Has the use of social networks – and more particularly Twitter – really reflected the population’s support for public health measures during the COVID-19 pandemic?
This is the question explored Hélène Carabinprofessor in the Department of Pathology and Microbiology of the Faculty of Veterinary Medicine and in the Department of Social and Preventive Medicine of the School of Public Health of the University of Montreal, and veterinary epidemiologist José Denis-Robichaudin collaboration with Erin Reesfrom the Public Health Agency of Canada.
To do this, they analyzed more than 40,000 responses obtained by the Angus Reid firm during 27 waves of surveys carried out from September 2020 to March 2022 among the Canadian population.
“We were looking to see how much people used Twitter [rebaptisé X depuis] and what was its impact on adherence to health measures by checking if they talked about the pandemic on this social network and, if so, if they commented on public health measures and in what tone they did so done, explains Hélène Carabin. At the same time, we wanted to understand how they behaved and whether there is a relationship between their behaviors and their use of social media.”
Limited use and influence
Hélène Carabin and José Denis-Robichaud
Credit: Courtesy
Furthermore, only 11% of people surveyed said they did not respect mask-wearing measures and 10.8% refused vaccination. People who used Twitter to express themselves about COVID-19 generally had a lower level of support for wearing a mask and getting vaccinated than those who did not use the platform, but these are individuals who were critical of health measures in their tweets adhered to them the least.
The study also highlights regional and linguistic differences. “There were fewer Twitter users in Quebec than elsewhere,” notes José Denis-Robichaud. But when they were angry about health measures in their tweets, it was reflected more in the behavior of French speakers than English speakers.
This is illustrated by the fact that, among people who criticized health measures, French speakers were seven times more likely not to wear a mask, compared to those who did not use Twitter. For their part, critical English speakers were four times more likely not to wear a mask.
Thus, Hélène Carabin, José Denis-Robichaud and the research team are the first to show that, on an individual level, there is a correlation between the way people reacted to health measures and what they said about Twitter.
Despite this discovery, the researchers are careful not to interpret the results too hastily. “The sociodemographic characteristics of social media users are not representative of the general population, which should encourage caution when using micromessages to evaluate behavior at the general population level,” warns Hélène Rifle.
To use or not to use social media during another pandemic?
The study opens up new perspectives for researchers, who are already considering what comes next.
“We want to know if, in the event of a new pandemic, it would be possible to use X or other social networks to see if and how people are talking about public health measures in order to estimate the rate of adherence to health measures,” they explain.
In this regard, they are working on improving micromessage analysis tools. José Denis-Robichaud presented the preliminary results of a research project at the international French-speaking Adelf-Epiter congress, which brought together French-speaking epidemiologists in the field in Limoges last July.
The objective was to evaluate the ability of an algorithm to read and classify messages published on X according to the tone of the messages and to compare the results with a classification carried out by non-specialists.
“The algorithm was good at classifying the types of public health measures discussed in the messages, but not at grouping them according to the tone or feeling expressed by people,” emphasizes José Denis-Robichaud. But recently we used version 4.0 of ChatGPT, which was found to be almost as good as humans at sorting messages based on tone.”
The next step is therefore to use these improved tools to analyze the impact of opinions expressed on X on the actual effectiveness of public health measures.
“In the event that health measures are severe and people do not want to adhere to them, we could see if this non-adherence influences the rate of transmission of the disease,” concludes Hélène Carabin. This is important for public health, as it would help predict the rate of transmission and could help design other approaches to improve acceptance of measures to mitigate the spread of the disease.”
