The far right in Spain: An ‘exception’ to what? Challenging conventions in the study of populism through innovative methodologies

ABSTRACT

The ‘exceptionalism’ of Spain’s lack of far-right populism has come to an end with VOX. This has spurred a debate that has converged upon another exceptionalism, that of Catalan secessionism, as the explanation of support for VOX. Challenging this consensus, this article revisits extant data to question the conclusions of past studies, framing their limitations within broader trends in populism studies that must be problematised. This problematisation is done through an innovative application of Essex discourse theory to mass-level data gathered in in-depth interviews with low-income VOX voters. The results propose a different interpretation of the success of VOX that links deprivation and values to economic policy, which had been previously discarded by studies framing VOX as exclusively ‘cultural’. Building on this analysis, this paper speaks more broadly to populism studies by proposing innovative methods, by transcending dominant conceptual schemes, and by indicating a need for alternative forms of theorisation.

KEYWORDS:

• Populism
• far right
• discourse
• VOX
• Spain

Introduction

For many years, Spain stood out as an exception to the widespread surge of far-right politics in the West. The recent and abrupt end to this exceptionality, brought by the surge in 2018 of VOX, that only a year later was the third largest party in Congress, was a most surprising event. The solution current scholarship has found has been to transform the prior exceptionalism into a new one: that of secessionism in the Catalan region, which frames VOX as a contingent nationalist reaction to a unique crisis (Rama et al., 2021; Turnbull-Dugarte, 2019; Turnbull-Dugarte et al., 2020; Vampa, 2020). But this consensus is problematic: first, because analyses of existing data have important limitations. Second, because the resulting theory does not offer an explanation of the motivations of a growing electorate.

Following Bourdieu’s (1979) classic piece on public opinion, I propose the problem with existing scholarship is it focuses too much on the snapshot instead of the movement of political opinion. Exploring movement – i.e. aggregate temporal developments as well as the meaning of people’s shifting behaviour – , as opposed to a tendency in the study of populism to operate with static accounts of mass identities (HoSang & Lowndes, 2019, pp. 156–163), is key to producing research that contributes a more rigorous and enriching perspective on the mass or ‘demand’ side of populism. In this paper I aim to do so, first, by means of a comparative consideration of the April and November 2019 general elections. Second through the application of Essex discourse theory to analyse citizen narratives. Ultimately, this dynamic approach will lead to problematising the most generalised theoretical frame in the literature, which is the economy-culture dichotomy: namely, the assumption that causal theories can pursue separately economic and cultural factors.

The main contribution of this paper is to offer an alternative explanation of support for far-right populism in Spain that challenges extant scholarship.¹ Focusing on understanding their puzzling yet critical success among deprived citizens, this paper also offers an original perspective on this core electorate of the far right and its links to economic policy. A further contribution consists in innovating the methodologies employed in populism studies more generally, and the final one is to harness the preliminary results of such an approach to problematise conceptual assumptions and theoretical tenets prevalent in the broader literature.

This paper proceeds as follows: the first two sections review the literature on the demand for far-right populist parties and on VOX, pointing to a series of shortcomings. Third, addressing support for VOX among the general population, I revisit the data employed in recent research to question the validity of the current consensus regarding the central relevance of the Catalan issue. Following up on this problematisation of previous analyses, and focusing on low-income VOX supporters, I then provide a preliminary explanation to rising support among this constituency that illuminates the statistical data better than the existing theory. I do so by analysing data generated by in-depth interviews through the lens of a discursive approach to populism that is largely disregarded in mainstream research (Katsambekis, 2022) but which I aim to show can be usefully revitalised if applied to the study of mass populism. Finally, I discuss the implications and contributions of my findings.

The demand-side of far-right populism: ‘Economy’, ‘culture’ and beyond

The study of far-right populism has shifted in the last two decades from a focus on actors and institutions to an increasing focus on the factors fuelling the demand for populist parties. Within this demand-side literature, the idea that ‘economic’ and ‘cultural’ explanations or factors can be meaningfully distinguished, and that they embody a fundamental, cross-culturally valid opposition, has structured the whole debate (Berman, 2021). Even if interactive models have recently become relevant (Gidron & Hall, 2020; Hopkin, 2020; Norris & Inglehart, 2019), these still reify the distinction between economic and cultural motivations. Overall, even if the utility of such a heuristic has been questioned (Bonikowski, 2017; Gidron & Hall, 2020), no explicit deconstruction has been carried out that may point to different analytical paths.

The shift of the scholarly gaze from parties and elites to the masses has not been accompanied by the development of more qualitative research, which is still today limited beyond the exploration of far-right activist groups. Furthermore, the most cited among the little ethnographic research that addresses ordinary people’s support for far-right populism (Cramer, 2016; Gest, 2016; Hochschild, 2016) lacks an engagement with theories of populism at all. This lack of qualitative engagement can explain that the demand-side literature should be prone to assumptions that are recurrently found to be problematic (Algan et al., 2017; Golder, 2016; Mudde & Rovira Kaltwasser, 2018), since survey data is not well-suited to determine the motivations and underlying meanings of opinions, which we should expect to be multifaceted. An example of this, which moreover suggests there is good reason to question the economy-culture dichotomy as reflective of distinct behavioural motivations, is how Hochschild’s (2016) ethnographic study of Tea Party supporters shows race in the US – supposedly ‘cultural’ – is actually made sense of through a discourse of distributive justice – supposedly ‘economic’. On the contrary, common practice in the literature – such as the conventional interpretation of anti-immigration attitudes as reflecting distinctly ‘cultural’ concerns (Algan et al., 2017) – bars such analytical paths and serves to sustain the economy-culture frame.

Even if lacking a direct engagement with the concept of populism, these sources are therefore important inspiration in motivating my approach, and provide a reference point throughout my analysis. The same applies to qualitative research in the field of European studies, which unlike populism studies experienced a significant qualitative turn in the 2000s (Duchesne et al., 2010) whose output showcases the capacity of focus-group and interview data to enrich as well as contest prevalent theories in the field. Specifically, studies focusing on ordinary citizens’ euroscepticism (e.g. Damjanovski et al., 2020; Díez Medrano, 2003; Duchesne et al., 2010; Meinhof, 2004; Van Ingelgom, 2014) provide another valuable reference point throughout this paper. These issues are linked to the fact that scholarship focusing on mass behaviour has been dominated by the ideational approach, which can be traced back to Mudde’s (2004, p. 544) seminal conceptualisation of populism as a ‘thin-centred ideology’. Throughout the years, the ideational school has developed into a strongly deductive programme (Dean & Maiguashca, 2020) that is exclusively quantitative and oriented to increasingly ambitious cross-cultural generalisation (Hawkins & Rovira Kaltwasser, 2019). This has contrasted strongly with the parallel development of post-structuralist scholarship within the Essex tradition, which understands populism not as an object (contra ideational scholarship) but as a discursive logic, whereby an accumulation of social demands unattended by the political system are harnessed to construct an overarching ‘people’ identity that demands representation. Accordingly, this approach advocates interpretive methods and is concerned with mapping how discourses structured around contested signifiers (the nation, the people, etc.) come to integrate discontent into anti-systemic political identities (Laclau, 2005, 2005/2007, pp. 83–100, 118–120).

Despite significant development, Essex scholarship has consistently focused on the analysis of elite discourses and has shied away from addressing the ‘demand’ side of populism.² Here I aim to show Essex theory can also offer a toolkit for the analysis of mass opinion that can serve as an ideal basis for a more dynamic, interpretive analysis – an analysis that transcends the economy-culture frame. To do so, I bring to the fore a case study which is intriguing because it fails to fit either the economic or the cultural explanatory schemes, as I will detail in the following section.

Revisiting the case of VOX: A review of extant literature in the light of broader scholarship

The general election of 10 November 2019 was a tight race for the reedition of a left-wing government in Spain. But what turned out to be the biggest surprise of the night was something else: the great success of VOX, who contest many of the consensuses constitutive of Spanish political life, came as a shock to many amid an escalating secessionist struggle in Catalonia which was already straining the political climate.

Precisely because this crisis was marking the political agenda at the time, it has been logical and common to propose the hypothesis that VOX voters were essentially reacting to contingent events. This position is reflected in the next-day press – both national (e.g. Cervilla, 2019; González, 2019) and international (e.g. Dombey, 2019; Minder, 2019) – , that attributed VOX’s success exclusively to the salience of Catalan secessionism, which would be mobilising people fearful of the unity of Spain being compromised. It also resonates with current scholarship on VOX: in the most extensive analysis to date, the key scholars in the debate have coalesced to author a book that reinforces the already building consensus that ‘[t]he main driver behind the system-rupturing rise of VOX was the secessionist process initiated in Catalonia in 2012’ (Rama et al., 2021, p. 141; see also Turnbull-Dugarte, 2019; Turnbull-Dugarte et al., 2020; Vampa, 2020). This thesis is further specified by noting that this ‘internal’ threat crucially combined with the ‘external’ one of immigration (2021, p. 93). The aim was to close the debate as regards the explanation for VOX’s rise by agreeing on a ‘cultural’ diagnosis whose clarity and comprehensiveness have received much praise by specialists including Mudde (Pimenta, 2023; Routledge, 2021). In what follows, I will therefore fundamentally engage with this book since it constitutes the most authoritative and elaborate version of the argument I contest, based on what is still the most updated electoral data.

This hypothesis is also appealing since it implies the ‘comforting’ idea that there could not possibly be 3.6 million Spanish citizens supporting VOX’s ideas in areas such as gender or taxation. That the rise of VOX relies on a reactive, protest vote rather than full endorsement: that is, were the Catalan conflict not to fuel at a certain time a demand for nationalist reaffirmation, such a party would not be electorally relevant because of its other policy proposals. The protest-vote hypothesis is reasonable because VOX is the party that most aggressively opposes both secessionism and regional decentralisation (Turnbull-Dugarte, 2019), and is in line with studies managing data from the previous general election (April 2019) that only found a significant correlation between voting for VOX and opposing regional autonomy (Turnbull-Dugarte, 2019; Vampa, 2020). However, and despite previous scholarship making a valuable contribution to our understanding of the phenomenon at hand, I propose a closer look at the data reveals key differences between the April and November general elections. It is evident that the Catalan conflict played a key role in placing VOX in the media (Mendes & Dennison, 2021), and that their electoral breakthrough can largely be explained as a protest-vote phenomenon (Turnbull-Dugarte, 2019). However, it is also true that support for VOX has consolidated in time on an electoral base whose characteristics indicate there is more to it than merely conjunctural, anti-Catalan identity politics.

First, one factor is key to understand VOX’s rise from a 10% to a 15% of the vote between April and November 2019: this difference is disproportionately grounded on a low-income electorate. By combining novel income data on electoral sections (INE, 2019) – generally not larger than 1,000 voters – with the voting record for those same sections (Ministerio del Interior, 2019), it is possible to map with great precision the relation between income and voter turnout. This new data shows VOX is the party that on average most percentage-points gained in each percentile of the sections below the 50th percentile, and both in absolute share and percentage-point-variation registers an inverse correlation between level of income and level of vote share, with a peak among the wealthier 10% (Sánchez, 2019). But this dynamism is obscured in Rama et al.’s analysis of the last election since it registers a positive association between income and vote, which is only applicable to an income group that can be equated to that last decile just mentioned (since in their analysis this category groups only 13.2% of the sample). As opposed to the perspective that VOX is ‘a party of the bourgeoisie’ (Rama et al., 2021, p. 81), the data shows in November 2019 VOX managed to become most successful among those most deprived by the consequences of the Eurozone crisis (Romanos, 2017): not just low-income voters but also young people (CIS, 2019a).

So what is it that has pushed disproportionately these constituencies to support VOX? The percentage of low-income vote seized – 61.2% of the party’s electorate is below median income, 36.5% identifies as low-middle class or below (CIS, 2019a) – is not an anomaly given the importance of this constituency among far-right parties generally (Afonso & Rennwald, 2018), but it is striking given they postulate a radically neoliberal economic programme (Ferreira, 2019). This programme not only stands out from the broader far-right family but also seems antithetical to the interests of these constituencies in a Southern debtor country (see Hopkin, 2020), who are just as predominantly concerned about economic issues as the broader sample – 48.2% and 42.6%, respectively, pointed to economic issues as the main problem in Spain (CIS, 2019a). All of which makes untenable an application of the conventional ‘economic’ explanatory theory that speaks of the losers of globalisation seeking economic sovereignty and chauvinist welfare.

Second, the relevance of the Catalan issue is measured by all studies using a binary survey item registering whether it had an influence at all on one’s electoral behaviour. This item gives relatively poor information since it does not force people to discriminate among preferences nor clarifies its electoral significance. However, more interesting items are available where the impact of the Catalan issue appears to be much smaller: on the one hand, Rama et al. (2021, p. 101) already note that VOX voters are surprisingly not more likely to point to the Catalan conflict as one of the three main problems in Spain. On the other, an item exists that singles out whether said influence determined or changed one’s vote choice (it is embedded within the influence item as a subquestion that actually taps into the electoral relevance of the issue). An item which is surprisingly ignored by the literature, yet registers a conceptually distinct type of response that ought not to be conflated with the broader notion of influence. Its consideration indicates the Catalan issue lost relevance significantly between the April and November elections – dropping from 47.4% to 27.4% of VOX voters signalling it as determining or changing their electoral choice (CIS, 2019a, 2019b). These figures indicate that even if the Catalan conflict certainly is key to understanding the possibility of VOX rising at a specific time, it is other issues that are actually motivating most of its voters to support them as they still do, and at electoral contests where the salience of the Catalan issue has been much lower (for instance, in the two regional elections held in 2022 VOX gathered 17.6% and 13.5% of the vote).

Third, these studies reproduce the more general problem noted earlier regarding the conventions that pervade the literature as to the validity of the economy-culture dichotomy. On the one hand, these studies have failed to integrate a nuanced understanding of socioeconomic deprivation. In the case of Rama et al. (2021, pp. 79–81, 84–87), an ‘economic’ explanation is discarded by considering only personal income and support for state intervention. However, populism studies have recently integrated from psychology that what matters is not objective measures of deprivation (e.g., income) but how individuals perceive their social position subjectively. The inclusion of measures of positional deprivation or subjective social status, which is defined as ‘the belief that the resources available in society are not distributed in a fair way’ (Bos et al., 2020, p. 7), has inspired a whole new strand of quantitative research that shows this factor to be a successful predictor at the individual level (Bos et al., 2020; Burgoon et al., 2019; Gidron & Hall, 2020; Kurer, 2020).

On the other hand, a further problem is these studies assume an unambiguous relation exists between the personal relevance of certain material conditions (declining income, destabilised life trajectories, etc.) and the way these are politicised. The ‘economic’ explanatory scheme assumes what attracts low-income voters disproportionately to populism is a nostalgia of post-war welfare regimes (e.g. Hopkin, 2020), and accordingly a lack of support for redistribution is taken here as proof that it is not ‘the economy’ (Rama et al., 2021, pp. 84–87). In parallel, the relevance of immigration is taken to point unequivocally to ‘the cultural’ (e.g. Inglehart & Norris, 2016) and so do Rama et al. (2021, p. 87), even if immigration survey items are impacted by a variety of issues that may also relate to material deprivation (Gidron & Hall, 2020). In the case of Rama et al., these assumptions are problematic even within the context of the data they harness: VOX is categorised as a cultural phenomenon, but the conventional ‘cultural’ thesis is not readily applicable given their young electorate, as well as the irrelevance of ethnic difference regarding immigrant attitudes (Rama et al., 2021, p. 99), which contradict key tenets within this explanatory scheme (see Norris & Inglehart, 2019). Moreover, national identification, which should be understood as the mirror of ethnic or cultural anxieties, is not statistically significant (Rama et al., 2021, p. 101).

Building on these considerations, this paper asks what explains support for VOX in the November 2019 general election. This is a research question that is explored in two interrelated steps: first, through statistical analysis concerning VOX’s full voter base. Second, through interpretive analysis of in-depth interview data, focusing on a specific subsample of that voter base (low-income voters) which bears greatest substantive and theoretical interest: because of its quantitative relevance in building VOX’s electoral success, and because of its interest as a research puzzle that problematises dominant conceptual frames, thus presenting an ideal base for an innovative approach. In turn, a more constrained target population seemed ideal for testing innovation in interpretive methods. In line with the puzzle presented, the research question in this second part is addressed by analysing how low-income voters account for their electoral behaviour and, at a more specific level, how support for radical neoliberal policy is made sense of in relation to personal experiences of deprivation.

Methods and data

This paper combines statistical and interpretive methods to address the gaps identified in extant research. Following a logic of complementarity rather than confirmation (see Small, 2011), I employ mixed-methods as an ecumenic approach that seeks to reach a better understanding of a single causal relation by combining the insights provided by each epistemological viewpoint. First, I replicate Rama et al.’s central model (2021, Chapter 4) to reexamine, in its own positivist terms, the established consensus that outrage at secessionism is the key driver of support for VOX. In doing so, I contest the statistical validity of their conclusions and question the idea that the debate might be ‘closed’. In turn, I link the problems in their argument to the conceptual assumptions held by scholars in the field more broadly, thus indicating the need to address the issue by means of a different approach. Following up on this necessity, I subsequently present the findings derived from interview data generated with low-income VOX voters in the months following the last general election (January to July 2020), selected on the account that their income was under the 33rd percentile. Original in-depth data, followed by an analytical strategy inspired in Essex theory, was necessary to address the research puzzle in a manner that transcended the conceptual and methodological issues identified in the literature. Therefore, whereas statistical analysis serves to indicate the need to revisit the case of VOX and to question established concepts and modes of research, interpretive analysis manages to reach into the dynamism of personal biographies and political opinions to offer an alternative, original answer to the research question (with a focus on the critical constituency of low-income voters). The two steps are not aimed at separately ‘testing’ or ‘triangulating’ the same social mechanisms, rather they are meant to combine to produce a convincing argument regarding the causes of a given phenomenon.

For regression analysis, I used data from the national post-electoral survey (CIS, 2019a) to replicate Rama et al.’s binary logistic regression on voting for VOX as the dependent variable, using the same variables and operationalisation except for that of income.³ Subsequently, I made two additions: first, the relevance of the Catalan issue in determining electoral choices was operationalised by transforming the binary indicator used by previous studies into a categorical conjunction of the two existing items I described before, so that the previous no influence (0) vs. influence (1) binary could be more granular: no influence (0) vs. influence in x way (1-5). Second, variables regarding economic concerns were introduced to explore this dimension beyond income: importance of economic problems in Spain, importance of economic problems personally, and subjective economic status. The first one captures the effect of selecting an economic issue (‘unemployment', ‘pensions', etc.) as the first in the list of the three issues that respondents are asked to name in an open-ended question regarding the main problems in Spain. The second one measures the same when the question is framed in terms of the problems that most affect one personally. The third is an imperfect approximation to the concept of subjective social status: whereas the question ‘How would you qualify your current economic situation?’ does not capture the broader and relational character of an ideal survey item (e.g. Gidron & Hall, 2020, p. 1035), it still privileges the subjective perceptions of people as opposed to objective measures.

As regards the discursive analysis of voter demands and narratives, this section is based on interview data with low-income VOX voters (N = 10). The interviews comprised two sections: a first one dedicated to discussing respondents’ life situations, professional histories and experiences of economic deprivation; and a second one dedicated to discussing their electoral behaviour and their opinions on Vox’s economic proposals. It was designed this way so that I could probe respondents’ opinions in the second section with references to opinions or concerns expressed in the first one, especially when incongruences were detected, so as to clarify the relation between their biographies and their politics. Provided rapport is not compromised, exerting this kind of pressure on respondents is especially useful in trying to uncover the assumptions on which their reasoning is grounded (Brinkmann & Kvale, 2015), and can also help in countering the bias of interview analyses to offer an excessively coherent picture of respondents (Lamont & Swidler, 2014).

Interviews lasted between 55 and 120 min. The fieldwork was conducted in two phases: a first one face-to-face in Barcelona in January 2020 (after which I adapted my interview strategy and made some minor modifications to the topic guide) and a second one remotely during the period of April-July 2020. All respondents lived in the metro areas of Barcelona (7) and Madrid (3). They were recruited via informal contacts (5), online forums (1), and snowballing (4),⁴ and all provided signed consent to participation which was approved previously by the Research Ethics Committee at my institution. While the initial objective was to reach a larger sample, sudden disruption caused by the pandemic made it harder to recruit within the timespan allocated to this project. According to feedback from declining respondents, the general difficulty of reaching far-right voters was worsened by the focus of interviews on experiences of deprivation, which was perceived as a sensitive topic. However, there is obviously no ‘ideal’ sample size, since the objective of this part of the research is to ascertain interpretively the connections between a range of opinions and political behaviour to a degree of saturation.

I considered the Essex approach to be suitable as a methodological lens for three reasons. First, in providing an exclusively formal template with which to map the construction of political identities – as a logic, not an object – , it is ideal to inductively inform the data that constitutes the research puzzle – including the survey data revisited – without conceptual impositions such as those inherent in mainstream populism approaches (Katsambekis, 2022).

This template is oriented to mapping dynamics of articulation: that is, how people’s political demands come to resonate, and eventually fuse discursively, with populist discourses which are equivalential chains of resignified demands that have in common their exclusion from a nominally democratic system – thus the claim to constitute ‘the people’. The glue for such an empty signifier as ‘people’, theorists argue, can only be one among those demands, which performs hegemonically as representative – and transformative – of the rest. Articulation and the construction of the populist frontier of antagonism are therefore concomitant processes where We-Them concepts are contingently solidified (Laclau, 2005, pp. 92–99, 102, 122). The opposite of articulation is dislocation, which is a conceptual way of grasping the subjective experience of crisis: a situation in which events destabilise people’s identities/discourses and force them to question them and to search for new ones which redress dislocation (Laclau, 1990, pp. 39–41) – potentially leading to populist articulation.⁵

Building on this exposition of the basic toolkit, a second reason why I argue it is suited to the research problem is that in shifting the unit of analysis from variables or groups to demands, it provides a perfect basis to explore the motivations of voters without a priori categorising them: demands are understood not as static givens but as dynamic discursive units, which can be transformed by the articulation of their particularity (e.g. racial animosity) with different political universalities (e.g., the people against federal intervention, echoing Hochschild's (2016) example). This epistemological openness goes beyond the aforementioned ‘interactive’ models to ground a true transcendence of the economy-culture dichotomy.

Third, since a relevant question is to understand the connection between trajectories of deprivation and changing voter preferences (the Bourdieusian movement), a theoretical scheme attuned to the analysis of dynamism is welcome in the light of the aforementioned tendency of research on populism to produce static representations of populist voters’ identities. Overall, these characteristics situate the Essex toolkit at the juncture of the need to evolve beyond conceptual shortcomings in the broader field and the analytical demands of a puzzling case study.⁶

A reappraisal of statistical evidence: Re-opening the debate

Following the procedure outlined in the previous section, I conducted a series of binary logistic models which are presented in Table 1. Model 1 is a replication of Rama et al.’s model, including their binary operationalisation of the influence of the Catalan issue. It regresses voting for VOX on national sentiment, church attendance, ideology, political interest, and the existence of concern regarding Catalan secessionism and immigration, besides standard sociodemographic variables (gender, age, education, income, class, type of community, and regional fixed effects). As described above, Model 2 substitutes the binary item on influence of the Catalan issue for a categorical operationalisation of the full variety of responses in the two embedded questions, so as to more precisely measure its relevance. Model 3 introduces the economic variables described above, and Model 4 combines all variables. Finally, I combine all statistically significant variables, besides control variables (Model 5), which allows for a much larger sample to be included.⁷ Whereas AIC and BIC tests reported in Table 2 show the expanded model (4) better fits the data than the original model (1), a Wald test simultaneous on all the variables dropped from Model 4 to Model 5 due to their low significance indicates their inclusion would not provide a statistically significant improvement of model fit (p-value = 0.8944). Despite being statistically non-significant, subjective economic position was kept because the test indicated it made a significant difference.

Table 1. Regression models on the total sample.

	Model 1	Model 2	Model 3	Model 4	Model 5
Gender	0.657	0.614*	0.675	0.631*	0.616*
	(−1.88)	(−2.10)	(−1.74)	(−1.97)	(−2.28)
Age	0.627***	0.642***	0.617***	0.627***	0.642***
	(−6.22)	(−5.97)	(−6.17)	(−5.99)	(−6.51)
Education	0.319***	0.326***	0.306***	0.307***	0.228***
	(−3.45)	(−3.40)	(−3.52)	(−3.53)	(−4.71)
Income group 2	1.269	1.274	1.334	1.345	1.493
	(0.86)	(0.86)	(1.01)	(1.01)	(1.45)
Income group 3	2.035*	1.947*	2.210*	2.120*	2.492**
	(2.17)	(1.98)	(2.34)	(2.15)	(2.86)
Subjective class 1 (mid)	2.529	2.570	2.647	2.659	1.328
	(1.67)	(1.63)	(1.74)	(1.67)	(0.56)
Subjective class 2 (mid-low)	3.534*	3.391*	3.408*	3.196	1.836
	(2.10)	(1.97)	(2.03)	(1.87)	(1.11)
Subjective class 3 (working-class)	4.745*	4.670*	4.667*	4.437*	2.751
	(2.54)	(2.37)	(2.52)	(2.29)	(1.76)
Subjective class 4 (poor)	5.435*	4.841	4.801	4.251	1.548
	(1.98)	(1.87)	(1.79)	(1.70)	(0.46)
Subjective class 5 (other/don’t know)	3.410	3.269	3.694	3.420	1.820
	(1.34)	(1.25)	(1.43)	(1.30)	(0.73)
Rurality	0.867	0.887	0.858	0.879	0.877
	(−0.67)	(−0.55)	(−0.71)	(−0.58)	(−0.64)
Church attendance	0.961	0.951	0.981	0.975
	(−0.47)	(−0.59)	(−0.22)	(−0.30)
Exclusionary national sentiment	1.045	0.928	0.998	0.893
	(0.17)	(−0.28)	(−0.01)	(−0.40)
Left-right self-placement	2.010***	2.082***	2.022***	2.095***	2.210***
	(11.26)	(11.14)	(10.97)	(10.97)	(14.05)
Political interest level 2	0.849	0.810	0.844	0.775	0.844
	(−0.43)	(−0.54)	(−0.43)	(−0.63)	(−0.46)
Political interest level 3	0.515	0.452*	0.505	0.424*	0.488
	(−1.74)	(−2.01)	(−1.71)	(−2.07)	(−1.89)
Political interest level 4	0.508	0.479	0.497	0.450	0.471
	(−1.61)	(−1.70)	(−1.60)	(−1.78)	(−1.80)
Immigration as one of the main problems	2.920***	2.914***	2.812***	2.794***	2.924***
	(4.35)	(4.22)	(4.11)	(3.97)	(4.56)
Influence of the Catalan issue	2.310***		2.215***
	(3.85)		(3.58)
Catalan issue as one of the main problems	1.422	1.317	1.350	1.279
	(1.31)	(0.96)	(1.09)	(0.85)
Relevance of the Catalan issue (Pro-independence)		0.505		0.474	1.238
		(−0.77)		(−0.87)	(0.22)
Relevance of the Catalan issue (Pro-dialogue)		1		1	1
		(.)		(.)	(.)
Relevance of the Catalan issue (Reinforce)		1.618		1.507	1.547
		(1.78)		(1.48)	(1.76)
Relevance of the Catalan issue (Vote)		1.039		0.884	0.809
		(0.05)		(−0.16)	(−0.31)
Electoral relevance of the Catalan issue (Change)		5.342***		5.362***	6.487***
		(5.65)		(5.67)	(7.04)
Subjective economic position			1.609	1.683	1.713*
			(1.64)	(1.82)	(2.07)
Economic issues as the main problem			0.668	0.630*	0.724
			(−1.83)	(−2.05)	(−1.60)
Economic issues as the main personal problem			1.065	1.125
			(0.28)	(0.51)
N	1705	1671	1661	1627	2386

Note. Regional fixed effects not included. Exponentiated coefficients; z statistics in parentheses.

* p < 0.05, ** p < 0.01, *** p < 0.001

Table 2. AIC and BIC tests.

Model	N	ll (null)	ll (model)	df	AIC	BIC
Model 1 (Rama et al., 2021)	1,705	−580.4763	−385.0876	38	846.1753	1052.945
Model 4 (full model)	1,627	−566.8635	−358.6581	44	805.3163	1042.674

In line with Rama et al., focusing on the final model, some control variables prove significant: age (0.64, p < 0.001) and educational level (0.23, p < 0.001) bear a negative effect. A higher income is statistically significant, yet only distinguishing those belonging to the higher third of the income table (2.49, p < 0.01). Turning to the political variables, both expressing concerns for immigration (2.92, p < 0.001) and placing oneself further to the right ideologically (2.21, p < 0.001) bear a significant and substantive positive association to voting for VOX.

Introducing a more nuanced measurement of the actual relevance of the Catalan issue produces an interesting result: as seen in Models 2, 4 and 5, it is not answering it had an influence that is statistically associated to voting for VOX, but more specifically answering it had an influence and that furthermore this determined or changed the respondent’s party choice.⁸ In other words, comparing across the available measures of the effect of the Catalan issue, only this category grouping specifically people that point to it as determining or changing their electoral behaviour can be said to capture a significant statistical relation between such contextual factor and the electoral choice of VOX voters (6.49, p < 0.001 in the final model). Conflating all the possible categories into a single yes/no influence question, on the contrary, is a statistical artefact that produces a significant coefficient by diluting a specific and conceptually distinct subsample. Together with the finding already cited from Rama et al. that the Catalan conflict is not particularly concerning in VOX supporter’s list of national problems, these results further clarify the salience of the Catalan conflict and imply that the issue, contradicting the main consensus found in the literature, did have a substantive effect but only bears explanatory power for a minority of VOX voters (27.5%) – as opposed to 54.4% if the binary influence measure is taken to be reliable. Which makes it necessary to reassess the widely-held idea, both in public and academic discourse, that the Catalan conflict is all there is to VOX, that VOX’s vote share is largely reflective of nationalist protest.

As regards the blindness of previous research to economic factors, Model 3 indicates the idea that it is ‘not the economy’ might be mistaken yet fails to offer conclusive evidence. According to the data, holding a negative evaluation of one’s personal economic situation is associated to an increase in the odds of voting for VOX (1.71, p < 0.05). However, this only holds at a standard significance level in the final model. The variable registering whether someone points to some economic issue as the main problem in Spain registers a negative association with voting for VOX (0.63, p < 0.05 in the full model), but its significance is lost in the final model. Therefore there is evidence to suggest subjective evaluations of one’s personal situation rather than objective measures, in line with recent research, are relevant in this case study. Nevertheless, as aforementioned, the variable used to tap into subjective status (opinion of one's economic situation) is an imperfect approximation. Capturing such complex evaluations might be best done through qualitative insight (Burgoon et al., 2019, pp. 54–55), to which I will turn in the following section.

Finally, addressing the issue of VOX’s low-income constituencies can be done by replicating Models 1–5 for a reduced sample that only considers voters below the median 2019 income (15,015 € (INE, 2020), approximated to the closest income group) because this is the way to capture that section of the population where according to Sánchez’s (2019) data VOX were comparatively most successful in the November election. The significant variables and their relative substantive impact remains virtually the same as in the full-sample models, except for those of subjective social class which can be seen as a product of the sample constraint. Overall, there are no differences with the original models which may hint at the distinct motivations of this section of the electorate, except for the weaker p-values which can be attributed to the reduced sample: gender and economic issues as the main problem become non-significant, and the p-values of education and immigration lose strength (Table 3).

Table 3. Regression models on the low-income sample.

	Model 1	Model 2	Model 3	Model 4	Model 5
Gender	0.762	0.679	0.776	0.678	0.647
	(−1.00)	(−1.35)	(−0.92)	(−1.33)	(−1.62)
Age	0.625***	0.647***	0.602***	0.618***	0.649***
	(−5.40)	(−5.10)	(−5.42)	(−5.28)	(−5.13)
Education	0.435	0.391	0.421	0.373	0.299*
	(−1.54)	(−1.86)	(−1.56)	(−1.92)	(−2.43)
Income group 2	1.301	1.298	1.384	1.404	1.485
	(0.92)	(0.90)	(1.10)	(1.13)	(1.39)
Income group 3	1	1	1	1	1
	(.)	(.)	(.)	(.)	(.)
Subjective class 1 (mid)	17.93**	18.44**	20.47**	21.49**	5.984**
	(3.06)	(3.12)	(3.11)	(3.18)	(2.62)
Subjective class 2 (mid-low)	19.83**	17.85**	19.21**	16.51**	5.711*
	(3.03)	(2.94)	(2.95)	(2.81)	(2.38)
Subjective class 3 (working-class)	40.54***	38.09***	41.62***	38.38***	13.29***
	(3.71)	(3.68)	(3.69)	(3.64)	(3.61)
Subjective class 4 (poor)	41.32**	34.92***	37.55**	31.48**	7.140*
	(3.28)	(3.29)	(3.09)	(3.15)	(1.96)
Subjective class 5 (other/don’t know)	28.75**	27.79**	30.65**	28.23**	8.986*
	(2.86)	(2.81)	(2.82)	(2.64)	(2.20)
Rurality	0.844	0.936	0.872	0.970	0.833
	(−0.64)	(−0.24)	(−0.49)	(−0.10)	(−0.68)
Church attendance	0.901	0.908	0.964	0.989
	(−0.99)	(−0.87)	(−0.33)	(−0.09)
Exclusionary national sentiment	0.841	0.715	0.835	0.714
	(−0.47)	(−0.88)	(−0.45)	(−0.82)
Left-right self-placement	2.006***	2.062***	2.006***	2.054***	2.196***
	(8.51)	(8.41)	(8.41)	(8.50)	(10.92)
Political interest level 2	1.045	1.038	1.051	0.972	1.142
	(0.08)	(0.07)	(0.08)	(−0.05)	(0.28)
Political interest level 3	0.504	0.485	0.480	0.428	0.477
	(−1.25)	(−1.37)	(−1.28)	(−1.57)	(−1.53)
Political interest level 4	0.522	0.534	0.492	0.457	0.401
	(−1.08)	(−1.08)	(−1.14)	(−1.31)	(−1.65)
Immigration as one of the main problems	2.551**	2.386**	2.473**	2.283*	2.395**
	(2.92)	(2.64)	(2.75)	(2.41)	(2.73)
Influence of the Catalan issue	2.572***		2.491***
	(3.59)		(3.36)
Catalan issue as one of the main problems	1.372	1.575	1.375	1.647
	(0.77)	(1.09)	(0.80)	(1.19)
Relevance of the Catalan issue (Pro-independence)		1		1	0.565
		(.)		(.)	(−0.81)
Relevance of the Catalan issue (Pro-dialogue)		1		1	1
		(.)		(.)	(.)
Relevance of the Catalan issue (Reinforce)		1.477		1.324	1.372
		(1.12)		(0.76)	(0.92)
Relevance of the Catalan issue (Vote)		2.080		1.932	2.307
		(0.83)		(0.70)	(0.93)
Electoral relevance of the Catalan issue (Change)		7.435***		7.808***	8.095***
		(5.20)		(5.44)	(6.09)
Subjective economic position			1.617	1.871	1.781*
			(1.46)	(1.90)	(1.99)
Economic issues as the main problem			0.796	0.750	0.846
			(−0.84)	(−1.02)	(−0.66)
Economic issues as the main personal problem			1.251	1.275
			(0.69)	(0.71)
N	1178	1148	1147	1117	1516

Note. Regional fixed effects not included. Exponentiated coefficients; z statistics in parentheses.

* p < 0.05, ** p < 0.01, *** p < 0.001

Overall, beyond the Catalan issue immigration seems to be the most widely relevant factor. Yet we cannot ascertain its underlying meaning, nor its relation to the relevance of subjective deprivation or the neoliberal economics of VOX. More fundamentally, the question as to why and how immigration is so relevant to these voters is significant because it deserves little discussion in Rama et al. even if it challenges the logic of a protest-vote hypothesis, meaning the motivations of voters are left largely unexplained. So the puzzle remains, and is particularly pressing with regard to the motivations of that substantive low-income electorate.

These results which reopen what seemed a closed debate, combined with the data discussed previously regarding the key relevance of a new low-income electorate, point to the interest of an interpretive approach to grasp the motivations of this section of VOX’s support base. The theoretical interest of such an approach is greater to the extent that there is a need, as I argued previously, to conduct research that questions prevalent categories and assumptions in the literature, which I have shown are reflected in the more specific debate on VOX. To this second level of analysis I turn now, focusing on a more constrained exploration of the motivations of low-income voters as the most intriguing constituency, to seek an explanation to the success of VOX that can make sense of available data better than previous theories.

Grounding VOX: Addressing the puzzle through in-depth data

What I will present now is a summary of the analysis of the interview data generated in my fieldwork in 2020. This part of my research, delving into the constituency of low-income voters, does not seek to make claims at the same level as the preceding regression analysis: as aforementioned, the objective is not to employ mixed methods to (dis)confirm or triangulate, but to complement previous data by exploring social meanings and processes within a subsample of theoretical and substantive relevance. This interest is justified because it concerns the core constituency of the far right yet in a paradoxical context that defies the way we traditionally theorise it, and it can therefore provide unique perspectives from which to reflect upon the broader phenomenon of working-class far-right constituencies (see Burawoy, 2009, pp. 42–44). Furthermore, given the key role of this segment of voters in the rise of VOX, it will provide a potential explanation for their overall success by illuminating the processes underlying statistical data which readily available theories cannot make sense of. The objective is, therefore, an analytical generalisation (Yin, 2018, pp. 20–21) that complements the previous statistical generalisation, taking into account that in-depth interview data is ideally suited to explore the processes and mechanisms underlying statistical correlations (Gerson & Damaske, 2021, pp. 34–35).

How did respondents account for voting for VOX?

To answer the first part of the research question I carried out a thematic analysis of the coded data to map the overarching social narratives that cut across respondents. In a second step, this thematic extraction of patterns was ordered and interpreted with the application of the Essex toolkit to reconstruct respondents’ perception of the political space, the place of their demands within it, and their relation to VOX’s discourse. I sought, so to say, to map the ‘present state’ of opinion and if/how it might reveal the anatomy of a populist articulation. In what follows, I will refer to respondents using pseudonyms; whenever I list several respondents, I will use only initials for the sake of readability.

The first relevant finding is respondents expressed concerns consistent with those of other far-right voters regarding immigration (e.g. Dörre et al., 2006; Hochschild, 2016), even if belonging to a supposedly ‘exceptional’ context. In line with the statistical analysis, much resentment was expressed towards immigrants, but not in a conventional ‘cultural’ sense (e.g. Rama et al., 2021, p. 87) but because they were perceived as undeserving recipients of welfare benefits:

There’s a lack of control since … since they started giving out national IDs, especially to people coming from South America […] there we lost control. Everything was expanded too much, later it collapsed – with the crisis – and … but I actually think this lack of control in Spain is something persistent. (David, male, 31, fixed-term employment)

David’s anxiety, condensed in the issue of citizenship rights, synthesises how respondents framed immigration basically as a problem of (re)distribution of resources – which as I will show involves much more than just ‘resources’.

Building on the last sentence quoted, this respondent hints at how alongside immigrants, as is also common for other far-right supporters (e.g. Hochschild, 2016, p. 35, 61), the category of the undeserving extends to other groups that appeared frequently across respondents’ accounts: lazy welfare recipients, beneficiaries of gender positive discrimination, regional governments that enjoy concessions by central government, and to the host of unproductive associations and state organisations (e.g. gender equality organisations) that were seen to proliferate as a wild garden of ‘ideological’ spoils. Respondents expressed in this way strong feelings of resentment linked to the privileges of other groups in society: type of narrative reflective of the notion of positional deprivation (Bos et al., 2020), and potentially illuminating the statistical relevance of subjective economic status. For most respondents, these forms of positional deprivation are associated to norms of positive discrimination supported by the left, which in turn are characterised sometimes as buenismo (literally ‘goodism’, meaning an unjustified excess of consideration for certain social groups).

Second, the data produced a more interesting finding that is not captured in the statistical analysis: besides feeling deprived in relation to other groups in society, respondents expressed great resentment towards the political ‘class’. This type of scheme therefore reflects a combination of both horizontal and vertical differentiation characteristic of the far right (Staerklé & Green, 2018): namely, antagonism towards both groups of fellow citizens and political elites, frequently because the latter are perceived to collude in the promotion of immigrants’ interests (Mudde & Rovira Kaltwasser, 2018). In line with what is expected of populist discourses, it was the vertical axis that defined the fundamental frontier of antagonism structuring respondents’ perception of the political field.

This frontier was surprisingly harsh: the political class was characterised by almost all respondents as an elite that is indistinguishable because it shares a common interest in controlling state structures to extract resources from citizens. Many respondents also perceived them as habituated to lying: the concept of intereses (‘interests’, used to refer to the hidden actual interests that move politics) was used by most respondents to describe the inner establishment workings.

However, respondents’ antagonising of the political class was broader than is generally assumed when speaking of elites in populism research. More specifically, respondents expressed strong negative opinions of the Estado de las autonomías – ‘state of autonomies’, the Spanish decentralised arrangement of political power in multiple Autonomous Communities – because it is perceived as a costly duplication of public offices that ‘only benefits politicians’ (Silvia, female, 23, student). This finding is in line with scholars highlighting the salience of opposition to territorial decentralisation among supporters of VOX (Arana, 2021; Vampa, 2020). Inasmuch as the structuring itself of the state is linked to politicians’ private interests, some respondents even reached the extreme of perceiving the state as patrimonializado (i.e. as literally belonging to the ‘estate’ of politicians).

This blurring of the boundary between the state and its transitory holders is easier to understand considering a democratic, welfare state only started taking shape in the 1980s, meaning it is harder to disentangle from its creators as an atemporal entity. A second relevant factor is the approximately 600 cases of political corruption unveiled in Spain since democratic transition in 1978 (Castañón et al., 2016), and in which many respondents found proof of the immoral ‘interests’ guiding the political class. Altogether, it is impossible for respondents to conceive of positive reform that does not entail drastic change within state structures:

We have to reduce the … structural cost of Spain: assistants, public positions and so on … all of that has to be cut. […] Because in the end what we’re doing is creating a state just to sustain them … The more years go by, the more there is of them … […] We have to … close or minimise the structure of the state for this to be under control. (Alberto, male, 49, permanent employment)

What Alberto illustrates is how, to the extent that the representative relation is broken, and the state and its regional governments, social services, etc. are distrusted as unreliable crony business, state minimisation becomes an end in itself for these respondents.

Previous studies found a significant correlation between voting for VOX and discontent with the political situation or dissatisfaction with democracy (Rama et al., 2021, p. 130; cf. Turnbull-Dugarte, 2019; Turnbull-Dugarte et al., 2020), but none tested political (dis)trust directly at the level of national elections. CIS data offers no specific item but a good proxy is to single out those pointing to ‘politicians in general’ or ‘the government’ as the main problem in Spain: while VOX voters in aggregate do present a higher frequency – 31.9% vs. 26.9% in the total sample, 28.4% vs. 24.2% in the sample below median income (CIS, 2019a) – , the testing of a dummy variable failed to produce a significant result when added to the previous analyses. A hypothesis one might venture is that, given an extremely high level of political disaffection exists in Spain, what distinguishes VOX supporters is not their frustration but its depth and meaning, which would determine in turn political mobilisation.

Accordingly, the prime demand on which all respondents agree is not welfare chauvinism, as is common for workers supporting the far right (Dörre et al., 2006), but a reduction of taxes: ‘it is true that we pay a lot of taxes, in general […] and we get less and less in return’ (Julián, male, 30, permanent employment). And this position is reflected in statistical data: when respondents are asked about state intervention, VOX voters are in line with other far-right parties in expressing the lowest support (Rama et al., 2021, p. 85); however, Rama et al., operating within the hegemonic economy-culture frame, can only see this an indication that the economic thesis of the ‘losers of globalisation’ is inapplicable. In presuming what the meaning of certain opinions must be, they are blind to the possibility of alternative discourses linking identities, demands and voter preferences. But from an in-depth exploration, it is this policy that seems to be the key demand representing and hegemonising all the rest (reduced immigration as a result of decreased welfare entitlements, opposition to PC and new gender norms for which funding must be cut, opposition to decentralised institutions as duplicating costs, etc.) linking them to a univocal policy solution that only VOX was putting forward: state-minimisation.

Following the Essex scheme, we must now turn our attention to the signifier(s) that thread all of these demands within the discourse exposed here as a clear, simple referent (i.e. the we side of the we-them populist frontier). In this case, who the people are for respondents was very clear: those who have higher ‘merit’ for various respondents, ‘those who are really hardworking’ (Julián, male, 30, permanent employment), those who have ‘cultura del esfuerzo’ (Alberto, male, 49, permanent employment) – literally ‘effort culture’, meaning a predisposition to work hard. Powerfully resonating with previous qualitative research in the field (Hochschild, 2016, pp. 71–72, 158; Kurer, 2020), what all respondents share is an identification as hardworking.

More specifically, revising the data units grouped under the hardworking theme, I realised what they express is a consistent norm as to how rights ought to be allocated, which is not according to citizenship or ethnicity but to one’s contribution; a morality or value-orientation perceived to have declined up to a point where nowadays, with the push of buenismo, ‘the policy has been established that everyone has a right to everything’ (Cristina, female, 53, permanent employment). This contributionist principle is congruent with their structural position since it is reminiscent of historical discourses of workers’ movements for welfare rights (see Claeys, 2015) (tellingly, some respondents referenced their parents as transmitting such value-orientations). But in this case such a claim for redistribution acquires an individualised form, coming close to the neoliberal vindication of the competitive, self-responsible individual (Brown, 2015): it represents not a victimised class but the moral currency of the ‘attitude of […] fending for oneself’ (Vanesa, female, 53, unemployed); that ‘like it or not, you have to manage on your own to get out of there’ (David, male, 31, fixed-term employment).

This conceptualisation is the legitimate, moral people they identify with. Even if they express signifiers and values resonant with traditional left-wing politics, these become linked to a demand for ‘more free market’ (Enrique, male, 20, student and part-time employed), because it is regarded as a space where the worth of effort is undistorted by the interests or buenismo guiding state intervention. Given their extremely negative perception of state intervention, the defence of the hardworking and the expansion of the market are one and the same thing: the signifier hardworking is the condensed representation of the fusion of a set of unattended demands with a single hegemonising demand (state-minimisation) that has come to channel and transform their meaning.

Overall, this analysis carves a feasible connection between subjective deprivation, immigration and ideology that addresses the first part of the research question and speaks back eloquently to statistical data. It also suggests, contradicting the literature’s consensus, that both economic position and economic policy had a great impact in determining support for VOX among this electorate, who as indicated above are primarily concerned with economic issues (CIS, 2019a); even if calling it ‘economic’ overlooks the fact that these issues only make sense when understood as embedded in cultural values, status concerns, and desires for social recognition – echoing recent findings in studies on the morality of economics (e.g. Fouksman, 2020).

The only point remaining is how the Catalan issue, which is taken to be so central in the literature, fits this discursive mapping. In line with my statistical analysis, the interview data supports my hypothesis that this factor has had a decreasing influence on electoral behaviour. Indeed, for all my respondents the importance of Catalan secessionism was at best secondary, and in half the interviews the issue did not even come up until I probed on it. When respondents talked about it, they predictably expressed outrage, but it was always other issues they were more concerned about. In other words, the Catalan crisis appeared not as a motivation for their electoral behaviour, but as an argument, a phenomenon they took as evident proof of how urgent their concerns were and of how precise VOX were in their diagnosis. Catalan secessionism was logically incardinated for them inside the broader discourse described above, where I already noted the expansion and tolerance of regional autonomy as a relevant grievance blamed on buenismo. Whereas its little actual relevance to these voters may seem surprising in light of how central it is to prior research, such mismatches between quantitative and qualitative (or mixed-methods) analyses are not surprising when the former impose predetermined categories, meanings, or assumptions upon the data. This kind of problematic is exemplified in some of the research on euroscepticism cited above (Duchesne et al., 2010; Van Ingelgom, 2014). Fundamentally, ‘[t]he salience of an issue cannot be equated with its presence alone’ (Van Ingelgom, 2014, p. 179). My findings also echo previous research which finds that what happens at the level of elite discourse can be deeply misleading as to the processes going on beyond the spotlight (e.g. Brubaker et al., 2006). In the case of these respondents, what is clearly the issue that VOX is best known for and takes precedence in their electoral manifesto (VOX, 2018, p. 2) is quite secondary and would not make them shift their politics were the Catalan issue to ‘disappear’.

Overall, one question remains: why this fixing of meanings and political frontiers, and not another? And how is all this related to their experiences of deprivation? What remains now is the question of how such demands, clearly in line with neoliberal policy tenets, have come to be paradoxically embraced by low-income voters, whose experience of deprivation in a Southern debtor country should hypothetically orient them the opposite way (Hopkin, 2020). To the issue of this ideational and behavioural movement I turn now.

How did respondents’ experiences of deprivation relate to their vote for VOX?

In the previous section I exposed how respondents are voting for VOX because their demands are best served by VOX’s contestation of key consensuses of the establishment: welfarism, PC, gender policy, immigrant rights, etc. What remains to be seen is how this articulation has developed from specific biographies and, more specifically, if and how this construction is grounded in respondents’ experiences of deprivation. While the previous analysis sought to describe the present state of opinion of respondents, what concerns me now is the connection between this and their life narratives (i.e. to discern a movement instead of a snapshot). Within those narratives, I sought to reconstruct the role deprivation played in the process that connects for each respondent experiences of dislocation – if these exist – with the creation of demands that finally came to constitute the populist frontiers that I have mapped in the previous section.

First, almost all respondents held a better socioeconomic position in the past and a majority narrate experiences of decline related to the impact of the 2008 crisis. A representative account is that of 53-year-old Cristina, formerly a high-earning professional in a real estate company, who has only recently managed to regain stable employment:

I have never been able to think about buying a home, never, never. Firstly because I preferred to invest in my son’s education, and secondly because I never knew what was going to happen, I never knew if someone was going to come around and raise taxes. Also now mortgages are easier, before they were so tough … you never know, and that uncertainty killed me.

Instability is associated both to the market and the state, since she feels ‘in the hands’ of both banks and politicians, a duality relevant for all respondents affected by the crisis. But it is not against the abusiveness of mortgage conditions that she is voting. After being fired in 2008, claiming unemployment benefits made her realise all her years of tax contributions did not pay off: while ‘a lot of money goes to cronies, they’re selling an image of a welfare state which in fact is not such, it’s not true. I don’t know anyone who can live on 600 euros [a month]’. Since then, every time her payroll is issued, she feels outraged at the amount of taxes listed, and this is what makes her ‘sick’, what moves her politically. Just as many other respondents, she expresses great resentment due to the continuous rise in taxes throughout the crisis. Speaking not of contributions but benefits, Gloria (female, 22, student and part-time employed) expresses the same kind of logic in a more relation manner:

my parents, at the time when they were doing worst, went to social assistance and what they saw was this, a load of immigrants, and all the aid going to them … If you live here, you’re Spanish, you’ve worked, you’ve contributed, you’ve paid your taxes, and there is no way of … not privilege, but it’s logical to think you will be helped out, but no, instead it’s going to people who … who really do nothing … it doesn’t make sense. It makes you feel stranded.

Paying attention to life trajectories is fundamental to flesh out the concept of dislocation which I indicated refers to experiences of severe destabilisation of notions of self. More importantly, people’s narratives offer a particularly privileged site to understand processes of identity reparation (Riessman, 2014) which is where we expect to map the logic of populism. In this case, dislocation has been caused by economic shocks, but also by the experience of the state as ‘a thorn in my side’ (Pedro, male, 58, unemployed) throughout this period of hardship in their lives. The question is why it is only the latter that has been politicised, while in Hochschild’s words they ‘accommodate the downside of the free market’ (2016: 124) – and even embrace it, as reflected in the previous section.

My argument is three key factors favoured this outcome: first, the lack of state capacity to redress the recession macroeconomically, or to aid them directly, put them in a situation where they had to struggle on their own. Whereas nothing is expected from ‘the market’, the state is subject to expectations upheld regardless of economic failure being a product of market dynamics. Failure to live up to said expectations could be the root cause of the destabilisation of traditional links between signifiers of working-class heritage and welfarism, making these signifiers capable of floating across a range of discursive formations because their contents had become undetermined upon being ‘abandoned’ by establishment discourses (Laclau, 2005, pp. 163–168). The ultimate proof of this abandonment for most respondents is that retirement pensions, which they value not as welfare but as due payment for their years of hard work, should currently be endangered due to ongoing fiscal and demographic strains. However, succeeding in their individual struggles has provided a platform from which to reconstruct their damaged status by drawing pride from endurance, as is explicitly put in many interviews, in line with their vindication of a hard-work morality. It is noteworthy that the EU, as a key actor behind the policy orientations that are responsible for these outcomes, is completely absent from this narrative except for a couple references that are largely positive. The fact that historically Spaniards have been exceptionally supportive of EU integration and linked it specifically to economic prosperity (Díez Medrano, 2003), and that said attitudes have not been significantly affected by the Eurocrisis experience (Real-Dato & Sojka, 2020), helps to understand the little salience of euroscepticism which is otherwise a common institutional focus for far-right populism (Halikiopoulou et al., 2012).

A second relevant factor is that the shocking amount of cases of political corruption aforementioned have mostly burst into the open after the crisis, increasing dramatically the awareness of citizens since then (Myśliwiec & Stachowicz, 2018). Along with the limitations of the state to counteract the crisis, corruption likely aggravated the experience of dislocation; at the same time, it provided a clear antagonist:

When you see MPs and politicians don’t have the same level of income tax, and […] they’re even exempt from paying for their phones. But what the fuck is this? I mean, what did they do the law for? For themselves only? I think it just all comes down to them saying: let’s just set up the law so that it benefits all of us here in government … at the expense of the working Spaniard. (Pedro, male, 58, unemployed)

What all respondents share is a personal testimony of how these years, while the corrupt political class ‘kept growing more and more and more’ (Pedro, male, 58, unemployed) in spite of the recession and the cuts to welfare, they had to put up with undeserved deprivation and managed without anyone’s help. This is something only deprived voters can wield politically: they have become entitled to contest the legitimacy of a political class that justifies its expansion on the merits of a welfare state they have discovered to be weak and unfair, and to contest the legitimacy of buenismo hegemony on the grounds that they have endured without asking for anyone’s pity. From their position of deprivation they have become legitimised to speak these truths, and therefore to ask certain questions that are chiefly aimed at immigrants, for the reasons spelled above: if we could endure on our hard work, why should others not? These concerns are exacerbated since a left-wing government took office in 2018 – the year of Vox’s first breakthrough. Following Pedro,

With all that’s happened in this country, and now we find that everything we have managed to achieve they are trying to take from us. That is why I support VOX: because I believe we’ve fought too hard for all this to disappear and … be diverted to people it shouldn’t be given too.

The depth of his words points to the relevance of applying a dynamic perspective which captures how the grievances fuelling VOX cannot be reduced to reactive dispositions, but rather consist of rich, personally significant discourses fermented in the long term.

A third enabling factor is the hegemonic capacity of VOX’s discourse within this dynamic frame: a discursive formation which was suited to incorporate their demands into an equivalential chain that neatly separates them from compatibility with the policies and state structures in relation to which they have developed a feeling of deprivation, and whose prevalence undermines the moral worth of their individual endurance. This is not to say that respondents should be constructed necessarily as self-interested rather than principled, as buying into ideas of collective governance solely to suit private-status concerns. I do not mean to imply a value-judgement that they do not actually believe in their political positions as sound and collectively beneficial, which is disconfirmed by the passionate and elaborate arguments of many respondents. But my analysis does indicate that politicisation stems primarily from personal experiences of dislocation, and that these in turn become prime arguments of ‘lifeworld’ knowledge, which is concerning to the extent that this makes political opinion resistant to counter-arguments of a theoretical or empirical nature.

The concurrence of these processes at a given historical moment is the endpoint of a narrative that connects back to the previous section’s discursive mapping: the point at which dislocated identities and unaddressed demands fix into new discourses that transform them in the process of articulation. This influence in many respondents’ thematic networks is explicit and manifest in many data units where they reproduce the same concepts used by VOX, such as ‘la España que madruga’ (Discurso de Santiago Abascal En Vistalegre Plus Ultra, 2019, pt. 26:21) – ‘Spaniards that wake up early’, meaning those that work hard, representative of a broader strategy to attract working-class voters.

Overall, applying the logic of Essex theory makes it possible to understand such displacements not as incongruent or irrational behaviours, but as the resignification of particular demands when these are disregarded by the system and become articulated into a discourse that imbues them with a new universality (i.e. the anti-systemic equivalence condensed in the defence of the hardworking) which is, in turn, deriving always its meaning from the particularity put forward by a hegemonising agent (i.e. VOX's neoliberal project) in opposition to a clear antagonist – immigrants, welfarist elites. Populist projects construct new political identities by adhering heterogeneous social demands through operations such as the one mapped here. Just as VOX might simultaneously try to articulate into the same neoliberal discourse, for instance, unattended conservative demands by associating LGBT expansion to state intervention in schools (e.g. Discurso de Santiago Abascal En Vistalegre Plus Ultra, 2019, pt. 14:20).

Conclusions

The main objective of this paper was to challenge the academic consensus built around the rise of VOX as an exclusively ‘cultural’ phenomenon, largely reducible to a reaction against minority national secessionism. To this end I extracted new insights from data previously used by existing scholarship, combined with new census data, to question the conclusions in previous research. Furthermore, I have provided original in-depth interview data that subsequently undermines the analytic premises on which that data had been interpreted. In parallel, constructively both parts work in complementarity to carve a different ‘story’ as regards rising support for VOX, as well as a different path to how mass populism can be studied.

Speaking to the debate on the demand-side of far-right populism, my discursive analysis shows the importance of ascertaining the meanings of citizens’ demands so as to integrate a mass perspective rigorously, which can in turn problematise concepts and assumptions ingrained in extant literature. In line with advocates of ‘economic’ explanations, these findings do support the idea that the changing economic environment is a necessary condition to understand the success of the far right – even in the case of Spain which had been previously framed as ‘cultural’. But an economic shock can only be understood as the primary cause of dislocation, not of the orientation of subsequent political preferences, which are determined by complex processes where the agency of ordinary people’s meaning-making efforts matters indeed. In the case of VOX, we find the ‘white working-class’ which constitutes the core base of far-right populism being consolidated in an unlikely scenario, and I propose that rather than an outlier this happens due to factors very similar to those fuelling the far right in otherwise very different countries (i.e. the politicisation of deprivation, its linking to migration), yet through logics and discourses of unexpected nature and implications. More specifically, I point to how material deprivation in the wake of an economic shock can become linked to neoliberal policy responses; but not indirectly as an unnoticed collateral for ‘cultural’ voters, which is a common trope (Afonso & Rennwald, 2018). Rather, I show a direct endorsement exists, built via elaborate narratives of dignity, morality and effort projected primarily upon immigrant others. Most concerningly, my analysis indicates a vicious spiral exists whereby deprivation derived from decreasing welfare expenditure and deregulation may actually foster support for further state minimisation among those most in need, which can then be most coherently channelled via far-right politics.

Moreover, this case study speaks to the broader literature on populism conceptually, methodologically, and theoretically. First, as is already reflected in the previous point, it problematises the generalised practice in populism research to reproduce unjustified assumptions that preordain the terms and scope of scholarship: the idea that there must be distinct economic and cultural explanatory paths, that these must map neatly onto specific phenomena, that these must represent a cross-culturally valid heuristic, that each must be linked to a bounded array of possible meanings. This is shown to be a wholly inadequate frame to make sense of political discourses that enmesh identity and deprivation, values and redistribution, morality and resources, whose analytical dissection would make unintelligible the rationales behind respondents’ behaviour. My findings therefore support previous calls to transcend rigid conceptual assumptions (Bonikowski, 2017; Gidron & Hall, 2020) as well as scholarship pointing to the need for a more interpretive approach to radical politics (e.g. HoSang & Lowndes, 2019).

Second, I do so by way of a preliminary application of new analytical methods that show valuable insights can be reached by applying systematically the Essex toolkit to mass-level data, pointing to how this scholarly tradition can be reinvigorated through innovative approaches. Indeed, it is an inductive focus on dynamic demands that has made it possible to open up new perspectives on an issue as extensively researched as the core constituency of the far right. In turn, besides animating more interpretive work on mass populism, I hope my mixed-methods design can prompt more ecumenic research and debate within a predominantly quantitative literature.

Finally, this paper also challenges populism scholarship at a theoretical level: problematising the concepts we use inevitably unhinges the structure of the theories built with them, allowing for fresh perspectives to emerge. In this case, my analysis proposes theoretical avenues that cannot fit the prevalent theoretical conversation because of a tendency to privilege clear, univocal explanations (either culture or economy), and a tendency to assume such explanations should be not contextual but cross-culturally valid, in line with an orientation to ambitious global theorisation (see Anastasiou, 2019) – rientations that might be the origin of the aforementioned conceptual issues. My point is this theoretical path fails to understand the nature of populism itself and of appeals to the ‘people’, whose conceptual emptiness is attuned to the de-structuralised politics of (post)modern societies (Laclau, 2005/2007, p. xi) and suited to the necessity of articulating increasingly heterogeneous constituencies and demands into operative political subjects. From this standpoint, what is relevant is not whether populism can be correlationally traced to a specific factor, but why and how multiple factors are constructed to produce populist articulations in the resonances between lay and elite discourse. What also becomes relevant is whether such articulation exists at all (i.e. whether populists merely surf discontent or actually build long-term identification, or both at once yet in relation to different constituencies). Whereas these are very much ignored questions in the literature, I have shown they can be critical to grasping the causes of populism as well as its implications.

I want to finish by highlighting Bourdieu’s (1979) point which started off this paper, and whose relevance is reflected in the results: the dangers of relying too confidently on data and methods that offer limited insight into the dynamism and embeddedness of public opinion. The salience and meaning of issues, as well as their interrelation, cannot be assumed but require in-depth exploration that seeks not merely description but causal processes. The difference is so significant that what seemed to be the mighty wind beating an otherwise calm sea of public opinion – the Catalan conflict – might be recast as a rock, relevant but nevertheless static, against which a silently building tide finally crashed – a tide already perceptible years before the Catalan conflict escalated (Alonso & Rovira Kaltwasser, 2015).

Acknowledgments

I want to thank Ellie Knott, Liene Ozoliņa, Maria Stapleton, and Oriol Bosch for their useful comments on earlier versions of this paper as well as for their wonderful support throughout the time I worked on this project. I also want to thank the editors and the two anonymous reviewers, whose dedicated engagement with the paper greatly improved it.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

The project that gave rise to these results received the support of a fellowship from ‘la Caixa’ Foundation [ID 100010434]. The fellowship code is [LCF/BQ/EU21/11890039].

Notes

1 I explicitly constrain my understanding of VOX as a case of far-right populism. I do so following the most comprehensive conceptual index of existing populist parties (Rooduijn et al., 2019), which defines the category following Mudde’s (2007) seminal conceptualisation of this party family as combining nativism, authoritarianism and populism (for more qualitative analyses of VOX’s discourse, see Ferreira, 2019; Marcos-Marne et al., 2021; Ramos-González & Ortiz, 2022). In this index, populism is conceptualised according to ideational theory which I will discuss later. My more constrained and explicit focus contrasts with a tendency to research and debate support for populism in general, and follows Mudde and Rovira Kaltwasser’s (2018, pp. 1674, 1678) indication that conflating ideological differences between left and right within the populist party family can endanger the validity of causal inference. Accordingly, I focus my literature review on debates regarding the demand for far-right populism, and throughout I point distinctly to the ways in which this paper contributes on the one hand to debates on the causes of far-right populist support, and on the other to inform theories and methods general to the study of populism more broadly.

2 To my knowledge, the only exceptions to this are two studies that also employ primary qualitative data, even if only as secondary data sources (Griggs & Howarth, 2000; Sik Ying Ho & Kat Tat Sang, 2000).

3 Given a categorical variable with 11 levels and a non-normal distribution, I opted for a 3-category simplification with equal-size groups as opposed to Rama et al.’s distribution of 3 categories with disparate frequencies. A check on the alternative model using this same distribution showed no significant effect on the results.

4 Further information regarding respondents is available in Appendix 1.

5 While this is certainly a very simplified presentation, it is also the aim of this paper to make more accessible in the form of a “toolkit” an approach that has frequently been perceived as obscure. For instance, recent attempts to reformulate the study of populism from a more sociological perspective (Bonikowski, 2017; Jansen, 2011) have failed to build on Essex theory yet have proposed ideas strikingly close to those of said theory.

6 The interview scheme is included in Appendix 2.

7 Detailed information of all variables introduced can be found in Appendix 3.

8 Note that the option pro-dialogue is omitted by the model because there is no single observation where a respondent is at once pro-dialogue and votes for VOX.

Appendices

Appendix 1.

Respondent profiles.

Table

RESPONDENT	AGE	GENDER	EDUCATION LEVEL	OCCUPATION	RECRUITMENT	INTERVIEW DATE
Martín	45	Male	Medium	Permanent employment	Acquaintance contact	8/1/20
Gloria	22	Female	High	Student and part-time employed	Acquaintance contact	8/1/20
Alberto	49	Male	High	Permanent employment	Acquaintance contact	9/1/20
David	31	Male	Medium	Fixed-term employment	Acquaintance contact	10/1/20
Vanesa	53	Female	High	Unemployed	Acquaintance contact	1/4/20
Cristina	53	Female	High	Permanent employment	Online	05-07/5/20
Silvia	23	Female	Medium	Student, formerly employed	Through Gloria	19/5/20
Pedro	58	Male	Medium	Unemployed	Through Cristina	11/6/20
Enrique	20	Male	Medium	Student and part-time employed	Through Silvia	15/6/20
Julián	30	Male	High	Permanent employment	Through Silvia	1/7/20

Appendix 2.

Interview scheme

I reproduce here the interview script as it was used in the second phase of fieldwork, with minor modifications resulting from the preliminary analysis of the data generated in the first phase:

1. Introduction, consent form

1. Age, gender, nationality.

2. Family situation.

3. Level of education, occupation, employment conditions. Experience of temporary/precarious employment or unemployment.

2. Material situation

1. How do you earn a living? What jobs have you had throughout your life?

   1. Has the recession affected your life? How have you lived through it?

   2. How did it make you feel?

   3. How about others in your same company/occupational group?

   4. What are your expectations for the future?

2. What public services do you use? Do you have an opinion on them?

   1. Do you think they have changed since the crisis?

   2. Consider asking about healthcare/transport/pensions/education/unemployment benefits depending on their life narrative.

3. VOX

• a. In general, why would you say you voted for VOX?

•  1. What reasons do you think were most important?

  •  1. Was that the reason why you changed from [previous establishment option].

  • 2. Was it the first time?

  • 3. Do you normally vote?

  • 4. What other parties have you voted for? Did you consider voting for them in this last general election?

  • 5. If Catalonia comes up as an issue > block 4.c.

• b. Have you read the manifesto or informed yourself about their proposals in some other way?

• c. What do you think about their proposals?

  1. Relate to proposals put forward in 3.a.

  2. What is it you like most/less?

• d. What do you think about their economic proposals?

  1. Relate to information put forward in 3.a or 3.c if applicable.

• e. [Inputs on public opinions regarding VOX’s economic proposals] So what do you make of all this?

  1. And what about [x] public service you care about? Do you think it could be maintained?

  2. And what about [x] issue you said worried you before? Do you think these proposals would help?

• f. Would you vote for VOX again?

  1. [Depending on the relevance of the Catalan issue] And if Catalan secessionism were resolved?

4. Possible extensions for topics not brought up if there is extra time [there is a series of topics that VOX gives importance to and haven’t come up in the interview …]

1. What is your opinion on immigration?

   1. What do you think should be done with MENAs (foreign unaccompanied minors)?

   2. Do you live in a neighbourhood with a lot of immigration? What is your experience?

2. What do you think about the law on abortion? And on gender violence?

3. What is your opinion/experience [if they live in Catalonia] of the Catalan struggle for Secession?

   1. For residents in Catalonia: do you fear a real declaration of independence? Have you felt insecurity? Have you been involved in violent/uncomfortable situations?

   2. What do you think of the reaction of PP/PSOE/Ciudadanos?

4. What do you think about same-sex marriage?

5. Closing

1. Something you want to add/ask?

2. [off the record] What do you think of the interview? Were you comfortable?

Appendix 3.

ummary of variables.

Table

Variable	Item (translated from Spanish)	Operationalisation
Vox vote	Would you tell me the name of the party or coalition you voted for in the general election to Congress held on November 10?	Binary (1 = yes, 0 = no)
Gender	[Not disclosed]	Binary (0 = men, 1 = woman)
Age	[Not disclosed]	Continuous
Education	Have you attended school or some other type of education?	Binary transformation of a 7-category variable (1 = higher education, 0 = all others)
Income group	In what level of this scale is your personal income, after tax, that is, your net income?	3-category transformation of an 11-point scale (0 = group 1, 1 = group 2, 2 = group 3)
Subjective class	Which social class would you say you belong to?	6-category transformation of a 13-category variable (0 = high, 4 = poor, 5 = other/don’t know)
Rurality	What city do you live in?	Binary transformation of a 7-point scale according to size
Church attendance	With what frequency do you attend mass or other religious services, besides occasions related to social ceremonies, such as weddings, first communions or funerals?	6-point scale (treated as continuous)
Exclusionary national sentiment	Which of the following statements would you say expresses better your feelings?	Binary transformation of a 6-category variable (1 = only Spanish, 0 = all others)
Left-right self-placement	When one talks about politics we normally uses the concepts left and right. In this card there is a series of boxes ranging from left to right. Using a scale from 1 to 10, in which 1 means left and 10 means right, which one would you place yourself in?	10-point scale treated as continuous
Political interest	Would you say, in general, that politics interests you a lot, quite, little or not at all?	4-category scale (1 = a lot, 4 = not at all)
Immigration as one of the main problems	What is, in your opinion, the main problem that exists currently in Spain? And the second? And the third?	Binary transformation of a 37-category open-response variable (1 = immigration among the 3 main problems, 0 = all others)
Influence of the Catalan issue	What has happened lately in Catalonia, has it had an influence on your electoral choice in the election of November 10?	Binary variable (1 = yes, 0 = no)
Catalan issue as one of the main problems	What is, in your opinion, the main problem that exists currently in Spain? And the second? And the third?	Binary transformation of a 37-category open-response variable (1 = Catalan issue among the 3 main problems, 0 = all others)
Electoral relevance of the Catalan issue	What has happened lately in Catalonia, has it had an influence on your electoral choice in the election of November 10? + How did it have an influence? [If they answer the issue did have an influence on their electoral choice]	Creation of a 6-category variable by fusing the information from two embedded items (0 = no influence, 1 = yes, it motivated pro-independence vote, 2 = yes, it motivated a pro-dialogue vote, 3 = yes, it reinforced my vote choice, 4 = yes, it kept me from abstaining, 5 = yes, it made me vote for those that can best stop secessionists/changed my vote to a party that better defends the unity of Spain)
Subjective economic position	How would you qualify your current personal economic situation: very good, good, average, bad, or very bad?	Binary transformation of a 5-category variable (1 = bad/very bad, 0 = all others)
Economic issues as the main problem	What is, in your opinion, the main problem that exists currently in Spain? And the second? And the third?	Binary transformation of a 37-category open-response variable (1 = unemployment/economic problems/problems related to the quality of employment/pensions/VAT/evictions/cost of energy/problems of sole proprietors/lack of investment in industry as number one problem, 0 = all others)
Economic issues as the main personal problem	And what is the problem that affects you personally the most? And the second? And the third?	Binary transformation of a 37-category open-response variable (1 = unemployment/economic problems/problems related to the quality of employment/pensions/VAT/evictions/cost of energy/problems of sole proprietors/lack of investment in industry as number one problem, 0 = all others)
Region fixed effect	[Derived from the question ‘What city do you live in?’]	19 binary variables (1 = yes, 0 = no)