Mapping public support for the varieties of differentiated integration

Abstract

This article maps and investigates public support for different types of differentiated integration (DI) in the European Union. We examine citizens’ preferences for DI using novel survey data from eight EU member states. The data reveals substantive differences in support for different types of DI. Factor analyses reveal two dimensions that seem to structure citizens’ evaluations of DI. The first dimension relates to the effect of DI on the European integration project, the second concerns the safeguarding of national autonomy. Citizens’ attitudes on this second dimension vary substantively across countries. General EU support is the most important correlate of DI support, correlating positively with the first and negatively with the second dimension. Our results underline that while citizens generally care about the fairness of DI, balancing out their different concerns can be a challenging political task.

Keywords

Core Europe differentiated integration factor analysis public opinion two-speed Europe

Introduction

In debates on the future of the European Union (EU), differentiated integration (DI) is regularly considered one possible option to overcome heterogeneity-induced gridlock. The introduction to this special issue summarizes the risks and opportunities that the academic literature attributes to DI (Schimmelfennig et al., 2023); however, we still know surprisingly little about public opinion on DI. Even though the literature discusses a variety of DI models and procedures, current knowledge about citizens’ support for DI remains largely limited to analyses of a single Eurobarometer item, inquiring about the support of a ‘two-speed’ Europe model of DI (De Blok and De Vries, 2022; Leuffen et al., 2022b). In this article, we, therefore, address a series of foundational questions about public support for DI. First, how supportive is the public when it comes to different types of DI? Second, can we identify underlying dimensions along which citizens form their preferences regarding DI? Third, how do different citizens–at the individual and national levels–diverge in their assessments of different types of DI? By shedding light on these questions, the article seeks to clarify whether we can investigate public support for DI beyond the deceptively simple question about a ‘two-speed’ Europe, yet without the need to analyze each of the various forms of DI completely separately.

We present newly collected survey data from eight EU member states representing close to two-thirds of the EU's population and report substantive variation concerning the support of different types of DI at the level of individual respondents as well as between member states. Furthermore, our novel “Comparative Opinions on Differentiated Integration” (CODI) dataset allows us to explore which aspects citizens take into account when evaluating DI. We do so in a series of preregistered empirical analyses. To dissect which latent dimensions underlie citizens’ evaluations of DI, we employ exploratory factor analysis. This reveals two latent dimensions of preference formation regarding DI. We interpret these two dimensions as relating to: (a) EU integration, and (b) national autonomy. In other words, when asked to evaluate different types of DI, citizens seem to consider how such instruments could affect both EU integration and national autonomy. Across the eight member states in our sample, we find a clear majority for DI models and procedures that can be related to the first dimension (a “core” Europe as well as a requirement for consent for member state opt-outs). In contrast, there is much more cross-national heterogeneity when it comes to models of DI associated with the second, national autonomy dimension, in particular the “à la carte” model. Furthermore, depending on citizens’ dispositions and national context, they diverge in their assessments of the two dimensions of DI. Most prominently, pro-EU citizens positively evaluate integrationist forms of DI but are more skeptical when it comes to using DI to preserve national autonomy. Thus, assessments of DI are related to more general questions about European integration (cf. De Blok and De Vries, 2022). In sum, we argue that the structure of public opinion reflects the fact that, depending on its specific form, DI may be perceived either as a driver or as a stumbling block toward more European integration. The article concludes with a discussion of the findings and an outlook on future research. In particular, we argue that more research into public support for DI is indispensable for evidence-based constitutional engineering in the EU.

Theorizing public support for the varieties of DI

Research on DI, defined as an incongruence of “the territorial extension of EU membership and EU rule validity” (Holzinger and Schimmelfennig, 2012: 292), has come a long way since the early debates on the topic in the 1970s. However, as various reviews of the extant literature (e.g., Gänzle et al., 2020; Schimmelfennig et al., 2023) make clear, there is still a scarcity of studies relating to public opinion on DI. This is troubling, as–especially in times of politicization–the legitimacy of European integration hinges on public support (Hobolt and Vries, 2016), and this should likewise apply to DI.

An obvious challenge to the study of public support for DI consists in the fact that “normal” citizens cannot be expected to have developed strong and stable preferences on the full complexity of possible DI models and procedures. Yet at the same time, they can be expected to care about the efficiency and legitimacy of current governance structures (cf. Schimmelfennig et al., 2023). As DI plays an important role in EU governance (Duttle et al., 2017; Leuffen et al., 2022a; Schimmelfennig and Winzen, 2020), it may also, in principle, be of interest to citizens; all the more, as it raises some important normative questions relating to (member state) equality, dominance, and autonomy (Bellamy et al., 2022; Eriksen, 2018; Lord, 2015).¹

In the following, we derive three hypotheses on who supports general notions of DI and then argue that we need to pay closer attention to public support for specific “grand” DI models. Finally, we describe additional forms of DI and how public preferences are structured around them.

Contextualizing and replicating the prior literature

Existing studies have largely focused on only one specific model of DI–the idea of a ‘two-speed’ Europe–for which support was measured in Eurobarometer surveys. Using this item, the literature has established that individual dispositions both concerning EU integration and social (in)equality may inform citizens’ attitudes about this particular model of DI. Leuffen et al. (2022b) argue that preferences on DI align with citizens’ dispositions toward societal differentiation. DI is favored by citizens holding a liberal economic ideology, while more equality-oriented citizens tend to oppose DI. Moreover, they reveal large regional differences in support of DI. Citizens in Southern European member states tend to oppose, while Northern and Eastern European citizens rather support a ‘two-speed’ EU. The authors attribute this pattern to the repercussions of the Eurozone crisis and hypothesize that Southern European citizens fear discrimination by more economically powerful states. De Blok and De Vries (2022) argue that more general attitudes toward European integration should also inform citizens’ support of DI. In particular, pro-EU citizens should appreciate types of DI that could solve stalemates and allow integration-friendly countries to move forward. However, given the limited measurement of DI preferences in these studies, it is unclear whether and how these findings generalize to more extensive measurements of DI preferences.

In fact, the academic literature distinguishes a variety of DI models. Most prominently, Stubb (1996) differentiates variation over time, space and issue; this distinction is taken up in Schimmelfennig and Winzen’s (2020: 17) triad of “multi-speed”, “multi-tier”, and “multi-menu”. The literature also discusses additional forms of DI, including “internal” and “external” DI (Leuffen et al., 2022a), “capacity” and “sovereignty” DI (Winzen, 2016), and “discriminatory” and “exemptive” DI (Schimmelfennig, 2014). The different forms of DI will be introduced in more detail below. At this point, it is important to stress that the academic debate acknowledges that these models vary in their impact on the unity of the EU and entail different consequences for the participating and non-participating states. Against this backdrop, it is important to learn whether academic distinctions and understandings of these different models of DI resonate with citizens. In other words, which citizens are in support of which kind of DI, and why?

As outlined above, existing research has pointed out alignments of both attitudes on European integration and of social dispositions with support for a ‘two-speed’ model of DI. In this article, we explore whether such relationships apply to DI in general, as postulated by Leuffen et al. (2022b), or whether they are limited to specific types of DI. In our pre-analysis plan, we, therefore, borrowed three hypotheses from the literature applied to DI at the general, still unspecified level.

The first hypothesis relates support for DI to attitudes toward European integration (De Blok and De Vries, 2022). Pro-EU citizens might support DI because they consider DI to be an instrument to overcome heterogeneity-induced gridlock. These citizens should prefer further integration by a subgroup of member states to an integration stalemate at the level of the status quo.

H1: Support for European integration correlates positively with the support of DI.²

Secondly, Leuffen et al. (2022b) link support for DI to support of liberal economic values, arguing that liberals could appreciate the efficiency-enhancing feature of differentiation. This is in contrast to more equality-oriented citizens who might consider DI a threat to the EU's unity and possibly solidarity.

H2: Support of liberal economic values correlates positively with support of DI.

Beyond individual-level attitudes, country-level factors may influence citizens’ support of DI. Citizens may not only consider the impact of DI on the EU as a whole but recognize that DI affects different member states in different ways. Indeed, Leuffen et al. (2022b) present strong regional variation in support for the ‘two-speed’ model. Citizens of Southern European member states displayed resentment about this concept, arguably because they fear the risk of being discriminated against. Leuffen et al. (2022b) link this aversion to DI to sociotropic concerns and a fear of being left behind, prevalent in countries heavily affected by the Eurozone crisis.

H3: Citizens in Southern European member states support DI less than citizens in other regions of the EU.

Public support for different models of DI

The previous hypotheses were related to DI in general. However, different types of DI follow different logics and may have different consequences. Citizens could therefore evaluate them differently. For instance, the “multi-speed” model, defined by Stubb (1996: 285) as a “[m]ode of differentiated integration according to which the pursuit of common objectives is driven by a core group of Member States which are both able and willing to go further, the underlying assumption being that the others will follow later”, suggests a possibility and desirability of re-establishing EU unity after a temporary period of differentiation. The “multi-speed” model thus makes it possible to overcome gridlocks without abandoning the goal of an “ever closer union”. Accordingly, the “multi-speed” model should enjoy greater support from integration-friendly as compared to Eurosceptic citizens, who may prefer permanent DI.

H4: Citizens who are more supportive of European integration support a “multi-speed” model of DI more than citizens who are less supportive of integration.³

The same pro-EU-minded citizens, in contrast, could oppose forms of DI which undermine the ties between the EU’s member states. Stubb (1996: 285), for instance, links a “variable geometry” conception to a “permanent or irreversible separation between a hardcore and lesser developed integrative units”. Thus, “multi-tier” models like “variable geometry” or “core” Europe may establish permanent forms of differentiation.⁴ An “à la carte” model is usually considered to be the most radical “multi-menu” form of DI granting member states strong freedoms to “pick-and-choose” from EU policies. Therefore, citizens in support of EU integration may be opposed to such forms of DI.

H5: Citizens who are more supportive of European integration support an “à la carte” model of DI less than citizens who are less supportive of integration.

Furthermore, picking up on the argument by Leuffen et al. (2022b), citizens with an economically liberal orientation can be expected to embrace flexibility. If choices are one's own responsibility, any kind of flexibility should be embraced by liberals as autonomy-enhancing. Dahrendorf’s (1979: 20) classic defense of a Europe “à la carte” makes this position very clear, when he argues that he has “often been struck by the prevailing view in Community circles that the worst that can happen is any movement towards what is called an Europe à la carte. This is […] odd for someone who likes to make his own choices […]”.

H6: Citizens who display liberal economic dispositions support an “à la carte model” of DI more than citizens who have stronger preferences for equality.

Additional forms of DI

So far, we have focused on DI in general as well as grand models of DI such as “à la carte”. However, the literature has identified other forms of DI, which also warrant investigation. For instance, Leuffen et al. (2022a) distinguish “internal” and “external” differentiation. The former category applies to intra-EU differentiation–the models discussed above fall in this category–whereas the latter type refers to a situation in which outside states participate in EU policies. The European Economic Area (EEA) is a paradigmatic case of external differentiation, granting full access to the Single Market to Norway, Iceland, and Liechtenstein (for other examples, see Lavenex, 2015 and Leuffen et al., 2022a). We do not possess strong theories concerning public support of external differentiation and refrain from formulating hypotheses on this topic. However, as it is one prominent subcategory of DI, it seems worthwhile to include it in our analysis.

Furthermore, both internal differentiation among EU member states and external differentiation require procedures and conditions that balance the interests of all states. These may entail, for instance, consent requirements for opt-outs or financial contributions by non-members. Assuming a general aversion to free-riding–people are generally assumed to care about fair and non-discriminatory procedures, also with respect to DI (cf. Leuffen, 2022)–we expect citizens to be more critical of unconditional forms of DI.

Finally, Schimmelfennig (2014) and Schimmelfennig and Winzen (2020) distinguish “exemptive” and “discriminatory” differentiation. While the former category refers to measures that are designed to protect member states’ interests or national sensitivities, the latter “pertains to the temporary or conditional exclusion of member states from participation in EU policies” (Schimmelfennig et al., 2023). “Discriminatory differentiation” may raise legitimacy concerns, due to possible violations of the principle of political equality (Leuffen, 2022). Discrimination is particularly problematic if it occurs based on arbitrary reasons; in the empirical analysis, we operationalize such arbitrary discrimination by being excluded based on national wealth. We expect strong opposition against such a measure. At the same time, there may arguably be legitimate reasons for excluding member states from specific policies; for instance, in the case of member states violating fundamental EU norms.

Why do citizens support DI?

The previous discussion suggests that DI may raise concerns for the EU as a whole, for those member states that wish to engage in more intense forms of integration as well as for member states that desire to opt out of some areas of integration. In fact, the normative literature on DI argues that DI needs to be fair for those member states who want to integrate further as well as for those member states which decide not to take part in an integration step (Bellamy et al., 2022; Leuffen et al., 2022a; Lord, 2015). In other words, DI needs to safeguard both the integration project and the autonomy of individual states.

We argue that these two dimensions should also influence citizens’ preference formation regarding DI. This raises the empirical question of how citizens weigh these dimensions when judging various models of DI. Are citizens’ attitudes toward these dimensions highly correlated, reflecting, for example, their general attitudes toward European integration? We explore this question, and the structure of the DI political space, through factor analysis in the empirical section of this article.

To sum up, we contend that citizens can be expected to evaluate different forms of DI differently; in particular, citizens are very likely to consider the background conditions and possible effects of DI, not least for their member states. This has important implications for the reform debate of the EU–without understanding citizens’ attitudes on such issues, it is hard to formulate evidence-based policy recommendations.

The CODI dataset

In this section, we describe our original survey data and introduce the measures we designed to investigate support for (different forms of) DI. Our dataset is based on an original online survey fielded in February and March 2021 in eight EU member states: Denmark, France, Germany, Greece, Ireland, Italy, Netherlands, and Poland. The sample includes states that differ by size and by geographical location (East, North, and South); together, the selected countries represent close to two-thirds of the EU population. There are around 1500 respondents in each country yielding a total maximal sample size of around 12,000. We used quota sampling with respect to national marginal distributions of age groups, gender, and sub-national regions. Furthermore, for some of our analyses, we employ weights based on population shares in the EU.

Table 1 lists the battery of survey questions designed to operationalize the key DI concepts and mechanisms outlined above. Our operationalizations aim to cover most of the relevant DI types and procedures, while at the same time keeping formulations simple enough to ensure that respondents can understand them. We developed each item in a larger research group and pre-tested them to ensure that there were no excessive “don't know” responses or obvious inconsistencies in response behavior across the items. Support for each statement is measured on a 1 to 5 Likert scale; an exception is the binary ‘two-speed’ Europe item, which we borrowed from the Eurobarometer.

Table 1.

Operationalization of DI concepts.

DI grand models
Core Europe	“The idea of building a core Europe, bundling the most integration-friendly states, is a good idea.”
À la carte	“Member states should generally be allowed to pick and choose from EU policies as they desire.”
Two-speed Europe (Eurobarometer)	“Countries which are ready to intensify the development of a common European policy in certain important areas (i) should do so without having to wait for the others or (ii) should wait until all Member States of the EU are ready for this.”
Opt-out and exclusion procedures
Consent for opt-outs	“Member states should be allowed to opt-out of specific policy areas only after receiving the consent of the other member states.”
No exclusion based on wealth	“No state should be excluded from common EU policies because it is less wealthy than most other EU member states.”
No exclusion based on breaching norms	“No state should be excluded from common EU policies even if it breaches some of the core norms and values of the EU.”
External differentiation
No external DI	“Non-EU states should generally be excluded from the participation in EU policies.”
Conditional external DI	“Non-EU states should be allowed to participate in selected EU policies if they adequately contribute financially to the EU.”

Preferences for a “multi-tier” (“core” Europe) conception are measured by support for the item: “The idea of building a core Europe, bundling the most integration-friendly states, is a good idea”. The “multi-menu” (“à la carte”) conception is captured by the statement: “Member states should generally be allowed to pick and choose from EU policies as they desire”. We have decided not to design a new item for the “multi-speed” model, but to instead replicate the binary Eurobarometer item: “As regards the idea of a ‘Two Speed Europe,’ which of the following comes closest to your personal preference? Those countries which are ready to intensify the development of a common European policy in certain important areas… (i) should do so without having to wait for the others or (ii) should wait until all Member States of the EU are ready for this”. This allows us to not only replicate the findings of previous studies working with the Eurobarometer item but also to validate the item in comparison with our own measures (see the Online appendix).

In addition to the grand models of DI, we inquired about respondents’ support of procedural rules concerning the granting of opt-outs, as well as the exclusion of member states based either on their wealth or their (dis)respect for fundamental EU norms. We asked the respondents to react to the statement that “Member states should be allowed to opt-out of specific policy areas only after receiving the consent of the other member states”. This item addresses support for institutional provisions designed to prevent a “free-riding” behavior of opt-out states (cf. Bellamy, 2019: 202). In other words, how do citizens trade off individual states’ sovereignty aspirations and community interests in the case of exemptive differentiation? We operationalize discriminatory differentiation with the statement: “No state should be excluded from common EU policies because it is less wealthy than most other EU member states”. Moreover, we investigate whether discriminatory differentiation should be used as a tool to sanction norm-violating states by asking about support for the statement: “No state should be excluded from common EU policies even if it breaches some of the core norms and values of the EU”. This item relates to the current debate on how the EU should respond to processes of democratic backsliding in member states (Bellamy and Kröger, 2021; Blauberger and Kelemen, 2017).

Finally, two items address external differentiation asking whether non-EU countries should be generally excluded from EU policies (“Non-EU states should generally be excluded from the participation in EU policies”) or whether they may participate in selected policies if they contribute financially to the EU budget (“Non-EU states should be allowed to participate in selected EU policies if they adequately contribute financially to the EU”). The latter item reflects the arrangements realized in the EEA.

Support for DI across Europe: Descriptive results

Having introduced the basic features of our survey data, we now descriptively explore support for the different DI items. Figure 1 shows support for our seven original items and the Eurobarometer's ‘two-speed’ Europe item across all countries. Here, we simplify the answer scale and collapse respondents who (strongly) agree or disagree, respectively, into one category; note that the ‘two-speed’ item is binary by design. It becomes clear that public preferences for all these types of DI are relatively pronounced: The modal response is always either (strong) agreement or disagreement. Furthermore, agreement or disagreement always approaches or exceeds the 50% mark, even though respondents had a “neither nor” option. This suggests that large portions of the population do not have difficulties forming an opinion on DI as presented by these items. This does not preclude that citizens’ interpretations of these concepts may diverge from expert assessments. At the same time, we also see that there is just a slight majority for the binary ‘two-speed’ item (perhaps exactly due to the lack of a middle category).

Figure 1.

EU-wide support for eight DI types.

Overall, we find public support for both an integration-friendly “core” Europe and an “à la carte” Europe. Moreover, a majority of respondents support the idea that opt-outs should only be allowed if the other member states consent. About 45% of our respondents support an “à la carte” conception of DI while about 27% signal opposition. The quite pronounced majority in favor of this conception surprised us, given that it may be considered particularly radical (Dahrendorf, 1979), and has been criticized with reference to democratic theory (Schimmelfennig and Winzen, 2020: 19). There is an extremely strong opposition to excluding member states for arbitrary reasons, namely for their level of wealth. On the other hand, a relative majority of respondents disagree that member states that breach fundamental norms should not be excluded. That is, they seem to be open to the possibility that member states may be excluded from policies in such cases. Finally, while a relative majority of respondents support generally excluding non-EU states from common policies, there also is clear support for allowing participation of non-EU states if these contribute financially.

In sum, there are pronounced majorities for DI procedures that seem to consistently follow a particular pattern: strict requirements for opt-outs, possible sanctions for norm breachers, participation of non-EU states only if they contribute, but no outright discrimination due to poor national finances. However, the relatively high support for both a “core” and an “à la carte” Europe is puzzling. The next section uses factor analysis to understand this puzzle. It suggests that there are two independent profiles behind these relative majorities supporting two different concepts, which we further explore in the subsequent sections.

Along which dimensions is support for DI structured?

How can there be strong support for both a “core” and an “à la carte” Europe? And how does support for these “grand models” relate to opinions on other more specific DI features? To answer these questions, we present the results from factor analysis. Exploratory factor analysis is a well-established tool to uncover a structure that parsimoniously summarizes empirical relationships between measured variables. In research on general EU attitudes, Boomgaarden et al. (2011) use this approach to suggest that there are as many as five different dimensions of opinion on the EU. In general, factor analysis assumes “that the observed (measured) variables are linear combinations of some underlying source variables” (Kim and Mueller, 1978: 8). It is important to emphasize, however, that “factor analysis does not tell the researcher what substantive labels or meaning to attach to the factors. This decision must be made by the researcher” (Kim and Mueller, 1978: 56).

We had preregistered to run “principal axes” factor analyses using an “oblimin” rotation assuming one and two factors, respectively. This decision was based on the conjecture that two dimensions, at most, should suffice to adequately summarize the structure of public DI attitude and that a three-factor model would not be stable, given that we work with only seven observed variables. Indeed, estimating a three-factor model leads to numerical problems that indicate exactly this problem. Nonetheless, the results, which we report in the Online appendix, are very similar to what we discuss in the following. In the Online appendix, we also closely replicate our results when including the ‘two-speed’ Europe item directly in the factor analysis, and we provide a series of preregistered tests that cast doubt on the usefulness of the ‘two-speed’ item more generally. We return to this topic in the conclusion and for now consider our seven original DI items only.

Table 2 shows the results from our factor analysis. We display estimated factor loadings (correlations). Clearly, the estimated factor loadings for the first factor are very similar in both the one-factor and the two-factor model. Therefore, we base our discussion on the two-factor model. We focus on items that have a loading of at least 0.3. This choice was not preregistered but reflects a rule of thumb formulated in the methodological literature (Hair et al., 2019: 151).

Table 2.

Results from a factor analysis on seven measures of support for DI.

	One-factor model	Two-factor model
	Factor 1	Factor 1	Factor 2
Core Europe	0.48	0.47	0.04
A la carte Europe	−0.13	−0.19	0.34
Consent for opt-outs	0.47	0.48	−0.06
No exclusion of less wealthy MS	0.32	0.32	0.14
No exclusion of norm-breaching MS	0.06	0.02	0.66
Exclude non-EU states from common policies	−0.05	−0.07	0.14
Non-EU states can participate if they contribute	0.36	0.36	0.01
Observations	12,092	12,092
Corr. b/w factors	NA	0.06
Eigenvalues	1.54	1.54 1.29
Variance explained	0.10	0.19

Note: Bold entries are factor loadings with an absolute size of at least 0.3.

In the two-factor model, the first factor correlates strongly with: (a) support for a “core” Europe of integration-friendly states, (b) support for the requirement that other member states consent to opt-outs, (c) opposition to excluding member states from common policies because they are poor, and (d) support for allowing non-EU states to participate in common policies if they contribute financially.

In sum, the first factor seems to capture divides about European integration. Respondents seem to care about whether DI strengthens rather than undermines European integration. Specifically, citizens either support or oppose the idea of an integration-friendly “core” Europe, limiting opt-outs and exclusions due to secondary criteria such as wealth, as well as the conditional participation of non-EU states. Perhaps surprisingly, opinions toward the exclusion of norm-breaking member states do not correlate strongly with this first dimension.

However, it is this item–opposition to the exclusion of norm-breaching member states–that correlates most strongly and positively (r = 0.66) with the second factor. Support for an “à la carte” Europe also correlates positively with this second factor. Therefore, we interpret this second factor as capturing opinions toward safeguarding national autonomy or respecting member state diversity. Interestingly, safeguarding national autonomy seems to constitute a second source variable and not the “negative pole” (Goertz, 2020: 81) of support for DI. Apparently, some citizens may very well support a strong “core” Europe (scoring high on the first factor), but at the same time be concerned about being “dominated” (Bellamy et al., 2022; Fossum, 2019; Leuffen, 2022) by other EU member states or “Brussels” (scoring high on the second factor), for example by being excluded from common EU policies. Support for an “à la carte” model of “pick-and-choose” naturally corresponds to the concern of being dominated, as it would allow member states to pro-actively avoid EU policies with which they are uncomfortable.

The overall correlation between the two factors is very small (r = 0.06). Again, this seems surprising, as one would expect citizens with high scores on the first factor (support for integration-friendly DI) to have low scores on the second factor (support for safeguarding national autonomy). However, guided by a series of preregistered hypotheses and regression analyses, we later provide evidence that EU membership support does indeed meaningfully explain positions on both dimensions simultaneously.

The results of the factor analysis are based on unweighted survey data, which are representative with respect to marginal national gender, age, and regional distributions. However, the analysis does not account for the varying population sizes of the sampled countries. In the Online appendix, we present results with observations weighted by the population shares of their countries. Overall, the results are very similar.

Finally, in further robustness analyses, we looked into potential cross-national heterogeneity in the factor analysis. Separate factor models for each country suffer from numerical fittings problems but confirm the structure of the first DI dimension. However, loadings are more variable for the second dimension. The Online appendix describes the results of “multigroup” factor models that similarly point toward some heterogeneity with respect to the second dimension, which we pick up in the discussion.

Support for important DI dimensions across countries

The results in the previous section were motivated by the fact that both a “core” and an “à la carte” Europe enjoy relatively high support among citizens. We established that two different DI support profiles or “packages” lurk behind this pattern. But this does not tell us how support for these two dimensions is distributed across countries. Therefore, guided by the results from our factor analysis, we here concentrate on the cross-national variance in support of DI. In the next section, we then consider individual-level regressions to show that on the individual level, support of EU membership consistently explains support on both DI dimensions.

The factor analysis suggested four aspects of DI to be most relevant for discriminating between supporters and opponents of DI: support for a “core” Europe and for consent for opt-outs correlate the strongest with the first factor, while support for not excluding norm breaching countries and support for an “à la carte” Europe correlate the strongest with the second factor.

In order to investigate more closely these four aspects, we create a net support measure. For each country, we subtract the proportion of respondents who (strongly) disagree with a DI rule from those who (strongly) agree. Whenever we report overall net support for the whole EU, this is based on applying population weights.

Figure 2 plots net support for the selected DI items across the countries in our sample. With respect to the first dimension of DI, the picture is remarkably clear. Overall, net support for a “core” Europe is large (25 percentage points) and it is positive in every country. Support for consent for opt-outs is even larger (29 percentage points). This shows that there is broad net support regarding the first, integration-friendly dimension of DI. This is consistent with the results from the factor analysis, which, however, did not tell us anything about absolute support levels.

Figure 2.

Net support for central items across countries.

Regarding the second dimension of DI support, there is strong net support for excluding countries from common policies when they violate fundamental EU norms (22 percentage points more respondents disagree than agree that this should not happen). There is, however, heterogeneity across countries. While large majorities in Denmark, Germany, France, and the Netherlands agree that countries could be excluded on these grounds, respondents in Greece, Ireland, Italy, and Poland are divided. It is possible that among the latter, respondents fear being excluded due to either breaching fiscal rules (Greece, Italy) or due to breaching democratic norms (Poland). While we have no specific data on these two lines of reasoning, the mean opposition to excluding countries because of low wealth is indeed also very high in these four countries (between 65% and 80%). At the same time, the four countries in which majorities support the exclusion of norm-breakers are all net contributors to the EU budget. Again, both lines of reasoning–the breaching of democratic norms or fiscal rules–may underlay these responses. In particular, citizens from net-contributing member states may find it difficult to accept that countries breaking democratic norms benefit from their financial contributions through the redistributive schemes of the EU budget. Possibly, those citizens may see financial sanctions as a legitimate instrument against norm-breakers (cf. Kelemen, 2020). However, our survey lacks items that directly address this interpretation.

At the same time, we also see relatively strong net support for the idea of an “à la carte” Europe (19 percentage points). But here again, there is strong variation across countries. While German and Irish respondents on net disagree with this idea, there is strong support for it in Greece, Italy, the Netherlands, and especially in Poland (with net support of just below 50 percentage points) as well as to a lesser degree in France. We suspect that country-specific policy or cultural considerations could explain these patterns, although it is not obvious how this particular pattern would emerge. On the side of “à la carte” opponents, it could be speculated that Germans have a greater normative-cultural preference toward a unified Europe and the Irish may resent the attempted “pick-and-choose” tactics of the United Kingdom before and after the Brexit vote. On the side of “à la carte” supporters, some Dutch respondents may prefer to abstain from further steps toward a fiscal union, whereas many conservative Poles might reject common approaches toward migration or minority rights. Taken together, this could explain why the factor models were more variable across countries with respect to the second DI dimension.

Most importantly, this analysis shows that there is no clear majority of countries that support the second empirical dimension of DI support: an “à la carte” Europe enjoys relatively high support while including norm-breaching member states does not. The first empirical DI dimension, in contrast, has clear overall net support in terms of its most important items. This again underscores that the factor analysis can only tell us which items can be effectively used to tell DI supporters and opponents apart concerning the two dimensions but does not inform us about absolute levels of support. Substantively, this shows that there is a surprisingly large majority for an integration-friendly type of DI, while a version of DI concentrated primarily on preserving national autonomy does not enjoy overall support.

Who supports which kind of DI?

We now test our original hypotheses about individual-level correlates of DI support and explore how the two empirical DI dimensions relate to other important variables. We had hypothesized that general EU support is associated with higher DI support, and especially with a “multi-speed” model; that economic liberals, too, are more likely to support DI, especially an “à la carte” Europe; and that Southern Europeans are less likely to support DI. Our results show that a simple one-dimensional conception of DI is limited and that it is worthwhile to distinguish different types of DI. Nevertheless, we also see that general EU support emerges as a central unifying correlate.

Table 3 shows results from linear regressions of support for various DI measures on these explanatory variables and a set of control variables.⁵ We use heteroscedasticity-consistent standard errors. Models 2, 3, and 4 were preregistered. We additionally explore support for an “à la carte” Europe and the second empirical DI factor.⁶ In the Online appendix, we also present a series of regressions that vary the set of control variables. These, too, were preregistered (except those for the second DI factor). With some minor exceptions mentioned below, the results of these regressions are the same as in Table 3.

Table 3.

Correlates of various measures of DI support.

	Dependent variable
	Support for core Europe	Support for À la Carte Europe	Support for two-speed Europe	DI factor 1	DI factor 2
	(1)	(2)	(3)	(4)	(5)
EU membership support	0.362***	−0.303***	0.041**	0.311***	−0.094***
	(0.031)	(0.033)	(0.016)	(0.020)	(0.023)
Liberalism	0.041***	0.057***	−0.015***	0.015**	0.051***
	(0.011)	(0.011)	(0.005)	(0.007)	(0.008)
EU south	0.187***	−0.023	−0.120***	0.090***	0.137***
	(0.025)	(0.027)	(0.013)	(0.017)	(0.019)
Trust in EU	0.389***	−0.191***	0.035**	0.283***	−0.016
	(0.033)	(0.034)	(0.016)	(0.020)	(0.023)
Left-right	−0.014**	0.045***	0.006**	−0.018***	0.027***
	(0.006)	(0.006)	(0.003)	(0.004)	(0.004)
Migration support	0.045***	−0.020***	0.004	0.048***	0.008*
	(0.006)	(0.006)	(0.003)	(0.004)	(0.004)
Growth over environment	0.004	0.033***	−0.0002	−0.011***	0.026***
	(0.005)	(0.005)	(0.002)	(0.003)	(0.004)
DI knowledge	−0.089***	−0.041**	0.002	−0.061***	−0.059***
	(0.017)	(0.018)	(0.009)	(0.012)	(0.013)
National identity	−0.246***	0.145***	−0.033*	−0.125***	0.031
	(0.034)	(0.037)	(0.017)	(0.021)	(0.025)
Male	0.052**	−0.067***	0.014	0.048***	0.042**
	(0.024)	(0.025)	(0.012)	(0.015)	(0.017)
Other gender	0.030	−0.061	−0.051	−0.354***	−0.069
	(0.162)	(0.342)	(0.150)	(0.126)	(0.193)
Age	0.0001	0.004***	−0.001**	−0.0003	0.003***
	(0.001)	(0.001)	(0.0004)	(0.001)	(0.001)
Migration background	0.016	0.035	−0.013	0.011	0.001
	(0.037)	(0.039)	(0.018)	(0.024)	(0.027)
University degree	0.078***	0.004	0.051***	0.029*	0.001
	(0.024)	(0.025)	(0.012)	(0.015)	(0.017)
Diff. paying bills: time to time	0.015	0.363***	−0.100***	−0.025	0.196***
	(0.041)	(0.043)	(0.020)	(0.027)	(0.030)
Diff. paying bills: almost never	−0.001	0.374***	−0.076***	−0.047*	0.196***
	(0.042)	(0.045)	(0.021)	(0.027)	(0.031)
Diff. paying bills: never	−0.056	0.316***	0.003	−0.077***	0.057**
	(0.038)	(0.040)	(0.017)	(0.024)	(0.026)
Constant	2.802*	−5.165***	2.261***	0.434	−5.761***
	(1.661)	(1.729)	(0.842)	(1.073)	(1.191)
Observations	7167	7164	6895	6856	6856
Adjusted R2	0.157	0.104	0.042	0.239	0.072

Note: *p < 0.1; **p < 0.05; ***p < 0.01. Entries are unstandardized coefficients from linear models with heteroscedasticity-consistent standard errors.

We start by discussing the role of general EU support (H1, H4, and H5). We find pronounced effects on support for a “multi-tier” core Europe conception and a somewhat weaker effect for a “multi-speed” DI model, confirming H4. In contrast, we find a strong negative effect for the “à la carte” model; EU supporters are less likely to support such an option as compared to Eurosceptic citizens, confirming H5. These findings are reflected in the results of models 4 and 5 (Table 3), which employ the DI factor scores as the dependent variables. The relationship is especially pronounced with the first, “inclusive core” Europe DI factor, where the coefficient translates to an approximately 0.5 standard deviations higher support. On the other hand, we find a negative effect for the second “national autonomy” DI factor indicating that EU supporters oppose DI models with a strong focus on this dimension. This corresponds with their opposition to the “à la carte” model. We can observe a similar pattern when looking at the preferences of respondents with an exclusively national identity, who oppose the first factor but react favorably to the “à la carte” option.

Taken together, these patterns bolster our interpretation of the first dimension being “integrationist” and the second dimension being “nationalist”. In sum, while there was little correlation between these two factors in the factor model, the regression analyses show that support for different DI models is empirically structured along general EU membership support. In our interpretation, citizens anticipate the effects of different DI models and evaluate these consequences in line with their more general preferences on European integration. Whereas Europhiles embrace the idea of strengthening a core Europe and limiting opt-outs, Eurosceptic respondents consider such a development a threat and instead prefer watering down European integration and preserving national autonomy, to the point of sacrificing fundamental EU norms. Finally, the results underline H4 and H5 in that EU supporters are skeptical toward a “multi-menu” (“a la carte”) but positive toward a “multi-speed” EU.

Concerning our other hypotheses on economic liberalism (H2 and H6) as well as regional differences (H3), the results are less clear-cut. Economic liberals appreciate a “core” Europe but show (slightly) less support for a ‘two-speed’ model; the latter finding stands in direct contrast to the results by Leuffen et al. (2022b). However, H6 is clearly confirmed: Economic liberals show considerably higher support for an “à la carte” Europe. Furthermore, exploratory regressions with the two different factor scores indicate that economic liberals are DI supporters. This is especially true for the second factor; the relationship with the first factor is less stable across models (see the Online appendix). In sum, the evidence for economic liberals points in the hypothesized direction. Nevertheless, the finding highlights once again that the ‘two-speed’ item is not an easily generalizable measure of more general DI preferences (see the Online appendix for further evidence).

Concerning differences between Northern/Eastern and Southern Europe, model 3 (Table 3) replicates the finding by Leuffen et al. (2022b): Southern Europeans (in our sample the French, Greeks, and Italians) are 12 percentage points less likely to support a ‘two-speed’ Europe. This is a very large difference that trumps any other variable's association with this outcome. On the other hand, Southern Europeans are marginally more likely to support DI as measured by our first DI factor (model 4, Table 3), but this relationship is not stable across models (see the Online appendix). Exploratory regressions suggest they also show higher support on both the “core” Europe item (model 1, Table 3) as well as the “national autonomy” factor (model 5, Table 3), although these relationships are weak. In addition, we find that Southern Europeans do not differ from other Europeans when it comes in support of an “à la carte” Europe (model 2, Table 3). Therefore, there is no clear evidence for H3, and it seems that regional differences vary greatly depending on which aspects or types of DI are considered.

In sum, in contrast to the strong and consistent results on general EU membership support, these tests suggest a weaker role for variables based on other political attitudes or regional differences. What remains are the strong regional differences on the ‘two-speed’ Europe item. Leuffen et al. (2022b) had suggested that Southern Europeans, suffering (pre-COVID) from the long shadow of the Eurozone crisis, possibly fear being “left behind” or discriminated against by DI. The Eurobarometer measure of DI might have accentuated that fear because it mentions “speed” and operationalizes opposition to DI as “waiting until all member states are ready”, possibly evoking ideas of a capacity-related DI. It is likely that the differential impact of the COVID-19 pandemic–hitting Italy earlier and harder–and the political conflict around the “Next Generation EU” recovery fund, which was (as in the Eurozone crisis) structured around a North-South cleavage, has stabilized or rekindled these fears. The alternative DI items in our survey contain no such framing, and the very small or zero differences between Northern/Eastern and Southern Europe concerning these measures are also consistent with this interpretation. Specifically, these results hint at Southern Europeans’ wish to belong to a “core” Europe and to not be discriminated against due to their economic situation. These questions highlight that researchers have to pay cautious attention to possible framing effects in the wording of survey questions if they are interested in (the causes of) cross-national differences in DI support.

Conclusion

This article empirically maps and investigates public support for different types of DI. The CODI survey contains a battery of innovative measures for different types of DI, and thereby allows us to provide, for the first time, insights into the dimensionality of public support for DI. Our starting point is the observation that the conceptual literature has long distinguished several diverging models or types of DI. However, our knowledge about public support of DI had previously been limited to just one Eurobarometer item, tapping into approval of a “two-speed” Europe model. Our results largely confirm the assessment that support for DI, indeed, depends on its specific design and that using a single survey measure is inadequate. DI preferences are multi-dimensional, support for DI varies across its different forms, and is explained by different individual-level characteristics.

A factor analysis suggests that public opinion on DI can be meaningfully summarized using two dimensions. We interpret the two dimensions to represent an “integrationist”, and a “national autonomy” related dimension of DI, respectively. Importantly, while there are relative majorities for central elements of both dimensions, we find that only the first package of DI measures enjoys strong support in every country within the sample, whereas the second exhibits stronger conflict lines. It should be noted that not all DI items load (strongly) on these factors; this is especially true for an item that inquires about the participation of non-EU countries in common policies. This indicates the need for future research to develop more rigorous and reliable instruments for measuring DI preferences. We suggest doing so based on the high-loading survey items and the two dimensions we have presented.

Our substantive interpretation of these findings is that the European public generally favors fair and rule-based differentiation. For instance, support for DI based on the notion of a “core” Europe is rooted in the idea that differentiation across member states via opt-outs should only occur based on consent. Moreover, the inclusion of third countries should be conditioned on their financial contributions. Thus, differentiation is, in principle, acceptable, as long as it is regulated fairly. When it comes to differentiation based on an “à la carte” Europe, we find less consistent support across countries; this model also bears a potential for conflict, given that more Eurosceptic parts of the public view it as a way to safeguard national autonomy. In fact, the support of this model is generally considered to be low among elites, arguably precisely for this reason.

Our results also indicate that a high degree of variance remains unexplained, especially between countries, when it comes to this second dimension of DI. The item on how to treat countries that breach fundamental norms points toward one potential explanation, as this has become a contentious topic, especially concerning Eastern European member states.

The measurement of DI raises the question of the reference category against which to compare DI. Do citizens compare DI against uniform integration–and if yes, at which levels of integration (the lowest common denominator or the positions defended by pro-European elites)–or against other forms of DI? Further research is needed to unpack these questions more systematically.

In political debates about constitutional reform and the future of the EU, DI is frequently proposed as a response to preference heterogeneity within the EU. However, while it may be an efficient solution to overcoming gridlock, it could lead to new problems, also from a public opinion perspective. Different citizens in different member states favor different forms of DI and thus new challenges to legitimacy may arise.

Supplemental Material

sj-pdf-1-eup-10.1177_14651165221127633 - Supplemental material for Mapping public support for the varieties of differentiated integration

Supplemental material, sj-pdf-1-eup-10.1177_14651165221127633 for Mapping public support for the varieties of differentiated integration by Julian Schuessler, Max Heermann, Dirk Leuffen, Lisanne De Blok and Catherine E De Vries in European Union Politics

Supplemental Material

sj-csv-2-eup-10.1177_14651165221127633 - Supplemental material for Mapping public support for the varieties of differentiated integration

Supplemental material, sj-csv-2-eup-10.1177_14651165221127633 for Mapping public support for the varieties of differentiated integration by Julian Schuessler, Max Heermann, Dirk Leuffen, Lisanne De Blok and Catherine E De Vries in European Union Politics

Supplemental Material

sj-r-3-eup-10.1177_14651165221127633 - Supplemental material for Mapping public support for the varieties of differentiated integration

Supplemental material, sj-r-3-eup-10.1177_14651165221127633 for Mapping public support for the varieties of differentiated integration by Julian Schuessler, Max Heermann, Dirk Leuffen, Lisanne De Blok and Catherine E De Vries in European Union Politics

Supplemental Material

sj-r-4-eup-10.1177_14651165221127633 - Supplemental material for Mapping public support for the varieties of differentiated integration

Supplemental material, sj-r-4-eup-10.1177_14651165221127633 for Mapping public support for the varieties of differentiated integration by Julian Schuessler, Max Heermann, Dirk Leuffen, Lisanne De Blok and Catherine E De Vries in European Union Politics

Supplemental Material

sj-r-5-eup-10.1177_14651165221127633 - Supplemental material for Mapping public support for the varieties of differentiated integration

Supplemental material, sj-r-5-eup-10.1177_14651165221127633 for Mapping public support for the varieties of differentiated integration by Julian Schuessler, Max Heermann, Dirk Leuffen, Lisanne De Blok and Catherine E De Vries in European Union Politics

Supplemental Material

sj-r-6-eup-10.1177_14651165221127633 - Supplemental material for Mapping public support for the varieties of differentiated integration

Supplemental material, sj-r-6-eup-10.1177_14651165221127633 for Mapping public support for the varieties of differentiated integration by Julian Schuessler, Max Heermann, Dirk Leuffen, Lisanne De Blok and Catherine E De Vries in European Union Politics

Supplemental Material

sj-r-7-eup-10.1177_14651165221127633 - Supplemental material for Mapping public support for the varieties of differentiated integration

Supplemental material, sj-r-7-eup-10.1177_14651165221127633 for Mapping public support for the varieties of differentiated integration by Julian Schuessler, Max Heermann, Dirk Leuffen, Lisanne De Blok and Catherine E De Vries in European Union Politics

Supplemental Material

sj-r-8-eup-10.1177_14651165221127633 - Supplemental material for Mapping public support for the varieties of differentiated integration

Supplemental material, sj-r-8-eup-10.1177_14651165221127633 for Mapping public support for the varieties of differentiated integration by Julian Schuessler, Max Heermann, Dirk Leuffen, Lisanne De Blok and Catherine E De Vries in European Union Politics

Footnotes

Acknowledgment

We thank two anonymous reviewers and the editor for their helpful comments. Moreover, we are grateful to Simone Cremaschi, John Erik Fossum, Magdalena Góra, Sandra Kröger, Frank Schimmelfennig, Mark Thatcher, Resul Umit, as well as participants at the ECPR SGEU conference in 2021, the ‘DI Garage' of the ECPR Research Network on Differentiated Integration, and various EU3D workshops, for their valuable input. Pascal Mounchid provided excellent research assistance. Work on this article was generously supported by the European Union’s Horizon 2020 Research and Innovation Programme under Grant Agreement 822419 and Grant Agreement 822304 as well as the Danish National Research Foundation, grant number: DNRF144.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article. This work was supported by the Horizon 2020 Framework Programme (grant numbers 822304 and 822419).

ORCID iDs

Julian Schuessler

Max Heermann

Lisanne De Blok

Catherine E De Vries

Supplemental material

Supplemental material for this article is available online.

Notes

References

Bellamy

(2019) A Republican Europe of States. Cosmopolitanism, Intergovernmentalism and Democracy in the EU. Cambridge: Cambridge University Press.

Bellamy

Kröger

(2021) Countering democratic backsliding by EU member states: Constitutional pluralism and ‘value’ differentiated integration. Swiss Political Science Review 27(3): 619–636.

Bellamy

Kröger

Lorimer

(2022) Flexible Europe: Differentiated Integration, Fairness, and Democracy. Bristol: Bristol University Press.

Blauberger

Kelemen

(2017) Can courts rescue national democracy? Judicial safeguards against democratic backsliding in the EU. Journal of European Public Policy 24(3): 321–336.

Boomgaarden

Schuck

Elenbaas

, et al. (2011) Mapping EU attitudes: Conceptual and empirical dimensions of euroscepticism and EU support. European Union Politics 12(2): 241–266.

Dahrendorf

(1979) In: Lecture

TJM

(ed) A Third Europe? Florence: European University Institute, 4–24.

De Blok

De Vries

(2022) A blessing and a curse? Examining public preferences for differentiated integration. European Union Politics 24(1): 143–163.

Duttle

Holzinger

Malang

, et al. (2017) Opting out from European Union legislation: The differentiation of secondary law. Journal of European Public Policy 24(3): 406–428.

Eriksen

(2018) Political differentiation and the problem of dominance: Segmentation and hegemony. European Journal of Political Research 57(4): 989–1008.

10.

Fossum

(2019) Europe’s Triangular Challenge: Differentiation, Dominance and Democracy. EU3D Research Papers.

11.

Gänzle

Leruth

Trondal

(2020) Differentiated Integration and Disintegration in a Post-Brexit Era. London: Routledge.

12.

Goertz

(2020) Social Science Concepts and Measurement. New and Completely Revised Edition. Princeton: Princeton University Press, 37–51.

13.

Hair

Black

Babin

, et al. (2019) Multivariate Data Analysis. Boston: Cengage.

14.

Hobolt

Vries

(2016) Public support for European integration. Annual Review of Political Science 19(1): 413–432.

15.

Holzinger

Schimmelfennig

(2012) Differentiated integration in the European Union: Many concepts, sparse theory, few data. Journal of European Public Policy 19(2): 292–305.

16.

Kelemen

(2020) The European Union’s authoritarian equilibrium. Journal of European Public Policy 27(3): 481–499.

17.

Kim

J-O

Mueller

(1978) Introduction to Factor Analysis. What It Is and How To Do It. Newbury Park: Sage.

18.

Lavenex

(2015) The external face of differentiated integration: Third country participation in EU sectoral bodies. Journal of European Public Policy 22(6): 836–853.

19.

Leuffen

(2022) Member State Equality and Procedural Fairness in Differentiated Integration: A Republican Perspective. In: Fromage

(ed) (Re-) defining Membership: Differentiation in and Outside the European Union. Oxford: Oxford University Press, 37–51.

20.

Leuffen

Rittberger

Schimmelfennig

(2022a) Integration and Differentiation in the European Union: Theory and Policies. Cham: Palgrave Macmillan.

21.

Leuffen

Schuessler

Gómez Díaz

(2022b) Public support for differentiated integration: Individual liberal values and concerns about member state discrimination. Journal of European Public Policy 29(2): 218–237.

22.

Lord

(2015) Utopia or dystopia? Towards a normative analysis of differentiated integration. Journal of European Public Policy 22(6): 783–798.

23.

Schimmelfennig

(2014) EU enlargement and differentiated integration: Discrimination or equal treatment? Journal of European Public Policy 21(5): 681–698.

24.

Schimmelfennig

Leuffen

De Vries

(2023) Differentiated integration in a multi-level European Union: Public preferences and institutional consequences. European Union Politics 24(1): 3–20.

25.

Schimmelfennig

Winzen

(2020) Ever Looser Union? Differentiated European Integration. Oxford: Oxford University Press.

26.

Schraff

(2022) Asymmetric ratification standards and popular perceptions of legitimacy. Journal of European Public Policy 29(3): 405–426.

27.

Stubb

(1996) A categorization of differentiated integration. JCMS: Journal of Common Market Studies 34(2): 283–295.

28.

Winzen

(2016) From capacity to sovereignty: Legislative politics and differentiated integration in the European Union. European Journal of Political Research 55(1): 100–119.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.01 MB

0.00 MB

0.02 MB

2.20 MB

0.31 MB