EU Accession and Political Trust

(1)

The Limited Impact of Becoming a Member State

Master Thesis Public Administration – Economics & Governance

M.J. Martens

s2634643

Leiden University

Faculty of Governance and Global Affairs, The Hague

9 June 2019

Thesis Supervisor - Dr. M. van Lent

Capstone Quantitative Research

Word Count – 17,084

(2)

1

Abstract

Since 2004, thirteen countries have acceded to the European Union. Scholarly insights prove how newly established democracies get consolidated and politically more powerful once becoming a member state, which raises the question whether citizens will trust their political system more once their country accesses. Using data from the European Social Survey, the relationship between EU accession and political institutional trust is analysed by using a regression discontinuity approach. This strictly distinguishes the more uncertain and subordinate candidate country phase from the phase where a country has actually acceded and is a member state in this powerful intergovernmental organisation. Regressions are performed on an individual country-level and a supranational cross-country level. Results indicate EU effects on political trust do vary between countries, yet a more general, slightly negative, trend is found for trust in institutions. Due to very low R-squared values, it can be concluded EU accession only has minor impacts on political institutional trust. This demonstrates political trust is an abstract concept, influenced by many covariates. Altogether, results should be interpreted with care due to both the low R-squared values and methodological difficulties.

1. Introduction

1.1 An Introduction to Accession Literature

European countries have experienced ever increasing levels of unification after the end of the Second World War. It was not until the early 1990s that many Central- and Eastern European countries (CEECs) became independent, after the fall of communism. These countries had aspirations to apply for European Union (EU) membership, and three major accession moments could be distinguished for these states. In 2004, a total of ten states joined the union, followed by two countries in 2007 and one in 2013 (precise overview given in Table 1, paragraph 2.2.1). Nowadays, nearly all of the newest EU member states are CEECs, with only the Southern European islands Cyprus and Malta as exceptions (ibid.). This process of fraternisation and unification, in Europe, is known as Europeanisation (ibid.; European Social Survey, 2018). Through this process, countries get more similar over time in terms of governance over time, because being part of the EU will affect a member state’s domestic politics. But before a new member state could access, it needs to comply to joint EU legislation known as the Community acquis (Gateva, 2019; Balfour & Stratulat, 2015). Enlargement of the Union, along with certain conditionalities for the countries to join, has always been a vital aspect of pursuing Europeanisation. In academic literature, enlargement of the European Union is even believed to be the most successful of the EU’s foreign policies. Especially the extensive enlargements of 2004 and 2007 are considered to be an overwhelming political success for the EU (Gateva, 2019).

Even today, the quest for a united Europe continues. By the late 2010’s, there are five countries in the process of joining the EU. Albania, Montenegro, North Macedonia, Serbia and Turkey hold the so-called candidate country status, which comes with the requirement to implement the Community

(4)

3

acquis. The EU will monitor this process and occasionally helps candidate countries to adopt the acquis (European Commission, n.d.; Schimmelfennig & Sedelmeier, 2004). However, the road to EU accession can be long, as Turkey’s long-running case illustrates (Hoekman & Togan, 2004). Is seems evident that on shorter term only the Balkan candidate states are truly eligible for full EU membership, as Turkey is recently struggling with democratic backsliding, combined with a growing resistance of several member states against Turkey’s accession (Ker-Lindsay, Armakolas, Balfour & Stratulat, 2017).

The acquis, being the EU’s legislation, has proven itself hard to adopt for countries more distant to the original member states. For the CEECs, major domestic changes had to be made, for instance in the policy domains of freedom, corruption and crime (Balfour et al., 2015). In fact, some of the new member states are still struggling with issues that should have been resolved through implementing the acquis. Examples are large-scale corruption and organised crime in Bulgaria and democratic backsliding in Hungary and Poland (ibid.; Gateva, 2019). Pridham (2000) endorses the difficulties in implementing the Community acquis, as it required making major domestic political changes in the former communist CEECs. However, for the new candidate countries, joining the EU may even be more difficult. Many of the original member states have ambiguous feelings about the two most recent enlargement rounds, on the one hand being a great success and on the other hand resulting in economic and political malaise. Also, the already acceded CEECs now participate in the Council voting process to welcome new member states, and without unanimity no new candidates are allowed. Altogether, this may result in an increasing accession threshold for the current candidate countries (Balfour et al., 2015; Gateva, 2019). Before the large fifth EU enlargement round of 2004, Pridham stated how democracy in the CEECs matures through the EU accession process (2002). After the actual accessions took place, the domestic impacts of joining the EU became measurable, and literature on this topic “has blossomed” and produced numerous valuable insights (Gateva, 2019). In the past years, case studies tend to show a pattern of democratisation in CEECs after EU accession, very much like Pridham’s early model (2002). However, precise shifts in domestic politics are nearly as diverse as the new EU member states are. As the new candidate countries prove, acceding countries all face different challenges that need to be resolved before the EU can be joined. During their term as candidate-country, they will all develop in quite another way but eventually comply to the same acquis benchmarks (European Commission, n.d.; Ker-Lindsay et al., 2017). Also, it is of importance to note that even though patterns of democratisation through EU accession can be distinguished, the significance of partisan politics seems to have declined over the years (e.g. Katsourides, 2003). These circumstances and developments make the precise domestic political changes in the process of joining the Union very hard to estimate, despite the presence of general trends.

1.2 Political Trust and its Relevance

Considering the large body of literature on political impacts of EU accession, this article shifts the focus towards its implications on trust in politics. To find out whether becoming a member state does develop citizens’ trust in their government in a certain way, the central research question is stated as:

“To what extent does joining the EU affect domestic political trust?”

This leaves an extensive scholarly debate about precise causal relationships of EU accession on domestic politics relatively untouched. However, trust in politics is considered a very important topic that heavily affects the domestic political sphere (Hetherington, 1998; Newton, Stolle, Zmerli, 2018). Societal problems often require governmental intervention, and governmental resources are provided by citizens, for instance through taxation. Hetherington (1998) explains how it takes a broad political support for a government in order to function well. Once citizens have to pay costs without experiencing benefits, it takes trust in domestic politics for the non-profiting individuals to still back the government’s policy. With high levels of political trust, individuals are likely to see the societal advantages a policy offers and not think of their government as wasteful in terms of resources (ibid.). Domestic political trust leads to compliance and therefore a more credible and stable state, which translates itself into a more favourable economic position on the global market (Borner, Brunetti & Weder, 1995).

In this article, two important and popular research topics are connected. Surprisingly, the specific link between EU accession and political trust seems, after reviewing a large body of scientific literature, to have never been investigated. Most closely related to this corner of research would be the article by Marien (2011), offering concrete guidelines to attempt a proper analysis of the concept of

(5)

4

political trust through using EU data. Further investigation of the precise impacts of accession are relevant, especially when it concerns an important outcome like political trust. At this moment, several candidate countries are close to joining the Union, while several other CEECs are still considering whether they should apply for EU membership. In this first group, predictions about the evolution of political trust after becoming a full member state could be made after thorough analysis. For the second group, the expected way in which political trust will develop could play a role in determining whether an EU accession application should be submitted.

In the next chapter, two exemplary frameworks of political trust are explained and operationalised for this research. This is followed by an overview of the history of EU enlargement and a framework describing three phases in the accession process. Combining the frameworks of trust and accession, hypotheses are described in section 2.3. The subsequent chapter on methodology describes how a regression discontinuity approach will be used in this quantitative research and what datasets will be analysed, along with conceived difficulties in using this approach. Both descriptive statistics and analysis results are presented in chapter 4, leading to an answer on the research question and reflection on the hypotheses, presented in the conclusion. Addressing limitations and guiding interpretation of the results, a discussion section is added to the concluding chapter.

2. Theory

In this chapter, both the independent variable “EU accession” as well as the dependent variable “political trust” are explained. An overview of academical literature on both topics is given, resulting in a conceptualisation suitable for a quantitative research approach. Due to its complicated nature and the desire to first know what kind of political trust will be measured exactly, the concept of political trust is explained first, in section 2.1. A predictive covariate, directly affecting the level of political trust, is described in paragraph 2.1.3. This is followed by a description and conceptualisation of the main independent variable of EU accession in section 2.2. In the concluding section of this chapter, EU accession is theoretically linked to political trust, thereby formulating hypotheses.

2.1 Political Trust

Political trust is a remarkable academic topic, as it is located on an intersection of political science, sociology and behavioural sciences. Levi and Stoker (2000) define the concept of trust in their article, even though it is a contested term. On an individual level, trust becomes relevant when an individual is exposed to an actor in charge, for instance an institution, that could betray and/or harm this person. Whenever the individual assumes the powerful actor involved will stay true to its promises, most illustratively not betraying citizens but following law and legislation instead, the levels of political trust in this actor are high. On the other hand, whenever individuals assume they cannot rely on the other actor and its promises, the level of political trust is low (ibid.). An addition to this concept is given by Bäck and Kestilä (2009). They link political trust to support for a government. As political trust behaves as a type of government support, a framework described by Easton (1965, 1975) could be applied. Within this framework, two theoretical considerations have to be made in order to classify the type of trust, to make further analysis possible (Marien, 2011). However, measuring the concept of “universal” political trust remains extremely hard, if not impossible. As trust is commonly measured through surveys, scholars often face difficulties in linking specific survey questions to this broad concept. In research aiming to give more practical and concrete recommendations about political trust, based on larger datasets, it is a common practice to reconceptualise the general term using Easton’s framework. This way, rather qualitative questions are operationalised for the quantitative analysis (ibid.).

2.1.1 Political Trust: Building a Framework

Easton is by many considered to be one of the grand theorists of political trust. The general principles of his 1965 framework are still applied nowadays (Bäck & Kestilä, 2009; Marien, 2011; Mattes & Moreno, 2018). Easton conceptualised support for politics, of which trust only is a smaller part. Through the years, several scholars have applied Easton’s concepts of support to political trust. Marien and Hooghe (2011) describe how citizens experiencing low levels of political trust are significantly more likely to accept or even conduct illegal behaviour, which means non-compliance to government rule. Also, when society’s level of trust in the domestic political system is low, effectiveness and legitimacy

(6)

5

of government action are questionable. Implementing law and legislation is harder in the political sphere this way, depicting a strong negative relationship of trust levels with government performance. Altogether, it is widely accepted to apply Easton’s framework of political support specifically for political trust (ibid.; Bäck & Kestilä, 2009; Marien, 2011). Trust, in fact, is a crucial aspect of government support. Without political trust, as Hetherington (1998) points out, politicians are unlikely to be able to operate, as society does not support them, making their rule lack legitimacy. In other words: no trust, no support, no government credibility. With low levels of voluntary compliance and inability to accomplish reforms, the level of trust may diminish further and further. This, once again, reduces the government’s capabilities: a system of negative feedback is established, resulting in an inevitable downward spiral until there is little or no political trust left (ibid.).

The importance of political trust has been acknowledged widely, yet the key question is whether all scholars have analysed the same type of political trust. Marien (2011) explains how political trust is both a broad and abstract or fuzzy concept. A scholarly debate about the measurability of political trust is still ongoing. While some scholars argue trust is measurable on a universal scale (e.g. Levi & Stoker, 2000), others disagree, arguing individual levels of political trust are incomparable. Marien (2011) therefore argues that new research should revisit the past and take into account Easton’s (1965) framework. With Easton’s framework as template, research results become comparable over studies (Marien, 2011). Articles on political trust do vary a lot and do not necessarily always measure the same kind of trust. A stunning amount of inconclusive and even conflicting results could be found, often caused by variance in research approaches and scholarly beliefs (ibid.). Therefore, Easton’s framework (1965), despite being old, is not yet obsolete to provide clarity to the concept of political trust.

Easton classifies political support using two dimensions: the object and the type of support (Easton, 1965, 1975; Norris, 1999). Applied to political trust, the object-dimension divides the contested concept of universal political trust into trust in several political objects. The objects, when conceptualised properly, are widely believed to be measurable, resulting in a scholarly preference to use only a smaller part of the total concept of political trust when researching (Marien, 2011). Individuals could support three aspects of a government or political system. Therefore, political trust is often conceived to be “a political attitude toward an object”, instead of the previously given definition of the assumption of not getting betrayed (Kumlin, Stadelmann-Steffen, Haugsgjerd, 2018, p.386). These three objects are defined as “political communities”, “political regimes” and “political authorities” (Easton, 1965, 1975; Norris, 1999). At the broadest, most basic level, there are the communities. Altogether, these cultural entities, consisting of shared convictions and beliefs, do form a state. They could best be described as foundation of political society, or citizens’ general identification with their nationality. These political communities are, to a large extent, independent from the way government in a country is organized and whomever is residing in office. However, it should be noted not to confuse political communities with nations, as many ethnic or socio-economic entities could coexist within a state, all functioning as communities (Norris, 1999). It are the political regimes giving substance to these communities. Regime is most often perceived to consist of all public institutions in society (ibid.; Marien, 2011). Institutions are mere tools to put the communities’ desires into practice. Examples of domestic institutions are the parliament, political parties, politicians and semi-political public actors such as the police and legal system. These actors, functioning on a national level, often are organised on a local level as well. For example, political parties and police units are present on a municipality level (Easton, 1965, 1975; Norris, 1999). The precise officials these institutions consist of are classified as “political authorities”. Being specific individuals one could trust or not, it is possible for individuals to have high levels of trust in the community or institution, without being confident about the individual politicians (ibid.; Marien, 2011).

The second classification level, as described by Easton (1965, 1975) is a distinction between “specific” and “diffuse” trust, making this a dimension of specificity. Certain political actors, under certain circumstances, could rely on having society’s confidence. Assuming that interaction with government is recurring periodically, trust is built over the years. This way, the political elite could deserve general support as exchange for well-conducted policies. However, the level of trust granted is specific for the political actor that brought the positively perceived policy to the citizens and is therefore called specific trust. Once a new official comes into office, he or she can only earn specific trust; it is not inherited from the predecessor, contrasting to the concept of diffuse trust (Bäck & Kestilä, 2009; Easton, 1965, 1975; Marien, 2011; Norris, 1999). Specific trust could also be applied on a regime- or

(7)

6

Figure 1. Easton’s framework (1965) and localisation of political institutional trust.

community-level, for example to reward a specific institution or socio-political group for shorter-term behaviour that benefitted the individual. This could be seen as a case of “quid pro quo” (Miller, 1971, p.201). Diffuse trust is of the long-lasting kind, to a great extent independent from policy output and the precise actors in charge (ibid.; Bäck & Kestilä, 2009; Easton, 1965, 1975; Marien, 2011; Wilkes & Wu, 2018). It applies to the “fundamental arrangements” of the political system itself (Bäck & Kestilä, 2009, p.173) or, how Wilkes & Wu (2018) explain it, it is a “moral” and “value-driven” orientation towards an object of Easton’s first classification (p.242). Diffuse trust is a more serious form of trust than its specific counterpart; a measure political actors can fall back on when they lack any degree of specific trust (Marien, 2011). In democracies, a certain level of distrust vis-à-vis politicians is desirable, because critical attitudes can encourage politicians to perform better. However, there always is a strong need for the fundamental diffuse trust (ibid.; Bäck & Kestilä, 2009). Specific trust is more likely to be associated on the level of “political authorities”, while diffuse trust is particularly present in “political communities” (Norris, 1999).

In academic literature, there is seemingly more emphasis on the second of Easton’s classifications. While nearly all scholars make a distinction between specific and diffuse trust, the objects of trust are often not properly defined. This is a problem, according to Marien (2011). Often, only trust in one political object is measured, yet the outcome is generalised to a universal concept of political trust. A close approximation to this concept could only be made by analysing conclusions from multiple articles, thereby covering a broad range of trusted communities, regimes and authorities/individuals. In individual articles, especially when using large surveys that have already been conducted, it could be more valuable to redefine the concept of political trust to one of its most applied sub-concepts (ibid.). Trust in political institutions, used as research topic by various scholars (see, e.g., Lühiste, 2006; Marien, 2011; Uslaner, 2018), proves itself the best way to measure levels of political trust quantitatively. As a more concrete type of trust, it is specifically meant to measure the level of trust in the already existing domestic institutions. Changes in this type of trust over time could have direct implications for government (Marien, 2011). Therefore the research question is slightly adjusted and now formulated as: “To what extent does joining the EU affect domestic political institutional trust?” Applying the first dimension of Easton’s framework (1965), trust in political institutions belongs to the object of political regime. As this object typically is defined by diffuse trust and specific trust in a more or less equal degree (Norris, 1999), a respondent’s general level of trust towards specific institutions could be measured (Marien, 2011). However, as many scholars argue, diffuse trust at all times remains the most important component for actors involved. Figure 1 summarizes Easton’s framework and locates trust in political institutions, which will be used as main dependent variable in this research.

Specific Trust

Dimension 1 – “objects”

Political Trust Diffuse Trust Political Communities entity-level Trust in Political Institutions Political Regimes institutional level Political Authorities individual level

Dimension 2 – “specificity”

(8)

7 2.1.2 Institutional Trust: Consolidating the Framework

With Easton’s framework as basic conceptualisation, a deeper theoretical analysis could be made. Norris (1999) made large theoretical contributions to the original framework; revising as well as expanding it. Norris did not revise much in Easton’s model, only renaming political authorities to “political actors”. However, she added to the original framework, by refining the object of political regimes into three parts, resulting in a total of five objects. In practice, lines between the 3- and 5-object models often are blurred. As Norris (1999) points out, respondents of surveys are unlikely to experience very divergent levels of trust for a specific object, and neither are they likely to only experience diffuse or individual trust. Large spillover effects do make political trust and all aspects of it complex and fuzzy concepts of their own (ibid.; Marien, 2011). Following this line of reasoning, it seems logical how the previously mentioned scholarly debate about a universal concept of political trust is still ongoing, despite the preference of many scholars to use Easton’s and Norris’ frameworks to a certain extent. But, as Norris explains, a theoretical distinction within the object of political regimes is justified based on results. Citizens, as respondents of surveys, “seem to distinguish between different levels of the regime” (Norris, 1999, p. 9). Exemplary results are the individuals believing in, trusting and supporting an institution’s values, while rejecting the way the institution works out in practice. As will be described later, this distinction is partly caused by a different balance of diffuse and specific trust within the objects.

Within Easton’s concept of political regime, Norris distinguishes the objects of “regime principles”, “regime performance” and “regime institutions”. These three classifications, originally meant for government support, could each be applied to trust in the same way as Easton’s framework. The most fundamental object in Norris’ regime-distinction are the regime principles. These can be perceived as the very basic values and principles in a political system. This object is comparable to the object of political communities, but specified to the context of a democratic system, thereby giving political substance to the communities. This object is very abstract and therefore not very popular in research on political trust. However, trust in regime principles sets the bottom-line for the more practical object of regime performance. This object is hard for respondents to grasp, but possible to measure because it could be defined as the support for the political system functioning in practice, without involving the precise trust in institutions (Norris, 1999; Marien, 2011). This makes regime performance seemingly function as trust in the way democracy (or another political system) works out in a country. Therefore, the concept of trust in regime performance could be linked to satisfaction. In most academic studies, satisfaction with the way democracy works in the respondent’s country is used as a way to measure this object (Marien, 2011). The third distinguished object within Easton’s concept of political regime, regime institutions, is the main and most popular component of political institutional trust (ibid., Norris, 1999). This object indicates society’s support for parliament, political parties, politicians, police and legal system as well as all other domestic public institutions, very much similar to the object of political regime as initially defined by Easton (Norris, 1999). As Norris points out, including more institutions gives a better approximation of trust in regime institutions. Examples of additional organisations that could be included are executive institutions, bureaucracy and the military (ibid., p.11). There are practical limitations, though, as no large survey measures trust in every single institution available, making research on regime institutions use only the trust in national institutions included in the survey data (Marien, 2011). As Marien states, “in new democracies, […] institutional trust influences representative and implementing institutions to the same extent” (p.38). This implies that trust in regime institutions does not vary much in the relatively newer democracies, like the Central and Eastern European countries. Therefore, measuring only fewer institutions could be legitimised for this study.

Yet, it is not plainly justifiable to measure trusts in various institutions and treat them like if they were one. Whereas the true political actors, like the government or politicians, could easily adjust to or create new circumstances due to their partisan nature, semi- or non-political government-related actors are known to be more rigid. The legal system and police, for example, do what they are trained to do based on domestic policies. This bureaucratic approach does not involve sympathising for certain partisan preferences. Government could introduce policies based on political convictions, but the implementing institutions will not be able to live up to them once they are not capable and motivated to do so (Olsen, 2009). This leads to the distinction between political and semi-political actors. Political trust tends to be higher for this second group of institutions, as they lack partisan politics, therefore being more accepted by society as a whole (Marien, 2011).

(9)

8

Figure 2. Norris’ framework (1999) and the dependent variables.

Norris (1999) emphasizes how many studies try to measure diffuse trust, while, like explained previously, this is hard or even impossible due to the fuzzy border between specific and diffuse trust. Trust in the object of regime institutions, in many surveys, is appointed as confidence in an institution; a term Norris uses herself. Even though trust and confidence are not necessarily the exact same thing, trust is the direct way to measure citizens’ confidence (Tolbert & Mossberger, 2006). In the survey used by Tolbert & Mossberger, respondents are even asked “how much trust and confidence” they have in a certain institution, resulting in one answer. Therefore, it implies that the concepts of trust and confidence are so closely related that they could be used alongside each other or used interchangeably (ibid.).

Norris’ additions to Easton’s framework are used to further adjust the scope of this article and define what type of trust will be analysed. This way, the abstract concept of political trust has been specified to an extent where opportunities for quantitative research emerge. After all, without clarity of which concept to measure, quantitative analysis would mean comparing “apples to oranges” (Marien, 2011, p.14). In most research, both satisfaction with democracy and trust/confidence in institutions are used. This corresponds to Norris’ objects of regime performance and regime institutions. Measuring both objects has advantages, as the object of regime performance is believed to be slightly more diffuse and regime institutions more specific. Through measuring the level of satisfaction with the way democracy works out in a country, an actual measurement within the universal concept of political trust is made, but being more distinct from the officials residing in office (Marien, 2011). It is still not possible to disentangle specific from diffuse trust, yet comparing both objects’ findings could sometimes lead to new insights, in their turn leading to new research opportunities. In many European studies on political trust, scholars focus only on the object of regime performance, falsely claiming it to cover the full meaning of political institutional trust (ibid.). Measuring multiple objects within the broader concept of political trust means the actual concept is approximated more closely, yet trust in distinct objects could contradict each other, resulting in inconclusiveness. Altogether, Marien (2011) concludes how analysing both the object of regime performance and regime institutions is concrete enough for a reliable quantitative approach, with institutional trust representing society’s perception of their political system’s performance, which in its turn practically is useful. After all, “institutions play an important role in shaping a democratic society” (ibid., p.16). Norris’ objects, along with the precise objects that would be used as dependent variables in this article are summarised in Figure 2.

Specific Trust

Dimension 1 – “objects”

Political Trust Diffuse Trust Political Communities entity-level

“Satisfaction with the way democracy works out”

Not measured

Dimension 2 – “specificity”

1. Parliament 2. Political parties 3. Politicians 4. Police 5. Legal system Regime Principles institutional values “Trust/confidence in specific institution(s)” Regime Performance institutional output Regime Institutions institutional design Political Actors individual level

(10)

9 2.1.3 Education as Covariate

In this article, individual findings are generalised to a societal country-level, and for a second analysis the same individual results are used for cross-country comparison. There are difficulties involved in doing this, as the data sampled on individual level may be unrepresentative for domestic society as entity, let alone cross-country analysis (Angrist & Piscke, 2014). This will further be elaborated in paragraph 3.2.1 of the methodology chapter. For larger-scale studies, is important to pay attention to existing covariates that are thought to affect the level of political institutional trust. A predictive covariate could be used to estimate the outcome of an analysis. By comparing the estimation to the actual data, the true impact of the independent variable could be approximated.

A predictive covariate is a characteristic with a strong heterogenous effect on the dependent variable(s), in this case the two objects of political institutional trust. Hooghe, Marien and De Vroome (2012) find that a respondent’s education level does accurately predict what trust levels he or she will have in politics. Attaining a higher level of education leads to increased cognitive abilities, which in its turn positively affects one’s political trust. In the research, political trust consisted of a slightly different four-item scale, yet it aimed to measure institutional political trust. Combined with the fact that trust in various institutions in newly consolidated democracies tends to be rather stable, education proves itself a good estimator for political institutional trust. In this research, education will therefore be used as independent covariate directly affecting political institutional trust, equally in the object of regime institutions and object of regime performance.

However, it should be noted there are many more covariates affecting political institutional trust. Some of these are basic respondent characteristics as well, such as gender (Hinojosa, Fridkin & Kittilson, 2017) and age (Mishler & Rose, 2001). Due to this article’s methodology, using a regression discontinuity approach, it probably is not necessary to control for more than one covariate. Yet it is of importance to notice the causal relationship between education level and the level of political trust is certainly not straightforward but, again, rather fuzzy instead.

2.2 EU Accession

In contrast to the complex concept of political trust, EU accession is a fairly easy concept to comprehend and measure. Accession to the European Union is concrete, but multiple phases in the process of joining the Union can be distinguished. This section will address the EU’s history of enlargement rounds, focussing on countries acceded in the the fifth and sixth round as well as describe and therefore conceptualise the process of joining the Union.

2.2.1 A Brief History of EU Enlargement

According to Gateva (2019), expansion has always been important to the European Union. Recovering from the Second World War, the first attempt of unification dates back to 1952, when the European Coal and Steel Community (ECSC) was founded. Europe was divided in two main areas, the capitalist West and communist East. Six western countries signed the Treaty of Paris: Belgium, France, Luxembourg, Italy, the Netherlands and West Germany. The cooperation was intended to share the countries’ coal and steel production, which would result in mutual economic advantages (Dedman, 2010). Nevertheless, coal and steel had been key resources in both World Wars. So, from a political point of view, the ECSC was a tool to assure all member states that none of the other participants would be able to secretly start a war (ibid.). This twofold path of economic and political incentives to join has been maintained from the very start. Also, there always has been a possibility for new member states to join. However, much like how the EU’s accession is still arranged nowadays, all member states needed to unanimously agree on another country’s membership to the Community (ibid., Gateva, 2019).

By signing the Treaty of Rome in 1957, the ECSC kept existing, yet cooperation between the states was expanded to many new sectors. Through establishing the European Economic Community (EEC), a customs union was established, further increasing economic collaboration. Later on, this did lead to the creation of a European single market. Meanwhile, still characterised by political intentions to avert a new World War, the newly founded European Atomic Energy Community (EURATOM) functioned as a tool to be sure none of the member states developed nuclear weaponry (ibid.). The initial six countries were now being referred to as member states of the “European Communities”. Even though it was possible for other countries to join the initial six, the vetoing structure became a serious barrier of entry. Most member states seemed eager to welcome Denmark, Ireland and the United Kingdom to

(11)

10

the Communities. These applicants were members of the newly founded counterpart of the EEC: “the European Free Trade Association” (EFTA). However, France that kept vetoing the UK’s accession, as president De Gaulle feared the British connections to the United States. (ibid.). It was not until 1973 that the first enlargement round took place. After more than a decade of negotiations, Denmark, Ireland and the United Kingdom were allowed to join the Communities. The main reason for this breakthrough was the French president De Gaulle leaving office, thereby clearly illustrating how important political relations have been for joining the EU (ibid.). During the 80’s and 90’s, more Western- and Southern European countries were welcomed to the Communities. Without facing major trade barriers for most goods, member states traded extensively with each other (ibid.). Politically, the Communities were known to strive for further integration. The political potential of the European Communities was clearly visible in countries in Portugal and Spain, which had been ruled by dictators for a long time. Regional policies, organised through the Communities, invested large sums of money in poor and lagging regions and helped democratising the new member states (ibid.). As a rule of thumb the more neutral states, less interested in unification, joined the EFTA. But, this offered a smaller customs union than the Communities. Due to facing a comparative disadvantage, several EFTA member states decided to change sides in the mid-1990s and acceded in the fourth enlargement round (ibid.). Their accession was arranged relatively quick, as there were ambitious expansion plans ahead.

With political victories of non-communist parties in Eastern Europe during the late 1980s, the demolishment of the Berlin wall in 1989 and the fall of the Soviet Union in 1991, the Communities had decided to enlarge eastwards. The member states signed the Treaty of Maastricht, which came into effect in 1993, initiating a whole new era of European unification (ibid.). It was not only out of solidarity that the eastward expansion was initiated: large political and economic interests are thought to have played a role. West- and East-Germany now were united again, which was perceived to be a dangerous thing by France. The old political pattern of the ECSC and European Communities, to prevent conflict amongst members, emerged once more. Many states were afraid a united Germany would “translate economic and financial power into political dominance” (ibid., p.105). The path leading to further European unification involved incorporating the Central and Eastern European Countries (CEECs). The Communities, through signing the Maastricht Treaty, turned into the European Union. This provided a vision and schedule for further European integration. The EU worked towards having its own currency, coordinating foreign and security policies and cooperating in combatting crime and terrorism. The Community acquis, mandatory for candidate countries to implement, included these matters and thereby set the benchmarks for new countries to join (ibid.; Kesner-Škreb, 2008; Gateva, 2019).

For the CEECs, it was not particularly easy to make the choice to apply for a candidate country status. Many of the citizens had experienced the Soviet Union or their own communist government as oppressors, resulting in a fear for “Euro-federalism” (Dedman, 2010, p.161). However, the Union offered good economic prospects, so membership was eventually seen as “a ticket to economic stability”, which could in its turn lead to more welfare (ibid., p.117). Therefore, governments of CEECs massively applied for EU membership between 1994 and 1996. The candidate country status was granted to every CEEC that did apply, without the citizens explicitly having to be in favour of joining the Union. This made the decision to join the Union, in many cases, a rather authoritarian decision. It was not until 2003, when EU membership was close to being realised for most of the CEECs, that referenda were organised to legitimise the decision to join the EU. Every CEEC accession referendum turned out in favour of accession, as the fear for Euro-federalism had depleted during the ‘90s (ibid.; Szczerbiak & Taggart, 2004). In 2004, the Community acquis had successfully been implemented by ten candidate countries. With EU support, major (political) progression was made in domains like freedom, jurisdiction, corruption and crime (Balfour et al., 2015). This was not easy to accomplish for many of the CEECs. Bulgaria and Romania, for example, failed to establish a well-functioning judicial system and did not reach the thresholds of fighting corruption and crime before 2004, which made the fifth enlargement round split into two parts (Commission of the European Communities, 2004; Gateva, 2019). It should also be noted Bulgaria and Romania were relatively poor countries, only having a GDP per capita of approximately 30% of the EU’s average. This is thought to have played an important role in their delayed accession as well (Dedman, 2010). With the fifth round being the largest expansion in EU history, the Union saw the number of member states increasing from 15 in 1995 to 27 in 2007. The sixth enlargement round, in 2013, added Croatia to this pool of countries. An overview of the new member states and date of accession is given in Table one for the last two EU enlargement rounds.

(12)

11 Table 1. EU enlargement rounds 5 & 6

Round Date of entry New member states Total

5.1 01/05/2004 Cyprus Lithuania Czechia Malta Estonia Poland Hungary Slovakia Latvia Slovenia 12 5.2 01/01/2007 Bulgaria Romania 6 01/07/2013 Croatia 1

Note: Enlargement round 5 was divided into two parts. Source: European Union, 2020

Somewhat unsurprisingly, the colossal fifth enlargement round caused “digestion problems” for the European Union (Balfour et al., 2015, p.1). France and Germany, the largest nations in the cooperation, lost part of their dominance to new member states. The EU enlargement policy is considered to be the EU’s most successful foreign policy, yet it has a “dual nature”, meaning that the EU affects new member states, but new member states also affect the EU and its policies (Gateva, 2019, p.2). At this moment, only the Balkan states seem to be able to join the EU on a relatively short term. Yet, complying to the Community acquis now plays a more important role than ever (Balfour et al., 2015). Even after having acceded, some of the CEECs have experienced backsliding to a political situation under the acquis benchmarks, like large-scale corruption and organised crime in Bulgaria and democratic backsliding in Hungary and Poland (ibid.; Gateva, 2019).

2.2.2 The Advantages of Acceding

Acceding to the EU, in the last two accession waves, has proven to be a hard process for some countries. It can be costly to implement the Community acquis, which raises the question what the effects of being a member state are and whether this results in net benefits. Schimmelfennig and Sedelmeier (2004) describe the “model of external incentives”, which could be used to explain how every acceding country is likely to experience a net benefit of joining the EU. Combined with Gateva’s concept of the dual nature of accession (2019, p.2), political implications for countries could, broadly, be described.

In the fifth enlargement round, it is important to distinguish the group of Central and Eastern European Countries (CEECs) from the Mediterranean islands Cyprus and Malta, as the political background of these countries differs significantly. Cyprus and Malta used to be crown colonies of the United Kingdom, until their independence in the 1960s. Partly due to their former ties to Western Europe, the countries did not sympathise to communist ideas during the Cold War and therefore were moderate capitalist states (Baldacchino, 2002; Sakkas & Zhukova, 2013). The first advancements towards EU membership in these countries were already made during the 1970s, and the islands are known to have always had a positive attitude vis-à-vis the Union (Dedman, 2010). In contrast to this, the CEECs were ruled by communist parties until the end of the Cold War and only got involved into (West-)European matters in the early 1990s. Oftentimes, the EU is thought to have been a last resort for these countries, as the fall of communism had rapidly caused political distress and insecurity (ibid.) According to Schimmelfennig and Sedelmeier (2004), states are rational actors, always striving for maximum benefits in the external incentives model. Under their pre-EU domestic political rule, the two scholars assume society’s needs are fulfilled, resulting in a “domestic equilibrium” (p.664). However, in the CEECs this balance was disrupted during the early 1990s, resulting in a quick decision of most governments to familiarise with an actor that was richer and more organised (Dedman, 2010). Still, by applying for EU membership the structure could remain unbalanced, as the EU maintains strict conditionalities to join the Union. These conditionalities are not necessarily aligned with the country’s domestic values, preferences and needs. Therefore, it would be irrational to join the EU voluntarily, so net advantages should be expected by the countries considering to accede. After all, partially distributing domestic political power to a foreign entity would not make sense without receiving benefits (ibid.).

(13)

12

According to the external incentives model, the European Union used multiple benefits to convince and reward countries. Roughly, three steps in the accession process do function as milestones for countries willing to join the EU. Firstly, a country has to associate itself with the EU. Usually, this is organised through signing an Association Agreement (ibid.). A variety of agreements is available, but every agreement in fact is a tool to tie a country to the European Union in some way. This does not mean the agreements exploit countries, as they offer mutual advantages. Most desired by many of the non-EU states are free trade agreements, partly or even fully giving access to the European single market (ibid.). Being allowed to join the European Economic Area (EEA) is conceived to be superior, as this makes the country an inherent part of the European single market. Such an arrangement was made with EFTA countries during the 1990s. The EEA is the world’s largest free trade area, so major economic opportunities emerge for new members (Dedman, 2010). Cyprus and Malta already received Association Agreements during the 1970s, but in the CEECs the much more rapid political shift partly made signing the Association Agreements and second phase in the process coincide.

This second phase is referred to as the candidate country phase (ibid.; Schimmelfennig et al., 2004). The CEECs were relatively easily allowed to enter this phase, as the European Council had decided so at the Maastricht Treaty summit (Dedman, 2010). For Cyprus and Malta, getting this status was more challenging as they were not allowed to become candidates until the enlargement strategy for the fifth round was complete (ibid.). With many of the CEECs anxious to become part of an entity again, it was not necessarily intrinsic motivation that played a role in their decision to apply for EU membership. But, following Schimmelfennig and Sedelmeier’s (2004) external incentives model, several external factors make potential member states see EU accession as something positive or beneficial. As mentioned previously, CEEC governments familiarised with the EU in a rather authoritative way (Dedman, 2010). This allows the government to closely analyse external incentives, without general opinion of the citizens being a major obstacle. In consolidated democracies, such an authoritative process would be a serious problem, but the CEECs were used to communist rule with a very dominant state, making the authoritative process of joining logical for the early 1990s (ibid.). Assuming that governments are rational actors, four factors could be distinguished in the external incentives model (Schimmelfennig et al., 2004). Firstly, assuming that the EU could provide economic and political rewards to a candidate country, as well as to actual member states, the expected rewards greatly define a country’s attitude vis-à-vis the Union. Two dimensions could be distinguished, namely the size of the rewards as well as the term before the rewards will be handed out to a country (ibid.). It is apparent how this factor played a very important role in the decision of the CEECs to accede to the EU. A favourable Association Agreement, giving access to the European Economic Area (EEA) was not signed by the EU until the countries agreed to become candidate countries, thereby giving the EU certainty that the CEECs would really join the Union. This way, CEECs perceived a large economic reward for temporarily sacrificing a part of their autonomy, as the Community acquis had to be implemented (ibid.; Dedman, 2010). The second and third factors refer to this acquis conditionality. The Community acquis provides clarity to countries, thereby encouraging them to comply to a set of rules; the benchmarks to join. EU officials regularly provided help during the implementation process, resulting in a lower threshold to join (Schimmelfennig et al., 2004). The second factor within the dimension of conditionality is the credibility of the EU to reject giving rewards when countries do not comply to the Community acquis. This clearly became visible by the delayed accession of Bulgaria and Romania. As fourth and last factor, the total adoption costs of implementing the acquis are of importance. Rational countries take into consideration both costs and benefits of EU membership, as it is mainly a net advantage of accession potential member states are looking for (ibid.). For the implementation costs, the pre-accession situation in a country plays an important role. The Community acquis, during the fifth enlargement round, consisted of 31 chapters. Together, these chapters cover “the whole of the legal system of the EU” (Kesner-Škreb, 2008, p.405), making the total implementation costs higher for countries that are less similar to EU member states. Candidate countries must accept the acquis as a whole and implement it as scheduled in the accession negotiations. In this process, the candidate countries are helped and supervised by EU officials. To help reduce the costs of EU accession, the Union provides funds to countries to contribute to the costs of implementing the acquis, thereby facilitating accession as well as making it more tempting to join the EU (European Council, n.d.; Kesner-Škreb, 2008; Schimmelfennig et al., 2004).

(14)

13

00

Figure 3. Conceptual model of EU accession.

The final, third, step in the progress of EU accession only is possible once the acquis is implemented. In this actual accession, a candidate country becomes a member state, which has major political implications. To start with, it should be noted every new member state except Cyprus organised a referendum to allow their citizens to make the final decision to join the Union. This occurrence is remarkable, considering it is not mandatory for a candidate country to organise a referendum at all (Dedman, 2010). According to Pridham, the candidate country phase is a “transition to democracy” and accession clearly indicates a shift to “consolidation of democracy” (2002, p.955). Therefore, for society, the precise moment of accession is a strong signal in CEECs that reliable democracy has been established, something that may have an impact on political trust. Secondly, the EU membership phase distinguishes itself from the candidate country phase through a mechanism what Gateva (2019) briefly appoints as “dual nature-effects”. Upon being a full EU member, supervision over the new member state declines, giving the country more political power and freedom. The country now is assured of a position in the EU, leading to a more certainty about having a favourable position in global politics. For some countries, like Hungary and Poland, the newly acquired position did lead to democratic backsliding, while in others it consolidated democracy (ibid.). However, institutions are believed to become more established through EU accession in every country, as a result of more extensive European collaboration and the increased political power. Collaborating in a supranational network like the EU causes the domestic institutions to interact with European institutions as equals, thereby triggering a learning process for both the domestic and EU institutions (Dedman, 2010; Gateva, 2019; Pridham, 2002; Zito & Schout, 2009). As for political institutional trust, it seems likely the actual accession would be a strong statement, concluding the more uncertain phase as candidate country and, at last, being an equal to the other EU member states, functioning in a powerful intergovernmental organisation. Therefore, it is assumed the new political situation will similarly show strong impacts of accession on political trust.

Altogether, the accession timeline is tripartite. Through an Association Agreement, non-EU countries could familiarise with the Union. For candidate countries, the external incentives model provides an explanation why they decided to apply for membership. Combined with Gateva’s “dual nature”-mechanism, it is apparent how full EU membership is thought to be the most rewarding phase, at least in political respect, for countries that joined the EU in the latest accession waves. This framework will be used in the next section to construct research hypotheses. A summary of EU accession is displayed in figure 3. In this figure, all democratic developments are related to the concept of regime performance, as elaborated in paragraph 2.1.2. The institutional developments, often also closely pertaining to democracy, are incorporated under the concept of regime institutions.

1. Association Agreement

Trade Agreement

3. Member State

2. Candidate State

EEA EEA Regime Performance ▪ Transition to democracy ▪ (for CEECs) Regime Institutions ▪ Community acquis ▪ Supervision & help

Regime Performance ▪ Democratic consolidation OR ▪ Democratic backsliding Regime Institutions ▪ Dual nature ▪ Political Power ▪ Institutional Consolidation Pre-EU situation “Rewards” - Size - Speed Conditionality - Clarity - Credibility Compliance to acquis Economic Political

(15)

14

2.3 Hypotheses

In this article, the phase as a candidate country is compared to the phase where the country is an actual EU member state. Therefore, the theoretical differences between those two phases are particularly important. Politically, acceding states move from a subordinary position to a situation where they have a vote in a very influential intergovernmental organisation, which is likely to translate itself into effects on citizens’ political trust levels. The fifth enlargement round shows how economic advantages were gained in a very early stage in the accession, indicating that being an actual member state instead of a candidate country did result in few economic changes.

When it comes to regime performance, measured as satisfaction with the way democracy works out, two scenarios are possible for countries. Firstly, countries are likely to get more democratically consolidated through EU accession. Pridham (2002) suggests that operating in “transnational networks”, as well as having successfully implemented the acquis, “is likely to have a reinforcing effect on democracy” (p.970). On the other hand, as pointed out by Gateva (2019), there has been democratic backsliding in two of the newly acceded member states. Altogether, the political impact of EU accession is highly context dependent and “questions could be raised about the limits of the EU’s transformative power” (ibid.). In short, based on the theoretical inconclusiveness, being a member state will not necessarily consolidate a country’s democracy. This could mean a significant causality is found on an individual country analysis level, measuring inhabitants of a country, but that a cross-country comparison will yield in not finding a significant causal impact of accession on trust in regime performance

Hypothesis 1A: On country-level, EU accession’s effect on trust in “regime performance” varies per country, yielding inconclusive results. (H1A)

Hypothesis 1B: On cross-country level, there is no significant impact of EU accession on trust in “regime performance”. (H1B)

As hypothesis 1B is a null hypothesis on its own, finding a significant causal relationship between accession and trust in regime performance will reject this hypothesis.

For regime institutions, quite another hypothesis emerges. Even though there is a dual nature of EU accession, both pathways lead to more extensive cooperation and/or institutional learning processes. Therefore it is likely that political institutions will consolidate through becoming a member state, resulting in better performance and higher levels of trust. As regressions will be conducted on two levels, the second hypothesis is split in two parts as well.

Hypothesis 2A: On individual/country-level, EU accession does positively affect trust in “regime institutions” in all countries. (H2)

Hypothesis 2B: On cross-country level, EU accession positively affects trust in “regime institutions”. (H2)

These hypotheses (2A and 2B) are rejected once no significant causality is proven on a cross-country level or whenever a significant negative effect of accession is found. Altogether, the findings will be used to attempt drawing conclusions about the total concept of institutional political trust. Based on the literature, an informal third hypothesis, about the total causal relationship of EU accession and institutional political trust, could be developed. Namely, the total impact of joining the EU on domestic political trust is likely to be small/limited but positive. This informal hypothesis states an expectation for the main research question.

3. Methodology

In this research, a multilevel regression discontinuity analysis will be conducted. Multilevel refers to the presence of both an individual country-level and a cross-country level analysis. However, on both levels of regression, individual observations will be used as units of analysis. Section 3.1 describes the datasets that will be used in this quantitative research and section 3.2 gives in-depth information about the way precise analysis will be conducted in the statistical software program Stata.

(16)

15

3.1 Data

Prominent quantitative datasets such as the World Values Survey (WVS) and European Social Survey (ESS) include questions that could be used to measure respondents’ political trust (for an overview, see Appendix A; European Social Survey, 2018; Inglehart et al., 2014), as well as various survey and respondent characteristics. Both the WVS and ESS offer data on some of the thirteen newest member states, yet the ESS is more consistent in including the new member states in its survey rounds. Also, the WVS lacks consistency in measuring respondents’ trust in regime performance throughout the survey waves, making it inferior to the ESS for this specific research (see appendices B and C). In this section, paragraph 3.1.1 operationalises the dependent variables, while paragraph 3.1.2 provides insights in the independent variables. All variables used in this research are summarized in Appendix E.

3.1.1 Dependent Variables

The European Social Survey’s questions about trust in various institutions do measure trust in regime institutions, while the question regarding satisfaction with the way democracy works out measures trust in regime performance (Marien, 2011). Both aspects, as explained in Norris’ theoretical framework of paragraph 2.1.2, are part of the broader concept of political institutional trust and when combined are a good attempt to measure this broader concept (ibid.; Norris, 1999).

Marien (2011) offers a concrete approach to measure trust in regime institutions by using the ESS. Like originally stated by Norris (1999), the best way to measure trust in this object is to combine trust in every public institution. However, insights provided by Marien do contradict this statement for relatively new democracies, as trust in domestic institutions turns out to be, to a large extent, independent of the exact institution (2011, p.25). By combining trust in domestic institutions as measured in the ESS, two distinct scales are made, namely trust in domestic political institutions and trust in domestic semi-political institutions. Trust in semi-semi-political institutions tends to be higher and more stable, according to Marien (2011), leading to the decision to distinguish political and less political institutions in this research. Based on the ESS questions, the first scale would include three items, namely “trust in parliament”, “trust in political parties” and “trust in politicians” (Appendix B). However, the first ESS round of 2002 did not yet measure trust in political parties. As this would lead to incomparable scales for 2002 and the other years, trust in political parties is excluded from the scale, making it consist of only two items (Marien, 2011). The code 𝑝𝑜𝑙𝑖𝑛𝑠 is assigned to this first scale, as overviewed in Appendix E. Just like the first scale, the scale about semi-political institutions, 𝑠𝑝𝑙𝑖𝑛𝑠, includes two items as well. “Trust in the legal system” and “trust in the police” both have the same weight and can therefore easily be rearranged into an 11-point ordinal scale, with value 0 being no trust at all and value 10 being complete trust. The same rescaling applies for 𝑝𝑜𝑙𝑖𝑛𝑠, to maintain an 11-point scale for all dependent variables (European Social Survey, 2018).

Trust in regime performance is measured through the ESS question “and on the whole, how satisfied are you with the way democracy works in your country?”. The explicit addition of the prefix “on the whole” implies that this question is meant to measure a very general attitude vis-à-vis the regime. Just like all items and scales about trust in regime institutions, outcomes are measured on an 11-point ordinal scale. For satisfaction with democracy, value 0 represents “extremely dissatisfied” and value 10 “extremely satisfied”. Trust in regime performance is a direct ESS value coded as 𝑠𝑡𝑓𝑑𝑒𝑚 and in this research recoded as regime performance, abbreviated 𝑝𝑒𝑟𝑓.

For both trust in institutions and regime performance, respondents have the possibility to refuse answering the question or to answer “don’t know”. As this results in missing values, unsuitable for analysis, the missing values are excluded from each corresponding regression. In Stata, this happens automatically, so no further steps should be taken into account in order to successfully exclude the missing values. As the two scales used both are dependent on outcomes for two institutions, slightly more data is excluded in the scales compared to trust in individual institutions.

3.1.2 Independent Variables

In the European Social Survey, data for twelve out of the thirteen newest member states of the fifth and sixth EU enlargement waves are included. Malta, also being the smallest EU member state, has not been included in any of the nine ESS rounds. These rounds, with the first wave of surveys conducted in 2002, are held biannually throughout Europe. However, the survey is not conducted in all countries every year, sometimes resulting in gaps in the data or the unavailability of data right before and/or after the

(17)

16

accession. This could be a problem because a comparison has to be made between the candidate country phase and member state phase. An overview of the precise participation for each ESS round is displayed in figure 4.

It is of importance to realise that ESS surveys are not all conducted simultaneously and neither is the conduction process spread over the entire span of two years. Instead, respondents are interviewed face-to-face over a timespan of approximately two months. The exact moment of measurement varies per country. The precise time of both the start and end of the interview are listed as descriptive variables in the dataset, which makes it possible to combine the date of the survey with the time the interview was conducted and use the result as a running variable, which is a fundamental necessity for a regression discontinuity approach. As respondents are questioned about the dependent variables relatively early in the survey, the most logical option is to use the starting time of the survey interview as running variable. This way, a timeline is created, which is useful for later analysis using the regression discontinuity approach. The running variable is coded as 𝑑𝑎𝑡𝑒, as shown in Appendix E.

Country ESS1 2002 ESS2 2004 ESS3 2006 ESS4 2008 ESS5 2010 ESS6 2012 ESS7 2014 ESS8 2016 ESS9 2018 Bulgaria Croatia planned Cyprus Czechia Estonia Hungary

Latvia planned planned

Lithuania planned Malta Poland Romania Slovakia planned Slovenia

Figure 4. Overview of European Social Survey participation for new EU member states.

Note: The blue line separates pre-EU and candidate country phases. The red line separates candidate country and member state

phases. Green (dark & light) cells indicate that the country is included in the given ESS round. White cells indicate that the country is not included in the given ESS round.

The red line as shown in Figure 4 indicates a candidate country’s accession to the EU, which is the main independent variable used in this research. With the running variable defined, the date and time of the interview can be compared to the date and time that the respondent’s country acceded as a member state to the EU. This way, a dummy variable indicating whether a country was an EU member state when the face-to-face survey started is added. This dummy, called 𝑒𝑢, is labelled 0 for the country not being a member state and 1 for the country having acceded.

For the analysis, the covariate highest achieved education level plays an important role as control variable, as briefly explained in paragraph 2.1.3. The ESS provides data on the respondents’ highest achieved educational level, formulated as the question: “what is the highest level of education you have achieved?” (European Social Survey, 2018). In every round, this results in various outcomes, varying between “not completed primary education” and “completed second stage of tertiary”. The ESS has recoded this into the variable 𝑒𝑖𝑠𝑐𝑒𝑑, as ISCED is the International Standard Classification of Education, therefore facilitating cross-country comparison (European Social Survey, 2018). This research recodes 𝑒𝑖𝑠𝑐𝑒𝑑 into the five-level ordinal categorial variable 𝑒𝑑𝑢. Practically, this means missing values as well as non-usable values will be excluded from the research. An overview of this recoding is given in Appendix D. Practically, missing values for 𝑒𝑑𝑢 will result in exclusion of data that would have been useful otherwise. Therefore, section 4.1 of the results-chapter will briefly overview the number of valid observations for all variables used in this research.

EU Accession and Political Trust - The Limited Impact of Becoming a Member State_x000D_