**1. Introduction**

Since the late 1980s, several right-wing parties have been growing in influence in Western Europe, leading to considerable interest from the media and academics. This political phenomenon is closely tied to the "third wave" of extremism in Europe (Acha 2017), as well as the so-called "populist Zeitgeist" (Mudde 2004). The growing importance of the family of far right parties has been spearheaded by the populist radical right, which has been its dominant and most successful expression (but not the only one) (Mudde 2007). While some see this phenomenon as one of the greatest threats to contemporary liberal democracies (Plattner 2010), others claim that it is given a disproportionate amount of academic and media attention, and warn against overestimating its success, since its political influence is limited (Mudde 2014).

While these parties were making significant progress in Europe1, three countries did not follow this trend: Spain, Portugal<sup>2</sup> and Ireland. Spain in particular has long been considered an outlier as it has had no significant radical right force, despite the presence of circumstances in which these parties were supposed to perform well: economic crisis, immigration, social protests, political discontent, etc., (Alonso and Rovira 2015). Many explanations were offered, from the influence of the legacy of the dictatorship (which would negate any possibility of a rise of the radical right) (González-Enríquez 2017), to the consolidated role of the center-right (Llamazares and Ramiro 2006;

<sup>1</sup> Despite significant individual victories, the true trend that characterizes the far right in Europe is the temporal and geographical variability of its performance over time. Authors such as Acha (2017), Art (2011) or Arzheimer (2009) have pointed out the bias in the literature, which focuses mainly on successful cases and ignores failures (which are more numerous and representative).

<sup>2</sup> In October 2019, the far right party CHEGA won a set in the parliament. However, the far right remains marginal in Portugal.

Llamazares 2012), not to mention the weakness and discredit of Spanish nationalism (Muñoz 2008). As such, a type of prescriptive discourse started to form around what became known as "Spanish exceptionalism", according to which Spain was refractory and immune to this political current.

More recently, VOX's sudden surge in the regional elections in Andalusia in December 2018 and its subsequent introduction to other arenas (at the European, national, regional and municipal level) has completely debunked so-called "Spanish exceptionalism". It is possible to allude to the end of "Spanish exceptionalism" because from the point of view of ideology, VOX shares common elements with other European radical right parties: ultra-nationalism, the idea of a threatened homogeneous national community, authoritarian, anti-immigrant and anti-feminist ideas, amongs<sup>t</sup> others (VOX 2018; Sánchez-Cuenca 2018; Anduiza 2018; Acha 2019a; Gould 2019)3. Thus, it is appropriate to place VOX within the current of the radical right, which accepts and operates within the framework of liberal democracy, though it opposes some of its basic principles (such as the rights of minorities) (Mudde 2007).

In a somewhat unusual electoral cycle (in which several elections have taken place in just a few months), VOX has achieved one of the milestones accepted in the literature to qualify as a success: achieving 5% of the vote in two or three consecutive elections (Art 2011) (see Table 1). In particular, the results in last national elections in November 2019 have been really notable: 15% of votes and 52 seats (5% more than its results in national elections of April 2019).


**Table 1.** VOX's results in elections (2014–2019).

Source: own compilation based on data of the Ministry for Home Affairs (2019).

The trajectory of VOX<sup>6</sup> was erratic until its first grea<sup>t</sup> electoral success in Andalusian elections in December 2018. These elections should be understood within the framework of the long hegemony of

<sup>3</sup> However, as noted by Acha (2019a), VOX has two characteristics that set it apart from dominant radical right expression in other European countries: first, the anti-immigrant component is comparatively much weaker (and in any case subject to the main pillar of ultra-nationalism and the defence of Spanish identity); secondly, it is not possible to clearly identify populist elements (people-centrism, anti-elitism or Manicheism, as noted by the ideational perspective (Hawkins and Rovira 2019).

<sup>4</sup> Elections took place in: Aragón, Principado de Asturias, Islas Baleares, Canarias, Cantabria, Castilla-La Mancha, Castilla y León, Extremadura, Comunidad de Madrid, Comunidad Foral de Navarra, Región de Murcia, La Rioja and Comunidad Valenciana. VOX did not stand for elections in Aragón, Islas Baleares, Comunidad Foral de Navarra and La Rioja.

<sup>5</sup> Elections took place in: Aragón, Principado de Asturias, Islas Baleares, Canarias, Cantabria, Castilla-La Mancha, Castilla y León, Extremadura, Comunidad de Madrid, Comunidad Foral de Navarra, Región de Murcia y La Rioja. VOX stood for elections in all Autonomous Communities.

<sup>6</sup> From the point of view of its organizational roots, VOX was officially registered as a political party in December 2013, when a small group of individuals with ties to the People's Party (PP) decided to create a new political alternative as a result of their dissatisfaction with what they perceived as the PP drifting to the center. The core of people who started VOX came from the PP's more conservative sectors. They were aligned with the former Prime Minister José María Aznar and the right-wing think-tank so-called Foundation for Analysis and Social Studies (FAES) (Casals 2014; Carmona et al. 2012).

the Spanish Socialist Workers' Party (PSOE), which had governed the Autonomous Community since 1982. The elections were called early by the president of the Community, Susana Díaz, as an attempt to make the most of the positive dynamic surrounding her party nationally (just two months earlier, Pedro Sánchez won the race to be Prime Minister through a vote of no-confidence, replacing Mariano Rajoy).

We have to consider two key facts in the national context that influenced the elections in Andalusia. On one hand, the center-periphery issue, which has historically been a key political issue in Spain (Moreno 1995; Linz and Montero 1999), alongside socio-economic ones, became even more important due to the increased territorial tension. One year before (1 October 2017), the Catalan government, led by pro-independence forces, pushed for an anti-constitutional referendum, and shortly afterwards (27 October) declared the unilateral and short independence of Catalonia (Barrio et al. 2018). As a result, the governmen<sup>t</sup> of Mariano Rajoy decided to apply article 155 of the Constitution, whereby the autonomous governmen<sup>t</sup> was shut down and the powers of the administration were transferred to the national government. At the same time, former Catalan president, Carles Puigdemont, fled from Spain, while members of his governmen<sup>t</sup> were detained to be put on trial for crimes of rebellion. On the other hand, in 2018, Spain experienced a rise in immigration, which led to increased politicization of this issue and increased media coverage (Acha 2019b). In this climate of growing tension, it is noted that VOX received more media coverage than other parties with a similar voting intention, approaching that of other consolidated parties (Olalla et al. 2019).

After the elections (see the results in Table 2), Juan Manuel Moreno Bonilla (PP) was appointed president of a PP and Cs coalition government, which was the first ever change in the governmen<sup>t</sup> in Andalusia. VOX's support was key in the investiture, through an agreemen<sup>t</sup> that met several demands of the radical right party (especially, the promise to abolish and cut down on public aid related to gender violence, LGBTQI rights and Historical Memory) (VOX 2019).


**Table 2.** Results of Andalusian elections (December 2018).

Source: own compilation based on data from Andalusian Government (2018).

Andalusia would be considered a suitable unit of analysis for studying VOX's electoral breakthrough for several reasons. Authors such as Acha (2017) or Pardos-Prado and Molins (2009) note the benefits of examining the radical right's performance at sub-national levels (regional or local). On one hand, Andalusia is the most populated Autonomous Community and the third in terms of GDP, according to the National Statistics Institute (2019a). At the same time Andalusia is one of the regions with lower GDP per capita: 19,132 € (Spanish mean is 25,854 €). Also, migration flows coming from Africa to Europe are notable in the south of Andalusia. Because of its demographic and economic importance, Andalusia is a crucial region for national governability and its internal dynamics have grea<sup>t</sup> implications. Given all this, the findings of this research might not be able to be extrapolated for the whole of Spain (but it is possible that some of these tendencies will have resonance in the national context).

Undoubtedly, VOX's arrival on the political landscape is a very significant event, not only for its potential direct impact, but also because of the way it may a ffect other political forces. At the same time, the emergence of the radical right party has important implications from the point of view of national

identity, insofar as it is an explicit expression of Spanish nationalism in the political landscape which, though present, had previously been more hidden (Sánchez-Cuenca 2018; Anduiza 2018). Due to its exclusionary nature, VOX introduces problems to democratic life, because, even though it accepts the framework of liberal democracy, it nevertheless opposes the rights of certain minorities. In this sense, it is important to examine in detail the vote for VOX, in order to rigorously understand the nature of this new phenomenon. In the following section some of the most relevant theories about radical right voting will be addressed, together with the research hypotheses for this analysis of VOX's vote in the Andalusian elections of 2018.

#### **2. The State of the Art Regarding Electoral Support for Radical Right Parties**

One of the most developed perspectives on the radical right is the view that highlights so-called demand explanations, which explore the attitudes, preferences and orientation of society and the electorate (Rydgren 2007). Great efforts have been made within this prolific line of investigation to analyze the electoral base of the radical right. Below, we will review some of the more established explanations on the matter and introduce the corresponding hypotheses on VOX.

First, one of the most widely recognized theories is the so-called "modernization losers" thesis. Originally put forward by Bell (1964), in its more up-to-date versions it is connected to different modern processes: globalization, risk society, post-Fordism, post-industrial society, etc., (Mudde 2007). This theory states that the least protected sectors (precarious or unskilled workers or the unemployed), who have seen their status and position in society lowered because of capitalist globalization, are attracted by the radical right (Rydgren 2007). This idea blends in with other theories on conflict between groups (Arzheimer 2009) and ethnic rivalry (Rydgren 2007), to the extent that it analyzes how these groups tend to blame minorities (immigrants and other differentiated ethnic groups) for socio-economic adversity and loss of status. The "modernity losers" thesis continues to dominate much of the field of study on the radical right, even though some of its faults have been pointed out: mainly, the difficulty of testing the impact of macro-structural economic factors on the performance of the far right and their inability to explain the variability of their success (Mudde 2007; Schwander and Manow 2017).

Related to this, it is to be noted that Spain—and Andalusia in particular—is one of the countries in Europe wherein the Great Recession has generated more negative consequences. Since 2008, the high levels of unemployment and social inequality have entailed a significant increase of precarious, lower socio-economic status groups (Carabaña 2016; González-Enríquez 2017). Considering the numerical increase of the so-called modernity/globalization losers, we can hypothesize that *those unemployed, with lower socio-economic status (in particular, unskilled workers) and low-incomes have a higher tendency to vote for VOX* (H1).

However, other approaches provide more nuance or minimize the impact of economic modernization and capitalist globalization, transferring the core of the explanation to cultural issues. From this point of view, it is not (or at least not merely) a material issue, but a symbolic one: thus, the radical right is a communitarian, traditionalist and authoritarian reaction to the libertarian, universalist and multi-cultural values of the New Left that arose in the mobilization cycle that started in 1968 (Bornschier 2010). In fact, Ignazi (1992) calls this process the "silent counter-revolution".

The far right is built on identity conflict and so-called ethnicist liberalism, differentialism or ethnocratic liberalism (Taguieff 1990; Griffin 2000). Its logic is based on defending and preserving different national traditions and cultures. Ultimately, it is an attempt to overcome classic racist and supremacist perspectives. Far from rejecting certain cultures and societies that are deemed inferior, it defends the equal value of all national cultural expressions. Nevertheless, differentialism is ultimately based on an exclusive and static idea that more or less openly discriminates and stands in the way of all kinds of cultural exchange. According to this theory, the far right's electoral base is characterized by having conservative, authoritarian and traditional opinions that are opposed to multiculturalism, and in particular the view that immigration is a threat to society and national identity (Rydgren 2007).

With regard to Spain, immigration from outside the EU has been quantitatively significant in certain periods of time: for example, in the 2000s Spain received half of Europe's immigration and foreign population, which rose from 637085 to 5648671 between 1998–2009, according to the National Statistics Institute (2019b). In 2006, 49% of Spaniards perceived immigration as society's biggest problem (Center for Sociological Research 2019a). Notwithstanding, this immigration has played quite a modest role in Spanish electoral behaviour and party competition. Authors such as Rinken (2016) or Ros (2018) show how during the Great Recession anti-immigrant attitudes not only remained stable, but also diminished at certain times. These findings are opposed to established theories about group competition, which predict an increase in anti-immigrant attitudes in economic hardship contexts (Allport 1954).

Nevertheless, it should be noted that the period preceding the Andalusian elections of 2018 was marked by a grea<sup>t</sup> social attention to immigration issue due the increase of the African immigration in the south Andalusia (González 2018). This does not necessarily mean the increase of total immigration to Spain or Andalusia, but rather the sudden politicization of the immigration issue. Considering this, an increase of anti-immigrant or authoritarian attitudes would have been expected regarding security, conceived of as the most important problem, since immigration used to be perceived as an issue that affects security and public order. In turn, African immigration—which is mainly Islamic—would be perceived as a threat to Catholic tradition. In this regard, Catholicism is a trait that has historically been linked to conservative ideologies in Spain (and to most of the far right family of parties) (Rodríguez 1992). At this point, our prediction about the link between the vote for VOX and Catholicism goes against some general findings, which note that those who are religious are more likely to vote for mainstream parties, instead for radical right parties (Arzheimer and Carter 2009). In short, what will be called the identity vulnerability hypothesis from now on can be stated as follows: *sectors with anti-immigration and authoritarian attitudes and Catholic religiosity are more likely to vote for VOX* (H2).

As an alternative to the two previous theories, we need to understand the impact of contextual socio-political factors (both nationally and in the region of Andalusia). It is necessary to consider the specific characteristics of Spanish electoral competition, not only influenced by the left–right axis, but also by the center–periphery cleavage (Moreno 1995; Linz and Montero 1999). The complexity of Spanish nation-building processes and the existence of several peripheral nationalist movements have led to the crucial importance of center–periphery cleavage. In recent years, tension regarding the political-territorial model has notably increased due to the mobilization of Catalan secessionist parties (the so-called "Procés"). In this regard, secessionist demands would affect those more centralist in Spain, who are also numerous among rightist-oriented citizens. Likewise, the 40 years of hegemony of the PSOE in Andalusia, together with a PSOE national government, would predict the activation of the right-wing vote against leftism. As Mudde (2007) notes, leftism is usually conceived of as an enemy in the radical right's discourse. Considering the above, we should attempt to examine to what extent ideological positions, preferences regarding the territorial model and opinions of leftist political leaders are good predictors of VOX's performance at the polls. According to this hypothesis, we expect that *ideologically conservative right-wing sectors, with centralist tendencies and a negative opinion of the president of the Regional Government of Andalusia and the Spanish Prime Minister, are more likely to vote for VOX* (H3).

In summary, the aim of this paper is to examine VOX's performance in the regional elections in Andalusia in 2018 to test two of the most likely hypotheses that explain the radical right vote: the losers of globalization thesis and the identity vulnerability thesis. In addition, we want to examine a third alternative hypothesis related to contextual socio-political factors. In short, we want to shed light on the factors that explain the sudden rise of the radical right in Spain from the point of view of electoral behaviour.

## **3. Methodology: Data and Hypotheses Operationalization**

In this section we provide information on the data and methods employed for testing hypotheses described above. In addition, we explain the operationalization of hypotheses: what variables are used and what results are expected to validate or reject the hypotheses.

In terms of the data, we used study no. 3236 of the Center for Sociological Research (2019b). This post-election survey, which is the main and best source of data on these elections, was carried out after the regional elections in Andalusia on 2 December 2018 (data collection occurred between 10 December 2018 and 3 January 2019). The survey utilizes random sampling and has N = 2913.

The method used in this study has been logistic regression7. The dependent variable in this research has been the share of the vote achieved by VOX (it should be noted that 122 respondents reported voting for VOX). To this end, the voting recall<sup>8</sup> at the regional elections in Andalusia in 2018 was recoded into a binary variable where voting for VOX was "1" and voting for another party or abstaining was "0". Regarding the explanatory variables, we used a total of 14 (see Table A1 in Appendix A), divided into four groups according to socio-economic status, identity dimension, contextual socio-political factors in Spain and, lastly, socio-demographic components. These groups correspond to the four logistic regression models carried out with a view to identifying the determinant factors in the vote for VOX at the regional elections in Andalusia.

Because of missing values of predictors and also to ensure comparability of explanatory power between models, we conducted multiple imputation to "fill in" the missing values using all the other information present in the dataset, increasing e fficiency and reducing the bias. Multiple imputation of missing values has been applied with multivariate regression modeling. Models using multiple imputation of missing values are shown in Table 3 (these results are essentially the same as using listwise deletion).

Model 1—used for testing the losers of globalization hypothesis (H1)—contains three variables: socioeconomic status, laboral situation and incomes. These variables are usually utilized in this kind of analysis (for example in Immerzeel and Pickup (2015)). We consider the losers of globalization those who have less economic capital, in other words: unskilled workers, unemployed and low-incomes people. Considering the increase of the losers of globalization in Andalusia before the aforementioned elections, we expect this group will be more likely to vote for VOX.

In turn model 2 examines H2 (identity vulnerability hypothesis) through three dimensions which are especially sensitive for the radical right: immigration, security and religion. In terms of operationalization, the first two variables show if immigration and security are considered to be the most important problem in Spain by respondents9. The other variable distinguises Catholic people from people considering themselves to be atheist, agnostic or of any other religion. In this sense, literature links voting for the radical right with anti-immigrant and authoritarian attitudes (Rydgren 2007);

<sup>7</sup> We conducted some robustness checks in the four regression models: both backwards and forwards introduction of variables. These results are coherent with the "enter method" used in this analysis and are available upon request.

<sup>8</sup> Self-reported vote is commonly used in political research, but is not exempt from biases (social desirability, for example), as noted by Brown-Iannuzzi et al. (2019).

<sup>9</sup> We recognize this is not the best way to measure anti-immigrant or authoritarian attitudes, but it is the only option provided by CIS 3236. We acknowledge this limitation. Undoubtedly, a better option is to measure the respondent's issue position using a 0–10 scale from fully in favour of restrictive policy on immigration to fully opposed to restrictive policy on immigration as does the European Election Studies, for example (see Schmitt et al. 2019).

moreover, the Spanish far right family of parties has historically been linked to Catholicism, so we predict that Catholic people will be more likely to vote for VOX (Rodríguez 1992).

Political contextual variables (preference for territorial model, ideological self-identification, conservative self-identification and evaluation of leaders) are used in the third model for testing H3 (socio-political contextual hypothesis). We decided to use the preference for the territorial model in view of the important role of center–periphery cleaveage in Spanish electoral competition. Moreover, current political and social polarization, because of secessionist demands in Catalonia must be taken into acccount. Also, it is necessary to consider the ideological dimension, so left–right scale and conservative self-identification are used. Evaluation of leftist leaders is used for exploring the possible existence of a kind of reaction against leftism.

Finally, the fourth model contains all the previous variables combined with socio-demographical variables: sex, age and studies. In this sense, previous findings have identified a certain profile of the radical right voter: male, young and with low levels of education (Arzheimer 2017).

## **4. Discussion of Results**

As detailed in the previous methodology section, we have created four logistic regression models (see Table 3) in order to verify the utility of the losers of globalization (H1), identity vulnerability (H2) and contextual socio-political factors (H3) hypotheses to explain the vote for VOX in the 2018 regional elections in Andalusia. In addition, we will introduce a fourth model with all the variables that explains these hypotheses, alongside the social and demographic control variables that tend to be introduced in these types of analyses (Arzheimer 2017).


**Table 3.** Logistic regression models of VOX's vote in Andalusian elections (2018).

Note: standard errors in parenthesis. \* *p* < 0.05, \*\* *p* < 0.01, \*\*\* *p* < 0.001. ˆ Reference category: upper class/upper middle class. Source: own elaboration based of CIS 3236.

On the one hand, the first regression model has tested hypothesis 1 (H1) according to which it would be expected that the radical right vote would be supported by the most disadvantaged social sectors who have su ffered at the hands of globalization and the Great Recession. Not only do the findings not support this hypothesis, but they also seem to sugges<sup>t</sup> the opposite. Thus, there are only significant results in the case of income: the 0.206 coe fficient sugges<sup>t</sup> that the higher the income of the electorate, the more likely they are to vote for the radical right party. At the same time, category unskilled worker has a −0.403 coe fficient (the orientation expected), while the labor situation (1 = unemployed) shows a negative coe fficient (contrary to H1). Clearly, there is empirical evidence to not accept the losers of globalization hypothesis (H1) in the case of VOX's electoral success in Andalusia. Furthermore, the results sugges<sup>t</sup> that the vote for the radical right party is accounted for to a significant extent by a comfortable socio-economic status and, at the same time, it is not attractive to the lower classes. Despite the negative impact of the Great Recession and the increase of the losers of globalization in Spain and Andalusia, these economically precarious groups do not seem to be attracted by this radical right option.

Model 2 has aimed to test hypothesis 2 (H2): *identity vulnerability hypothesis*. Two factors have significant e ffects that operate as expected: those who see immigration as the most important problem and those who identify as Catholic (compared to those who do not) are more likely to vote for VOX. In particular, immigration issues have had a strong e ffect on the VOX vote (*p* < 0.001 and coe fficient 1.283), in line with the findings of previous literature about radical right voting (Arzheimer 2017). As pointed out previously, both issues—immigration and Catholicism—are important ideological pillars of the radical right party (though compared to other similar parties, anti-immigration ideas are qualitatively and quantitatively less important to VOX) and seem to be important for the electoral base as well. On the other hand, there are no significant e ffects in relation to security being perceived as society's biggest problem.

The findings regarding the contextual factors hypothesis (H3) are presented in regression model number 3. In the case of the Spanish territorial model, the e ffect is positive (1.210) and statistically significant in the expected way, which can be interpreted as greater likelihood to vote for VOX as positions are more centralist. This is not surprising if we consider, on the one, hand, the party's emphasis on the territorial issue and its repeated appeals to eliminate the Autonomous Communities, and on the other, the polarized climate about territorial issue because of the Catalonian conflict. The same happens with ideological self-identification, which has a significant e ffect (0.284 coe fficient): those located further to the right on the ideological scale are more likely to vote for VOX. On the other hand, the opinion of voters on Pedro Sánchez—appointed Prime Minister a few months before the elections in Andalusia—is significant (−0.287 coe fficient), revealing an inverse correlation with the vote for VOX (that is, the worse the evaluation of Sánchez, the better likelihood to vote for VOX). This, together with the absence of significance of the opinion of voters on Susana Díaz (president of Andalusia), appears to sugges<sup>t</sup> that the vote for VOX should be fundamentally interpreted within a national framework.

In view of all the above, the results are especially interesting in model 4, which incorporates all of the variables from the previous three models, plus some socio-demographic control variables. This model's explanatory power is considerable, accounting for 33.6% of the vote for VOX (R2). Therefore, it ostensibly improves on the explanatory power of model 3 (R<sup>2</sup> of 23.6%) and is much better than the first two models, which border on being statistically invalid.

First, the statistical significance of incomes disappears in model 4, while surprisingly this variable has little significance in a positive way (0.730 coe fficient)10. Secondly, it should be noted that the significance of the immigration and religion variables is zero in this fourth model. Furthermore, the significance and e ffect of the question of the territorial model, ideological self-identification and

<sup>10</sup> It should be noted that the significance disappears when using listwise deletion.

opinion of Pedro Sánchez variables remain and become more important in the hypothesized direction in the case of ideological self-identification. In addition, there is a significant negative effect in VOX voter age (−0.026 coefficient) which coincides with the previous findings which have been identified as the young profile of the radical right's vote. As Arzheimer notes: "many studies demonstrate an effect of age, with younger voters being more likely to vote for the extreme right. Presumably, this age group is less firmly attached to the stablished parties, has a more intensive sense of ethnic competition, is subject to lower levels of social control and more prone to experiment with their vote" (2017, p. 386). The VOX vote appears not to be affected by the sex variable, though in other cases the literature has found a male bias in the vote for these types of parties (Givens 2017). Figure A1 in the Appendix A shows the average marginal effects (AMEs) of independent variables (CI 95%) for model 4. AMEs represent the marginal effect of each covariate on the likelihood of vote for VOX, keeping all other covariates constant.
