The Baskerville's dog suddenly started barking: voting for VOX in the 2019 Spanish general elections

The electoral success of the new populist radical right-wing party, VOX, which achieved an unprecedent electoral result in the Spanish general elections of April 2019, brought an end to Spain’s exceptional status as a country free of the radical right. This article asks: who votes for VOX? Empirically, we present the first assessment of electoral support for VOX at the national level. Relying on national post-electoral survey data, our results show that the electoral profile of VOX’s supporters differs from that of populist radical right-wing parties from the rest of Europe. Support for VOX, much like the voters of their European contemporaries, tends to be markedly higher amongst males; economic status, however, has the reverse effect than that observed elsewhere on the continent, with individuals on the higher end of the income distribution more likely to have voted for VOX in the April 2019 general elections. Importantly, we establish that national identity plays a large role in explaining support for the new radical rightwing challenger and that the effect of identity is conditioned by negative evaluations of the political situation in Spain. ARTICLE HISTORY Received 13 February 2020 Accepted 5 June 2020


Introduction
Since the latter part of the twentieth century, a number of Western democracies, particularly those in Europe, have witnessed the rise of radical right-wing parties. Whilst radical political challengers on the right remained within the political periphery during much of Europe's post-war history, the recent fourth wave (Mudde 2019) of the far right in the new millennium has brought populist and non-populist radical right-wing parties into the focus. The contemporary ascent of the radical right has been most acute in the aftermath of the Great Recession and the Eurozone crisis, which accelerated the emergence of political challengers (Hobolt and Tilley 2016;Kriesi and Pappas 2016;Morlino and Raniolo 2017).
Despite being one of the European countries to have endured some of the harshest economic and social consequences of the financial crisis, Spain, alongside Portugal, remained free of radical right, constituting the so-called Iberian exception (Alonso and Rovira Kaltwasser 2015). In effect: in contrast to the post-crisis developments in many other European countries, Spain did not witness the emergence or re-establishment of an electorally successful radical right-wing party (Lisi, Llamazares, and Tsakatika 2019). The sharp increase in the radical right across Western Europe after the Great Recession, coupled with the lack of strong populist radical right challengers in Spain and Portugal, represented something of a paradox and led some scholars to assess why the dog failed to bark beyond the Pyrenean Mountains (Alonso and Rovira Kaltwasser 2015).
Spain's exceptionalism came to an end in 2018 when the populist radical right-wing challenger, VOX, gained electoral representation in the regional parliament of Andalucía, which spurred it onto further electoral successes at the national level in Spain's general election of April 2019. This paper contributes to our understanding of the electoral bases of VOX by providing the first study of the individual-level determinants of its support at the national level. By doing so, we (1) bring the case of the new populist 1 radical right party VOX to the stream of research of like-minded parties in other European countries; (2) highlight the main differences of VOX with other contemporary radical right parties in Europe and beyond; and (3) analyse in detail the interplay of two factors, nationalism and critical evaluations of the political situation, which had been so far studied independently but not in their interaction.
Empirically, we rely on national post-election survey data to ask: who votes for VOX? Our findings show, much in line with the territorial conflict thesis posited by Turnbull-Dugarte (2019a), that national identity plays a central role in explaining VOX's electoral support. Specifically, we show that adopting a more (Spanish) nationalist as opposed to a plurinational or a more regional identity increases the probability of voting for the radical right in Spain, but that this effect is conditioned by individuals' political evaluations. Moreover, by focusing on the interaction effect between evaluations of the political situation and national sentiment, we find that the propensity to cast a vote for VOX in the April 2019 Spanish general elections increases among those with stronger national sentiments and negative political considerations, whereas national identity exhibits no effect on the probability of voting for VOX for those who hold a neutral or positive assessment of the political situation. This clarifies that VOX's support is not just an expression of Spanish sentiments but also of political resentment.

VOX and the emergence of the radical right in Spain
Before the pre-emptively called regional elections in Spain's most populace autonomous community, Andalucía, no radical right-wing party had achieved more than one per cent of the votes cast (Alonso and Rovira Kaltwasser 2015) in any Spanish election. In Andalucía, the new populist right-wing challenger party, VOX, obtained 11.1 per cent of the votes and 12 of the 109 seats in the regional parliament. 2 This entrepreneurial success was repeated during the Valencian regional elections in April 2019, where VOX took home 10.7 per cent of the vote share and 10 out of 99 seats. More recently, the party consolidated its political position as a successful radical right-wing party during the country's general election which took place in the same April. 3 Nationally, VOX obtained more than 10 per cent of the votes and 24 out of 350 seats. In fact, and after more than six months without an inaugurated Government, in the most recent Spanish general elections, which took place on November the 10th, VOX increased their vote share by 5 percentage-points, becoming the third largest party with 52 seats.
This fifteen per cent vote share places the party's success on an above average comparative footing with other radical right-wing parties that have emerged on the continent during the contemporary wave of radical right-wing party success in the last decade. The average vote share of a given radical right-wing party during the 2010-2018 period was slightly lower than eight per cent (Mudde 2019).
Initial political commentary has attempted to clarify the political discourse and ideological position of VOX (Ferreira 2019) as well as identify the party's main supporters (Turnbull-Dugarte 2019a). It is clear that VOX belongs to the radical right. Mudde (2019) makes the important distinction between parties that fall on the radical right and the extreme right. Whilst the latter are opposed to democratic regimes, the former remain supportive of democracy per se but advocate policies that would seek to curtail the liberal components that democratic regimes facilitate. VOX can be categorized as a radical right party in that it seeks to operate within Spain's representative democratic institutions. The party's ideological identification as a radical right-wing party is made self-evident in its programmatic platform. Economically, the party adopts a conventionally conservative agenda that promotes a focus on market liberalism, reduced state intervention and cutbacks to the social welfare state. The policy debates that form the heart of the party's communication strategy and those that it seeks to adopt as their main ideological heuristic fall within the libertarian vs authoritarian cleavagealso referred to as the green/alternative/libertarian vs traditional/authoritarian/nationalist (GAL-TAN) or socio-cultural axis (Bakker et al. 2015;Polk et al. 2017).
Amongst the catalogue of positions that VOX seeks to promote, some of the most salient and most radical include the advocation of the dissolution of Spain's devolved communities and the establishment of a single centralized state government; opposition to same-sex marriage and a bill that would protect the 'natural family'; and the reform of the country's abortion laws (VOX 2018). They also seek to put an end to gender violence protection laws and reinstall the former family violence laws instead. The party very much adopts a gendered view of society, which is an increasingly important issue amongst the European radical-right (Miller-Idriss and Pilkington 2018), and advocates for the protection of traditional gender norms whilst protesting against what it pens as radical left-wing feminism .
Moreover, and much in line with their European contemporaries, VOX also seeks to own the political space left vacant by the mainstream right Partido Popular (PP) and co-opts the anti-immigration rhetoric (Vampa 2020;Dennison and Mendes 2019) which has been prevalent amongst radical right-wing parties in a host of other European states (Betz 1994;Art 2011;Zhirkov 2014;Mudde 2007Mudde , 2019, particularly in response to the ongoing migrant crisis. Both Turnbull-Dugarte (2019a) and Ferreira (2019) argue that alongside the territorial conflict, nativism forms a central part of VOX's platform: it is explicitly islamophobic and advocates both xenophobic and ethnonationalist policies (Ferreira 2019, 87-88). On the immigration question, VOX rejects the disembarkation of rescued migrants in Spain, calls for the deportation of all undocumented immigrants, as well as for reforming the welfare system which, they argue, incentivises immigrants to come to Spain (VOX 2018). Critical positions against the EU certainly run deep within the party too, not unlike some of its European contemporaries (Pirro and van Kessel 2017;Pirro and Taggart 2018), although the party does not encourage in any way Spain's exit of the community polity. It adopts a soft Eurosceptic (Taggart and Szczerbiak 2002) position that is hostile to the European model, which it views as a threat to the sovereignty and autonomous decision-making power of Spain (Mudde 2019, 40-41) and blames the former establishment and governing parties (PP and PSOE) for having provided the EU with too much say over domestic affairs.
VOX's ideological position across the economic and cultural cleavages is visualized in Figure 1. Relying on expert survey data from the Global Party Survey (GPS) dataset (Norris 2020), it is clear that VOX's political offering is distinct from that of Spain's other four main parties: the conservative PP, the social-democratic Partido Socialista Obrero Español [Spanish Socialist Workers Party] (PSOE), the liberal Ciudadanos [Citizens] (Cs), and the populist radical left party Unidas Podemos [Together We Can] (UP); and seeks to fill the ideological space that is closer to the polar ends on the economic right-wing and conservative space. Note also that expert evaluations highlight the importance of populism for Spain's new populist right-wing party, consistent with the categorization of the PopuList dataset (Rooduijn et al. 2020). Not only is VOX more reliant on populist rhetoric than the party's contemporaries within the Spanish party system (it has as populist rhetoric score of 9.16 on a scale from 0 to 10 4 ), but populism is also identified as a core and salient aspect for the party.
The spatial placement of VOX within the extreme ends of the economic and cultural axis alongside the significant role populism plays in its political offering makes VOX, as Norris argues, a 'classic profile' example of a populist radical right-wing party (Norris 2020). Figure 2 compares the spatial position of VOX's ideological offering on the economic and cultural dimension, as well as that of the populism scores, to other populist radical right-wing parties from Western Europe. As illustrated, whilst VOX's ideological offering is actually more radical on both the socio-cultural and economic dimensions in comparison to the Alternative for Germany (AfD), France's National Rally, and the Austrian Freedom Party (FPÖ), its populist rhetoric and the saliency of the same is equitable to that of its populist radical right-wing peers.
The first empirical contribution to shed light on the factors that explain VOX's electoral success amongst voters comes from Turnbull-Dugarte (2019a), who argues that, in contrast to the electoral determinants of support for populist radical right-wing parties in the other European states (Rooduijn 2018), concerns over immigration are not associated with voting for VOX in Spain (but see Dennison and Mendes 2019) . Turnbull-Dugarte (2019a) claims that VOX's emergence can be attributed to the ongoing constitutional crisis in Catalonia. In October 2017, a coalition of the Catalan separatist parties in government in Catalonia called for an Independence Referendum without the authorization of the Spanish Government. This unauthorized public consultation triggered the beginning of a symphony of political events that have left a deep mark on Spanish politics, including the national government's ill-advised use of police force in Catalonia in an attempt to block people from taking part in the consultation; the application (for the first time) of Article 155 of the Spanish Constitution allowing the central government to suspend devolved government in the region; and the beginning of an ongoing judicial process in which senior members of the Catalan government are accused of rebelling against the state. 5 VOX materialized in the political storm that began in the aftermath of the Catalan crisis and focused much of its campaign efforts on mobilizing support for a tough stance on Catalan separatists. The party was also extremely critical of the PP-led government's handling of the situation and actually filed a legal suit against the Spanish premier accusing him of neglecting his oath-sworn duties as Prime Minister to defend the constitution by allowing the Catalan referendum to take place (Sangiao 2018). Importantly, Turnbull-Dugarte (2019a) attributes the electoral success of democratic Spain's first radical right-wing party not to a shift in the long-term electoral preferences of voters, but to the ability of a political entrepreneur to increase the political saliency of the Catalan question and the importance of Spanish nationalism. We should emphasize that, since 2017, VOX acted as a private prosecutor against the Catalan independence leaders in the case of the Procés (Sangiao 2018). This gave notoriety to the party within the media and fuelled VOX's appeal among former PP supporters. José María Aznar, former Spanish President and ex-leader of the PP criticized, in several public interventions, the PP's position on the Catalan conflict and placed himself in favour of the postulates defended by VOX (imprisonment of independence leaders, suppression of Catalan autonomy and illegalization of Catalan independence parties).
Whilst we have some information about the electoral profile of VOX, these pilot assessments rely on a post-electoral survey from the Andalusian elections alone and may not be reflective of the wider voter profile of VOX's voters at the national level. Of note is that the socio-demographic profile of Andalusian citizens is notably distinct from that of the wider population. The electorate of Spain's southern region tends to fall on the lower end of the county's income distribution, 6 reside in a geographical area that is dependent on the agricultural industry and which has, until December 2018, been governed at the regional level by an uninterrupted series of left-wing governments led by the PSOE with various junior coalition partners. We do not yet know what are the most relevant factors to understand why a voter, in the April 2019 general elections, opted to support VOX instead of sticking to their former vote choice, which in the previous national election of 2016 had mainly benefitted the PP and, to a lesser extent, Cs.
Hence, we seek to gain a better understanding of the recent VOX phenomenon in Spain by investigating (1) the factors that explain the probability to cast a vote for VOX in the long run; and (2) the characteristics that account for the vote for VOX in the recent April 2019 Spanish general elections. Empirically, we examine national post-electoral survey data from the Centro de Investigaciones Sociológicas (CIS) to establish the determinants of support for VOX across two different indicatorsthe self-reported probability of voting for VOX; and retrospective vote choice.
We structure the article as follows. In the next section we discuss the literature on the predictors of electoral support for radical right-wing and populist parties and formulate some initial hypotheses that build upon this literature. We then present the data and methods, providing a description on the estimation techniques applied before engaging in a discussion of the main results and outlining some brief conclusions.
Understanding support for radical right-wing parties Until 2018, some investigations have focused on understanding why, as opposed to countries such as Greece and Italy, who had been also severely hit by the Great Recession, Portugal and Spain did not experience the emergence of extreme, radical or populist forces on the right of the ideological spectrum. Alonso and Rovira Kaltwasser (2015) advance two main reasons for the anomalous absence of the radical right in Spain. For these authors, a combination of a restrictive (greatly disproportional) electoral system, the persistence of a traditional cleavage structure where the established parties generated and maintained strong ideological and partisan links with voters, and the ability of the main right-wing party, the PP, to cater to the electoral demands and ideological preferences of the most conservative Spanish constituents, acted as a barrier for the radical right's success. Additional reasons may include the division of Spanish radical right (Llamazares and Ramiro 2006) and the legacy of General Francisco Franco's Dictatorship (who ruled Spain from 1939 to 1976), which could have worked as a barrier for the success of extreme right parties (Torres 2016).
Halikiopoulou and Vasilopoulou (2018) went a step further to explain why in Greece a party such as Golden Down (XA) obtained, in the September 2015 elections, close to seven per cent of the vote share, whereas in the Iberian Peninsula radical right-wing parties remained largely unsuccessful. In Greece, the declining levels of trust in state institutions resulted in an overall crisis of democratic representation, party system collapse and the success of extreme or populist right parties, such as XA and the Independent Greeks (ANEL). In Spain, however, the economic crisis gave place to frustration and mistrust in political institutions, but these were channelled by other political entrepreneurs, a leftist and populist party, Podemos [We can], and a self-penned liberal party, Ciudadanos. Electoral support for these new parties, however, was driven by political dissatisfaction rather than voters' discontent with the economic situation (Orriols and Cordero 2016; Marcos-Marne, Plaza-Colodro, and Freyburg 2020). Crucially, however, economic crises, alone, do not necessarily facilitate the rise of the extreme right (Halikiopoulou and Vasilopoulou 2018, 26).
Political challengers' success, regardless of their ideological colour, depends on their ability to monopolize a niche issue that is not owned by any of the established parties and to occupy it credibly (van de Wardt, De Vries, and Hobolt 2014;Hobolt and Tilley 2016). In Spain, neither the economic, political, or the refugee crises gave place to the emergence of radical right-wing party. The first crisis, that of the economy, was channelled by Podemos (Bosch and Durán 2019); the second, that of politics and democracy, was channelled by both Podemos and Ciudadanos (Orriols and Cordero 2016;Rodríguez Teruel and Barrio 2016;Rama and Reynaers 2019), and the third, the refugee crisis, did not have a large impact in Spain, at least, until very recently. 7 Rather, the window of opportunity for a successful Spanish radical right-wing party was provided by a major territorial crisis (Turnbull-Dugarte 2019a), intensified in 2017 after the celebration of the unauthorised Catalan Referendum of Independence (which took place on 1st October) and the ensuing temporal suspension of Catalan autonomy (approved by the Spanish Senate on 27 October). The unsuccessful attempt of Mariano Rajoy's government to prevent the consultation undermined the credibility of the PP as a party suitable to fight for the union of Spain. Under the leadership of Santiago Abascal (a Basque politician who began his political career in the PP as an MP in the Basque regional parliament), VOX emphasized the territorial dimension in its discourse (Anduiza 2018), presenting itself as the only party able to 'defend Spain' (Abascal 2018). However, apart from the territorial/ national conflict, other factors could help to understand the vote for VOX. In order to identify these factors, we formulate some hypotheses, based on the literature on populist radical right-wing parties.

Hypotheses to explain the vote for VOX
Although it is not possible to single out a common sociodemographic profile for all radical right right-wing parties (Rooduijn 2018), certain patterns are certainly more prevalent than others. It is well established that these parties are considerably less liberal in the domain of family relations (Akkerman 2015) and defend more traditional gender roles (Mudde and Rovira Kaltwasser 2015) than their non-radical counter-parts (and a fortiore, than left-wing parties). Whether these positions curtail (Campbell and Erzeel 2018) or not (Immerzeel, Coffé, and van der Lippe 2015) support among women, and irrespectively of the role played by political socialization (Spierings and Zaslove 2017), socioeconomic positions and other factors, the fact is that a gender gap in the voting constituents of the populist radical right is recurrently found: men have a higher propensity to vote for the far right than women (Harteveld et al. 2015;Spierings and Zaslove 2015;Stockemer, Lentz, and Mayer 2018). Research also tends to disclose a metropolitan versus rural cleavage in the success of radical right-wing parties, with stronger support in rural settings (De Lange and Rooduijn 2015;Fitzgerald 2018), and education, also, tends to be negatively associated with voting for the radical right (Van Hauwaert and Van Kessel 2018). Regarding age, the evidence is less clear. It is plausible to expect that, as older citizens are less educated and have more traditionalist views than the younger generations, they form part of the so-called losers of globalizationthose who are unqualified and also identify strongly with their national community (Kriesi, Grande, and Lachat 2008, 8) and thus may be more likely to be attracted by the discourses of the radical right and vote for them. However, it is also true that younger electors have less clear political preferences and weaker party identification, and thus, on top of displaying more volatile voting patterns (Dassonneville 2013) and a higher tendency to experiment, they could have a higher propensity to vote for the radical right (Han 2016;Arzheimer 2018). In this line, the empirical link between age and support for the radical right or other closely related parties remains mixed (Werts, Scheepers, and Lubbers 2013, 194-95;Van Elsas 2017, 74).
H1 (demographic thesis): Men, those living in rural areas and with lower education levels will be more prone to cast a vote for VOX.
The losers of globalization thesis argues that the populist radical right gains votes from people who are marginalized because of changes in socioeconomic circumstances such as globalization and deindustrialization (Betz 1994) and increased automation (Im et al. 2019), which makes them romanticize the pre-globalization age that many radical right parties promise to reinstall (Steenvoorden and Harteveld 2018). Thus, feeling that traditional parties on the left and the right are no longer able to improve their economic situation (McGann and Kitschelt 2005), the losers of the new occupational structure are more inclined to listen to whomever promises to address their concerns. Although radical right parties do not have a discourse focused on economic insecurity, by highlighting the economic competition of immigrants, they can attract the vote of the losers of globalization. In a multi-level analysis across sixteen countries, Han (2016, 59) demonstrates that 'among the individual-level variables, men, the young, and the poor are more likely to support RRPs [radical right-wing parties]'. Additionally, Mols and Jetten (2017) have shown that individuals' perception that the national economy is performing poorly also tends to lead voters into the arms of populist and radical right-wing parties. We, therefore expect lower income status and negative economic evaluations to correlate positively with support for VOX: H2 (economic loser thesis): Those with lower income and who perceive the economy to be performing poorly will be more prone to cast a vote for VOX.
On the other hand, studies on both populism and the radical right have underlined how populist radical right-wing parties' anti-elite rhetoric attracts those voters less satisfied with the traditional political parties, as well as those most critical with the way in which political institutions work. In general terms, people who distrust the political establishment are especially prone to vote for populist parties (Bowler et al. 2017). Lubbers and collaborators, for example, find that 'people who are more dissatisfied with democracy are more likely to vote for extreme right-wing parties' (Lubbers, Gijsberts, and Scheepers 2002, 353). Furthermore, dissatisfaction with democracy may also capture authoritarian attitudes, which have been documented to foster voting for the radical right (Donovan 2019). As Orriols and Cordero (2016) have demonstrated to be the case for the electoral success of Podemos and Ciudadanos in the 2015 elections, 'while Podemos switchers were mainly politically disaffected left-wing voters, electoral support for Ciudadanos came from younger and ideologically moderate voters who had lower levels of political trust' (p. 469). We expect the same direction for VOX with political challenger status and an anti-establishment rhetoric likely to attract support from those who are politically dissatisfied.
H3 (political dissatisfaction thesis): Those with a higher level of political discontent and who are more dissatisfied with the way democracy works will be more prone to cast a vote for VOX.
Given the central role that the nation plays for radical right-wing parties (Llamazares 2012), we expect that individuals who attach importance to the nation and who hold a stronger Spanish national identity will be more likely to vote for a party with a nationalist or nativist agenda vis-à-vis those with more fluid or regional identities. 8 H4 (nationalist thesis): Those with a stronger identity attachment to the Spanish nation-state will be more prone to cast a vote for VOX compared to those with a more plurinational or regional identity attachment.
Finally, if Spain has remained free of a radical right-wing party until now, it is likely the case that nationalist identification alone does not explain why voters have run towards the radical right. There are reasons to believe that nationalist voters may feel attracted to the radical right when the boundaries of the national community have been contested (Pertwee 2016). We maintain that nationalist sentiment has been activated by the ongoing political crisis engendered by the territorial conflict in Catalonia. As a result, we argue that the impact of nationalist (Spain-centric) identification on support for the new party will be greatest among those who hold a negative evaluation of the political situation of Spain at the time. Since the territorial conflict over Catalonia played a salient role in the political debate among political parties at the time, negative political evaluations will capture those who are unhappy with the Catalan situation. Indeed, the primary (and shared) message amongst opposition parties to the Sánchez-led PSOE government during the period immediately before the elections were called, was a focus on defending the unity of Spain and bringing an end to what they penned the crisis of Spanish unity.
H5 (nation in crisis thesis): National identity will increase electoral support for VOX most amongst those who evaluate the domestic political situation negatively.

Data and empirical approach
We study the support for VOX in the 2019 April general elections in Spain from two complementary perspectives using data of the Spanish post-electoral survey (CIS 2019). This questionnaire was run by the Centro de Investigaciones Sociológicas (the Spanish national, public institution responsible for survey data collection at the national level; CIS). The sample consisted of 5,943 interviews (out of 6,000 designed interviews) among the Spanish voting age population with the right to vote in general elections (i.e. Spanish nationals aged 18 years or more) who were residing in Spain (including the two autonomous cities of Ceuta and Melilla in Africa). 9 We model electoral support for VOX using two distinct measures. The first perspective aims to capture the long-term support for VOX, the dependent variable being the selfreported propensity to vote for VOX, customarily referred to as the 'probability to vote' (PTV) question. As Torcal (2019a) claims, focusing on the PTV allows us to investigate the dynamics of change beyond the specifics and idiosyncratic elements of a given point in time, and may be therefore more suitable for the extrapolation of future trends than a static picture. 10 PTV is indicated via an eleven-point ordinal variable running from 0 (would never vote for VOX) to 10 (would always vote for VOX). Despite the ordinal nature of the PTV variable, we estimate an ordinary least squares (OLS) regression given the implications of the parallel odds assumption. Employing OLS using bounded data is not without its limitations, however, given that it can yield out-of-range estimations. For robustness, we apply a number of alternative estimation techniques (full output reported in the appendix). 11 Of importance is that relying on the OLS model which we report here does not condition our findings.
The second perspective captures the short-term and election-specific support for VOX in the 2019 April general elections, and the dependent variable is vote recall (1 = voted for VOX, 0 = voted for any other party); abstainers, null and blank votes are recoded as missing. Given the dichotomous nature of the data, we estimate a binary logistic regression model.
As to the independent variables, we have selected a series of indicators that capture the concepts in our five primary hypotheses by pooling on the established empirical operationalisations within the literature. In order to test H1 and H2, we include a number of measures that capture both the demographic and socioeconomic profile of voters. We include indicators of age (in years); age squared; gender (1 = female, 0 = male); income (1 = 900€ or less, 2 = from 901€ to 1,800€, and 3 = more than 1,800€); education level (1 = less than lower secondary, 2 = lower secondary, 3 = upper secondary, and 4 = university degree), and an indicator of the increasingly important urban/rural divide, size of place of residence (1 = urban dweller, 0 = rural). In addition to the objective measure of socioeconomic status, we also include a binary measure of individuals' sociotropic perception of the economic situation (1 = negative evaluation of the economy, 0 = neutral or positive economic evaluation).
Turning to H3, we incorporate two additional variables to assess the impact of political evaluations on support for VOX. These include (i) the binary indicator, political discontent (1 = political situation bad, 0 = political situation neutral/good); and (ii) dissatisfaction with democracy (10 = very dissatisfied with democracy, 0 = very satisfied). To test the nationalist hypothesis (H4), we include an indicator that measures respondents' self-reported identification with the national identity of Spain: the dichotomous variable national identification (1 = identifies mainly or exclusively as Spanish, 0 = plurinational or regional identity). A plurinational or regional identity signals that the individual identifies in equal measure, or more, with the identity of their autonomous community in comparison with that of the nation state. In order to observe whether nationalist identities exhibit an incremental effect on support for VOX when respondents adopt a negative assessment of the political situation (H5), we include an interaction term between national identification and political discontent.
Alongside the explanatory variables, we also include a battery of controls. Firstly, we control for ideological identification, ideological self-placement is an ordinal variable that measures respondent's positions on the left-right (1 = left, 10 = right) dimension. Two controls are added to measure levels of political involvement: political interest (1 = high political involvement, 0 = low); and whether individuals followed campaigns on social networks (1) or not (0).
Given the persistent influence of religion in shaping electoral outcomes in Spain, churchgoer (1 = Catholic who attends religious services once a week or more, 0 = not religious, not Catholic or less frequent attendant) is also controlled for. Finally, we cater to the potential role of political incentives having an impact on strategic voting and include a control for the district magnitude (3 = 10 or more seats, 2 = 6 to 9 seats, 3 = 5 seats or less). Summary descriptive statistics for the variables included in the model are exhibited in the appendix (Table A1). 12 Although the survey used is representative of the territorial and socio-demographic characteristics of respondents, it does not match the effective vote distribution of the 2019 April elections (for instance, in the questionnaire, 31.45 percent of respondents claim to have voted for the PSOE and 11.65 declared having abstained, whereas the effective figures were 20.36 and 28.24 respectively). Given that the questionnaire does not provide weights, we have create a weighting variable and apply weights in all the analyses to redress the effective distribution of voting behaviour. The results shown here are robust to changes in the models' specifications. 13

Analysis
We study voters' support of VOX from two different perspectives: (1) the probability to vote for VOX, ranging from never to always; and (2) the probability of having voted for VOX in the last general elections. Table 1 reports the main results of the analysis: the left-hand panel regresses the selfreported PTV for VOX whilst the right-hand panel estimates the logistic regression on vote recall for VOX. Model 1 and Model 3 provide an empirical test for H1-H4, whilst Model 2 and Model 4 provide a test for the conditional hypothesis, H5.
In terms of PTV, the results provide only limited support for H1. In line with demographic assessments (H1) of support for populist radical right-wing parties across Western Europe, women are substantially less disposed to support the party, and we observe a curvilinear relationship between age and support for VOX with middle-aged individuals exhibiting the strongest support for the party. However, in direct contrast to Robust standard errors (two-tailed) in parentheses. *** p < 0.01, ** p < 0.05, * p < 0.1 † Pseudo R-squared in case of models 3 and 4.
our hypotheses and existing support for the radical right in other states, instead of rural and low educated voters it is actually urban dwelling and higher educated voters who report a higher PTV for VOX. Interestingly, we do not provide any evidence to support the economic loser thesis (H2). There appears to be no relationship between income levels or sociotropic assessments on the state of the economy and support for the radical right in Spain. Whilst the significance of the effect is less robust (p < 0.1), there is a positive relationship between income status and support for VOX. This is somewhat at odds with the comparative findings observed across the rest of the continent and suggests that the economic constitution of the radical right's electorate in Spain is, in fact, distinct from the stereotypical economic profile of radical right supporters.
Conversely, however, the results provide substantive support for the political dissatisfaction thesis (H3). Both increases in one's level of dissatisfaction with democracy and having an overall negative assessment of the political situation currently facing Spain are associated with a significant increase in the PTV for VOX. Finally, nationalism clearly had an influential role in predicting support for VOX (H4), with identification with a Spanish identity (vis-à-vis a more plurinational or a regional identity) exhibiting a substantively important and statistically significant increase in the PTV for VOX. This result is consistent with the those presented by Sánchez-Cuenca (2019) on the bases of aggregatelevel information of the November elections, and with the findings shown by Torcal (2019aTorcal ( , 2019b using individual-level data from a private pre-election survey. Amongst the battery of controls, we observe results that largely conform with our expectations. Increased ideological identification with the right and frequent church attendance are positively associated with the PTV for VOX. Being politically interested and actively following political parties on social media are correlated with an increased PTV for the right-wing challenger. These findings are congruent with our expectations given initial evidence on the political interest of VOX's voters (Santana and Rama 2019), the highly effective social media-focused communication strategy adopted by the party (Viejo and Alonso 2018) and the superior efficacy of political entrepreneurs in Spain on social networks (Anduiza, Cristancho, and Sabucedo 2014;Turnbull-Dugarte 2019b).
But do these determinants of the PTV yield a similar effect on the probability of actually voting for the party? Model 3 reports the output from the binary logistic regression estimation predicting respondents' 2019 vote choice. On the whole, the results of Model 3 echo those presented in Model 1. Women were significantly less likely to have voted for the new party whilst those at the higher end of the income distribution are more likely to have voted for VOX. Notably, socioeconomic perceptions don't appear to have played a role in explaining vote choice for VOX in 2019, whilst the same is not true for political indicators. Both increasing dissatisfaction with democracy and negative political evaluations demonstrate a significant and incremental effect on the probability of having voted for VOX. Finally, as hypothesized and observed in the case of the PTV, nationalist identification increases the likelihood of voting for VOX.
What mechanism activates nationalist sentiment and support for VOX? Turnbull-Dugarte (2019a) maintains that the party's electoral success in Andalucía was likely triggered by the political crisis engendered by the territorial conflict brought about by debates over independence in Catalonia. The author finds that those who favour a more consolidated and centralized form of government in Spain are both substantially and significantly more likely to support for the new right-wing challenger. We hypothesize, however, that identifying with a centralized and uniquely Spanish identity is only associated with support for VOX amongst those who are discontent with the political situation. Figure 3 presents the output from the multiplicative interaction effect between nationalist sentiment and individuals' assessment of the state of the political situation facing the country (H5). The illustration suggests that the role of nationalist sentiment is conditioned by evaluations over the political situation. Indeed, identification with national identity visà-vis a more plurinational or a regional identity exhibits no effect on the probability of voting for VOX for those who hold a neutral or positive assessment of the political situation. The same is not true, however, of those who evaluate the political situation negatively. Nationalist identification, in this case, increases the probability of voting for VOX by 7.5 percentage points which is a politically substantively increase.
The conditionality of nationalist identification here is important. It suggests that identifying more with the national identity of Spain vis-à-vis the autonomous alternatives does not, in isolation, explain electoral support for VOX, but rather only those voters who share these sentiments whilst at the same time being troubled by the political situation in the country are drawn to the new political challenger. In other words, nationalism is not enough; it has to be activated by a political conflict that brings the mobilizing potential of the territorial conflict (over Catalonia) to the fore. Though, empirically, the validity of our finding is restricted to the Spanish case, it suggests interesting avenues of research for our broader understanding of support for the radical right: is the relationship between nationalist identification and voting for the radical right also moderated by the evaluation of the political situation in other European countries?

Conclusions
The electoral success of VOX as the first populist radical right party to gain parliamentary representation signals a turning point in Spain's brief democratic history since the fall of the Franco Regime. VOX emerged not only as a party that ruptures the far right-free status quo within the national parliament but has also proved itself to be politically relevant, supporting a number of right-leaning coalitions across Spain's multiple layers of governance at the regional and municipal levels.
In this article, we set out to understand the individual-level determinants of support for the new right-wing challenger. Our results contribute to fill a gap by including the case of Spain in the stream of research on the rise of populist radical right-wing parties in Europe. We document several similarities with the typical electoral profile of populist radical right support observed in other Western European democracies, mainly in terms of sex, ideology and national sentiment. The likelihood of voting for VOX is higher among middle-aged men, right-leaning voters, catholic and frequent church goers as well as those who identify most with the Spanish nation state and who hold a negative evaluation of the political situation. Yet we also uncover notable patterns of divergence in terms of geography and, to some extent, income and education. VOX's constituents are more bourgeois, with the party banking significantly more votes among urban residents, those with higher secondary education, and citizens with high income levels. Our study, however, is not without its limitations. Of significance, is that we are unable to provide a direct empirical test of the role of immigration concerns or attitudes towards EU integration in our models given the lack of concrete survey instruments that measure these attitudinal variables in the CIS post-electoral survey. Future work may seek to revisit these questions.
Finally, and of paramount importance to understanding the electoral success of VOX, is the effect of nationalism. We demonstrate that those individuals who adopt a Spanish national identity vis-à-vis those who hold a more plurinational or a more regional identity are substantially and significantly more likely to report an increased probability to vote for VOX or to report having voted for the party within the most recent general election. Importantly, we also observe that the impact of individuals' national identity is conditioned by evaluations of the political situation. Specifically, nationalist sentiment only played a significant effect on electoral support for VOX amongst those who viewed the political situation to be negative. In other words, nationalism only increases support for the new populist radical right-wing party in Spain when individuals view the political situation to be negative. Given that the main political conflict facing voters during the 2019 election was that of the ongoing Catalan crisis, there is a clear link between negative political evaluations and the territorial conflict. Albeit clear, this link is not perfect: individuals who perceive national identity as contested are likely to assess the political situation negatively, but not everyone assessing political situation negatively will do so because of a contested national identity. It is noteworthy, however, that the Catalan crisis was one of the most salient political issues at the time. Whilst our theoretical argument is consistent with the data (nationalism only exhibits and independent effect amongst those who are politically dissatisfied), the data does not provide for an explicit and direct test that this is the result of the territorial conflict. Those who are more prone to view their national identity in a centralized way are, therefore, driven to VOX when they feel the political situation is bad, possibly because they consider that the political situation places the unity of Spain at risk. In Spain, nationalism, when brought under threat, prevails.
These results also raise important questions beyond the Spanish case. In many European polities, subnational minorities may be perceived to pose a threat to national unity, both in the West (the UK, Belgium and even France may constitute good examples) and in the East (Hungary or Poland, for instance). To what extent is the support for the radical right enhanced by a combination of nationalist sentiments and critical evaluations of the (national) political situation? Notes 1. We identify VOX as populist whilst acknowledging a lack of consensus in the literature regarding the categorization of Vox. Turnbull-Dugarte (2019a) and Mudde (2019) argue that populism forms a core element of Vox's political offering; see also Zulianello (2020), but this is challenged by Ferreira (2019). Relying on a detailed discourse analysis of VOX's electoral programme, Ferreira argues that, although populist rhetoric is present, it is not a central aspect.
Expert survey data classifies VOX as both populist and radical. The PopuList dataset (Rooduijn et al. 2020) identifies populist parties as 'parties that endorse the set of ideas that society is ultimately separated into two homogeneous and antagonistic groups, 'the pure people' versus 'the corrupt elite,' and which argues that politics should be an expression of the volonté générale (general will) of the people', and the experts of this survey categorize VOX as both a radical-right-wing and populist party. A dichotomous operationalization (i.e. populist vs non-populist) is, however, also problematic (Norris 2020). For that reason, we rely on data which operationalizes populism as a matter of degree (see Figure 1). 2. In December 2018, VOX was a very young, but not a newly born party. VOX had been created in 2013 as a spin-off of the People's Party (PP). Before the 2018 Andalusian elections, it had participated in the 2014 European elections and the 2015 and 2016 general elections (alongside several regional elections) without gaining any seats. 3. For an overview see Simón (2020). 4. Populist rhetoric and populist saliency scores are measured via the following survey instruments [emphasis in the original]: (i) 'Parties can also be classified by their current use of POPULIST OR PLURALIST rhetoric. POPULIST language typically challenges the legitimacy of established political institutions and emphasizes that the will of the people should prevail. By contrast, PLURALIST rhetoric rejects these ideas, believing that elected leaders should govern, constrained by minority rights, bargaining and compromise, as well as checks and balances on executive power. Where would you place each party on the following scale?' (0 -Strongly favours pluralist rhetoric; 10 -Strongly favours populist rhetoric). (ii) 'And how IMPORTANT is populist rhetoric currently for each of the following parties?
Where would you place each party on the following scale?' (0 -No importance; 10-Great importance).
5. Ultimately, the Spanish Supreme Court ruled that nine of the twelve Catalan separatist leaders were guilty of sedition and the misuse of public funds resulting in prison sentences ranging from nine to thirteen years, depending on the individual. 6. Andalucía, after Extremadura, is the region with the second lowest gross domestic product per capita (Cordero, Fernández-Esquer, and Rama 2019). 7. Until 2014, Spain had 0.1 asylum applicants per one thousand inhabitants; in 2015 and 2016, the figure rose to 0.3; in 2017, to 0.8 and in 2018, to 1.2. In the UE, the peak was observed before, in 2015 and 2016 (2.6 and 2.5, respectively). Source: EUROSTAT.
8. In empirical terms, we measure the concept of nativism with the Linz (1973) questions of national identity. Thus, and due to the absence of questions regarding immigration feelings in the CIS survey, we use this indicator as the best proxy to nativism. We opt for including the category of 'more Spanish than from the Autonomous community' with the 'only Spanish' option under the category of 'nationalist identity' to avoid a strongly unbalanced distribution of our key independent variable. 9. Sampling was polyetapic and stratified by conglomerates, with random proportional selection of the primary units (509 municipalities of the 50 provinces and two autonomous cities) and the secondary ones (sections), whereas the tertiary, final units (individuals) had to satisfy age and gender quotas. The sample is nationally representative, and the sampling error is 1.3% for a 95.5% confidence level and P = Q. Interviews were administered face-to-face and the fieldwork took place from May 10th to May 25th 2019. This is the unique national post electoral survey available, as the last post electoral survey (November 2019) carried out by CIS is not publicly accessible. 10. PTV is also useful as it allows us to model support for VOX whilst reducing the risk of social expectation bias skewing self-reported vote choice. Voting for parties that are penned as radical and/or extreme may be deemed socially undesirable, leading to voters misreporting their real vote choice. Including models that estimate the PTV therefore serves as an additional validity test for self-reported vote choice. 11. Our results remain robust to both ordinal logistic regression models as well as fractional logit models -following the approach advocated by Papke and Wooldridge (2008) and recoding the PTV to a 0-1 scale. 12. The table reports the unweighted count of observations, mean, standard deviation, minimum and maximum. It also shows the variance inflation factors (VIFs). All the VIFs lie well below the values that would call for concern regarding potential problems of multicollinearity, implying that the variables can be simultaneously introduced into the models. 13. Additional models have also been run that include further controls (civil status, subjective social class, and the belief that democracy is not the best system) or alternative specifications of the variables employed (age groups instead of age and its square, and ordinal instead of binary specifications of political interest, sociotropic economic evaluation, and political evaluation), yielding qualitatively the same results. As mentioned before, although it would have been interesting to include indicators related to immigration, European integration, women's issues, the rights of homosexuals, or climate change, the CIS survey did not include these questions in the post-election questionnaire.