The sociodemographic profile of populist citizens

A classic literature maintains that citizens with an affinity for radicalism are typically older men with lower levels of education and a precarious job situation (Betz, Reference Betz1994; Kriesi et al., Reference Kriesi, Grande, Lachat, Dolezal, Bornschier and Frey2008). More recently, some scholars have tried to assert this same sociodemographic profile applies to populist citizens (Goodhart, Reference Goodhart2017; Inglehart and Norris, Reference Inglehart and Norris2016). For both strands of scholarship, the overall argument resembles the so-called ‘losers of modernization’ thesis, which contends that certain segments of society demand economic protection and the recovery of national sovereignty as they suffer from the transformations associated with globalization. Because of this, disadvantaged groups subsequently turn to political actors who wish to undo these societal changes and promise to insulate them from such developments. Most commonly, various scholars studying Europe argue this stimulates (right-wing) populist mobilization (Inglehart and Norris, Reference Inglehart and Norris2016; Goodhart, Reference Goodhart2017; Eichengreen, Reference Eichengreen2018; Mouffe, Reference Mouffe2018). A paradigmatic example in this regard was the 2016 Brexit referendum, which mobilized an underlying divide between socially liberal cosmopolitans (typically younger and well-educated) and older, less educated social conservatives (Hobolt, Reference Hobolt2018). Similarly, across Latin America, debates surrounding the Washington Consensus divided electorates and eventually contributed to the emergence of (populist) radical left forces toward the end of the 1990s (Weyland et al., Reference Weyland, Madrid and Hunter2010; Levitsky and Roberts, Reference Levitsky and Roberts2011).

Nevertheless, the literature raises noteworthy concerns about the ‘losers of modernization’ thesis. Scholars looking to specifically relate the thesis to populism are (too) quick to assume there is a uniform group of citizens who might or might not share populist proclivities due to modernization dynamics and who look to support populist parties. This line of reasoning, at least as a cross-national expectation, is highly problematic, however. We can identify conceptual, theoretical, and empirical concerns in this regard.

First, it confuses populist voters with populist citizens. Nevertheless, as many scholars have shown, the populist potential vastly outweighs the populist vote in any society, and we must therefore carefully distinguish between the two. Second, there is no theoretical reason why economic modernization would stimulate populism by itself. As Mudde and Rovira Kaltwasser (Reference Mudde and Kaltwasser2018: 8) argue, ‘this theory goes back decades […] but in almost all previous cases, it explained nationalism or the radical right, not populism per se’. Third, and closely related to the previous argument, empirical evidence for the ‘losers of modernization’ thesis remains – at best – ambiguous. For example, Remmer (Reference Remmer2012) shows that increasing support for leftist populist leaders in Latin America has been driven by economic improvement, not deterioration, and the subsequent improvement of subjective well-being (cf. also, Wiesehomeier and Doyle, Reference Wiesehomeier and Doyle2013). Something similar can be said of Europe, where indicators of absolute deprivation, such as low income and unemployment, and indicators of relative deprivation, such as socioeconomic inequality, only partially explain support for the populist radical right (Arzheimer, Reference Arzheimer, Backes and Moreau2011; Spier, Reference Spier2010). Not by chance, some scholars argue that the rise of populist forces has less to do with objective changes and more with subjective feelings, such as social status perceptions (e.g. Gidron and Hall, Reference Gidron and Hall2017) and nostalgic deprivation (e.g. Mudde and Rovira Kaltwasser, Reference Mudde and Kaltwasser2018).

Even more, the electorates of populist forces are not as uniform as the ‘losers of modernization’ thesis would lead us to believe. Rooduijn (Reference Rooduijn2018) demonstrates there is no such thing as the populist voter in contemporary Europe, since those who back populist parties are not always male, lower educated, older, and unemployed, nor do they share opinions on issues such as European integration, immigration, and redistribution. Similarly, Rooduijn et al. (Reference Rooduijn, Burgoon, Elsas and van de Werfhorst2017) and Rooduijn and Burgoon (Reference Rooduijn and Burgoon2018) find that populist electorates across European democracies are quite diverse in terms of their socioeconomic status. In addition, the broader literature highlights a clear distinction between left-wing and right-wing populist electorates. The former typically attracts young individuals with secondary and high education living in urban areas (Rodríguez-Teruel et al., Reference Rodríguez-Teruel, Barrio and Barberà2016; Ramiro and Gomez, Reference Ramiro and Gomez2017), whereas the latter typically engages lower educated men with a working-class background (Arzheimer, Reference Arzheimer2009, Reference Arzheimer, Backes and Moreau2011; van der Brug and van Spanje Reference van der Brug and Spanje2009; Rydgren Reference Rydgren2012). Van Hauwaert and van Kessel (Reference Van Hauwaert and Kessel2018) further confirm that left-wing and right-wing populist forces across Europe draw support from different constituencies.

It is, thus, not surprising that populist electorates and supporters show quite some heterogeneity in terms of their sociodemographic features. Yet, considering the broader populist potential, the question remains whether this also translates to populist citizens. That is, can we identify populist citizens by a shared set of characteristics? For instance, Elchardus and Spruyt (Reference Elchardus and Spruyt2016) suggest that populist citizens in Flanders tend to be lower educated (and male). Van Hauwaert et al. (Reference Van Hauwaert, Schimpf and Dandoy2019b) confirm this for regional populist demand across Europe and additionally highlight the interrelation with deteriorating economic development, stronger regional identities, and increased distance from the capital. We wonder, however, to what extent these observations are generalizable to other countries, even across world regions. For the moment, there is no such indication and we have no conclusive reasons to expect that populist citizens share any specific sociodemographics across countries and between continents (except maybe lower educational attainment).

The political profile of populist citizens

The ideational interpretation of populism relies on the construction of a morally homogeneous but sociologically heterogeneous understanding of ‘the people’. That means populist citizens are not inevitably leftist or rightist and their ideological affinities are not path-dependent. Of course, populist forces almost always appear attached to other sets of ideas (the host ideology) that are crucial for mobilizing larger sections of the population (Mudde, Reference Mudde2004; Stanley, Reference Stanley2008).Footnote ³ For instance, inclusionary left-wing forces usually combine populism with socialism to depict ‘the pure people’ as the socioeconomic underdogs, while exclusionary right-wing actors typically combine populism with nativism to portray ‘the pure people’ as the only ones that should make up the nation-state (Mudde and Rovira Kaltwasser, Reference Mudde and Kaltwasser2013). Therefore, in terms of ideology or policy positions, we expect populist citizens to be rather heterogeneous, even when populist forces dominate only one side of the political spectrum.

The populist set of ideas is an expression of citizens’ discontent with the establishment because of the latter’s unwillingness to take into account ‘the silent majority’. In effect, the presence and activation of populist ideas at the mass level are related to the growing tension between responsiveness and responsibility: the more traditional political actors behave responsibly toward the international markets and supranational institutions, the less they can respond to citizens (Mair, Reference Mair2009). Citizens perceive these conditions, base their experiences on them, and formulate interests, which – in turn – guide their political behavior. This is not a minor point, because it brings to light that those who support populist ideas do not only challenge the current state of affairs, and therefore ‘the establishment’, but are also interested in the political situation and engage in discussions about the problems faced by society. While recent generations are more critical and unhappy with democratic functioning, and therefore have a higher populist potential, Dalton (Reference Dalton2013) argues they remain strongly interested in and well-informed about politics. Van Hauwaert and van Kessel (Reference Van Hauwaert and Kessel2018) conclude that populist individuals should not be thought of as apathetic individuals, but rather as politically attentive citizens. With that in mind, we expect that populist citizens will collectively be interested in politics, perhaps even more than their non-populist peers.

This is emblematic of the active political aptitude of populist citizens. Therefore, we also anticipate an empirical relationship between partisan affinity and populist attitudes. Given that the populist set of ideas helps to give a voice to those who do not feel properly represented and are at odds with ‘the corrupt establishment’, we expect populist citizens to be less likely to identify with or show sympathy for one of the crucial actors in representative democracy: political parties. Roberts (Reference Roberts, Kaltwasser, Taggart, Espejo and Ostiguy2017) argues that populism usually depicts political parties as self-serving entities that act against the will of ‘the pure people’. Following the depolicitization of politics (and the rise of identity politics), mainstream politicians and parties have become increasingly similar, thereby making it easier to portray them as a homogeneous elite, regardless of their ideological profile and sociological background.

Populist individuals, thus, have little motivation to feel affectionate toward parties, let alone show partisan support. Meléndez and Rovira Kaltwasser (Reference Meléndez and Kaltwasser2019) provide a first empirical assessment of this argument by examining the link between political identities and populist attitudes in Chile. Their contribution reveals that citizens who have a positive identification with established political parties tend to reject populism, while those who have a negative identity toward established political parties are coalesced by populist attitudes. As this goes to the heart of populism, we expect populist citizens across countries, who are more critical and – per definition – think of the elites as ‘bad’, to be less likely to position themselves close to or be sympathetic with a party.

The democratic profile of populist citizens

To a great extent, the growing interest in populism relates to the debate about its consequences for democracy. On the one hand, populism can be a democratic threat, since it supports a dualistic worldview that promotes polarization and the moralization of the political debate. Müller (Reference Müller2016: 3–4), for instance, maintains that ‘the idea of single, homogenous, authentic people […] is a dangerous fantasy, because populists do not just thrive on conflict and encourage polarization; they also treat their political opponents as “enemies of the people” and seek to exclude them altogether’. On the other hand, populism can also be a democratic corrective, since it helps to give voice to excluded sectors and construct a common identity against the establishment. For example, Mouffe (Reference Mouffe2018: 5) claims that (left-wing) populism, ‘understood as a discursive strategy of construction of the political frontier between “the people” and “the oligarchy”, constitutes, in the present conjuncture, the type of politics needed to recover and deepen democracy’.

In line with this, an extensive scholarship argues and shows that the populist set of ideas is not hostile to democracy per se but is rather at odds with the existence of unelected actors and institutions that can make decisions against the will of ‘the pure people’ (Mudde and Rovira Kaltwasser Reference Mudde and Kaltwasser2012; Huber and Schimpf Reference Huber and Schimpf2016a, b, Reference Huber, Schimpf, Heinisch, Holtz-Bacha and Mazzoleni2017; Huber and Ruth Reference Huber and Ruth2017). After all, the very notion of democracy means that political power comes from the ‘demos’, implying that the ultimate political authority is vested in the people and not in unelected bodies composed of experts. To paraphrase the work of Canovan (Reference Canovan1999: 10), populism is about enacting the redemptive side of democracy: the mobilization of popular enthusiasm around the idea that ‘the people are the only source of legitimate authority, and salvation is promised as when they take charge of their own lives’.

When existing political forces and institutions fail to represent citizens’ interests, the corresponding gap between ‘what is’ and ‘what should be’ the will of the people can strain core democratic tenets and put representative principles under stress. This is in line with the functionalist account of democracy conceptualized by Easton (Reference Easton1965) and the conflict Mair (Reference Mair2009) saw between responsive and responsible government. In this light, citizens employ populist ideas to express their discontent with the existing state of affairs and their willingness to repoliticize problems that are – intentionally or unintentionally – not being addressed by the establishment (Rovira Kaltwasser, Reference Rovira Kaltwasser2014). Citizens who adhere to these ideas typically hold that the democratic system is working improperly as ‘the people’ are not being heard and have the impression that ‘the elites’ only care about themselves (Hawkins et al., Reference Hawkins, Carlin, Littvay and Kaltwasser2018a).

Based on this understanding of populism, we argue that populist citizens are not necessarily authoritarians. Rather, populist citizens are more likely to be democrats who feel slighted by democratic functioning and the lack of (accurate) representation. They are supportive of democratic principles, but largely dissatisfied with their practical implementation. As Mudde (Reference Mudde2004: 562) states, ‘[…] after years of reading and hearing about dysfunctional national and supranational democracies, more and more people have become both sensitised to the problem, and convinced that things can and should be better’. Such ‘critical citizens’ can be a resource for the improvement of democratic processes and structures (Klingemann, Reference Klingemann and Norris1999; Dahlberg et al., Reference Dahlberg, Linde and Holmberg2015). Alternatively, such ‘disillusioned citizens’ can also undermine governmental effectiveness, weaken institutions, erode faith in democracy, and lead to crisis and breakdowns (Stoker, Reference Stoker2006). In line with the threat/corrective interpretation of the populism vs. democracy relationship, that means such ‘dissatisfied democrats’ (to use a more value-neutral term) can either challenge democratic stability or promote democratic reform. While their presence is undeniable, their relationship with populist attitudes remains unclear.

Both these interpretations are largely theoretical, without much empirical support. Vehrkamp and Wratil (Reference Vehrkamp and Wratil2017) provide an initial account. They empirically demonstrate that German populist citizens ‘are not enemies of democracy, but are rather disappointed democrats. The populist criticism of democracy is by no means a radical criticism of the system itself. […] Populists in Germany are certainly much more dissatisfied with the functioning of democracy, but they have not – or at least not yet – turned their backs on democracy as a political system’ (Reference Vehrkamp and Wratil2017: 25). We have similar expectations across countries. Increasing gaps in mass–elite congruence and democratic representation provide a fertile breeding ground for the proliferation of populist ideas at the mass level regardless of the effectiveness of the democratic system, the economic situation, and/or the electoral presence of populist forces.

In sum, recent scholarship examines a wide variety of characteristics that we might attribute to populist citizens, yet the patterns of variance between countries and across world regions remain – at best – underdeveloped. We anticipate the sociodemographic profile of populist citizens to be rather unique to the context and the situation, and therefore relatively idiosyncratic. We also expect high levels of heterogeneity in the political profile of populist citizens because political identifiers are typically context-specific. We, most importantly, expect extensive homogeneity in the democratic profile of populist citizens between countries, regardless of world region, because populism and democracy are what we can refer to as ‘necessarily accompanying factors’. In the follow sections, we examine these expectations empirically.

The dependent variable: Populist citizens

We harmonize eight separate items (see Table 1) that have previously been used to measure populist attitudes (e.g. Van Hauwaert and van Kessel, Reference Van Hauwaert and Kessel2018).Footnote ⁴ By selecting these eight items, we tap into the three dimensions of the ideational approach to populism, namely the extent to which citizens are of the opinion that (a) ‘the people’ is a virtuous and homogeneous entity, (b) portray themselves at odds with ‘the elite’, and (c) advocate popular sovereignty set against elitist rule. Across all countries, individual items consist of 5-point Likert scales, with higher values indicating higher levels of populist attitudes (cross-country Cronbach’s alpha = 0.83). We estimate a latent populist construct through item response theory (IRT). This particular technique has a number of advantages over the more traditional estimation methods.Footnote ⁵ For each step of the empirical analysis, we estimate the populism scale across the set of included countries.Footnote ⁶

Table 1. Question wording for populist items

Note: We include the country-specific wordings of the populist items in Section B of the supplementary materials.

Independent variables and predictors

We examine the intrinsic characteristics of populist citizens across different countries through three specific profiles, namely sociodemographic, political, and democratic. For each profile, we use a set of variables. Considering we harmonize several datasets, we standardize the indicators with different question wordings or answer categories between countries.Footnote ⁷

First, we include five traditional variables to gauge the typical sociodemographic profile of populist citizens, namely gender (0 = female), age, age squared, educational attainment, capital region (0 = no), and main professional activity. The original education variables are recoded into three categories: no secondary education (reference category), completed secondary education, and completed university. The capital region variable refers to the broader (and available) administrative region that includes the capital city of each country. The original ‘main professional activity’ variable is recoded into five categories: employed (reference category), studying, unemployed, housework, and retired. These variables are available for all countries.

We use three variables to provide insights into the political profile of populist citizens. First, we use a standardized ‘left-right self-identification’ variable as a summary item of an individual’s political positioning. Second, we include a standardized ‘political interest’ variable (higher values = more political interest). Third, we gauge respondents’ ‘partisan affinity’ with a dummy variable (0 = no affinity).Footnote ⁸ The availability of these variables differs between countries, so we model each variable separately.

Finally, we use two separate variables to tap into the shared democratic profile of populist citizens across countries. First, we use a standardized version of the ‘satisfaction with democracy’ variable (higher values = more satisfied). Second, we complement this with the so-called ‘Churchillean democracy question’, that is, a standardized variable asking whether democracy is better than any other form of government (higher values = stronger agreement). Both tools are simultaneously available in the nine European countries, Turkey, and Chile.

Exploring patterns of variance: The characteristics of populist citizens

We further explore the variation from Figure 1 and scrutinize the populist demand and its origins by conducting a set of multivariate analyses that examine the different profiles we theorized above. In a first step, we only probe sociodemographic variables to describe typical populist citizens and acquire insights into their sociological characteristics. Table 2 presents the pooled results for all countries (column 1), as well as the pooled results for the European and Latin American clusters separately (columns 2 and 3, respectively).Footnote ⁹

Table 2. The sociodemographic profile of populist citizens – pooled models

Note: Standard errors in parentheses, * P < 0.05. All models include country-fixed effects.

The most general model in Table 2 suggests that – across world regions – populist citizens tend to be male, older (although this tends to be less prolific as respondents get older), lower educated, not in the capital region, and unemployed. To a certain extent, this corresponds to recent literature suggesting that populist potential is prominent among ‘modernization losers’ (Goodhart, Reference Goodhart2017). It means that if we focus just on the average for all countries and without undertaking a more fine-grained analysis of regional differences, this generic sociodemographic profile describes more populist citizens.

Yet, given that our collection of countries is remarkably diverse in terms of sociodemographic composition, we are of the opinion that a more detailed analysis is in order to obtain an accurate picture. Therefore, to examine how homogenous – and robust – these results really are, we assess what unites and differentiates populist citizens in Europe and Latin America in the second and third columns of Table 2, respectively.Footnote ¹⁰ The pooled regression for Europe suggests the previously highlighted profile applies to populist citizens in Europe (Spruyt et al. Reference Spruyt, Keppens and Droogenbroeck2016; Castanho Silva et al. Reference Castanho Silva, Vegetti and Littvay2017; Van Hauwaert and van Kessel, Reference Van Hauwaert and Kessel2018; Stanley, Reference Stanley2019). From both a theoretical and empirical perspective, this is largely unsurprising, considering that most studies of populist potential only focus on European countries.

In Latin America, there appears much more heterogeneity in terms of what sociodemographics describe populist citizens. We only find evidence that populists typically live outside the capital region and tend to be retired. This is equally unsurprising as populist leaders and their corresponding regimes usually have been able to mobilize long-standing resentments in rural populations and the non-cosmopolitan countryside against urban elites (Canovan, Reference Canovan1999; Taggart, Reference Taggart2000). This type of conflict is not new in Latin America, a region characterized by large socioeconomic disparities that, in turn, provide a fertile breeding ground for widespread resentment of the ruling elites. This also explains the lack of support for a relationship between (lower levels of) education and populist attitudes. Education-based categories are not necessarily politicized status markers across Latin America, rendering it plausible that educational differences are not as pronounced when compared to the European context.

When comparing the sociodemographic profiles of European and Latin American populist citizens, we notice that living outside a country’s capital region is the only shared identifier of populism – or rather, populists – across world regions. This, of course, does not mean populists only live in rural areas or are per definition less cultured and provincial. Our finding merely highlights it is more likely that, on average and across world regions, living outside the capital region serves as an identifier of populist citizens. Living outside capital regions renders citizens more prone to populist proclivities, something that reinforces the argument that cosmopolitanism is an antithesis of populism. After all, populist forces usually argue that the elite represents the ideas and interests of privileged sectors of society, that is, the so-called ‘high culture’, which more often than not disdain the values and opinions of the ‘low culture’ (Ostiguy, Reference Ostiguy, Kaltwasser, Taggart, Espejo and Ostiguy2017).

Building on this, we turn to the political profile of populist citizens. We examine political positioning, political interest, and partisan affinity in two steps. First, we provide pooled models for these separate variables across world regions. Second, we specify the same three models for Europe and Latin America separately. Like in Table 2, such an approach provides us with important insights into how united or divided the political features of populist citizens are between and across world regions.

The pooled models from Table 3 indicate that populist citizens tend to identify more with the left, be more politically interested than the average citizen, and not sympathize with any parties (regardless of the existence of populist parties). While these latter two results are not surprising considering some of the literature’s recent findings concerning political identifiers of populist potential (Van Hauwaert and van Kessel, Reference Van Hauwaert and Kessel2018; Meléndez and Rovira Kaltwasser, Reference Meléndez and Kaltwasser2019), the former is likely a construct of aggregation, as classic literature firmly posits that populism is not necessarily left or right (Mudde and Rovira Kaltwasser, Reference Mudde and Kaltwasser2012). The potential interaction of these political variables with contextual factors is something to consider. With that in mind, we further disaggregate the three models according to world region.Footnote ¹¹ Here, we notice the political profile of populist citizens is quite distinct within each world region. At least with the data available to us, this inhibits the creation of a composite political profile of populist citizens. Nonetheless, we can highlight a number of interesting findings.

Table 3. The political profile of populist citizens – pooled models across world regions

Note: Standard errors in parentheses, * P < 0.05. All multicountry models include country-fixed effects. All pooled models include the nine European countries and Turkey. Pooled models 2 and 4 also include all five Latin American countries. Pooled model 3 also includes Costa Rica. All models include the sociodemographic variables specified in Table 2. We refer to section F of the supplementary materials for full models.

Table 3 suggests populist citizens across Europe tend to be more interested in politics and unsympathetic to political parties. Neither of these observations is surprising and they remain largely in line with recent findings (Webb, Reference Webb2013; Dahlberg et al., Reference Dahlberg, Linde and Holmberg2015; Bowler et al., Reference Bowler, Denemark, Donovan and McDonnell2017; Van Hauwaert and van Kessel, Reference Van Hauwaert and Kessel2018) The data does not provide any conclusive evidence regarding political interest and party apathy of populist citizens in Latin America. Both in Europe and Latin America, model 2 finds populist citizens – on average – tend to identify with the left side of a one-dimensional ideological spectrum. We remain cautious interpreting this finding (and drawing inferences from it), as populism is – by definition – neither left-wing nor right-wing in nature. It would be overzealous to interpret these results as most citizens with populist affinities being leftist. In line with the European populism literature, we believe this finding illustrates country-level heterogeneity (this is further substantiated when we look at country-specific results).Footnote ¹² This is in line with recent findings from Latin America (Andreadis et al., Reference Andreadis, Hawkins, Llamazares, Singer, Hawkins, Carlin, Littvay and Kaltwasser2018; Bornschier, Reference Bornschier, Hawkins, Carlin, Littvay and Kaltwasser2018), as well as previous scholarship on the correspondence between the exhaustion of the conservative modernization project and the rise of radical leftist populism in Latin America (Weyland et al., Reference Weyland, Madrid and Hunter2010; Levitsky and Roberts, Reference Levitsky and Roberts2011).

As a final step, we examine the democratic profile of populist citizens by exploring the role of two specific variables, namely democratic satisfaction and democratic support. Table 4 provides pooled and country-specific results for both indicators.

Table 4. The democratic profile of populist citizens

Note: Standard errors in parentheses, * P < 0.05. All models include country-fixed effects and the sociodemographic variables specified in Table 2. We refer to section F of the supplementary materials for full models.

First and foremost, we notice that the direction and significance of both democracy coefficients remain stable throughout all models. In essence, this indicates an important and unseen degree of homogeneity in the democratic profiles of populist citizens, both between world regions and across countries. The democratic satisfaction coefficient is systematically negative throughout the different models, meaning populist citizens tend to be more dissatisfied with democracy – regardless of their national context. The democratic support coefficient is systematically positive, which suggests that populist citizens are more likely to prefer democracy, despite all its flaws, to any other form of government (such as authoritarianism).Footnote ¹³

The two indicators together highlight that populist citizens across countries adhere to democracy as a political system but are disgruntled with how the democratic regime functions in their countries. This confirms that populism and democracy are not inevitable opposites (Mudde and Rovira Kaltwasser, Reference Mudde and Kaltwasser2012; Rovira Kaltwasser, Reference Rovira Kaltwasser2012) and, to some extent, substantiates populism as a ‘pathological normalcy’, rather than a ‘normal pathology’ (Betz, Reference Betz1994; Mudde, Reference Mudde2004, Reference Mudde2010). It further shows that populist citizens cannot be reduced to alienated voters, authoritarians, or anti-establishment electorates. Rather, those who support populist ideas are more accurately described as ‘dissatisfied democrats’.

These results reinforce the argument that the gap between democratic ideals and their practical implementation fosters the rise of populist ideas. Previous scholarship describes this difference between what the principal (the citizen) wants and what the agent (the government) delivers as ‘agency loss’ or an inevitable level of incongruence (Grossman and Hart, Reference Grossman and Hart1983). In this case, however, considerable agency loss could indicate that populist citizens reject the notion of Schumpeterian democracy, namely democracy as a division of labor where citizens elect representatives and subsequently trust them to be responsible (Mair, Reference Mair2009).

The resulting democratic deficit, by consequence, indicates that citizens clearly want to express their discomfort with the existing state of affairs, exercise their civic duty, and actively impact decision-making. We, thus, expect there to be a positive correlation between the (perceived) democratic deficit and populist attitudes. To further examine this, we calculate the democratic deficit by subtracting the democratic satisfaction score from the democratic support score.Footnote ¹⁴ We interact the subsequent variable with country dummies and then regress this on populist attitudes. Figure 2 plots the average marginal effects of these interactions.

Note: Point estimates with 95% confidence intervals. The underlying models include the sociodemographic variables specified in Table 2. We refer to Section F of the supplementary materials for full models.

Figure 2. Average marginal effects of democratic deficit on populist attitudes.

As expected, Figure 2 indicates the effect of the democratic deficit variable is consistently significant and positive. This means citizens tend to be more populist when there is an increasing gap between pragmatic and normative opinions of democracy. We, thus, find initial evidence that populist citizens are not per definition looking to overthrow democracy, but actually favor some form of overhaul within the democratic framework. Even though the analysis relies on fewer countries than the previous models (due to the availability of the democracy items), we are confident the results translate beyond these contexts, as the countries included in this step of the analysis range from Chile to Turkey, as well as various countries covering a broad geographical spectrum within Europe.