A 60 billion tweet study of anti-vaccine misinformation
Provaxx profiles often refer to news media and science sites, while antivaxx profiles much more often refer to YouTube.
One of the reasons why Denmark has lifted all corona restrictions is our high vaccination adherence. About 81 percent of the population has been vaccinated, and 61 percent have received a third shot.
The vaccines, especially the third shot, appear to provide moderate protection against infection, and high protection against serious illness and death.
But why are there some who do not want the vaccines, that virtually all medical science agrees are safe?
WHO has proclaimed vaccine scepticism as one of the biggest threats to global health. A contributing reason why someone is reluctant to get vaccinated is a lack of confidence in vaccines.
There are many different factors that affect the trust in vaccines. Among other things, misinformation on social media can contribute to mistrust of health authorities and create a false picture of the relationship between the pros and cons of vaccines.
In our latest research, we - Sune Lehmann, professor at DTU and I - using a powerful computer have trawled through 60 billion tweets to improve our understanding of today's vaccine scepticism on social media.
Anti- and pro-vaxxers draw on very different sources
Using newly developed methods within ‘deep learning’ and ‘natural language processing’, we taught a computer to identify which attitude to vaccines was expressed in a given tweet.
From there, we were able to identify those users who consistently express strong attitudes for and against vaccines. We use the terms ‘provaxx’ and ‘antivaxx’ to describe these profiles, which make up 45 percent and 3 percent of the profiles in our data, respectively.
To understand the vaccine debate, we then examined which sources these profiles shared information from when talking about vaccines. We did this by creating a computer program that followed links in tweets written by pro- and anti-vaxxers and recording which domain the link pointed to.
It turned out that the two 'sides' of the debate draw on very different resources:
Provaxx profiles often refer to news media and science sites, while antivaxx profiles are far more likely to use links to YouTube videos and to sites known for spreading false news and conspiracy theories.
12 people are behind the majority of the misinformation
We then grouped the sources into a few categories. Using the online Media bias/fact check service, we identified sites that are known for sharing pseudoscience and conspiracy theories.
In addition, we grouped news sites (CNN, Fox, The Guardian, and so on), and Facebook and Instagram as social media. YouTube got its own category as it makes up a large number of links.
Sites that sold products related to medicine or health, we categorized as commercial.
We could see that anti-vaccine profiles far more often share links to pages dealing with pseudoscience and conspiracy theories, which is not so surprising as research shows that tweets with anti-vaccine messages often contain conspiracy theories.
Additionally, antivaxx profiles very often link to commercial sites that offer alternative health products for sale. This is more surprising, as vaccine scepticism is often due to fears of financial conflicts of interest.
In fact, studies show that the majority of vaccine-related misinformation comes from a small group of just 12 people, several of whom make a profit on their venture.
For example, Joseph Mercola has made hundreds of millions selling vitamin supplements and other health products, while spreading misinformation about vaccines and conventional medicine.
Echo chambers confirm our worldview
It is a well-known phenomenon that people interact most with people whose opinions are similar to their own. This 'echo chamber effect' is also well documented on Twitter.
We could confirm that the vaccine discussion on Twitter also largely takes place in echo chambers, where people primarily interact with users who share their views.
We similarly discovered that the sources of information that people are exposed to in their social networks depend heavily on their attitude towards vaccines. The more resistance to vaccines a user expressed, the further from the normal was the media image they could see shared from their social circle.
We call this phenomenon 'epistemic echo chambers' - that is, an echo chamber where (almost) all the 'knowledge' one is presented with confirms the perception one already has.
Common responsibility to combat misinformation
As vaccine misinformation emerges on social media, much of the responsibility falls on the tech giants. But the media also has a responsibility not to report, or give unnecessary speaking time to, medical misinformation.
In Denmark, the media often draw a false balance between views by giving equal, or even more, speaking time to views that are very far from the scientific consensus in the field, and which have at times been directly misleading.
A slightly older example from home is the TV 2 documentary 'The vaccinated girls', which led to a drastic reduction in the number of HPV vaccinations. Researchers have estimated that the misleading TV 2 documentary will be to blame for about 100 cases of cancer, as well as 25 deaths.
An example of better handling of dubious information is a new analysis, where, among others, Jonas Herby from CEPOS claims to show that restrictions do not reduce infection - a claim that goes against all existing knowledge about epidemics. The analysis has not been assessed by peers but has subsequently received harsh criticism from researchers with an understanding of the subject.
While some media outlets - such as Fox News, which has often carried misinformation about the pandemic - have uncritically passed on the results, most media outlets have chosen to consult researchers who have expertise in epidemiology and infection models, and who have typically criticized the non-peer-reviewed working paper in strong terms, both in Denmark and internationally.
If we return to the vaccines, the most common arguments against vaccines are not substantiated in the scientific literature.
Therefore, those covering the topic should not hold up medical information and misinformation as equal views.
- 'Characterizing polarization in online vaccine discourse — A large-scale study', PLOS ONE, 2022, DOI: 10.1371/journal.pone.0263746
- 'Effectiveness of a Third Dose of mRNA Vaccines Against COVID-19–Associated Emergency Department and Urgent Care Encounters and Hospitalizations Among Adults During...', Morbidity and Mortality Weekly Report (2022), DOI: 10.15585/mmwr.mm7104e3
- 'COVID-19 Incidence and Death Rates Among Unvaccinated and Fully Vaccinated Adults with and Without Booster Doses During Periods of Delta and Omicron Variant Emergence', Morbidity and Mortality Weekly Report (2021), DOI: 10.15585/mmwr.mm7104e2
- 'What factors promote vaccine hesitancy or acceptance during pandemics? A systematic review and thematic analysis', Health Promotion International (2021), DOI: 10.1093/heapro/daab105
- 'Quantifying the rise of vaccine opposition on Twitter during the COVID-19 pandemic', Journal of Communication in Healthcare (2021), DOI: 10.1080/17538068.2020.1858222
- 'Adapting and Extending a Typology to Identify Vaccine Misinformation on Twitter', American Journal of Public Health (2020), DOI: 10.2105/AJPH.2020.305940
- 'Covid-19 vaccine hesitancy on English-language Twitter', Profesional de la información (2021), DOI: 10.3145/epi.2021.mar.12
- 'Understanding vaccine hesitancy around vaccines and vaccination from a global perspective: A systematic review of published literature, 2007–2012', Vaccine (2014), DOI: 10.1016/j.vaccine.2014.01.081
- 'Birds of a Feather Tweet Together: Integrating Network and Content Analyses to Examine Cross-Ideology Exposure on Twitter', Journal of Computer-Mediated Communication (2013), DOI: 10.1111/jcc4.12001
- 'The online competition between pro- and anti-vaccination views', Nature (2020), DOI: 10.1038/s41586-020-2281-1
- 'A Dynamic Model of Vaccine Compliance: How Fake News Undermined the Danish HPV Vaccine Program', Journal of Business & Economic Statistics, DOI: 10.1080/07350015.2019.1623045
This article was originally published on our Danish sister site Forskerzonen. Translated by DTU Compute.