Crude, anonymous, partisan, sectoral and anti-elitist. Electoral disinformation in Spain (2019-2022)

The aim of this research is to analyse the electoral misinformation circulating in the public space as a whole during campaign periods. The sample is made up of 481 rebuttals published by the verification media Maldita.es , Newtral , Efe Verifica and Verificat in relation to 409 pieces of misinformation during the six campaigns carried out in Spain during the political cycle that began in the general election in November 2019, which also includes the regional elections held in Galicia, the Basque Country, Catalonia, Madrid, Castilla y León


Introduction
The appeal of analysing the spread of "deliberately" false information in the public sphere (Salaverría et al., 2020, p. 2) increased with Donald Trump's rise to power in 2016 (Orbegozo-Terradillos; Morales-i-Gras; Larrondo-Ureta, 2020).Subsequent research has shown that during the US electoral campaign, 10% of the tweets contained hoaxes or conspiracy theories, and 15% presented extremely biased information (Bovet;Makse, 2019).However, if there is one time when the spread of hoaxes reached a peak, it was in the early 2020s when, in the context of the COVID-19 pandemic, misinformation on health, science, and health policy issues increased dramatically (Brennen et al., 2020).
The phenomenon of disinformation has been present in politics for decades (Salaverría et al., 2020).However, the consolidation of digital platforms has made it easier for citizens with ideological interests (Herrero-Diz; Pérez-Escolar; Plaza-Sánchez, 2020) to share any kind of negative information originated by elites (Chadwick;Vaccari, 2019), and to be self-sufficient in the production and distribution of potentially viral messages.The content of these messages is twisted to display misleading information and offer political actors the opportunity to open up a range of nodes through which to disseminate disinformation, fuelling populist drifts and employing increasingly crude and critically-charged language (Ernst et al., 2017), which proliferates in crises (Pérez-Curiel; Velasco-Molpeceres, 2020).
Political disinformation is a social problem, as it promotes a clear polarisation in vote casting, a weakening of democratic health (Pira, 2019) and an increase in distrust towards the media and political actors as a whole (Casero-Ripollés; Doménech-Fabregat; Alonso-Muñoz, 2023).Citizens consider this to be a problem that puts the democratic system at risk.In this sense, Herrero-Diz, Pérez-Escolar and Plaza-Sánchez (2020) note that 79% of the Spanish population recognises their difficulty in differentiating between false and untrue content.Faced with the growing concern about this phenomenon, information verifiers, media and social researchers have tried to alleviate the situation, delving deeper into the search for disinformation and identifying what type of disinformation is promoted, through which channels and in what intensity (Almansa-Martínez; Fernández-Torres; Rodríguez-Fernández, 2022).However, as Casero-Ripollés, Doménech-Fabregat and Alonso- Muñoz (2023) show, people still perceive a clear lack of resources to solve the problem, and they blame the media and political actors themselves for not being forceful in eradicating disinformation in the digital sphere.
Disinformation research is becoming increasingly widespread in the social sciences, especially about the COVID-19 pandemic (García-Marín; Salvat-Martinrey, 2021).Even on occasions when the object of study is political, its centrality lies in the investigation of the narrative surrounding the coronavirus (Pérez-Curiel; Velasco-Molpeceres, 2020), with little research seeking to respond to the typology and discourse of disinformation in campaign periods (Rodríguez-Hidalgo; Herrero; Aguaded, 2021), where disinformation plays a fundamental role in the deconstruction of the political image.
The aim of this research is therefore to analyse the electoral disinformation circulating in the public space as a whole during the campaign periods.The sample is made up of 481 rebuttals published by the verifying media Maldita.es,Newtral, Efe Verifica and Verificat on 409 pieces of disinformation covering the six campaigns carried out in Spain during the political cycle that began with the general election in November 2019.Specifically, the study analyses the scope of generation and/or dissemination of disinformation, its format, its issuers and sources, its protagonists, its themes, and its discourse, with particular attention paid to populism and negativity.

Background and disinformation typologies: sources, formats and spheres of dissemination of hoaxes in the contemporary public sphere
The internet was the main access point used by citizens during the 2020 state of emergency to obtain information related to the COVID-19 pandemic (Brennen et al., 2020).Despite the efforts made by governmental organisations, which invested large amounts of money on online information resources, and by social networks themselves, which introduced special information services such as the COVID-19 Info Center proposed by Facebook (Nielsen et al., 2020), the spread of hoaxes in the digital public sphere was inevitable.
Social networks themselves were the main channel for transmitting disinformation about the coronavirus (Nielsen et al., 2020).Published studies on this phenomenon (Magallón-Rosa; Sánchez-Duarte, 2021; Rodríguez-Andrés, 2018) point to the consolidation of disinformation in modern Western democratic societies not being understood without taking the formal characteristics of the media and social context into account (Pérez-Curiel; Velasco-Molpeceres, 2020).Thus, before the advent of the internet and dig-Within the hybrid media ecosystem misinformation circulates faster and spreads more widely ital platforms, disinformation circulated more slowly, as it was propagated in a local context (Gutiérrez-Coba; Coba-Gutiérrez; Gómez-Díaz, 2020).The epicentre of its origin and its reproduction format was reduced to fake news and oral information.
In contrast, within the hybrid media ecosystem (Chadwick, 2013), misinformation circulates faster and spreads more widely (Vosoughi;Roy;Aral, 2018).At a time of decline in journalism (Calvo; Cano-Orón; Llorca-Abad, 2022), when the traditional role of the media as information watchdogs is being diluted, social networks, with their innovative and user-friendly design, provide users with digital tools that allow them to create disinformative content.The lack of digital literacy required to create hoaxes (Cerdán-Martínez; García-Guardia; Padilla-Castillo, 2020) has favoured the implementation of cheapfakes (Gamir-Ríos; Tarullo, 2022), i.e. disinformation created in an amateurish way by users using their own mobile devices.
Despite pressure from different spheres (Weidner; Beuk; Bal, 2020), social networks have not implemented effective measures against disinformation and are the ideal space to disseminate hoaxes.In this scenario of information confusion, the work carried out by fact-checking platforms is increasingly necessary (Ramon-Vegas; Mauri-Ríos; Rodríguez-Martínez, 2020).Some studies have pointed out that these agents are key in the collection, verification and dissemination of falsehoods circulating in the public sphere (Ramon-Vegas; Mauri-Ríos; Rodríguez-Martínez, 2020).Others have taken a more nuanced view, finding that in periods of open crisis, as in the case of COVID-19, citizens tend to give more credibility to the traditional media, especially television (Masip et al., 2020).
The social impact that the phenomenon of disinformation has had in recent years (Calvo et al., 2022) has led many researchers to take an interest in studying the anatomy of hoaxes.To this end, several works published to date have studied the composition of the disinformation collected by the various verifiers, and the object of analysis ranges from the format, through the sources, the protagonists, and the degree of alteration of the hoaxes.In terms of the main format of the disinformation, the hoaxes that have circulated about COVID-19 have been mainly textual, both exclusively and combined with other formats (Peña-Ascacíbar; Bermejo-Malumbres; Zanni, 2021;Salaverría et al., 2020).
In the political sphere, Paniagua-Rojano, Seoane-Pérez and Magallón-Rosa (2020) noted the prevailing use of text on social media during the 10 November 2019 election, and López-Martín, Gómez-Calderón and Córdoba-Cabús (2023) found that the majority of political hoaxes debunked by International Fact-Checking Network (IFCN)-accredited verifiers between 1 January and 31 March 2022 featured plain text (32.56%), while 30.23%contained the misinformation in the text embedded into the image, and 19.77% of unretouched photographs contained purely textual falsehoods.
Following that which has been described in this section, we offer the following research question: RQ1: In what sphere, who are the protagonists and their attributes, and with what format, sources and degree of alteration is political disinformation disseminated in electoral campaigns?

Us versus them: the construction of populist disinformation agendas
The overabundance of information resulting from the new communicative paradigm, in which the combination of media, political, and citizen logic is fundamental in creating a vision of social reality (Casero-Ripollés, 2018), causes major issues to be displaced to a position far removed from popular interest, while sensationalism, demagogy, and emotionality occupy a privileged space (Orbegozo-Terradillos; Morales-i-Gras; Larrondo-Ureta, 2020).Information chaos is the product of a modern, liquid society (Bauman, 2002), in which objective facts have less and less influence and propagandistic narrative practices condition political news (Bañuelos-Capistrán, 2020).
Around this idea, Pérez-Tornero et al. (2018) coined the term "media neopopopulism" to refer to a populism supported by the spectacular nature of politics, the erosion of the classical media, the need to bring leaders with media charisma to power and, as has been said, the hatching of emotionality to the detriment of the dissemination of logical issues (Orbegozo-Terradillos; Morales-i-Gras; Larrondo-Ureta, 2020).In climates where ideological polarisation is the norm (Llorca; Fabregat-Cabrera; Ruiz-Callado, 2021), "the constant deepening of divisions, the theoretical rejection of any kind of negotiation, and the constant hammering of conspiracy theories" (Waisbord, 2020, p. 272) Disinformation does feed into the polarising strategy of populism are, according to Arias-Maldonado (2016), the basis for constructing the stereotype of the antagonistic "other", i.e., a corrupt elite constantly confronting a pure and noble people (Savarino, 2006).
The construction of this antagonistic vision of reality can be seen in the communicative strategy followed by far-right activists and groupings in Germany, through the use of disinformation with an "unhygienic" theme focused on discrediting the government's management of COVID-19, the latter being understood as a fraction of the "elite establishment" (Vieten, 2020, p. 12).In Europe, Pérez-Curiel and Rivas-de-Roca (2022) identified conspiracy theories, corruption and the economy as the main themes that followed a strategy framed around criticism of the elite and defence of the commons, and which contained hoaxes that came to be propagated through mainstream media.
The negative connotation that has surrounded contemporary populism is therefore evident.In his work, Laclau explained that no political action distances itself from populism, as the construction of the people and social division are the "political acts par excellence" (Laclau, 2005, p. 195).This assumption conditions all definitions that have tried to offer empirical knowledge about populism, being diverse, broad and ambiguous (Groshek;Engelbert, 2013).This phenomenon is not just therefore an ideology or political regime (Mouffe, 2019), but a particular form of political discursivity (Lava-Santos, 2023).
From the perspective of Aalberg et al. (2017), this conceptualisation of populism rejects any pejorative and authoritarian connotations and can occur in left-wing or right-wing groupings (Jagers;Walgrave, 2007).From this premise, it is interpreted on one hand that populist movements can emerge in historical periods marked by economic and political crises, and as proof of this are the extreme right-wing nativist parties such as the National Front (France) or the Northern League (Italy).On the other hand, it is also deduced that parties that were already present in political life, as well as the social audience itself, are impregnated with the identifying elements of this populist discourse (Lava-Santos, 2023).
While populism does not need disinformation to reach certain social sectors, disinformation does feed into the polarising strategy of populism (Waisbord, 2020).Deploying an "us/them" binary narrative (Wondreys; Mudde, 2022), disinformation actors often employ this dialectic by avoiding political debate (Rivas-Venegas, 2021), claiming the need to recover popular sovereignty, disowning groups and collectives considered demeaning, and repeatedly appealing to patriotic and cultural values (Engesser;Fawzi;Larsson, 2017;Ernst et al., 2019).
Based on the cognitive bias hypothesis (Saiz-Vélez, 2020), according to which information on political issues affects individuals more than any other issue (Orbegozo-Terradillos; Morales-i-Gras; Larrondo-Ureta, 2020), several authors have sought to ascertain which populist discursive elements mark the political agenda of candidates in electoral campaigns.
As recent evidence, Lava-Santos (2021b) observed in the electoral campaign for the 2021 Catalan Parliament election that defending the commons and the ostracism of immigrant groups are presented in issues related to health, education, or work, while criticism of the elites and the defence of popular sovereignty are framed in ideological and personal issues.Internationally, Pérez-Curiel and Domínguez-García (2021) showed that Trump's criticism of elites is conditioned to campaign-related and general ideological issues, while patriotic values are introduced in sectoral issues.
However, the populist behaviour of political hoaxes spread in the digital sphere has not been analysed in detail.We therefore propose the following research question: RQ2: What populist discursive elements are most frequently employed in the various disinformative political issues in electoral campaigns?

Negative disinformation campaigns: from the unofficial to the political
When the First World War ended, the Bolshevik political police first used the term 'disinformatzia' (Rodríguez-Andrés, 2018) to describe actions aimed at preventing the establishment of the communist regime in Moscow.Linked to its warlike origin, Rodríguez-Andrés (2018, p. 235) stresses that any disinformative element aims to "attack the one who is considered as an adversary".In politics, the communicative practice characterised by criticising and attacking the rival acquires the name of negative campaigning.Far from being a novelty (Greer;LaPointe, 2004), its consolidation as an electoral strategy dates back to the 1950s, with the birth of television.Nowadays, this type of campaign has been strengthened thanks to the exceptional space for expression created by social networks (Cabo; García-Juanatey, 2016).
Both concepts -disinformation and negative campaigning-are closely related, especially in the consequences they can have on society (Aguerri; Miró-Llinares, 2023), but they also present differences.Disinformation is based on the use of lies (Aguerri; Miró-Llinares, 2023), its intention can be offensive (Montes, 2022) by discrediting the rival, but it can also have the purpose of distorting reality in favour of the disseminator of the hoax (defensive disinformation).On the contrary, negative campaigns can be based on falsehoods as part of their discursive resource, or can simply be constructed through pejorative lexis (Aguerri; Miró-Llinares, 2023) using truthful reasoning.In any case, criticism of the rival includes all forms of attack, regardless of its level of unfairness, dishonour, irrelevance, or manipulation (Walter;Vliegenthart, 2010).
While it has been shown that the introduction of disinformative agendas can manipulate public opinion and condition citizens' voting decisions (Pira, 2019), one of the questions that has been addressed is the effect that the dissemination of a negative campaign can have on the electorate.According to García-Beaudoux and D' Adamo (2013), using the communicative strategy of attack can increase citizens' interest in the political debate.However, negative campaigning also presents risks for the political actors who adopt it (Joathan; Alves, 2020).García-Beaudoux and D'Adamo (2013) distinguish three; -"boomerang effect", as the saturation of negative information produces a rebound effect that ends up hurting the attacking actor.-"victim syndrome", in which citizens perceive an attack as unfair and generate positive feelings towards the attacked candidate.-"double damage" effect occurs when the negative campaign affects both the attacker and the attacked.
To avoid these potential risks, political actors begin to rely on the actions of anonymous groups or individuals not linked to official political coalitions (Joathan;Alves, 2020).This gives rise to the concept of unofficial negative campaigning, in which political cyber-violence plays a fundamental role (Villar-Aguilés; Pecourt-Gracia, 2020) and its development thanks to the trolling subculture of "cyber-ghettos" (Johnson; Bichard; Zhang, 2009) and pseudo-media that disseminate "poor quality" information (Cervera, 2018, p. 9).
Another aspect to consider is the behaviour of negativity emanating from platforms such as Telegram (Tirado-García, 2023) or Twitter (Lava-Santos, 2021b).Among the main results obtained, it is seen that discrediting candidates' programme proposals, ideological components or personal traits is regularly reproduced on social networks.Even in the study of negative campaigning on Facebook during the 2016 election, Marcos-García, Alonso-Muñoz and López-Merí (2021) showed that the use of negative campaigning was based on personal, ideological and previous management criticisms of the actors concerned.Based on this premise, we present the following research question.
RQ3: Which social actors are making the attacks and what is the behaviour of these critics?

Materials
This paper analyses the disinformation related to the six electoral campaigns that took place in Spain during the political cycle that began in the wake of the general election held on The composition of the research corpus went through three stages.The first was the capture phase, which consisted of using the web-scraping tool Octoparse to download the 1,042 texts published during the periods analysed on the websites: -

Disinformation agendas can manipulate public opinion and condition the citizen vote
The reason for the inclusion of Maldito Dato and Newtral fact-checks is that the aim of the research is not limited to the analysis of disinformation disseminated on social networks, but extends to that circulating in the public sphere as a whole; therefore, the materials should not be restricted to the rebuttal of hoaxes, but should also include the verification of statements made by the political class and pieces published in the media.
The second -the narrowing phase-involved selecting only the 669 publications strictly related to electoral campaigns.
To do so, we took into consideration the typology of issues developed by Mazzoleni (2010) based on Patterson, which distinguishes between political issues, policy issues, personal issues and campaign issues.
The third definition -the results of which are shown in Table 1-involved discarding publications that referred to more than one piece of misinformation, as this was a study of misinformation and not of verifiers' publications.This reduced the sample to 481 pieces of disinformation referring to 409 problematic materials, a figure that has been taken as the basis for calculating the percentages in the presentation of results.

Method
The research applies the classic methodology of content analysis (Krippendorff, 2004;Neuendorf, 2016) to the 409 pieces of misinformation that make up the sample.To this end, a codebook of thirteen variables was drawn up, as shown in Table 2, and articulated around five study dimensions: scope of dissemination, format, epistemological authority, prominence, subject matter and discourse.The first dimension comprises only the Scope of dissemination (v1) variable that, based on the distinction of interrelated spheres in political communication proposed by Mazzoleni (2010), categorises misinformation according to whether it has been generated and initially disseminated by: a) members of the political class (political system), regardless of whether they have done so in institutional, media or social network environments; b) journalists or the media (media system), regardless of whether they have done so in the media products themselves or on social networks; and c) identified or anonymous citizens (citizen-elector system), on social networks.
The second dimension studies the format of problematic materials disseminated in the public sphere through five variables.The analysis of this dimension does not apply to disinformation corresponding to the political and media sphere, as its formal characteristics -mainly statements and texts with the appearance of news-are so homogeneous that they would distort the results.
-On one hand, the dichotomous Presence of text (v2) and Presence of multimedia resources (v3) variables analyse the inclusion of these elements in the disinformation disseminated; while the Disinforming element (v4) variable, also dichotomous, is interested in which of them is the source of the deception.-On the other hand, and only applicable to disinformation distributed via multimedia resources, the Type of resource (v5) and Degree of alteration (v6) variables classify the multimedia elements according to their format and degree of manipulation, which distinguishes between existing when it has been shared without alterations, reconfigured when it has been altered, and fabricated when it has been created ad hoc.
The definition of the variables in this dimension is based on the cheapfakes-deepfakes spectrum proposed by Paris and Donovan (2019) and the complexity curve proposed by Gamir-Ríos and Tarullo (2022).
The third dimension analyses the epistemological authority of disinformation by combining two variables.
-On one hand, the Issuing Instance (v7) variable categorises the origin of the disinformation based on various categories established after an initial study of the research corpus and whose definition takes into account the typology of disinformation producers proposed by Tucker et al. (2018) in their review.-On the other hand, the Source (v8) variable analyses the attribution of disinformation based on the typology proposed by Salaverría et al. (2020): a) anonymous, when the source is not mentioned; b) fictitious, when the identity of the source is invented; c) impersonated, when the disinformation is attributed to an existing source but which was not involved in it; and d) real, when the identity is correctly attributed to an existing source, although the content is false.
The fourth dimension observes the protagonist of disinformation through the combination of two variables.
-On one hand, the Protagonist (v9) variable classifies the person or entity that is the object of disinformation based on a catalogue of options drawn up after an exploratory approach to the corpus and structured into two levels; the main one distinguishes between: a) no protagonist or citizens in general; b) institutions; c) political class, which distinguishes between non-state level parties (PANE) and state level parties (PAE); d) social agents; e) collectives; f) agents with public relevance; g) others.
-On the other hand, the Attribute (v10) variable analyses the characterisation of the objects of disinformation on a traditional three-grade scale: negative, neutral and positive.
Finally, the fifth dimension deals with the subject matter and discourse of disinformation.Based on Mazzoleni (2010), the Theme (v11) variable distinguishes between ideological (political issues), sectoral (policy issues), personal (personal issues) or electoral (campaign issues).Assuming that disinformation always presents populist features, since, as Waisbord (2020) states, it is part of its core strategy, the Populist strategy (v12) variable takes into account the populist discursive features defined by Engesser et al. (2017) for political messages and differentiates between: a) defence of the people, understood as the citizenry as a whole; b) attack on political, economic or media elites; c) vindication of national sovereignty in the face of globalisation or the interference of external agents; d) ostracism of those who are different or of minorities (such as migrants or the LGBTQI+ community) and e) appeal to patriotic values.
The composition of the corpus went through a capture stage, another delimitation stage and a final definition stage Finally, the Type of attack (v13) variable combines and adapts the classifications of García-Beaudoux and D' Adamo (2013) and Valera-Ordaz and López-García (2014), and distinguishes between: a) no attack; b) personal, including private life, competence and appearance; c) ideological, referring to non-sectoral political issues; d) managerial, linked to the sectoral results of holding public office of an executive nature; e) programmatic, related to electoral proposals; f) background and track record, not linked to management; g) association with groups considered denigrating by broad sectors of the population; and h) electoral, referring to the campaign itself, to incidents that occurred during its course or to suspicions about the election results.
The coding was carried out by one of the three authors after having conducted a test with another, independently and separately, on 12.5% of the randomly selected misinformation sample (n=52).The Krippendorff's Alpha coefficients obtained, shown in Table 2 and calculated using ReCal software (Freelon, 2013), corroborate the reliability of the results.

Analysis of the scope of dissemination
The study of the scope of dissemination of disinformation (v1), reflected in Table 3, shows that the problematic materials disproved by the four Spanish verifiers about the six electoral campaigns that constitute the object of study come mainly from the citizen system (56.5%),having been generated and disseminated on social networks or messaging applications by people with no public relevance, or anonymously.The second space with the highest concentration of misinformation is the political system (40.1%),which includes statements made by the political class, regardless of whether they have been made in electoral environments -such as rallies or debates-, institutional environments -such as parliaments or seats of government-, media environments -such as interviews-, or digital environments -such as social networks.The media space, referring to journalists and the media, both in networks and in the media products themselves, occupies a residual place in the generation and circulation of disinformation (3.4%).The correlation by the campaign is similar in the general, Galician and Basque, Catalan and Madrid elections, while in the Castilian-Leonese and Andalusian elections, disinformation promoted by the political system predominates.

Format analysis
Analysis of the format of the 231 pieces of misinformation corresponding to the citizen sphere shows, as seen in Table 4, that practically all (98.3%) are disseminated using text (v2) and that the presence of multimedia resources (v3) is also in the majority, although in a smaller proportion (69.7%).In terms of the typology of these resources (v5), graphic-textual compositions stand out, present on 24.2% of occasions; photographs, on 22.5%; and videos, on 13%.The element on which misinformation is based (v4) is, above all, text (64.9%, n=150), while misinformation through multimedia resources is a minority (35.1%, n=81).In turn, the observation of the degree of alteration (v6) of the disinformation originating in the civic space that has multimedia elements (n=161) reveals the predominance of pre-existing materials shared without alteration (51.6%), as opposed to those created ad hoc (40.4%) and those reconfigured (8.1%).The cross-analysis of variables 5 and 6, reflected in Table 5, shows that the most frequent combinations of multimedia resources and degree of alteration are pre-existing photographs (31.7%), fabricated graphic-textual compositions (26.7%), pre-existing videos (14.3%) and reconfigured compositions (7.5%), which are generally based on the incorporation of typographic elements into real photographs.

Analysis of epistemological authority
The most frequent sources of disinformation (v7) are anonymous accounts or accounts without public relevance, which generate half of the problematic material debunked by the verifying media (51.1%); the right-wing political class generates a quarter (25.4%); and the left-wing political class is responsible for 14.2%.In terms of sources (v8), the majority of misinformation comes from real sources (46%), although almost as much does not explicitly state its provenance (41.1%).Misrepresentation constructed by impersonating sources represents 11.7% of the sample, while those that invent them occupy a residual place (0.2%).
The cross-analysis of both variables, shown in Table 6, shows that the most frequent combinations of issuing body and source are disinformation generated by anonymous or non-publicly relevant accounts that do not make their source explicit (40.1%), those disseminated in a real way by the right-wing political class (24.7%), those disseminated in a real way by the left-wing political class (13.9%), and those disseminated by anonymous or non-publicly relevant accounts through the impersonation of their sources (8.3%).
e320506 Profesional de la información, 2023, v. 32, n. 5. e-ISSN: 1699-2407 10  The combined observation of both variables, shown in Table 7, shows that the most frequent protagonists and characterisations are the left-wing political class treated negatively (31.1%), the right-wing political class characterised in the same way (15.6%),migrants portrayed in a similarly adverse light (9.3%), the right-wing political class linked to positive attributes (7.3%), and the non-state political class portrayed in a detrimental light (6.4%).
All the protagonists of misinformation are characterised in a predominantly negative way, but the political class ascribed to the right or centre-right PAEs is, among the most frequent actors, the one that obtains the highest positive treatment, 30.9%, compared to 15.6% of the PANEs and 7.5% of the left or centre-left PAEs.

Thematic and discursive analysis
The most frequent thematic macro-category of electoral disinformation (v11) is that corresponding to political-sectoral issues (53.8%),followed by political-ideological issues (26.9%),those referring to the campaigns themselves (13.7%) and those linked personally to the candidates (5.6%).In turn, the most common populist discursive feature (v12) is the attack on the elites (51.6%), followed by the defence of the people (23.5%), the marginalisation of those who are different (18.8%) and the appeal to patriotic values (5.6%).No disinformation has been detected in which the predominant populist discursive feature is the vindication of national sovereignty in the face of globalisation or the action of external agents.
The cross-analysis of both variables, shown in Table 8, shows that the most frequent combinations of populist themes and discourse are, with very similar frequencies, the defence of the people on sectoral issues (18.3%), the attack on the elites on ideological issues (18.1%) and the attack on the elites on sectoral aspects (17.8%).They are followed by marginalising the different on sectoral issues (13.9%) and attacking elites on issues related to campaigns or their results (10.5%).
Attacking the elites is the most common populist discursive strategy in political issues (67.3%),campaign issues (76.8%) and personal issues (91.3%); in contrast, in policy issues, defence of the people predominates (34.1%), although there is also a significant presence of attacks on the elites (33.2%) and the ostracising of minority groups (25.9%).Finally, the presence of direct attacks on the political class (v13) is present in 46% of the disinformation analysed (n=188).Among those with such negativity, those containing ideological attacks (34.6%) and management attacks (30.3%) stand out, although those attacking through references to personal attributes (10.6%) and with the dissemination of suspicions about the course of the campaigns or election results (10.1%) are also noteworthy.As Table 9 shows, 50.5% of the attacks come from the public sphere, 45.7% from the political sphere and only 3.7% from the media.The most frequent combinations are management attacks generated from the political sphere (26.1%), ideological attacks arising from the civic (20.2%) or political (12.8%) sphere, and personal attacks circulated from the citizen system (9.6%).

Discussion and conclusions
The results of the study show that, although most of the hoaxes circulating during the electoral campaign are disseminated by citizens through social networks (RQ1), official political communication also has disinformative features.In line with Herrero-Diz, Pérez-Escolar and Plaza-Sánchez (2020), hoaxes find their natural ecosystem in the virtual environment, so, logically, more than half of the campaign disinformation verified by fact-checkers circulates on digital platforms (Herrero-Diz; Pérez-Escolar; Plaza-Sánchez, 2020).The use of persuasive and propagandistic rhetoric based on fallacies (Pérez-Curiel; Velasco-Molpeceres, 2020) has always been present in political discourse, but content analysis has confirmed that politicians intensify the use of disinformation in their statements during electoral periods.
Continuing with RQ1, the predominance of disinformative formats containing textual elements has been corroborated in all campaigns.Likewise, the presence of multimedia elements, such as photographs or videos, which have not undergone any alteration, beyond the insertion of an accompanying text including the disinformation, is common.The preponderance of falsehoods in which the disinformation resides in the text and not in the audiovisual material confirms the findings of research by López-Martín, Gómez-Calderón and Córdoba-Cabús (2023), Gamir-Ríos and Tarullo (2022), Gutiérrez-Coba, Coba-Gutiérrez and Gómez-Díaz (2020) and Salaverría et al. (2020), who point to cheapfakes as the main problem of political disinformation (Gamir-Ríos; Tarullo, 2022).Of course, there are sound and graphic-textual hoaxes that obey compositions with a relatively low degree of complexity (Gamir-Ríos; Tarullo, 2022).
In addition to the difficulty of detecting impersonated sources in the digital environment (Salaverría et al., 2020), there are also a number of anonymous individuals or individuals with no public relevance who support and disseminate the deception itself.Possibly, as it is a common citizen who is the source of the dialectical usurpation, they manage to capture the attention of other users, who in turn virtualise the same content, believing that the political actor is the author of the information.
The loss of rigour and the deficient work of the journalistic contrast of information has become a social and Image 6. Example of disinformation with the presence of anti-elitist attacks as a populist discursive feature.
The groups that most frequently lead misinformation are left-wing politicians  et al., 2020).These statements contradict what is presented in this study, as only four fake news items have been verified, which seems to indicate that these channels take extreme caution with content and ensure that the material is truthful during campaign periods.
Looking at the protagonists of the hoax and its attribute, the bulk of hoaxes about possible election rigging and hoaxes about subsidies received by immigrants is striking.Reinforcing the conclusions obtained by Gamir-Ríos, Tarullo and Ibáñez-Cuquerella (2021), producing mistrust regarding the democratic system (Almansa-Martínez; Fernández-Torres; Rodríguez-Fernández, 2020) and immigration is a communicative practice that occurs frequently in the electoral campaign, and which accentuates a populist discourse warning of a dangerous otherness that threatens the people.Furthermore, disinformation to the detriment of left-wing political actors and the benefit of the right-wing political class points to, on one hand, the fact that the main producers of disinformation are right-wing activists and politicians (Chadwick; Vaccari, 2019; Gamir-Ríos; Lava-Santos, 2022).On the other hand, the argument that disinformation in campaigning can be used to offend rivals or in self-defence is strengthened (Montes, 2022).
Focusing on RQ2, it can be observed that disinformation circulating during the campaign recurrently deals with sectoral and ideological issues, leaving campaign-related issues and personal aspects of candidates in the background.Another distinctive feature is that criticism of elites is present in all thematic categories, including personal aspects.
Likewise, when disinformation disseminators present a thematic focus on issues such as health, education, labour and immigration, they construct a populist discourse in defence of the people.In turn, a populist interest in marginalising immigrant and feminist groups is observed when the hoaxes present in the verifiers have an ideological and confrontational theme, something that fits with what has been observed in previous studies (Lava-Santos, 2021a; Pérez-Curiel; Domínguez-García, 2021).
Finally, the findings from the cross-referencing of the variable's scope of dissemination and type of attack (RQ3) reveal a common axis in the criticisms made by politicians and citizens.Both social agents misinform by attacking the ideological aspects of their respective rivals.On the contrary, the communicative strategy carried out in the political sphere ratifies the systematic use of disinformation that criticises the previous management of the adversary, while the citizens collective utters a greater personal discrediting in the codified hoaxes.These results contradict the conclusions provided by Tirado-García, (2023) and Marcos-García, Alonso-Muñoz and Casero-Ripollés (2021), who state that the negative campaigning present on the social networks of politicians usually focused on discrediting the personal characteristics of the candidates.
The limitation lies in the selection of a sample collected exclusively from the rebuttals provided publicly by the four verifiers.A large number of hoaxes spread during the campaigns analysed may have been propagated by other media, which has prevented the collection of a larger amount of disinformation that may have gone unnoticed among the fact-checkers.In this sense, the results should be considered ascribed to the study variables.One of the main limitations of disinformation research at an international level lies in the need to support learning about fact-checking techniques among academics and researchers, which, in addition to serving as a training element, would make it easier to identify hoaxes circulating in the public sphere and digital politics, within the timeframe of different campaigns.
Future research could provide a comparative study of the phenomenon in upcoming electoral campaigns.It would also be of interest to carry out a comparison between the populist discourses, themes and negativity present in the disinformation circulating in different countries, which would provide an overview of the quality of their media systems.Furthermore, an analysis of the influence of electoral disinformation on citizens during the same campaigns studied in this paper would complement the results.The attack on the elites is the most common populist discursive strategy in ideological, sectoral and personal issues Disinformation often resides in the text, demonstrating how easily a hoax can be created and disseminated

References
10 November 2019*.It does so by studying 409 pieces of disinformation debunked in 481 publications by Maldita.es,Newtral, Efe Verifica and Verificat; the four Spanish media accredited by the International Fact-Checking Network.The investigation of disinformation based on disinformation published by verifying media is a well-established methodology in Spain (Almansa-Martínez; Fernández-Torres; Rodríguez-Fernández, 2022; Gamir-Ríos; Lava-Santos, 2022; Gutiérrez-Coba; Coba-Gutiérrez; Gómez-Díaz, 2020; Herrero-Diz; Pérez-Escolar; Plaza-Sánchez, 2020; Noain-Sánchez, 2021; Peña-Ascacíbar; Bermejo-Malumbres; Zanni, 2021; Salaverría et al., 2020) and internationally (Brennen et al., 2020).In addition to the rebuttals linked to the campaign that started the cycle, the materials include those referring to the campaigns for the elections to -the Parliament of Galicia and the Basque Parliament held on 12 July 2020; -the Parliament of Catalonia on 14 February 2021; -the Assembly of Madrid on 4 May 2021; -the Parliament of Castilla y León on 13 February 2022; -the Parliament of Andalusia on 19 June 2022.The timeframe analysed in the six cases includes the official electoral campaign periods -eight days for the general election and fifteen days for the others-, the reflection and voting days, and the day after the elections.The composition of the timeframe coincides with that usually studied in research on digital political communication in electoral campaigns(Gamir-Ríos et al., 2022;Lava-Santos, 2021a).

Image 4 .
Example of disinformation by the left-wing political class.Image 5. Example of disinformation by the right-wing political class.

Table 1 .
Details of materials by campaign and verifying media

Table 2 .
Variables and categories of the content analysis Defence of the people / Attack on the elites / Claim for sovereignty / Marginalisation of the different / Appeal to patriotic values 13.Type of attack α = 0.785 No attack / Personal / Ideological / Management / Programmatic / Background and track record / Association with denigrating groups / Electoral

Table 3 .
Spheres of generation and dissemination of electoral disinformation

Table 4 .
Formats of electoral disinformation generated and/or disseminated at the citizen level

Table 5 .
Degree of alteration of the multimedia resources distributed in the citizen system according to their type

Table 6 .
Source of electoral disinformation according to the issuing body

Table 7 .
Protagonists of electoral disinformation according to their attributes

Table 8 .
Themes of electoral disinformation according to the populist discursive strategy employed

Table 9 .
Negativity of electoral disinformation according to its scope of dissemination