Where does ethnic concentration matter for populist radical right support? An analysis of geographical scale and the halo effect

Abstract

It is often hypothesised that the share of the population in an area belonging to an ethnic minority group positively influences the support for populist radical right-wing parties among native residents. However, empirical tests of this relationship have yielded mixed results, which may be a result of the wide variety of geographical scales at which ethnic concentration has been measured. Furthermore, it may be that it is the spatial distribution of minorities within the residential area that matters for radical right support, rather than their overall group size. The present study examines these issues by constructing egohoods and halos of varying sizes around respondents' homes. Connecting survey data from the Netherlands Longitudinal Lifecourse Study to detailed geographical data on ethnic concentration, it is found that support for the Dutch PVV is high in areas with low shares of minorities and decreases in areas with higher minority shares, up to a tipping point when minorities make up around 25% of the population. When shares of ethnic minorities become even larger, we tentatively conclude that support for the PVV increases again. This observed U-shaped pattern is consistent across distance-based egohoods ranging in radii from 200 to 5000 m, population-based egohoods with between 4000 and 120000 inhabitants, and administrative neighbourhoods, districts, and municipalities. Additionally, this study found that, in urban areas, native residents of relatively homogenous neighbourhoods whose surrounding area – the ‘halo’ – harbours a pronounced cluster of minority residents are more likely to support the radical right.

Introduction

Populist radical right-wing parties (PRRPs) have won considerable shares of the vote in recent elections all over Europe (e.g. Inglehart & Norris, 2016). The rise of these parties has stimulated research and debate among social scientists who aim to explain the popularity of PRRPs (Golder, 2016; Inglehart & Norris, 2016; Rydgren, 2007), whose ideological core has been defined as a combination of nativism, authoritarianism, and populism (Mudde, 2007). One line of research has focused on how the residential environment of voters affects PRRP support (e.g. Teney, 2012; Van Gent, Jansen, & Smits, 2014; Van Noord, De Koster, & Van der Waal, 2018; Van Wijk, Bolt, & Johnston, 2019). A contextual characteristic that has got particular attention is that of the concentration of ethnic minorities in the residential environment. It is often expected that the share of the population in an area belonging to an ethnic minority group positively influences PRRP support among native residents (e.g. De Blok & Van der Meer, 2018; Green, Sarrasin, Baur, & Fasel, 2016; Kaufmann, 2017; Savelkoul, Laméris, & Tolsma, 2017). However, tests of the relationship between ethnic concentration and PRRP support have yielded mixed results.

One potential explanation of this inconsistency is the wide variety of geographical scales at which this relationship has been examined. Studied localities range in size from countries (e.g. Arzheimer, 2009), to administrative regions (e.g. Green et al., 2016), to municipalities (e.g. Rink, Phalet, & Swyngedouw, 2009), to neighbourhoods (e.g. De Blok & Van der Meer, 2018). It has been recognised that scale may be an important aspect of the relationship between ethnic concentration and PRRP support (Biggs & Knauss, 2012; Bowyer, 2008; Kaufmann & Goodwin, 2018). However, until now almost all studies suffer from the limitation that they examined the impact of ethnic concentration aggregated to one or at best two administratively defined geographical units. This has restricted the inferences that can be made about the important geographical dimension of scale, but also about the impact of the spatial distribution – e.g. segregation – of ethnic groups within and between different geographical units of analysis. So far, there is only limited insight into the question where ethnic concentration matters for PRRP support (cf. Sharkey & Faber, 2014).

Relationships between ethnic concentration and PRRP support have been explained from the theoretical mechanisms of perceived ethnic threat and interethnic contact (e.g. Green et al., 2016; Savelkoul et al., 2017). Residents of areas with high shares of minorities are expected to be more likely to support a PRRP, because the presence of minorities in the residential environment heightens feelings of threat among majority members (e.g. Green et al., 2016; Savelkoul et al., 2017). On the other hand, living in an area with many minorities is expected to increase opportunities for (positive) contact between the majority and minority population. Positive interethnic contact has been hypothesised to reduce outgroup prejudice and as a result decrease PRRP support (e.g. Biggs & Knauss, 2012; Savelkoul et al., 2017). What is often missing from these discussions is an explicit consideration of the geographical scale at which these mechanisms are likely to manifest themselves. As social interactions decay with distance, many daily interactions take place within the local neighbourhood (e.g. Onnela, Arbesman, González, Barabási, & Christakis, 2011). In contrast, most corroborative evidence for the threat mechanism has been found at relatively large scales, such as countries (e.g. Arzheimer, 2009) and large voting districts (e.g. Bowyer, 2008; Ford & Goodwin, 2010; Green et al., 2016). The relative relevance of the contact and threat mechanism may thus depend on the geographical scale under consideration (Biggs & Knauss, 2012; Bowyer, 2008). The potential relevance of geographical scale is supported by the findings of a recent meta-analysis by Kaufmann and Goodwin (2018), which included studies that have examined the relationship between ethnic concentration on the one hand and opposition to immigration and PRRP support on the other. They found that the geographical scale at which ethnic concentration is measured is an important predictor of differences in findings. Ethnic concentration more often increased anti-immigration attitudes at the very small and at larger scales, whereas in geographical units with 5000–10000 inhabitants (e.g. large neighbourhoods) ethnic concentration was more frequently associated with reduced opposition to immigration (Kaufmann & Goodwin, 2018). However, studies in this meta-analysis have examined one or at best two spatial scales at the same time. Furthermore, only a small number of studies have examined the effects of ethnic concentration in very small localities with fewer than 1000 residents.

Most studies on the relationship between ethnic concentration and PRRP support have operationalised ethnic concentration by using administrative areal units. Using administrative units to investigate the link between ethnic concentration and support for PRRPs has benefits, as these units often correspond to socially relevant areas with which people may identify. However, a well-known issue in geographical research is that of the modifiable areal unit problem (MAUP; Openshaw, 1984), which indicates that the estimated impact of contextual characteristics may be influenced by the shape and scale of the aggregation units that are used (Andersson & Malmberg, 2015; see also Kwan, 2012). Administratively defined geographic units are furthermore ill-suited to assess whether effects differ across spatial scales, because there are only a limited number of different administratively defined units within specific countries and administrative units of the same type (e.g. municipalities) may vary widely in both population and area size themselves. To tackle this issue and to examine where, at which geographic scale, ethnic concentration matters, this study defined the local environment as an egohood (Andersson & Malmberg, 2015; Hipp & Boessen, 2013), which is an egocentric neighbourhood around a respondent's home that can be constructed for widely varying sizes (based on area or population size).

In addition to effects of ethnic concentration, threat and contact theory also imply an effect of the spatial distribution of ethnic minority groups. According to what has been labelled as the ‘halo effect’, PRRP support would be higher when ethnic minority shares are higher in the environment surrounding the local residential area than in the local residential area itself (Martig & Bernauer, 2018; Rydgren & Ruth, 2013). Especially these voters would experience threat that cannot be mitigated by contact at the local level. Empirical tests of halo effects are scarce, however, and we are the first to distinguish between halo effects that are produced by a difference in minority group size between the local residential area and the surrounding area (the 'difference hypothesis') and halo effects that are induced by the clustering of minorities in the surrounding environment (the 'clustering hypothesis'). Consistent with our egohood approach, we will examine both general halo effects and clustering effects at a variety of geographical scales.

The focus of this contribution is on the Partij voor de Vrijheid (Party for Freedom; PVV hereafter), a Dutch political party that is generally classified as populist radical right (Bakker et al., 2015; Vossen, 2011). The PVV was established in 2006 by its leader and only member Geert Wilders. The biggest success of the PVV was in the 2010 national elections, when the party won 15.45% of the vote. In subsequent elections, the party again won considerable shares of the vote (10.08% in 2012 and 13.06% in 2017), establishing itself as one of the major players in Dutch politics.

Previous studies on the relationship between ethnic concentration and PRRP support have often been based on ecological analyses of election results (e.g. Bowyer, 2008; Martig & Bernauer, 2018; Rydgren & Ruth, 2013), which introduces well-known risks associated with ecological fallacies (Robinson, 1950). As both party preferences and turnout rates may vary between ethnic groups as well as across localities (e.g. Fieldhouse & Cutts, 2008), ecological studies cannot determine the percentage of PRRP voters among the native population in each geographical unit (Savelkoul et al., 2017). This makes it impossible to distinguish compositional and contextual effects. To empirically examine scale and halo effects of the relationship between ethnic concentration and support for the PVV, we combined individual-level survey data from the Netherlands Longitudinal Lifecourse Study (NELLS) and contextual data on ethnic concentration with detailed geographical resolution obtained from Statistics Netherlands.

To sum up, our contribution focuses on two research puzzles. First, we investigate how the relationship between ethnic concentration and PRRP support varies across geographical scales. Second, we examine how differences in ethnic concentration between the own residential area and surrounding areas influence PRRP support. Our approach builds on previous studies by (1) using individual-level data to overcome risks associated with ecological fallacies; (2) examining the ethnic concentration in egohoods rather than administratively defined units to overcome problems associated with the MAUP (Andersson & Malmberg, 2015); (3) testing ethnic concentration and halo effects at a wide variety of geographical scales, including that of the very small locality; and (4) distinguishing between two specifications of the halo effect, namely the difference hypothesis and the clustering hypothesis .