Introduction

Recently, the European Union (EU) has been facing a number of challenges that have called into question its very existence, including issues related to its economic vision and borders protection, as well as the ongoing COVID-19 pandemic and Ukrainian war.

Despite the EU’s successes in promoting political stability and involving countries on the periphery, these crises have highlighted an underlying institutional fragility. As a result, there has been a decline in institutional legitimacy and an increase in Euroscepticism, as evidenced by the decline in citizens’ trust in EU institutions (Berg 2019; Eurofound 2022; Roth et al. 2022).

In this context, it is important to note that while the UK is an outlier when it comes to exit scepticism (Hobolt 2016), there are different reasons why people in different countries way wish to leave the EU (De Vries 2018; Ejrnæs and Jensen 2019, 2021). The UK’s 2016 vote to exit the EU was the first real example shedding light on the consequences of the intensifying Eurosceptic sentiment (Vasilopoulou 2016), and since then, concerns of a potential ‘contagion’ in other countries (Walter 2020, 2021) have persistently spread to the point of worrying Brussels seriously.

Several scholars have paid attention to the Eurosceptic political parties which have obtained strong support in national and European elections, becoming important political actors against EU integration policy (Szczerbiak and Taggart 2008; Taggart and Szczerbiak 2004). Additionally, recent studies have highlighted the important role played by non-party groups and (un)civil society in Euroscepticism (Ruzza 2009). Furthermore, although Eurosceptic parties have contributed to the Eurosceptic framing, the public perception of these matters is crucial in understanding the current state of Euroscepticism within the EU (De Vries 2018; Williams and Spoon 2015).

In this new scenario, the economic, immigration, COVID-19 crises and Ukrainian war have served as pressure tests for EU–member state relationships, and trust in the EU has declined, raising concerns about potential negative outcomes for EU stability. Thus, public opinion towards EU membership is becoming increasingly important as a barometer of national elite consensus and of policy competition (De Vries 2018).

In the light of these developments, this article aims to analyse citizens’ attitudes towards the EU using data from the European Election Studies (EES) Voter Study 2019. The main goal is to explore how Euroscepticism is taking root in public opinion and how EU citizens would react in the face of a possible referendum to leave the EU, similar to the one held in the UK. Although the Brexit referendum provides context, this study does not directly analyse its impact. Specifically, the research question being addressed is: What are the key factors that influence public attitudes towards the European Union and how do these attitudes vary across different EU member states in relation to the potentiality of a withdrawal referendum?

This study takes into account the multidimensional construct of Euroscepticism, which consists of different attitudes and behaviours towards the EU (Boomgaarden et al. 2011). Additionally, the variations in the dimensional structure of EU attitudes across different countries are assumed to reflect country-specific variations in the strength of the different dimensions (De Vreese et al. 2018). Furthermore, recent studies have shown that factors such as political ideology, trust in EU institutions, and perceptions of national identity influence variations in public preferences for future EU scenarios across countries (Goldberg et al. 2021a, b). Overall, this study aims to identify the main clusters of European citizens improving existing literature by providing a more in-depth understanding of the factors that influence public attitudes towards the EU and how these attitudes vary across different member states in relation to the potentiality of a withdrawal referendum.Footnote 1

This article is divided into four sections: the first analyses Euroscepticism in public opinion; the second describes the data and method used to select the variables for categorical principal component analysis (CATPCA), considering the components for a cluster analysis of European citizens; the third aims to understand different countries’ approach to a referendum on an exit from the EU; and the fourth section highlights the conclusions.

The public Euroscepticism

Euroscepticism, initially viewed as a characteristic of British politics (Harmsen and Spiering 2005; Leconte 2010), has developed alongside the EU integration process, representing a contingent or (un)qualified opposition to it (Taggart 1998).

This opposition is better defined in party-based studies on Euroscepticism, where Taggart and Szczerbiak (Szczerbiak and Taggart 2008; Taggart and Szczerbiak 2004, 2018) distinguish between hard and soft Euroscepticism. Hard Euroscepticism encompasses political parties that oppose EU membership and reject the integration process, while soft Euroscepticism there are parties which are not against EU membership but specifically oppose some policies that undermine the national interest.

In this context, political parties’ strategies to mobilise citizens’ opinion of European integration have become a primary differentiator between pro-EU political parties and Eurosceptic ones. While the former play a vital role in the existence of the integration model, Eurosceptic parties often adopt anti-Europeanism as a policy strategies for political competition, effectively adopting a new ideological approach (Flood and Soborski 2017; Leconte 2015). Anti-Europeanism aligns with the broader anti-establishment model, characterised by a mistrust of traditional political institutions and a desire for more direct forms of democracy and increased citizen involvement in the political process (Mudde and Rovira Kaltwasser, 2017). This model has gained popularity in recent years, particularly in the wake of the global financial crisis, as many citizens have come to feel left behind by the traditional political establishment (Inglehart and Welzel 2005). This kind of approach is often associated with the rise of populist movements and parties in Europe and North America, as well as the emergence of leaders who present themselves as outsiders and reject the traditional political establishment. Furthermore, this model is also associated with the decline of traditional political parties and the rise of new political actors such as populist parties, which are often Eurosceptic and anti-immigrant (Fitzi et al. 2018).

In recent decades, a number of significant EU-centred surveys, such as Eurobarometer, EES, have enabled us to gain a deeper understanding of public Euroscepticism, focusing on the essential attitudes towards the EU institutions. Among the factors that seem to drive Euroscepticism, there are feelings about national politics and institutions, distrust of supranational institutions, fears about the impact of the integration process on national identity, and personal interest-based utilitarianism (Hobolt and De Vries 2016). Among all, the utilitarian aspect appears to be more closely correlated with the European integration process, although exclusive national identity also prejudices European citizens against it (McLaren 2007). The utilitarian explanation for public support for EU integration has its origin in 1990s studies, which interpreted individual cost-benefits analyses as indicating a strong correlation between higher socio-economic levels and support for the EU integration process (Anderson and Reichert 1995; Gabel 1998). Indeed, a first block of studies has highlighted how the EU commonly was not perceived as a significant threat to the national and cultural identities of each country, confirming the key role of economic costs and benefits in public opinion of the European integration process (McLaren 2002). As such, Europeans who had faith in a robust economic future accepted European integration as a positive process, while those with a negative view adopted a Eurosceptic approach. Moreover, recent studies have emphasised the importance of education level as a factor in EU support (Hakhverdian et al. 2013). However, more generally, there has been a wider decline in EU trust in recent years (Berg 2019; Eurofound 2022; Roth et al. 2022). This decline in support is particularly notable during two acute crises: the 2007–2010 economic crisis and the subsequent austerity measures implemented against member states, and the 2015–2016 refugee crisis, which highlighted the limitations of European governance in dealing with the large influx of immigrants from Africa and the Middle East.

During the global economic crisis, although a steep increase in Euroscepticism was registered in the countries most touched by the crisis, the economic determinants haven’t become an important cause for Euroscepticism. Instead, factors such as national identity and political institutions played an important part in explaining public Euroscepticism (Serricchio et al. 2013). In fact, the relationship between national identity and support for integration has been a subject of previous research, and it has consistently been found that identity-related factors are more influential than economic factors in sha** public opinion (McLaren 2006). It should be noted that these factors are often interrelated and that the relationship between national identity and support for integration can be complementary. The general assumption is that the supranational nature of the European Union may erode national identity, and therefore, individuals with a strong sense of national identity tend to be less supportive of the EU integration process (Carey 2002; Hooghe and Marks 2009).

In contrast, recent studies on the Eurozone have shown a shift towards a more utilitarian approach in public opinion, where the economic cost–benefit has become a more important factor in determining support fort the EU (Hobolt and Wratil 2015). Other scholars have highlighted the role of economic conditions in sha** citizens’ attitudes towards EU institutions during the Great Recession (Gomez 2015). As a result, economic factors such as unemployment and high interest rates have been strongly associated with lower levels of support for the EU. Moreover, the crisis has made Europeans more responsive to economic issues—although this perception is more reliable in countries with fewer welfare protections than ones based on intensive social policy (Anderson and Hecht 2014)—with potentially destabilising consequences in the southern regions (Bosco and Verney 2012) since the turning point of Maastricht (Verney 2011).

A second crucial moment in European history appeared to shift the centre of gravity of support for national identity factors. In 2015, the EU was characterised by the European refugee crisis, which reached a pressure point in the two-year period 2015–2017, causing significant strain on Europe’s borders and highlighting the EU’s inability to manage the emergency. This situation was closely linked to non-economic factors, such as security or cultural/national identity, that could be affected by the perceived threat of immigration. Indeed, as previously discussed, these aspects have a high likelihood of changing citizens’ attitude towards EU institutions (McLaren 2002, 2004). A strong correlation exists between Eurosceptic positions and negative sentiment towards immigrants, as demonstrated by De Vreese and Boomgaarden (2005). Additionally, the exclusive perception of national/local identity is likely to result in the development of Eurosceptic positions (Hooghe and Marks 2004).

Despite the correlation between critical attitudes towards immigrants and Euroscepticism, the refugee crisis of 2015–2017 surprisingly did not result in an increase in anti-immigrant attitudes or critical views towards the EU, as highlighted by Daniel et al. (2019). However, in certain southern member states, the refugee crisis led to an increase in distrust towards the benefits of EU membership. For instance, in Italy, the Dublin regulation’s asylum rules were widely perceived as too costly in comparison to the benefits of EU membership (Dixon et al. 2018; Geddes and Pettrachin 2020).

While a growing body of literature has attempted to define the individual role for EU integration support between a utilitarian and/or identitarian model, public opinion is also heavily influenced by national political contexts (Hobolt and De Vries 2016). Indeed, the national context serves as a benchmark for how citizens evaluate EU integration, and several scholars have emphasised how the EU integration process was so disconnected from citizens’ reality that the construction of their opinion about the EU was entrusted to proximity sources such as national politics (De Vries and Edwards 2009) or media influences (Azrout et al. 2012; De Vreese and Boomgaarden 2006; Hooghe 2003).

From this perspective, two main approaches exist to the relationship between the national and European levels. On one side, it is argued that citizens use national institutions as a benchmark to evaluate European integration and therefore, domestic factors become more important than European ones due to a lack of knowledge about EU institutions (Anderson 1998). As a result, Europeans form attitudes towards the EU by using nation-states as a reference point (Kritzinger 2003). On the other side, studies have shown an inverse relationship between national evaluation and EU support, where citizens who are dissatisfied with national policies have a higher level of support towards EU policies and vice versa (Sánchez-Cuenca 2000). This same benchmark system is applied to the quality of democracy at the national level, with citizens who have positive evaluations of their national democracy having lower support for EU integration (Rohrschneider 2002).

Several factors influence public views on EU integration, each with different impacts on public Euroscepticism. These determinants, which can be categorised as utilitarian, identitarian or national, can have a varying effect in specific historical, economic or national contexts.

Given the impact of national evaluation on EU support and the role of direct participation tools, such as referendums, in sha** the political bond between citizens and institutions, it is evident that the relationship between the national and European levels is complex and multifaceted. It must be said that in the past national governments have voluntarily committed to holding referendums on European issues, despite not being legally obligated to do so. Scholars have analysed the political strategy behind these decisions, which in most cases were driven by national policies linked to domestic issues (Oppermann, 2013).

Throughout the EU history, some citizens have had opportunities to express their views on European integration through referendums. Over 50 referendums on European issues have characterised European politics, such as the treaty reforms (Mendez et al. 2014). However, there are two main views on the role of EU referendums in the political connection between citizens and institutions. Some scholars have argued that EU referendums are equivalent to second-order ballots linked to domestic political issues (Franklin et al. 1995), while others have claimed that electorates make their choices based on their own attitudes towards the EU (Hobolt 2009).

In this regard, this paper aims to understand Europeans’ view of the EU by examining a hypothetical vote in a referendum on exiting the EU.

To do so, the paper analyses the Eurosceptic component in all member states using the EES survey data set through CATPCA, a method for examining categorical (ordinal and nominal) data, which is increasingly being used among (Teney et al. 2014; Benasaglio Berlucchi 2022; De Luca 2022; Gómez-Balcácer et al. 2022). This study employs CATPCA to classify the component structure of citizens’ attitudes by considering the ordinal and nominal characteristics of the data collected in the EES survey. This approach reduces the data, including significant variables, into one or more components that can be used in future studies.

Methods and research design

This paper is based on the 2019 European Election Study (EES) Voter Study, which was carried out in 28 EU member states following the European Parliament elections (Schmitt et al. 2020). The survey was primarily administered online and consisted of over 100 questions. Respondents were selected randomly from access panel database using stratification variables, resulting in a sample size of 26,538 individuals, with roughly 1,000 interviews conducted in each EU member state except Cyprus, Luxembourg and Malta, where 500 interviews were conducted.

This study focuses on analysing 21 variables from the EES data set in five distinct areas: a) national politics focus (1 to 6); b) European politics focus (7 to 11); c) citizens’ attitude towards general policy issues (12 to 17); d) citizens’ attitudes towards politics (18 and 19); and e) sociodemographic characteristics (20 and 21) (see Table A in supplementary material).

The initial sample of 26,538 cases was reduced to 16,546 after eliminating cases for the following reasons. The question items characterised an interval scale with values ranging from 2 to 11, inclusive of the options ‘don’t know’ and ‘missing values’. However, cases containing ‘don’t know’ and ‘missing values’ were eliminated as they posed a hindrance to the statistical analysis, given that such variables cannot be ordered. This choice resulted in the purification of the data set, rendering the variables ‘ordinal’. With the remaining 16,546 cases, a CATPCA with optimal scaling was applied to convert the ordinal variables into metrics. Subsequently, a cluster analysis was conducted on the identified factors, followed by a bivariate analysis exploring the relationship between sociopolitical variables and the potential for a referendum regarding an exit from the EU.

The purpose of this analysis is to determine the indicators of citizens’ attitudes in the EES sample towards national and European governance, as well as towards politics in general. Additionally, the validity of the structure of these dimensions will be examined, and profile of distinct citizen groups will be identified through cluster analysis. The significance of the identified clusters in relation to sociographic and political variables will be assessed, with a focus on understanding the impact of the proposed ‘EU exit’ referendum. Finally, the distribution of the clusters at the national level will be evaluated.

In this study, the CATPCA was characterised to analyse the data set in order to uncover the underlying components of the qualitative and ordinal observed variables. The decision to use CATPCA for data analysis was based on several objectives, characterised as follows: a) reducing the complexity of the data; b) investigating the relationships between the 21 observed variables; and c) identifying the latent structure present within the data set.

CATPCA utilizes optimal scaling procedures to ‘enhance’ the qualitative variables and analyse the main components. The values for the structure have a monotonic function for each type of nominal (dummy) and ordinal variable observed with the order of the modes preserved. This aims to maximise the variance between the alternative modes, assuming certain assumptions.

The analysis was performed with 100 iterations, and satisfactory level of Cronbach’s alpha (the highest reaches 0.798 with over 19 per cent of variance explained) is presented in the analysis (Table 1); using the Kaiser rule, six components were extracted. The study also verified eigenvalues through parallel analysis, which resulted in six components; the Scree plot showed four components.

Table 1 Model summary
Full size table

The analysis suggests that the optimal solution comprises five components, which should be evaluated through the interpretation of the factor coefficients. An oblique rotation method (e.g. Promax) was subsequently applied, taking into account the possibility of a correlation between the components/factors (partly due to the similarity of some variables). Bartlett’s test was significant (p = 0.000), and the KMO index (Kaiser–Meyer–Olkin measure of sampling adequacy) was 0.773, indicating that the factorial model is appropriate for the data analysis. In terms of the communalities (which indicate the proportion of individual indicator variance that remains explained by the factorial model despite the reduction to ‘p’ components and the loss of information that follows), for almost all the indicators the value exceeded the critical threshold of 0.50. The five-component extraction, in fact, allows 53.2 per cent of the variance to be explained in cumulative terms.

Upon initial analysis of the pattern coefficient matrix (Table 2), the five-component solution does not display significant cross-loading. Component 1 appears to represent Eurosceptic claims, with strong and negative representations of all variables focused at European level and with a utilitarian aspect: EU membership is undesirable; there is no trust in the EU institutions and how they generally operate; the UK is seen in a better light after the Brexit referendum. An identity aspect is added to these variables, with full, active support for a restrictive policy on immigration. Component 2 seems more focused on the national benchmark of a pessimistic view of past and future domestic economic trends and perceives the government’s results more negatively and does not trust the National Parliament. Component 3 is more focused on the highest level of political interest, specifically related to the ability to access more political information in a democratic country. The last two components show slightly interpenetration through cross-loading values. However, Component 4 is closely associated with conservatism, displaying four key characteristics: an older demographic, strong support for restricting privacy rights to combat crime, opposing same-sex marriage and a belief in the importance of economic growth over environmental protection. Component 5 is more focused on financial issues, opposing state intervention in the economy and the redistribution of wealth, and representing the highest levels of standards of living.

Table 2 Pattern matrixa
Full size table

As a result, the five components have been renamed as follows: Euroscepticism (1); National pessimism (2); Knowledgeability (3); Conservatism (4); Economic liberalism (5).

The obtained and standardised factorial scores were employed in a cluster analysis using the k-means cluster method with principal component analysis (PCA) factors as input to define the profiles of European citizens through their attitudes towards the previously analysed aspects. This approach aims to maximise the similarity of elements within the groups and the dissimilarity between them. The k-mean method is a non-hierarchical approach that allows for the maximisation of both internal and external variance (within and between groups). An initial cluster range (from 2 to 7) was determined using the values ​​of the F test. The interpretability of clusters was based on the final group centres (averages of the clusters, taking into account the grou** variables), the number of clusters, their homogeneity, and the calculation of each pseudo-F.

Base on the evaluation criteria, the most reliable solutions were found to be those with two , three, and four clusters. The pseudo-F values decrease as the number of clusters increases ​​(cluster 2 = 3704.57; cluster 3 = 3809.58; cluster 4 = 3523.93). Therefore, the four-cluster solution was chosen, taking into account its higher homogeneity. The four clusters are presented in Table 3.

Table 3 Final cluster centres
Full size table

Cluster 1 exhibits a strong sensitivity towards ‘Knowledgeability’ (score of 0.30, positive), but relatively low sensitivity towards ‘Conservatism’ (score of1.26, negative). This cluster can be characterised as ‘Progressive’.

Cluster 2 shows sensitivity towards both ‘Conservatism’ (score of 0.73, positive) and ‘Knowledgeability’ (score of 0.64, positive) and comparatively low sensitivity towards ‘Euroscepticism’ (score of − 0.58, negative). This cluster can be described as ‘Moderate’.

Cluster 3 displays a strong sensitivity towards ‘Economic liberalism’ (score of 0.90, positive) and a weak relationship with ‘Knowledgeability’ (score of − 0.79). This cluster can be labelled as ‘Homo economicus’.

Cluster 4 exhibits a high sensitivity towards ‘Euroscepticism’ (standardised score of 1.44, positive, the highest among the four clusters) and, although at a lower level compared to the European level, towards a negative domestic vision (score of 0.88, positive, the highest among the three clusters). However, this cluster displays relatively low interests in ‘Economic liberalism’ (score of − 0.48). This cluster can be classified as ‘Eurosceptic’.

The data was limited to a single national case and the four clusters were analysed in conjunction with several key descriptive variables from the EES questionnaire, including sociographic factors (age, education, area of origin and social class), political positions (left–right), and positions on a potential exit from the EU referendum.

The clusters profile and the member states

The classification of the four clusters enables the examination of various aspects of each group to shed light on how Europeans’ attitudes are associated with certain sociopolitical traits. Specifically, Table 4 displays a bivariate analysis of the four clusters against two types of variables: sociographic (such as gender, level of education, urban–rural dimension, and social class) and political (such as left–right self-placement and hypothetical vote for an EU exit referendum.

Table 4 – Cross-tables: European clusters vs sociopolitical variables (%)
Full size table

In terms of the sociographic variables, there is no significant difference in gender distribution, with males being more represented due to the structure of the sample. However, a higher female representation is evident in the ‘Progressive’ and ‘Homo economicus’ groups (around 46 per cent). Age, as a metric variable, positively characterises ‘Conservatism’ and negatively ‘Economic liberalism’ component. This results in the ‘Moderate’ group being composed of older citizens, while the ‘Homo economicus’ group consists of younger ones. ‘Progressive’ is made up of young citizens, and ‘Eurosceptic’ of older ones, although to a lesser extent.

In regard to years of full-time education, two critical observations arise. Firstly, the ‘Progressive’ and ‘Homo economicus’ groups have the lowest number of citizens with ‘low education’ (2.7 per cent), while the ‘Eurosceptic’ group has the highest (6.2 per cent). Secondly, ‘Progressive’ and ‘Moderate’, which are primarily anchored to the ‘Knowledgeability’ component, have the highest percentage of citizens with medium–high education levels (although ‘Moderate’ has the lowest rate of citizens still studying). On the other hand, ‘Eurosceptic’ ha the lowest percentage of highly educated citizens (45.9 per cent).

In terms of occupation, the highest rates of working-class (25 per cent) and lower-middle-class citizens (26.4 per cent) are found in the ‘Eurosceptic’ group. ‘Progressive’, ‘Moderate’ and ‘Homo economicus’ groups consist mainly of medium–upper class citizens. In the urban–rural dimension, the most significant difference is between two ‘Progressive’ group, which primarily belongs to large towns, and the ‘Eurosceptic’ group, which primarily belongs to middle–small towns or rural areas.

The political aspects of the sample is characterised by a higher placement of citizens towards the left and centre-left for the ‘Progressive’ group, gradually transitioning through the ‘Moderate’ and ‘Homo economicus’ groups, which occupy the more central areas of the left–right political spectrum, and ultimately reaching the extreme right ‘Eurosceptic’ group. Although ‘Moderate’ group also contains a high percentage of right placements.

The sample demonstrates a clear majority in favour of remaining in EU in a hypothetical referendum, with the highest percentages being in the ‘Moderate’ and ‘Progressive’ groups at 92.1 and 88.7 per cent, respectively. However, two factors contribute to a better understanding of the hypothetical referendum. Firstly, while the majority of the ‘Homo economicus’ group favours remaining in the EU (66.3 per cent), this group also contains the highest percentage of individuals who do not have a clear stance, indicated as ‘Don’t Know’ (19.5 per cent). This may be due to a lack of understanding of the EU or self-interest-based primarily on economic considerations. This influences the decision to stay in the Union until the benefits outweigh the costs; the risk is that a reversed trend may prompt a change in these unsettled positions. Second, the ‘leave’ position is prevalent in the ‘Eurosceptic’ group (64.4 per cent), which is unsurprising given their previously demonstrated negative attitudes towards the EU.

However, the hard-soft dimension, which can be seen as a form of ‘exit scepticism’ (De Vries 2018) or ‘hard Euroscepticism’ (Ejrnæs and Jensen 2021), could also be relevant. In a group primarily characterised by Euroscepticism, the decision to remain within the EU (22.5 per cent) may include soft Euroscepticism.

The other groups also contain significant portions of citizens who are willing to leave the EU, but these decisions are not solely based on Euroscepticism, but rather on ideological or economic factors. The data show that currently, just over 64.4 per cent of the ‘Eurosceptic’ group hold a ‘hard’ Eurosceptic position, while 22.5 per cent hold a ‘soft’ Eurosceptic position opposing exit from the EU. However, a high percentage of citizens in this group do not choose either option (13.1 per cent).

To summarise, the ‘Progressive’ group has a slightly higher female presence, higher levels of education, a tendency to reside in large urban centres, and a tendency to vote for centre-left parties with a higher percentage of citizens in favour of remaining in the EU. The ‘Moderate’ group mainly resides in medium-sized urban areas and consists of an average-income middle class, occupying mainly centrist positions, and has the highest percentage of pro-EU citizens. ‘Homo economicus’ group resides in small- to medium-sized towns, is mainly centre and centre-right, and generally favours remaining in the EU, but has the highest percentage of individuals who would not participate in a hypothetical referendum. Lastly, the ‘Eurosceptic’ group has the lowest levels of education and moderate incomes, resides in peripheral areas away from centre-left political views, and holds a clear negative stance against Europe, but also includes individuals who may vote to remain in the EU.

The weight of individual clusters in different European countries and the role of the Eurosceptic cluster in EU member states should also be taken into consideration. Indeed, the EU average of the Eurosceptic cluster is 19.6 per cent (Table 5).

Table 5 European clusters in member states (%)
Full size table

The analysis of the data presents an interesting examination of the formation of two groups that are more or less homogeneous. On the one hand, there are the 14 countries, including Belgium, Bulgaria, Croatia, Cyprus, Czech Republic, Finland, France, Greece, Italy, Latvia, Slovenia, Slovakia, Sweden, and the UK, which display varying levels of Euroscepticism but have a higher-than-average percentage of Eurosceptics. The countries are arranged in Fig. 1 based on their Eurosceptic cluster, starting with those that have the highest difference between their levels of Euroscepticism and the European average, to those with values that are lower than the European average. Notably, the presence of a strong ‘Homo economicus’ is prevalent in countries with higher Euroscepticism rates, while the countries with fewer Eurosceptics have higher levels of ‘Progressive’ and ‘Moderate’ values.

Fig. 1
figure 1

Difference from EU average (%)

Full size image

However, only eight of the countries pose a concern to the EU with their high levels of Euroscepticism, including Belgium (27.2 per cent), Bulgaria (27 per cent), Czech Republic (26.9 per cent), France (27.9 per cent), Greece (26.6 per cent), Latvia (29 per cent), Sweden (26.2 per cent) and, despite everything, the UK (35.7 per cent). The Czech Republic and Latvia also display high levels of ‘Homo economicus’, which was previously considered the greatest unknown in comparison to a potential EU exit referendum.

On the other hand, Austria, Germany, Hungary, Luxembourg, Malta, Poland, Romania, and Spain seem to strongly support the pro-EU dimension through a majority in both clusters that ensure remaining in the EU. Thus, among the two opposing blocs, there are countries with higher percentages of Eurosceptics and others with more manageable levels. Geographically, the northern countries of the Scandinavian peninsula, Finland, and Sweden, have the highest percentage of Eurosceptics. Meanwhile, central and eastern European countries, led by Bulgaria, the Czech Republic and Latvia, hold the lead in Euroscepticism. In continental Europe, Belgium and France have high levels of Eurosceptics, while only Greece and parts of Italy display high levels in southern Europe, which has been a long-standing concern for the EU. Among the latter, Malta, Portugal and particularly Spain seem to have consolidated a group of strong EU supporters.

Conclusion

The aim of this study was to examine the correlations between the attitudes of European citizens’ towards Euroscepticism and their potential responses to a hypothetical exit referendum akin to the UK. The study aimed to identify the major clusters of attitudes through an evaluation of the perceptions of both European and national aspects.

The analysis revealed a clear pattern of Euroscepticism across the sample population, which was frequently associated with a negative view of national politics. This sentiment appeared to be in opposition to the components of ‘Knowledgeability’ and ‘Conservatism’, while showing a lack of sensitivity to ‘Economic liberalism’ (see Table 4). This suggest that ‘Euroscepticism’ constitutes a distinct cluster, referred to as the ‘Eurosceptic’ group, albeit a smaller one, which is also closely tied to national factors.

A smaller subset of the sample displayed characteristics of the ‘Eurosceptic’ cluster, including low levels of education, residency in small- to mid-sized town or rural areas, a lower-middle social class, and close proximity to the centre-right political spectrum. In this context, a referendum result in favour of exiting the EU appears to be a remote possibility. The majority of the European population appears to be inclined towards a ‘remain’ vote, with two significant clusters, the ‘Progressive’ and the ‘Moderate’, demonstrating almost unanimous support for this position. There was, however, a degree of confusion regarding the stance of the ‘Homo economicus’ group. The ‘Eurosceptic’ cluster was driven by the ‘hard’ dimension, highlighting the entrenched nature of this position within a portion of the population, a trend that requires continuous monitoring in future studies. It is important to note, indeed, that the proportions willing to vote to leave the EU may not be static. Campaigns during referendums can significantly shift public opinion, as was the case in the UK. Furthermore, the mere existence of a proportion willing to vote 'leave' does not necessarily indicate that a referendum will be held.

The study also identified differences in the level of Euroscepticism among the member states, with Belgium, Bulgaria, France, Greece, Sweden, and particularly the Czech Republic and Latvia displaying higher levels of Euroscepticism. Conversely, Austria, Germany, Hungary, Luxembourg, Malta, Poland, Romania, and Spain exhibited greater pro-European sentiment.

The study found that Euroscepticism has undergone shifts in recent years, with some countries exhibiting an increase in their Eurosceptic dimension, while others, such as Italy, have experienced a decrease. This variation was attributed to a combination of factors, including utilitarian considerations for evaluating EU membership and supranational democracy, identity factors such as the immigration crisis, and the interplay between the European and national level, where the national benchmark, both economic and political, plays a crucial role in sha** the Eurosceptic position.

The distrust towards European integration has been a fluctuating phenomenon. At times, it manifests as a direct criticism of the national political elite and particularly towards EU institutions, with the notion of an exit referendum being proffered as the panacea for all issues. It is worth noting that this analysis constitutes a snapshot of the European sentiment prior to the COVID-19 pandemic and the Ukrainian conflict. These crises present a novel avenue for analysing public Euroscepticism in order to gain a deeper insight into the future attitudes of citizens towards the European Union. For example, the COVID-19 crisis has different crucial premises which could register a further impact on attitudes towards the EU—several member states have expressed doubt about the functioning of EU institutions during the emergency. Scholars highlighted that, among the most affected countries, for example, the Italian people felt abandoned by EU institutions and other member states, becoming increasingly Eurosceptic because of how the virus was handled (Russack and Blockmans 2020).

During the COVID-19 emergency, some national governments achieved higher levels of popularity by frequently criticising the EU as inadequate for the task. Furthermore, several countries also tried to organise a group of solidaristic members against EU decisions about coronavirus. All these decisions have impacted public opinion about EU membership with essential changes in each relationship. Indeed, several governments have supported the concept of future pledge referendums about EU issues which have transformed the before/after COVID era into an interesting moment in understanding attitudes towards the European integration process.

Furthermore, the Russian military attack on Ukraine in February 2022 had significant repercussions for the EU. The European Peace Facility, which had only been part of the European budget since 2021, was swiftly redirected in its entirety to financing the purchase of Ukrainian weapons, and these arms were supplied by EU member states (De Vries 2023). Moreover, the EU responded by imposing economic sanctions, banning Russian TV channels, and implementing a partial embargo on Russian oil. Theoretically one might expect an external military threat could lead to increased integration within the EU. However, the EU rapidly saw a resurgence of old divisions over who would bear the financial burden (De Vries 2023). Furthermore, there is a significant propensity among EU member states to accommodate immigration from Ukraine and integrate refugees into their respective labour markets. Additionally, there is a widespread political endorsement in EU countries for Ukraine’s admission into the EU (also the Republic of Moldova, and Georgia made significant strides in their aspirations to become members of the EU). Nonetheless, several scholars have noted that incorporating Ukraine into the EU may result in various issues, such as the unequal distribution of immigration across individual member states, particularly in those countries that do not enforce transition periods (Welfens 2023).

A new cooperation step is arrived also at the 2022 Danish referendum. Indeed, Denmark held a referendum in June 2022 on their opt-out clause about the Common Security and Defence Policy (CSDP) of the EU. The outcome of the vote was ground-breaking, as 67% of the Danish population chose to join the E’'s CSDP, ending their 30-year opt-out stance and their previous resistance to further EU integration. This decision allows Denmark to participate in EU military operations and collaborate in military capability development and acquisition. Similarly, Finland and Sweden’s recent abandonment of military neutrality and joining NATO reflects a similar trend.

Now that the stakes of cooperation have been raised beyond the level of the individual state due to the external threat, the question remains how the EU will shape its future and organise itself, and how its member states will respond.