Navigation – Plan du site

Beyond the Anglo-Saxon Experience: City-Suburban Political Polarization in Brussels, Belgium

La polarisation politique de la périphérie bruxelloise au regard du modèle anglo-saxon
Filip De Maesschalck
p. 319-339


Les conséquences politiques et électorales de la périurbanisation sont récemment réapparues en tant que sujet de recherche dans la communauté universitaire anglo-saxonne. Ces travaux concernent presque exclusivement des agglomérations anglo-saxonnes et cet article veut élargir le champ spatial d’étude en analysant les impacts électoraux et politiques de la périurbanisation à Bruxelles, la plus grande région urbaine en Belgique. Utilisant des données et des enquêtes électorales, l’article montre une polarisation entre la ville et la périphérie. Comme dans les pays anglo-saxons, la ville-centre vote de plus en plus à gauche. Cette ville-centre s’appauvrit si bien qu’elle devient répulsive pour les périurbains. Ils cherchent à défendre leur territoire contre les désagréments réels ou supposés de la ville qui est identifiée aux étrangers et à la criminalité. Toutefois, on ne peut pas assimiler les processus à ceux observés dans les villes anglo-saxonnes. Dans ces dernières, l’indépendance fonctionnelle croissante du territoire périurbain en l’absence de politique de redistribution rend possible une vie séparée de la ville-centre. À Bruxelles, c’est différent. La ville-centre reste le plus important réservoir de main-d’œuvre et de services pour l’ensemble de l’agglomération. En ce sens, Bruxelles apparaît comme représentative de beaucoup de villes de l’Europe de l’Ouest. Toutefois, cela n’a pas abouti à un soutien financier de la couronne périurbaine à la politique sociale. Cela a abouti plutôt à une volonté des périurbains d’utiliser la ville en toute sécurité, sans beaucoup se soucier des problèmes des habitants de la ville-centre. Enfin, l’analyse montre également que les évolutions électorales peuvent être différentes à l’intérieur de la couronne périurbaine quand les histoires politiques sont différentes.

Haut de page

Texte intégral

The american experience: the development of suburban defensive localism

1Academic interest in the relationship between suburbanization and the vote is largely limited to the U.S., and was already present in the 1950s and the 1960s. The debate revolved around the question of whether and why the fast suburbanization of the U.S. induced a shift to the Republican Party, while the cities remained loyal to the Democratic Party. In the following decades, work on the subject is rare, partly because these early studies are not consistent in showing the existence of an independent spatial effect. In recent years, the electoral and political impact of suburbanization has regained some academic interest again in the U.S. [Hodge and Staeheli, 1992; Gainsborough, 2001a]. In particular, a growing divide between the city and the suburbs has been observed, where a conservative and neo-liberal suburban area increasingly contrasts with a left wing urban centre, favouring redistribution on the national scale. Still, the policy aspects of this divide have received more attention in the literature. In this way, suburban secession movements from the central city have been analysed [Keil, 2000 ; Defronzo Hasselhoff, 2002], and, more generally, the possibilities and limitations of regional cooperation in fragmented metropolises [Abbot 1997 ; Gainsborough 2001b].

2The study of Gainsborough (2001a) is the most elaborated research on the topic in the U.S. so far. In a review of the American literature, the author observes that contemporary political science has largely ignored the relationship between modern suburbs and political behaviour. Moreover, social scientists who do argue that the suburbs are politically important not seem to account for the fact that the same conclusion was rejected by political scientists in the 1960s and 1970s. To address this gap, the author explores both historical and contemporary suburban political behaviour in the metropolitan zones of the U.S. It appears that, from the end of the 1980s onwards, party choice and party-identification is consequently influenced by the suburban location after controlling for individual variables. Similarly, in the most recent period, support for federal social government spending is significantly lower among suburbanites than among city dwellers.

3To explain this recent upsurge of explicit suburban political attitudes and behaviour, Gainsborough (2001a) uses the concept of suburban defensive localism, referring to the development of suburban defensive reactions to the impoverished city and even the desire to fence itself off from it. This defensive localism has its sources in two processes. A first one is the crisis of the cities in the 1970s and the 1980s, which has resulted in suburban anti-urban feelings and defensive reactions to it. The cities in this decade came increasingly to embody, in the minds of the suburban dweller, all that was bad. This means that suburban residents do not appear to view their fortunes as being tied to those of nearby city dwellers, and they do not respond to the plight of the city with increased support for programs that might benefit urban residents. On the contrary, where the differences between city and suburban fringe are large, attitudes and behaviour tend to be more polarized. Gainsborough (2001a, p. 95) states :

4"The more the fortunes of the city and surrounding suburbs diverge, the more likely suburban residents are to favour Republican candidates for offices and to oppose spending on federal programs. This finding supports the argument that the distinctive political attitudes of suburbanites are, in part, a reaction against the conditions that exist in the nearby city."

5A second process that explains the recent upsurge of a defensive localism is the increasing self-sufficiency of the suburban fringe. The suburbs are no longer merely residential areas, but have become equally concentrations of jobs, recreation, and services :

6"it is these very changes that allow today’s suburbanites to live their lives insulated from the needs of cities and their residents – and that may have been, then, a necessary ingredient in the emergence of a distinct suburban politics."[Gainsborough, 2001a, p. 19].

Research outside the United States

7Outside the United States, research on the relationship between suburbanization and the vote is scarce. Cox (1968) tried to introduce the topic in England, starting from the observation that almost all the work on electoral differences between suburbs and city was carried out by American researchers in the American context. He shows an independent spatial effect of the suburban living environment on voting behaviour in the London suburban area and suggested further research in England. Despite the following academic discussion about Cox’s study [Taylor, 1969 ; Kasperson, 1969 ; Cox, 1969 ; Biel, 1972] – partly methodological in nature but ultimately dealing with the existence of a suburban effect – the study of the topic has not continued. At the most, it has been mentioned as a potential influence on the vote [Crampton, 1984 ; Johnston, 1986]. Recently, the operation of these processes in Great-Britain has been analysed again by Walks (2005b), who had already explored electoral and political polarization between central city and suburban fringe in the Canadian context [Walks, 2004a ; 2004b ; 2005a]. Walks (2004a, p. 290) states with regard to Canada that
"this research has thus found a pattern of city-suburban polarization similar to that documented in the US, despite the relative health of Canadian inner cities (…). This provided further support for the proposition that there is something inherent in the relationship between suburbanization and political ideology that operates independently of national context or social segregation."

8This also makes the study of these processes relevant outside the mentioned countries. In Australia, there has been some attention for the topic [Forrest et al., 1984 ; Johnston and Forrest, 1985], but outside the Anglo-Saxon world, it has rarely been thoroughly explored. Yet, Ostendorf (1987) observes political polarization between central city and suburban fringe in the urban regions of the Netherlands, which Passchier and van der Wusten (1990) confirmed. Recently, Hoffman-Martinot and Sellers (2005) have brought together a range of studies on metropolitanization and political change in Western Europe. Though it remains unclear whether an independent suburban effect has come into existence, they still conclude that :
"our findings about the consequences of metropolitan change for politics remain more provisional than the (…) other conclusions. Yet the evidence for widespread new cleavages between more leftist central cities and conservative suburban areas is already considerable."

9In addition, somewhat independent from the Anglo-Saxon research tradition, also the French literature starts to focus on the suburban vote, in particular since the presidential elections of 2002. These elections showed a clear contrast between more left-wing cities and suburban zones where the extreme right was very successful. Ravenel et al. (2003) build on earlier work on urban-rural differences to explore these electoral differences, and conclude that a ‘nouveau clivage’ [Ravenel et al., 2003, p. 482] is developing. After controlling for variables such as education and profession, spatial differences continue to exist between left-wing central cities and a suburban environment where the extreme right has mobilized successfully. Suburban inhabitants vote against the real or perceived annoyances (crime, poverty, foreigners) coming from the urban field that they have chosen to leave, and the physical distance to the city supports these negative representations. These processes are confirmed in later studies [Bussi et al., 2004 ; Rivière, 2008], though the latter author stresses the importance of social-economic characteristics explaining the spatial differences in voting. Unlike many US cities, French central cities still show a concentration of highly educated individuals and white collar professions, the suburban fringe concentrating relatively more blue-collar workers.

10Though the French urban socio-spatial structure is not representative of many Western European cities, the limited attention for the topic of urban-suburban electoral and political polarization in continental Western Europe is certainly related to the less pronounced suburbanization in these countries. However, suburbanization in Belgium is well developed in comparison with other Western European countries such as Germany, the Netherlands, and even Great Britain [Kesteloot, 2001]. Suburbanization is the spatial expression of mass consumption, and has been encouraged by government in all these countries. The development of owner-occupied housing outside the cities – offering more space and cheaper land prices – was an element in the expansion of the consumption package, and in addition caused the promotion of other consumption goods, the car being the most essential. In Belgium specifically, suburbanization has since long been encouraged by policy measures not only for economic but also for political motives. The ruling political parties have considered housing construction outside the cities as an excellent countermeasure against socialist mobilizations in the cities. Especially the Catholic party and, after WWII, its successor the Christian-democratic party have been active in this respect [De Decker et al., 2005]. As a consequence, the hypothesis of the existence of an electoral and political divide between city and suburban fringe in the Belgian context is legitimate.

11This suburbanization is, however, mainly residential. Contrary to much U.S. cities, Belgian central cities remain the main provider of jobs and suburban commuting to the central cities is still dominant. The development of these different socio-spatial configurations of cities is elaborated by Kesteloot (2005), who distinguishes between the model of the topological city, characteristic of the U.S., and the dramatic city, representative for the Western European core countries. In the U.S. topological cities, interaction between the urban poor and the suburban dwellers disappears through the functional development of the suburbs. In the dramatic city, interaction remains through the persisting process of commuting. The question arises, then, whether and how electoral and political city-suburban polarization manifests itself without self-sufficiency of the suburbs, which, following Gainsborough (2001a, p. 19), may be ‘a necessary ingredient in the emergence of a distinct suburban politics’. In this paper, polarization between city and suburbs is analysed in the Brussels urban region, which is by far the largest Belgian urban region with almost twice the population of the second largest urban region, Antwerp. Brussels is a strong example of a dramatic city, its central employment and services being quite robust through the city’s role as Belgian and European Capital City. Furthermore, the linguistically, socio-economically, and consequential institutionally complex structure of the Brussels urban region makes it possible to verify the uniformity of the city-suburban divide in different contexts within the same urban region.

12The first part considers the suburbanization of Brussels. The second part focuses on electoral and political polarization. The existence of city-suburban electoral polarization is verified by means of correspondence analysis on aggregate election results. Making use of party platform analysis, it is examined whether the observed developments can be accounted for by the same processes as those suggested in the Anglo-Saxon world. Subsequently, it is shown that not all parts of the suburban area experience similar electoral developments, which are associated with different party political histories and mobilization. Finally, applying multi-level analysis to political attitudes available on the individual level, it is revealed that attitudinal, rather than electoral, city-suburban polarization can exist in a uniform way in the whole of the suburban area. In this way, an independent spatial effect of the suburban living environment can also be verified for the whole of the suburban area.

The suburbanization of Brussels

13Suburbanization processes and electoral developments are analysed in the former province of Brabant (Fig. 1). This area contains the greatest part of the Brussels urban region, as delimited by Van der Haegen et al. (1996), together with the smaller urban region of Louvain. The latter is strongly affected by the Brussels developments and forms an enclave in the Brussels range of influence. This range of influence can be considered to stretch out as far as the commuter zone, containing municipalities that are not considered to be a part of the urban region but where at least 15 % of the working population works in the Brussels agglomeration (being the continuous built-up area that starts in the central city).

14The successive Belgian federalization operations since the 1970s included the establishment of three regions with extended powers: the Dutch-speaking Flemish Region, the French-speaking Walloon region, and the bilingual Brussels Capital Region. The former province of Brabant was situated in three Belgian Regions, which made policy conflicts inevitable. Consequently, in 1995, the former province was split up along the language border to form Flemish Brabant and Walloon Brabant. In the Brussels Capital Region, the provincial powers were assigned to the existing institutions in the region. This former province counts almost 2.4 million inhabitants, nearly one fourth of the Belgian population.

Figure 1. The Brussels region and the former province of Brabant

Figure 1. The Brussels region and the former province of Brabant

Source : Van der Haegen et al., 1996.

15Suburbanization has been very pronounced in this area. The Brussels Capital Region has known a severe population loss ; the opposite is true for the provinces of Flemish and Walloon Brabant. Between 1968 and 1998, the migration balance of the Brussels Capital region was negative, with a loss of nearly 125.000 people or 12 % of its original population ; in Flemish and Walloon Brabant, the migration balance was positive and amounted to almost 230.000 people or 21 % of its original population. Since suburbanization is mainly limited to middle and higher incomes, an income reversal between city and suburban fringe has occurred. In the beginning of the 1960s, the mean income was still highest in the Brussels Capital Region. Now the Brussels Capital region has become very impoverished. The reverse has happened in the rest of Brabant (Fig. 2).

Figure 2. Population and income development in the former province of Brabant

Figure 2. Population and income development in the former province of Brabant

Source : National Statistics Office, 1955 - 2005.

16The income reversal is not always perfectly mirrored by population change. The migration balance in the Brussels Capital Region remained positive until the late 1960s, and this is related to the massive influx of foreign workers. The Belgian internal migration balance was already negative in these years [Jouret, 1972]. These foreign workers were mostly employed in low-wage jobs, and as a consequence the mean income in the Brussels Capital Region was – compared with the provincial mean – already decreasing. Similarly, the positive migration balances in recent years are due to strengthened foreign immigration [de Corte et al., 2003]. However, the Belgian internal migration balance is less negative than before. This is related to gentrification processes. Following Van Criekingen and Decroly (2003), gentrification takes the form of marginal gentrification in cities like Brussels, ranking relatively low in the world city hierarchy. It is a temporary answer of young, non-familial, and highly educated households, mostly of middle class origin, to unstable familial and labour market positions. Indeed, there is a growing time interval between leaving the parental home and the starting on the labour market, on the one hand, and a stable familial and labour market position, on the other hand. And, as Van Criekingen and Decroly (2003, p. 2455) state,
"Living conditions supplied by inner-city neighbourhoods are particularly suited to the specific social reproduction needs of young adults in both familial and professional transitional positions."

17Usually, these groups will leave the city again and will be replaced by new such households. In other words, it is not a phase in a stage model of gentrification, but a separate process. This process of marginal gentrification does not result in strongly rising income levels. Marginal gentrifiers, connected to their labour market positions, mean an influx of cultural capital rather than financial capital. Moreover, these gentrification processes do not signal the end of suburbanization. The crisis years of the 1980s meant a temporary weakening of suburbanization, but the process regained strength later on (Fig. 2). However, the weakest economic sectors, often employing foreign workers living in the central city, were hardest hit by the crisis, and this still impedes geographical mobility for these groups.

18As a consequence of these suburbanization processes, commuting to Brussels increased enormously after the Second World War. The share of commuters in the labour market of the Brussels Capital Region rose from 20 % in 1947 to more than half in 1991. In 2001, it was already 60 %. In absolute figures, the number of people commuting to the Brussels Capital Region rose continually from 114.000 in 1947 to 314.000 in 1991. In 2001, this figure slightly decreased. However, unlike the American experience, this does not mean that the suburban municipalities are becoming more self-sufficient. The number of inhabitants of the banlieue - the outer zone of the urban region most influenced by suburbanization [Van der Haegen et al., 1996] - working in the banlieue itself is continuously declining, and more than half of them still work in the Brussels Capital Region or in the neighbouring municipalities.

Electoral and political polarization in the Brussels urban region

19For the analysis of the impact of suburbanization on electoral geography, we used election results for the Chamber of Deputies from 1977 until 1999 and survey data from 1999, the only year for which survey data were available at a lower geographical level than the province for the whole study region. Before 1977, electoral developments were mainly influenced by the growth of the language parties. By 1977, the electoral support for all these parties is on the decline. The traditional parties not only partly complied with their demands through the federalization of Belgium, they also incorporated language topics in their ideology, and eventually split up along the language cleavage in the 1960s and 1970s [Van Haute and Pilet, 2006 ; Vandermotten et al., 2001]. Because of these splits of parties within the same ideology – together with the existence of a lot of small political parties – the electoral results are not used on the level of political parties, but are aggregated by political tendency on the basis of party platforms.

  • 1 In the latter case, it concerns a dimensional interpretation rather than a scalar interpretation on (...)

20The electoral developments are analyzed at the level of electoral cantons, the lowest geographical level for which Belgian election results are available, by means of correspondence analysis. Correspondence analysis is an exploratory technique [Greenacre, 1984]. It creates a number of synthetic dimensions on the basis of a frequency table, in this case cross-tabulating electoral cantons and political tendencies, taking the different weight of the cells into account. The successive dimensions embody the largest possible share of the total inertia, the deviations from the expected values under independence. These dimensions can be visualized as bi-plots on which both electoral cantons and political tendencies can be positioned as points. The centre of these planes represents the average situation. Overall interpretation of a dimension is possible through an evaluation of the position of the different points, in particular its direction from the centre. When electoral cantons are positioned close to each other, they show similar voting behaviour. When political tendencies are positioned close to each other, they have a common spatial profile. Finally, when an electoral canton and a political tendency are located close to each other, the political tendency in question is over-represented in the respective electoral canton.1 This technique has been used for the analysis of electoral data at a certain point in time [Blasius and Greenacre, 1998]. In a longitudinal study, however, one can consider the values of an electoral canton in the different years as different rows in the frequency table. In this way, it is possible to follow the time-paths of the electoral cantons in a bi-plot, of which the dimensions can be interpreted by means of the fixed position of the political tendencies.

21The first dimension of the correspondence analysis accounts for 47.4 % of the total inertia and shows opposition between parties that can be considered as emanations of traditional cleavages, on the one hand (language parties, communists, Christian-democrats), and new parties, on the other hand (green parties, extreme right parties, and to a lesser degree the renewed liberal parties). The second dimension accounts for 23.9 % of the total inertia and shows an opposition between pillarized parties (social-democrats, Christian-democrats and to a lesser degree liberals) and non-pillarized parties (language parties, extreme right parties, green parties, other parties). Pillarization points to the historical compartmentalization of society by certain political parties - the pillarized parties, in which party organizations controlled almost every aspect of the life of the people belonging to the respective pillars or compartments [see Deschouwer and Lucardie, 2003]. Fig. 3 (on top) shows the bi-plots of these first and second dimensions. The time-paths of the most central urban electoral cantons are indicated (on the left), together with the adjacent and strongly suburbanized cantons, which are all located in Flemish Brabant (on the right). The location of the indicated cantons is shown on Fig. 4.

Figure 3. Time-paths of urban cantons (left) and Flemish suburban cantons (right) on the three main dimensions of the correspondence analysis

Figure 3. Time-paths of urban cantons (left) and Flemish suburban cantons (right) on the three main dimensions of the correspondence analysis

POLITICAL PARTIES : gre = green, exr=extreme right parties, lan= language parties, soc= social democratie, com= communist parties, chr=Christian-democratic parties, lib=liberal parties, oth=other parties

Figure 4. Urban and Flemish suburban electoral cantons in the former province of Brabant

Figure 4. Urban and Flemish suburban electoral cantons in the former province of Brabant

Source : Staatsblad, 1977.

22The time-paths of the central urban cantons on these dimensions differ from those of the Flemish suburban cantons. Within both groups however, time-paths are very similar. The central urban cantons all start from a depillarized situation. This is particularly due to the strength of the language parties in this area of severe language conflict, where the mobilizing efforts of these parties were most successful. The gradual weakening of the language parties, reflected in the movement on the first dimension, went together with a clear movement towards the pillarized parties on the second dimension. This is related with the successful strategy of the traditional parties to incorporate the language cleavage into their ideology and structures. From the 1980s onwards, the growth of new parties – mainly green and extreme right parties – is reflected in a clear movement on the first dimension, and lead to a renewed depillarization on the second dimension. In the Flemish suburban cantons, the movement to new, non-pillarized parties is observable too, but depillarization has been happening far more gradually.

23Thus, though the trajectories of both groups are different, the central urban as well as the Flemish suburban cantons move to the new, non-pillarized parties. This development seems to indicate that both areas will evolve to the same point. However, the third axis still accounts for 11.1 % of the total inertia and indicates that this will not be the case. It is a clear left-right opposition, in which the left is determined by communists, social-democrats, and green parties, and the right mainly by Christian-democrats and the extreme right parties. Fig. 3 (below) shows the bi-plots of the first and third dimensions and the time-paths of the central urban cantons (on the left) and Flemish suburban cantons (on the right). The opposite development of both groups is clear. The central cantons remain on the left, the Flemish suburban cantons stick to the right, and the difference between both groups strengthens. At the end of the 1990s, the Flemish suburban cantons are most to the right of all the cantons of the former province of Brabant. The opposite is true for the central canton of Sint-Gillis, the only one completely situated in the urban centre. The general twists in both trajectories are related to national party political developments. The most striking twists are observable in the crisis years of the mid-1980s, when the opposition of social-democrats was very successful in mobilizing against the governing Christian-democrats, which lost a lot of votes.

Defensive localism in a context of persisting commuting

  • 2 In 2004, Vlaams Blok was forced to dissolve, as a series of its constitutive organizations were con (...)

24The above analysis shows that the Flemish suburban environment, which is traditionally Christian-democrat, is now turning to the extreme right. As a consequence, the main extreme right party in Flanders, the Vlaams Blok (VB)2, has experienced a spectacular electoral growth since its establishment in 1978 and became, at the latest Flemish elections, the second largest party, after the declining Christian-democrats. This does not match with most research on the electoral success of the extreme right parties. This research mostly focuses on the urban environment where the extreme right exploits immigrant scapegoat discourses of the impoverished inhabitants [see Schuermans and De Maesschalck, 2009]. However, using the concept of suburban defensive localism [Gainsborough, 2001a], the success of these parties in suburban districts becomes interpretable. The (VB) uses a virulent anti-urban discourse that identifies cities with foreigners and crime. An analysis of the federal party programmes of the VB (Vlaams Blok 1978 ; 1981 ; 1985 ; 1987 ; 1991 ; 1995 ; 1999), and in particular their views on the city, shows the gradual association the party makes between cities, foreigners, and crime, and the growing attention for these themes. In their first manifestoes, cities, foreigners, and crime were not the main topics. Emphasis was mainly on the traditional themes of the extreme right of that time : Flemish nationalism, solidarism – a system similar to corporatism – and ethical values as a counterforce against Soviet imperialism and American moral decay. In 1985 foreigners, or "guest workers" as they were still called, got more attention and were explicitly linked to cities. The Vlaams Blok (1985 : 2) stated :
"The presence of a large number of guest workers in a limited number of confined spaces (mostly in certain quarters of big cities) will create ever larger and more terrifying problems. (…) the feasible colonization of Flanders and Europe by Africans and Asians is unacceptable for the Vlaams Blok."

25In 1987, street crime was introduced as a major topic, and was also explicitly linked to the cities :
"Crime is growing. In ever more quarters of our big and medium-sized cities, it is certainly not advisable to venture on the streets in the dark."[Vlaams Blok, 1987, n.p.].

26However, the presence of immigrants was not yet explicitly linked to it. In 1991, the VB eventually put the blame for crime (in cities) on foreigners :
"Large groups of Turks, Moroccans and the like stick together in our big cities. Second and third generation youngsters are uprooted and form an enduring source of discontent and crime."[Vlaams Blok, 1991, p. 15].

27The triad cities-foreigners-crime is eventually constructed, and is increasingly stressed in the following years. In 1995, the first year in which the manifesto treats a large spectrum of elaborated themes, the chapters on foreigners and crime encompass 15 % of the manifesto ; in 1999 it was 30 %. Moreover, these themes often return in other chapters, such as the chapter on social policy and the chapter on elderly people.

28This anti-urban discourse that associates cities with foreigners and crime caught on in the suburban environment. In particular, The VB now exploits the suburban anti-urban defensive reactions facilitated by the impoverishment of the city and the inflow of foreigners into the city. This intense suburban process of identifying the Belgian cities with foreigners and crime, and oppositely nurturing of the safe and white suburban environment, is clearly shown in qualitative research on the basis of focus groups [Schuermans and De Maesschalck, 2009]. At the same time, the party stresses the importance of urban security through the hard approach of severe punishment, zero-tolerance, more police patrols, and the permanent expulsion of criminal foreigners. In contrast with many American cities, the majority of suburban inhabitants in Belgium still have to commute to the city for work, as is the case in many cities of Western Europe [see Kesteloot, 2005]. As a consequence, the promise of more urban safety without requiring tax-burdening solutions for urban social problems is an attractive option for the suburban dweller.

29Next to the growing mobilization around the topics of foreigners and crime, increasingly associated with the city, the VB has continued to mobilize around Flemish nationalism. The VB itself has made reference to its main topics by what it calls ’the three Vs’ : Vlaanderen, Vreemdelingen en Veiligheid (Flanders, foreigners and security). However, the migrant issue and, to a lesser degree, crime remain the most important motives to vote for the VB, Flemish nationalism playing a minor role [Swyngedouw, 2001]. Moreover, in Brabant, these topics have become linked. One of the main themes the VB put forth during the latest local elections in Brabant was the resistance against what the party calls ‘Brusselization‘ [see Laeremans, 2007]. This concept unites the diffusion of the perceived dangers of the city (foreigners and crime) into the neighbouring municipalities with the diffusion of the French language.

The different party political heritage in Walloon-Brabant

30Though suburbanization in the Walloon part of the Brussels Urban Region has been pronounced, the time-paths of the Walloon-Brabant electoral cantons are very different from their Flemish-Brabant counterparts. Fig. 5 shows the time-paths of the Walloon Brabant electoral cantons on the first and second dimensions (on the left) and on the first and third dimensions (on the right) of the correspondence analysis (for the location of these cantons, see Fig. 4). Like the Flemish suburban cantons, these cantons move towards the new parties on the first dimension and gradually depillarize on the second, though also here a reversion of these trends is observable in the 1980s because of the integration of the language cleavage in the ideology of the traditional parties. However, depillarization is clearly less pronounced in this region. At the same time, all the Walloon Brabant cantons remain situated on the political left on the third dimension, though the most suburbanized cantons south of Brussels have moved from the utmost left to the utmost right.

Figure 5. Time-paths of the Walloon cantons on the three main dimensions of the correspondence analysis

Figure 5. Time-paths of the Walloon cantons on the three main dimensions of the correspondence analysis

POLITICAL PARTIES : gre = green, exr=extreme right parties, lan= language parties, soc= social democratie, com= communist parties, chr=Christian-democratic parties, lib=liberal parties, oth=other parties

31This seems to point to a weak, if any, suburban anti-urbanism in the Walloon suburban fringe. The illustrated developments, and their differences with the Flemish suburban cantons, are fundamentally related to the different political history of Flanders and Wallonia. While Flanders is historically a Catholic and Christian-democratic region, Wallonia has a social-democratic past. Though Walloon Brabant does not belong to the big Walloon industrial basins, it has known a lot of small-town industrial developments and was strongly influenced by the socialist mobilizations in these industrial basins [Vandermotten et al., 2001]. The social-democratic party still has a considerable mobilizing power in Wallonia, which is less influenced by depillarization than Flanders [Coffé, 2005]. At the same time, as a consequence of the split of the parties along the language cleavage, the party political landscape in Wallonia differs from that in Flanders. In particular, the extreme right is virtually non-existent in Wallonia. In contrast with the VB in Flanders, the French-speaking extreme right is very fragmented and badly organized. Yet, Coffé (2005) clearly shows that the electoral potential for the extreme right is as large in Wallonia as in Flanders, electoral success being prevented by the weak organizational capacity of the French-speaking extreme right and the mobilizing powers of the other parties, in particular the social-democratic party.

32This means that suburban anti-urbanism can exist in the suburban areas of Wallonia, but does not (yet) translate into election results. To account for the different party political structures in Flanders and Wallonia an analysis can be carried out of anti-urban attitudes rather than voting behaviour, on the basis of survey data. In this way, language cleavages do not directly influence the analysis and a possible opposition between city and suburban fringe can be uniformly analyzed for the whole former province of Brabant. Moreover, as it concerns data on the individual level, it is possible to verify whether this opposition is the result of an independent effect of place, rather than the different population composition of city and suburban fringe.

33Furthermore, this analysis can also account for the party political differences between the Flemish suburban fringe and Brussels. Though French-speaking and Dutch-speaking political parties co-exist in the Brussels Capital Region, the region is mainly French-speaking, what could contribute to the weak position of the extreme right in Brussels. In Antwerp, the largest Flemish city, the extreme right has always been very strong in the central city, exploiting immigrant scapegoat discourses of the impoverished inhabitants. However, the same processes of urban-suburban polarization can be observed in Antwerp [see De Maesschalck, 2007]. The social-democrats have become the largest party in the central city again, while the VB is strongly suburbanising and is now the largest party in different suburban municipalities.

Anti-urban attitudes in the whole suburban fringe

34Since 1991, the political science departments of several Belgian universities have co-organized a panel study after each federal election, probing the political attitudes and behaviour of the Belgian population. In the following analysis, only the ’1999 General Election Study’ (ISPO/PIOP 2002) is used, since this alone locates the interviewees at a lower geographical level than the provincial (i.e., the municipal) for the three Belgian regions. With the exception of one study [Lubbers et al., 2000], the Belgian surveys were never used for spatial analysis at a lower geographical level than the provincial.

35In this paper, a multi-level analysis of anti-urban attitudes will be performed, separating the influence of individual-level characteristics and the impact of place. In this way, it is possible to assess whether these attitudes are indeed more widespread in suburban municipalities than in the urban centre, controlled for individual-level characteristics. Multi-level analysis can be considered as a form of regression analysis. After the determination of a general regression equation, the remaining variance is partitioned into an individual variance and an inter-group variance, the groups (in this case, municipalities) being defined in advance. The individual variance points to differences between individuals, the inter-group variance to differences between municipalities. Subsequently, this inter-group variance can be used to derive separate regression equations for the different groups. The estimates of the parameters of these separate regression equations are influenced by the general equation : the less precise these so-called shrinkage estimates, the more they are influenced by the general equation. Thus, ’the shrinkage estimates ‘borrow strength’ (…) from all the data simultaneously’ [Jones, 1991, p. 151]. At the same time, the estimates of the parameters of the general regression equation are a weighted means of the least-squares estimation regressions of the groups, but weighted in such a way that the most reliable regressions have the largest impact. This means that multi-level analysis models the different levels at the same time.

36Anti-urban attitudes can only be measured indirectly, since direct attitudes towards the city are not included in the survey. Based on the above discourse which equates the city with foreigners and crime – and is supposedly more widespread in the suburban fringe – we combined the attitudes towards foreigners and crime as an indicator of anti-urbanism. This combined scale is a clear reflection of the anti-foreigner and hard approach towards urban crime discourse around which the extreme right mobilizes. Though such variables are also used to operationalize a cleavage universalism-ethnocentrism [see Van Haute, 2008], it is our contention that they are strongly linked to anti-urbanism, as a consequence of the strong interplay between ethnocentrism, attitudes towards crime and anti-urbanism. As also shown in the US context [Henderson, 2006], Schuermans and De Maesschalck (2009) describe the strong identification of foreigners and crime with the city in Belgium. The suburban environment, on the other hand, is considered to be ‘white’ and safe, leading to defensive reactions against the real or perceived urban annoyances.

  • 3 The thirteen constituent scales each are an evaluation, by means of a value from one to five, of ce (...)

37In particular, we combined thirteen scales to measure ethnocentrism, on the one hand, and attitudes towards crime, on the other. The scale has a minimum of thirteen (no anti-urbanism) and a maximum of sixty-five (strong anti-urbanism).3 It clearly concerns one underlying and uni-dimensional concept, as is indicated by the value of Cronbachs alpha. This scale of reliability has a value between zero and one, and can be considered as a measure of common variance [Bland and Altman, 1997]. While 0.70 is accepted as a sufficient value for reliability, the combined scale we constructed has a value of 0.93.

38Using this combined scale as dependent variable, two multilevel analyses are performed. In a first analysis, only one independent variable is used, the net household income, which is allowed to vary at the municipal level. The net household income is a broad, synthetic variable controlling for the influence of a certain class membership. The use of only one independent variable enables the visualization of the results, in particular the regression lines of the different municipalities. In a second analysis, more independent variables are added, which control for the individual characteristics usually considered as influential in electoral and attitudinal research. These are – next to income – education, occupation, age, and gender. While the income variable is split into twenty-eight categories of EUR 500, education has only five (lower, lower secondary, higher secondary, higher, university), just as occupation (semi-routine and routine occupations, lower supervisory and technical occupations, small-scale employers and independent workers, intermediate occupations, managers and professionals), the latter based on the British social economic classification. The age variable is continuous. In the former province of Brabant, 1105 people in seventy-two municipalities were interviewed. Eventually, thirty-nine municipalities with a total of 956 interviewees were retained for our analysis, on average twenty-five interviewees per municipality.

39Table 1 gives the output of the first multi-level analysis. The upper table shows information on the general equation. The (positive) intercept and the (negative) income coefficient are significant at the 0.01 level. This means that the level of anti-urbanism decreases as the income level rises. The lower table is more important considering the possible existence of an independent spatial effect. It splits the residual variance into the individual variance and the group variance on the municipal level. The individual variance has a value significant at the 0.01 level. The residual variance on the municipal level is split up into the variance of the intercept, the variance of the income coefficient, and the covariance between both. Only the first is significant at the 0.01 level and has a value more than half of the variance at the individual level. In other words, it shows that much of the residual variance is situated on the municipal level.

Table 1. Scale of anti-urbanism: effects and variances of multi-level analysis (independent variable income)

Effects, general regression equation (SAS – PROC MIXED)


Standard Error

t Value

Pr > |t|











Residual variances (SAS – PROC MIXED)


Standard Error

Z Value

Pr Z

Group : intercept





Group : covariance





Group : income










40Fig. 6 visualizes this municipal variation, showing the regression lines of the different municipalities. On this graph, the municipalities belonging to the Brussels Capital Region, the city of Louvain, and the other – mostly suburban – municipalities are, respectively, indicated. It is clear that the municipalities of the Brussels Capital Region, together with the smaller city of Louvain, in general have a lower value on the scale of anti-urbanism than the other, mostly suburban, municipalities. In addition, the regression lines of the urban municipalities descend less or even ascend as the income level rises, though caution has to be made here since the group variance of the income coefficient was not significant.

Figure 6. Anti-urban attitudes in the municipalities of the former province of Brabant

Figure 6. Anti-urban attitudes in the municipalities of the former province of Brabant

41Yet, the spatial logic is stronger than a simple opposition between urban and suburban municipalities. The municipalities with the most pronounced regression lines are located on the utmost ends of the urban-suburban divide. The four municipalities with the highest intercepts, and therefore the most negative attitudes, are among the wealthiest suburban municipalities of the country (respectively Lasne, Braine-l’Alleud, Rixensart, and Waterloo). It concerns spatially connected municipalities in Walloon Brabant on the south side of the Sonian Forest, which stretches from the east of the Brussels Capital Region to the Walloon Region and has stimulated early and very pronounced suburbanization. On the other hand, the two municipalities with the lowest intercepts, and therefore the most positive attitudes, are very central and impoverished urban municipalities of the Brussels Capital Region (respectively Sint-Gillis and Elsene). The municipalities of the Brussels Capital Region with intercepts higher than the general equation all are located in the relatively wealthy and more suburban eastern part of the region.

42These spatial differences remain when more control variables are added, further reducing the residual variance. Table 2 shows the results of this second multi-level analysis. The lower table reveals that, at least for the intercept, the spatial differences still are significant at the 0.01 level. The upper table shows that education and, to a lesser degree, occupation, replace income as a significant explanatory variable. At the same time, the age variable is also significant at the 0.01 level. This means that the older the interviewees, the more negative the measured attitudes.

Table 2. Scale of anti-urbanism: effects and covariances of multi-level analysis (larger set of independent variables)

Effects, general regression equation (SAS – PROC MIXED)


Standard Error

t Value

Pr > |t|































Residual variances (SAS – PROC MIXED)


Standard Error

Z Value

Pr Z

Group : intercept





Group : covariance





Group : income










43Thus, the opposition between city and suburbs turns out to be very clear. The analysis revealed that the discourse connecting anti-foreigner stances with a hard approach to urban crime, which we operationalized as a measure of anti-urbanism - starting from the mutual relationships between ethnocentrism, attitudes towards crime and anti-urbanism - showed a clear urban-suburban pattern. These spatial differences remain when controlling for a series of individual characteristics. The most suburbanized municipalities have the highest values on the scale of anti-urbanism. Moreover, the municipalities with the largest intercepts are located in Walloon Brabant. This means that anti-urban attitudes, like in Flanders, are also observable in the Walloon suburban areas. However, it does not (yet) manifest itself in election results because of the different party political structure in Wallonia and in Flanders.

44The most central urban municipalities have the lowest values on the scale of anti-urbanism. Here, the additional role of gentrification processes can be discerned. Indeed, the low intercepts of the municipalities of Elsene – where the French and Dutch speaking Brussels Universities (ULB and VUB) are located – and Sint-Gillis can be related with gentrification processes, which are most marked in these municipalities [Van Criekingen, 2005]. The attitudinal and electoral consequences of gentrification were already examined in Canadian cities by Ley (1994 ; 1996). Ley stated that leftist and libertarian attitudes and electoral behaviour are associated with the new cultural class, which is part of a larger new middle class of experts and managers in the advanced service sector in the big cities in post-industrial society. These are highly educated groups employed in the arts, media, education and public sectors, which have a preference for living in the urban centres of big cities. Recently, Walks (2006) has found similar mechanisms in Toronto, where gentrifiers’ preference for inner-city living was part of identity construction processes associated with leftist attitudes and behaviour. Ley’s observations are also confirmed by Hamnett (2000 ; 2003), who links these processes to world cities as defined by Sassen (1991). Following Ley, he also states that when gentrification takes on other forms, it is possible that the new cultural class is pushed aside by higher income groups of the advanced service sector. These groups are more concerned about maximizing the value of their property investment, are risk averse and, consequentially, more conservative.

45However, following Van Criekingen and Decroly (2003), gentrification takes on the form of marginal gentrification in Brussels. Like stated earlier, this is a usually temporary immigration of young non-familial households – mostly of middle class origin – which find in the urban housing market a solution for the growing time gap between leaving the parental home and the attainment of a stable family and income. Marginal gentrifiers are mainly employed in the communication, the cultural and the (para)public sectors, and are very comparable to the groups Ley described in the Canadian cities. This means that marginal gentrification in the Brussels context would also strengthen the left-oriented parties in the central cities. Indeed, the 1999 sample reveals that highly educated young people between twenty-one and thirty-five without children in the Brussels Urban Region have significantly more positive attitudes towards cities, position themselves significantly more to the left, and, as a consequence, vote relatively more on social-democratic and green parties than their counterparts in the other municipalities.

46Finally, the group of naturalized foreigners is able to influence the election results in the central city, foreigners having no voting rights on the regional and national level. Sandri and De Decker (2008) demonstrate that Muslims, the most important group among the Belgians of foreign origin in Brussels, vote mainly for parties on the left, consistent with their socio-economic position. Because of abstinence, the electoral and political influence of new Americans in the central cities of the U.S. is weaker than their absolute number could involve [Logan and Mollenkopf, 2003]. However, voting in Belgium is still compulsory. At the same time, the nationality legislation has been increasingly liberalized ; the successive Belgian governments chose to encourage integration of foreigners through granting the Belgian nationality rather than granting voting rights. The naturalization legislation has been influential in the Brussels Capital Region in particular [Kesteloot et al., 1999]. Jacobs (2004) estimated the number of naturalized foreigners in this region at 10 %.

Conclusion and discussion

47Electoral and political polarization between city and suburban fringe is documented in the U.S. and the Anglo-Saxon world, where a conservative and neo-liberal suburban area increasingly contrasts with a left-wing urban centre that favours redistribution. In particular, Gainsborough (2001a) identifies the growth of a suburban defensive localism. Enhanced by the urban crisis and increasing suburban self-sufficiency, support for redistribution with the impoverished cities has weakened and is combined with defensive reactions against these cities, even resulting in attempts at suburban secession. This paper aimed at broadening the scope of this research and explored the political and electoral consequences of suburbanization in Brussels, Belgium. The Brussels urban region is strongly marked by suburbanization processes until the present day but is, as a strong representative of many Western European cities, not characterized by functional independence of the suburbs.

48Electoral polarization between urban centre and suburban fringe was clearly observable in Brussels. While the political left is increasingly successful in the urban centre, the extreme right is successful in the Flemish suburban fringe and joins the traditional Christian-democratic presence in this area. As a result, it is now the second largest party in Flanders after the declining Christian-democrats. The extreme right propagates a virulent anti-urban discourse, identifying the cities with foreigners and crime, and warning about the diffusion of these perceived urban dangers. At the same time, the party proposes a hard approach to urban crime. This catches on in the suburban fringe and can be associated with defensive localism mobilizing the anti-urban attitudes of the suburbanite and the fear of the perceived dangers of the city. However, contrary to the U.S. situation, the self-sufficiency of the suburban zone is not increasing and most suburbanites still have to commute to these cities for work. Rather than resulting in support for solutions to urban social problems, a suburban discourse emerges focusing on a safe city for the suburban user and hence a preference for a hard approach to urban crime, without much reference to the interests of the present inhabitants of the city. This means that the city-suburban divide also manifests itself without marked functional independence of the suburbs, though in an adapted form. Likewise, the interests of the city dwellers are non-essential for the suburban citizen, but the city has to be made safe for the suburban city user without requiring tax-burdening solutions for urban social problems.

49However, in the Walloon part of the suburban fringe, electoral polarization could not be observed. Since the split of the main political parties along the language cleavage, the party political structure differs fundamentally on both sides of the language border, which makes comparison difficult. In particular, the political extreme right is virtually absent in Wallonia, while the social-democratic party has still quite some mobilizing power there. By means of survey data, it was possible to compare the presence of anti-urban attitudes for the whole study area uniformly. Moreover, using a multi-level analysis, it was possible to control for a range of individual characteristics. This analysis revealed that the discourse connecting anti-foreigner stances with a hard approach to urban crime showed a clear urban-suburban pattern, after controlling for a range of individual characteristics. The more pronounced the suburbanization in a municipality, the more negative the measured attitudes are. The more central the municipality, the more positive these attitudes are. In this way, the municipalities with the most pronounced negative attitudes were located in the Walloon suburban area. This does not seem to manifest itself (yet) in election results, but does point at the potential of a party mobilizing around these anti-urban attitudes in this region. More in general, it points to the importance of party political history and mobilization in the construction of the city-suburban electoral divide.

50The left-wing inclination of the central cities is likely to be strengthened by the growing number of naturalized foreign workers and their descendents. The living conditions of these groups are on average very poor and the available research points to left-wing voting behaviour. Unlike the U.S. electoral system, voting is compulsory in Belgium, and therefore these groups can significantly influence the election results. In addition, gentrification is influential in the left-wing inclination of the central cities. Gentrification takes the form of marginal gentrification in cities like Brussels, ranking relatively low in the world city hierarchy, meaning a temporary immigration of young, highly educated, non-familial, middle-class households that consider the urban environment as fitted reproduction setting for transitional stages in work and household positions. Anti-urban attitudes are less developed among these groups, and they vote more for left parties than their counterparts in the suburban fringe. This means that gentrification in Brussels is not (yet) an immigration of very wealthy, risk averse groups willing to protect their property investment. The latter process can be associated with more conservative attitudes and connected with what Smith (1996 ; 2002) calls the revanchist city : the process in which the space of the weaker groups on the housing market is systematically curtailed with the aim of capital accumulation.

51Thus, the effect of marginal gentrification on electoral geography can be considered, to a large extent, as a consequence of selective migration patterns of the middle class. The available data do not permit the verification of this for the suburban attitudes and voting behaviour. It is not clear whether this concerns an effect of self-selection rather than an independent impact of the suburban environment. Walks (2006) showed that self-selection of the middle class is an important factor explaining the left-wing attitudes of the gentrifiers in Canadian cities, being less so for the right-wing attitudes of the suburbanites. In this area, an independent impact of the suburban environment seems to play a larger role. The author unravelled the possible processes producing these impacts, though these mechanisms seem to be associated in complex ways. In the U.S. context, Gainsborough (2001a : 62) suspects that the impact of the suburban environment is more important than self-selection, seeing the decisive role of policy and estate agents in the suburbanization process. Moreover, the author states, it actually is an artificial distinction :
"Even if causality does work in the opposite direction – that is, even if distinctive attitudes cause people to move to the suburbs - these attitudes are still unavoidably affected by the existence of suburbs as a locational choice. (…) The existence of suburbs makes defection from the problems of the city possible by reinforcing the idea that the problems can be escaped without serious investment in solving them."

52The latter arguments are convincing in the Belgian context, where policy has largely influenced the development of suburbanization. Finally, the causality of the urban-suburban political and electoral divide does not relate directly to the suburbanization of politics. Whether self-selection is important or an effect of the suburban environment itself, the growing (enfranchised) population in the suburbs vis-à-vis the central cities enhances the electoral and representational potential of the suburbs. Moreover, political parties of whatever political hue are not indifferent to this growing electoral potential. This is documented in the U.S. and Britain [Gainsborough, 2001a ; Walks, 2005b], but parties are also adapting in Belgium to what they perceive as the suburban preferences, and these developments even affect the parties of the left, as is clearly the case with the Flemish social-democratic party.

53The political and electoral impact of suburbanization in Belgium is not limited to Brussels. De Maesschalck (2007) shows similar patterns of suburbanization and electoral-political polarization in the Antwerp urban region, though the results could not be verified on the individual level because of data insufficiency. Even in France, where impoverishment of the central cities as a consequence of suburbanization is relatively weak, such processes are demonstrated [Ravenel et al., 2003 ; Bussi et al., 2004]. This adds further evidence to the observation that
"there is something about the suburbanization process that leads residents of inner cities and suburbs to adopt divergent political viewpoints, regardless of local or national context." [Walks, 2005b, p. 515].

54At the same time, the differences with the Anglo-Saxon experience and the internal differences within the Brussels Urban Region must provoke further research on the electoral and political impacts of suburbanization, and, more in general, the relations between urban space, political convictions, and electoral reality, in different national and local contexts, also outside the Anglo-Saxon world.

Haut de page


ABBOTT C. (1997), The Portland Region : Where City and Suburbs Talk to Each Other – and Often Agree, Housing Policy Debate, vol. 8, n° 1, pp. 11-51.

BIEL H.S. (1972), Suburbia and Voting Behavior in the London Metropolitan Area : an Alternative Approach, Tijdschrift voor Economische en Sociale Geografie, vol. 63, n° 1, pp. 39-43.

BLAND J.M. & ALTMAN D.G. (1997), Statistics Notes : Cronbach’s Alpha, British Medical Journal, n° 314, pp. 572-572.

BLASIUS J. & GREENACRE M. (1998) (ed.), Visualization of Categorical Data, London, Academic Press.

BUSSI M., FOURQUET J. & RAVENEL L. (2004), Élections régionales de 2004 : notabilités traditionnelles et nouvelles maîtrises des territoires électoraux, Revue française de science politique, vol. 54, n° 4, pp. 639-667.

COFFÉ H. (2005), Do Individual Factors Explain the Different Success of the Two Belgian Extreme Right Parties, Acta Politica, vol. 40, n° 1, pp. 74-93.

COX K.R. (1968), Suburbia and Voting Behavior in the London Metropolitan Area, Annals of the Association of American Geographers, vol. 58, n° 1, pp. 111-127.

COX K.R. (1969), Comments in Reply to Kasperson and Taylor, Annals of the Association of American Geographers, vol. 59, n° 2, pp. 411-415.

CRAMPTON P. (1984), Spatial Polarisation of Political Representation in Great Britain 1945-1979, Geography, n° 302, pp. 28-37.

DE CORTE S., RAYMAEKERS P., THAENS K., VANDEKERCKHOVE B. & FRANÇOIS G. (2003), Onderzoek naar de migratiebewegingen van de grote steden in de drie gewesten van België, Brussel, Cel Grootstedenbeleid.

DE DECKER P., KESTELOOT C., DE MAESSCHALCK F. & VRANKEN J. (2005), Revitalising the City in an Anti-Urban Context : Extreme Right and the Rise of Urban Policies in Flanders-Belgium, International Journal of Urban and Regional Research, vol. 29, n° 1, pp. 152-171.

DEFRONZO HASELHOFF K. (2002), Motivations for the San Fernando Valley Secession Movement : the Political Dynamics of Secession, Journal of Urban Affairs, vol. 24, n° 4, pp. 425-443.

DE MAESSCHALCK F. (2005), “Een spiegel van de maatschappij ? De woonplaats van de Belgische parlementsleden in kaart gebracht, 1961-2004”, in F. Witlox & V. Van Acker (ed.), Mobiliteit, maatschappij en milieu in kaart gebracht, Gent, BEVAS-SOBEG, pp. 63-69.

DE MAESSCHALCK F. (2007), “De stad versus de suburbane rand. De electorale geografie van de suburbanisatie in het Antwerpse stadsgewest”, in N. Van Nuffel (ed.), Van Christaller tot Wallerstein, Zelzate, Nautilus Academic Books, pp. 51-66.

DESCHOUWER K. & LUCARDIE P. (2003), Partijen en partijsystemen in Nederland en Vlaanderen, Sociologische Gids, vol. 50, n° 2, pp. 131-155.

FORREST J., HAY A.M. & JOHNSTON R.J. (1984), In What Ways are the Suburbs Different ? A Note, Politics, vol. 19, pp. 97-101.

GAINSBOROUGH J.F. (2001a), Fenced Off. The Suburbanization of American Politics. Washington, D.C., Georgetown University Press.

GAINSBOROUGH J.F. (2001b), Bridging the City-Suburb Divide : States and the Politics of Regional Cooperation, Journal of Urban Affairs, vol. 23, n° 5, pp. 497-512.

GREENACRE M. (1984), Theory and Applications of Correspondence Analysis, London, Academic Press.

HAMMETT C. (2000), “Gentrification, Postindustrialism, and Industrial and Occupational Restructuring in Global Cities”, in G. Bridge & S. Watson (ed.), A Companion to the City, Oxford, Blackwell, pp. 331-341.

HAMMETT C. (2003), Gentrification and the Middle-Class Remaking of Inner London, 1961-2001, Urban Studies, vol. 40, n° 12, pp. 2401-2426.

HENDERSON J. (2006), Secessionist Automobility : Racism, Anti-Urbanism, and the Politics of Automobility in Atlanta, Georgia, International Journal of Urban and Regional Research, vol. 30, n° 2, pp. 293-307.

HODGE D.C. & STAEBELI L.A. (1992), Social Relations and Geographic Patterns of Urban Electoral Behavior, Urban Geography, vol. 13, n° 4, pp. 307-333.

HOFFMANN-MARTINOT V. & SELLERS J. (2005) (ed.), Metropolitanization and Political Change, Wiesbaden, VS Verlag für Sozialwissenschaften.

ISPO / PIOP (2002), 1999 General Election Study Belgium. Codebook : Questions and Frequency Tables, Leuven, ISPO – K.U.Leuven / Louvain-La-Neuve : PIOP – U.C. Louvain.

JACOBS D. (2004), “The Challenge of Minority Representation in Brussels”, in G. Aubarell, C.A. Nicolau & A. Ros (ed.), Immigració i questió nacional. Minories subestatals i immigració a Europa, Barcelona, Editorial Mediterrània, pp. 51-64.

JOHNSTON R.J. (1986), Places and Votes : the Role of Location in the Creation of Political Attitudes, Urban Geography, vol. 7, n° 2, pp. 103-117.

JOHNSTON R.J. & FORREST J. (1985), Spatial-Structural Effects and the Geography of the Flow-of-the-Vote at the 1991 New South Wales State Election, Australian Geographe, vol. 18, n° 2, pp. 286-290

JONES K. (1991), Specifying and Estimating Multi-Level Models for Geographical Research, Transactions of the Institute of British Geographers, vol. 16, n° 2, pp. 148-159.

JOURET B. (1972), Définition spatiale du phénomène urbain bruxellois, Bruxelles, Éditions de l’Université de Bruxelles.

KASPERSON R.E. (1969), On Suburbia and Voting Behavior, Annals of the Association of American Geographers, vol. 59, n° 2, pp. 405-411.

KEIL R. (2000), Governance Restructuring in Los Angeles and Toronto : Amalgamation or Secession ?, International Journal of Urban and Regional Research, vol. 24, n° 4, pp. 758-781.

KESTELOOT C. (2001), “Verstedelijking in Vlaanderen : problemen, kansen en uitdagingen voor het beleid in de 21e eeuw”, in F. De Rynck (ed.), De eeuw van de stad. Voorstudies, Brussel, Ministerie van de Vlaamse Gemeenschap, pp. 15-83.

KESTELOOT C. (2005), “Urban Socio-Spatial Configurations and the Future of European Cities”, in Y. Kazepov (ed.), Cities of Europe, Oxford, Blackwell, pp. 123-149.

KESTELOOT C., PELEMAN K. & VAN DER HAEGEN H. (1999), The Changing Geography of Foreigners in Belgium 1991-1999 : the Impact of Population and Nationality Changes, Tijdschrift van de Belgische Vereniging voor Aardrijkskundige Studies, vol. 68, n° 2, pp. 313-348.

LAEREMANS B. (2007), Verbrusseling : tegengaan of ondergaan ?, Brussel, Egmont.

LEY D. (1994), Gentrification and the Politics of the New Middle Class, Environment and Planning D : Society and Space, vol. 12, n° 1, pp. 52-74.

Ley D. (1996), The New Middle Class and the Remaking of the Central City, Oxford, Oxford University Press.

Logan J. & MOLLENKOPF J. (2003), People and Politics in America’s Big Cities, New York, Drum Major Institute for Public Policy.

LUBBERS M., SCHEEPERS P. & BILLIET J. (2000), Multilevel Modelling of Vlaams Blok Voting : Individual and Contextual Characteristics of the Vlaams Blok Vote, Acta Politica, vol. 35, n° 4, pp. 363-398.

OSTENDORF W. (1987), “Culturele differentiatie binnen Nederland : van territoriale binding naar sociaal-ruimtelijke uitsortering”, in H. Van der Wusten (ed.), Postmoderne Aardrijkskunde. De sociografische traditie voortgezet, Muiderberg, Coutinho pp. 168-179.

PASSCHIER N.P. & VAN DER WUSTEN H. (1990), “The Electoral Geography of the Netherlands in the Era of Mass Politics, 1888-1986”, in R.J. Johnston, F.M. Shelley & P.J. Taylor (ed.), Developments in Electoral Geography, London, Routledge, pp. 39-59.

RAVENEL L., BULÉON P. & FOURQUET J. (2003), Vote et gradient d’urbanité : les nouveaux territoires des élections présidentielles de 2002, Espace Populations Sociétés, n° 3, pp. 469-482.

RIVIÈRE J. (2008), Le vote pavillonnaire existe-t-il ? Comportements électoraux et positions sociales locales dans une commune rurale en cours de périurbanisation, Politix, vol. 83, n° 3, pp. 23-48.

SANDRI G. & DE DECKER N. (2008), “Le vote des musulmans”, in P. Delwit & E. Van Haute (ed.), Le vote des Belges, Bruxelles, Éditions de l’Université de Bruxelles, pp. 39-53.

SASSEN S. (1991), The Global City : New York, London, Tokyo, Princeton, Princeton University Press.

SCHUERMANS N. & DE MAESSCHALCK F. (2009), Fear of crime as a political weapon : Explaining the rise of extreme right politics in the Flemish countryside, Social and Cultural Geography, forthcoming.

SMITH N. (1996), The New Urban Frontier : Gentrification and the Revanchist City, London, Routledge.

SMITH N. (2002), New Globalism, New Urbanism : Gentrification as Global Urban Strategy, Antipode, vol. 34, n° 3, pp. 427-450.

SWYNGEDOUW M. (2001), The Subjective Cognitive and Affective Map of Extreme Right Voters : Using Open-Ended Questions in Exit Polls, Electoral Studies, vol0.20, n° 2, pp. 217-241.

TAYLOR P.J. (1969), Causal Models in Geographical Research, Annals of the Association of American Geographers, vol. 59, n° 2, pp. 402-404.

VAN CRIEKINGEN M. (2005), “Mobilités Résidentielles Intra-Urbaines et Inégalités Sociales à Bruxelles”, in F. Witlox & V. Van Acker (ed.), Mobiliteit, maatschappij en milieu in kaart gebracht, Gent, BEVAS-SOBEG, pp. 253-259.

VAN CRIEKINGEN M. & DECROLY J.-M. (2003),Revisiting the Diversity of Gentrification : Neighbourhood Renewal Processes in Brussels and Montreal, Urban Studies, vol. 40, n° 12, pp. 2451-2468.

VAN dER HAEGEN H., VAN HECKE E. & JUCHTMANS G. (1996), De Belgische stadsgewesten 1991, Statistische studiën, vol. 104, pp. 3-42.

VANDERMOTTEN C., DECROLY J.-M., DESSOUROUX C. & ROUYET Y. (2001), Permanences et ruptures dans la géographie électorale de la Belgique, Belgeo, vol. 2, n° 1-2, pp. 7-38.

VAN HAUTE E. (2008), « Ethnocentrisme versus universalisme : des comportements discriminants ? » in P. Delwit & E. Van Haute (ed.), Le vote des Belges, Bruxelles, Editions de l’Université de Bruxelles, pp. 97-110.

VAN HAUTE E. & PILET J.-B. (2006), Regionalist Parties in Belgium (VU, RW, FDF) : Victims of Their Own Success ?, Regional & Federal Studies, vol. 16, n° 3, pp. 297-313.

Vlaams Blok (1978), Verkiezingsprogramma 1978,. Brussel, Vlaams Blok.

Vlaams Blok (1981), Wat wil het Vlaams Blok. Programma Vlaams Blok, Brussel, Vlaams Blok.

Vlaams Blok (1985), Programma Parlementsverkiezingen 1985, Brussel, Vlaams Blok.

Vlaams Blok (1987), Programma Parlementsverkiezingen 1987, Brussel, Vlaams Blok.

Vlaams Blok (1991), Uit zelfverdediging. Verkiezingsprogramma 1991, Brussel, Vlaams Blok.

Vlaams Blok (1995), Nu Afrekenen. Verkiezingsprogramma 1995, Brussel, Vlaams Blok.

Vlaams Blok (1999), Baas in eigen land. Verkiezingsprogramma van het Vlaams Blok, Brussel, Vlaams Blok.

WALKS R.A. (2004a), Place of Residence, Party Preferences, and Political Attitudes in Canadian Cities and Suburbs, Journal of Urban Affairs, vol. 26, n° 3, pp. 269-295.

WALKS R.A. (2004b), Suburbanization, the Vote, and Changes in Federal and Provincial Political Representation and Influence between Inner Cities and Suburbs in Large Canadian Urban Regions, 1945-1999, Urban Affairs Review, vol. 39, n° 4, pp. 411-440.

WALKS R.A. (2005a), The City-Suburban Cleavage in Canadian Federal Politics, Canadian Journal of Political Science, vol. 38, n° 2, pp. 383-413

WALKS R.A. (2005b), City-Suburban Electoral Polarization in Great Britain, 1950-2001. Transactions of the Institute of British Geographers, vol. 30, n° 4, pp. 500-517.

WALKS R.A. (2006), The Causes of City-Suburban Political Polarization ? A Canadian Case Study, Annals of the Association of American Geographers, vol. 96, n° 2, pp. 390-414.

Haut de page


1 In the latter case, it concerns a dimensional interpretation rather than a scalar interpretation on the basis of weighted Euclidian distances.

2 In 2004, Vlaams Blok was forced to dissolve, as a series of its constitutive organizations were convicted for racism. Consequently, the party re-established itself under the new name Vlaams Belang. However, its political staff is unaltered and its political programme is only refurbished to prevent new convictions.

3 The thirteen constituent scales each are an evaluation, by means of a value from one to five, of certain propositions. Originally, one stood for ‘completely agree’ and five for ‘completely disagree’, but this scale was reversed so that a high score on the combined scale corresponds to a high degree of anti-urbanism. Eight scales measure attitudes towards foreigners: ‘Belgium should not have guest workers’,’ Immigrants cannot be trusted’, ‘Guest workers threaten work of Belgians’, ‘Guest workers exploit social security’, ‘Muslims a threat to our culture’, ‘Repatriate guest workers when less jobs’, ‘No political activities for immigrants’, ‘More conditions to become a Belgian’, ‘Five scales measure attitudes towards crime: ‘Zero tolerance in unsafe neighbourhoods’, ‘Young criminals same penalties as adults’, ‘Take a firm line with petty theft’, ‘Criminals have too many rights’, ‘Prison sentence should be full length’. (ISPO, PIOP, 2002)

Haut de page

Table des illustrations

Titre Figure 1. The Brussels region and the former province of Brabant
Crédits Source : Van der Haegen et al., 1996.
Fichier image/jpeg, 904k
Titre Figure 2. Population and income development in the former province of Brabant
Crédits Source : National Statistics Office, 1955 - 2005.
Fichier image/jpeg, 1,5M
Titre Figure 3. Time-paths of urban cantons (left) and Flemish suburban cantons (right) on the three main dimensions of the correspondence analysis
Légende POLITICAL PARTIES : gre = green, exr=extreme right parties, lan= language parties, soc= social democratie, com= communist parties, chr=Christian-democratic parties, lib=liberal parties, oth=other parties
Fichier image/jpeg, 960k
Titre Figure 4. Urban and Flemish suburban electoral cantons in the former province of Brabant
Crédits Source : Staatsblad, 1977.
Fichier image/jpeg, 836k
Titre Figure 5. Time-paths of the Walloon cantons on the three main dimensions of the correspondence analysis
Légende POLITICAL PARTIES : gre = green, exr=extreme right parties, lan= language parties, soc= social democratie, com= communist parties, chr=Christian-democratic parties, lib=liberal parties, oth=other parties
Fichier image/jpeg, 380k
Titre Figure 6. Anti-urban attitudes in the municipalities of the former province of Brabant
Fichier image/jpeg, 1,3M
Haut de page

Pour citer cet article

Référence papier

Filip De Maesschalck, « Beyond the Anglo-Saxon Experience: City-Suburban Political Polarization in Brussels, Belgium », Espace populations sociétés, 2009/2 | 2009, 319-339.

Référence électronique

Filip De Maesschalck, « Beyond the Anglo-Saxon Experience: City-Suburban Political Polarization in Brussels, Belgium », Espace populations sociétés [En ligne], 2009/2 | 2009, mis en ligne le 01 avril 2011, consulté le 28 mars 2017. URL : ; DOI : 10.4000/eps.3737

Haut de page


Filip De Maesschalck

Institute for Social and Ecionomic Geography
KU Leuven
W de Croylaan, 42
3001 Leuven

Haut de page

Droits d’auteur

Licence Creative Commons
Espace Populations Sociétés est mis à disposition selon les termes de la licence Creative Commons Attribution - Pas d'Utilisation Commerciale - Pas de Modification 4.0 International.

Haut de page
  • Logo Université de Lille 1 - Sciences et technologies
  • Logo CNRS - Institut des sciences humaines et sociales