: Conflict configurations at the transnational and domestic level

In this paper we aim to map public preferences in the domain of asylum policy in order to explore the various lines of conflict and their intensity in the aftermath of the refugee crisis. Taking as a starting point the EU polity’s two-level structure, we study both transnational lines of conflict looking at polarisation between countries with regard to the major asylum-related policies that have been proposed or adopted during the different phases of the refugee crisis, and domestic lines of conflict looking at the polarization between various groups within countries in terms of their general support for migration and party allegiances. Utilising an original cross-national survey fielded in 16 EU member states, we find that conflicts surrounding asylum policy are more intense at the domestic level between supporters and opponents of migrations, than at the transnational one between various types of countries. Moreover, such conflicts are being structured around the relocation of asylum applicants debate (involving quotas or compensation), while other policies involving external or internal bordering (such as the EBCG and general border closures) or externalisation (such as the EU-Turkey deal) are comparatively less polarising. Generally, our results show that the conflict potentials of immigration policies, rather than being fully mobilised or alleviated, are still large and have markedly increased, especially in the destination states of north-western Europe over the last few years, with implications for the options available to policy makers.


Introduction
The EU polity has a two-level structure which invites political structuring at both the supranational level of the EU and the national level of the member states.Similar to comingtogether federations, the conflict structure at the EU level is dominated by the territorial dimension.This dimension produces two lines of conflict: a vertical one, focused on the powers of the polity centre vis-à-vis those of the member states, and a horizontal one, revolving around the specific interests of the member states (Kriesi, Ferrera, and Schelkle 2021).But the European integration process does not only pit countries against the European centre and against each other, it also pits citizens with diverging views of this process against each other within each country.Viewed from the perspective of the general public's attitudes on the EU asylum policy, we can analyse the extent to which citizens from different countries are divided between themselves, and how they are divided among themselves within each country.We shall first analyse the transnational conflicts between citizens from different countries, and then focus on the conflicts between citizens within countries.
In the domain of asylum policy, we expect the transnational conflicts to be closely related to the country-specific experiences in the refugee crisis and in the years following the crisis.By contrast, we expect the within-country conflicts to be rooted in a broader divide between cosmopolitans and communitarians, which is based on structural developments that go beyond the experience of the refugee crisis.In terms of horizontal transnational conflicts, we expect the perspective of the general public to be shaped by the type of states they are living in: 'frontline'; 'transit '; 'open destination'; 'closed destination'; and, 'bystander' states.Based on the countries' policy heritage, their geographical location on general migration trajectories in Europe, as well as their immediate crisis experience we distinguish between five types of states: frontline states, transit states, destination states, further divided into two sub-setsrestrictive and open destination states, depending on their institutional and political openness towards incoming refugees, and bystander states.According to the Dublin regulation, countries of first arrival are responsible for any asylum-seeker entering the EU through their territory.In the refugee crisis, this regulation shifted the obligation of accepting and integrating refugees to the southern European frontline states, Greece and Italy in particular, where the majority of refugees first arrived in the EU.But the bulk of the refugees did not stay in these frontline states but continued their journey towards the north of Europe.On their way they travelled across the transit states such as Hungary and Austria.Open destination states such as Germany and Sweden were the principal destination states during the refugee crisis, while closed restrictive states such as France and the UK remained largely untouched by the inflow of refugees.In addition to these four types of member states, there is a residual category -the bystander states, a category for countries that were not directly concerned by the crisis.Among these states were several eastern European countries, as well as Ireland and Portugal.The experiences of these different types of countries during the crisis are expected to have shaped citizens' asylum policy preferences.Second, beyond the general country types, we especially expect the policy positions adopted by policymakers during the crisis to have shaped citizens' policy preferences, as it is well known that policymakers and their parties are opinion forming actors of great importance (Zaller 1992, Druckman, Peterson, and Slothuus 2013, Slothuus and Bisgaard 2021).We expect the citizens of frontline states to oppose the citizens from transit, destination and bystander states, because their countries would benefit most from a reform of asylum policy designed to increase transnational burden sharing.At the same time, we also expect the citizens of the Visegrad 4 countries -Poland, Hungary, Czech Republic and Slovakia -to be the most divergent from those in frontline states, since they have been most mobilized by policymakers against policies designed to increase transnational burden-sharing during the refugee crisis.Given the great impact of the mobilization of the V4 on the populations in eastern Europe, it is likely that the citizens of other eastern European bystander states will share the positions of the citizens in the V4 countries.
Turning to within-country conflicts, the European integration process can be viewed as part and parcel of a larger process of globalization, which re-structures national politics in terms of a new structuring conflict (or cleavage) that opposes cosmopolitans-universalists to nationalists-communitarians.The new structuring conflict raises fundamental issues of rule and belonging and taps into various sources of conflicts about national identity, sovereignty, and solidarity.The emerging divide concerns above all conflicts about the influx of migrants, competing supranational sources of authority, and international economic competition.Scholars have used different labels to refer to this new structuring conflict -from 'GAL-TAN' (Hooghe, Marks, and Wilson 2002), 'independence-integration' (Bartolini 2005), 'integrationdemarcation' (Kriesi et al. 2008), 'universalism-communitarianism' (Bornschier 2010), 'cosmopolitanism-communitarianism' (Zürn and Wilde 2016), 'cosmopolitanism-parochialism' (De Vries 2017) to the 'transnational cleavage' (Hooghe and Marks 2018) and the cleavage between sovereignism and Europeanism (Fabbrini 2019: 62f.).However, what they all emphasize is that the new divide constitutes a break with the period of 'permissive consensus' and that conflicts over Europe have been transferred from the backrooms of political decisionmaking to the public sphere.At the same time, the new conflict leads to a renaissance of nationalism (and a de-solidarization process between nation-states) and a politicization of national political, economic, and cultural boundaries.
These authors agree that the new divide is above all articulated based on two types of issues: immigration and European integration and that it mainly concerns cultural-political, and not economic aspects of these issues.For multiple reasons -programmatic constraints, internal divisions, incumbency, and so forth-the mobilization potential created by this new conflict has been neglected and avoided (de-politicized) by the mainstream parties (De Vries and van de Wardt 2011; Green-Pedersen 2011; Hooghe and Marks 2018; Netjes and Binnema 2007;Sitter 2001;Steenbergen and Scott 2004).Consequently, voters turned to new parties with distinctive profiles for their articulation.Over the past decades, it was first the cosmopolitan side which mobilized.In the aftermath of the "cultural revolution" in the sixties and seventies, radical left and green parties mobilized the social-cultural segments of the new middle class in the name of cultural liberalism, environmental protection, and multiculturalism.The "cultural revolution" also transformed the social-democratic parties, which, in the process, have become essentially middle-class parties in almost all countries of Western Europe (e.g., Gingrich and Häusermann 2015;Kitschelt 1994).
In a second wave of mobilization starting in the 1980s and 1990s, it was mainly radical right which have mobilized the heterogeneous set of the losers of globalization (Kriesi et al. 2008) and their concerns about immigration and European integration.These parties were mainly newly rising challengers, but in some countries such as Austria and Switzerland, they consisted of transformed established centre-right parties.These parties all endorse a xenophobic form of nationalism that can be called 'nativist' (Mudde 2007), claiming that states should be inhabited exclusively by members of the native group (the 'nation').Accordingly, the vote for these parties has been shown to be above all an anti-immigration vote Oesch 2008), and, to some extent a vote against Europe (Schulte-Cloos 2018; Werts, Scheepers, and Lubbers 2013) and against the cultural liberalism of the left that has increasingly shaped western societies (Ignazi 2003;Inglehart and Norris 2016).
The green parties on the one hand and the radical right parties on the other hand mainly rose in north-western Europe.They have become established forces in the national party systems of the respective countries, even if, for various reasons, the radical right broke through in some of them belatedly.In southern Europe, until recently, with the exception of the Italian Lega Nord (Betz 1993), radical right parties have not been able to get a foothold.The impact of the new conflict has been more limited in the countries of southern Europe -for reasons that have to do with their political legacy (long-lasting authoritarian regimes and strong communist parties, i.e. a strong 'old' left), with their having been emigration countries until more recently, and with the fact that the return to Europe after the authoritarian period was perceived as a return to western civilization (Diez Medrano 2003).However, under the impact of the combined economic and political crises that shook southern Europe in the more recent past (Hutter and Kriesi 2019), new parties of the radical left (but hardly any green parties) have surged in Greece, Spain and (to a more limited extent) in Portugal.More recently, parties of the radical right also rose in Italy (Lega) and Spain (Vox).In central-eastern Europe, both types of radical parties have so far been rather weak or transient, due to the Communist heritage and the low level of institutionalization of the party system.Instead, in this part of Europe, we have witnessed a radicalization of mainstream parties -of the centre-right (e.g., in Hungary (Fidesz), Poland (PiS), and the Czech Republic (ODS)) or the centre-left (e.g., in Romania (PSD)), which have defended positions previously adopted by the radical right in Western Europe.
At the domestic level, we expect that the conflicts are indeed shaped by attitudes about immigration and European integration and that these attitudes are most clearly articulated by the parties taking a nationalist position (the radical right and the conservative-nationalist right in some countries), on the one hand, and those taking a cosmopolitan position (the Greens and the radical left), on the other hand.Overall, we shall show that domestic conflicts are more polarizing than transnational conflicts, which is to suggest that the potential for further transnational conflicts is, indeed, quite large.In general, the opponents to immigration are crucial for asylum policymaking: if they dominate in some member states, they can induce their governments to legitimately block transnational burden sharing.In line with this argument, we shall see that the more restrictive policies are more likely to be supported than policies that aim at transnational burden sharing.

Measurement
We use data collected as part of an original cross-national survey fielded in 16 EU member states in June-July 2021, covering all the five types of states we are interested in 1 .The national samples were obtained using a quota design based on gender, age, area of residence, and education and consist of around 800 respondents per country, amounting to a total of 13,095 respondents.The survey's larger scope was the study of attitudes related to the multiple crises that have hit the EU since 2008 (such as the financial and sovereign debt crisis, Brexit, COVID-19) and within this scope, the survey included a section focusing specifically on the refugee crisis.This section consisted of multiple items ranging from attitudes towards migrants and immigration more generally, to performance evaluations of the national governments and the EU in the refugee crisis, to evaluations of specific policies proposed or adopted during the refugee crisis.Additionally, the survey included a host of general political attitudes, enabling our in-depth analysis of the conflict configurations surrounding policies in the refugee crisis.The timing of the survey in the aftermath of the refugee crisis also provides us with two advantages.First, it allows us to compare all the policies that have been proposed or adopted during the different phases of the crisis.Second, rather than measuring agreement with these policies at the peak of the crisis, when respondents might be biased in favour of one policy or another due to contingent considerations, asking them about their evaluations of policies in the aftermath of the crisis allows for a more considerate assessment of these policies.In what follows, we describe the items used in detail, as well as the measures employed for systematically comparing conflict configurations between and within countries.
To measure attitudes towards policies, we include a series of six items tapping into agreement with all major types of policies that have been proposed or adopted at the EU level, but also policies adopted by member states.The EU policies taken into consideration are the 1) relocationrelocation quota, requiring countries to accommodate a share of asylum applicants, 2) relocationrelocation compensation, requiring countries to pay compensation to other countries that accommodate asylum applicants, 3) external bordering through EBCG, investing in reinforcing external borders by reinforcing the border and coast guard, 4) Dublin regulation -requiring refugees to be accommodated by the country through which they first entered Europe and in which they were first registered, and 5) externalization, pursuing deals with third countries (such as Turkey and Libya) via financial and other incentives.To this we add a sixth category concerning international policies of member states: 6) internal border control, i.e., reinforcing countries' internal borders by improving border surveillance, building fences, or pushing back migrants by force.The entire battery of items included in the survey can be found in the Appendix with all items measured on scales from 0 to 10.
For measuring immigration attitudes, we use a series of eight items tapping into views about the impact of immigrants in several areas (economy, culture, criminality, overall quality of life) and into the degree to which each country should allow various groups of people to come and live in the country (same race/ethnic group, different race/ethnic group, poorer countries outside Europe, poorer countries inside Europe).This combination of items for measuring immigration attitudes has already been applied in a cross-national setting in the framework of various waves of the European Social Survey (ESS).As the items are all related conceptually and load unto a single factor with Eigenvalue of higher than 1 (see Appendix), we combine them into a single index of pro-and anti-immigration attitudes.Beyond immigration attitudes, we also expect party allegiance to be an important driver of within-country conflicts on policy.For measuring party allegiances, we use a standard vote recall question and re-code parties in our 16 countries into eight different party families: radical left, green, social-democrats, liberal, conservative-Christian-democratic, radical right, other, and non-voters.Finally, we also include Euroscepticism measured by a question on whether European integration has gone too far or should be pushed forward.
We examine descriptively the conflict configurations in four different ways.First, we present the kernel-smoothed distributions of the policy-specific attitudes in the different countries and domestic groups.Second, we estimate levels of policy polarization across different groups in terms of country types, pro-/anti-immigration attitudes, party family, and Euroscepticism and focus the bulk of our analysis on summary polarization measures across these groups.This polarization measure is based on the Kolmogorov-Smirnov (KS) statistic (see Marsaglia, Tsang, and Wang 2003;Siegel 1956: 127-136), which quantifies the distance between the empirical distribution functions of two samples.Our choice for KS statistic is guided by three arguments: first, since we cannot assume a specific shape (e.g.normal) of the distributions of policy agreement across the different groups, this statistic offers a distribution-free alternative to other, parametric measures of distance (e.g.Bhattacharyya distance); second, the KS statistic can be used as a metric which means it is symmetric (distance between distribution A and B is the same as distance between distribution B and A) and has a finite range from 0 to 1, with higher values indicating larger distances between the compared distributions; third, the statistic detects a wider range of differences between two distributions than simply comparing summary statistics such as the mean or the median.Third, we attempt to reduce the complex conflict-configurations by relying on multi-dimensional scaling procedures (mds).These procedures are designed to place the different entities (in our case member states, as well as social groups defined by their immigration attitudes and partisanship), in a low-dimensional (typically two-dimensional) space.The distances between the entities in the resulting space reproduce their policy-distances as closely as possible.The substantive meaning of the spatial dimensions lies in the eyes of the beholder; one relies on the raw data to come up with an interpretation of the dimensions, but of course this is more art than statistics.Finally, we also use ordinary regression analysis to show how the two types of conflicts relate to each other.

Transnational polarization
To explore the horizontal line of conflict between member states among the citizens we start by looking at the distribution of support in our main country types in order to examine the direction of the attitudes towards selected policies (Figure 1).Generally, regardless of policy type, we notice that the public in frontline and transit states differs the most from the public in other states in terms of policy support.With respect to relocation (Figure 1a), in frontline states the attitude distribution is heavily skewed in favour of the relocation quota; this is unsurprising since relocation policies would alleviate their immediate burden.By contrast, in transit states the public is most opposed to the relocation quota, as they are neither immediately affected by the problem pressure nor ultimate destinations of the migrant flows.The distribution of support is very of similar in destination states, of both the closed and open kind, and in bystander states with respondents being somewhat more positive towards the policy, but with a large neutral share of respondents.With regard to the Dublin regulation (Figure 1b), again unsurprisingly, respondents in frontline states are the ones most opposed to it, followed by the transit states.By contrast, bystander and destination states are rather neutral.Finally, Figures 1c and 1d indicate that external bordering via the reinforcement of the EBCG and externalization via deals with third countries are the least polarizing policies among the citizens with similar distributions across all country types, all heavily skewed towards neutral-positive attitudes.Transit states are the only ones that slightly diverge in the sense that they have even a higher share of positive attitudes towards these policies than the other country types.Indeed, in line with the visual insights from Figure 1, the relocation policies (quota and compensation) have been the most contested between member states, followed closely by the Dublin regulation.By contrast, internal bordering and externalization appear as the least divisive issues between member states at the demand level.This difference in the divisiveness of policies on the demand-side closely follows the patterns on the supply-side and the actual policy outcomes of these proposals.While internal burden sharing based on quota and compensation proposals had failed with countries being highly divided on the issue, externalization based on deals with third countries (such as the EU-Turkey agreement) was eventually (one of) the arguably successful policies.Therefore, the EU-Turkey episode, which dominated most of the peak phase of the crisis and was the single most politicized policy decision taken during this crisis, left a positive legacy among the public -most likely due to its successful implementation: in the aftermath of the crisis, externalization to third countries appears as the least polarizing option on the demand side.
Beyond these general patterns, countries also diverge according to their type and centrality in the crisis.While the distance measure used here does not tell us the direction of the country-specific deviations (pro or against the policy) from the mean, we can interpret these deviations based on the insights from Figure 1.With regard to relocation, we see that several frontline states (Italy and Greece), bystander states (Latvia and Poland), and transit states (Hungary) appear to be most polarized.As already apparent from Figure 1, it is above all citizens in Italy and Greece, as the most affected frontline states, who favour these policies as they would reduce their immediate burden, whereas bystander and transit countries are the most opposed to these policies.Going beyond our country types, we see more specifically that not all transit and bystander states are polarized to the same degree.Together with Latvia, Hungary and Poland stand out the most.This indicates that the pattern observed at the level of decisionmakers during the crisis, when the resistance of the Visegrad group (V4) was formed against relocation, persists among the citizen public in the aftermath of the crisis.
Among the destination states, public opinion in Germany is the most transnationally polarized with respect to relocation, even if to a lesser extent than public opinion in Latvia, Poland, and Hungary.This is unsurprising given the centrality of Germany in the relocation debate.With regard to the Dublin regulation, the countries whose positions stand out the most are Greece and the UK, the former suffering directly from its dysfunctionality, whereas the latter, a geographically insulated, closed destination state, benefited most from shifting the burden to any other state along migration routes.Finally, with regard to internal and external bordering (EBCG) and externalization, we see smaller deviations with most countries having similar distributions in terms of agreement with these policies, with the exception of some bystander states (in particular Portugal) which seem to deviate the most when it comes to agreement to these issues.
In Figure 2 we examine the transnational conflict configuration via multi-dimensional scaling in a bi-dimensional space determined by attitudes towards relocation (x-axis) and attitudes towards border control (y-axis).This representation of the transnational conflict configurations confirms that the relocation policy is structuring the space the most.We have less variation among the member states on the border control dimension and hardly any with regard to externalization.The horizontal alignment of member states in terms of relocation shows three clusters of countries.On the left-hand side, opposed to relocation, we have members of the Visegrad group -Hungary and Poland (joined by Latvia) -as the most vocal opponents of relocation, whereas on the pro-relocation right-hand side, we have the two frontline states most heavily hit by the crisis -Italy and Greece.

Domestic conflict configurations: immigration attitude and policy (selected countries)
We study the domestic conflict configurations from two perspectives.On the one hand, we focus on the configurations defined by immigration-related attitudes, on the other hand, we analyse the conflicts between party families.The configurations between groups with pro-and anti-immigration attitudes define the political potentials for the mobilization by the political parties.These conflicts between attitudinal groups remain latent as long as they are not mobilized by political actors.Among possible political actors, we only study parties.However, parties are among the key actors when it comes to the mobilization of immigration related attitudes.The divisions between attitudinal groups is expected to be larger than the corresponding polarization between parties, as parties offer bundles of issue positions and immigration is only one of many relevant issues.

Distribution of immigration attitudes
For our study of the refugee crisis, it is above all immigration-related attitudes which can be expected to determine the policy-specific substantive demands.Consistent with earlier work, these attitudes vary considerably across countries as well as across time, which allows for context-specific politicization of the underlying structural conflict between cosmopolitans and communitarians in each of the different member states.We shall first consider the policyspecific conflict configurations in the 16 countries based on the immigration attitudes, before presenting the respective conflict configurations based on the partisan preferences of the voters in the different member states.
Based on our factor for immigration-related attitudes, we have created three categories of citizens: those opposing immigration, those having a rather neutral attitude with respect to immigration, and those favouring immigration3 .Table 2 presents the immigration attitudes by member states, which are ordered from the country most opposed to immigration to the country most favourable to immigration.These distributions reflect the situation in summer 2021.Overall, there is a slight plurality of 42.8 percent of citizens favouring immigration, compared to 36.5 percent opposing it.The countries differ, however, considerably in this respect.Importantly, there are a number of countries, where pro-immigration groups constitute a minority, while a plurality of the citizens oppose immigration.Importantly, the rank order of the countries in Table 2 does not align well with the different types of states we have distinguished throughout this study based on their experience during the refugee crisis.Thus, among the member states most We have also created a factor for immigration attitudes which is directly comparable to the factor that we obtain based on ESS-data.The ESS-data cover the period 2002-2018 for most of our 16 countries allowing us to compare the current immigration attitudes to the attitudes reaching back to 2002. Figure 3 presents the development of immigration attitudes over time.In this figure, the countries have been grouped according to their over-time patterns.The first graph includes three open destination states -Germany, the Netherlands and Sweden, a closed destination state (France), and a bystander state (Finland).The support for immigration has varied across these five countries in the past, but in all these countries it has collapsed in the last few years.The collapse occurred after 2018, i.e., at a moment when the refugee crisis was already a past memory.The collapse was most striking in Sweden, which used to be by far the country most favourable to immigration.By summer 2021, the support for immigration in Sweden has converged with the support in Germany, Finland, and the Netherlands below the mid-point of the scale.Table 3 shows that the collapse in Sweden has occurred across the political spectrum, even if the radical left has proved to be somewhat more resistant to the general movement against immigration than the rest of the parties.At the same time, the share of the radical right, the party most opposed to immigration has more than doubled in Sweden.

Figure 3: development of immigration attitudes over time, mean factor scores by country
The second graph in Figure 3 shows the countries, where the support for immigration has been rather stable or has improved more recently, albeit from very different levels.This is a mixed group of countries which includes bystander states (Ireland, Poland, and Portugal), frontline states (Greece and Italy), a transit state (Hungary) and a closed destination state (UK), but not a single open destination state.In two of these countries (Hungary and Italy), support for immigration reached a low point in 2016, at the height of the refugee crisis, from which it recovered in the more recent past.The contrasting developments in the two sets of countries led to a convergence of immigration attitudes in the countries under study: the standard deviation of the country means fell from .37 in 2018 to .30 in 2021 4 .

4
Between 2002 and 2018, the corresponding standard deviation had always varied between .32 (2002) and .41 (2008, 2010, 2016).To account for these contrasting developments, we have calculated the correlation between the share of the citizens in a given country that considers immigration one of the most important problems facing their country and/or the EU and the level of immigration attitudes in 2021: this correlation is negative and substantial (-.71), which means that the greater the salience of immigration in a given country in 2021, the lower the support for immigration.The refugee crisis has been most salient in open destination and transit states.
Asked which crisis before the Covid-19 pandemic had been the greatest threat for the survival of the European Union -the refugee, financial, poverty/unemployment, or Brexit crisis, 41 percent of the citizens in open destination and 43 percent of those in transit states mentioned the refugee crisis, compared to only 21 percent in frontline states, 28 percent in restrictive destination states and 30 percent in bystander states.Since the refugee crisis, the salience of immigration issues has, if anything, increased once again: not only roughly one third (32 percent) considered the refugee crisis as the most threatening crisis retrospectively, by summer 2021 almost half (47 percent) of the citizens in our 16 countries considered immigration as one of the most important problems facing their country and/or the EU.The salience of immigration had increased in all countries except Sweden and Germany (the two most important destination countries in the crisis), Austria and Hungary (the transit states) and Poland (a member of the V4), where it had already been very high previously.

Policy support by immigration attitude
Table 4 presents the domestic policy-specific polarization between pro-and anti-immigration groups.The policies are arranged from left to right as in the previous table.As can be seen, similar to the transnational level, the relocation quota (and the related compensatory measures) are the most polarized policies.External and internal border control measures are also highly polarized, while the Dublin regulation and even more so externalization are less polarized among attitudinal groups.Compared to the conflict configurations between countries, the level of polarization is, however, generally considerably higher between the attitudinal potentials within the member states.This means that the latent conflict potential has not been fully mobilized in transnational conflicts.As we shall see, even at the domestic level, this potential has not been fully mobilized.Looking at country differences, there is a strong possibility of conflict between pro-and antiimmigration groups with respect to relocation quotas in some countries.Thus polarization between attitudinal groups is highest in France, a restrictive destination state, and in the transit and bystander states.They are somewhat lower in the open destination states and much lower in the frontline states of Greece and Italy.As is illustrated by Figure 4a for some selected countries, pro-immigration groups are generally in favour of relocation quotas, which means that domestic polarization is high where anti-immigration groups oppose such quotas.With the exception of frontline states like Greece and Italy, this is the case in all types of countries.Citizens who are in favour of immigration see quotas as a possible measure to accommodate refugees in an equitable way.Citizens who are opposed to immigration do not wish to adopt policies, such as relocation quota, which allow refugees to stay in Europe.The anti-immigration citizens in frontline states are an exception, most likely because their countries would benefit from relocation schemes.
With regard to the Dublin regulation (Figure 4b), the positions of the pro-immigration groups are not quite clear: large parts of these groups take a neutral position in all types of countries.Even the opponents of immigration are somewhat uncertain about this regulation, but clear-cut minorities among them support it in destination and transit states (UK, Sweden and Austria as examples), where the regulation is intended to keep refugees out, and oppose it in frontline states (Spain and Italy), where the regulation is intended to keep refugees in the country, and in France (whose opponents to immigration behave in this case like opponents in frontline states).By contrast, with regard to border control measures, the position of pro-immigration groups is not so clear, while they are generally supported by opponents of immigration are, as is illustrated by Figures 4c and 4d.Externalization (Figure 4e), finally, is generally supported by both groups, but to a somewhat greater extent by the opponents to immigration, especially in destination states like the UK or Sweden.Overall, this analysis clarifies that it is the opponents to immigration who will be decisive for some policy options in the EU member states: they oppose relocation quotas and, in frontline states, the Dublin regulation, which creates potential obstacles for these solutions.Given that they constitute large minorities or even a plurality in many countries -above all in transit states, in Latvia, Greece, and France, but also in open destination states like Sweden and the Netherlands-the governments of the respective member states are legitimately opposing these policy proposals.By contrast, the opponents to immigration are much more favourably disposed to externalization and internal and external border controls.While the proimmigration groups are not as supportive of the latter policies, they are not clearly opposed to them, which makes this type of solution potentially more consensual.
In addition to the immigration attitudes, we have also analysed the political potential of Euroscepticism.The corresponding polarization indicators can be found in Table A2 in the Appendix.As this table shows, the attitudes with regard to European integration are as polarizing as are immigration attitudes.The twin issues -immigration and European integration -solicit similar conflict configurations in the member states, which is why we do not pursue the European integration attitudes any further here.

Policy support by party family
Partisan conflicts are the most likely venue for the articulation of conflicts about refugee-related policy episodes in member states.Table 5 presents the overall polarization between voters from different party families with respect to the six policies in comparison to transnational polarization and domestic polarization by attitudes.As expected, attitudinal groups are more polarized than are political parties.In particular, the partisan conflicts are more attenuated with regard to relocation, but also with regard to border control.In contrast, there are few differences between attitudinal and partisan polarization concerning the Dublin regulation and externalization.However, even if they are less polarized than the attitudinal groups, note that policy-specific partisan conflicts are still a lot more polarized than the corresponding transnational conflicts, which is to confirm the critical role of domestic opposition to EU policy proposals.Considering the country differences in detail, with respect to relocation quotas, the partisan conflict remains intense between the left and the right in all countries except frontline states.This is shown by Figure 5a, where we present the distribution of policy-specific attitudes for the centre-left (social-democrats) and the radical right (or the conservative right in countries without a significant radical right) for some selected countries.There is also a reduced but still important conflict with respect to border control (not shown).The radical right is embracing border control internally and externally, while the centre-left is not adopting clear-cut positions in this regard.Greece is exceptional to the extent that, in this country, not only the radical right but also the centre-left is in favour of the reinforcement of the external borders, while it is the radical left (Syriza) which opposes this measure to some extent.By contrast, with respect to the Dublin regulation and externalization, we do not find any attenuation of partisan conflicts compared to attitudinal polarization.In line with the previous results, the left is uncertain about this regulation, while the radical right tends to embrace it in destination and transit states, but opposes it in frontline states (Figure 5b).Externalization, which was the least contested between attitudinal groups turns out to be more contested between parties than between attitudinal groups in closed destination states, transit states, and Poland (not shown).In destination states, the right is somewhat more in favour of externalization than the left.By contrast, in frontline states, there is hardly any difference between the two opposing sides, as they both tend to support externalization to the same extent.
Overall, we can conclude that domestic partisan polarization between left and right, while less pronounced than attitudinal polarization, is still very intense.Moreover, there are less differences between the policy domains in terms of partisan polarization than in terms of attitudinal polarization.Finally, partisan polarization is particularly pronounced in the closed destination states.Let us first consider the relocation quotas and the corresponding compensation proposals: here, all three factors strongly contribute to the conflict.The pattern of results is very similar for the two types of proposals.First, the attitudinal conflict is the main driver of these attitudes, even if we control for partisan and country effects.People who support immigration are in favour of quotas, people who oppose immigration are against them.The very strong effect of immigration attitudes implies that the partisan mobilization here has been weak, and this issue could become much more politicized in the future.This is to suggest that, given the widespread opposition to immigration across Europe, it is a very bad idea to further pursue policies involving quotas and related proposals.In partisan terms, with the exception of the radical right, there are few differences between party families with respect to quotas.It is the radical right which gives political voice to the opposition to quotas.The only exceptions to this pattern occur in Hungary and Poland, where Fidesz and PiS, officially two conservative parties, are even more opposed to quotas than the radical right.In terms of between-country differences, Italy and Greece are the two nations that really stand out.Italy and Greece have-by far-the highest support for quotas.This is not simply a front-line country effect, as support for quotas is significantly lower in Spain.
Internal border controls and the reinforcement of external borders (EBCG) are also strongly associated with immigration attitudes, but these policies are preferred by immigration opponents.Accordingly, parties on the right are more supportive of such policies than parties on the left.For these policies, however, Fidesz and PiS do not stick out as much as they did for quotas and compensations.There are hardly any country differences with regard to internal border controls, except that the British and the Romanians perceive them in a somewhat more positive light than the other Europeans, and the Portuguese are somewhat more critical in this respect.Country differences are also more contained in terms of reinforcing external borders, but populations of destination and bystander states tend to be slightly more critical of such policies than Greeks, Italians, and Germans.
In contrast to the four previous policies, assessments of the Dublin regulation are hardly associated with immigration attitudes in general.Partisan differences are also generally rather small.With regard to this policy, country-differences dominate: all countries, even Spain, are more in favour of this regulation than the citizens of the two frontline states most hit by the crisis, and the Hungarians.Finally, as we have already seen before, externalization is least structured by the three effects we are considering here.It is slightly more favoured by people holding pro-immigration attitudes.Liberal and conservative parties are somewhat more supportive of such policies and there are no systematic country-patterns.
We have run separate regressions with an interaction term to account for possible different effects of immigration attitudes in frontline states.Figure 7 presents these differences for the six policy proposals.Two results stand out: on the one hand, the effect of immigration attitudes on the policy assessment is clearly reduced in the frontline states for quotas and compensatory measures, because, as we have seen, even those who oppose immigration are also rather in favour of quotas.On the other hand, while immigration attitudes have no effect on the assessment of the Dublin regulation in most countries, this regulation is clearly more accepted by people holding pro-immigration attitudes in frontline states.Next, we present the joint distribution of conflict configurations based on multidimensional scaling (mds).While the regression approach analyses the configurations policy by policy, mds-techniques allow for a configurational analysis that takes into account all the policies at one and the same time.We first present the combination based on immigration attitudes (Figure 8) before we turn to the combination based on partisan conflicts (Figure 9).The configuration based on attitudes has a dominant horizontal dimension representing the major policies that have been adopted during the crisis -relocation quota and internal and external border control measures, and a secondary vertical dimension representing above all the failed Dublin regulation.The most consensual policy -externalization -hardly contributes to the structuring of the joint space, nor does the Dublin regulation contribute to the structuring of the joint space with party families.As we have seen, supporters of immigration tend to be in favour of quotas and against border controls, while opponents of immigration tend to be against quotas and in favour of border controls.The attitudinal divide clearly trumps the divide between member states, which again documents that the potential for further politicization has not yet been fully exploited by the political forces in Europe.The divide between member states is secondary to the attitudinal divide, which is reflected by the fact that each attitudinal camp is further divided into two groups of countries, with the eastern European supporters and opponents of immigration forming separate groups which are less favourable to the major policies than the western and southern Europeans.On the vertical dimension of the attitudinal space, which mostly represents the Dublin regulation, the opponents to immigration from the southern European frontline states form a separate cluster: they take a middling position on the main dimension, mainly because they are less opposed to relocation quotas than opponents of immigration in other countries.At the same time, they are the group which is most opposed to the Dublin regulation.By contrast, those who oppose immigration in destination states like the UK, Germany or Sweden are the groups most in favour of this regulation.Note that the second dimension does not contribute much to the structuring of the space in terms of immigration attitudes.As for the combination of conflicts between partisan families with transnational conflicts, the dominant horizontal dimension is the same as in the previous graph, but the vertical dimension is not so much related to the Dublin regulation.Instead, it refers to aspects of border control which do not always go together with positions on quotas in some countries.On the horizontal dimension, in most of the countries the radical right is opposed to the left (radical left, greens and social democrats).Importantly, the radical right also includes the conservative parties in Hungary (Fidesz) and Poland (PiS).The conservative parties are marked in bold in the graph in order to show that they are spread considerably across the horizontal axis.While most of them are located in the middle of the space, with the Austrian conservatives closest to the cluster of the radical opponents of burden-sharing, note that the German CDU as well as the Portuguese conservatives (together with some liberal parties) are part of the left cluster which is favouring burden sharing.The conservative/Christian-democratic parties have reacted quite differently to the refugee crisis in the different countries (Hadj Abdou, Bale, and Geddes 2021; Abou-Chadi, Cohen, and Wagner 2021), which is reflected in their voters' policy positions as we can see here.On the other hand, the radical right in the two frontline states (Greece and Italy) is not part of the radical opponent cluster, but situated in the middle of the space, given that it is also rather favourable to quota schemes.On the vertical dimension, there are party families in some countries which differ with respect to the positions on border control -some oppose some aspects of border controls, while others generally support border controls.In the group opposing border controls we find Portuguese parties across the entire spectrum and centre-left and centre-right parties from eastern Europe, while the group supporting border controls includes mainly right-wing parties from the frontline states and the UK, but also the centre-left party from Greece and the German liberals.

Conclusion
In terms of transnational conflicts, we have found the expected opposition between the frontline states (Greece and Italy), on the one hand, and the V4 countries (augmented by eastern European bystander states) on the other hand.The contrasting stance of the policymakers from these countries during the refugee crisis is reflected in their voters' positions.Citizens from western European destination, transit, and bystander states generally take more moderate positions on the main dimension of conflict, which is defined by relocation policies.
At the domestic level, we found the expected opposition between nationalists and cosmopolitans which are politically articulated by the radical right and some nationalistconservative parties on the one side, and by the left and some parties of the mainstream right on the other side.We found that the same dimension structures the debate at the national and at the EU level.The domestic polarization appears to be more intense than the transnational one, especially in terms of immigration attitudes.When analysing the combined transnational and domestic conflict configuration, this is reflected in the greater structuring capacity of domestic conflicts: transnational conflicts appear as secondary to the domestic attitudinal conflicts, where they form a subdivision of the two attitudinal camps, and they are also secondary to the domestic partisan conflicts, where they divide the partisan camps with regard to some aspects of the border control policies.The transnational conflicts are ultimately rooted in the domestic conflict structure of the member states, where the opponents of immigration constitute the critical factor.In some key countries, they make up a plurality or even a majority of the population which is mobilized by radical right and nationalist-conservative parties, depending on the country.
The implications for European policymakers in the domain of asylum policy are quite clear.The conflict potentials of immigration policies have not yet been fully mobilized.They are very large and they have markedly increased in the destination states of northwestern Europe over the last few years.This means that policymakers are facing very strong constraints in terms of what is possible in this policy domain.As long as the critical underlying attitudinal potentials are not fully mobilized and as long as the parties mobilizing the opponents to immigration do not constitute the dominant coalition partner in government, joint solutions at the European level remain possible even in the most contested policy domains.However, when opponents to immigration become dominant in a given country and the parties mobilizing them become the dominant coalition partner or the exclusive governing party, as has been the case in Hungary and Poland (and other eastern European countries), the respective member states can legitimately prevent joint solutions, even if such solutions are supported by most of the other member states and, above all, by the frontline states.Given this state of affairs, relocation schemes are not a politically feasible option today.The Dublin regulation benefits from the fact that the voters even in the frontline states do not seem to be aware of what this policy exactly implies.However, voters in frontline states are well aware that their burden is not sufficiently shared by the other member states.Finally, the more restrictive policies of border control and externalization receive more support.Externalization policies are least contested.

Figure 1 :
Figure 1: policy-specific distribution of support, by country type a) Relocation quota

Figure 2 :
Figure 2: Transnational conflict configuration according to citizens' policy positions in the refugee crisis: mds-solution

Figure 4 :
Figure 4: Policy support by immigration attitudes a) Relocation quota: support

Figure 5 :
Figure 5: Policy support by party family a) Relocation quota

Figure 6 :
Figure 6: Transnational and domestic conflict configurations according to citizens' policy positions in the refugee crisis: OLS-regression coefficients

Figure 7 :
Figure 7: the effect of immigration attitudes on the six policy positions in frontline states and other states

Figure 8 :
Figure 8: Transnational and domestic conflict configurations according to citizens' policy positions in the refugee crisis and immigration attitudes: mds-solution

Figure 9 :
Figure 9: Transnational and domestic conflict configurations according to citizens' policy positions in the refugee crisis and party families: mds-solution 1)

Table 1 :
transnational polarization by policy and country, Kolmogorov-Smirnov statistic 2

Table 2 :
immigration attitudes by country (ordered by share against)

Table 3 :
The case of Sweden

Table 4 :
Domestic polarization between pro-and anti-immigration groups, by policy and country, Kolmogorov-Smirnov statistic 5

Table 5 :
Comparison of overall polarization, transnationally and domestically by attitudes and party family, across policies: Kolmogorov-Smirnov statistic