How ordinary people view Muslim group rights in Britain, the Netherlands, France and Germany: significant ‘gaps’ between majorities and Muslims?

ABSTRACT Taking four countries—Britain, the Netherlands, France and Germany—with distinct state approaches and public debates over accommodating Muslims, we study the views of ordinary people from the majority and Muslim populations on Muslim group rights. We compare their responses to questions on mosque-building, teachers wearing religious symbols, and religious classes in schools, to determine whether there is a significant ‘gap’ between the majority and Muslim minorities. We find highly significant ‘gaps’ between the majorities and Muslims over Muslim group rights in all countries, with the majorities less supportive. Importantly, it is a shift by the majority population against Muslim group rights that produces this ‘gap’ as the question moves from provision for Christians to Muslims, while Muslims hold similar views over rights for Christians and their own religion. In Britain and Germany, the two countries where church/state relations privilege Christianity over other religions, majorities especially support Christian over Muslim group rights. The British findings are remarkable, because a country which substantially grants and has the most supportive public debate for Muslim group rights, produces the largest ‘gaps’ between its majority and Muslims. We think this is due to political context, where in contrast to the Netherlands, there is no outlet for political opposition to Muslim group rights.


Introduction
Across Europe, and in contrast to North America, states' attempts to accommodate Muslims are the definitive feature of 'multicultural' controversies (Joppke and Torpey 2013;Koopmans 2013). This has a factual basis. 1 It also places the state's efforts to accommodate Islam as a minority religion at centre stage. Much research has focused on cross-national variations in the legal and institutional incorporation of Islam (Fetzer and Soper 2005;Joppke 2009), and whether religion is a 'bridge' or a 'barrier' to integration, to use Foner and Alba's (2008) formulation. Some argue that Islam is a 'barrier' and source of conflict because European societal institutions and national identities remain anchored to an important extent in Christianity and do not make equal room for Islam (e.g. Zolberg and Woon 1999, 7;Foner and Alba 2008, 374). Charles Taylor makes a similar claim about the Christian foundations of liberalism (1994,62): 'as many Muslims are well aware, western liberalism is not so much an expression of the secular, postreligious outlook that happens to be popular among liberal intellectuals as a more organic outgrowth of Christianity'. Against this, other authors (see especially Joppke 2009) have documented how European liberal nation-states notwithstanding their different institutional traditions for accommodating religions and their distinct national identities, have importantly extended Muslim group rights, largely because their legal and constitutional institutions uphold liberal norms, often in the face of public opposition. For Joppke and Torpey (2013, 141-142), public sentiments should not be confused with public institutions, which due to the prevalence of liberal norms cannot operate unchallenged on an ethnocentric basis, so that: 'religion, particularly Islam, may still be more "barrier" than "bridge" to including immigrants in Europe but only as a matter of mentalities, not of institutions'.
So far, there has been virtually no research on how the institutional accommodation of Islam, mediated through political controversies, relates to the views of ordinary people living in that country. While different degrees of institutional accommodation have taken place, and public controversies have occurred in all countries, though to different degrees and focusing on different issues, how does this relate to what people from the majority and Muslim minority populations think about extending Muslim group rights?
This article is one of the first to address the important question of what ordinary people think about the extension of group rights to Muslims by recourse to original empirical evidence on their opinions and on political contestation that they see in public debates. Group rights exhibit two features: first, if granted the group right goes beyond the set of common civil and political rights of individual citizenship that are protected in all liberal democracies; second, if realised, the group right constitutes the recognition and accommodation by the state of the distinctive identity and need of the minority group . Examples include policies that allow exemptions from existing rules and obligations, state support for separate institutions, special facilities in mainstream public institutions, such as schools and media, representation rights for ethnic and religious organisations, and affirmative action programmes. The study is cross-national comparative, including the Netherlands, Britain, France and Germany. These four countries have the most sizeable Muslim populations of immigrant origin in Western Europe (Buijs and Rath 2002) and importantly, have implemented distinct state approaches for extending rights to Muslims based on their specific church-state relations (Koenig 2007).
We examine how ordinary people respond to questions on issues about mosque-building, teachers wearing religious symbols, and religious classes in schools. Our primary aim is to examine whether there are significant 'gaps' in opinions between the majority populations and Muslims themselves. This matters because it indicates whether there is consensus or a potential for conflict along Muslim/non-Muslim lines among ordinary peoples' perceptions. As van der Noll and Saroglou state (2015,221): 'Studying public opinion … gives an indication of exclusion and oppression in everyday encounters, as well as whether there is a basis for current and future legislation concerning the accommodation of Islam.' It allows us to see whether a strong line of demarcation is constructed by the ways that majorities and Muslims think about these issues that might impact upon the social relationships between them. Of course, mutual tolerance works from both sides. So it matters where the opinions of Muslims and majorities stand relative to one another over the issues, and in relation to the degree and form of institutional accommodation of Muslim group rights in their respective country.
Given the important cross-national variations in the degrees and ways the state has accommodated Muslim group rights in the four countries, we examine the majority/Muslim 'gaps' within countries and interpret our findings in relation to their specific national contexts. An important part of this context is provided by the national public debates over Muslim group rights. First, public debates carried by the mass media are the location for controversies over Muslim group rights that result when political actors challenge a state's policy. Second, the conceptual and cognitive frameworks that ordinary people have for interpreting group rights will be shaped, but not determined, by public debates, because this is the agenda-setting supply-line of political information that is publicly visible and accessible in a society. 2 We undertake an analysis of national public debates in order to have a contextual indicator for the degree and form of contestation over Muslim group rights that is visible to people in their respective country and that may potentially shape their opinions. In this way, we interpret our findings on the 'gaps' between majorities and Muslims, by relating them back to a state's policies for granting Muslim rights, and the visible public controversies between political actors over them.
In the next section, we outline the basic differences in their respective church-state relations that have provided the institutional frameworks for accommodating Islam. After this we undertake a claim-making analysis of public debates over group rights for Muslims, in order to determine the degrees and forms of contestation in the countries. This matters because it indicates how politics in the public domain mediates the issues of Muslim group rights, so that they are visible to ordinary people. Especially important is the degree of public support or opposition for Muslim group rights from actors representing the state, political parties and majority civil society. Then we turn to the primary focus, which is an analysis of responses by members of the public to survey questions on the presence of minarets, whether teachers should be allowed to wear (Christian or Islamic) religious symbols in schools, and whether religious classes should be allowed in schools for Christians or Muslims. Here we examine the 'gap' in opinions between the majority and Muslims within each country. We take this as an indicator for the socio-cultural distance between the groups over the extension of Muslim group rights. In addition, our Muslim sample is composed of four groups with origins in different countries (former-Yugoslavia; Pakistan; Morocco; Turkey) to see if there is variation among Muslim minorities. The cross-national comparative research design allows us to interpret the findings on public opinion in relation to their distinct national institutional approaches for accommodating religious rights for Muslims, and their respective public debates for legitimating them. Finally, the concluding discussion accounts for the 'gaps'.

State accommodation of Muslim group rights
Religion is centrally important to understanding European countries' controversies over their attempts to 'integrate' Muslims (Bader 2007;Foner and Alba 2008;Joppke 2009;Laurence 2012;Brubaker 2015). First, although European societies consider themselves broadly secular, Christian religions play influential institutional social and political roles, that matter irrespective of the actual number of practicing worshippers. These institutional arrangements of church/state relations define pre-existing conditions and the political environment into which immigrant religions have to negotiate a space for their community. Second, religious identification is a belief system that can shape an individual's core identity, opinions and political behaviour. A religious migrant may consider practicing religion as a sacred duty that cannot be compromised. While states consciously try to shape civic identifications through integration policies, they do not to the same degree for migrants' religious identifications, not least because the liberal state upholds individual freedom of religious practice. Third, the nature of the immigrant religion is likely to influence the degree to which migrants adapt or resist when faced by the dominant culture. In this respect, the public duties of worship that are associated with Islam can be more obtrusive and visible, and less easy to accommodate within the public life of a western society, than those of immigrant religions where worship takes place mostly in private.
The sizeable literature on cross-national variations emphasises how historical resolutions of church-state conflicts have shaped the accommodation of Islam as a minority religion in Europe (Fetzer and Soper 2005;Statham et al 2005;Koenig 2007;Soper and Fetzer 2007;Laurence 2012). There is considerable agreement on the prominent features that define the church/state separations in our four countries. France, the Netherlands, Britain and Germany represent four distinct institutional types of religious accommodation, here understood as 'opportunity structures' (Koopmans and Statham 2000), that importantly influence the degree and form to which Islam has been incorporated through an extension of rights to Muslims. Here we briefly present the basic cross-national variations in the countries' institutional traditions for church/state relations, to provide context for informing the subsequent analyses.
France, under the concept of laïcité, is the archetypal secularist case, where there is a strict statechurch separation that provides little space for religion in public life and institutions. The French state and public institutions are committed to secularism and even the role of Christian churches is restricted in the public sphere. As a consequence of laïcité, the French state has been resistant to the idea of separate institutions, such as schools, for religious groups, and displays of religiosity in public environments. In France there is a broad consensus that civil servants should not display visible markers of religion. In a context where even Catholic institutions, such as schools, receive far less state support and recognition than elsewhere, Islam has faced difficulties in finding an institutional foothold within a restrictive state framework (Laurence 2012).
The Dutch case stands in stark contrast to French laïcité. In the Netherlands church/state relations evolved around the logic of 'pillarisation' as a consequence of ideological struggles between Catholic, Protestant and secular groups in the late nineteenth century. 'Pillarisation' consists in a denominational segregation of society where religious or ideological groups have the right to establish their own social infrastructure with state support. Hence Dutch church/state relations follow a tradition for: noninterference of the state in religious self-governance, which was broadly defined to include religious schools, hospitals, cultural and welfare institutions, and a range of other sectors. The compromise also entailed full state funding-on an equal basis for all denominations-for these sectors. (Koopmans 2013, 155) Although actual 'pillarisation' died out in the 1960s, its imprint is still influential as a logic embedded in institutional arrangements and law, and serves as a reference point for how to accommodate Muslim rights (Carol and Koopmans 2013, 171). In this context, Muslims and other 'newcomer' religious minorities have found it relatively easy to claim group rights granted to other religious denominations, while the state has traditionally refrained from preventing the expression of minority religions in public institutions.
Britain has an official state Christian church that is privileged over others. The Church of England is led by the Queen, as Head of State, while more than 20 Anglican Bishops sit in the second chamber the House of Lords. In Britain, the rights and privileges that the state grants the Church of England are not automatically extended to other religious groups. Nonetheless, a pragmatic form of accommodation has proceeded, with the state relatively willing to grant rights to newer religions, a process that is supported in a paternalistic way by the Church of England: 'importantly, the presence of an established church and its close links with politics and public policy in Britain encouraged Muslim groups to look to the state for recognition of their religious rights and public policy needs' (Soper and Fetzer 2007, 936). Overall, this has provided some considerable degree of parity between religions over time, while elites have been relatively supportive of demands to extend rights to Muslims on an equal basis to other minority religions, though full parity has not been achieved (Fetzer and Soper 2005;Statham et al 2005).
In Germany, state recognition is extended to several Christian (especially Catholic) and Jewish religious denominations as public corporations, a formal status that entails privileges, including to receive Church taxes collected by the government, organise religious education (RE) in state schools, and provide social welfare services (Soper and Fetzer 2007). Crucially, the German state has so far not been willing to extend the public corporation status to Islam that it has afforded the Christian and Jewish denominations. This requirement for formal status has proven to be a barrier that has made it relatively difficult for Muslims to gain group rights (see e.g. Laurence 2012; Carol and Koopmans 2013). In addition, a strong imprint of Christianity remains in German liberalism that has been less accommodating than the establishment Church variant in Britain. Joppke makes this point with regard to the banning of the headscarf in Germany (2009,123): This is a case where Muslims are really excluded from a certain 'Christian-occidental' self-definition of the state, simply because one cannot be Christian and Muslim at the same time. This is the identity that transpires in the headscarf laws of the Catholic-conservative Länder.
Another factor in Germany is the difficulty for migrant minorities (and Muslims) to gain access to formal citizenship relative to France, Britain and the Netherlands . Muslims are less able to demand parity of treatment to others when significant numbers of them remain formally non-nationals and lacking in political leverage.
We see the distinct church/state relationships have been institutionalised and normatively justified as 'opportunity structures' (Koopmans and Statham 2000) that shape, though do not predetermine, to what degree and how countries have been willing to grant religious rights to Muslims. In this sense, France, the Netherlands, Britain and Germany offer four different models of institutional and discursive frameworks ('opportunities') for Muslim group rights. Comparatively, the Dutch and the British church/state traditions have been more open to accommodating Islam as a new minority religion. In particular, the Dutch form of group-based pluralism inherited from 'pillarisation' allows a greater and relatively equal recognition of minority religions and favourable opportunities for Muslim groups to stake their claims. Britain's relative openness and accommodating approach is more elite-led and top-down with the aim of avoiding conflict with religious minorities. The Church of England has importantly supported Muslim rights, but retains its position at the top of the hierarchy. In Germany, a combination of high formal barriers to state recognition and status and relatively low political influence due to high barriers for migrant populations to gain access to citizenship is not conducive for Muslims. While in France, Muslims face very high barriers to religious rights, because Christian churches have historically been denied many privileges and rights within a context of strict state secularism. This is supported by the empirical findings of the Immigrant Citizenship Rights Indicators (ICRI) which uses a battery of 14 indicators for each country's policies for extending rights to Muslims-for example, allowing Islamic religious classes in state schools; allowing Islamic call to prayer-that are coded on a five-point scale (−1 to +1, with +1 indicating the most accommodating policy) (see, Carol and Koopmans 2013). According to these ICRI indicators, in 2008, the Netherlands (0.79) and Britain (0.64) were relatively expansive for granting religious rights to Muslims, while Germany (−0.11) and especially France (−0.25), were relatively restrictive (see also, Koopmans 2013, 156).
In sum, there are two important dimensions of cross-national variation: the degree to which Muslim group rights are accommodated; and the degree to which Christian religions are privileged over others. First, the Netherlands and Britain have been relatively much more accommodating to Muslim group rights than France and Germany. The second important variation is that in Britain and Germany, the Christian churches are substantially privileged relative to other religions. By comparison, the Netherlands and France treat religions in a relatively more equal way, notwithstanding that the Dutch are inclusive and the French exclusive.

Political contestation over Muslim group rights in public debates
The relationship between a state's institutional and legal approach to accommodating Islam and the way that ordinary people view the issues is not direct but mediated. The public debates carried by the mass media are an especially important source of communication that shapes, but does not determine, how people interpret public problems. 3 Here we compare the national public debates, so that we have an indicator for the contestation between political actors over Muslim group rights that is visible to ordinary people. We apply the same public claim-making approach (Koopmans and Statham 1999) as the large body of research that uses mass-mediated public debates as a primary source of data for analysing political contestation (see also, Koopmans et al. 2005;Koopmans and Statham 2010;Statham and Trenz 2013).
Given the high threshold of media selectivity, public debates are dominated by elite and organised political actors. So studying this form of contestation gives an indicator for the cues that ordinary people face from their respective national political arenas, when they interpret and make political sense of a public problem. In this perspective, a country's church-state framework matters, because (Carol and Koopmans 2013, 166), These differential institutional contexts define opportunity structures for claim-making about Islamic religious rights, and explain why certain rights are highly controversial in one country, while they are hardly debated at all in other countries, either because they are consensually accepted, or because they are consensually rejected.
There is not space to undertake a full analysis of public debates, so we draw on earlier findings from EurIslam (Carol and Koopmans 2013), but add a new analysis that compares the overall the degrees and forms of contestation in the four countries' national public debates over Muslim group rights. This matters for our study because it allows us to relate the way that (mostly elite) political actors contest Muslim group rights back to ordinary people's opinions on the same type of issues. Specifically, we want to examine to what degree the states' respective approaches to Muslim rights have faced public challenges, from (a) political party competition and (b) mobilisation by collective actors and groups from civil society. This provides a basic idea of how much, in which direction, and from which arena of politics, a state's orthodoxy for extending Muslim rights has had its legitimacy visibly challenged by political actors.

Approach, method and data
In the claim-making approach, the unit of analysis is a claim-making act: A political claim-making act is a purposeful communicative action in the public sphere. Claim-making acts consist of public speech acts that articulate political demands, calls to action, proposals, or criticisms, which, actually or potentially, affect the interests or integrity of the claimants or other collective actors. (Koopmans and Statham 2010, 55) We use an original sample of 'claims' data about Muslims and Islam drawn from five newspapers per country (varied by political affiliation, left/right, and type, broadsheet/tabloid), 4 for a 10-year period, 1999-2008, prior to the survey fieldwork.
A random sample of articles with 'claims' about Muslims/Islam was retrieved for each country by using keyword searches for Islam*/Muslim*/Moslem*/mosque/imam/Qur'an (Quran, Qur'an, Koran)/headscarf/ burqa (burkha, burka, burqua)/minaret (or their respective equivalents in national languages). Claims were included regardless of the actor type making the claims (governments, political parties, courts, Muslim organisations, employers, churches etc.) and coded by trained research assistants using a uniform codebook. 5 Detailed information is coded on a number of variables for a claim, including: time and place; claim-making actor; action form; addressee; substantive issue of claim; position of the issue for Muslims/Islam; frame. The resultant data set has more than 4000 claims covering a wide range of issues relevant to Muslims/Islam.
For this article, we restricted the sample to 'evaluative claims' . Evaluative claims are where actors try to push the public debate over Muslim group rights in a decisive direction, and are an important indicator for contestation. These are claims where an actor makes a claim that, if realised, clearly expresses support for, or opposition to, an extension of rights for, or the position of Muslims. Technically, this meant excluding all claims coded 'neutral' or 'ambivalent' (0), but including those whose realisation implies some deterioration in the rights or position of Muslims (coded −1), and those implying an improvement (coded +1). From this, we calculate a mean score that indicates an actor's aggregate position over Muslim group rights, ranging from −1 restrictive to +1 expansive, and it is possible to reconstruct each actor's overall position, relative to other actors.

Analysis
For this analysis, we retrieved all evaluative claims on group rights for Muslims/Islam from the samples for Britain, the Netherlands, France and Germany. We use the same definition as Carol and Koopmans (2013, 166-167), whose study uses the same data set: Claims about religious rights then contest entitlements regarding the performance (e.g. to be buried according to Islamic prescriptions) or non-performance (e.g. dispensation from mixed swimming classes) of certain actions for religious reasons, or they are about entitlements that require others to perform (e.g. to create prayer spaces in schools) or refrain from performing (e.g. not to depict the Prophet Mohammed) certain actions for religious reasons.
The study by Carol and Koopmans (2013, 178-180) showed that there are striking cross-national differences in the Muslim rights issues that become publicly controversial and that these relate to the respective state forms of accommodation of Islam. In the relatively more restrictive countries, Germany and France, they found that six-tenths of claims over Muslim rights strongly focused on issues of 'the headscarf' and the presence of mosques and minarets. By contrast, these issues were present but less prominent in controversies in the more accommodating Netherlands and Britain, accounting for about a fifth of claims. In Britain and the Netherlands, controversies about fullface covering (burqa/niqaab) and Islamic schools, as well as a lot of specialised issues, such as sharia law, were present that were virtually absent in France and Germany. Overall, Carol and Koopmans (2013, 186) find that settlements reached over basic rights such as headscarves and mosque construction in the Netherlands and Britain, have not quelled the controversies, but that the debate has shifted onto new issues, and more 'obtrusive' and fundamental forms of demands for Muslim rights, such as full-face covering.
Bearing in mind that these distinct issue contents show that the national debates over Muslim rights occur on relatively different substantive trajectories, we now turn to examine the overall degrees and forms of contestation in the four countries, based on the evaluative claims.
For group rights issues, there is a permanent numerical imbalance between the non-Muslim majority and Muslim minority, which means that the state steps in and has a contradictory role. On one side, the state has duties to protect and uphold the rights and identities of the minority, particularly if they are citizens, while on the other, majoritarian politics is based on the idea that state sovereignty derives from popular will, which legitimates the notion that majority's opinion should be the basis for decision-making. This contradictory role of the state makes it a central actor and target in the contentious politics over Muslim group rights, because it has to make decisions while facing the mobilised demands of Muslims, on one side, and those by political actors from the majority society, on the other. For this reason the decisive political conflict lines to study are the triad of actor-relationships between state-majority; state-Muslim and majority-Muslim. Table 1 shows the basic contours of public contestation over Muslim group rights in the four countries. Actors are aggregated into four basic types: state and judiciary actors; legislative and political parties; 'native' civil society organisations and groups; and Muslim/Islamic organisations and groups. State and judiciary includes all institutional actors operating on behalf of the state administration, government and judiciary. State actors 'go public' to legitimate, justify and re-constitute the 'orthodox' policy stance. Legislative and political parties covers all actors who are acting in their public representative capacity as a legislature or political parties. 6 Contestation over Muslim rights can come in the form of party competition over the state orthodoxy. In 'native' civil society organisations and groups, we include all public actors, organisations and groups that are non-Muslim. This includes Churches, employers, and racist and anti-racist organisations and groups, among many others, and it indicates the aggregate level of support for Muslim rights mobilised by civil society from the majority population. Last, Muslim/Islamic includes organisations and groups that have a collective identity that is explicitly representative of the Muslim community. For each country, the first column shows a collective actor's share, and the second its mean position (−1 restrictive to +1 expansive), in claim-making for Muslim religious rights. In Britain, we see that the state takes a strong advocacy position (+0.80) supporting Muslim rights in a context where these are significantly granted. This orthodox state position stimulates few claims from political parties (7.5%), and is strongly supported by organisations and groups from the 'native' majority (37.6%; +0.83), and also by Muslim organisations and groups (22.6%; +1). Overall, the public debate that is visible in Britain shows few signs of opposition and provides strong advocacy supporting Muslim group rights.
In the Netherlands, where Muslim rights are also significantly granted, the state's position is supportive (+0.37) but less decisively so than in Britain. We see that the state orthodoxy has faced an important oppositional challenge from party politics (25%; −0.44) in the period 1999-2008. Closer inspection shows that this strong opposition came from the Party for Freedom (Partij voor de Vrijheid, PVV). Since 2005, this radical right party associated with Geert Wilders has campaigned strongly on an explicitly anti-Islamic agenda, for example, against the 'Islamisation of the Netherlands', and won 24 seats in the 2010 general election. However, this challenge to the state's orthodoxy on Muslim rights is not replicated by mobilisation of organisations and groups from Dutch civil society. 'Native' civil society organisations and groups mobilise significantly and support Muslim rights (33.3%; +0.42). Finally, Muslim organisations and groups' advocacy (17.3%; +1) challenges the state to do more.
Turning to France, we see that Muslims themselves have a strong presence in the debate and advocate strongly for extension of rights to Islam (39.3%; +0.93). As a result of the relatively open access to national citizenship for migrants in France , there a large number of French citizens who are Muslims. Possessing citizenship seems to empower French Muslims to make demands for an extension of rights for their community. Also, in the context of a state orthodoxy of laïcité that is restrictive to religious rights, we find little overt opposition to expanding Muslim rights. The French state (19.2%; +0.31), parties (11.9%; +0.63) and 'native' civil society (29.6%; +0.35) relatively support Muslim rights. This shows a degree of willingness from the French state and representatives of the majority population for some extension of Muslim rights, albeit from a situation where these are not already granted to anything like the same degree as Britain and the Netherlands.
In Germany, the state is the dominant actor in the public debate over Muslim rights and neither advocates nor opposes an extension (42.2%; −0.05), a position that is less supportive than the states in the other three countries. Political parties (16.1%; −0.06) and 'native' German civil society (24.1%; −0.08) take positions that are identical to the state orthodoxy. This shows that there is on balance little contestation against the German state orthodoxy from the representatives of majority society. Relative to the other countries, Muslims in Germany find less potential support from allies among 'native' civil society and the state and little overall support for an extension of Muslim rights. In a country where the Muslim population of migrant origin faces much higher barriers to access citizenship than in the other three countries , and the state prioritizes Christian religions, Muslims organisations and groups stand alone in their advocacy for Muslim rights (+0.54), while some even publicly oppose this extension. 7 Overall, we find a public debate in Britain where the state, and majority and Muslim civil society organisations and groups are strong advocates of Islamic rights. In the Netherlands, mobilised opposition by a radical right party challenges the state and native civil society, who support Muslim rights, but relatively less so than Muslim organisations and groups themselves. In France, Muslims have a strong presence in the debate over Muslim rights, and receive some encouragement from the state and French civil society organisations and groups. Finally, in Germany, the state is a strong presence in the debate, and along with political parties and native civil society organisations, is neutral over extending Muslim groups rights. It seems that Muslims in Germany have not yet gained a foothold in debates over their own access to group rights.

'Gaps' between majorities and Muslims over group rights? 8
Individual-level data on public opinion is often used to test explanatory models for inter-group relations and hostility to outgroups of immigrant origin (e.g. Strabac and Listhaug 2008;van der Noll and Saroglu 2015) or from other religions (Koopmans 2015). By contrast, our research objective in analysing cross-national survey data is to verify the size of the 'gaps' in opinions between the non-Muslim majority and Muslims, empirically. This gives an indicator for the socio-cultural distance between the majority and Muslim minorities in the respective countries based on their relative degree of acceptance of Muslim group rights.

Approach, method and data
This article uses data from our original survey conducted within the EU Framework seven project EurIslam 9 (see introduction to this special issue) to gauge the way that ordinary people view Muslim rights issues. The data set was collected through Computer Assisted Telephone Interviews in 2010/ 2011. The survey oversamples people with a migrant background and includes more than 7000 interviews. In each country, interviews were undertaken with migrant minorities from the most important Muslim origin countries (former-Yugoslavia; Turkey; Pakistan; Morocco) in the countries of settlement that we investigate. This 'Muslim' sample was drawn by onomastic procedure, that is, common family and first names were sampled from the latest electronic phonebooks, then respondents were screened to see if they, or at least one of their parents, were Muslims and from our selected countries of origin. Bi-lingual interviewers were used so that Muslim respondents could choose the language of their settlement or origin country. In addition, we collected a sample of the majority population who are not Muslims. This 'Majority' sample was collected randomly in each country. Efforts were made to ensure that there was a gender balance in all samples. 10 From the large number of survey questions, we selected those about three issues that relate explicitly to an extension of religious group rights, and which have been the subject of public controversies about the relationship between the state and Muslims in Europe: (1) allowing the construction of mosques with visible minarets; (2) allowing religious symbols on schoolteachers' clothing and (3) the provision of RE in the school curriculum. For the issues about schoolteachers' clothing and provision of RE, we are able to compare responses about the religious rights for Christianity and Islam.
We took a research decision to analyse variations between groups within a country, instead of comparing groups regardless of country. The thinking here is that our respondents live in four distinct national legal, policy and interpretative contexts for understanding Muslim group rights. This has been established empirically by studies on church/state relations (see earlier), including those using indicators (see ICRI above), and on public discourse (see Carol and Koopmans 2013 and above). Because there are significant national differences for accommodating Muslim group rights, opinions that respondents express for or against them mean something substantively different depending on their respective national country context. A member of the public living in Britain, whose child may go to a school where a teacher wears a religious headscarf, is institutionally and legally empowered to do so, and where if politicians speak about this at all, do so in supportive tones, is clearly relating to a different conceptual and institutional world than someone in France, where all this would be an inconceivable and legally impossible. The 'normal' in Britain and France is very different precisely because of their distinct approaches to accommodating Islam as a minority religion. This means that respondents to our survey questions place themselves on a scale of agreement/disagreement for a proposition over religious rights within a context that is national. For this study, we consider that the scales for agreement/disagreement in the survey questions are best employed measuring the relative differences in opinions between individuals (from the majority and minority groups) within a country, because the scales do not substantively mean the same thing across the four countries.
In addition, we recognise that 'Muslim' covers a variety of national backgrounds, ethnic, religious and racial groups. Although our primary interest is the 'gap' between the majority and Muslim groups, we try to avoid the common pitfall of lumping all Muslims together, by also allowing for examination of the variation between groups of Muslim immigrant origin from four countries of origin (former-Yugoslavia, Pakistan, Morocco and Turkey). This enables us to examine the differences in opinion between the majority and each of the four Muslim groups, as well as identifying variations between Muslim groups, where these are also relevant to the discussion.

Analysis
To analyse the responses we conducted five (group: native; ex-Yugoslavian; Moroccan; Turkish; Pakistani) × two (gender: male; female) two-way analyses of covariance (ANCOVA) controlling for age, educational level and income, for each country. We include age, education, and income as covariates, and gender as an independent variable, to see whether our findings hold, regardless of generational differences, whether people are better educated or not, or how much they earn, and whether gender matters.
In ANCOVA the test of whether groups' means are the same is represented by the F-ratio and an associated significance value. A first general finding is that gender does not affect results, while group belonging matters a great deal in explaining variance. In the 20 tests undertaken (i.e. analyses within each of four countries for five survey questions on the three issues), we found gender to be significant at the p < .05 level in only three instances. By contrast, F tests showed that group variance by group (native, ex-Yugoslavian, Moroccan, Pakistani, Turkish) was very highly significant at p < .001 in each of the 20 tests undertaken (see bottom row in relevant tables below). For this reason, we focus in the following on group differences within the four countries, respectively.
We include tables that show the means adjusted for the effects of covariates, and level of significance of Bonferonni-corrected pairwise comparisons of these means. Specifically, our tables show the adjusted means and significance of the differences between the native group and the four groups of Muslim origin, since these were much greater than the differences between the four groups of Muslims.

Allowing minarets on the skyline?
A first clear indicator for the acceptance of Islam as an immigrant religion concerns whether people think that it should be possible to build mosques with minarets. While Church spires are a common feature on the skyline across Europe, the question arises whether Muslim minorities should be granted the right to have mosques that make minarets visible in the physical public space of their country of settlement. Accepting minarets on the skyline is a visible symbolic marker for including Islam within a country's public space.
Before examining respondents' opinions, it is first necessary to provide a basic idea of the crossnational variations in states' permitting the construction of visible minarets. There is no space for detailed coverage, but we use the information provided and publicly accessible by the ICRI indicators on national policies for extending group rights to Muslims. We use the 2008 ICRI data, which is the closest time-point to when our survey was held. The ICRI findings on the countries' policies for allowing the construction of mosques with minarets show that laïc France is the least accommodating compared to Britain, the Netherlands and Germany, though it notes that the German case has moved in this direction only just before 2008, relative to Britain and the Netherlands. By 2008, this gives the following outcomes: Britain had about 1500 visible mosques serving a Muslim population of 1.5-2 million, the Netherlands about 20 for 850,000-1 million, Germany between 40 and 160 for 3.5 million, while by contrast France had only about 15 for 4.5 million Muslims.
Our survey question is: 'Several aspects of state-religion relations have recently been under discussion in <France/Germany/the Netherlands/the UK>. Do you agree or disagree with the following statements? The construction of minarets should be allowed on <French/German/Dutch/British> territory.' Table 2 shows the main findings of the Bonferonni-corrected pairwise comparisons. The scores show the adjusted means for a group's responses on a four-point scale, ranging from 1 'agree strongly', to 2 'agree', to 3 'disagree', to 4 'disagree strongly'. A mean above 2.5 rising towards 4 indicates a group's increasing disagreement, while one below 2.5 decreasing to 1, indicates increasing agreement. The columns show the results within each country. The 'Majority' row gives the adjusted mean score for the non-Muslim sample. The subsequent four rows show adjusted means for the four Muslim groups (ex-Yugoslavian; Pakistani; Moroccan; Turkish), respectively. The * symbol shows when this opinion is highly significantly different from the 'majority' opinion (at *p < .05, **p < .01, *** p < .001). The 'Muslim' mean is a score calculated from the means of the four Muslim groups (weighted) per country. Last, the 'Gap' from the Majority to the Muslim is calculated by the majority mean opinion minus the Muslims' aggregated mean (per country). An overall plus 'gap' figure shows the degree to which that Muslims are more in agreement with a proposition than the majority.
First, we see that all four groups of Muslim origin (ex-Yugoslavian; Pakistani; Moroccan; Turkish) are highly significantly different (p < .001) from the majority in every country. This shows a dissensus among respondents over having minarets in their country that runs along Majority/Muslim lines. Differences between the majority and Muslims are very much significantly greater than differences between the groups of Muslims. Among Muslims, we see that groups from former-Yugoslavia are closer to the majority view in each country, but still highly significantly different from the majority.
Second, from the aggregate Muslim adjusted mean, we see that Muslims 'agree' that minarets should be allowed in all countries. They agree more strongly in France (1.71) where this happens least, and Germany (1.75) where it has started only since 2002, compared to the Netherlands (1.98) and Britain (1.88) where this has been permitted for relatively longer.
Third, turning to the majority population, while the Dutch (2.42), German (2.39) and French (2.45) are close to 'neutral' and just on the scale towards agreement, the British non-Muslims (2.93) clearly 'disagree' with minarets. Turning to the overall Majority-Muslim 'gaps' this produces, we see that the cultural distance is smallest in the Netherlands (0.44), and then Germany (0.64), compared to France (0.74). Broadly this follows what we would expect from the three countries' policies for accommodating Islam. However, the striking exception is Britain, which has the largest Majority-Muslim 'gap' (1.05) over this Muslim group right, despite being a country that has significantly allowed minarets. This large 'gap' in Britain comes from an unexpectedly strong majority opinion that is against minarets, compared to the majorities in other countries, while the British Muslim opinion is as we would expect similar to that in the Netherlands, the other country where minarets have been relatively numerous and visible on the skyline for some time. It is striking that our respondents from the British majority population express clear disagreement with the accommodating approach that political elites have taken over this issue, not least when we recall that Muslim group rights were also least contested in British public debates. This finding seems to indicate a gap between British policies and elites' public stances that are relatively supportive of Muslim group rights, on one side, and the opinions of ordinary British non-Muslims, on the other.

Allowing teachers to wear religious symbols in schools?
It is well established that states have historically used education systems as an important agent for nation-building (Gellner 1983), and that state schools remain important agents for socialising the next generation into the values, identities and ideas of who belongs to a national community. This is also why schools have regularly been the institutional location for public and legal disputes over the place of religion in society, generally, and over the place of Islam. Conflicts over whether pupils or teachers can wear veils or crucifixes in the state-controlled environment of the public school have been a regular field for the so-called 'culture clashes', where religious values, and in particular the values of Islam, are seen to challenge those of the state. Teachers perform a special role in schools as public servants acting on behalf of, and as employees of, the state. Hence the degree to which teachers are banned from wearing religious symbols in conducting their professional roles indicates a state's position over accommodating a religion. Regarding actual situations in our four countries, again, the ICRI indicators demonstrate crossnational variations in policies. In 2008, no religious symbols at all were allowed in French public primary and secondary schools. Britain and the Netherlands were more accommodating. Teachers were permitted to wear the Islamic veil, while Christian religious symbols were already accepted parts of state education. In Germany, Christianity was strongly present in German education, while teachers had been banned from wearing the veil in a number of States, although the Federal Court had ruled at the Federal level that there was no clear legal basis for this in 2003.
While the minarets question looked only at opinions on a provision for Muslim group rights, we have two questions on teachers' attire and religious symbols: one on Christian symbols and clothing, and the other on the Islamic veil. This allows us to compare respondents' opinions over provision for the majority religion and for Islam. It provides a more nuanced understanding of the majority and Muslim minority groups' views over religious rights, for Christian religions and Islam, relative to one another.
After the same introduction used for the minaret question, the survey queries whether respondents agree or disagree with two statements: (a) Teachers in public schools should not be allowed to wear visible Christian symbols such as a cross or a nun's habit. (b) Teachers in public schools should not be allowed to wear a Veil.
Note that the questions are worded negatively, so that agreement indicates an opinion against religious symbols. However, to follow the same interpretive logic as Table 2 on allowing minarets, and subsequent tables, we adjusted the data in the Table 3(a) and 3(b), so that a mean above 2.5 moving towards 4 indicates increasing disagreement with an extension of religious rights, and below 2.5 towards 1, increasing agreement. Table 3(a) shows findings on teachers wearing symbols and clothing associated with the majority Christian religions. First, we see the strong imprint of laïcité in the opinions of the French majority and Muslim minority relative to those in other countries. The French majority respondents (2.93) have the strongest disagreement with teachers being allowed to wear Christian symbols of all groups, and are the only group apart from Moroccan Muslims in France (2.60), and former-Yugoslavian Muslims in Britain (2.69), who register on the disagreement side of the scale. By contrast, in the countries where the state incorporates Christian religions to a greater degree, and allows teachers to wear clothes signifying Christianity, respondents from the majority are clearly against banning Christian symbols. This applies to the Netherlands (2.30), but to a greater degree to Germany (2.23), and especially Britain (1.90), the two countries where the state significantly privileges Christian religions.
Second, turning to Muslim respondents, their overall positions in countries where the state incorporates Christianity, the Netherlands (2.07), Britain (2.29) and Germany (2.20), 'agree' with permitting Christian symbols for teachers' attire. This Muslim support for Christian symbols in state schools is most likely because Muslims understand that an expansion of group rights for Islam is often best legitimated by a demand for parity with the majority religion. It is harder for liberal states to deny rights to some that that are already extended to others (Joppke 2009). Even in laïc France, the overall Muslim mean (2.48) is equivalent to 'neutral', showing neither support nor opposition to Christianity in schools in a context where it is denied. So overall, we witness no Muslim opposition to the actual (or in France possible) state accommodation of Christianity in this form. This goes against the idea of a 'Christianity versus Islam' boundary along religious divisions driven and constructed by Muslim minorities. On the contrary, Muslims' relative support for Christian religious rights is indicative of a more pragmatic approach to their own religious accommodation.
Regarding the 'gaps' between the majority and Muslim minorities over Christian group rights, it is only in France that the symbolic presence of Christianity in schools clearly divides the majority and all four resident Muslim minority groups. Here the strong imprint of laïcité in the French majority's opinions, who on aggregate 'disagree' with teachers wearing Christian symbols in schools, means that former-Yugoslavians, Pakistanis and Turks are highly significantly different from the majority, while Moroccans just miss significance at the .05 level (p = .07).
However, when we turn to opinions over teachers wearing the Muslim veil, we see from Table 3 (b) that there is a clear dividing line between the majority population and Muslim minorities in all four countries. The opinions of all four groups of Muslim origin in all four settlement countries are highly significantly (p < .001) different from the majority view, with the exception of former-Yugoslavians in the Netherlands and Germany. Among Muslims, Table 3(a) and 3(b) show that those who most likely came as refugees from the ethno-religious wars that tore Yugoslavia apart, tend to be less in favour of religious rights, than Muslims with origins in Pakistan, Morocco and Turkey. Table 3. Agreement/disagreement with teachers being allowed to wear (a) visible Christian symbols in country of residence, by group and (b) Veil, by group (scale 1-4: agree strongly 1; agree 2; disagree 3; disagree strongly 4).

Netherlands
Britain Note: Group of predominantly Muslim origin significantly different from 'native' at *p < .05, **p < .01, *** p < .001-Bonferroni pairwise comparison. + Moroccan in France just misses significance (.07). Table 3(a) and 3(b), we can unpack some important features of the changes in opinion that occur as the question shifts from group rights for Christians to Muslims. First, the overall aggregate Muslim means are remarkably similar for teachers' wearing Christian or Islamic symbols across all countries (Christian v. Islam: NL: 2.07 v 2.05; UK: 2.29 v 2.33; F: 2.48 v 2.45; D: 2.20 v 2.23). This shows that Muslims agree to accommodating religious symbols in state schools for their own and the majority Christian religions to the same degree. Again, we think that this similar level of support for their own minority religion and Christianity comes from the importance of demanding parity with a majority religion for advancing Muslim group rights. However, it demonstrates a second important feature of the significant 'gap' between the majority and Muslims over teachers wearing the veil, namely, that it is produced by a change in the opinions of respondents from the majority populations as the question shifts from Christian to Muslim rights. It is the majority populations' opposition to the veil that drives the relationship and produces the significant 'gaps'. Only in the Netherlands (2.48) where teachers can wear the veil does the majority remain 'neutral' overall, while in Germany (2.76), and especially in France (3.16) and Britain (3.64), the majority agrees strongly with banning teachers from wearing the veil. This may not be a surprise for the France, where the veil is banned for teachers, nor Germany where there is a mixed policy, but where again the veil is banned in a number of States. However, the British majority's very strong opposition is exceptional in that it goes directly against the country's policies for not banning teachers' from wearing the veil.

By comparing
Here it is worth pointing out that the shift in majority opinions when the question moves from Christian to Muslim rights is largest in Britain (1.74) and then Germany (0.53), compared to the Netherlands (0.18) and France (0.23). Britain and Germany are the two countries whose church/ state accommodations especially privilege Christian religions over other religions, while the Netherlands is relatively more equally accommodating to all religions, and France equally unaccommodating to all. From this, it seems that granting special privileges to Christian religions over others provides legitimacy for majorities to also discriminate in their views and support provision for Christians, but not Muslims. For Britain, the argument is that maintaining the status of a privileged state Church, the Church of England, does more to uphold a sense of Christian privilege in the minds of the majority than the extension of Muslim group rights does to undermine it. The British majority clearly agree with teachers being allowed to wear Christian symbols and attire, but disagree strongly with an extension of this right to Muslims, even though this is what actually happens. In Germany, where Muslims communities have less of a foothold in society, generally, because of restrictive citizenship, and Christian churches are clearly granted privileges that are not extended to Islam, this discriminating stance of the majority in prioritising Christian rights is less surprising.
In sum, the British findings are especially striking. The British majority's strong agreement with banning teachers from wearing the veil produces a large 'gap' between Muslims and non-Muslims (1.31) that is almost twice the size of that in France (0.71), where the laïcité conditions make teachers wearing a veil an absolute non-starter. We should recall that this occurs in the context of a British public debate where extending Muslim group rights is significantly supported and largely uncontested. There seems to be a large gap between the favourable elite-dominated British public debates and the oppositional views of ordinary people from the majority over extending religious rights to Muslims.

Allowing Religious Education in schools?
A final query to test opinions on whether group rights should be included within the public institutional fabric of state provision addresses RE in state schooling. We replicate the design applied for teachers' clothing and religious symbols and ask questions that allow comparison between provision for the Christian and Islam religions. In part, this is to verify whether earlier findings hold, given that references to the veil might have provoked atypical emotive responses from respondents, or that the negative wording of the question might have produced distorting effects. The focus on the place of majority and minority religions in the curriculum of state education goes to the core of questions about the public and institutional incorporation of religion, not least because it is through their education systems that states seek to promote and generate their preferred values and national identities.
With regard to the factual state of affairs facing our respondents, again, ICRI is our source for the basic variations in national policy contexts. In 2008, Britain, the Netherlands and Germany allowed provision for Christian RE but had not extended this to Islamic RE on an equal basis. In Britain, parents were able to request that local councils on RE install Islamic classes. This has led some state schools to offer Islamic RE in areas with a dense Muslim population. Similarly, in the Netherlands, there is a partial form of acceptance. In Germany, there had only been a few pilot projects for Islamic RE classes by 2008, with the main barrier being the subordinate position of Islam relative to the Christian religions within the state's hierarchy of religious recognition. In France, the exclusion of all religious instruction from state schooling meant that the provision for Islamic RE was a nonstarter.
The survey asks whether respondents agree or disagree with two statements: (a) Public schools should offer Christian religious education for those who want it.
(b) Public schools should offer Muslim religious education for those who want it.
Table 4(a) shows the adjusted means for groups' opinions over Christian RE in state schools. In line with the findings on religious attire, the main dividing line is between laïc France and the three countries where state education includes Christian Churches to a greater degree. First, the impact of laïcité is clear on the French majority's opinions (2.80), they are the only group who on aggregate 'disagree' with Christian RE for those who want it. By contrast, all four groups of Muslim origin in France agree with the proposition to a degree that is highly significantly different (p < .001) from the majority. This produces a 'gap' between the French majority and Muslims (0.71) that is significantly wider than in the Netherlands (0.18), Britain (0.10) and Germany (0.22), where there is a relative majority and Muslim consensus in favour of this right for Christians. Support for Table 4. Agreement/disagreement with schools offering (a) Christian religious education and (b) Muslim religious education to those who want it in country of residence, by group (scale 1-4: agree strongly 1; agree 2; disagree 3; disagree strongly 4).

Netherlands
Britain Christian religious instruction in schools is especially strong among the German majority (1.57 D 1.79). This confirms the earlier finding that Muslims tend to support Christian rights, probably thinking that this is more conducive for a spill over of rights to their own religion. Again, this indicates a pragmatic stance, with no evidence of a sizeable Muslim opposition to Christianity simply because it is a different religion anywhere. When the question shifts to state provision of RE for Muslims (Table 4(b)), we also find a similar overall pattern to the earlier findings on teachers' wearing religious attire. First, there are clearly divided opinions in all countries between the majority and Muslims. With the exception of former-Yugoslavians in the Netherlands, all groups with Muslim origins in all countries are highly significantly (<.001) different in their views to the majority, and more in agreement with allowing Islamic RE.
Once more we see that Muslims in the Netherlands, France and Germany, hold similar views on RE provision for Christians and Muslims. It is only in Britain where Muslims shift in their views (0.36) to be more in agreement with this right for themselves than for Christians. Also, among the Muslims groups, the former-Yugoslavians are relatively less in favour of this Muslim group right than the others, but with the exception of the Netherlands, much closer to the Pakistani, Moroccan and Turkish groups than the majority.
Again following the pattern we observed earlier, it is the majority groups' relative shift in opinion against provision for Muslims compared to Christians that leads to important 'gaps' between the majority population and Muslims. Also this shift in majority opinions is greatest in Britain (Christian 2.15 v. Muslim 2.75 = 0.60) and Germany (Christian 1.57 v. Muslim 2.17 = 0.60), the two countries that privilege Christian religions in their state accommodation over other religions. The shift is less pronounced in the Netherlands (0.24), where state accommodation generally treats religions more equally, and France (0.21), where they are all relatively equally excluded.
The overall findings bear some imprints of the countries' respective forms of Church-state accommodation. French majority-supported state laïcité leads to a wide Majority/Muslim gap (1.01) and few prospects for an extension of religious rights to Muslims. The French majority are unlikely to support rights for Muslims that are denied to Christians. In the Netherlands, where the state treats religions relatively equally and RE for Muslims is partially accommodated, the Majority/Muslim gap (0.37) remains significant, but is the smallest of the four countries. In Germany, state inclusion of (especially) Christian religions but not Islam, leads to an overall strong majority support for religious provision compared to other countries, but also a relatively strong reaction by the majority against specific provision for Muslims (German 'gap': 0.46). We also see a similarly strong reaction by the British majority respondents when the question switches from provision for Christians to provision to Muslims. In the British case, this is supplemented by an assertiveness of Muslims for an extension of rights to them relative to Christians.
The British findings are remarkable. For a country with some degree of state accommodation for religious classes for Muslims, it is striking that the British majority shows a clear opposition to this policy and makes a very clear distinction in its support for Christian rights and opposition to extending the same rights for Muslims. Conversely, British Muslims support Christian rights, but advocate support for their own religious rights to a greater degree, no doubt encouraged to expect 'parity' from an institutional system that goes a long way to providing it on many issues. As a consequence, we witness a polarisation between the opinions of the majority and Muslims in Britain as the question shifts from Christian to Muslim provision. This results in a striking 'gap' (1.06) along ethno-religious lines between the British Muslims and non-Muslims, driven from both sides. It is indicative of a clear dissensus in the views of ordinary people and a potential for conflict between the British majority and Muslims over Muslim group rights. Recall also, that all this occurs in a country with relatively accommodating policies and a highly supportive public debate for extending rights to Muslims, and where there is little evidence for mobilised public opposition to Muslim group rights by political parties or from civil society.

Discussion
The primary aim of this article was to examine how ordinary members of the public view the state's accommodation of Muslim group rights within their respective countries. We found that regardless of the degree to which, and how, states have accommodated Islam, there are highly significant 'gaps' between the opinions of the majority populations and Muslims over what the provision of Muslim group rights for should be, respectively, in all four countries. A wide socio-cultural distance exists between non-Muslim majorities and Muslims over these issues, with Muslims being clearly more in favour of an extension of rights for their group, while facing varying degrees of opposition from the majority. In everyday life, ordinary Muslims and non-Muslims seem to be no closer to a consensus over the 'fair' limits for accommodating Islam.
This 'gap' persists in the Netherlands and Britain, which shows that significantly extending Muslim rights does not seem to bring majorities and Muslims to a consensus.
Also the 'gap' holds and is highly significant between the majorities and each of the four groups of Muslims from different countries of origin (former-Yugoslavia, Pakistan, Morocco and Turkey) in each country. 11 Conversely, differences between the four Muslim groups were seldom significant. Given this degree of internal consensus among different Muslim groups relative to their dissensus with their respective national majorities, it does seem possible to speak of a 'Muslim' opinion over Muslim group rights within a country.
We were able to gain access to important information on the nature and source of the 'gap' by comparing the majorities' and Muslims' opinions on state provision for Christians versus Muslims. First, we found that Muslims tend to support state provisions for the majority Christian and their own religion to the same degree. This is most likely because Muslims see better opportunities to advance their own group rights through a context that supports all religious group rights. Second, by contrast the majority populations tend to support state provision for Christian rights substantially more than for Islam. Importantly, it is the majorities' opinions that change and turn against group rights as the demand shifts from provision for Christians to Muslims. This shift in majorities' opinions drives the relationship and results in a significant gap between non-Muslims and Muslims. Our findings demonstrate that majority opinions matter a great deal in determining the potential for controversy over Muslim group rights.
Here there is a crucial methodological point. We were able to reveal the importance of majorities in generating the 'gap' only because our sample included Muslims and non-Muslims, and our research design allowed comparison between responses about the majority Christian religion and Islam. Future surveys need to include these considerations into their sampling and design. If we had not included a majority sample, then we would have lost the important reference point for understanding the problematic and which provided our key findings. 12 Regarding cross-national variations, we found that the shift in majority opinions for supporting Christian versus Islam provision was greatest in Britain and Germany, the two countries where state accommodation clearly prioritises Christian religions above others. The shift was markedly less in Netherlands, which is relatively more accommodating across religions, and France that is relatively equally restrictive to all. This is not to suggest that people in Britain and Germany see themselves as more Christian than those in the Netherlands, but that they find greater legitimacy for placing Christianity over Islam and in doing so they repeat the discrimination within their country's approach to religious accommodation. While many members of the native population are not practising Christians, or even religious, they will have grown up and been socialised, for example, in schools, where the specific national church-state accommodation was embedded into their experiences of public life. For Germans and Brits, the primacy of the Christian religions is part of the national heritage, values and identity that has accompanied them on their journey through life. In this sense, their opposition to extending group rights to Muslims can be a preference for maintaining the status quo (against change) and need not be an indicator for outright hostility against Islam as a religion or Muslims. This majority stance does however run against the views and expectations of Muslims, especially in Britain, where the state-often encouraged by its own Church of England-has significantly extended religious rights to Muslims and publicly legitimated the idea that this will continue.
Notwithstanding this explanation, the strength of British majority opposition to Muslim group rights is exceptional and runs through our findings. The British findings completely contradict the notion that accommodating policies lead to closer opinions between the majority and Muslims. Britain has accommodating policies for Islam and a highly supportive and consensual public debate that legitimates expansive Muslim rights. However, this is combined with a majority public opinion that strongly opposes Muslim rights to a degree that it produces the widest 'gaps', even compared to laïc France.
There is no evident problem with the British data. It was collected in the same way and with the same rigour as the other data sets, it meets the quality standards of other comparable projects, and the findings have been controlled for age, education and income. Nor is the make-up of our sample of 'Brits' importantly different from the general population (see Carol 2013). Although the opinion data is limited because it presents only a 'snapshot' at one time-point, there was no exceptional event during data collection that would account for a strong majority opposition.
Accepting the British findings as a 'fact', how do we explain the apparent 'gap' between supportive policies and public debates on one side, and an oppositional majority public opinion on the other? We think that political context matters and refer back to our findings on the degree and form of contestation within public debates.
Our starting thesis is that the public presence of a recognisable anti-Islam party has a sort of pressure valve 'effect', i.e., it specifies issues in a way that makes positions and outcomes visible and clear in the public domain, which has the effect of bringing the opinions of the rest of majority society together against the anti-Islam stance.
Here the comparison between Britain and the Netherlands is instructive. The Dutch public debate is less supportive of Muslim rights and more contentious. Anti-Islamic mobilisation by Wilders' PVV political party shifts the debate in a restrictive direction, but the state still publicly legitimates Muslim group rights and has an ally in organised Dutch civil society. Turning to public opinion, this produces a 'gap' between the Dutch majority and Muslims that is the smallest in spite of the impact of Wilders' mobilisation. When people can hear a virulent anti-Islamic political 'voice' (Wilders' PVV) that is highly visible, publicly resonant, and has a degree of legitimacy through its presence in the institutional polity, many sections of majority society will distance themselves from this actor, because of the extreme views it expresses. Mainstream political actors will not want to be viewed in the same camp as a party that is explicitly anti-Muslim and expresses 'Islamophobic' views. This leads to a broader range and differentiation of political reasons for opposing group rights, including more moderate ones, and a different set of available cues for the public to understand the issue.
In Britain, it seems that in the absence of a mobilised anti-Islamic political actor, a groundswell of resentment has built up among the British majority over the significant extension of Muslim group rights. Without the trigger of an anti-Islamic party on the political landscape, there is a lot of low level undifferentiated resentment expressed against Islam that is manifest in everyday life, but largely ignored by political elites and unrepresented in public debates. In this context, perceptions that elites are not responding to majority opinions, for example, in claims that the government's 'political correctness' leads to policies that privilege Muslims over majority interests, may also lead to Islam becoming a focus for broader anti-elite sentiments. At present these remarks remain speculative. To go further, we would need more data and systematic research on the relationship between public debates and opinions over Muslims group rights.