Constraining dissensus and permissive consensus: variations in support for core state powers

Abstract Public opinion regarding EU integration has become increasingly important in light of both increasing public scepticism to integration and the EU’s gradual shift towards integration of core state powers (CSPs). This article contributes to an emerging literature studying support for integration of core state powers by investigating whether such integration, given its greater potential for polarisation and activation of nationalist identities, is more likely to be opposed than regulatory integration. The results show that some CSP integration may meet greater support than regulatory integration. They also show that a larger number of exclusively national citizens support externally, rather than internally, oriented CSP integration. This suggests that what policy area a policy belongs to may be more important for predicting contestation of its integration than whether it is a core state or regulatory power.

the nature of these powers makes it likelier that their integration will polarise public opinion and activate nationalist identities and identity-based contestation of integration (De Wilde and Zürn 2012;Hooghe and Marks 2009: 13). However, the literature on the politicisation and contestation of EU integration does not sufficiently investigate whether all core state power integration is similarly likely to be opposed, and if it always meets greater opposition than traditional regulatory integration. This article seeks to answer the following question: 'Is core state power integration always opposed to a greater degree than regulatory integration?' This question is important because public opinion may be one reason why for instance defence and military policies have not been integrated to any great degree, despite the functional benefits of such integration (Bremer et al. 2020: 57). Using repurposed Eurobarometer data from 2013 to 17, this article makes two contributions to the existing literature on core state power integration and the politicisation of European integration that followed the ratification of the Maastricht Treaty in 1992: I first show that some core state power integration enjoys greater public support than regulatory integration, even if the former is typically considered more likely to be contested. Second, I show that a larger number of citizens identifying exclusively with their nation-states support the integration of core state powers that are externally, rather than internally, oriented. These variations are found even if exclusively national citizens are theoretically more likely than those with more inclusive national identities to equally oppose both forms of core state power integration.
The results suggest a nuanced picture of support for core state power integration, with important implications for the EU's continuing integration of core state powers: Since opposition to integration appears to vary by policy area, and not by whether policies are regulatory or core to state functioning, EU leaders may be more able to integrate core state powers without meeting contestation than is often presumed. The same leaders may thus be shielded from 'Juncker's Curse' , the name given to a phenomenon in which voters punish politicians for unpopular but necessary reforms (Bremer et al. 2020;Buti et al. 2009), even when advocating for shifting the power to make decisions in particularly salient policy areas from the national to the EU level.
This article contributes to the literature on public support for the new policy integration after Maastricht and the inherently multi-faceted nature of support for European integration (Boomgaarden et al. 2011;de Vries and Steenbergen 2013;Kanthak and Spies 2018). However, the results are arguably relevant for theorising about international policy integration outside of the EU context. The reason is that they indicate that citizens with strong attachment to their nation-states may primarily oppose supranational integration of core state powers where they feel national autonomy is most important, such as in the provision of social welfare and domestic security.

Conceptualising differentiated public opinion
As Kanthak and Spies (2018) show, the literature investigating public support for EU integration has traditionally been bifurcated, with one strand focussing on diffuse support for the European Union (such as Foster and Frieden 2021;Gabel 1998;Harteveld et al. 2013;Hooghe and Marks 2005) and another on support for specific policies such as fiscal and monetary integration (Bechtel et al. 2014;Hobolt and Wratil 2015) or common security and defence policies (Schilde et al. 2019;Schoen 2008). Broadly speaking, the literature finds three main motivators of diffuse and specific support for European integration: Utilitarian considerations, elite cues and different configurations of individual identities (Foster and Frieden 2021;Gabel 1998;Gabel and Palmer 1995;Harteveld et al. 2013;Hobolt and Wratil 2015;Hooghe and Marks 2005;Kuhn and Stoeckel 2014). However, with some exceptions (Boomgaarden et al. 2011;de Vries and Steenbergen 2013;Kanthak and Spies 2018), what motivates varying support for EU policies, and especially the integration of core state powers, is still under-investigated.
Core state powers can either be conceptualised as powers that directly or indirectly relate to a state's capacity for coercion, such as powers of taxation, law enforcement or border control (Genschel and Jachtenfuchs 2016: 43), or as 'essential competences and essential powers of public authority' (Kuhn and Nicoli 2020: 7). I find the latter definition more useful because it captures also non-resource based CSPs and will use this definition throughout the paper.
Despite a large literature investigating both diffuse support for the EU and specific policy integration, little is known about whether opposition to core state power integration is universally greater than to regulatory integration. This question has become increasingly important in light of the functional need for such integration revealed by the sovereign debt and migration crisis (Kuhn and Nicoli 2020). Integrated migration policies could for instance allow Schengen frontier countries like Italy greater control over their own borders. Similarly, smaller member states might be more able to exercise their external powers as sovereign actors if backed by integrated military capacities. If making up their minds solely based on utilitarian calculations, we would thus expect citizens to express support for some core state power integration.
However, even where CSP integration may potentially benefit EU member states it is a theoretically reasonable assumption that it will be met with greater opposition than regulatory integration. This is because the expansion of the scope of EU integration, into policy areas at the core of domestic politics, that core state power integration represents is more likely to polarise public opinion, be mass politicised and activate nationalist identities than what will be the case for less salient regulatory integration (De Wilde 2011;De Wilde and Zürn 2012;Hooghe and Marks 2009;Kriesi 2016;Zürn 2019). This article investigates the empirical support for this assumption by comparing public support for several instances of core state power integration to support for one example of regulatory integration, digital single market policies. While using only one regulatory policy as a baseline makes it more difficult to generalise the results to other policy areas, it is arguably enough to answer my research question: Finding that some core state power integration is supported to a greater degree than even one regulatory policy would show that CSP integration is not universally opposed to a greater degree than regulatory integration. I hypothesise the following: H1: Citizens will express lower levels of support for integration of core state powers than regulatory policies. Some literature suggests diverging preferences for CSP integration, with citizens supporting core state power integration where it benefits them or their countries (Freudlsperger and Jachtenfuchs 2021: 124). Bremer et al. (2020) similarly show that while right-wing populist voters consistently oppose core state power integration, the extent to which they do so varies by policy. However, Bremer et al. do not compare support for CSP and regulatory integration. Furthermore, their data is restricted only to the situation in April 2018 and in eleven of 27 member states. This article, in contrast, uses Eurobarometer surveys fielded between 2015 and 2017 to test the assumption that citizens are more likely to oppose integration of a range of CSPs than a regulatory policy like digital single market policies. The key contribution of this paper is thus to compare support for CSP and regulatory integration, using data from several years and all member states. However, the limited time frame means we cannot automatically assume that the results constitute a general trend.
One reason to expect the policy-specific differences found by Bremer et al. (2020) to be representative of a broader trend is that core state power integration can constrain either the internal or external autonomy of member states. For instance, previous literature finds consistently high support for policies such as common defence policies, often more so than for economic integration (Eichenberg and Dalton 2007;Schilde et al. 2019). Support for integration of externally oriented CSPs, which constrains the member states as foreign policy actors, is thus likely to be greater than for internally oriented core state power integration that constrains domestic state autonomy.
This may be because externally oriented CSP integration could also constitute a 'rescue' of nation-states (Milward 1999) by strengthening the military and diplomatic capacity of member states and thus their autonomy to act in the foreign realm. Internally oriented integration, however, is more likely to be seen solely as an imposition on autonomy in areas that are among the most salient to domestic politics, such as fiscal and migration policy. Domestically oriented CSP integration is thus arguably more likely to polarise popular opinion and activate patterns of identity-based contestation, with opposition as a likely result. However, as the case of EU defence policy shows, there may be regional variations to this pattern (Peters 2014). For instance, citizens of Nordic countries may be less supportive of both common monetary and defence policy, as evidenced by the Danish opt-out from common security and defence policies (CSDP) (Adler-Nissen 2014) and the Swedish non-adoption of the euro, than other West Europeans. I hypothesise the following: H2: Citizens' support will be lower for internally oriented core state power integration than externally oriented core state power integration.
This external-internal dimension may be more or less salient depending on configurations of national identities. To test whether different conceptions of national identities produce different levels of support for internal or external core state power integration, the second step of my research design uses Eurobarometer data from 2013 to 2017 to compare support for two sets of core state powers among those identifying only with their nation-states. I first measure support for externally oriented common foreign and defence and security policies, before comparing this support to that for two instances of internally oriented economic core state power integration: economic and monetary union (EMU) and mutualised sovereign debt in the form of Eurobonds.
While external CSP integration could potentially expand state autonomy and thus be more attractive than internally oriented CSP integration to all citizens, I believe exclusively national citizens are less likely than those with more inclusive identities to evaluate integration according to such a logic. Rather, because they are both likely to see the nation-state as the legitimate locus of national decision-making, especially in policy areas where salience is high, and to be particularly critical of the EU's constraint on national sovereignty in salient policy areas (Hooghe and Marks 2009), they are likely to be similarly critical of both internal and external CSP integration. This is because the two forms of integration are similar in how they both shift powers of decision-making in policy areas core to statehood from the national to the EU level.
This similarity also leads me to believe that internal and external CSP integration will have a similar capacity to awaken the kind of nationalist identification that has been found to be key for mobilising Eurosceptic sentiments among exclusively national citizens (De Wilde and Zürn 2012; Hooghe and Marks 2009). I therefore hypothesise: H3: Support for internally and externally oriented core state power integration will be largely equal among citizens with exclusively national identities.
My study thus contributes to the literature on core state power integration by investigating both whether there is a systematically lower level of support for core state power integration than regulatory integration, and whether there is divergence in support between internal and external CSP integration, either in the population as a whole or among those who identify solely with their nation-states.

Study 1
My first analysis mimics, using observational data, the structure of a within-subjects experiment, where each respondent is sequentially exposed to a control condition and every treatment condition (Choi 2021). I create a factor of all policies, with digital single market policies as the baseline. This factor, which features both the regulatory policy and each core state power as levels, is my independent variable. The dependent variable is a dummy variable indicating support or opposition to each of them. Each coefficient thus shows the relative change in the probability that a policy will be supported when compared to digital single market policies. This design lets me test both the hypothesis that support for core state power integration will be systematically lower than for an instance of regulatory integration, and whether the extent to which this is the case varies between different types of CSPs.
While digital single market policies may be more salient today than other regulatory integration, the proposed policies mainly have the goal of regulating and eliminating barriers to cross-border commerce in the digital marketplace. This makes them more similar to the regulatory and market harmonising integration that has traditionally been core to EU policy integration than to the integration of essential functions and powers of member state governments that is so characteristic of the EU's recent turn to core state power integration (European Commission 2015; Majone 1999). While the ideal would be a composite index that averaged the levels of support for several regulatory policies the data do not include questions about other regulatory policies than digital single market policies. This makes it more difficult to generalise the results to all instances of regulatory integration.
My data come from 2015 to 17 Standard Eurobarometer surveys and include, as for study 2, every member state but Croatia, which entered the EU in 2013. Croatia was excluded from the data to include only member states that have been in the EU since at least the 2004 enlargement. Post-stratification weights are implemented in all samples. While it is difficult to make temporal generalisations from such a short time series, necessitating further research, these surveys are selected because they let me compare support for several core state powers, including both EMU and Eurobonds, to support for a regulatory policy. The lack of questions specifically measuring support for regulatory integration before 2015 makes it difficult to use a similar design for 2005-15, while the absence of questions about Eurobonds after 2017 makes it difficult to extend the study to the present while comparing support for both EMU and Eurobonds.
To account for individual-level effects specific to each respondent as well as specific effects of country and time, I use three-way fixed effects. I also cluster the standard errors on country, year and subject (Mummolo and Peterson 2018;Zeileis et al. 2020). This has the benefit of controlling for all relevant variables at both the individual and contextual level. While I would have ideally preferred to establish what individual-level variables are most strongly associated with support for specific instances of core state power integration, this is impossible when including subject-specific fixed effects.

Dependent variable
The dependent variable is support or opposition to integration, coded as a dummy where 1 indicates support and 0 opposition to the policy in question. Using a long dataset in which the respondent's value on the dependent variable is recorded for all included policy areas, I analyse whether the same individuals express different levels of support for all policies. This lets me establish whether respondents are, on average, consistently less supportive of core state power integration than regulatory integration.

Independent variables
As my first analysis is essentially an analysis of whether the same people express different levels of support for individual policies, I implement fixed effects for years, countries and subject-specific characteristics (Mummolo and Peterson 2018). The policies serving as my 'experimental conditions' are support for EMU, common defence and security policies, common energy policy, common foreign policies, common migration policies, Eurobonds, and free movement of persons inside the EU. The baseline question is one asking whether respondents support digital single market policies. This policy is less likely to be politicised and to mobilise nationalist identities. All independent variables use the wording 'Please tell me, for each statement, whether you are for or against it' , with the relevant policy area featuring in the second part of the sentence. The possible responses are For, Against or Don't Know. Don't Know are throughout coded as missing and removed from the analysis.
While energy policies may appear less essential to the state than other policies, energy policy has important implications for member states' economies, social conditions and security (Hofmann and Staeger 2019;Natorski and Surrallés 2008). Energy policy may thus have a geopolitical and economic salience in many states that makes it reasonable to treat it as a core state power.
These core state powers are included because they feature in surveys that let me both compare support for CSP integration and regulatory integration and measure variations in support for a very broad range of core state power integration: While Eurobonds and EMU constrain states' previously exclusive right to issue debt and currency, which has been a particularly salient question in the years following the sovereign debt crisis, common energy policies impact an economic core state power unrelated to the state's historically exclusive right to mobilise fiscal resources. Common defence and foreign policies, on the other hand, constrain states' external autonomy, while common migration policies and the right to freedom of movement between member states for EU citizens curtail member states' ability to regulate access to their territories for both EU citizens and non-EU migrants. The included variables thus capture support for the integration of core state powers that adhere closely both to the resource mobilisation definition used by Genschel and Jachtenfuchs (2016) and the definition of CSPs as essential powers and functions of governments used by myself and Kuhn and Nicoli (2020).

Results and discussion
As Figure 1 (with coefficients reported in online appendix item A1.5, supplementary materials) clearly shows, respondents express very different levels of support for core state power integration. I reject H1, as some core state power integration clearly enjoys greater support than digital single market policies. I also reject H2, that internally oriented core state power integration meets greater opposition than externally oriented core state power integration, as the variations do not systematically relate to the orientation of integration. For instance, support for common defence policies is greater than for digital single market policies, while the opposite is the case for common foreign policies. Free movement for EU citizens also enjoys greater support than all other policies surveyed, even though it constrains national sovereignty in the highly politicised field of migration control.
My results seem instead to suggest that a more persistent divide exists between economic and non-economic core state power integration, with the former being most strongly contested. A robustness test analysing support for EMU, CSDP and common foreign policies in the years 2005-19, reported in A2.1, confirms that EMU has met the greatest opposition of the three policies in the years after the Eastern Enlargement, with common foreign policies being more strongly opposed than common defence policies. This confirms the results shown by Figure 1 and suggests that a divide in support between integration of economic and non-economic policies has been a persistent feature of public opinion regarding CSP integration.
I also analyse data from 2015 to 2019, as Standard Eurobarometer surveys in this period include all relevant variables other than one measuring support for Eurobonds (results shown in online appendix items A2.3 and A2.4). The patterns found are very similar to the main analysis, supporting the rejection of both H1 and H2. Figure 2 shows broadly similar patterns at the regional level: Though there are some regional variations, EMU and Eurobonds seem to produce the greatest opposition in most regions. There is furthermore no consistent pattern of greater support for externally rather than internally oriented integration across the regions. This shows that the results are robust to regional variations.
However, Figure 2 also reveals some of the regional variations hinted to by the literature. For instance, Nordic citizens appear more sceptical of core state power integration across the board, as evidenced by their strong opposition to CSDP, which enjoys much greater support in other regions. They are also equally likely to oppose EMU and common foreign policies.
Another interesting variation is found in the regional levels of support for common energy policies. Citizens of countries belonging to Central and Eastern Europe appear less likely to support common energy policies than citizens of other regions. One possible explanation is the history of many of the Eastern Enlargement countries as post-Soviet republics. This political history may make them more willing to chart a different path than the rest of the EU in the field of energy policy. Additionally, concerns regarding the economic viability of specific sectors in the relevant countries, such as Polish concerns regarding the viability of their coal industry (Ćetković and Buzogány 2019) can make it more likely that such integration will be opposed in this particular region. This shows how regional contexts are likely to play a role in producing opposition to CSP integration. Figure 3 shows the results of a robustness test of regional support for CSP integration using data from 2015 to 19. I find that the patterns are largely similar between the main results and the robustness test (coefficients are reported in A2.5).
Based on this I draw three preliminary conclusions: Rejecting H1, I show that citizens do not necessarily express greater opposition to core state power integration than regulatory integration. This nuances a key assumption of the literature on politicisation of EU integration, which is that core state power integration is more likely than regulatory integration to be contested because it requires ceding sovereignty to the EU level in policy areas that are core to statehood and particularly salient to domestic politics (De Wilde and Zürn 2012; Hooghe and Marks 2009). I also reject H2, that internally oriented core state power integration will consistently meet greater opposition than external core state power integration. Instead, a more salient divide seems to exist between economic and non-economic core state power integration. The greater opposition to economic CSP integration could stem from how increasing integration of monetary and fiscal policies has been heavily contested both during and after the sovereign debt crisis that struck the Eurozone periphery countries. My main analysis does not cover the Eurozone crisis years. However, the frequently emotive discourses about how the crisis should be solved (Schmidt 2014) makes it likely that this contestation would lower support for the EU's economic framework also in the years covered by my study.

Study 2
My second analysis uses Eurobarometer data from 2013 to 2017 to compare support for external and internal CSP integration among citizens with exclusively national identities. It thus tests H3, which states that exclusively national citizens will express largely similar preferences towards both external and internal CSP integration, by comparing exclusively national citizens' support for EMU, Eurobonds and common defence and foreign policies. As mentioned, I believe exclusively national citizens will express similar levels of support for internal and external CSP integration because they are likely to consider the loss of formal sovereignty due to integration similarly problematic for both externally and internally oriented core state powers. This analysis proceeds in two steps: I first test, using a regression model that includes both a measure of national identity and relevant control variables, whether the effect of national identity varies across policies. However, because these effects may be sensitive to variations in the underlying levels of support for each policy, the second step of the study compares the percentages of exclusively national citizens expressing support for each of the policies. Thus, study 2 investigates both variations in the effect of identity on support for external and internal CSP integration, as well as what shares of exclusively national citizens express support for each of the four policies.
The four policies mentioned are particularly well-suited to investigating how the orientation of integration shapes public support for it because they are the instances of CSP integration for which support is surveyed least likely to be seen as imposing dual constraints on states' autonomy: While policy integration related to migration, energy and freedom of movement for EU citizens is likely to be seen as constraining what policy outcomes governments can seek both domestically and in their interactions with other states, economic and monetary union and debt mutualisation through Eurobonds are instances of integration less likely to be seen as constraining the state externally. Conversely, common foreign and defence policies clearly constrain EU member states externally but are arguably less likely to be seen as doing so domestically.

Dependent variables
The first dependent variable of study 2 is support for European monetary union (EMU). The variable measuring support for EMU asks respondents whether they favour or oppose common economic and monetary policies: 'Please tell me, for each statement, whether you are for or against it: A European economic and monetary union with one single currency, the euro' . The three possible responses are For, Against or Don't Know.
The second variable, support for Eurobonds, defined by the Eurobarometer as 'common European borrowing' , uses a Likert scale with possible responses being Strongly in favour/Fairly in favour/Fairly opposed/Strongly opposed/DK. The phrasing of the question is: 'Thinking about reform of global financial markets, please tell me whether you are in favour or opposed to the following measures to be taken by the EU: The introduction of Eurobonds (European bonds) (INT.: IF NEEDED, PRECISE "COMMON EUROPEAN BORROWING")' . I recode the variable as a dummy, with 'strong' and 'fairly strong' support coded as 1 and 'strong' and 'fairly strong' opposition as 0. This makes it possible to compare the coefficients to those of the other models, whose dependent variables are dummies.
The wording of the question for the dependent variable measuring support for common defence and security policies (CSDP) is: 'Please tell me, for each statement, whether you are for or against it: A common defence and security policy among EU member states' . The question related to the fourth dependent variable, support for common foreign policy (CFP), is worded 'Please tell me for each statement, whether you are for it or against it: A common foreign policy of the Member states of the EU' . The response categories for these variables are also For, Against or Don't know. For all variables, 'Don't know' is treated as missing data.

Variable of key interest
In order to operationalise exclusively national identity, the key theoretical concept of study 2, I create a dummy variable coding everyone who states that they identify exclusively with their nation-states with 1 and everyone else 0. This operationalisation is well-established in the EU public opinion literature (Hobolt and Wratil 2015;Hooghe and Marks 2005). This study focuses on the effect of territorially exclusive identities because the distinction between territorially exclusive and partially territorially inclusive identities has been found to be more important for predicting Eurosceptic sentiments than variations between identities featuring both national and European components (Hooghe and Marks 2009).

Control variables
I include a range of control variables. First, I include trust in the EU and national governments. Trust in the EU is a proxy for diffuse support for EU membership, commonly seen as a predictor of specific support for EU policies. Trust in national governments is included as a control variable for the same reason, as many rely on heuristics relating to their trust in national institutions when making up their minds about EU institutions (Harteveld et al. 2013). Trust in the EU and the national government is not a perfect proxy for diffuse support, but it may be seen as a reservoir of good will towards both institutions that may come close to diffuse support (Armingeon and Ceka 2014: 88-96). A better measure of support for the EU would be a question asking either whether respondents believe the EU is a good thing for their country or one asking if their country benefits from membership. However, these questions have not consistently featured in Standard Eurobarometer surveys in the relevant years.
I also include variables for gender, age and education level measured as the age at which the respondent finished schooling. To operationalise ideology, I use a 11-unit scale with 0 indicating far-left positioning and 10 far-right positioning. Finally, I include socioeconomic variables such as dummies for whether the respondent is a manual worker, as well as an indicator of how respondents perceive the national economy. These are found to predict support or opposition to EU integration (see for instance Carrubba and Singh 2004;Gabel 1998;Gabel and Palmer 1995;Hobolt and Wratil 2015;Hooghe and Marks 2005;Schoen (2008).
The measure of perceptions of the national economy combines respondents' perception of the state of the national economy and labour market. A principal component analysis (Hayton et al. 2004) suggests that the variables constitute one dimension. This is strengthened by a Cronbach (1951) test of reliability, as the index has an alpha of 0.8.

Model
As the baseline levels of support for the four policies are likely to be different for each country in any given year, I use multilevel models with random country-year intercepts to account for these variations. The model comparing support for the four policies is formalised as: The level of support Y for individual i in year t for any of the four policies thus becomes a function of a country-year specific constant (β 0j ), socio-demographic characteristics (Χ 1 ), levels of trust in the EU (Χ 2 ), evaluations of the national government and economic situation (Χ 3 ) and a dummy for exclusively national identity (Χ 4 ). The control variables are added to allow for increased confidence that we are measuring the effect of identity, as they reduce the potential for confounding due to omitted variable bias.
In all models, units with missing data on any modelled variable are removed from the analysis via listwise deletion.

Limitations
There are limitations to this design. Omitted variable bias may still be an issue, as relevant control variables may have been omitted from the analysis, potentially leading to spurious correlations. I have, as described above, sought to minimise this risk by including a broad range of relevant control variables. In addition, the period of the analysis, 2013-2017, is one in which the EU was marked by what may be termed a poly-crisis (Gänzle et al. 2019), in which debates surrounding EMU and common EU migration policies were likely to be polarised due to external crises. However, my ability to expand the timeframe of this analysis is somewhat limited, as both the specific policies for which support is measured and the wording used to do so has changed periodically. This makes comparison across longer periods difficult.
Missing data may also be problematic, particularly if missingness on important variables is highly correlated with ideology or income. There are more missing data on the variable measuring support for Eurobonds than on the other dependent and independent variables used in the study (see online appendix items A1.1-2 for a graphical overview). However, as the missingness is consistently high across the full range of values for both income and ideology, bias due to missingness is unlikely to be a large problem. The fact that there are only moderate levels of missing data on the other variables, whether dependent or independent, and that this missingness does not vary strongly as a function of ideological orientation or income, supports the assumption that bias due to missingness will generally be a limited concern in this study. Table 1 shows that the effect of identity varies across the full range of policy issues (see online appendix for predicted probability plots derived from Table 1 together with results from all robustness tests). This weakens H3, by hinting that exclusively national citizens may have very different attitudes towards external and internal CSP integration. The robustness tests confirm that the effects are similar when one omits controls for trust in the EU, trust in the national government and sociotropic evaluations of the economy (see A2.10-11). They thus appear robust to bias caused by either potential endogeneity or multicollinearity.

Results and discussion
However, because the observed variations may be driven by underlying differences in support for each policy, I also compare what percentage of exclusively national citizens express support for each instance of core state power integration. This is needed to conclusively accept or reject H3. Figure 4 shows the levels of support for each policy among exclusively (A) and inclusively (B) national citizens. The figure shows that exclusively national citizens express diverging attitudes towards internal and external CSP integration: While 65 percent express support for a common defence policy, only 51 percent do so for EMU. For common foreign policies and Eurobonds, the numbers are 60 and 13 percent. The fact that support for external CSP integration is greater among exclusively national citizens than support for internally oriented integration allows me to reject H3. On the other hand, chart B shows that inclusively national citizens express similar levels of support for all CSP integration other than Eurobonds. The same is found across all regions (see A2.16-19). Exclusively national citizens thus express more nuanced preferences for core state power integration than my H3 assumes: Rather than expressing similar attitudes towards both forms of integration, a greater share of exclusively national citizens seem to prefer core state power integration that is externally, rather than internally, oriented. This may be because they, if presented with a trade-off between expanding de facto autonomy of the state in areas where it may be limited and ceding its formal sovereignty, may place lesser weight on formal sovereignty than I hypothesised. This is a particularly plausible explanation for the higher level of support for common foreign and defence policies, as member states seeking to strengthen themselves as foreign policy actors may find their ability to do so hampered by limited military and diplomatic resources. The theoretical mechanism that I saw as likely to produce greater support for external CSP integration in the broader public may thus help explain the preferences of a group that I instead assumed would express almost equal levels of support for both external and internal core state power integration. My results thus hint at the need for more nuanced theoretical assumptions about the relationship between identity and support for core state power integration. Future research will also need to investigate whether these divergences are a persistent feature of public opinion, applicable even when the EU's economic policies are less strongly contested.
The analysis also yields two additional insights: First, we find approximately the same level of support for Eurobonds among inclusively and exclusively national citizens. A reasonable assumption is that politicisation of mutualised debt, through the framing of mutualised debt and bailouts as a 'rescue' of other countries (Schmidt 2014), would activate nationalist identities and in-group identification to a greater degree than EMU. This could arguably lead to greater opposition to Eurobonds among those with exclusive national identities compared to those with territorially inclusive identities. That there are nevertheless small differences between the two groups' support for this policy suggests that Eurobonds have characteristics that make them equally controversial in both groups. Understanding what these characteristics are is an important task for future research.
We also find less support for common foreign policies than common security and defence policies among exclusively national citizens, even if the differences are small. One explanation may be that common foreign policies encompass a broader range of policies than common defence policies. They could thus lead to integration in areas where countries have more variable real autonomy, and where it might also be greater than what is the case for military policies. This could again lead to them being perceived as more problematic impositions on national sovereignty than common defence policies.
One should not generalise from common foreign and defence policies to all externally oriented core state power integration, nor from EMU and Eurobonds to all internally oriented CSP integration. However, the greater support for external rather than internal oriented core state power integration hints that individuals with exclusively national identities are more likely to support core state power integration that can be seen as strengthening real national autonomy, even if it constrains member states' formal decision-making authority in the same fields.

Conclusion
The shift from regulatory and economic integration to one encompassing the integration of powers and functions core to state functioning has been one of the most important developments of the post-Maastricht Treaty EU. However, we still know little about whether integration of different core state powers is opposed to different degrees, and why this may be. As these policies, as a group, have a salience that makes it likely for them to be mass politicised and contested on the basis of identity, understanding whether they are universally contested to a greater degree than regulatory integration is an important question (De Wilde 2011; Hooghe and Marks 2009;Zürn 2019).
In contrast to what I hypothesised, I find that support for some CSP integration, such as common defence policies and freedom of movement for EU citizens, is greater than for regulatory integration, even if the former may be assumed to be more amenable to public contestation. I also find that a greater share of exclusively national citizens express support for externally, rather than internally and economically, oriented CSP integration. This could be explained by how internally oriented CSP integration constrains national autonomy in areas with both higher salience and greater de facto national autonomy, such as fiscal and monetary policies. External core state power integration, on the other hand, might expand states' capacity to act externally where their individual resources might otherwise have been limited, for instance in the field of defence policy. Such expansion of real, national autonomy may be more attractive to those who feel strongly attached to their nation-states.
The results suggest that we need to nuance our assumptions about what EU policy integration is most likely to be contested by EU citizens: It does not appear to be the case that highly salient core state power integration, in which policies core to statehood are integrated at the European level, will necessarily be contested to a greater degree than regulatory integration. We also need a more nuanced understanding of how national identity relates to opposition to core state power integration: My results show that even exclusively national citizens, who are theoretically likely to have similar attitudes towards both internal and external core state power integration because they both require member states to cede formal sovereignty to the EU in highly salient policy areas, have nuanced attitudes towards such integration. This is evident in their stronger preference for external, rather than internal, CSP integration. Integration is thus more likely to be contested where the EU seeks to further integrate core state powers belonging to the domestic realm. This is a potential challenge to an EU that has long sought to harmonise for instance member states' fiscal and migration policies. Such integration is both highly salient and likely to be seen as an imposition on a state's domestic autonomy.
Understanding these nuances is also important for our understanding of how support for the EU's CSP integration is structured and what international policy integration is likely to be supported outside of the EU context. However, further research is needed to establish whether similar patterns exist when using other regulatory policies as a baseline, and when comparing support for other external and internal CSPs.
My results point to two fruitful avenues of future research: The first is the need to nuance our understanding of what drives support for the integration of certain core state powers. Given that Eurobonds are almost equally likely to be supported by citizens with exclusively and inclusively national identities, it seems clear that there are distinct and different causal mechanisms at work for each policy. Disentangling these causal mechanisms requires a policy-level approach to analysing support for core state power integration. This analysis should combine data describing elite discourses regarding integration of specific core state powers with survey data measuring popular support for integration of the same core state powers. This is potentially useful because elite cues are important predictors of public support for European integration and because the elite preferences for core state power integration that can be revealed through party manifestos themselves act as drivers of core state power integration (Genschel and Jachtenfuchs 2016;Steenbergen et al. 2007).
The second question is whether these patterns, which suggest a congruence between public opinion for EU integration and functionally differentiated integration, equate to support for institutionalised differentiated integration. The question of what individuals actually support differentiated integration and why is still an under-investigated issue (de Blok 2021; Leuffen et al. 2022). Thus, investigating whether there is support for mechanisms of differentiation and what drives it is a key area of future public opinion research.
A nuanced understanding of what drives opposition to core state power integration is finally important for understanding when such integration will be seen as legitimate by European citizens. This could make it less likely for functionally needed integration to be halted by a constraining dissensus (Hooghe and Marks 2009;Hutter and Grande 2014). Understanding the limitations of the constraining dissensus is thus important to both policy-makers and scholars of European integration.

Disclosure statement
No potential conflict of interest was reported by the author.