Introduction

Why does the character of political conflict vary from country to country? Possible answers include distinct social divisions, ‘politics following policy’ so that national policy traditions structure politics, or nationally specific conflict because of the interplay between society and policymaking. Each of these answers is part of a distinctive tradition in country comparative studies. For instance, Dolezal et al (2012, p. 94) echo a long-standing political sociological argument when they point out that ‘social class’ forms the base for political positions that divide ‘winners’ and ‘losers’ of globalization. By contrast, in their presentation of country-specific citizenship policy typologies Koopmans et al (2005) stand in a long-running tradition in which historically institutionalized policy trajectories strongly determine the nature of political debate. A different tradition is followed by Van der Brug and Van Spanje (2009) who highlight that of the many social differences and conflicts that could be politicized, political parties can attend to only one or two at a time and strategically select those that fit their profile. Consequently, certain ‘issues are organized into politics, while others are organized out’ (Schattschneider, 1960, p. 69), creating systemic national variation in the issue priorities of political parties. It is of major societal importance to understand, however, under which circumstances or for which political issues these research traditions offer theoretically valid explanations, as the structure of the relationship between society and politics is crucial in determining the quality of representative democracy. We seek to contribute to this major theoretical debate and focus on a very specific aspect of the politics on the migration issue.

In this article, we examine the circumstances under which certain sections of society (and not others) become politically meaningful categories: as objects of policy, as a source of political contestation or as a base for political mobilization. Among social categories, those associated with immigration have recently become central for party-political conflict. We examine whether politicians differentiate among immigrant groups, and if they do whether they usually divide immigrant groups by country of origin, on the basis of their administrative status, or by religion. While our general model can be applied to each of the groups mentioned, the focus is on the difference between Muslim-migrants and other immigrant groups for reasons of research design. There is more intense politicization associated with this (religious) category compared with other (administrative) distinctions among immigrants (as reflected in a growing academic literature on the subject, for example, Fetzer, 2005; Zainiddinov, 2013; Cinalli and Giugni, 2013; Dancygier, 2013; Helbling, 2014b; Strabac et al, 2014). It is thus more likely to find evidence for the theoretically plausible causes outlined with a focus on Muslim immigrants as objects of political claims. These political claims are used as evidence of differentiation and that Muslim immigrants are constituted as a particular group. Our research question is: Under which circumstances do politicians differentiate Muslim immigrants – rather than national or other groups – as objects of political claims?

We relate to the discussion on the origin of political opposition and assess the relative importance of social-structural differences, political institutional structures and (party) political strategies for the construction of socio-political categories in general, and religious categories in particular. Seven European countries are examined between 1995 and 2009: Austria, Belgium, Ireland, the Netherlands, Spain, Switzerland and the United Kingdom. We describe the different ways in which people with a migration background are portrayed as a distinct political category: as an immigrant group from a particular country, as a group denominated by religion (usually Islam), with reference to race or ethnicityFootnote 1 or based on their administrative status. We focus on claims about immigration and civic integration because this is where the construction of immigrant groups is most likely to occur. While the examination of group classifications is not a new field of study, we include a relatively broad range of explanatory factors and mediating factors that vary over time, across countries and between political claims. This allows us to assess the relative importance of several factors, and potentially produce better-specified models, or at least suggest the appropriate level at which future research should construct theories of the politicization of immigration in particular or the character of political conflict in general.

Theory

Our starting point is the observation that there are systematic differences in the categorization of immigrant groups (Koopmans et al, 2005, pp. 107–145; Koopmans, 2007a, pp. 701–702). In very broad terms, these differences in the relevant discursive identities and social categories may be caused by strategic choice on the part of political actors, and by opportunity structures that vary across countries, and over time. We present a parsimonious model that includes factors relating to political behaviour, factors that are more structural in nature, and more contingent factors such as specific events. We propose four hypotheses about the constitution of (Muslim) immigrant categories as a distinct category of political contestation – as objects of policy, but also as claims-makers. These hypotheses pertain to social-structural change, political institutions, party political strategies and other country- and time-specific phenomena. These factors are influenced by the immediate context of the political debate.

Fundamentally different assumptions regarding the origin of political conflict underlie each of the hypotheses presented. In very broad terms, first, in a political sociological view, ‘real’ socio-economic circumstances are regarded as ultimately determining what happens in politics. For instance, Lipset and Rokkan (1967) point to industrialization and economic growth as a prerequisite for the political conflict between labour and capital (see also: Lipset, 1960). Second, Lowi (1972, p. 309) posits that certain ‘policy conditions underlie political patterns’. The way policies are organized such as by regulation or redistribution, and the language employed in policymaking set narrow contours of political possibilities. The assumption that policy and political arenas are strongly related is also common in country-comparative studies of historical institutionalists. They highlight the way in which national path dependencies create distinct national ‘types’ of policy environments and associated politics (for example Thelen, 2004). Third, in contrast to these socially and policy deterministic views, others emphasize independent dynamics of politics: political conflict as mainly structured through the interaction of political actors in a relatively ‘independent’ political domain. Central to this view is that ‘the reduction of the number of conflicts is an essential part of politics’ (Schattschneider, 1960, p. 64), and that political parties are in the continuous business of strategically selecting and deselecting issues: Party strategy determines the character of politics.

Political classification of immigrant groups

Several immigration-related collective (political) identities or social categories may be expressed in the public sphere. We follow Koopmans and Statham, 1999 and Koopmans et al, 2005 who identify four types of categorization: administrative status categories, racial categories, religious categories and categories drawing on country of origin or ethnic identities.Footnote 2 The first category includes undifferentiated identities, such as being an immigrant or belonging to a minority group, but also administrative status like being an asylum-seeker or a foreign citizen. The second category covers racial identities such as Black or Asian, and refers to differences based on blood relations. It has an Anglo-American tradition, considered politically correct in the United Kingdom (for example, Aspinall, 2002), but regarded inappropriate in large parts of continental Europe. The third category highlights the religious affiliation of immigrants. Because of our focus on claims about immigration and integration, we reduce this category to Muslims in the empirical analysis. In the fourth category, ties with the country of origin are the basis for political mobilization. In sociological terms, we refer to ‘categories’ rather than ‘groups’ (Jenkins, 1997), although we acknowledge that the two may be reinforcing to some extent. While we formulate our expectations specifically for Muslim migrants, at an abstract level all expectations should also hold for the other categories mentioned. For instance, the specification of ‘asylum seekers’ rather than any other category may be more likely in countries with relatively large numbers of migrants, with ‘civic’ citizenship policy traditions and with strong anti-immigration parties. For sake of clarity and brevity, we do not fully develop such a non-Muslim specific framework but we discuss and present a brief example of such a model in the supplementary material.

The first reason why immigrant categories differ can be found in enduring social trends. Here we assume that ‘real-world’ differences between societies or economies ultimately underlie political developments. Politics varies between countries because it is related, directly or via ‘grievances’ expressed through public opinion, to different socio-economic circumstances (Lipset, 1960; Kriesi et al, 2012). This political sociological argument can also be applied to immigrant groups, namely that the absence, presence or size of immigrant groups can be expected to influence whether they are politicized. Without immigrants there can hardly be political contestation of immigration issues and immigrant categories. All the countries included in this article are ‘receiving countries’, albeit at different rates and with very different immigration histories. The demographic composition of the immigrant population also varies significantly across the countries studied, notably also with regard to Muslim immigrants (Pew Research Center, 2011, pp. 121–124). The different immigration histories and resulting composition of the immigrant population should produce distinct patterns in the occurrence of Muslim identities.

Population hypothesis:

  • The size of the Muslim immigrant population in a country is positively associated with the proportion of claims about Muslim immigrants.

A second hypothesis deals with country differences in citizenship policy types. All social categories and identities are regarded as part of broader discursive, political, historical and institutional structures that facilitate or restrict the use of one category over others. It is assumed that on-going status-quo policy programmes determine the contours and language of political debates rather than anything else. Koopmans et al (2005, p. 142) find a ‘continued relevance of national integration politics’ and substantial cross-national differences in the use of immigrant categories (Koopmans and Olzak, 2004; see also: Koopmans and Muis, 2009). These immigrant categories may be linked to citizenship policies or ‘models of integration’ (limitations of which are discussed, for example, in Joppke, 2007; Bertossi, 2011). These models persist over time and are historically institutionalized legacies of earlier policy decisions (Duyvendak and Scholten, 2011, p. 333). Koopmans (2007a, pp. 701–702) notes that the differences in the collective identities on which migrants mobilize ‘are largely independent of the composition of the immigrant populations, and rely instead on the dominant discourses on citizenship and immigrant integration, which emphasize national origin in Germany, race in Britain and delegitimize differentialist identities in France’. Accordingly, the public expression of religious identities should be associated with the extent to which citizenship configurations allow for the expression of group right (‘cultural pluralism’ rather than ‘cultural monism’, Koopmans et al, 2005, pp. 51–73, p. 115). This is facilitated in the Netherlands, has some policy legitimacy in the United Kingdom, but is not recognized in Switzerland.Footnote 3

Citizenship hypothesis:

  • Religious identification is more common in countries with ‘cultural pluralist’ citizenship traditions, such as the Netherlands, than in countries with ‘cultural monist’ traditions, such as Switzerland. We therefore expect relatively more claims about Muslim immigrants in countries with cultural pluralist traditions than in countries with cultural monist traditions.

We focus on differences in policy traditions rather than discursive structures (for example, Medrano, 2003), because discursive structures are very difficult to disentangle from the actual categories of immigrant groups (but see Cinalli and Giugni, 2013). This means that while conceptually different, the independent and dependent variables are in our view empirically too closely connected to be differentiated in a meaningful manner. In addition, we focus on ‘citizenship tradition’ as a part of the generic opportunity structure rather than specifying opportunities per actor-type such as immigrant organizations or anti-immigrant parties. We think that actor-specific opportunity structures are not immediately useful for the question at hand, given that there are also different opportunities within actor types. As a consequence, such distinctions are difficult to consider in the empirical analysis.

Our third hypothesis considers strategic considerations of political actors who seek expressions that are favourable to their political positions and arguments. There are at least three relevant strategic mechanisms available to them. First, political actors need the news media for ‘validation’ or ‘standing’ as recognized actors, as Gamson and Wolfsfeld (1993, p. 116) write: ‘receiving standing in the media is often a necessary condition before targets of influence will grant a movement recognition and deal with its claims and demands’. This mainly applies to (new) social movements but probably also applies to (new) political parties, such as anti-immigrant parties. The validation of immigrant groups and anti-immigrant parties is a prerequisite for polarized politics on the issue – without actors, no politics. Second, political actors pursue a ‘framing’ strategy aimed at ‘structuring the world so you can win’ (Riker, 1996, p. 9). Actors emphasize certain attributes of issues and de-emphasize others in such a way to gain maximum public support for their argument. Third, as noted by Schattschneider (1960, p. 2), ‘the central political fact in a free society is the tremendous contagiousness of conflict’. This implies that when some actors are in conflict – for instance, anti-immigrant parties and pro-migrant groups – other actors will easily get involved, such as major government parties. These other actors are likely to use the pre-existing terms of the debate and may adopt positions close to potential ‘winners’ of the debate (Van Spanje, 2010).

The most important actors in this regard are political parties. We know that party strategies regarding the categorization of immigrant groups are shaped by the nature of party political mobilization and party political relationships within the party system (Green-Pedersen and Krogstrup, 2008; for example Van der Brug et al, 2009). While several aspects of the party system play a role – including its dimensionality and polarization – the electoral success of issue owners, in this case anti-immigration parties, seems to be a critical factor. Their presence in national legislatures should make it more likely that immigration is politicized in general (Dolezal et al, 2012), and more likely that Muslim immigrants are considered a distinct category within the immigrant population.

Anti-immigration party hypothesis:

  • The electoral success of anti-immigration parties can be expected to be positively related to the proportion of claims about Muslim immigrants.

Major political and non-political events, abroad and in the country, can affect the ways in which immigrant groups are referred to in public debates (for example, Olesen, 2007). In policy studies, this would be referred to as events that trigger ‘punctuated’ political change (Birkland, 1997; Jones and Baumgartner, 2005; Walgrave and Varone, 2008). This is especially the case for events with great resonance in public and political debate. A corollary is that violent Muslim extremism in New York in 2001 and later in Madrid and London provided a trigger for political actors to differentiate among immigrants, and to specifically focus on Muslims as a subgroup of immigrants. This can mean a punctuation of policies or new wave of politics on integration issues. In all countries under consideration, we should therefore find fewer references to Muslim immigrants before 2001 than after (Dolezal et al, 2012).

Terrorism hypothesis:

  • It can be expected that there are more claims about Muslim immigrants after 2001 than before 2001.

We argue that these potential explanations are mediated by the context of specific political debates, which limit the immigrant categories that can plausibly be used. For instance, it normally makes little sense to talk about illegal migrants in the context of political debates on civic integration policies, because illegal migrants are usually excluded from such policies. The first mediating factor relates to the topic at hand: We expect more frequent use of status categories on migration topics, and more frequent use of religious categories in claims about civic integration. The arguments also matter: For similar reasons we expect that the arguments used to justify the position of the political actors are related to immigration classifications (for example, Helbling, 2014a).Footnote 4 When political actors provide instrumental, pragmatic and usually policy-related arguments, we expect them to refer to policy status categories. In cases where political actors justify their position with reference to their own or others’ identity or culture, they are likely to use religious or ethnic migrant categories. This resembles the typology used by Cinalli and O’Flynn (2014) where they differentiate types of arguments, such as the more inwardly directed identity arguments and arguments that acknowledge the other like appeals to the common good. We also include the type of claims-maker as a mediating factor. We expect government actors to use administrative categories more often, whereas political parties and other actors probably use more society-oriented categories, such as those associated with religion, country of origin or race/ethnicity. While these are rather broad categories, we do not have an a priori reason to expect other specific types of actors to use certain classifications.

We acknowledge that the three outlined mediating factors associated with the direct political context merit fundamental theoretical grounding including in-depth specification of causal mechanisms. Our theoretical focus is on variation in the broader country context rather than proximate factors associated with the political debate. Proximate factors alone will probably produce relatively underspecified models and ultimately provide unsatisfactory answers to our research question. For instance, if the policy topic at hand strongly determines Muslim-oriented claims-making, we will immediately have to deal with the follow-up question, why topics vary between the countries studied. Furthermore, we find that the theories at the ‘meso-level’ of the political debate are relatively underdeveloped, making it more difficult to deductively design the research. We therefore rely on inductively identified categories, and in the conclusion we discuss this as an area for future research. Figure 1 summarizes the expected relationship between the factors outlined.

Figure 1
figure 1

Conceptual model of immigrant categorization.

Full size image

Data and Methodology

We use a large-scale claims analysis of newspapers in seven countries (Austria, Belgium, Ireland, the Netherlands, Spain, Switzerland and the United Kingdom) between 1995 and 2009. This is a period when immigrant numbers – including Muslim immigrants – increased substantially in some of the countries covered, particularly Spain and Ireland. As a result, the data include sufficient variance to study the dynamics of claims-making, and are likely to allow inferences beyond the time covered. These countries have varying proportions of Muslim immigrants, distinct citizenship traditions and differ in the electoral success of anti-immigrant parties. The countries thus provide variance in the explanatory variables outlined in the preceding section. For instance, the share of Muslims in the population is relatively low in Ireland and Spain, while their share in Austria and the Netherlands is relatively high. With presence in all countries, the potential for politicizing Muslims is given – especially at times when immigration and integration are high on the political agenda (see Van der Brug et al, 2015 for detailed country descriptions and patterns of politicization). The selection of the time period makes it possible to compare patterns of claims-making before and after 2001.

Our approach to data collection is similar to other political claims analysis projects such as MERCI (Koopmans et al, 2005), EURISLAM (Carol et al, 2009), LOCALMULTIDEM (Cinalli and Giugni, 2007; Cinalli and Giugni, 2011) and EUROPUB (Koopmans and Statham, 2010). Similar to these projects, we refer to political claims-making as the ‘the purposive and public articulation of political demands, calls to action, proposals, criticisms, or physical attacks, which, actually or potentially, affect the interests or integrity of the claimants and/or other collective actors’ (Koopmans et al, 2005, p. 254). Political actors continuously make claims. These could be observed, for example, in police records, press statements or parliamentary debates. Practically all studies, however, rely on newspapers to document claims by various types of political actors (for example, Beyeler and Kriesi, 2005; Koopmans, 2007a, 2007b; Leifeld and Haunss, 2012). Earlier studies point out that claims selection bias is a relatively limited problem (Earl et al, 2004; Mügge, 2012). We include claims irrespective of their form or the actor who makes them, but in the analysis focus on claims where an object actor is specified.

We use a sample of newspaper articles from both quality and tabloid newspapers,Footnote 5 drawn using the same random sample of 700 days in each of the countries studied, yielding 7029 claims. The articles were selected by browsing through the physical newspapers and the coding procedure was centralized (coders from different countries were trained collectively and could coordinate their coding decisions among themselves). The number of claims per country ranges from 614 in Ireland to 1319 in the Netherlands, with a mean of 1004 claims (see Berkhout, 2012 for a discussion of inter-coder reliability).

As is common in claims analysis, we make a distinction between the subject actor (that is, the claims-maker), the object actor, the topic of the claim, as well as the frame used to justify the claims. The subject actor is the organization or its representative making a claim about immigration and integration. The object actor describes the group potentially affected were the claim to become reality. This group can be a specific immigrant group or any other specific section of the population (but can remain unspecified as in claims assumed to affect ‘the public’ in general). The subject actor may relate to the object actor in a positive or negative way. If a claim is classified as being about a religious group, this means that were the particular claim to become reality, a sub-group based on religion would be affected. To illustrate this, a newspaper article noted that regional Green parties see racism behind a dispute regarding the wearing of headscarves. In this article, the Green parties are the subject actors, Muslims wearing the headscarf are the object actors and with racism moral principles are invoked as justification. The claim concerns civic integration.

Of course members of groups identified by religion also have a particular legal or administrative status, but the claims-maker in this instance chose to highlight religion. Put differently, the claims in the news are used as evidence of how immigrant groups are constituted, because in each instance that a claims-maker speaks of Muslim immigrants, he or she could also have referred to them in terms of country of origin (for example, Turkey), residence permit (for example, a person with a short-term permit), being a foreign citizen or indeed in a way that does not identify the person as an immigrant (for example, a young mother). It is in this sense that we can talk about immigrant and minority groups being politically constructed, with claims-making reflecting active choices.

We focus on claims about immigration and integration in a relatively broad sense. This includes government activities about the entry and exit of people from the country, the general policy direction, issues of border control and actions related to illegal entry. We also cover social, cultural and economic conflicts, as well as issues related to social cohesion if they involve immigrants. In this context, we cover policies on targeted integration, language and citizenship programmes, and issues on how immigration affects existing policy. Activities, problems and social contributions of immigrant communities are included.

Findings

Claims-making about immigration and integration is dominated by government and party actors (Van der Brug et al, 2015). As can be seen in Table S1 in the supplementary material, government actors tend to make most claims about immigration and integration, and they tend to make claims that are positive about Muslim immigrants. By contrast, political parties tend to make more negative claims about Muslim immigrants. In the United Kingdom and Ireland journalists are more common as claims-makers than in the other countries. Non-governmental organizations and other civil-society organizations are also important claims-makers in all countries covered. With just 1 per cent of claims in the news, anti-immigrant actors play only a marginal role in claims that appear in newspapers (compare Meyer and Rosenberger, 2015). Similarly, with 2 per cent of claims, Muslim organizations are not among the major claims-makers on immigration and integration, not even in claims where Muslim immigrants are affected – in which case 16 per cent of claims are made by Muslim organizations.

We begin with a description of immigrant categorization by country and over time, showing how the different claims-makers jointly refer to different immigrant groups. This provides the context for the analysis in the subsequent section on how Muslim immigrants appear in claims. Table 1 gives the percentage of claims by object actor; the first column in the table gives the distribution of claims about the different groups for all seven countries jointly. Claims by all actor types are combined, and most claims are made about administrative status groups (first four rows of the table). Claims referring to religion, race/ethnicity or country of origin are far less common. Highlighted in grey is the use of religious categories, the variation of which we treat in the subsequent section. To some extent the numbers in Table 1 are in line with the specificities of the migration profile of the countries under study (Van der Brug et al, 2015). While the use of administrative categories is dominant in all countries, it is the relative attention to the administrative subcategories that seems related to immigration patterns. We find, for instance, large proportions of claims about ‘illegal migrants’ in Spain and about ‘labour migrants’ in Ireland. The same numbers also indicate some support for the idea that citizenship traditions affect immigrant identities. We find, for instance, a high proportion of race/ethnicity classifications in the United Kingdom. The relatively low percentages for religion indicate that for non-Muslim immigrants, religion tends not to be a salient characteristic used by claims-makers. Indeed there are only very few claims with reference to other religions present in our sample of claims (N=17).

Table 1 Percentage of claims about different immigrant groups
Full size table

Figure 2 shows how the use of Muslim immigrants as object actor in claims-making has changed, showing three periods of 5 years. With the exception of Spain and Ireland, we see the proportion of claims about religious groups increase substantially over the years. For example, between 1995 and 1999, just 1.5 per cent of claims in Austria concerned religious groups. Between 2005 and 2009, this proportion has increased to 13.1 per cent of claims: a ninefold increase. Britain differs from the other countries in that ethnic and racial categories are also used regularly, especially in the first period covered here. Initially, between 2000 and 2004, we observe a marked increase of administrative status categories in Britain, and more recently an increasing importance of claims about religious groups. In the third period, there are more claims about religious groups than about ethnic and racial groups, indicating that the developments in Britain increasingly resemble those of the other countries under study.

Figure 2
figure 2

Proportion of claims about Muslim immigrants by country and over time.

Notes: Excludes claims with no recorded object actor, 5-year periods are averaged (1995–1999, 2000–2004, 2005–2009).

Full size image

The changes over time in Figure 2 highlight a dynamic nature not apparent in country typologies. The rise of Muslim immigrants as a category of immigrants is not in line with the categories that are commonly associated with different traditions of citizenship policies. The figures suggest that there is indeed merit in assessing the influence of the factors discussed in the theory section.

Figure 3 shows the proportion of claims with Muslim immigrants as object actor by the factors mentioned in the conceptual model in Figure 1. The mediating factors are shown in the upper half of the figure. These are variables that are associated with the immediate context of the political debate. The bars in the lower half refer to country and time differences. There is substantial variation in all factors of interest, generally confirming the expectations outlined above. To start with the mediating factors, government and parliamentary actors are less likely to make claims about Muslim immigrants than other actors. We generally record only very few claims by anti-immigrant parties (supplementary Table S1), and do not report them separately so as not to misinterpret small numbers. Of all government claims, around 5 per cent have Muslim immigrants as object actors, compared with 9 per cent for other actors. This is a significant difference in the propensity to make claims about Muslim immigrants as can be seen by the confidence intervals that do not overlap. It confirms that government actors are more likely to use administrative categories than other actors.

Figure 3
figure 3

Proportion of claims about Muslim immigrants by claims-maker, argument, topic, country and time.

Notes: Given are 95 per cent confidence intervals of the mean (n=7114). Anti-immigrant parties are not shown separately because they are responsible for only a very small proportion of claims (compare Table 1).

Full size image

When actors use arguments of collective identity, in 19 per cent of the cases they refer to Muslim immigrants. This is far more than when claims are justified with other arguments and strongly supports that a political justification referring to identity also leads to group classifications based on cultural identity, in this case Muslim. In claims made about immigration – as opposed to integration –, we find far fewer claims about Muslim immigrants. The results indicate that studies of ‘new’ political conflicts in Western Europe need to use a broad range of topics under migration issues or risk serious under-specification of the causes and effects of such new conflict dimensions. As also visible in Table 1, there are substantial differences between countries in the use of Muslim immigrants as a political category in political claims. The lowest two bars in Figure 3 show that there is a substantial increase in the claims about Muslim immigrants after 2001. The strong variation in the variables related to the immediate context of the political debate (actors, arguments, topics) indicates that country differences, such as citizenship configurations, only partially explain the types of migrant categories used.

In Table 2, we present five logit regression models in order to assess the effect of all variables simultaneously rather than relying on bivariate associations. The dependent variable is the use of Muslim immigrants as object actor in political claims. Models 1–3 include factors related to the immediate political debate. In Models 4 and 5, we add contextual country and time-level variation. With data on only seven countries, we have insufficient statistical leverage to assess country-level variables. We therefore capture country-level variation through country dummy variables, and below assess whether the country differences observed in these models are associated with the country characteristics hypothesized to cause these differences (namely the population of Muslim immigrants and the seat share of anti-immigrant parties). This approach is more adequate than cross-tabulating country proportions of claims about Muslim immigrants because it allows us to account for variation caused by other explanatory variables associated with actors, topics and arguments used.

Table 2 Claims-level logit regression on Muslim immigrants as object actor
Full size table

Model 1 shows that, compared with the reference category government actors, other organized actors are more likely to refer to Muslim immigrants. While statistically significant, substantially this is only a small effect: The type of actor explains only a very small part of the variation (pseudo R 2=0.01). In Model 2, we add the types of arguments used. This increases the explanatory power to 0.06. As mentioned earlier, the use of identity arguments increases the likelihood that actors make claims about Muslim immigrants. This is a substantial effect and clearly shows that ‘argumentation’ is a central instrument for political actors. Following Cinalli and O’Flynn (2014), this can be interpreted as inwardly oriented arguments being important in the case of Muslim immigrants, but our analysis does not allow for a more in-depth exploration of this view. Studies like theirs are necessary to differentiate between different claims in terms of quality and impact – something the data at hand does not allow. In Model 3, we additionally account for the topic discussed. If the topic of a claim is civic integration, it is far more likely that Muslim immigrants are referred to than when immigration issues are debated. The intermediate variables derived from the properties of the claim itself explain jointly about 19 per cent of the variation in the propensity to make claims about Muslim immigrants.

Model 4 also includes country dummies, and the difference in the country coefficients is broadly consistent with Figure 3. Austria is an exception: In Figure 3, Austria ranks below Belgium in the use of claims about Muslim immigrants, whereas in Model 4 it ranks above Belgium. This suggests that the differences between countries in Figure 3 are partly caused by the nature of the debate (actors, arguments and topics); after controlling for these factors, country differences become more pronounced. In Model 5, we include a dummy variable to test for a 9/11 effect. Looking at the pseudo R 2, this marginally improves the explanatory power of the model, but the topic of a claim and the country still provide the largest contribution to explaining the use of claims about Muslim immigrants.

Why do we find that in some countries – especially Switzerland and Austria – claims are more likely to be about Muslim immigrants than in other countries? Figure 4 and Figure 5 present the coefficients of Model 5 as bar charts, ranked by the proportion of Muslim immigrants in the population and by the seat share of anti-immigrant parties, respectively (supplementary Table S2).Footnote 6 We have selected Ireland as the reference category because, with a relatively small Muslim community and no anti-immigrant party interpretation is made simpler. Figure 4 suggests that in countries with a comparatively large share of Muslim immigrants, such as the Netherlands and Austria, we do not find more frequent references to Muslim immigrants than in countries with comparatively small shares, such as Switzerland and the United Kingdom. Especially the Swiss case indicates the weakness of the hypothesis that the number of Muslim immigrants should lead to frequent claims about Muslim immigrants. Despite the relatively small share of Muslim immigrants we find that claims in Switzerland are more likely to refer to Muslim immigrants than claims in other countries. Indeed, over time the Muslim population in Switzerland has remained relatively stable, but claims about Muslim immigrants have increased significantly. By contrast, Figure 5 shows that in countries with substantial anti-immigrant parties in parliament there are strong country effects for the use of claims about Muslim immigrants. With the exception of the United Kingdom, the rank order of the coefficients is the same as that of the seat share of anti-immigrant parties. This strong and consistent effect is in line with the literature on strategies of political parties. It seems that the presence of an anti-immigrant party ‘validates’ the category of Muslim immigrants, though anti-immigrant parties are not prominent claim-makers themselves.

Figure 4
figure 4

Country coefficients of Model 5, ranked by Muslim population.

Notes: Odds ratios of logit model, ranked by average proportion of Muslim immigrant population (in brackets).

Full size image
Figure 5
figure 5

Country coefficients of Model 5, ranked by share of anti-immigrant parties.

Notes: Odds ratios of logit model, ranked by average seat share of anti-immigrant parties (in brackets); anti-immigrant parties are listed in the supplementary material.

Full size image

Discussion and Conclusion

In this article, we examined what kind of immigration-related categories are used in political debates about immigration and civic integration. We are interested in the circumstances under which a religion-based immigrant categorization occurs, using claims-making as evidence for such categorization. We focus on Muslim immigrants since international terrorism and anti-Islamic political mobilization make them the most likely case for differences in categorization associated with new conflict dimensions. This provides us with leverage to assess potential explanations for country differences in the political use of immigration-related categories. On the basis of theories from distinct research traditions, we expected these differences to depend on three factors: on citizenship policies, on the socio-structural composition (the proportion of Muslim immigrants in the population) and on political strategies of political parties – particularly anti-immigration parties.

With regard to citizenship traditions, claims-makers in all countries mostly refer to very broad, administrative, status-related categories such as immigrants or asylum-seekers. They differ in the use of the smaller racial/ethnic, religious and country-of-origin categories. Although we find significant differences across countries, these differences are only partially associated with commonly used citizenship typologies. For instance, Muslim categories are relatively common in the, according to Koopmans et al (2005), moderately cultural pluralist United Kingdom and the culturally monist Switzerland, whereas the limited recognition of group rights in these countries led us to expect prominent use of racial and country-of-origin categories, respectively. This article demonstrates that there are substantial changes over time in that the use of religious categories has increased in all countries under study – except in Spain and Ireland. This suggests that the use of immigrant categories in the public sphere is more dynamic than can be captured in country types that change only very slowly (compare Ruedin, 2015). The observation of similar trends in different countries also suggests that the countries are more similar than any typology would present them (also typologies pertaining to other differences between countries, such as the media system: Hallin and Mancini, 2004). A strong relationship between the policy arena and the political arena could be found for administrative immigration categories, but we do not find that the (non-)recognition by administrative groups spills over into political debates (or the other way around). In other words, while we have no grounds to challenge any typology of immigration and integration policy, we do not find that it is strongly related to political contestation on these issues: Policy and political arenas seem to operate in relative isolation.

Turning to socio-structural factors, we find only partial confirmation of the expectation that the presence of a substantial Muslim minority translates into the politicization of Muslim immigrants. On the basis of Figure 4, our impression is that this works as a threshold effect: In countries with under around 1 per cent of Muslims, such as in Ireland and Spain, Muslim immigrants are rarely the object of claims; in countries with more substantial minorities (more than around one per cent), such as the United Kingdom and the Netherlands, Muslim immigrants are common objects of political claims. Studies on sub-national units may be useful to examine this proposition further. However, the variation in claims about Muslim immigrants in the countries with a substantial Muslim minority is unrelated to the actual size or growth of the Muslim minority over time. This suggests that theories emphasizing the social base of politics provide conditional explanations. A social base is a necessary but not sufficient explanation for, in this case, immigrant categorization.

In countries where anti-immigrant parties have substantial parliamentary presence, Muslim immigrants are more frequently the objects of political claims. This is what can be expected on the basis of the literature on party political strategies (for example, Van Spanje, 2010), and suggests that immigrant classification is predominantly an actor-driven phenomenon conditional on the party system. Anti-immigrant parties seem to drive Muslim categorization when they are formally and informally able to participate in electoral competition. The presence of an anti-immigrant party with ‘standing’ in the news media seems to successfully ‘frame’ migration in identity terms and produce a ‘contagion’ of other actors in the use of Muslim immigrants as objects in claims-making. The contagiousness of conflict affects non-anti-immigrant political parties and actors, in the sense that these actors also make more claims that potentially affect Muslim immigrants. With the small numbers of claims recorded for anti-immigrant parties in the data available, we cannot provide a formal test for the interaction between anti-immigrant and other actors. In more general terms, various strategic and actor-related factors seem to promote (or restrict) the political translation of social differences. If we assume, as Kriesi et al (2012) do, that anti-immigrant parties are a political manifestation of the ‘new cultural cleavage’ between so-called winners and losers of globalization, then this new social cleavage – while perhaps uniting ‘losers’ – divides Muslim immigrants from other immigrant groups.

The analysis of the immediate political context of claims-making suggests that this could be a more fruitful level for theory formation than the country level discussed earlier. At the level of topic fields, we found an exceptionally clear difference between immigration and integration as topics on the one hand, and the categorization of Muslim immigrants as a distinct group on the other hand. There is also a strong correlation between the use of identity frames and references to Muslim immigrants. In theoretical terms, macro-level factors, such as socio-structural and political institutional variables seem to be only part of the story, and need to be combined with lower-level variables such as those associated with the arguments used (for example, Cinalli and O’Flynn, 2014) at the actor level, but also at the level of topic domains. Such a combination would take into account the emergence and decline of issues on the policy agenda of countries; it seems that the agenda determines politics. In future research, it is thus important to carefully incorporate expectations about topic-level variation in political behaviour. As a minimum, one should try to incorporate variation in the topic selected for research, like the difference between immigration and integration (and sub-topics) we used in this article. Following Lowi (1972), such meso-level theory formation has a small tradition in studies of public policy.

Our analysis of the use of Muslims as an immigrant category has shown that the political debate is more volatile, and thus probably more responsive to immediate concerns, than suggested by the sole analysis of changes of policies and policy traditions. It seems that for Muslim immigrants in Europe, becoming a distinct social category is, to say the least, probably a mixed blessing. On the one hand, with becoming a distinct social category comes an increased likelihood of recognition. This can mean increased autonomy in aspects important to the group. On the other hand, this distinct social category is predominantly shaped by actors other than Muslims, meaning that the boundaries and meaning of being Muslim are to a large extent outside the control of Muslims.