2.1. The state of the art

EU public opinion research focusses predominantly on the question of the legitimacy of the EU in the eyes of the European people, their assessment of their countries’ (prospective) EU membership and their evaluation of EU policy. This body of research has greatly enhanced our understanding of EU public opinion, yielding insight into the attitudes and perceptions of citizens across member states: In addition to revealing that peoples´ demographic characteristics—like age, educational level, social–economic status, geographical location and gender—affect their attitudes towards the EU, a range of theoretical perspectives are shown to offer explanations for peoples’ support for the EU (Bauer, Reference Bauer2020; Hobolt & de Vries, Reference Hobolt and de Vries2016). Many of the demographic explanations, for instance, rely on a utilitarian explanatory mechanism: When EU policies are (seen to be) beneficial to citizens or their countries, they tend to support and trust the EU (Dijkstra et al., Reference Dijkstra, Poelman and Rodríguez-Pose2020). Moreover, the extent to which people hold national or European identities as well as the cues they take from politicians, media or their peers are also shown to play a role (Aichholzer et al., Reference Aichholzer, Kritzinger and Plescia2021; Armingeon & Ceka, Reference Armingeon and Ceka2014; Mikulaschek, Reference Mikulaschek2023; Schlipphak & Treib, Reference Schlipphak and Treib2017). In addition, there is an emerging literature that indicates that emotions affect peoples’ support for, and trust in the EU (Boomgaarden et al., Reference Boomgaarden, Schuck, Elenbaas and De Vreese2011; Garry, Reference Garry2014; Vasilopoulou & Wagner, Reference Vasilopoulou and Wagner2017). Finally, preferences with regard to European integration align with other attitudes like globalism and migration and are part of a new political cleavage in society (Hooghe & Marks, Reference Hooghe and Marks2018; Hooghe et al., Reference Hooghe, Marks and Wilson2016; Treib, Reference Treib2021). Over time the field has become more diversified in terms of its explanans of interest. As more and different data-sources became available—like the Special Eurobarometer, the Europarlemeter and European Value Studies—scholars were able to refocus their questions to specific policy areas or current events. In addition, the field slowly moved away from simple “for or against” questions, to study different forms of EU support and Euroscepsis (De Vries, Reference De Vries2018).

2.2. Research gaps

However, fundamental questions remain unanswered, specifically with regard to the nature of peoples’ beliefs regarding, and attitudes towards the EU. As a polity that is quite distant to its constituents, EU citizens often have little interest in, and knowledge of the EU, its structure or policy making system (Clark, Reference Clark2014; Hix, Reference Hix2015). Theories across the fields of behavioral economics, public opinion and social psychology suggest that under these circumstances, citizens’ attitudes are more likely to be based on mechanisms like heuristic thinking, cue-taking or attribution rather than on the rational processing of information, argumentation and reflection (Antonakis & Atwater, Reference Antonakis and Atwater2002; Armingeon & Ceka, Reference Armingeon and Ceka2014; Brosius et al., Reference Brosius, van Elsas and de Vreese2020; Hobolt & de Vries, Reference Hobolt and de Vries2016; Hooghe & Marks, Reference Hooghe and Marks2016; Kahneman, Reference Kahneman2011; Popper, Reference Popper2013; Shamir, Reference Shamir1995; Van Esch & Steenman, Reference Van Esch and Steenman2025). This raises the question (Converse, Reference Converse2006; Wilker & Milbrath, Reference Wilker and Milbrath1970) to what extent citizens actually hold a coherent, ideologically informed set of beliefs regarding the EU.

This question was first posed by Converse in the context of US public opinion in the 1960s (Converse, Reference Converse2006; Wilker & Milbrath, Reference Wilker and Milbrath1970). Converse argues that in contrast to the political elite, the majority of the general public does not possess a coherent set of political beliefs and tend not interpret politics through an ideological lens. Focussing on liberalism and conservatism, he shows elites to have structured and coherent ideological beliefs while citizens often provide incoherent or random responses to political survey questions. Notwithstanding the controversy of this finding (Norpoth & Lodge, Reference Norpoth and Lodge1985; Wilker & Milbrath, Reference Wilker and Milbrath1970), the question of whether people hold coherent, ideological beliefs remains relevant in the context of the EU.

Noting the limited knowledge of, and interest in the EU of many Europeans, some EU scholars follow Converse in questioning whether the attitudes citizens report during public opinion polls actually represent pre-existing beliefs (Aldrin, Reference Aldrin, Gaxie, Hube and Rowell2011; Gaxie, Reference Gaxie, Gaxie, Hube and Rowell2011). Others draw attention to the saliency of ambivalence and indifference in peoples’ attitudes towards the EU (De Vries, Reference De Vries2013; Lutz, Reference Lutz2021; Steenbergen & Brewer, Reference Steenbergen, Brewer, Saris and Sniderman2004; Stoeckel, Reference Stoeckel2013; Van Ingelgom, Reference Van Ingelgom2014) and report that around 30–40 per cent of citizens are ambivalent about the EU (Stoeckel, Reference Stoeckel2013). Such ambivalence may stem from how the subject is interpreted. Lutz, for instance, finds that ambivalence regarding free movement in EU results from the fact that respondents value their own free movement differently than that of others (Lutz, Reference Lutz2021). The discussion of these questions, however, remains more limited and less fundamental in EU Studies than in the field of US politics.

2.3. The role of methods

This may be partly due to the sources and methods used in the domain. First, given the size and nature of the EU and the diversity of its member states, gathering representative data is difficult and expensive. Scholars have therefore relied predominantly on the Eurobarometer, a bi-annual pan-EU public opinion poll held by the EU institutions. The near monopoly and political nature of this source limits the kind of questions that scholars can study (Aldrin, Reference Aldrin, Gaxie, Hube and Rowell2011; Nissen, Reference Nissen2014). Although EU scholars are increasingly sourcing data independently, even large collaborative efforts like the European Value Studies remain more limited in scope than the Eurobarometer.

Second, regardless of the precise source, EU public opinion research has been dominated by survey studies. While this method provides an efficient and tested way to study attitudes, it is less suited for capturing peoples’ underlying and pre-existing belief systems, or to explore their consistency (Glynn et al., Reference Glynn, Herbst, Lindeman, O’Keefe and Shapiro2018). Traditional survey studies separate attitudes from the wider belief system they derive from and obscure the interrelation between attitudes people hold on different topics. Moreover, especially when the questions pertain to issues that are not close to peoples’ daily lives or when they have little knowledge of the subject, surveys risk measuring artificially constructed rather than pre-existing attitudes. Also, the top-down nature of the method—with its predefined questions and limited response options—provides respondents with limited space to convey their own thoughts (Aldrin, Reference Aldrin, Gaxie, Hube and Rowell2011; Gaxie, Reference Gaxie, Gaxie, Hube and Rowell2011; Glynn et al., Reference Glynn, Herbst, Lindeman, O’Keefe and Shapiro2018; Wilker & Milbrath, Reference Wilker and Milbrath1970; Zaller & Feldman, Reference Zaller and Feldman1992).

Finally, surveys are not well-equipped to study ambivalence. Ambivalence is generally defined as judging something to be positive and negative at the same time, which is hard to capture with a survey design (Norpoth & Lodge, Reference Norpoth and Lodge1985; Stoeckel, Reference Stoeckel2013). Scholars have found ways to get around this by developing specific sets of questions, using proxies or other methods like survey experiments or integrating qualitative techniques like focus groups and interviews (De Vries, Reference De Vries2013; Lutz, Reference Lutz2021; Steenbergen & Brewer, Reference Steenbergen, Brewer, Saris and Sniderman2004; Van Ingelgom, Reference Van Ingelgom2014). However, more information and other tools are needed to develop a better understanding of ambivalence (Stoeckel, Reference Stoeckel2013). Applying the technique of CM may provide an answer to this call.

3.1. The pros and cons of using CM to study public opinion

The specific nature of CM has several benefits over public opinion polls and may provide a useful methodological complement to the domain of public opinion research. This is especially the case when exploring the nature and coherence of peoples’ belief systems (Axelrod, Reference Axelrod1976b; Van Esch & Snellens, Reference Van Esch and Snellens2024; Yang & González-Bailón, 2017; Young, Reference Young1996).

First, as indicated above, standard survey questions are necessarily narrow in scope, are formulated top-down by the researcher and allow respondents only a few possible response options. Vignette studies or conjoint survey experiments offer respondents more leeway to speak their minds but to a more limited extent than CM. Even while working with preset concepts, CM allows respondents a lot of freedom to express their ideas. For instance, when offered a set of 30 concepts respondents could potentially draw 1800 different relations (30 concepts * 29 concepts * types of relations—positive/negative) and combine these to create an even higher number of different maps.

In addition, the bottom-up and flexible nature of CM reduces the risk that the attitudes it reveals are artificial constructs rather than reflect pre-existing beliefs, and allows respondents to express themselves in ways that researchers may not have imagined (Aldrin, Reference Aldrin, Gaxie, Hube and Rowell2011; Gatto & Panarello, Reference Gatto and Panarello2022; Gaxie, Reference Gaxie, Gaxie, Hube and Rowell2011; Glynn et al., Reference Glynn, Herbst, Lindeman, O’Keefe and Shapiro2018; Nissen, Reference Nissen2014). Finally, respondents are free to ignore topics they do not have an opinion on, or to draw a CM with only a few concepts or relations. Our study shows that people also take this liberty: The minimum number of relations our respondents drew was one, the maximum 21.

Finally, unlike most other methods by which scholars may obtain deeper insight in citizens’ beliefs—like interviews and focus groups—CM can be applied in large-N research: The data for the maps can be gathered in an (online) survey setting and automatically stored. In addition, software has been developed to analyze the data in an efficient manner (Van Esch, Snellens et al., Reference Van Esch, Snellens, Dekker, De Gooijer, Lobée, Patist, Van Riet, Rozendaal, Tichattibin and Van der Wulp2024; Van Esch, Treep et al., Reference Van Esch, Treep, Moopen and Snellens2024). Finally, because of its focus on belief systems rather than single attitudes as well as the freedom it allows respondents to express their ideas in paradoxical ways, CM can be used to study ambivalence. In fact, as will be illustrated below the technique offers different ways to operationalize and measure ambivalence.

Using the CM method for public opinion research also has drawbacks. First, if incorporated in a large-n context it cannot escape the problems associated with eliciting attitudes from respondents at a distance: The concepts with which to draw the maps may be interpreted differently by different people. Moreover, combining negative concepts and negative relations is problematic in CM research as this may result in relations that are the equivalent of double negative statements which are difficult to interpret by both respondents and researchers (Boukes et al., Reference Boukes, Van Esch, Snellens, Steenman and Vliegenthart2020). Third, the technique is not well-known among respondents. Although maps generally correspond well with what respondents intend to express (Hodgkinson et al., Reference Hodgkinson, Maule and Bown2004; Vo et al., Reference Vo, Poole and Courtney2005), the task makes some respondents insecure about whether they are fulfilling the task correctly. Finally, as a novel approach to public opinion research, the ways to analyze the data on a larger scale and regarding questions about ambivalence in particular are still being developed (Van Esch, Treep et al., Reference Van Esch, Treep, Moopen and Snellens2024; Young, Reference Young2001).

4.1. Case selection and data collection

To answer these questions, this study will explore the belief systems of 504 Dutch citizens regarding the Eurozone crisis. The Eurozone crisis constitutes a highly salient case that features in many public opinion studies. Moreover, as a founding member of the EU and EMU the Netherlands is often included in studies of EU public opinion. This makes this case both a typical and a hard case for our study (Yin, Reference Yin2009): Given that the case has been explored extensively, it will be hard to come up with new insights (sub-question one). Moreover, the complexity of the issue area increases the likelihood that citizens’ attitudes are based on cue-taking, heuristics, or attribution, making it less easy to refute the thesis of Converse (sub-question two). In addition, the Netherlands was less affected by the Eurozone crisis than other member states, reducing the chance for its citizens to hold strong ideological attitudes towards the subject. Finally, the complexity of the crisis increases the likelihood that people’s beliefs will be ambivalent, making it a typical case to study sub-question three thereby allowing us to explore the phenomenon of ambivalence further.

The data for this study was gathered as part of a larger research project including nine EU member states (Denmark, France, Germany, Hungary, Ireland, Italy, the Netherlands, Spain, and the United Kingdom). For each of these countries, in the spring of 2017, a representative sample of approximately 500 respondents completed an online survey comprised of the CM task and a short list of background questions (Van Esch et al., Reference Van Esch, Steenman, Joosen, Brand and Snellens2018). To elicit respondents’ cognitive maps on the Eurozone crisis, a purpose-built web-based software tool (DART) was used. This tool was built to provide a simple user-friendly online facilitator to guide respondents through the task of drawing their cognitive maps, that can be integrated into online survey software. It guides respondents step-by-step through the CM task providing instructions for each step (see Appendix 3). It also links the CM data automatically to the responses to the survey questions to facilitate the analysis of the data.

After a brief instruction on how to complete the CM task, respondents were asked to choose a maximum of seven out of 34 key concepts that to them were most relevant to the Eurozone crisis.Footnote ¹ These 34 concepts are derived from a list of the most salient concepts in the cognitive maps on the eurozone crisis of the Dutch prime ministers and central bank governors who served during the crisis. To make the concepts from the leaders’ CMs easier to grasp and to avoid jargon, concepts with similar meanings were collated, and the most technical terms were reworded into everyday language (see Appendix 3). Following CM standards, negatively worded concepts were avoided as much as possible (Boukes et al., Reference Boukes, Van Esch, Snellens, Steenman and Vliegenthart2020). After selecting the concepts, respondents were presented on a virtual whiteboard on which their selected concepts were randomly spaced. The concept “crisis-free Eurozone” was included automatically in the map as the central concept of this study.

The respondents were then asked whether and how they perceived these concepts to be related by drawing arrows between them, following the freehand drawing approach (Drury et al., Reference Drury, Pfaff and Klein2020; Hodgkinson et al., Reference Hodgkinson, Maule and Bown2004; Knox et al., Reference Knox, Furman, Jetter, Gray and Giabbanelli2024; Markíczy & Goldberg, Reference Markíczy and Goldberg1995). This approach is seen to be more intuitive, less time consuming and tedious by respondents. In case of probing the general public for their beliefs regarding a topic they have little interest in, this is important for it prevents respondents from skimming over the task or dropping out. In addition, Hodgkinson et al., find this approach to produce maps that respondents feel represent their ideas adequately (Hodgkinson et al., Reference Hodgkinson, Maule and Bown2004).

The respondents in our study had the choice between positive (green) and negative (red) arrows. They were allowed to draw arrows between a set of concepts in both directions and contrary to common practice, to draw both a positive and negative relation between the same set in the same direction (see Figure 2). The latter option offers respondents the opportunity to assert conflicting thoughts, the most strict indicator of ambivalence (see below). These options were, however, not explicitly disclosed to the respondents in order not to influence their natural thought process. In addition, we opted not to ask respondents to weigh the relations as is common in Fussy CM studies in order not to complicate the task in our non-supervised setting (Knox et al., Reference Knox, Furman, Jetter, Gray and Giabbanelli2024).

Figure 2. Operationalization of ambivalence as conflicting relations.

In the second part of the study, the respondents were asked a series of survey questions regarding their background, media use, political preferences and trust in a number of national and European institutions. Moreover, a question was included regarding their trust in the EU, using the wording of the Eurobarometer to make our findings comparable to most studies on EU public opinion. However, in order to be able to compare the CMs of different subgroups, we used a 5 point Likert scale from “very little trust” to “very much trust” and an additional “don’t know” option (score of 6) rather than the bivariate “tend to trust,” “tend not to trust” that the Eurobarometer uses.

4.2. Operationalization

In order to analyze the data, a collective Dutch cognitive map was composed on the basis of all the individual maps via the “aggregation method.” This involves merging the individual maps through common concepts (Vo et al., Reference Vo, Poole and Courtney2005). The positive and negative relations were merged into a single arrow and given the color (green/red) of the value of their sum (positive/negative). Information on the number of positive and negative relations was retained to enabled detailed analysis of the CMs (see Table 3 and the Appendix 1). The aggregation method results in complex CMs and makes it hard for individual respondents to recognize their own beliefs. However, in the context of this study this method seems the most appropriate as others require respondents to collaborate or are focussed on studying system dynamics. The aggregation method is also easy to apply in an online setting and mimics the way in which attitudes are aggregated in the democratic process (Knox et al., Reference Knox, Furman, Jetter, Gray and Giabbanelli2024; Vo et al., Reference Vo, Poole and Courtney2005). Nonetheless when analyzing the collective maps, one must remain cautious that some of the results could be an artifact of the aggregation method.

Table 1. List of terms and measures used in CM analysis

In order to explore to what extent the method of CM can credibly capture the belief system of the Dutch respondents with regard to the Eurozone crisis and lead to new insights (sub question 1), their collective CM was converted into a graph. In addition, several basic CM descriptive measures like the weight and the sign of the relation and saliency of the concepts as well as the evaluation of concepts were calculated using the Cognitivemapr R-package and displayed in the graph (Van Esch, Snellens et al., Reference Van Esch, Snellens, Dekker, De Gooijer, Lobée, Patist, Van Riet, Rozendaal, Tichattibin and Van der Wulp2024; Van Esch, Treep et al., Reference Van Esch, Treep, Moopen and Snellens2024).Footnote ² The visualizations of (excerpts of) the graph enables the narrative analysis of the collective map providing insight respondents’ belief system with regard to the Eurozone crisis. By comparing the findings of this analysis with an overview of existing insights in the attitudes of the Dutch regarding the Eurozone crisis, the potential added value of CM research is explored. As existing research has a strong focus on trust in the EU, the analysis of the CM will focus on the relation between citizens’ beliefs on the Eurozone crisis and their trust in the EU.

The second sub question asks whether CM can provide insight in the extent the thesis that citizens’ beliefs are less ideologically informed than those of the elite holds in the field of EU public opinion (Converse, Reference Converse2006). Since in the context of the Eurozone crisis, the ideologies that Converse focusses on—conservatism and liberalism—are of limited relevance, our analysis will focus on two rivaling economic ideologies that are widely seen as having played a crucial role in the EU and Dutch political and public debate on the Eurozone crisis (Dullien & Guérot, Reference Dullien and Guérot2012; Hall, Reference Hall2014; Schäfer, Reference Schäfer2016; Van Esch, Reference Van Esch2014): Ordoliberalism and Keynesianism. The Ordoliberal paradigm is characterized by a belief in the primacy of price stability, which may be ensured by pursuing austerity and denouncing monetary financing. For Keynesians, economic growth and employment take precedence and economic stimulation is advocated to promote these goals during economic downturns. Keynesians are also more favorable to monetary financing and for a central bank to act as a lender of last resort.

To measure the extent to which our respondents’ beliefs are ideological inspired, the measures “paradigm support” and “instrument support” are used. Paradigm support reveals to what extent the beliefs in the CM indicate support for a particular ideology or policy paradigm. Instrument support measures the extent to which respondents support different types of policy instruments (Van Esch & Snellens, Reference Van Esch and Snellens2024). The list of concepts used in our study contains references to three types of policy instruments used during the Eurozone crisis: reform of the European Economic and Monetary Union, stronger fiscal regulation and structural reforms. Paradigm support measures actors’ philosophical beliefs, while instrument support focusses on the more everyday practical and operational translation thereof. Including both acknowledges the empirical finding that citizens’ ideological beliefs seem to encompass both, and that actors may at the same time have different ideological leanings at the philosophical and at the operational level (Feldman & Zaller, Reference Feldman and Zaller1992; Zaller & Feldman, Reference Zaller and Feldman1992). As data on both measures is also available for the Dutch prime minister, Mark Rutte, and governor of the Dutch central bank, Klaas Knot (Van Esch & Snellens, Reference Van Esch and Snellens2024), their scores will be compared to those of our respondents. This provides a first indication as to the value of Converse’s thesis in the context of EU politics.

The final sub question asks if using CM may provide a deeper understanding of the extent and ways in which their beliefs regarding the EU are coherent or ambivalent. Ambivalence is generally defined as holding opposing evaluations towards the same object (Lutz, Reference Lutz2021). However, this general definition allows for a more and a less strict interpretation. In its most strict definition, ambivalence indicates that someone sees something to be both true and not true (Lavine et al., Reference Lavine, Johnston and Steenbergen2012). The second more lenient operationalization of ambivalent beliefs is more commonly used and defines ambivalent objects to have both positive and negative aspects or consequences (De Vries & Steenbergen, Reference De Vries and Steenbergen2013; De Vries, Reference De Vries2013; Lutz, Reference Lutz2021; Steenbergen & Brewer, Reference Steenbergen, Brewer, Saris and Sniderman2004; Stoeckel, Reference Stoeckel2013). There are no established measures for these definitions of ambivalence in the literature on CM, but the technique offers ways to measure both.

In the context of a cognitive map the idea of ambivalence as “conflicting beliefs” translates to drawing a both a positive and a negative relation between two concepts (Axelrod, Reference Axelrod1976a, see Figure 2). Whether a CM contains contradictory beliefs is visible in the map or the list of its relations. In order to establish the extent to which a CM holds contradictory beliefs, we calculated the percentage of conflicting relations of the total relations in the CM.

The second definition of ambivalence translates in concepts having both positive and negative consequences. In Figure 1, for instance, European cooperation has the positive effect of fostering a crisis-free eurozone as well the negative effect of leading to more bureaucracy. To establish whether a belief indeed carries different positive and negative implications, it is important to consider the belief system as a whole and first determine the value of each concept in its context rather than simply count the number of outgoing positive or negative relations. In Figure 1 for instance, European cooperation has two positive outgoing relations. However, since the concept bureaucracy holds a negative value to this respondent, European cooperation has both a positive and a negative effect and thus fits the second definition of ambivalence.

In addition to categorizing a concept as ambivalent or not, we also determined the extent to which a concept is ambivalent by factoring in how many positive and negative consequences it has (cf. Stoeckel, Reference Stoeckel2013). In Figure 1, for instance, European cooperation has both a positive and negative consequence, while liberalization has two negative and one ambiguous consequences. Liberalization is thus seen as largely negative and less ambivalent than European cooperation. To calculate the ambivalence score of a concept, we established the number of consequences and type of consequences (positive, neutral or negative) for all the concepts that have effects (outdegree > 0). We then calculated its ambivalence using the following function:

\begin{equation*} \textit{Ambivalence}\; \textit{score}\; (node)= 1-\Big(\sqrt (\sum \textit{consequences}^{2}\Big)\Big)/\textit{outdegree}\Bigg) \end{equation*}

To be able to compare the CMs, the average ambivalence of all ambivalent nodes was established. In addition, we calculated the ambivalence score as a percentage of all the cause concepts (outdegree > 0) in a map (see Appendix 2).Footnote ³

4.3. Description of the data

The level of trust in the EU among our respondents is largely comparable to that reported in the Eurobarometer that was published around the same time as our survey (EB 87, May 2017). Instead of 45 per cent, 42 per cent of our respondents tend not to trust the EU (Likert score 1-2) while 20 instead of 49 per cent univocally trust the EU (Likert score 4-5). Using a 6-point scale reveals that 29 percent of our respondents are ambivalent about their trust in the EU (Likert score of 3), which suggest that providing respondents with only two answer options—as the EB does—distorts respondents’ attitudes. Finally, the number of people in our sample who do not know whether they trust the EU is at 2 per cent smaller than in the Eurobarometer.Footnote ⁴ During the CM task, the 504 Dutch respondents drew a total of 2.783 arrows, an average of 5.5 relations per person using an average of 5.4 concepts. All 34 available concepts were used at least once.

5.1. The Dutch take on the Eurozone crisis

Research in EU studies has provided significant insights into the impact of the Eurozone crisis on people’s trust in the EU using survey methods (Armingeon & Ceka, Reference Armingeon and Ceka2014; Bauer, Reference Bauer2020; Dijkstra et al., Reference Dijkstra, Poelman and Rodríguez-Pose2020; Foster & Frieden, Reference Foster and Frieden2017). The literature, for instance, shows that the level of trust in the EU significantly declined over the course of the crisis from 65% in 2007 to 36% by 2015 (Foster & Frieden, Reference Foster and Frieden2017). This decline was particularly pronounced in countries with excessive budgetary deficits and that were subjected to austerity measures imposed by the EU. Conversely, citizens in countries that were less affected by the crisis—like the Netherlands—retained higher levels of trust in the EU (Foster & Frieden, Reference Foster and Frieden2017, see Figure 3). Exploring the reasons behind these trends, economic factors such as unemployment, economic stagnation, industrial decline, and the imposition of structural adjustment programs are identified as key contributors to the erosion of trust in the EU. In addition, citizens’ attitudes towards the EU correlate with their assessment of their national government’s crisis management. In contrast to citizens in highly affected countries, citizens in less affected countries show more confidence in their national government than the EU. Those in crisis-hit nations tend to prioritize public investment and economic stimulus over austerity measures, while citizens in other countries prefer spending cuts and austerity. Interestingly, citizens from crisis-affected countries demonstrate a higher acceptance of strict fiscal rules enforced by the EU, such as automatic fines for excessive public spending. In general, citizens evaluate the EU’s effectiveness based on general evaluations of the economy rather than on personal experiences (Bauer, Reference Bauer2020).

Figure 3. Trust in the EU in the Netherlands (Bauer, Reference Bauer2020; EB 85, EB 86, EB 87).

The Netherlands was only marginally hit by the eurozone crisis. While GDP remained relatively low in comparison to other member states, so did its governmental debt and deficit and unemployment rates. As a result, the Netherlands was not subjected to externally imposed structural adjustment programs but the government itself did impose austerity measures (Picard, Reference Picard2015; Van Esch & Snellens, Reference Van Esch and Snellens2024). At the start of the crisis, however, a large majority of Dutch respondents perceived the crisis as having a (fairly) significant impact on the national economy. In contrast to the hawkish, ordoliberal tone of the Dutch public debate and positions of the political elite (Bijsmans, Reference Bijsmans2021; Verdun, Reference Verdun2022), a majority of Dutch respondents supported providing financial help to other EU countries (Eurobarometer, 2010; Kleider & Stoeckel, Reference Kleider and Stoeckel2019). In addition, they favored coordinated EU action over individual national responses even though the majority did not feel the euro helped mitigate the crisis. Finally, little over half of Dutch respondents favored EU member states to invest rather than to reduce public spending. They also showed strong support for a range of EU measures to prevent future crises like stricter regulation of financial markets, the creation of a European monetary fund, penalizing non-compliance with EU rules on debt and deficit and coordinating economic, budgetary and financial policies (Eurobarometer, 2010, 2011a, b). Little is known about the Dutch attitudes in later years of the crisis and to our knowledge no studies exist that explore the role of ambivalence in citizens’ beliefs regarding the Eurozone crisis or the Dutch in particular.

5.2. What the Dutch believe: Governmental debt and bureaucracy

To explore what insights can add to these findings by applying CM, the most salient relations from the aggregated CM of the Dutch respondents are visualized in a graph (see Figure 4). In addition, the most salient concepts and relations including the weight (w), value (v) and sign of the relations and the goal-orientation (go) of the concepts helps to explore the content of CMs (see Tables 3 and 4). Analyzing this data leads to some new and surprising findings: While respondents were specifically asked and primed to draw their belief system regarding the Eurozone crisis, the most salient (s) concept in the collective CM is the term “bureaucracy” (s: 412, see Table 3).Footnote ⁵ Ideas that pertain more directly to the crisis like for instance the global financial crisis (s: 322) are much less salient. Bureaucracy is seen mainly as a cause (go: −0,38) and perceived as negative. The connection between bureaucracy with other concepts in the maps reveals why this is the case. First and foremostly, bureaucracy is seen as the main contributor to the Eurozone crisis (negative relation to crisis-free Eurozone, w: 60, v: -30), Moreover, it is perceived to fuel Euroscepsis (w: 36, v:2), has a negative effect on European cooperation (w: 14, v: -12) and diminishes trust in the EU (w: 23, v: -21).Footnote ⁶ Delving into the details also provides insight into why and how our participants feel bureaucracy is connected to the Eurozone crisis: In their minds it dissuades compliance in general and with EU budgetary rules in specific (w: 13, v: -9; w: 12, v: -4) as well as the reform of EMU (w: 7, v: -5). In addition, they feel it reduces the stability of the Eurozone (w: 10, v: -8), diminishes market trust (w: 7, v: -5) and has negative influence on economic development (w: 10, v: -8) and sound governmental finances (w: 7, v: -5).

Figure 4. Excerpt from collective CM of all respondents.

Other key concepts in the CM point to two alternative narratives regarding the crisis. The first narrative revolves around the global financial crisis which is the second most salient concept in the belief system of our respondents and is—as another major and direct cause of the Eurozone crisis—also evaluated negatively (w: 38, v: -24). In the minds of our respondents the global financial crisis also caused a decline in market trust (w: 20, v: -16) and the stability of the Eurozone (w: 16, v: -14) while fueling Euroscepsis (w: 17, v: 5). However, this is not the dominant narrative. Analysis of the other key concepts reveals that they interpret the crisis primarily as a governmental debt crisis. Concepts like “sanctions for member states that do not comply to the EU budgetary rules” (s: 278), “sound governmental finances” (s: 253) and “compliance with EU rules for governmental budgets” (s: 231) are all part of the top ten most salient concepts in the CM and evaluated positively (see Table 2). In fact, all three concepts are seen as major preventors of, or solutions to the Eurozone crisis, with a combined effect that far exceeds that of the global financial crisis or bureaucracy. Moreover, there is little disagreement among the Dutch respondents that enforcing sound budgetary policies would have a positive effect on the Eurozone (see columns 5-6 in Table 3 and Appendix 1). The conclusion that the framing of the Eurozone crisis as a governmental debt is dominant, is further supported by the low saliency of concepts that would point to the interpretation of the Eurozone crisis as a more structural economic crisis like employment (s: 142), economic growth (s: 128), domestic structural reforms (s :62), sustainable social security systems (s: 57) or liberalization (s: 49, see Appendix 1).

Focussing on the perceived consequences of the Eurozone crisis, the most salient effect of a crisis-free Eurozone in the eyes of the Dutch is the somewhat tautological conclusion that it would (re)establish stability with the Eurozone (w: 32, v: 28). Solving the crisis is also seen to increase trust in the EU (w: 31, v: 19) and foster further European cooperation (w: 29, v: 19). These more general effects outweigh the positive economic effects of a crisis-free Eurozone in the mind of the respondents, like strengthening the economy of the EU (w: 23, v: 21), increasing market trust (w: 22, v: 14), improving government finance (w: 22, v: 8), economic growth (w: 18, v: 6) and development (w: 20, v: 14) and employment (w: 10, v: 4). The perceived effects of the crisis, thus focus on structural issues like stability and trust rather than on the economy. In line with this, only a few of our respondents draw a link between the crisis or its consequences and the general interest or their self-interest (s: 52; s: 51).

Finally, it is clear from the state of the art in EU Studies that the Eurozone crisis had a negative effect on peoples’ trust in the EU. This connection also exists in the minds of the Dutch citizens in this study. In their CM the effect of the crisis on EU trust is seen as stronger than on other perceived consequences of the crisis (w: 31, v: 19). Moreover, a similar idea is conveyed through the perceived link between the concepts “successful Economic and Monetary Union” and “stability in the Eurozone” and trust in the EU (w: 8, v: 8; w: 10, v: 6). Other factors that negatively impact trust in the EU are bureaucracy (w: 23, v: −21) and Euroscepsis (w: 15, v: −15). As indicated above, these two factors are mutually reinforcing in the minds of our respondents, fortifying their negative impact on trust. In addition, as bureaucracy is also seen to significantly hinder the Eurozone’s ability to become crisis-free (w: 60, v: −30), it also indirectly limits trust in the EU in the eyes of the Dutch respondents.

The final remaining question is whether the difference in EU trust among the respondents is also linked to differences in their belief systems. When conducting a quick comparison of the maps, several things stand out: Those with low and ambivalent trust in the EU identify the same top ten concepts to which they generally assign the same value. The collective CM of the group with high trust diverges from this and they are the only group for which the concept EU trust falls just outside their top ten of most salient concepts. Moreover, in the view of the respondents with high trust in the EU, the Eurozone crisis only has a weak effect on EU trust (w: 3), weaker than the reverse effect of trust on the crisis (w: 5). Answering this question would require a more in-depth analysis and the development of new ways to facilitate the comparison of CMs in a more in-depth fashion.

Table 2. Top ten most salient concept in the aggregated CM of all respondents

Table 3. Top ten most salient relations in the aggregated CM of all respondents

5.3. The ideology of the masses

The second sub question asks whether CM may help to answer one of the classic questions in the domain of public opinion, the extent to which citizens’ attitudes are actually grounded in an ideologically inspired belief system (Converse, Reference Converse2006; Wilker & Milbrath, Reference Wilker and Milbrath1970). To answer this question the “paradigm support” and “instrument support” of our respondents is are calculated and compared to that of two key members of the Dutch elite: the Dutch prime minister, Mark Rutte, and governor of the Dutch central bank, Klaas Knot (Van Esch & Snellens, Reference Van Esch and Snellens2024).

Table 4. Paradigm support for the different groups of Dutch citizens in comparison to the Dutch political and financial leadership

Table 4 shows the level of paradigm support in the cognitive maps of all our respondents, the different subgroups and the two Dutch leaders. In correspondence to the findings from the narrative analysis, the analysis shows a strong support among our respondents for the Ordoliberal policy paradigm. This support seems independent of their trust in the EU. Interestingly and in sharp contrast to the thesis put forward by Converse, the belief system of our respondents is even more strongly Ordoliberal than that of the Dutch elite: The maps of all subgroups contain about three times more arguments in support of an Ordoliberal approach to the crisis than a Keynesian approach, while about half of the arguments of the leaders correspond to a more Keynesian ideology.

Table 5 shows the level of support for the different policy instruments. In line with their perception that the Eurozone crisis is a governmental debt crisis and corresponding to their Ordoliberal ideology, the participant in our study support stronger EU fiscal regulation over structural and EMU reforms with the exception of the group that do not know whether they trust the EU. They also generally prefer structural over EMU reforms with the exception of the group with ambiguous trust in the EU. However, the major dissent is between the Dutch citizens en their leaders, who both show more support for structural reforms than the Ordoliberal measure of imposing stronger fiscal regulation.

Table 5. Instrument support for the different groups of Dutch citizens in comparison to the Dutch political and financial leadership

5.4. Ambivalence: Beyond to believe and not to believe

Finally the findings of our analysis confirms the conclusions of earlier studies that ambivalence is an integral, and relevant aspect of public opinion that deserves further exploration. Focusing on ambivalence in its uncommon, strict definition reveals that of 50 of our 504 respondents took the liberty to draw at least one pair of conflicting relations (see Appendix 2).Footnote ⁷ In total these citizens drew 72 pairs of conflicting relations. Seventy percent of this subgroup drew only one conflicting pair of relations, three respondents indicated that they felt strictly ambivalent about all the relations in their CM (see Appendix 2). Overall, the share of strictly ambivalent relations for the maps of these 50 respondents ranges from around 11 to 100 percent. These results indicate that even in its most strict definition, ambivalence is a relevant construct when studying public opinion.

The concepts that are most often seen to have conflicting consequences are crisis-free eurozone (10 times), closely followed by bureaucracy (7 times) and global financial crisis (6 times). There is a considerable variety in the conflicting implications these concepts are perceived to have. Bureaucracy, for instance is seen to have both a positive and negative effect on Euroscepsis, crisis-free Eurozone, liberalization, compliance with EU rules for government budgets and the stability of the Eurozone. In almost 85 percent of cases a particular conflicting relation is drawn by only one respondent. The most frequently mentioned strict ambivalent relation is that between bureaucracy and Euroscepsis, whereby three respondents felt bureaucracy both fostered and reduced Euroscepsis. These examples suggest that it would be hard for these respondents to form an attitude or policy preference with regard to the topic of bureaucracy, indicating the broader relevance of studying strictly ambivalent beliefs for political science.

A more commonly used and less strict measure operationalizes ambivalence as holding opposing evaluations towards the same object. In the context of CM this translates in concepts having both positive and negative implications. Almost 31 percent (155) of maps in our set fulfill this condition and identify positive and negative consequences for one or more concepts in their map. This adds up to a total of 309 ambiguous concepts (see Appendix 2). This percentage corresponds quite closely to the level of ambivalence found in survey-based studies of EU public opinion.Footnote ⁸ Our findings, however, confirm the relevance of also assessing the extent to which respondents’ beliefs are ambivalent (cf. Stoeckel, Reference Stoeckel2013), for the level of ambivalence per concept and per map exhibits a high degree of variability. In 67 percent of our cases the ambivalent concept has an equal number of positive and negative consequences, indicating maximum ambivalence (score of 1). In these cases, respondents would again have a hard time asserting a preference or attitude on the subject at hand. In the remaining 33 percent of cases, the respondent’s assessment of the concepts is skewed towards positive or negative, which suggests that they would be able to form a clear opinion on the subject.Footnote ⁹

The concept that is most frequently perceived as fully ambivalent (equal number of positive and negative effects) is crisis-free eurozone (46 times), closely followed by bureaucracy (32 times), compliance with EU rules for government budgets (18 times) and sound government finances (small debt/deficit) (16 times). When reviewing their average ambivalence score, however, none of these concepts make the top ten. Moreover, when we also take into account the cognitive maps without any ambivalent beliefs (ambivalence score 0), liberalization, subsidiarity and the European digital market come out as the concepts respondents are most ambivalent about. Compliance with EU rules for government budgets, bureaucracy and sound government finances (small debt/deficit) closely follow (see Table 6). The concepts innovation and pragmatism are the only concepts seen as non-ambiguous by all respondents.

Table 6. Top ten ambivalent concepts (overall)