Affective polarization is the emotional distance between positive feelings toward people who sympathize with our political ideas, and the rejection of those who have different opinions (Orriols 2021). Therefore, it refers to an emotional separation that appeals to the feelings aroused in people rather than reason (Miller 2021). Academic consideration of polarization has too often been limited to aspects such as ideology or the political party system, neglecting the study of the affective or socio-emotional features of this behaviour (Shereikis 2020). However, the latter are currently seen as fundamental aspects, especially considering recent polarizing trends in which the crossover between political identity and social identity has gradually increased. The result is a hostile and fragmented electorate that huddles together in radicalised parties, or at opposite extremes of the ideological scale (Iyengar, Sood and Lelkes 2012; Mason 2013).

The problem is that excessive polarization can directly impact our social behaviour as well as our political attitudes (Shereikis 2020); this is because simply believing that we belong to a specific group, or identifying with it, necessarily involves distancing ourselves from another group (Huddy, Mason and Aarøe 2015), and adversaries are then considered the enemy. Thus, growing polarization in the electoral scenario has led to a two-fold conflict: the electorate’s rejection of voters of the opposing party; and a far more negative view of the situation when their party is not in power (Crespo Martínez, Rojo Martínez and Mora Rodríguez 2021). However, depending on the society under analysis, this statement needs to be qualified (García-Marín, Luengo and de Blasio 2021), since the impact of affective polarization will not be the same across all party systems, and multi-party contexts have only been addressed in the last decade (Gidron, Adams and Horne 2020). Such contexts involving multiple parties often include problematic factors such as party hooliganism and possibly even the confusion of affective polarization with negative partisanship (Crespo Martínez, Mora Rodríguez and Rojo Martínez 2024).

The theoretical underpinnings of this type of polarization stem from social identity theories (Iyengar et al. 2012) based on a psychosocial process in which reality is reduced to two conflicting and exclusionary ways of thinking, and the stances of groups to which one does not belong are considered harmful (Villa Gómez et al. 2020). Consequently, positive bias is prevalent toward the in-group, and negative bias is directed at the out-group (McCoy, Rahman and Somer 2018), which arouses tribalistic dynamics (Torcal 2023). This situation is nothing less than the consequence of attempting to change the collective vision by using specific practices and communication techniques in an attempt to standardize citizens’ perspectives on public matters so that their viewpoint is in line with that of the dominant elite (Silva 2004).

In examining the reasons for this rise in socio-emotional polarization, research by several authors (Iyengar et al. 2012; Levendusky 2013; Prior 2007) has shown that the media contributes to this increase in polarization. In fact, the question has been raised as to whether consumption from certain media outlets fuels this phenomenon within a political context where extreme opinions seem to garner more attention. Many theorists posit that voters who share journalists’ political leanings are increasingly attracted to ideological content that consolidates their partisan stance, thereby contributing to affective polarization (Prior 2013).

The current proliferation of new media and the growing number of platforms give people the chance to exert more control over their media diet, generally speaking, and more ways of avoiding counter-attitudinal messages (Festinger 1957; Sears and Freedman 1967; Frey 1986). Some experiments, such as those conducted by Johnson and Arceneaux (2013), have proven that media with specific ideologies polarize attitudes significantly more when survey respondents are forcibly exposed to them (Tsfati and Nir 2017). Additionally, partisan media is more frequently consumed by audiences with an affinity for either end of the ideological spectrum. Levendusky (2013) discovered that consuming like-minded media consolidates attitudes and radicalizes audiences. Even when people do consume media with opposing ideas, it does not usually alter their point of view. This is mainly due to the fact that when an audience is exposed to media of various stances, it tends to ignore the information presented and its source, yet by contrast, exposure to like-minded media strengthens listener confidence in their own values and political beliefs. This phenomenon is directly linked to affective polarization, since the ability to interact selectively with like-minded media enables consumers to self-validate, often at the expense of political figures or partisans they dislike (Shereikis 2020). This could even result in erroneous perceptions of these adversaries (Levendusky 2013; Yudkin, Hawkins and Dixon 2019).

The schism that defines Spain’s left–right and centre–periphery politics and media exposure to like-minded viewpoints, or in other words, a non-diverse media diet, tends to polarize individuals, as these factors reinforce intergroup identity (Remiro 2021). At the same time, individuals tend to consume information in line with their own political preferences (Iyengar and Hahn 2009). However, studies on the centre–periphery division indicate that this particular discord might exacerbate affective polarization, which underscores the role played by identifying with a certain group as a potential source of this type of polarization (Padró-Solanet and Balcells 2022).

Guess (2021) reached a similar conclusion, given that most US individuals restrict their media diet within the broad ideological spectrum, which increases affective polarization of audiences. He also found an overlap of nearly 65% between the media diet of Democrats and Republicans in 2015, and around 50% in 2016. Other authors have also found a significant relationship between both variables. For example, Browning and Sweetser (2020) revealed that a more diverse media diet leads to less political polarization in individuals.

In addition to the influence of ideological biases, audience segmentation or segregation can also affect the consumption of political content (Berrocal Gonzalo, Waisbord and Gómez García 2023). This division of audiences with similar traits into groups can create an “echo chamber” effect and an increase in affective polarization (Pariser 2011). It is common practice among audiences to receive their information from news sources that broadcast ideas similar to their own (Perryman 2017).

One sociodemographic variable that might have a significant impact on affective polarization to a greater or lesser extent is gender, given that women display a higher level of polarization than men. This gender gap (Gillion, Ladd and Meredith 2020; Ondercin and Lizotte 2021) could be the result of different ideological positions and/or opinions about certain issues (Mayordomo 2021). Affective polarization also seems to decrease as academic levels increase (Clavero Mira, Ortiz García and Olaz Capitán 2023). The reason is that people who have pursued higher education can more easily understand arguments for and against issues compared to individuals with lower educational levels (Abramowitz 2010), as the latter take a more emotional approach based on identifying with their own group’s position (Pérez Zafrilla 2023). Moreover, some studies on the situation in Spain suggest that greater interest in public affairs might go hand-in-hand with a rise in the consumption of channels such as La Sexta, regardless of people’s ideological position. In short, the level of interest in politics could be a determining factor in the choice of media (Valera-Ordaz 2023).

The aim of this research is to discover how diversity in an individual’s media diet, as well as the consumption of certain media, influences affective polarization. The analysis focuses specifically on the media diet of the survey respondents during the national election campaign in Spain on 23 July, 2023 (23J). Given that the objective was to analyze the media used rather than how often the information was consumed, the main Spanish television news channels and digital newspapers were examined.1 A general analysis of polarization and ideological segregation levels in the Spanish media system has also been conducted.

Considering the main objective, the research questions to be addressed are as follows:

Q1. Are Spanish television and newspaper audiences ideologically segregated and polarized? In which media is this segregation and polarization most evident?

The hypotheses to be verified are as follows:

H1a. There are significant differences in the ideological self-placement of the audiences of diverse media.

To address the first research question regarding the general analysis of media subsystems, data was taken from the Pre-Electoral Study No. 3411 on the 2023 General Elections by the Centre for Sociological Research (CIS), completed from 8 to 27 June, 2023. As this study has a large sample size (29,201 survey respondents) and includes relevant variables, it was used as a reference source to gain a more precise understanding of the diverse and fragmented media situation. Using this data, an ANOVA test was conducted to calculate the ideological segregation of the H1a audiences. The results show statistically significant differences in the ideological self-placement of audiences of different media (see Tables 2 and 4).

An additional aim was to clarify the influence of a diverse media diet. Therefore, to address the main objective of this research, data was obtained from the Third National Survey on Political Polarization in Spain, conducted by CEMOP (the Murcia Centre for Studies on Public Opinion), between 10 and 21 July, 2023.2 This survey provided further insight into which media each respondent relied on for information about the campaign. It also includes a question (the feeling thermometer), used to calculate the dependent variable (individual affective polarization), which was not possible with the CIS study.

To measure the present study’s dependent variable (individual affective polarization), the unweighted version of the formula proposed by Wagner (2021) was used to estimate the spread in the scores assigned by the feeling thermometer to the four main political parties of the Spanish system: PP, PSOE, Sumar and Vox. The elements of the Wagner formula are as follows: p is for the party, i is for the voter, and likeip is for the like-dislike score that the individual assigns to each party. The Wagner formula proposed is as follows:

Next, the following question was taken from the Third National Survey on Political Polarization in Spain by CEMOP (2023), which was used to apply the formula to the research: “In Spain, there are several political parties or coalitions representing the population’s various leanings. On a scale of 0 to 10, where 0 means having feelings of ‘antipathy and rejection’ toward this party or coalition, and 10 means having feelings of ‘sympathy and adherence’, how do you feel about the following political parties or coalitions that I am going to name?”

Finally, to calculate diversity in an individual’s media diet, Questions 1 and 2 of the aforementioned CEMOP survey were used: Firstly, “Which television channel or channels do you use to follow the election news?” Secondly, “Which newspaper or newspapers, either in paper or digital format, do you use to obtain news on the elections?” These questions were used to reveal the television channels and newspapers that the respondents claimed to rely on for obtaining news about the 23J elections. Construction of the variable followed the logical order below:

i)

Each media was classified into a space (left, right, or centre) based on their audience’s ideological position, which were assigned according to the analysis of the CIS data in Study No. 3411. To that end, a bivariate cross-tabulation was conducted using question 3aR (“Which print or digital newspaper do you read the most to follow political and electoral information?”), question 3bR (“Which television channel do you watch the most to obtain political and electoral information?”), and the individual ideological self-placement scale of Pre-Electoral Study No. 3411 of the 2023 general elections.

The coding of the dummy variables3 in the linear regression is as follows (CV denotes a control variable and IV indicates an independent variable) 4:

- Diversity of the TV diet (IV): either diverse, or this media is not used to follow news about the campaign: 0; non-diverse: 1.

The reason why the media mentioned above were chosen over others is mainly due to these channels and newspapers being further away from the centre on the ideological scale, as detailed in the results section, considering the average self-placement of the audiences. As such, these media were selected based on their status as influential actors in the Spanish media system. Media diets from other channels and newspapers with audiences closer to central positions were also taken into account in order to control this effect.

Media outlets in Spain mostly consist of private business groups. However, there are state-owned audio-visual corporations and news agencies as well, with the foremost being Radiotelevisión Española (RTVE), the main public broadcaster at a national level. The historic justification for the presence of a public station is the social value of some of the roles performed by public media in providing information, as well as educational and cultural broadcasting, yet it is obvious that these initiatives coexist with political communication, and even propaganda, by those in authority (Pérez García 2023).

The Spanish media system mainly consists of large groups. The most prominent are the following: Grupo Planeta (press: La Razón and shareholders of Atresmedia); Atresmedia (TV: Antena 3, La Sexta); Grupo Prisa (Press: El País); Mediaset (TV: Telecinco, Cuatro); Vocento (Press: ABC, La Verdad, Las Provincias, etc.); Unidad Editorial (Press: El Mundo); Grupo Godó (Press: La Vanguardia, TV: 8TV); and finally, Prensa Ibérica (Regional Press). In addition to the evolution of pre-existing media and the fact that they have joined the Internet, specific digital media have also been founded. The main ones in Spain are El Confidencial, El Español, elDiario.es , Okdiario and Libertad Digital (Pérez García 2023).

Regarding the relationship between the media and those in authority, a classic study by Mancini and Hallin (2004) describes Spain as having a “polarized pluralism” model, which is a system that includes certain common interests between political parties and the media. The main feature of this model is the relationship between the political system and the media which, being a close association, results in highly partisan media outlets.

Below are details of the Pre-Electoral Survey No. 3411, carried out by CIS, which was used to determine the political positions of the various media according to the ideological profile of the citizens who consume their content. Establishing a self-placement mean among the consumers of each media outlet has allowed us to discover the average political position of each media’s consumers, thereby enabling an analysis of the polarization of the system as a whole.

The selection criterion for television and newspapers outlets is based on media audience ratings, according to the 2023 General Media Framework Study by the Association for Media Research (AIMC 2023). In that study, the most recent of its kind, the press and television media with the largest cumulative audiences were chosen due to their status as the most prominent media in Spain. The other selection criterion is based on sample size. Consequently, the selected media are those with a larger total number (N); this ruled out those with a smaller sample size, as this could have distorted the central position measurements.

As shown in Table 1, the television channel whose viewers are furthest to the right is Antena 3 (6.08), while the consumers of Telecinco (5.47) are in the centre of the ideological scale. By contrast, the channels to the left of the scale are La Sexta (3.58) and TVE 1 (4.55). The channel from RTVE (TVE 1) is slightly to the left at the moment, while those of the Mediaset Group (Cuatro 5 and Telecinco) are mostly in the centre of the self-placement scale. The public channel studied (TVE 1) is presumably closer to the ideological position of its respective government. As a result, it could be assumed that public channel viewers change depending on the “leanings” of whoever is in power.

Analysing the standard deviation of each media outlet is also relevant, as it provides information about whether the viewers have greater or lesser ideological homogeneity. This, in turn, highlights the media that have a more multi-party orientation that can be viewed by people with more diverse opinions, compared to those that lead to more closed communities. The viewers of La Sexta (sd=2.321) and Antena 3 (sd=2.358) show less ideological diversity. By contrast, there is considerable diversity among individuals who consume Telecinco (sd=2.920) and TVE 1 (sd=2.625). It would be reasonable to assume that the channels followed by people with a more similar ideology might coincide with more neutral communication strategies, which can rarely be consumed by people who are further away from these positions, as this would risk a high level of dissonance. Consequently, they would be communication channels that are more polarized and militant, or closer to an “echo chamber” than plurality, which could result in segregated environments.

Finally, people who read the newspapers ABC (7.05) and El Mundo (6.23) are clearly to the right on the ideological scale in the Spanish media system. Conversely, readers of El País (3.78) and La Vanguardia (4.34) are more to the left, on average, and are considered benchmark outlets for this group of people.

If we look at standard deviation, the two print media with the largest readership, El Mundo (1.937) and El País (2.032), display less ideological diversity among their readers, or in other words, they generate more segregated communities. These outlets have a strong feature of political symbolism, as well as an editorial and news position that is clearly significant for the public. In contrast, readers of La Voz de Galicia (2.586) and El Correo (2.311) are more ideologically diverse.

We conclude this section with a calculation of the polarization levels of each media subsystem. To this end, we have applied the classic formula of Sani and Sartori (1980), as their calculation can measure the distance between two groups located at opposite ends of a given scale. These authors only included prominent parties to calculate systemic polarization. In our case, we have applied this strategy by focusing on nationwide media with the highest market share. The Sani and Sartori formula provides an estimate of the elasticity of space between media extremes. It can be located close to 0 (i.e. there is limited or no distance or polarization between the groups at each extreme), or close to 1 (i.e. there is maximum polarization between the groups at each extreme) (Crespo, Rojo and Mora 2021: 82). In the case of television, if we consider the two opposite extremes of the Spanish media subsystem, based on media outlets with the strongest impact (La Sexta, more to the left, and Antena 3, more to the right), we obtain a polarization level of 0.27.6 For the print media, this rises to 0.36, considering El País to be more to the left and ABC more to the right.7 The relationship between polarization and fragmentation cannot be overlooked in the case of newspapers. Diversity in the print media subsystem offering creates small business niches, which might foster reader loyalty through messages that are ideologically limited, but which are extremely emotional and satisfying for highly motivated audiences. This would not be feasible in ecosystems with a more reduced offering, or less spatial elasticity, such as television. Furthermore, the digitisation of print media has lowered barriers, which are still considerable in other ecosystems, thus making it easier for newcomers to enter the market. This has led to new, less mainstream voices competing directly with traditional broadsheets, thereby forcing the actors involved to reconsider whether centre-focused or catch-all strategies are a good idea.

We know that print media is a much more polarized media subsystem than television. However, are the media as a whole more polarized than the party system itself? Based on data from the aforementioned Pre-Electoral Survey No. 3411 of the CIS, we calculated the Sani and Sartori polarization index of the Spanish party system by using the parliamentary context after the elections of 23 July, 2023, as a reference. The aim was to compare this political polarization with the polarization of the various media subsystems. We calculated the average ideological self-placement of the multiple categories of the Vote+Sympathy variable. According to the results, the political party furthest to the left is EH Bildu (M=2.21, N=327, sd=1.723), and the party furthest to the right is Vox (M=7.59, N=2,617, sd=2.026). These parties were selected over others due to the fact that their average ideological self-placement was the closest possible to the extreme ends of the ideological scale. Furthermore, this choice was made according to the ideas of Sani and Sartori (1980), since both parties are open to coalition and blackmail, or in other words, they are either significant at a governmental level in the domain where coalitions are decided, or they offer serious competition to the opposition, or both. With regard to the party system overall, the polarization index is 0.60.8 According to these data, we can conclude that both television and print media are less polarized than the party system itself. This finding is relevant, since the perception that the most prominent, nationwide media outlets are extremely influential actors that mobilize people with regard to public opinion (Innerarity Grau 1998) is a misconception, according to our research. Although their impact is still evident in the results of this study, this does not mean that their scope exceeds that of the Spanish party system. Nevertheless, the polarizing role of the media as a whole is noticeable, and the idea that they display a reality that is substantially different from the country’s competing political parties deserves further study.

After offering a general analysis of the Spanish media system, which forms the basis of how citizens stay informed, we have analyzed the features of this consumption, which might explain the reasons why affective polarization occurs. Firstly, we have addressed the influence of a diverse media diet, not merely as a numerical question, such as the number of media outlets, but as a substantive issue as well. In other words, we have explored whether or not individuals consume a variety of media with different ideological leanings. Secondly, we have also examined the specific influence of Spanish media at the extreme ends of the subsystems (see Table 1 and Table 2: La Sexta and Antena 3; El País and ABC).

Furthermore, we developed a linear regression model that includes the news diets of television and newspapers regarding the election campaign, as well as the consumption of certain channels or media outlets that cover election-related news, controlled by sociodemographic variables and ideological extremism, which has become one of the explanations for affective polarization for most academics (Rogowski and Sutherland 2016). Several authors have confirmed this trend, highlighting that the relationship between election campaigns and affective polarization is mostly linked to ideological polarization (Hernández, Anduiza and Rico 2021 Rojo and Crespo 2023).

Therefore, the model used is as follows:

The results of the model (see Table 5) show that a diverse media diet has a significant influence on the level of affective polarization, which was expected (less diversity, more polarization). This effect appears in the media diets of both television and newspapers, even when controlling for ideological extremism (β4=.142, p<0.01; β5=.087, p<0.01). Furthermore, La Sexta is the only channel that clearly affects polarization levels, or in other words, consuming this channel is directly related to a sharp increase in the level of individual affective polarization. These results are consistent with those of Rojo, Crespo and Mora (2023), who confirmed that left-wing polarized people consumed content from La Sexta more than other channels, and the consumption of this channel, as seen in a discriminant analysis, was precisely one of the features that differentiated this group from right-wing polarized people.

Moreover, the other results of the variables introduced into the model are consistent with the findings obtained from the existing literature, which are as follows: Women have higher polarization levels. Ideological polarization is a key explanatory factor of affective polarization. People with higher educational levels tend to have lower levels of affective polarization (Clavero Mira, Ortiz García and Olaz Capitán 2023).

As mentioned above, non-diversity in television consumption is essential for explaining affective polarization of individuals. The continual intake of media that is close to one’s own position reinforces attitudes that can exacerbate people’s partisan biases. Moreover, this is even true when the influence of ideological polarization is considered. Conversely, however, non-exposure to this type of stimuli or exposure to diverse stimuli with counter-narratives can be a depolarizing mechanism. The influence of a non-diverse media diet is especially strong in television, which is still the main medium for understanding political operations in Western democracies. However, just how long this will last is an ongoing debate.

Nevertheless, regarding the calculation of ideological polarization, which is based on the ideas of Sani and Sartori (shown above), print media has a considerably higher polarization index than television. This leads us to consider factors that might cause this divergence between the two media, especially those related to the profile of their consumers. Television has been considered a persuasive tool for decades (Vázquez Toledo 2005), and despite the appearance of digital media and other new technology on the scene, television has continued to evolve, and its ability to influence has not abated (Campos Freire 2015). Unlike print media, television has one particular strength that continues to the present day – it is still the most powerful means of disseminating audio-visual messages. By enticing viewers with images, it manages to convey the idea that its broadcasts reflect reality, which is why it is so effective in shaping public opinion (Fernández Torres 2005).

While television seems more prone to generating affective polarization in audiences, the findings for newspaper consumption are similar, yet less intense. Newspapers have a smaller and more sophisticated audience, and they confirm what Lazarsfeld, Berelson and Gaudet (1948) stated about the effects of the media, generally speaking. According to these authors, the media foster the consolidation of individuals’ prior preferences when people come into contact with their content. Therefore, it is quite possible that newspaper readers tend to choose the press outlets they plan to consult in advance, which means that merely reading these publications might indicate polarization in individuals even before they consult the political news therein. This could explain why non-diverse consumption of print media leads to affective polarization in the audience. Nevertheless, this polarization is lower than that of television.