Introduction

The ongoing rise of contemporary ‘new populism’ (Taggart 2000) has transformed not only social and political systems but also political geography in Europe. Nevertheless, relevant accounts of the geography of populism and political discontent or resentment (Köppen et al. 2021; Bourdin and Torre 2023; de Lange et al 2023) have started circulating only in the last few years, especially after Trump’s election and Brexit, which were driven by peripheral, rural and less densely populated areas (Jennings and Stoker 2019). This means that the literature on the ‘places of populism’ has flourished with a delay compared to research on populist parties and voters, which has mostly focused on national-level explanations, glossing over the relevance of places. So much so that in a 2019 editorial of ‘Political Geography’ one could read that ‘there is no substantive debate about the place of populism in geography, how populism should be spatially thought through, or how populist electoral politics should be theorized in relation to other long-standing discussions in our discipline’ (Lizotte 2019, p. 71).

However, following the renewed heed to places and local contexts in studies of populist and radical right phenomena (Fitzgerald 2018; Dvořák et al. 2022; Harteveld et al. 2022; Patana 2022; Arzheimer and Bernemann 2023), there is now substantial acceptance of the popular thesis of the places ‘left-behind’ (Gordon 2018; McKay 2019) or ‘that don’t matter’ (Rodríguez-Pose 2018; McKay et al. 2021). This influential line of argument suggests that the surge of populist parties symbolizes the retaliation of lagging towns and regions, which feel abandoned by policymakers after socio-economic changes brought about by globalization, de-industrialization and post-industrialization. The polar opposite of such declining ‘places of populism’ would be large and thriving cities, which are conceived of as insensitive to populist appeals. Therefore, the territorial dimension of the cultural cleavage between populist and pluralist–libertarian orientations (Norris and Inglehart 2019, p. 51) would simply pit dynamic and booming cities against struggling rural and small town areas.

Although this thesis is appealing and overall valid, by accepting it ‘blindly’ we run the risk of falling into determinism or providing a limited explanation. In fact, the ‘left-behind’ thesis seems to ignore, first, that populist (radical right) parties are strong in some rural areas but not in others, and the same applies to cities (Harteveld et al. 2022); second, that there can be substantial variation in political attitudes at the within-city level too, as populist/radical right pockets may exist in clearly ‘not-left-behind’ metropolises (Huijsmans et al. 2021, p. 11; Essletzbichler and Forcher 2022; Crulli and Pinto 2023). Furthermore, as a consequence of urban sprawl, changes in urban planning, and more broadly the developments that have taken place in the transition from industrial to digital societies, the same division between ‘cities’ and ‘non-cities’ has lost much of its meaning (Simon 2008). Hybrid spaces that can hardly be classified as either ‘cities’ or ‘countryside’ now exist inside major metropolises (Damon et al. 2016). Political divides have emerged within large metropolitan areas (Sellers et al. 2013; Crulli 2022, 2023) that actually comprise suburban (or peri-urban) and rural places too, in addition to proper urban environments (Ströbele 2017). Therefore, focusing solely on the simple division between thriving ‘cities’ and lagging ‘non-cities’ may be improper.

Moreover, even by limiting our scope to the common features emerging from different interpretations of the concept (Rovira Kaltwasser et al. 2017), populism remains a wide umbrella, under which different actors and ideas can be found. Populism involves an appeal to ‘the (will of the) people’, opposition to ‘the establishment’ or the ‘élites’, and a rejection of pluralism (Mudde 2004). But beyond this ‘thin’ ideological core, populism comes up in a variety of shapes, including left-, right-wing, and more ‘ambiguous’ or ‘valence’ ones (Zulianello 2020; Taggart and Pirro 2021). Hence, we can expect different populisms to be successful in different settings. The radical right version of populism, represented by populist radical right (PRR) parties combining populism with nativism and authoritarianism (Mudde 2007), is the most prevalent in Europe (Mudde and Rovira Kaltwasser 2013; Crulli and Viviani 2022; Rooduijn et al. 2023). Therefore, singling out which of its ideological tenets are more strongly associated with geographical divides is paramount. So is understanding whether PRR voters are more geographically condensed in comparison with the populist electorate at large.

Against this backdrop, the following research question still needs consideration: [RQ1] Is contemporary European populist and radical right politics stronger in (inner) cities, in the suburbs, or in rural territories? This issue becomes even more important from a historical perspective. Whereas classic cases of populism in the nineteenth and twentieth centuries were clearly associated with either urban or—more often—rural milieus (Canovan 1981; Taggart 2000; Strijker et al. 2015), contemporary European populist parties do not normally seek the vote of specific social classes residing in urban, suburban, or rural environments.Footnote 1 Still, their electoral outcomes appear to be geographically clustered and heterogeneous, which makes the puzzle of the geography of populism intriguing and complex. Furthermore, answering this research question is critical when it comes to populist and radical right attitudes. Except for some thorough single case studies (Arzheimer and Bernemann 2023; Huijsmans 2023),Footnote 2 the strand of research on this topic has not gone much beyond demonstrating that living farther away from the country’s capital (Van Hauwaert et al. 2019; Rovira Kaltwasser and Van Hauwaert 2020) or in disadvantaged peripheral regions (Dvořák et al. 2022) predicts stronger populist attitudes. Therefore, there is still a lack of nuanced territorial analyses.

From such broader research question [RQ1] stem more specific, albeit not less relevant, ones: [RQ2] Are there differences in the way populism and its sub-type on the radical right are spread across European cities, suburbs, and countryside? [RQ3] Does urban/suburban/rural residence matter for populist and radical right orientations on its own, or is it simply a proxy for the extent to which one’s region is ‘left-behind’?

The paper addresses all these questions by resorting to the last round of the European Social Survey (ESS), conducted between 2020 and 2022. Through this dataset, I aim to gauge whether living in different places—big cities, suburbs or outskirts, towns or small cities, villages, and countryside—predicts different levels of support for populist and radical right parties and ideas, even controlling for a wide battery of sociodemographic, political, and contextual variables. Although researchers have used previous ESS waves to investigate the urban–rural divide in the political attitudes of European citizens (Maxwell 2019; Kenny and Luca 2021), at the time of writing, this is the first comprehensive analysis of the relationship between place of residence and both populist and radical right voting patterns and attitudes of contemporary (i.e. 2020s’) Europeans.

In the remainder of the paper, I first provide a historical–geographical outline of how populism was and appears to be connected to rural, urban, or suburban settings. This helps to better spell out the research expectations. Next, I illustrate the research design, discussing the data and measures used. I then present the empirical investigation, before concluding with a discussion of how the paper’s findings fit into and contribute to the current strand of research on populism and its geography. By emphasizing that the urban–rural divide is primarily linked to thick ideological components—authoritarianism and nativism—rather than the thin populist underpinning, the paper invites comparativists to pay greater attention to the radical right, rather than focusing on populism itself.

Populism in rural, urban, and suburban areas: a historical–geographical overview

The first acknowledged historical examples of populist movements are found outside Europe and are closely tied to rural areas. Both the Narodnichestvo in 1860–1870s Russia and the People’s Party in 1890s USA were forms of rural or agrarian populism, albeit very different ones (Canovan 1981; Taggart 2000). Narodnichestvo was the ideology of a Russian group of intellectuals—the Narodniks, usually translated as ‘Populists’—who had a romantic view of the lifestyle of Russian peasants and wanted them to rebel against the tsarist regime. Paradoxically, then, this early populist phenomenon was in fact ruled by an intellectual élite. The US People’s Party, instead, was a mass and bottom-up force, representing the political platform for the radical claims of farmers in the South and West of the USA. These two different historical cases highlight that populism of the origins was a sort of radicalism in rural areas. The intertwining of populism and rural politics has historically been so strong that Margaret Canovan, in her pioneering book on the topic, distinguished between two main types of populism only: ‘political’ and ‘agrarian’ populism (Canovan 1981, p. 13).

Connections between (right-wing) populism and the rural world appear strong today as well, especially after Brexit and Trump’s win, which have drawn public attention to the renewed urban–rural conflict (Jennings and Stoker 2019). So much so that, at the end of 2016, the Guardian claimed that Brexit and Trump were evidence that the ‘power has leaked from cities to the countryside’, because ‘cities may dominate our culture, but a backlash against liberal values and multiculturalism has been led by rural and small-town voters’ (Beckett 2016). This awareness has also grown in academic debate. The possible relationships between rural protest groups and populist parties have been analysed in an edited volume by Strijker et al. (2015). More recently, Scoones and colleagues (2018) called for a new research agenda investigating the rural roots of authoritarian populism around the world; how rural areas shape and are shaped by it, and what alternatives may be put into place to counter it. Still, true cases of agrarian/rural populist parties have been extremely rare in post-WWII Europe (De Lange and Rooduijn 2015).

Anyway, populism has not remained confined to the countryside. Before the current rise of populist parties—especially ‘radical right’ (Mudde 2007) or ‘exclusionary’ ones (Mudde and Rovira Kaltwasser 2013)—in Europe, populism marked twentieth-century Latin America. When used to describe Latin American politics of the last century, populism ‘refers less often to agrarian movements than to urban political parties which are either headed by a charismatic leader or are catch-all in their support’ (Canovan 1981, p. 138; see also Strijker et al. 2015, p. 26).

Coming to contemporary Europe, another cornerstone of the pertinent literature, Paul Taggart’s ‘Populism’, stated in its introduction that the ‘new populism’ of European politics is ‘fundamentally urban’ (Taggart 2000, p. 6). Nonetheless, there are both theoretical and empirical reasons to posit that the territorial character of contemporary European populism is bewildering.

The complex geography of contemporary European populism

While detecting the social and territorial roots of classical examples of populism in either rural or urban settings is straightforward, doing so with contemporary European populism is much more complex. I argue that there are three main sources of this complexity (Fig. 1).

  1. (I)

    The first, structural one is that with globalization, de-industrialization and post-industrialization, the rigid distinction between ‘urban’ and ‘rural’ areas has lost much of its relevance. Economic geography, urban sociology, and political ecology have already underlined ‘the growing real-world limitations of traditional concepts of a simple rural–urban dichotomy’ (Simon 2008, p. 167; see also Sellers et al. 2013; Ströbele 2017). Hence, the need to explore urban processes and spaces through new analytical lenses (Dear and Flusty 1998), or to focus on new types of place, rather than traditional constructs such as ‘regions’ (Knutsen 2011), ‘rurality’, or ‘peripherality’ (de Lange et al. 2023). In fact, spaces of a hybrid nature, which are neither city nor countryside, such as suburban and peri-urban zones, have surfaced within large metropolitan areas (Simon 2008; Damon et al. 2016). These structural transformations have made the classical urban–rural cleavage (Lipset and Rokkan 1967) much more blurred, not only in economic but also in sociological and political terms. In this regard, some authors believe that the mix of proximity and separateness of peri-urban fringes from proper urban centres can foster resentment towards the latter, and help to explain the strength of some populist movements and parties in such areas (Bourdin and Torre 2023). In the same vein, it has been suggested that suburban dwellers seem to support populist (radical right) parties to ‘reclaim urban space for daily use and as a defensive strategy in view of metropolitan change’ (van Gent et al. 2014, p. 1775).

  2. (II)

    The second source of complexity is related to the socio-territorial bases of populist parties in contemporary Europe, which are strikingly variegated. I mentioned that nineteenth-century American and Russian populisms had little in common except being phenomena of rural areas, whereas twentieth-century Latin American populism was a movement for the urban working classes, associated with urban parties and leaders (Canovan 1981; Taggart 2000; Strijker et al. 2015, p. 26). Contemporary European populism, instead, appears as the political representative of the ‘losers’ of globalization, or those ‘culturally insecure’ people ‘who feel they have suffered transnationalism’ (Hooghe and Marks 2018, p. 115). According to scholars of the influential ‘neo-cleavage theory’ (Marks et al. 2021), contemporary European populism is the expression of people located on the losing side of a new value-based cleavage, pitting beneficiaries against victims of modernization and globalization. Nowadays populism would thus represent a cultural—more than an economic—backlash produced by the ‘excesses’ of post-materialism (Norris and Inglehart 2019). For our purposes, this implies that, although some social classes—e.g. the manual working class (Bornschier and Kriesi 2012)—are overrepresented among the populist electoral base, contemporary European populism does not refer to specific social classes either in cities or in the countryside. In fact, according to De Lange and Rooduijn (2015), out of almost 70 populist parties that existed in Europe over the 1980–2013 period, only two could be defined as ‘agrarian populists’.Footnote 3 Therefore, despite the recent successes of the Dutch Farmer-Citizen Movement (BBB), ‘the age of agrarian populism seems to be over’ in Europe (Strijker et al. 2015, p. 32). To sum up, since the losers of globalization—typical populist voters—are mainly defined in a cultural sense, and populist parties do not appeal exclusively to urban or rural sectors, defining contemporary European populism in itself as either ‘urban’ or ‘rural’ is risky, and eventually inadequate.

  3. (III)

    The third source of complexity is interconnected with the previous one and has to do with populism’s ideological versatility. Although populism in the old continent is predominantly radical right-leaning, populist parties exist across the whole left–right political continuum. While sharing a populist worldview, the thicker ideological contents shaping such worldview are massively different between radical right-wing, left-wing, and valence populism (Zulianello 2020). Should we expect the same drivers for all parties and ideas within the populist spectrum—and thus posit similar associations with place of residence regardless of the specific subset? or should we expect different spatial patterns depending on the thicker ideological elements with which populism is combined? Some clues come from voting results, especially in countries with diverse types of populist parties equally capable of gathering remarkable shares. Two examples are Italy and France. The three Italian populist parties—the two on the radical right, League (Lega) and Brothers of Italy (FdI), and the more ‘chameleonic’ Five star Movement (M5S)—have attracted voters from very diverse territories. While the League has typically been stronger in less densely populated towns and rural areas, and more recently also in the suburbs of big cities (Passarelli and Tuorto 2018; Bazzoli and Lello 2022; Crulli 2022), only in the last election has FdI assumed a similar ‘village-oriented’ profile (Chiaramonte et al. 2022). In contrast, the M5S has gained the most in larger cities, performing worse in more rural municipalities (Levi and Patriarca 2020). Similar remarks apply to the French populist vote. Here there are two very strong populist parties: one on the radical right, the National Front/Rally (FN/RN), and one on the radical left, the France Unbowed (FI) (Rooduijn et al. 2023).Footnote 4 While the FN/RN has found its strongholds in peri-urban and rural environments, the FI has been more popular in urban settlements, especially large urban conurbations. This has given rise to a ‘polymorphic’ geography of populism (Bourdin and Torre 2023, p. 225). Therefore, geographical divides may be tied to differences between varieties of populism. The radical right-wing variant seems to be more prevalent in rural environments compared to populism at large. This might be due to its authoritarianism and nativism. As for the former, PRR parties tend to idealize the past, along with traditional lifestyles and national culture, all of which are typically valued more substantially in the countryside (e.g. Buzogány and Mohamad-Klotzbach 2022). Such a weltanschauung may also lead to calls for safeguarding traditional sectors, and agriculture primarily. In fact, the already mentioned agrarian populist party BBB is considered far-right by the authoritative PopuList (Rooduijn et al. 2023). Likewise telling is that the flare-up of European farmers’ protests against costly measures of climate protection in early 2024 has found the overarching backing of PRR parties. As regards nativism, the emerging strand of research on ‘populism and localism’ suggests that right-wing populists—and not other types—‘discursively localize nativist sentiments through idealistic visions of the local community’ and sometimes employ ‘a romanticized vision of rural life and community’ (Chou et al. 2022, pp. 137; 133).

Fig. 1
figure 1

Populism geography in contemporary Europe: sources of complexity and research expectations

Full size image

Summarizing all these considerations, the following expectations on correlations between domicile and support for populism and radical right politics can be formulated:

(Exp. 1) The rural-urban divide in contemporary European populism is not a ‘continuum’: variations and nuances exist across places falling between ‘urban’ and ‘rural’, such as suburbs or outskirts.

(Exp. 2) The divide between cities and suburban-rural areas is more linked to radical right votes and sentiments (i.e. thick ideological elements) than to thin populism itself.

Research design

I assess the strength of populist and radical right parties and attitudes across urban, suburban, and rural areas through an individual-level analysis relying on the last round of the European Social Survey (ESS), as released in July 2023. Using the 10th wave of the ESS allows us to carry out a very large-N investigation founded on current data from 23 countries all over Europe.

In the ESS, the distinction between types of places is based on respondents’ perception of the area where they live. Respondents were asked ‘Which phrase on this card best describes the area where you live?’, and they were given the following options: (1) ‘A big city’, (2) ‘The suburbs or outskirts of a big city’, (3) ‘A town or a small city’, (4) ‘A country village’, (5) ‘A farm or home in the countryside’Footnote 5 (see Appendix 1 for the frequency distribution). Although people may be inaccurate when describing the type of place they inhabit (e.g. Huijsmans 2023, p. 8), the fact that respondents, not the interviewer, identify the type of residency through their own description should not be understood as a limit. Indeed, not only it has been found that people’s awareness of their daily life contexts impinge upon their political preferences (Mancosu 2019). Above all, recent research has stressed that people’s feelings and evaluations of their local contexts are crucial in determining discontent with politics (McKay 2019), lack of political trust (McKay et al. 2021), a ‘sense of powerlessness to affect politics’ (Boswell et al. 2022), or populist orientations (Arzheimer and Bernemann 2023; Huijsmans 2023).

Dependent variables

The ESS contains questions that can be used as proxies for all three pillars of PRR ideology: populist, nativist (i.e. anti-immigrants), and authoritarian attitudes.Footnote 6

I selected two questions to capture populist attitudes, five questions related to nativism, and three questions regarding authoritarian positions (Table 1). I recoded all these questions, so that higher values reflected higher levels of populism, nativism and authoritarianism. Then, I created three summary indexes of populist, nativist, and authoritarian attitudes, by adding the respective recoded variables, dividing the results by the number of variables, and rescaling the results to a [0, 1] range. Cronbach’s α is 0.71 for the populism index, 0.84 for the nativism index, and 0.61 for the authoritarianism index. These values are acceptable, i.e. they indicate that the items are sufficiently closely related to construct scales from them. However, to further justify the creation of the three indexes, I conducted an exploratory factor analysis, which confirmed the existence of the three expected latent dimensions of populism, nativism, and authoritarianism (Appendix 2). The three indexes serve as the first three dependent variables for our series of country fixed effects regressions (models 1–3).

Table 1 ESS questions employed to create the summary measures of populist, nativist, and authoritarian attitudes
Full size table

The other outcome variables are represented by a dichotomous variable indicating whether the respondent is a populist voter or not (model 4) and another binary variable indicating whether the respondent is a PRR voter or not (model 5). Populist parties, and the subset of PRR parties, were identified through an accurate cross-check of the most popular and reliable classifications of populism in contemporary Europe (Zulianello 2020; Taggart and Pirro 2021; Rooduijn et al. 2023). This led to the exclusion of interviewees from Latvia, Iceland, Ireland, Montenegro, North Macedonia, and Serbia, as these countries are not covered by all these classifications. The final sample included 65 parties in 23 countries (Appendix 3).

Independent variable

I constructed the main predictor for the individual-level analysis using the ESS item called ‘Domicile’, which is aimed to measure the geographical residence of respondents. As a first step, I aggregated self-declared residents of suburbs/outskirts and towns/small cities into the same category, labelled as ‘Towns and Suburbs’. I also aggregated self-declared residents of villages and countryside into the same category, which I dubbed ‘Rural areas’. Therefore, the main predictor for the first empirical part is a categorical variable with three possible outcomes: 1 = big cities; 2 = towns and suburbs; and 3 = rural areas.Footnote 7

As a second step, I employ the original five-category ‘Domicile’ variable as independent, to verify whether further nuances arise by breaking down the urban–rural continuum into more categories that fall between the two extremes. Based on the first identified source of complexity in populism geography (and expectation 1), it is likely that this is the case.

Control variables

I run country fixed effects regression models controlling for several types of factors:

  • Sociodemographic variables, namely gender, age, level of education, occupational status, household’s income;

  • The respondents’ feelings about their household’s income;

  • Left–right self-placement;

  • Other predictors regarding political orientations that may have a strong impact on populist/radical right support, according to previous studies. Specifically: (dis)trust towards the national parliament and (dis)satisfaction with democracy (e.g. Van Hauwaert and van Kessel 2018) (Appendix 1 presents the descriptive statistics of all variables used in the statistical models).

Analysis

Descriptive statistics

I start by exploring the descriptive statistics of the three summary measures to provide an overview of how populist and radical right attitudes correlate with place of residency in contemporary Europe. Figure 2 plots the boxplots and mean scores of the indexes across geographical categories. PRR attitudes are overall stronger in less densely populated areas. However, the gaps are not substantial, and, above all, we must differentiate between ‘pure’ populist attitudes on the one hand, and radical right orientations on the other.

Fig. 2
figure 2

Boxplots and averages of the summary indexes of populist, nativist, and authoritarian attitudes: comparison between self-reported residents of big cities, towns–suburbs, and rural areas. Note: descriptive statistics calculated using the analysis weight (anweight) available in the ESS. Red lines mark the overall means

Full size image

Differences in populist attitudes between big cities and both towns/suburbs and rural areas are not statistically significant. As for nativism, there is a tiny but statistically significant gap between residents of towns/suburbs and big cities (p valueFootnote 8 = 0.002). Instead, residents of rural areas are almost 0.07 points more nativist than big city dwellers (p value < 0.001). These gaps are modest, but still, non-negligible. Clear divides between types of residencies also emerged in terms of authoritarianism. Big city residents are less authoritarian than residents of towns/suburbs and rural areas by 0.026 and 0.05 points, respectively. The t-tests confirm that both differences are statistically significant with a p value < 0.001.

To sum up, populist attitudes are very widespread across contemporary Europe, as indicated by the mean and median of the populism measure being close to 0.8 in each subgroup. Such attitudes are not restricted to specific locations like the fringes of cities or the countryside. On the contrary, urban, suburban, and rural residents hold similar levels of populist attitudes. Therefore, the common belief that populist attitudes are more prevalent in certain places such as urban or rural peripheries needs problematizing. By unpacking the other two pillars of PRR ideology, a different scenario emerges. On the nativism and authoritarianism scales, residents of cities, suburbs, and the countryside all exhibit average scores below the midpoint. This implies that Europeans, irrespective of their residential context, do not hold radical right-leaning views on average. Nevertheless, residents of small towns, suburbs, and rural areas appear to have stronger nativist and authoritarian attitudes than those in big cities.

Regression models

To explore whether these residency-based divides can account for differences in populist and radical right support even when controlling for sociodemographic, political, and contextual variables, I estimate a series of country fixed effects regression models. A first look at the coefficient plots of the models (Fig. 3) reveals that, once again, we must distinguish between populist and radical right attitudes and votes.

Fig. 3
figure 3

Coefficient plots of country fixed effects linear regression models explaining populist, nativist, authoritarian attitudes (models 1–3) and logistic regression models predicting vote for populist and PRR parties (models 4–5). Notes: Regression models run using the analysis weight (anweight) available in the ESS. ‘Big cities’ is the baseline category. 95% Cis displayed. Appendix IV reports the full regression outputs and graphs presenting the results as predictive margins (models 1–3) and predicted probabilities (models 4–5)

Full size image

Living in both towns/suburbs and rural areas predicts statistically significantly higher populist views, but the effect sizes are very small, with coefficients around 0.01 compared to a constant of 0.78 (model 1). The rift between large cities and suburban or rural areas is more apparent when it comes to radical right sentiments. As we may have expected after the descriptive overview, the clearest divide between types of residencies is the one associated with a nativist mindset (model 2). Residing in towns/suburbs or rural areas predicts higher levels of nativist attitudes, with coefficients of 0.024 and 0.042, respectively, against a constant of 0.295. The model explaining nativist attitudes has also the best goodness of fit (adjusted R2 = 0.31). Differences between locations are statistically significant and remarkable in terms of authoritarianism as well. Model 3 estimates higher authoritarian attitudes in towns/suburbs (coefficient of 0.015) and rural areas (coefficient of 0.017) compared to the base category of big cities. Finally, models 4 and 5 suggest that residing in towns/suburbs and even more so rural areas enhances the probability of voting for populist parties slightly, whereas for PRR parties notably.

All these findings can be deemed reliable, as I analysed a very large sample of respondents from 23 countries and controlled for a wide range of covariates. Still, they may be affected by the way I constructed the independent predictor, i.e. by collapsing the five categories of the original ‘Domicile’ variable into the customary categories of ‘city’, ‘towns/suburbs’, ‘countryside’. As the next section emphasizes, more complexity surfaces when disaggregating the main predictor into more places that lie in between the urban–rural dichotomy.

More granular patterns (hidden) in between the urban and the rural

Disaggregating the three categories of the previous independent variable into the five originally provided by the ESS—big city, suburb/outskirt, town/small city, village, and countryside—unveils a more nuanced and fragmented picture.

From the descriptive statistics of the three summary indexes sorted by these five categories, I found confirmation that the highest levels of nativism and authoritarianism are in country villages (Fig. 4). Conversely, there is almost no difference in the endorsement of populist attitudes across domicile types. The second-highest mean score is found in the ‘farm/home in the countryside’ category for the nativism scale, and in towns/small cities for the authoritarianism scale. The most remarkable difference compared with previous statistics is that suburbs/outskirts of big cities have the lowest average score on all three scales. This indicates that aggregating residents of towns and suburbs/outskirts within the same category may have been inadequate.

Fig. 4
figure 4

Boxplots and averages of the summary indexes of populist, nativist, and authoritarian attitudes. Comparison between self-reported residents of big cities, suburbs/outskirts, towns/small cities, villages, and countryside. Note: descriptive statistics calculated using the analysis weight (anweight) available in the ESS. Red lines mark the overall means. (Color figure online)

Full size image

Consequently, I replicated the regression analysis using the predictor with the original five categories (Fig. 5). It turned out that country village is the only residency type significantly associated with higher levels of populist attitudes (model 1). When it comes to populist voters, model 4 suggests that dwellers of both towns/small cities and villages are more likely to vote for populist parties compared to those residing in big cities.

Fig. 5
figure 5

Coefficient plots of country fixed effects linear regression models explaining populist, nativist, authoritarian attitudes (models 1–3) and logistic regression models predicting vote for populist and PRR parties (models 4–5). The five-category ESS ‘domicile’ variable is used as main independent variable. Notes: Regression models run using the analysis weight (anweight) available in the ESS. ‘Big cities’ is the baseline category. 95% Cis displayed. Appendix IV reports the full regression outputs and graphs presenting the results as predictive margins (models 1–3) and predicted probabilities (models 4–5)

Full size image

Turning to radical right attitudes and voters (models 2, 3, 5), the findings are much more in line with the regressions illustrated in the previous section. Both nativist and authoritarian attitudes and the PRR vote become stronger moving out from big cities to suburbs/outskirts and small towns, while living in more remote rural areas predicts the strongest commitment to radical right ideas and parties. Once more, the influence of living in rural places seems decisive above all for nativist beliefs. ‘Country village’ and ‘farm/home in the countryside’ have the highest coefficients among all variables included in model 2, except education level.

Summing up, the distribution of populist orientations is spread out among diverse types of residencies. A continuum running from allegedly less-populist cities to allegedly more-populist urban and rural peripheries is not detectable—which confirms expectation 1. Hence, it is important to differentiate also between towns and suburbs, two locations that I previously merged into the same category. On the other hand, such a continuum more clearly exists in terms of radical right attitudes and votes. Both become stronger the more we move towards less densely populated places—in line with expectation 2.

Purely a matter of places ‘left-behind’?

We answered research questions 1 and 2, by demonstrating that different places have different associations with populist, and much more so, radical right politics, with the latter being stronger in the countryside. However, considering the leitmotif of the places ‘that don’t matter’, we cannot rule out beforehand that rural residence is simply a proxy for regions ‘left-behind’. Hence, we need to scrutinize whether residency impinges upon radical right orientations independently of the economic development of the respective region [RQ3].

To test this, I introduced an additional contextual variable in the models, by taking advantage of the ESS item ‘region’, which specifies the respondent’s region of residence at different NUTS levels depending on the country.Footnote 9 Using this item, I matched individual-level data with Eurostat data on the region’s economic development.Footnote 10 Specifically, I considered the change in regional GDP per capita from 2011 to 2021, expressed in Purchasing Power Standards. I assigned each respondent a value ranging from 1 to 3, depending on whether s/he resides in a region belonging to the first, second, or third tertile of the respective distribution. This allows us to distinguish between dwellers of areas characterized by low, medium, or high GDP per capita growth (see Appendix 5 for the frequencies).

As Fig. 6 bears out, the region’s economic development has a significant influence on authoritarian and nativist positions only. Residing in regions with higher GDP per capita growth predicts substantially weaker radical right sentiments. Even more noteworthy for this paper’s argument is that the correlations between domicile and populist/radical right politics are not altered by the introduction of this additional covariate. This suggests that while regional economic conditions matter for the spread of nativism and authoritarianism, they do not absorb the role of sub-urbanity/rurality.

Fig. 6
figure 6

Coefficient plots of country fixed effects linear regression models explaining populist, nativist, authoritarian attitudes (models 1–3) and logistic regression models predicting vote for populist and PRR parties (models 4–5). A categorical variable measuring GDP per capita growth in the respondents’ region is added as an additional covariate. Notes:Regression models run using the analysis weight (anweight) available in the ESS. ‘Big cities’ is the baseline category 95% Cis displayed. Appendix V reports the full regression outputs

Full size image

As a further step, we can also gauge how radical right attitudes vary between rural areas in stagnant and growing regions (as well as big cities in stagnant and growing regions, suburbs in stagnant and growing regions, etc.). To do so, I ran the same country fixed effects models again, this time by interacting residency with the economic conditions of the respondent’s region.

The interaction term leads to different findings for nativism and authoritarianism. As for nativism, the gap between urban areas in stagnant and thriving regions is much larger than that between rural areas in stagnant and more affluent regions (Fig. 7). Put differently, there is not much difference in terms of nativist attitudes between rural areas in worse off and better off regions, while the lowest levels of anti-immigrant sentiments clearly lie in big cities with medium/high GDP per capita growth. This reinforces the idea that rurality correlates with stronger nativist positions independently of the place’s socio-economic conditions. Results from the model explaining authoritarianism are less solid in this case. The only reliable indication is that both big cities and farms in regions with higher GDP per capita growth are less authoritarian than their worse off counterparts.

Fig. 7
figure 7

Interactions between residency and economic development in the respondent’s region: the effect of GDP per capita growth on radical right attitudes at different residency types Notes: Plots extrapolated from models in Table S11 (Appendix 5)

Full size image

Robustness checks

I tested the validity of this paper’s results by conducting several robustness checks for the main models already discussed. First, I performed the analyses again by running multilevel models with two more country-level predictors. The firs one is a dummy variable indicating whether the respondent’s country is in Western Europe (WE) or in Central–Eastern Europe (CEE). Controlling for this is relevant because CEE countries are characterized by a more unstable structure of political conflict and a different cleavage politics (Hutter and Kriesi 2019). For instance, the urban–rural cleavage has traditionally been more pronounced in CEE (Strijker et al. 2015, p. 27). The second predictor is another dichotomous variable indicating whether a populist party has been in government at least once in the respondent’s country. As can be gleaned by comparing Figs. 5 and 8, different model specifications yielded highly similar conclusions.

Fig. 8
figure 8

Coefficient plots of multilevel linear regression models explaining populist, nativist, authoritarian attitudes (models 1–3) and logistic regression models predicting vote for populist and PRR parties (models 4–5). Notes:Regression models run using the analysis weight (anweight) available in the ESS. ‘Big cities’ is the baseline category. 95% Cis displayed. Appendix VI reports the full regression outputs

Full size image

Secondly, I reran the main models by splitting the sample between WE and CEE (Fig. 9). Nativism has the most consistent relationships with residency across both regions. In terms of attitudes, the most remarkable variation is the strong association between authoritarianism and rural settings in CEE, which is not observed in WE. In terms of voting, we also detect a considerable correlation between country villages and PRR voting in CEE, which is absent in WE.

Fig. 9
figure 9

Coefficient plots of country fixed effects linear regression models explaining populist, nativist, authoritarian attitudes, and logistic regression models predicting vote for populist and PRR parties. Sample split between Western Europe and Central–Eastern Europe. Notes:Regression models run using the analysis weight (anweight) available in the ESS. ‘Big cities’ is the baseline category. 95% Cis displayed. Appendix VI reports the full regression outputs

Full size image

Lastly, I performed several other regression analyses using different operationalizations of the dependent variables. Specifically, I ran ten separate country fixed effects ordered logistic regressions, one for each single item related to either populism, nativism, or authoritarianism (first three columns of Fig. 10). I also constructed two alternative scales to use as dependent variables in other two regression models. The first one combines all radical right beliefs (i.e. five nativism items and three authoritarianism items: α = 0.8), whereas the second one combines all ten items related to PRR attitudes (α = 0.77) (right column of Fig. 10). Among populist attitudes, anti-elitism is more correlated with low-density areas compared to people centrism. No remarkable differences arose in how the five nativism-related items relate to residency. As for authoritarianism, only obedience and respect for authorities and loyalty towards the country’s leader have statistically significant relationships with place of residence. Finally, the models explaining the two summary measures reassert that PRR orientations are stronger in the countryside.

Fig. 10
figure 10

Coefficient plots of country fixed effects ordered logistic regression models predicting attitudes related to populism (1), nativism (2), and authoritarianism (3), and linear regression models explaining summary measures of radical right and PRR attitudes (4). Notes:Regression models run using the analysis weight (anweight) available in the ESS. ‘Big cities’ is the baseline category. 95% Cis displayed. Appendix VI reports the full regression outputs

Full size image

Although interesting variations emerged between European regions and single proxies for either populism, nativism, or authoritarianism, none of these robustness checks altered the overall validity of previously presented findings and arguments.

Discussion and conclusion

Differently from historical cases of populism in the nineteenth and twentieth centuries, contemporary European populist parties do not usually target specific socio-economic classes residing in either urban, suburban, or rural environments. Still, they appeal to grievances of people and places that feel ‘left-behind’ by policymakers and alleged ‘élites’. Knowing this does not mean that the main research question we have addressed here is to be taken for granted: Is contemporary European populist and radical right politics stronger in (inner) cities, in the suburbs, or in rural territories?

In fact, differences in populism between big cities and both suburbs/outskirts and towns are not statistically significant in our main statistical model (Fig. 5). On the other hand, there is a gap between big cities and country villages, but it is not as pronounced as might be assumed based on common sense and journalistic accounts. Whereas previous related research on the link between residency and socio-economic/political values or immigration attitudes found that the urban–rural polarization was best understood as a continuum, ‘running on a gradient from inner cities to suburbs, towns, and the countryside’ (Maxwell 2019; Kenny and Luca 2021, p. 566), this study does not support such an interpretation in terms of populist attitudes nor votes. High levels of populism cut across residency types. However, in purely descriptive terms, ESS respondents in suburbs/outskirts are, on average, the least populist, nativist, and authoritarian compared to those living in other types of places (Fig. 4). Hence, we need to pay more attention to the residential contexts in between the ‘urban’ and the ‘rural’ in order to tease out the role of geography in spreading populist and radical right orientations. In stylized terms, dwellers of a new suburban area next to a metropolis will arguably differ in their voting behaviour from their counterparts in a small town surrounded by rural landscapes. Therefore, it is time to more fully move beyond ‘the often overly simplified urban/non-urban geographical lexicon of populism’ (Chou 2020, p. 1107).

On the other hand, a clearer and greater divide between cities, suburbs, towns, and rural areas emerged in terms of radical right votes and attitudes (i.e. anti-immigrants and authoritarian ones). This result has broader implications for research on populism in general (not only populism geography), as it underscores the need to reflect more carefully upon the drivers of different types of populism. For instance, it has recently been reasserted that ‘in Europe, there is little doubt that nativism is the most fundamental ideological tenet of the Populist Radical Right’ (Rovira Kaltwasser and Taggart 2022, 4). This paper corroborates this claim, by showing that nativist orientations—and, to a lesser extent, authoritarian ones—are much more associated with territorial divides compared with populism per se. In fact, there is a rather clear trend of radical right attitudes and votes increasing steadily as the place of residency becomes less dense and more rural. This is not entirely a matter of places’ composition, or the fact that individuals with specific backgrounds tend to self-select or sort into certain places. Nor is it purely a tale of places ‘left-behind’. Controlling for a wide gamut of individuals’ socio-economic characteristics and the economic development in their region does not alter the positive correlation of suburban and especially rural residence with nativism, authoritarianism, and PRR voting. On the contrary, the difference in nativist beliefs between rural areas in stagnant and growing regions is much narrower than that between big cities in stagnant and growing regions. All this buttresses the notion that there is something in sub-urbanity, and, even more so, rurality that substantially matters for holding anti-immigrant stances. Indeed, nativism is also correlated with domicile types in a very coherent manner across Western and Central–Eastern Europe.

The takeaway is that geographical divides ‘activate’ more through the thick components of populist radical right ideology compared to the thin populist underpinning. Therefore, future studies on the geographical polarization of politics may turn their attention to radical right phenomena, more than to purely populist ones. Drawing out the limitations, this research certainly does not allow one to contend that ‘residence creates attitudes’. What the statistical models have shown is that it is not only the type of people who live in more or less rural areas who ‘make’ them. Nonetheless, establishing a causal and independent effect of residency on attitudes requires panel data that track individuals moving from one area to another (and ask them the reasons behind their mobility). Therefore, a question mark still hangs over the following question: Do individuals develop stronger radical right sentiments because they reside in more remote rural regions? Or do they remain in/move to such areas because they already hold higher nativist and authoritarian attitudes, which make them feel uncomfortable in cosmopolitan big city centres? As this question remains open to scrutiny, addressing the causal mechanisms underlying the relationship between residency and attitudes through innovative survey designs or experiments could represent a fruitful research agenda for political scientists and sociologists.