Information disorders and misinformation

For several years now, society has been facing new communicative scenarios in which misinformation, or so-called fake news, are the protagonists. Electoral processes, referendums, social and health crises have been affected by these information disorders (Wardle and Derakhshan, 2017) that seek to generate social upheaval to benefit their promoters. The complexity of this mediamorphosis lies in the difficulty in determining the impact and characteristics of these intentional cascades of content, which do not always viralize false information, but also biased and malicious content that intoxicates the public sphere (Del-Fresno-García, 2019).

The interest in this phenomenon has fueled the publication of numerous works focused on understan- ding its architecture and particularities through the classification of false content. Wardle (2017) esta- blishes seven types of disinformation: (1) satire or parody (2) use of inaccurate contexts; (3) impostor content (4) fabricated content (5) false connections (6) false context, and (7) manipulated content. Sa- laverría et al. (2020) point out that this categorization, although useful, mixes models with disinforma- tion procedures, as well as the taxonomy of false content offered by Tandoc Jr., Ling, and Lim (2017) on which they clarify that it includes variants that are more persuasive than false, such as advertising, public relations, and propaganda content. However, Tandoc Jr., Ling, and Lim (2017) delimit this per- suasive force by establishing two levels (high and low) on two variables: “factuality” or the degree to which the contents are based on facts; and the intention of the promoter, which alludes to the objective that the deception seeks.

It is observed that intentionality is one of the fundamental premises when establishing taxonomies such as the one delimited by Althius and Haiden (2018), which establish four levels very similar to those previously exposed: satire, bullshit, propaganda, and disinformation. Also, based on motivation, Maas- berg, Ayaburi, and Yoris (2018) identify eight types of content: real news, fake news, satire, bias, cons- piracy theories, rumor, state-sponsored propaganda, unintentional misinformation, false statements, and sensationalist publications.

In Spain, the study and general categorization of disinformation have focused on its impact on the communication of organizations (Rodríguez-Fernández, 2019a; Martín-Herrera and Micaletto; 2021) and on political communication, addressing post-truth (Rodríguez-Fernández, 2019a; Martín-Herrera and Micaletto; 2021). Ferrandíz; 2019) or the categorization of hoaxes in electoral campaigns (Pania- gua; Seoane and Magallón, 2020).

The presence of false attributions of actions, images, false statements, or photomontages, among other false content, is also observed in these classifications. In a more specialized way, deepfake taxonomies are beginning to be registered (Gómez-de-Ágreda; Feijóo and Salazar-García, 2021), more complex to verify as they are created with artificial intelligence.

It is also worth paying attention, in the case of political communication, to the web pages that support this false content. These “disinformation media” are “web pages that adopt the aesthetics and form of the media to spread propaganda and whose false, distorted, or evaluative content contributes to bene- fiting a political or social cause” (Rodríguez-Fernández, 2021).

They are the protagonists of information disorders by generating an alternative reality that is magnified by the interrelationships with related pages and with interactions on social networks. According to a Digital New Deal study (Kornbluh and Goodman, 2020), participation and engagement with these pages increased by 102% since the 2016 US presidential election. In their digital strategy, social ne- tworks are fundamental vehicles of diffusion but also the link-building work with similar pages that contributes to creating and reinforcing social bubbles that promote polarization (Nikolov et. al., 2015; DiFranco; Gloria-García, 2017).

Thus, the disinformation of extreme right-wing ideologies causes more adherence and, therefore, more viralization than those of the extreme left, registering an average of 426 interactions per thousand followers per week (Edelson et al., 2021). Achieving high viral content or manipulating social interac- tions is also an objective of manipulation and a fast track to position issues in public debate.

Fact-checkers acknowledge finding more misinformation in politics, followed by hoaxes about immi- gration —in different situations, such as events or crises, but always from hate speech— and, lastly, health and science (Rodríguez-Fernández, 2019b). It should be noted that in crises, users tend not to check the information and believe exclusively what they see on the internet (Bairong and Zhuang, 2017).

This situation of social weakness is taken advantage of by the promoters of this type of practice to give credibility to their messages. In fact, although it is in 2010 when the first fake news about electoral processes in Massachusetts began to be recorded, it is not until 2014 when the first intentional infor- mation begins to be observed that blames immigrants for being carriers of the Ebola virus and which also serve as a pretext for some North American politicians to debate closing borders (Waldrop, 2017).

Disinformation during the COVID-19 pandemic

The COVID-19 pandemic has generated a historical social situation that contributes to a notable in- crease in misinformation related to the virus. “We are not only fighting against an epidemic but also against an infodemic” declared in February 2020, the Director-General of the World Health Organi- zation (WHO), Tedros Adhanom. The term “Infodemic” has been used since then to identify disinfor- mation that refers specifically to the virus, but also to content that has been used as a pretext to affect other governments, political leaders, or institutions.

The research works carried out during the pandemic have contributed to delving into some key aspects of this type of disinformation, such as the issuers, the most used channels, the intensity, or the type of viralized content depending on the social context. Gallotti et al. (2020) developed during the first phase (January-March 2020) an infodemic risk index, based on the analysis of 112.6 million tweets. They concluded that G8 countries such as Russia or Germany presented a higher infodemic risk than develo- ping countries such as Thailand or the Philippines and observed that the average infodemic decreased as the number of infections increased. Also in this first stage, Pérez-Dasilva; Meso-Ayerdi, and Men- diguren-Galdospín (2020), observed the conversation on Twitter, noting that some institutional actors, such as Donald Trump, then President of the United States, contributed, supported by related media, to the spread of disinformation. The authors also point out that although Italy and Spain were two of the most affected countries, they did not have a special role in the global conversation.

In Spain and since the start of the pandemic, Maldita has detected 1,027 lies, false alerts, and misin- formation about the coronavirus (Maldita; 2021). The content and intensity have been adapted to the evolution of the pandemic itself. López-Pujalte and Nuño-Moral (2020) perceived an increase in ho- axes after the declaration of the State of Alarm (March 14th), reaching higher values between March 20th and April 11th. Observation also made by Sánchez-Duarte and Magallón (2020) on the study of verifications that goes from 32.5% before confinement to 67.5% in the month after.

From the self-perception of citizens, Losada, Rodríguez-Fernández, and Paniagua (2020) collect du- ring the first phase of confinement (March and April 2020) that 93% of those surveyed (N: 1,823) claimed to have received fake information or hoaxes. A very similar percentage to that registered by Fernández-Torres, Almansa-Martínez, and Chamizo-Sánchez (2021), in which 92% of respondents ac- knowledged having received disinformation (N: 1,115), identifying WhatsApp as the preferred chan- nel (86%), followed by Facebook (58%), and Twitter (31%).

These channels, even in the same order, are also collected by Salaverría et al. (2020) who conducted a content analysis of verifications (N: 292) between April 20th and 26th, categorizing much of it as deception (64.4%), followed by decontextualization, and exaggeration (17.1% respectively) and, to a lesser extent, jokes and memes (1.4%). Regarding the most widely distributed types of misinforma- tion, and also on the verification study (N: 166), Sánchez-Duarte and Magallón (2020) determined that 24.7% were hoaxes about prevention and measures, respectively, 21% did allusion to hoaxes about contagion, and 28.9% to other issues. Similarly, Castro et al. (2020) concluded that one in five hoaxes dealt with the prevention or cure of the coronavirus, argued on the principle of medical authority, and fundamentally promoted the use and consumption of substances of natural origin. Conclusions were similar to those obtained by Moreno-Castro et al. (2020) who, after analyzing text and video chains shared on WhatsApp, identified that, in the majority, the subject presented themself as medical person- nel (36.5%) or as anonymous person (30.9%).

The medical community has been openly concerned about the impact of misinformation. According to data from the III Study on Hoaxes in Health, 92% of the health professionals surveyed believe that the crisis has increased the number of hoaxes, and 3 out of 4 claims to have seen their patients worried about fake news. More disturbing are the data linked to treatments, since 65% of those surveyed have attended consultations about treatments for covid without scientific evidence (Ramírez; 2020). De- nialism and anti-vaccine movements seem to have had, precisely with the disinformation of political intention, quite a leading role in this crisis.

Magallón (2020) points out that during the first stage of the crisis, a greater presence of hoaxes about contagion stood out, while as the confinement was relaxed, political misinformation abounded. It can- not be ignored that the pandemic has been an argument used to create narratives that affect certain countries and groups. EUvsDisinfo (2020a) identified that, among the narratives about COVID-19 with a political focus, those that exposed that the United States had created the virus, that the European Union was not facing the crisis, attacks on China as the alleged creator of the virus, or the idea that the pandemic was part of a global conspiracy, among others, stood out. The European Union also warned about false theories that pointed to Jews as the creators of the virus (La Vanguardia; 2020).

Vaccines have also been the protagonists of disinformation in the pandemic. The European External Action Service exposed that foreign state actors have “sensationalized and misrepresented information on the safety of vaccines manufactured in the West and have fueled anti-vaccination movements within the EU” (EUvsDisinfo, 2020b). The G7 countries have also publicly stated their intention to develop common guidelines to combat Russian and Chinese disinformation aimed at generating mistrust about electoral processes and vaccines (Euronews; 2019).

Organization and discourse of anti-vaccine movements

Far from the digital sphere, the first anti-vaccine movements were born in the second half of the 19th century when the United Kingdom promoted several laws to force its population to be vaccinated against smallpox. Then, as incipient pressure groups, the Anti-Vaccination League, created in 1853, and later, in 1867, the Anti-Compulsory Vaccination League, were formed. Between 1870 and 1880, numerous pamphlets, books, and journals against vaccination were published. Their influence began to spread to other European countries and in 1872, vaccination rates in Stockholm had fallen by 40% while in Sweden the decline reached 90%. Finally, the pressure of these first anti-vaccine movements forced the British Government to issue, in 1898, a new Vaccination Law in which sanctions were eli- minated and a conscientious objection was included for the first time, allowing parents who did not believe that the vaccination was effective or safe to obtain an exemption certificate. The situation was also replicated in the US where anti-vaccine leagues also managed to repeal anti-vaccine laws (Wolfe and Sharp, 2002).

What were originally constituted as pressure groups are now denier movements that threaten herd im- munity through a discourse focused on the discussion of the safety and effectiveness of vaccines. The ultimate goal is to generate mistrust around health personnel and official information sources, mainly the Government and pharmaceutical companies (Santillán and Rosell, 2019).

Schmitd and Betsch (2019) identify five basic tricks used by deniers in the construction of their argu- ment: (1) resort to fake experts (such as a healer), (2) appeal to conspiracies, (3) ask for the impossible, for example, that vaccines are one hundred percent safe and effective, (4) use the logical fallacy or the false dilemma, in which two points of view are presented as the only alternatives of thought, and (5) self-selecting the data, a practice also known as cherry-picking, in which information is selected based on argumentation, choosing data, examples, and particular cases while avoiding information that does not meet the argumentative objective.

Among the most popular arguments is a study published in 1998 by Dr. Andrew Wakefield that asso- ciated the triple viral vaccine (measles, rubella, and mumps) with autism. The study has been refuted as it was shown that its results were biased since, at that time, the author had applied for a patent for a measles vaccine and he intended to discredit the triple viral in parallel. The journal withdrew the publi- cation from its repository, the rest of the authors retracted the work, and even Wakefield himself lost his medical license due to falsifying evidence (Salleras, 2018). However, the false correlation is still valid.

The pandemic has fueled the disinformation activity of these movements and the social media accounts of anti-vaxxers increased their followers by at least 7.8 million during 2019. It is estimated that 31 million people follow them on Facebook and 17 million people subscribe to similar accounts on You- Tube. This translates into a significant economic benefit for social networks: it is estimated that the an- ti-vaccine movement could generate annual revenues of one billion dollars for social media companies (Burki, 2019). In December 2020 alone, when the vaccination program was launched at the European level, fact-checkers issued 128 articles in 16 countries (Sessa, 2021). Likewise, recent research from the Center to Counter Digital Hate found that just 12 anti-vaxxers are responsible for nearly two-thirds of related misinformation on social media. It is estimated that they reach 59 million followers through four fundamental networks: Twitter, Facebook, YouTube, and Instagram (McNamara, 2021).

Topic of hoaxes

On both platforms, the issue related to the pandemic that has been denied the most is that of vaccines. Thus, hoaxes about vaccines account for 54% of all the lies spread about COVID-19 in the case of Newtral and 48% in the case of Maldito Bulo.

The two platforms also coincide in the second most detected COVID-19 issue: that of masks. But from then on, so many similarities are no longer detected since it is the aid that ranks third in the case of Newtral and the origin of the virus in Maldito Bulo. Curiously, neither Newtral publishes fact-checks about the origin of the virus nor Maldito Bulo on aid.

Virus variants; the virus in other countries; measures allegedly taken by the authorities; alleged state- ments by political leaders; alleged remedies, medical treatments, and consequences of the use of PCR are the main issues on which the fact-checks analyzed with COVID-19 as the protagonist deal.

Source: Own elaboration.

Source: Own elaboration.

However, most of the hoaxes detected by the fact-checkers are not about COVID-19 but various other topics. Both Newtral and Maldito Bulo coincide in detecting fake content linked to national and inter- national political debate: more than half of hoaxes unrelated to the pandemic are related to political debate.

Besides this topic, Maldito Bulo dismantles fake news about phishing, vishing, or scams; related to so- cial networks and troll accounts; violence (sexism, terrorism, etc.); immigration, racism, xenophobia; 5G, technology, and robotics; health; education; security forces and bodies; and others, like nature. For its part, Newtral denies, besides hoaxes about the political debate, others that deal with education; health; security forces and bodies; lies related to social networks; phishing, and courts.

Source: Own elaboration.

Source: Own elaboration

Protagonists of the analyzed fact-checkss

The main fake news detected have as protagonists the Government of Spain (more than 8%), regional governments, and other institutions. In other words, the hoaxes deal with these institutions and fake decisions or fake measures are attributed to them. Between the three, they account for almost a quarter of all fake information, specifically, 22.8%. Its prominence is maintained both with news about CO- VID-19 and other topics.

As for the political leaders, Pablo Iglesias is the one who stars in the most hoaxes, perhaps because the study period coincides with the end of his position as Vice President of the Government when he pre- sented himself as his party’s candidate in the regional elections in Madrid. In second place are Pedro Sánchez and Santiago Abascal, who star in the same number of fake news.

The high number of hoaxes (more than 4%) about international political leaders, from different insti- tutions, is striking. Likewise, companies, entrepreneurs, and business associations stand out, as they are the protagonists in 10% of the fake content detected. Hoaxes about companies generally address issues unrelated to the pandemic.

To a lesser extent, hoaxes that have as a protagonist religious organizations and religious leaders; fe- minist organizations; LGTBI+ collective; immigrants; NGOs; other third sector organizations; mutual and private insurance; hospitals, health residences and organizations; educational centers, teachers, students; and nature (especially animals) have also been dismantled.

Table 3. Protagonists of hoaxes.

Source: Own elaboration.

Origin of the hoax

Anonymous people or organizations are the creators of 30% of the hoaxes, unknown people or organi- zations of 20%, and famous people of almost 10%. In other words, half of the misinformation detected has been created by people who are not socially known or by people whose identity is unknown. The same thing happens with hoaxes outside the pandemic as with misinformation about COVID-19. As for the hoaxes created about COVID-19, it should be noted that sometimes the origin lies with denialist organizations and people (mainly, hoaxes about vaccines and fake news about masks).

It is also striking that more than 15% of hoaxes have emerged in the media. Curiously, Maldito Bulo is the platform that supervises the media the most since Newtral has not detected any hoax in the analyzed period.

For their part, according to what was published by the fact-checkers, it is the right-wing and far-right parties, parties currently in opposition in Spain, who have created the most hoaxes in the analyzed pe- riod. However, the political parties, as well as the leaders of the formations that make up the coalition government in Spain, are the ones that have spread the fewest hoaxes.

Finally, another significant fact is that in 15% of the cases, Newtral and Maldito Bulo do not specify the origin of the hoax in their fact-check. In this percentage, therefore, who is behind these fake con- tents is unknown.

Table 4. Origin of the hoax.

Source: Own elaboration.

Transmission and intent of the hoax

In more than 90% of the hoaxes detected by the fact-checking platforms, the content is clearly contrary to some person or organization or contains insults, bad words, or foul language against those people or organizations, that is, it is disinformation that pursues the discredit of said personalities or institutions. This negativity is greater in the case of hoaxes about the pandemic (91%), especially in the case of those detected by Newtral (95%).

In only 1.5% of cases, fake content refers to beneficial consequences or notable actions of a person or group, that is, they can be classified as positive. For their part, 7.8% present a neutral tone, limiting themselves to offering data, without entering into evaluations, and without falling into foul language or insults.

Where does the hoax spread and circulate? The results show that social networks are the main transmi- tters of disinformation. In 40% of the cases, it is indicated that the hoax is circulating through various social networks. Regarding the networks, the results show that Twitter is the network in which fake content is spread the most (20%). The 13% of hoaxes that circulate through the media and the 13% of hoaxes that circulate through WhatsApp are also very significant.

It is surprising that 8% of the hoaxes detected were distributed by phone calls and that there is more dismantled fake telephone information than there was distributed by email (7%). The phone hoaxes have been detected, in their entirety, by Maldito Bulo. These fake news stories mainly refer to scams in the form of phishing and vishing.

To a lesser extent, fact-checking platforms have also dismantled lies circulating on Facebook, instant messaging (not WhatsApp), Instagram, and YouTube. In only 4% of cases was it not specified where the hoax had been distributed.

Source: Own elaboration.

What days of the week do they post the most fact-checks? The results show that Tuesdays, followed by Mondays and Wednesdays, are the days when more fake news is uncovered, although, on weekends, both Saturdays and Sundays, the platforms continue their work of highlighting the circulation of hoa- xes

In this sense, there are some differences between Newtral and Maldito Bulo. While Newtral shares more fact-checks on Tuesdays and Wednesdays (Wednesdays especially refer to hoaxes on topics other than the pandemic), Maldito Bulo does so on Mondays and Tuesdays.

Hoax language

The majority of fake news detected on both platforms is based solely on text: in 41% of cases, the hoax only contains text. The second most used option is the text and image option (27%), which is located especially in the verifications shared by Maldito Bulo. In 18% of cases, the denied news is distributed in video format, with Newtral detecting this type of content to a greater degree.

To a lesser extent, the platforms also dismantle hoaxes in audio (almost 8%) and memes/still images (5.5%). In some cases, Newtral and Maldito Bulo do not indicate the language of the hoax, although this happens very rarely (less than 2%).