at a time, we assume that they will accept the majority opinion reaching them in the earliest step. Our modified voter model enables the possible flip of users’ opinions at later steps when they are exposed to information dominantly from the opposite side. We simulated the number of people eventually belonging to both sides and refined the model by adjusting the parameters so that the real data could be reproduced better (Sano and Torii, 2021). As people do not tend to change their minds easily once they have believed in something, it is critically important to convey scientifically correct information to the public as quickly as possible.
Conceptual diagram of the information diffusion simulation on the network. It shows how information originating from nine influencers from each of the two camps is transmitted to direct supporters who retweet the post directly in Step 1, to secondary helpers in Step 2, and to others in Step 3 and beyond. The model is based on the voter model, which assumes that those who receive information will accept the majority opinion of the information conveyed in the earliest step, and is modified to include the possibility of switching sides if the opinions from the opposite camp are dominantly received in later steps, while also taking into account their own previous opinions.
Next, we repeated the simulation by adding hypothetical scenarios to the constructed model, searching for the conditions that would best increase the power of Group A and C. We found that rather than increasing the frequency of direct supporters’ retweets or the number of secondary helpers, increasing the number of direct supporters of influencers who disseminate scientific information, i.e. adding new user connections, was the most effective way to strengthen the group (Sano and Torii, 2021). Scientists must cooperate to strategically deal with social media, maybe asking students for help, as well as science communicators and general science enthusiasts for participation in the dissemination of information. Although it is not easy to intervene in the other group, it is important to make efforts to deny scientifically incorrect information as early as possible to prevent its spread. Useful information must be sent out promptly by individual groups of scientists without waiting for the experts to reach a consensus as a unified opinion, and their public communication must be continued with strengthened influential power.
We interviewed a few scientists who were influential on Twitter during the period of 1 to 3 months immediately after the nuclear accident and asked them to describe the situation, the reactions of people around them, and their opinions about the dissemination of information by scientists. A medical professor leading a radiation therapy team at a university hospital and a physics professor specialised in nuclear physics both gained a few hundred thousand followers after the accident. The former’s activity, however, faded out after a month, and as a result, the influential scientist who continued to disseminate information was limited only to the latter physicist who was struggling alone on social media. Interviews revealed how much effort was needed to continue dispatching useful information on social media. In addition to the burden of preparing accurate statements being difficult to balance with daily work, the experts in the line of fire had to be prepared not only to be criticised and called a ‘purveyor of government opinion’ but also to face the risk of threats and lawsuits. Not only in social media but also in television and other media, true experts were forced to refrain from speaking out due to criticism from the public and restraint from their organisations, and as a result, self-claimed experts with personal or political intentions had their chance to speak out and gained popularity.
RECOMMENDATIONS AND PROBLEMS
We have compiled the following recommendations (abridged for this article) for science and risk communication in the new era, based on our research results described in the previous section, as well as on our study of fact-checking Chernobyl-related tweets (Uno et al., 2023) and the experience of our collaborators in Web-based information transmission (Kono et al., 2022) after the Fukushima accident.
What is scientifically correct is not always conveyed to the public.
Scientists must abandon their illusion that the public will be convinced if they explain scientific facts logically. Serious and formal contents will not reach them.
Unified opinion will not be trusted.
Citizens’ anxiety was caused by the government’s emphasis on the unified opinion, which was (suspected to be) biased information.
Visualise (the distribution of) expert opinions.
People are relieved by visualisation and presentation of various (although sometimes contradictory) opinions, rather than by experts’ unique voice.
Scientifically correct information must be sent out at the earliest timing.
Early arrival of information will determine public notion for the successive years.
Incorrect information must be fact-checked and corrected promptly.
Otherwise, this wrong information will spread quickly and prevail for a long time.
The evidence and judgement process should be included in the message.
Our analysis has shown that messages with grounding facts are more convincing and better accepted than those without any mention of evidence.
Information dissemination needs to be strategic.
Comprehensive and repeated communication through various media is important.
Public trust is essential.
People watch the whole personality of the sender to judge trustworthiness.
Influencers and supporters must unite and collaborate.
Scientists must recognise the importance of social interaction. Absence of expert opinions will allow the rise of self-professed specialists and lay critics.
Protect science communicators to ensure their safety and freedom of speech.
Our interviews revealed that the scientific influencers were exposed to frequent slanders and even threats.
Interactive and dialogue-based dissemination is important.
Accepting public feedback on websites and answering questions are important.
Attend to the readers’ anxious emotions and attractive narratives.
Analysing public response and attention helps to understand the problem.
The scientific community and academic societies must be prepared for crises.
Preparation in normal times is important.
Education in radiation sciences is needed in schools and for older generations.
Subjects learned at school are trusted and will become the basis of people’s judgement.
Science and politics must cooperate in public communication.
Pipes of communication must be secured between science and politics. Politicians’ decisions must be transparently based on scientific knowledge.
Science and media (mass media and social media platforms).
Misleading communication and the problem of mass media giving the same weight to major and minor scientific opinions need to be addressed.
CONCLUSIONS
Experts tend to believe that people will understand if they explain scientific facts and that a unique voice as a result of experts’ consensus is important in order not to confuse the public. But our studies have shown that the experts’ unified opinion would not be trusted because the public suspects that they are sending out biased information with the intention to calm people down. The government and the scientists have lost public trust once they have been suspected of being biased and disorganised in their communication. People are relieved by the visualisation and presentation of various (although sometimes contradictory) opinions, rather than by experts’ unique voice. This notion, already commonly known in the field of STS (science, technology, and society) studies, is unfortunately not yet popular among radiation specialists.
We have shown that information dissemination needs to be strategic especially in the new era of social networking, because influencers, who play a key role in the diffusion of information, have a large influence on information networks. Once the opinions of a small number of influencers have spread and dominated, it is very difficult to change the public notion afterwards, regardless of whether they are scientifically sane or not. Some people may be trapped in the filter bubble and echo chamber of the closed network community. Our simulations have revealed that it is critically important on social media to disseminate correct information to as many people as possible at the earliest timing. It is also necessary to counteract scientifically incorrect information, fact-check it, and disseminate it promptly. The scientific community and academic societies must be prepared for crises even in normal times and must work to build cooperation networks among scientists in a variety of fields and to help increase the number of experts who communicate with society.
Footnotes
ACKNOWLEDGEMENTS
This work was supported by the Research Project on the Health Effects of Radiation organised by the Ministry of the Environment, Japan. The authors are grateful to other members of the project for their collaboration and research findings.