What’s the difference?
A study to what extent knowledgeable
citizens are better able to distinct between
national and European Union politics in
forming their opinion
Thesis submitted in partial fullfillment of the requirements for award of the degree of Master of Science in the field of Public Administration
Master Thesis
Sjoerd Aelen
Name: Sjoerd Aelen
Student number: S1440128 Supervisor: Dr. D.D. Toshkov
Faculty: Governance and Global Affairs
Programme: Master Public Administration, Economics and Governance What’s the difference?
A study to what extent knowledgeable citizens are better able to distinct between national and European Union politics in forming their opinion
Abstract
This thesis is written to come to a better understanding of the framework for trust in the European Union. It uses data from the European Social Survey, round 9 (2018). Trust in the national parliament is considered the most important predictor for trust in the European Parliament. This research specifically looks at the moderating effect of “Cognitive Mobilization” (CM) within this framework. To reflect CM, we use a person’s level of education, interest in politics and consumption of news on politics and current affairs.
Our assumption is that the more knowledgeable a person is, the better he or she is able to distinguish between national and European Union politics when forming an opinion. We measure this using CM as moderator within the relationship between trust in the national parliament and trust in the European parliament. We expect that the correctional effect of CM is more intense for people that are higher educated, interested in politics and follow politics by watching, reading and listening to the news.
We find empirical evidence that CM indeed moderates this relationship. Interest in politics is found to be the most important factor. Consumption of news has no effect at all. We also find that the effect of education is different between member states: the highest educated in relative corrupt societies are more likely to trust the European Parliament than their counterparts in relative “clean” countries.
Index
Chapter 1: Introduction ... 5
Chapter 2: Literature review and hypothesis development ... 7
2.1 What is institutional trust? ... 7
2.2 Institutional trust within the multi-level structure of the European Union ... 9
2.3 The relationship between trust in the national Parliament and trust in the European Parliament ... 11
2.4 The relationship between cognitive mobilization and trust in the European Parliament ... 12
2.5 The relationship between micro and macro-level factors and trust in the European Parliament ... 15
Chapter 3: Theory ... 17
3.1 Hypothesis development ... 17
3.1 Conceptual framework ... 19
Chapter 4: Research design ... 20
4.1 Data sources ... 20
4.2 Operationalization of variables ... 21
4.3 Methodology ... 22
4.1 Validity and reliability ... 22
4.2 Case selection ... 22
4.3 Descriptive statistics ... 23
Chapter 5: Analysis of empirical findings ... 27
5.1 Basic regression with five main independent variables ... 27
5.2 The moderating effect of Cognitive Mobilization ... 30
Chapter 6: Conclusion ... 34
Chapter 1: Introduction
From the standpoint of trust, it is historically seen remarkable that six West European states that used to be at war initiated The European Coal and Steel Community (ECSC) in 1952 and the European Economic Community (ECSC) in 1958. A time when trust between elites must have been at an absolute low in Europe. The literature on European Integration often attributes this to the efforts of leading European politicians (the elite-level) and a number of background conditions: American hegemony that supported democratic consolidation of weak political regimes in Western Europe, the North Atlantic Treaty Organization (NATO) and a long period of economic growth following World War II (Engelbrekt et al., 2018, p.4).
In these years, different political elites learned to work together and completed the single market, the euro, the Schengen passport-free zone and a successful policy of enlargement (Engelbrekt et al., 2018, p. 5; Checkel & Katzenstein, 2009, p. 21). Different scholars describe this as years of more or less constant negotiation that fostered trust and an ability to solve problems at the elite level. However, according to Engelbrekt et al., (2018, p. 3), it is also clear that nothing similar has taken place among the broader population.
Concerns about political trust seem to be more present today than ever before. Trust is under severe pressure due to several crises Europe is facing today: it is put under stress by the migrant and refugee crises, the lingering euro crisis, a multitude of security crises, ‘Brexit’ and it must be acknowledged that the US administration under President Trump has taken a much more critical stance towards the European Union (Engelbrekt et al., 2018, p. 2). Most recently we can also add the financial consequences of the coronavirus to this, which seems to lead to a “competition” between the north and south.
This thesis is written in order to come to a better understanding of the framework for trust in the European Union, by specifically focusing on one of its institutions: the European Parliament. We specifically zoom in at the moderating effect of “Cognitive Mobilization” (CM). The idea that CM has a moderating effect between trust in the national Parliament and trust in the European Parliament has gained acceptance of different scholars (Muñoz 2017; Armingeon et al., 2017). New to this study, is that it tries to reflect levels of CM differently than was done in
numbers of correct answers to three simplistic questions on knowledge about the EU from the Eurobarometer, this research reflects the level of cognitive mobilization in a person’s level of education, self-reported interest in politics and consumption of news on politics and current affairs.
The process of European integration comes with difficult to understand multi-level governance structures. This makes it more and more difficult to explain how decisions are made and what the balance of interests were. This research tries to better clarify the evaluative framework that citizens use by specifically focusing on the functioning of CM. Results contribute to a better understanding of commonalities and differences between member states. This is valuable information for both policy-advisors and political scientists.
The expectation is that the more knowledgeable a person is, the better he or she is able to distinct between the National parliament and the European Parliament in his or her evaluation. The main research question of this thesis is:
“How does political support work in multi-level settings: is the proxy effect of trust in the national parliament on trust in the European parliament less present for people with higher
levels of Cognitive Mobilization (CM)?”
Next to this, we look at the “corrosive effect” of corruption on trust in the national parliament. In order to do this, we made two groups from our whole sample with the five relative most corrupt member states, and the five relative most ‘clean’ member states. This is done, in order to see if effects differ when corruption is taken into account. All hypotheses are presented in chapter 3.
This study does not have the pretention to present a complete framework for trust in the European Parliament. It is stringently focused on CM as a moderator. This research applies a simple linear regression model with data from the European Social Survey, round 9 (2018).
Chapter 2: Literature review and hypothesis development
The concept of institutional trust is central to this study. We discuss this concept in relation to the unique multi-level governance structure of the European Union. From this literature review, we derive four explanations for trust in the European Parliament: 1) trust in the national parliament; 2) the level of cognitive mobilization; 3) the level of perceived domestic corruption; 4) socio-economic factors.
The following literature review first explains the function of institutional trust in liberal democracies. Then, it explains the functioning of trust within the multi-level governance structure of the European Union. Then, we follow with five explanations for trust in the European Parliament which are derived from literature. Finally, this chapter presents hypotheses based on this literature review.
2.1 What is institutional trust?
There are many conceptualizations of trust in the social sciences. Most popular in economics, political science and sociology is to approach trust as a rational choice theory (RCT). RCT represents a broad class of models that relies on assumptions of an individual’s rationality. RCT assumes that individuals can make calculated decisions bases on costs, benefits, and probabilities while using all available information. This makes it possible for the individual to choose between different options for the action with the highest net benefit (Cook & Santana, 2018). Weaker versions of RCT acknowledge cognitive limitations, restrictions on the availability of information and the existence of social structural constraints. This is known as “bounded” rationality (Simon, 1955). Different studies in the field of comparative political science show us that trust is generated through an evaluative process rather than being deeply culturally embedded. Therefore, we can understand institutional trust within a rationalist framework (e.g. Hakhverdian & Mayne, 2012; Levi & Stoker, 2000; van der Meer 2017). Institutional trust is a complex phenomenon, but there seems to be some minimal consensus about its meaning. From a perspective of the broader population, institutional trust can be understood as citizens’ support for political institutions such as the government and parliament in the face of uncertainty about or the vulnerability to the actions of these institutions (van der
that case, trust is central in the capacity to reach agreements between international agencies or nations (Engelbrekt, Bremberg, Michalski, & Oxelheim, 2018, pp. 1-3). Trust is the expectation that the object of trust will produce positive outcomes. It can be conceptualized in a graded fashion or more dichotomously (one either trusts or distrusts) (Levi & Stoker, 2000, p. 476).
Trust judgments inspire courses of action and are therefore considered building blocks of social cooperation. Trust relationships are highly efficient since they do not involve high costs of monitoring and sanctioning (Levi & Stoker, 2000; Warren 2017, p. 1). Governments that enjoy a high level of trust by their citizens tend to work more efficient, and their decisions enjoy a higher level of legitimacy (Arnold et al., 2012, p. 1). In the remainder of this literature study, we will focus on institutional trust from the citizens’ perspective.
From an administrative point of view, institutional trust is desirable in both democracies and non-democracies. From a liberal democratic perspective, one would call for an active and vigilant citizenry that can scrutinize policies and if necessary, assert control over the government (van der Meer 2017, p. 6). High levels of political trust may imply uncritical attitudes and reduce political interest. “Non-trusting citizens nevertheless strengthen
democracy, if only because their lack of trust strengthens engagement” (van der Meer 2017, p.
6; Butler, Giuliano, & Guiso, 2016). The relationship between trust and democracy is, therefore, a complex phenomenon and balance.
To maintain democratic representation, it is important to prevent mistrust from spreading into cynicism. “Disenchanted citizens may decide to withdraw from politics altogether – resulting in
even more disenchantment – or provide fertile ground for the emergence of anti-system political parties” (van der Meer & Zmerli, 2017).
Concerns about institutional trust in the Western world are not new. Since the mid-1970s, numerous scholars started to write on the erosion of institutional trust (Offe, 1972; Miller, 1974; Crozier et al., 1975). One scholar that early envisaged that trust would play a decisive role in the process of European integration was political scientist Karl W. Deutsch. He examined the historical evolution of federated and amalgamated communities in order to attempt to find
the same community and came to trust and understand each other, prospects for successful integration would be greatly enhanced (Engelbrekt et al., 2018, p. 2).
2.2 Institutional trust within the multi-level structure of the European Union
Local institutions are almost in every country, and for a great share of the population, more trusted than national institutions. The position of an institution within the overall structure,
ceteris paribus, affects citizens’ trust. The logic behind this is that small political units are more
responsive and more open to participation than larger ones. However, this is as well contrasted with the idea that one can expect more impersonal management from larger political units that might lead to more integrity, less nepotism and more competent politicians (Muñoz, 2017, p. 70; Oliver 2000).
The pattern described above is not found for the EU. Some citizens have more trust in the supranational layer of the European Union, while others put more trust in their national institutions. This has to do with national characteristics and the unique multi-level governance structure of the European Union.
Muñoz (2017) criticizes contemporary literature on institutional trust. His main critique is that research is too often restricted to the analysis of attitudes towards nation-state political institutions in isolation. He finds this unsatisfactory because it omits key theoretical and empirical questions. First of all, not all institutions are equal. They differ in size and visibility what determines their relative position within a multi-level political system. Second – and maybe most important - citizens do not deal with a mere juxtaposition of institutions, but with complex and entrenched governance structures in which various institutional layers relate to each other (Muñoz, 2017, p. 69). Therefore, it is important to take the unique multi-level structure of the European Union into account within this line of research.
The European Union did not start within the complex multi-level structure as it is functioning today. This merely developed since the 1980s which can be considered as a period of politicization. Decisions more and more started to affect the future course of the European Communities and therefore became matters of public debate. Nowadays, the European Union is considered a sui generis multi-level polity, that has a big impact on citizens’ everyday lives
The structure of the European Union does not assign exclusive policy competence or a stable hierarchy of political authority to any level. It requires multiple political actors at different levels to continuously negotiate, deliberate and implement policies (Cengiz, 2015, pp. 826). The European Union has supreme law-making and adjudicating powers, and enforcement at the national levels, but policy-making is also an interplay between the national and European level. For example, at the Council of Ministers. (Deetham, 2013, pp. 279-281).
The structure of the European Union clashes with the traditional liberal/republican legitimacy model that takes the nation-states’ institutional structure as their main template and equate power-holders’ legitimacy with their democratic credentials (Cengiz, 2015, pp. 829). However, according to Habermas (2012, pp. 234), this does not directly have to be a problem in post-modern societies, according to which deliberative politics acquires its legitimating force from the discursive structure of opinion and will-information. Deetham (2013, pp. 279-281) agrees by saying that legitimacy in the European political space should be seen as an interactive process. The legitimacy of states is sometimes reinforced and sometimes compromised. Therefore, studies that separate the audiences for legitimacy claims neatly between state elites for the EU level and citizens (the broader population) for the national level cannot be sustained (Deetham, 2013, pp. 279-281).
Arnold et al., (2012, p. 3) find it essential to assess how citizens directly evaluate the institutions of the European Union within this multi-level structure, especially in the light of debates on the perceived democratic deficit and the persistence of Euroscepticism. They criticize earlier studies stating that they too often focus on determinants of support for the European Union, overlooking critical aspects such as the trustworthiness of the institutions of the European Union. Arnold et al., (2012, p. 5) says that trust in EU institutions can be used as a central indicator for support – and therefore legitimacy – of the European Union. Since 2012, studies on public opinion about the European Union use institutional trust as a central indicator, rather than general support for membership of the EU.
2.3 The relationship between trust in the national Parliament and trust in
the European Parliament
Trust in the national parliament
Elections for the European Parliament are often considered as ‘second-order’ national elections, presenting voters with little stimulus other than their habitual national party context (van der Eijk, Franklin & Marsh, 1996). Studies show that there is low salience of European issues with the public and that elections are largely driven by domestic factors (Arnold et al., 2012).
Few Dutch citizens knew names of European party-leaders during the 2019 EP elections, and only 29% felt involved in the European elections. This is considerably lower than at the national elections (77%) (Peeperkorn, 2019). Therefore, the assertion is that trust in the national Parliament (first-order) influences trust in the European Parliament (second-order), even though, these Parliaments ostensibly function on different levels. (van der Eijk et al., 1996). Scholars agree that on the individual level, trust in the National Parliament is one of the most important predictors for trust in the European Parliament (e.g. Muñoz et al., 2011). This does, however, not preclude that this is different on the country level. Whether this relationship is positive or negative is even contested. This is the discussion between the congruence and
compensation hypothesis.
The congruence hypothesis is that citizens extrapolate their increasing distrust in national institutions to EU institutions (Torcal et al., 2018). This is a spillover effect from the national level to the EU level. This theory is mainly developed by Anderson (1998) and assumes that citizens generally have low levels of interest in European politics and therefore a low level of information and knowledge on the EU. It regards the nation-state as a central player: orientations are created on the national level and are dependent on national factors (Kritzinger, 2003). This logic is considered the strongest predictor by several authors (e.g. Harteveld, van der Meer & de Vries., 2013; Brosius et al., 2018). Interesting to the congruence effect is that predictors of trust are not inherent to the EU itself and therefore hard to influence. The congruence effect is related to a national context theory.
The compensation hypothesis is that citizens who mistrust national institutions put more trust in EU institutions. Citizens then see the EU as it where some sort of lifebuoy that can mitigate national shortcomings. This effect comes up when citizens are dissatisfied with their national government. This can be due to several factors such as inefficiency, high costs, or corruption. (Muñoz et al., 2011; Arnold et al., 2012; de Wilde et al., 2015). The mechanism underlying this compensation hypothesis is that citizens of inefficient and ill-functioning member states will perceive lower costs and higher benefits of transferring sovereignty to the EU. Following this theory, opinions on the National Parliament does not function as a proxy for citizens opinion on the European Parliament. On the opposite, citizens who are happy about the functioning of their government will probably feel unsure about transferring sovereignty (Muñoz et al., 2011, p.5).
2.4 The relationship between cognitive mobilization and trust in the
European Parliament
Cognitive mobilization
The third predictor to be discussed is the level of cognitive mobilization (CM), put forward by Inglehart (1970). He states that citizens in postindustrial societies are generally better educated and have increased access to information which constitutes their political participation. The process of CM leads to more independent citizens and the ability to relate to remote roles and situations. Citizens then possess more political resources and skills that enable them to deal with the complexities of politics and reach their own political decisions without cues (Dalton, 2007, p. 276). CM is a comprehensive concept for which different indicators are used. Originally, Inglehart and Dalton used the variables ‘level of education’ and ‘interest in politics’ to create an index. Other scholars also used variables such as ‘frequency of political discussion’ (Alaminos & Penalva, 2012).
Janssen (1991, p. 467) later found that the process of European integration is experienced less threatening for people who better understood it. This implies a positive relation between CM and trust in the European Parliament. In order to measure CM, we use the two original predictors implied by Inglehart. These are the respondent’s ’level of education’ and ‘interest in politics’.
Theories of CM were originally used to do studies on habitual party cues and alignment. The studies of Muñoz (2017) and Armingeon (2017) give us reasons to believe that this theory can also be used to see to what degree habitual cues and alignment are present in the relationship between trust in the national parliament and trust in the European Parliament. Therefore, we repeat that CM not only has a direct effect on trust in the European Parliament but also has a moderating effect between trust in the national Parliament and trust in the European Parliament.
Cognitive mobilization: level of education
Within the concept of CM and the related theory by Inglehart (1970), we would expect that education boosts trust in the European Parliament. However, studies show scattered findings. Arnold et al., (2012) show that these inconsistencies of the effect of education are dependent upon micro and macro-level variables. Arnold et al., (2012) find that highly educated citizens tend to trust the European Parliament less. The authors say that this might be due to the fact that educated citizens are more aware of institutional shortcomings, such as the democratic deficit of the European Parliament. Other studies show that the effect of education on trust in the European Parliament is moderated as well. This will be discussed in the next paragraph.
Cognitive mobilization: Interest in politics
CM implies a psychological involvement in politics so that latent abilities are applied to political decision making (Dalton, 2007, p. 277). Therefore, political engagement or political interest is often used when constructing an index for CM (e.g. Dalton., 2007; Arnold et al., 2012). Arnold et al., (2012) found that citizens that are more interested in politics and engage in political activities are more likely to trust the European Parliament.
The conflicting findings of Arnold et al., (2012) between ‘political interest’ and ‘education’ are remarkable. Especially given the long line of research on CM. This could mean that either one of these is not a good indicator for the underlying concept of CM, or that the relationship between trust in the national Parliament and European Parliament is changing over-time and across Europe. This is probably more likely, knowing that the European Union experienced a process of politicization since the 1980s and nowadays has a unique multi-level structure.
CM as a moderator between trust in the national Parliament and trust in the European Parliament
According to Muñoz (2017) and Armingeon et al., (2017), the lack of information and complexity of multi-level systems make citizens judge second-order institutions based on their attitudes towards first-order institutions. This is especially the case for citizens with low levels of CM. This means that the relationship between trust in the national Parliament and trust in the European parliament is moderated by individual levels of CM. Karp, Banducci & Bowler (2012) earlier on found a similar effect
These studies on the moderating effect of CM use two simple questions on objective knowledge as indicators for CM. These are derived from the Eurobarometer. Armingeon et al., (2017) used the questions: 1) is Switzerland part of the EU? 2) How many member states does the EU have? One could question the strength of these proxies. The questions ask for simple facts and are quiz-like. Though, their findings show that CM moderates the relationship between ‘trust in national parliament’ and ‘trust in European parliament’.
Cognitive mobilization: news consumption
Besides a self-reported interest in politics and the level of education, levels of cognitive mobilization can be reflected in news consumption as is done by Karp et al., (2013). It is, however, important to know that different media sources influence opinions differently. Traditional media, in general, determines more optimistic views while people exposed to new media such as Twitter and Facebook likely take a more critical stance towards the EU (Conti & Memoli 2017, p. 41).
2.5 The relationship between micro and macro-level factors and trust in
the European Parliament
Torca et al., (2018) found that the effect of extrapolation of trust by citizens is not constant over time, but dependent on national economic and political conditions. This is in line with the idea that orientations towards the European Union are created on the national level and dependent on micro and macro-level factors (e.g. Kritzinger, 2003; Eurofound, 2018). In this section, we will focus on perceived domestic corruption and take other national factors into account as well.
Perceived domestic corruption
Most mentioned macro-level factor that influences trust in institutions is the level of perceived domestic corruption. Domestic corruption erodes trust in the national political system. Therefore, it increases the willingness of citizens to turn to the European Union as an alternative to the Nation-state (Arnold et al., 2012). However, it was also found that this effect of ‘perceived domestic corruption’ on trust in EU institutions vanished when the global financial crisis hit Europe in 2008, and strong austerity measures were taken (Obydenkova & Arpino, 2017).
Hakhverdian & Wayne (2012) found that the effect of corruption on institutional trust is moderated by the level of education. They found that in corrupt societies, the most educated are more distrustful towards their national institutions than the least educated. The given explanation is that “higher educated are better able to identify practices that undermine the
smooth functioning of a democratic society and are normatively more troubled by such practices” (Hakhverdian & Wayne, 2012, p. 6). The authors call this the “corrosive effect” of
corruption on political trust. On the opposite, they also found that trust in institutions in relative “clean” societies increased with education. This study looked on a domestic level at 21 European democracies. Their dependent variable was a combination of trust in five domestic institutions. This shows that education fulfils a broader role within the rational evaluative framework of trust in institutions. This study of Hakhverdian & Wayne (2012) shows that the level of education both as a predictor and moderating variable seems to work different between countries.
Other national socio-economic factors
Other mentioned explanations for trust in the European institutions found by scholars are socio-economic circumstances, voice-and accountability (functioning of national democracy), perceived social-tensions and satisfaction with the quality of government (Harteveld et al., 2013; Roth, Gros, & Nowak-Lehmann, 2014; Muñoz, 2017; Karp et al., 2003).
Freitag and Bühlmann (2009, p. 1556) state that citizens who live in countries where authorities are seen as incorruptible, institutions reduce income disparities and where political interests are proportionally represented are more likely to trust one another.
Chapter 3: Theory
The literature review tells us that the support for the European Union is mainly expressed through trust in its institutions by European citizens. Therefore, the theory of this study takes trust in the European Parliament as its dependent variable.
Based on earlier studies by scholars, we assume a rational evaluative framework for trust in institutions, meaning that individuals are able to make independent trust judgements based on what they experience, hear and process. To this, we acknowledge bounded rationality, meaning that they can’t process all information and therefore use cues and proxies in reflecting their opinions.
This brings us to the idea that within the complex multi-level governance structure of the European Union, people who are better informed are better able to make ‘pure’ rational evaluations, meaning that they make less use of cues and proxies in forming their opinion. This would mean that the effect of trust in the national Parliament on trust in the European Parliament -the strongest predictor according to earlier studies- is less present for the ‘pure’ rationalist.
We assume that the effect between trust in the national Parliament and trust in the European Parliament is moderated by the level of respondents’ “cognitive mobilization” (CM). Based on the literature review, we expect that this concept is reflected in 1) the level of education; 2) the level of interest in politics; 3) consumption of news on politics and current affairs.
3.1 Hypothesis development
Different studies show that the level of ‘trust in the national Parliament’ is the most important predictor for ‘trust in the EU parliament’. However, it is assumed that the strength and direction of this relationship are dependent on micro and macro-level factors. Before we come to this, we first formulate the hypothesis in which ‘trust in the national Parliament’ is a primary explanatory variable:
A second direct explanatory variable for ‘trust in the European Parliament’ discussed, is the ‘level of education’. Whether this relationship is positive or negative is, however, contested. The following hypothesis is formulated:
H2: “There is a positive effect of education on trust in the EU parliament.”
Most studies on trust in European institutions apply a rationalist framework in which institutional trust is generated through an evaluative process rather than being culturally embedded. Studies show that it difficult for citizens to differentiate between the European Parliament and the national parliament when evaluating their performance. This is due to a combination of the multi-level structure of the European Union and bounded rationality and the fact that policy is often an interplay between the national and European level.
Following earlier studies, we assume that the relation between ‘trust in national parliament’ and ‘trust in European parliament’ is moderated by the level of ‘cognitive mobilization’. Citizens with a higher ‘cognitive mobilization’ are better able to evaluate the functioning of the national Parliament and the European Parliament separately. Earlier studies that looked at this moderator used questions on objective knowledge about the EU as a proxy to measure CM. We think that the predictors “level of education”, “interest in politics” and “news consumption” are better ways to reflect levels of cognitive mobilization and formulate the following hypotheses:
H3: “The effect of trust in the national Parliament on trust in the EU parliament is less present
for high educated citizens.”
H4: “The effect of trust in the national Parliament on trust in the EU parliament is less present
for citizens that are more interested in politics.”
H5: “The effect of trust in the national Parliament on trust in the EU parliament is less present
for citizens that watch, listen and read more news on politics and current affairs.”
The last hypotheses that will be tested in this study are based on the results of the study of Hakhverdian et al., (2012). These authors found that in corrupt societies, the most educated are more distrustful towards their national institutions than the least educated. They call this
corrosive effect is present in two most relative corrupt societies and two most clean societies for our dataset. We formulate the following hypotheses:
H6: “In relative corrupt societies, high-educated put more trust in the European Parliament than
in their national Parliament.”
H7: “In relative corrupt societies, high-educated are more distrustful towards their national
Parliament than the least educated.”
3.1 Conceptual framework
This research is retrospective and focused on the causal mechanism between ‘trust in national parliament’ on ‘trust in EU parliament’ (X-Y). By taking into account the effect of the moderating variable ‘cognitive mobilization’, we come to the following conceptual framework:
Independent Levels of national corruption Independent Cognitive Mobilization Controls Gender, Age, Social trust, Satisfaction with the
Independent Trust in the national Parliament Dependent Trust in European Parliament Moderator Cognitive Mobilization
Chapter 4: Research design
This section describes the data and methodology used in this study. First, data collection will be described. Second, the main variables and control variables used in the models will be shortly discussed. An overview of all variables is presented in appendix A. Third, the methodology of this study including applied models, are presented. Finally, this chapter will give an overview of relevant descriptive statistics.
4.1 Data sources
This research is a single-case study (on the EU) and mainly used data drawn from the European Social Survey (ESS) round 9 (2018). The ESS is a cross-national survey measuring attitudes, beliefs and behavior of the various European populations. The survey is held every two years since 2001. The ESS questionnaire is conducted by trained staff. Interviews are conducted face-to-face. The dataset used consists of 31.024 individual respondents.
Not all European countries are included in every ESS round. The last ESS round in 2018 covered 19 European countries of which 16 are members of the European Union. This study is focused on these 16 member-states. ESS round 9 (2018) does not include Spain and Portugal. Therefore, ESS round 8 (2014) is used for one model in the data analysis chapter. This is older data but makes it possible for us to see if results differ when these member-states, who in 2014 put more trust in the European Parliament than in their national government, are included as well. The ESS is a social-survey study. Questions are focused on respondents’ individual opinion. This makes it possible to discover and compare differences and similarities between European countries on social themes. “Media use and trust”, and “politics” are two themes that are repeated in every round. This makes the ESS very suitable in order to answer the research question of this study.
Next to the ESS, the Corruption Perception Index (CPI) from Transparency International is used to reflect levels of corruption in different countries. This index is published annually and based on expert opinion and surveys. This is the only country-level variable used in this study.
4.2 Operationalization of variables
Table (1) below, presents all variables that are included in the different models of the data analysis. The main concepts follow from the literature review. These are institutional trust and cognitive mobilization (CM). Institutional trust is split out between the expected main independent variable (trust in the national Parliament) and dependent variable (trust in the European Parliament). Following the theoretical framework, CM consists of respondent’s level of education, self-reported interest in politics and consumption of news about politics and current affairs (reading, listening and watching). The table (1) also presents two other individual socio-economic factors that are used in the data analysis and the variable CPI which is used in one model as an alternative to perceived transparency of political decisions.
Table 1: Description of variables used in this research
Concept Variable Indicator Measurement
Institutional trust Trust in the European
Parliament Ratio 10 (complete trust) 0 (no trust at all) – Trust in the national
Parliament
Ratio 0 (no trust at all) – 10 (complete trust)
Cognitive Mobilization
(CM) Level of education Nominal 27 country-specific options. Rescaled to four different category levels. Interest in politics Ratio 1 (very interested) –
4 (not at all interested) News consumption on
politics and current affairs Ratio Watching, reading or listening on a typical day in minutes
Individual socio-economic
factors (controls) Gender Nominal Male - Female Social trust Ratio 0 (can’t be too careful) – 4
(most people can be trusted)
Age Nominal Years old
Satisfaction with the
economy Ratio 0 (not at all satisfied) – 10 (completely satisfied)
Macro-level factors Corruption Perception
4.3 Methodology
This research is explanatory and uses a large N-design. Linear regression is applied as a statistical method. Its goal is to find out whether the proxy effect of trust in the national parliament on trust in the European parliament is less present for people with higher levels of Cognitive Mobilization (CM). This expectation is based on different theories by scholars, as discussed in the literature review. Next to this, it is expected that this effect of extrapolation is not constant over time but dependent on national factors. Therefore, we built a multi-level dataset with data on both individual and national level. The first reason to do this is that it allows us to control for macro-variables. Secondly, a multi-level analysis makes it possible to test for heterogeneity by specifying cross-level interactions (Steenbergen & Jones, 2002). In other words: we can compare different countries with each other by making subpopulations.
4.1 Validity and reliability
In this research, we introduce a comparison between the five most ‘clean’ and five ‘most corrupt’ societies in our tables by splitting the sample and run the same regression analyses. It is essential to be aware of the fact that levels of corruption are not the only difference that characterize these two groups of countries. Other differences are that the ‘most corrupt’ are almost all located in Eastern Europe, are relative new member states and are net contributors to the EU-budget (except for Italy) (Buchholz, 2020). The five ‘most clean’ are almost all located in Western Europe and are net contributors. These are all declaring factors for differences between these two groups of which the effects could be captured in the estimates we present. Variables used differ in each table. This is because in large N-designs, including too many variables in one regression could cause an overfit, meaning that models could describe random noise rather than the actual relationships between variables.
4.2 Case selection
The European Social Survey is a cross-national survey, carried out by national coordinators. This brings an extra layer of complexity as is discussed by Lynn et al., (2004). The sampling design produced by national coordinators are checked upon by a special ESS sampling-team. Fundamental criteria are that the population is equivalent in each nation and a strict
nationality. The requirement for a minimum effective sample size is 1.500 interviews (Lynn et al., 2004, p. 5-10). A drawback is that not every country participates in every ESS round. This is why ESS round 9 (2018) only involves data from 16 member EU member states.
4.3 Descriptive statistics
Figure (2) below visualizes trust in the national Parliament and European Parliament based on ESS round 9 (2018). Countries are sorted from left to right by levels of the Corruption Perception Index 2018.
Figure 2: trust in the National and European Parliament, sorted by the Corruption Perception Index (source: European Social Survey, 9, 2018; Transparency International 2018)
This figure (2) shows averages to the question ‘On a score of 0-10, how much do you personally
trust each of these institutions?’. The figure (2) shows that Bulgaria, Cyprus, Poland and Slovenia
are amongst the countries with least trust in their national parliaments (all below 4). Austria, Germany, Finland and The Netherlands are amongst the countries with most trust in their national Parliaments (all above 5). Countries with high differences in trust between the national and European Parliament are Austria (1,03), United Kingdom (0,81), Cyprus (-1.02) and Poland (-0,82). The trendlines show that levels of trust in the European Parliament are over-all
0,0 1,0 2,0 3,0 4,0 5,0 6,0 7,0 Bulgari a Hunga ry Italy Cyprus Czechi a Polan d Slove nia Franc e Estoni a
IrelandBelgiu m Austr ia Germa ny United King dom Neth erlan ds Finland
Trust in the National and European Parliament
Trust in National Parliament 2018 Trust in European Parliament 2018
hand side). Countries with more trust in their national Parliament than the European parliament are mainly located on the right-hand side. Czechia being an exception to this. The United Kingdom stand out, having both low levels of trust in their national and European Parliament, which of course very well could be due to struggles around Brexit. Important to note is that all EU member states are reflected in this figure since not all are included in ESS Round 9 (2018).
Figure (3) below shows that for all countries included in this study, high-educated put more trust in the European Parliament than low-educated. The countries are sorted from left to right, based on the Corruption Perception Index 2018. Trendlines are added to visually check for a relationship between trust in the European Parliament and levels of corruption.
Figure 3: Trust in the European Parliament across Educational levels, sorted by Corruption Perception Index (source: European Social Survey, 9, 2018; Transparency International 2018)
The figure (3) illustrates that Trust in the European Parliament increases, as corruption decreases. It is interesting to see that the difference between the two trendlines increases, which implies that differences in trust between the group of the highest educated and lowest educated disappears as corruption increases (left-hand side).
0,0 1,0 2,0 3,0 4,0 5,0 6,0 Bulga ria Hunga ry Italy
Cyprus Czechi a Polan d Slove nia Franc e Estoni a Irelan d Belgiu m Austr ia Germa ny Unite d King dom Neth erlan ds Finlan d
Trust in the European Parliament accros
Educational levels
Highest educated Lowest educated
The figure (4) above illustrates that for the five most corrupt countries, trust in the European Parliament increases with every educational level. Trust in the national parliament is between 3.4 and 4.0 for all levels of education and does not seem to show an increase with every level. The figure below (5) illustrates that for the five most ‘clean ‘countries’, trust in in the national parliament increases with every educational level. Trust in the European Parliament is between 3.7 and 4.6 for all levels of education.
0,00 1,00 2,00 3,00 4,00 5,00 6,00 7,00
Less than lower
secondary Lower secondary postsecondaryUpper and BA and higher
Trust per educational level in five most 'clean' countries
0,00 1,00 2,00 3,00 4,00 5,00 6,00Less than lower
secondary Lower secondary postsecondaryUpper and BA and higher
Trust per educational level in five most corrupt countries
Trust in national parliament Trust in European Parliament
Figure 4: Trust in the national and European Parliament per educational level for the five most corrupt countries (source: European Social Survey. 9. 2018)
Figure 4 shows that in relative corrupt societies, high-educated are not more distrustful towards their national parliament than the least educated. This means that we can already reject hypothesis 7 by visual inspection. Figure 4 also shows that in relative corrupt societies, high educated put far more trust in the European Parliament than in their national parliament. This contrasts to figure 5 with the five most “clean” societies where this works the other way around. This is in line with hypothesis 6.
Figure (6) below shows averages to the question: “On a typical day, about how much time do
you spend watching, reading or listening to news about politics and current affairs? Please give your answer in hours and minutes.” The figure (6) shows that Italy, Belgium and Hungary stand
out as countries where people consume lots of media on politics and current affairs. Visually compared to levels of trust in the European Parliament from figure 2, relations between these two are not directly recognized on the country level.
Figure 6: Consumption of news on politics and current affairs, sorted by Corruption Perception Index
(source: European Social Survey, 9, 2018; Transparency International 2018)
These statistics show a large spread in the consumption of political news per country. Therefore, it can be expected that the effect of news consumption on trust in the European Parliament in the regression analyses will be weak or even insignificant.
0 20 40 60 80 100 120 140 160 180 200 Bulga ria Hunga ry Italy
Cyprus Czechi a Polan d Slove nia Franc e Estoni a Irelan d Belgiu m Austr ia Germa ny Unite d King dom Neth erlan ds Finlan d
Watching, reading or listening to news about
politics and current affairs on a typical day in
Chapter 5: Analysis of empirical findings
This section presents the results of different regression analyses in four tables. In these tables, subsets are made for 1) All 16 countries included in the dataset; 2) The five most corrupt countries in the dataset (Bulgaria, Hungary, Italy, Cyprus and Czechia); 3) The five least corrupt countries in the dataset (Finland, the Netherlands, the United Kingdom, Germany and Austria). All models have trust in the European Parliament as the dependent variable. We have chosen to exclude income as a control variable since including this variable would cause the total number of observations to drop with 5225. This is probably because not all respondents were willing to share information on their income.
Subsection 5.1 presents the regression analysis of the key variables used in this study. Subsection 5.2 includes tables with an interaction term between trust in the national Parliament and variables that reflect levels of CM.
5.1 Basic regression with five main independent variables
This subsection presents the results of the most basic regression analysis conducted for this study. These can be found in table 2 on the next page. It examines the relation between trust in the European Parliament and the variables ‘trust in the national Parliament’, ‘Education’, ‘Interest in politics’. ‘News consumption’, “Corruption” and individual control variables. This subsection is related to hypotheses one and two.
This R-squared level in table 2 varies from 0.315 for the most corrupt countries to 0.466 for the least corrupt countries. An R-squared level of 0.466, means that around 46.6% of the variation in the dependent variable is due to variation in the independent variables included. The R-squared level for all 16 countries is roughly the same as in other studies that look at explanations for trust in the institutions of the European Union (e.g. Arnold et al., 2012). Important to note is that the standard errors are higher in model 2 and 3. This is due to a combination of higher variance in the dependent variables and a smaller number of observations.
increase with one point on a scale of 0 to 10 in trust in the national Parliament results in an increase of 0.528 points for trust in the European Parliament. This empirical evidence is in accordance with earlier findings of Muñoz et al., (2011). The standard error is only 0.006, which suggests that this estimate is very accurate. This result is in support of the first hypothesis: there is a positive and significant effect of trust in the national parliament on trust in the European parliament. We will see this effect as well in the other upcoming tables.
Table 2: Basic regression with five main independent variables
Independent variable Model 1 (Std. error) Model 2 (Std. error) Model 3 (Std. error) All 16 countries 5 most corrupt 5 least corrupt Main variables of interest
Trust in National Parliament .528 *** (.006) .485 *** (.014) .537 *** (.011)
Education (Reference: less than lower secondary)
Lower secondary -.224 *** (.052) .176 (.114) .109 (.082)
Upper and postsecondary -.219 *** (.053) .039 (.116) -.111 (.082)
BA and Higher -.001 (.055) .364 *** (.125) .027 (.084) Interest in politics .06 *** (.015) .092 *** (.035) .093 *** (.027) News consumption 0 (0) 0 * (0) 0 (0) Corruption .022 *** (.001) -.011 ** (.005) -.111 *** (.011) Individual-level controls Female .25 *** (.025) .082 (.057) .431 *** (.042) Social trust .096 *** (.006) .111 *** (.012) .095 *** (.011) Age -.014 *** (.001) -.003 (.002) -.023 *** (.001)
Satisfaction with the economy
.078 *** (.006) .064 *** (.015) .089 *** (.011)
Constant .761 *** (.095) 2.002 *** (.322) 3.071 *** (.272)
Mean dependent variable 4.453 4.306 4.595
R-squared 0.368 0.315 0.466 Standard deviation dependent variable: 2.530 2.677 2.430 Number of observations 27941 6277 7436 *** p<.01, ** p<.05, * p<.1
All three models in table 2 show a small effect of interest in politics on trust in the European Parliament. News consumption is insignificant (model 1 and 3) or has no real effect at all (model 2). This implies that news consumption is an incorrect variable for the underlying concept of CM. The effects of corruption are negligibly small, but it is interesting that they vary from positive (model 1) to negative (model 2 and 3).
We expected to see the estimate for education to become more positive with every increase in the level of education. The results show different outcomes for all three models. For the overall sample (model 1), we spot a negligible difference between lower secondary and upper and postsecondary with a p-level lower than 0.001, indicating relative to the reference category. The estimate for BA and higher educated is even insignificant. The estimates change when groups are split-up between the five most and the five least corrupt countries. The estimate for the group of highest educated people in the most corrupt countries (table 2, model 2) is significant and strong (0.364 points) with a p-level lower than 0.001. This contrasts to the same group of highest educated in the 5 least corrupt countries. This means that when you live in Bulgaria, Hungary, Italy, Cyprus or Czechia and have obtained a bachelor’s degree or higher, you are more likely to trust the European Parliament than your peer In Finland, the United Kingdom, the Netherlands, Germany and Austria.
The effects of education in model 2 and 3 in table 2 are heterogeneous, which means that the effect is different, in different populations. This is why, in relation to the second hypothesis, we can only state that there is a positive and significant effect of education on trust in the European parliament for countries in model 2. This effect is not imperceptible for the overall sample and the group of countries in model 3.
Table 2 presents a stronger negative association between national perceived corruption and trust in the European Parliament for model 3 than it does for model 2. One must understand that this does not automatically mean that people living in countries included in model 3 find corruption more worrisome. Because countries are already split-up based on corruption, chances are high that effects of corruption are already captured in other estimates.
Most control variables for the models in table 2 are significant, with a p-level lower than 0.01. What stands out is the highly significant association for gender with trust in the European Parliament. Women living in Finland, the United Kingdom, the Netherlands, Germany and Austria put relatively more trust in the European Parliament than men. There is no difference of trust in the European Parliament between man and women for Bulgaria, Hungary, Italy, Cyprus or Czechia. Based on table 2, we can state that young people relatively put more confidence in the European Parliament. One year older makes people -0.014 points trust less
5.2 The moderating effect of Cognitive Mobilization
This subsection presents the results for the regression analyses that look at the moderating effect of CM. These tables can be found on the next pages. Results in table 3, 4 and 5 are related to the main research question of this study that looks at the moderating effect of CM, which is captured in hypothesis three and four.
Cognitive mobilization: level of education
Table 3 on the next page presents the results of the regression analyses with an interaction between trust in the national parliament and educational level. If all 16 countries are taken into account (model 1), results show small differences between educational levels. All three levels make less use of trust in the national Parliament as a proxy for trust in the European Parliament than the reference category, but we do not spot a stronger negative association for every increase in educational level since the association is less intense for upper and post-secondary educated. We can conclude that the proxy effect of trust in the national parliament on trust in the European parliament is least for the highest educated people in all 16 countries. Model 2 shows a more negative interaction for every lower educational level. This implies that in these countries, the proxy effect is least present for the least educated. This is remarkable because it is contrary to our theory. At the same time, the results of the interaction effect are insignificant for model 3, which makes it impossible to make a good comparison between the most and least corrupt countries. This is why we can only draw the conclusion that the proxy effect is least present for the highest educated group of people in the whole sample.
Cognitive mobilization: interest in politics
Table 4 on the next page presents the results of the regression analyses with an interaction between trust in the national parliament and interest in politics. This interaction shows a negative and significant association for all three models. This supports hypothesis four. For the first model in table 4, the result means that as interest in politics increases with 1 point on a
Table 3: Interaction for Trust in National Parliament with education
Independent variable Model 1 (Std. error) Model 2 (Std. error) Model 3 (Std. error) All 16
countries corrupt 5 most corrupt 5 least
Main variables of interest
Trust in National Parliament .567 *** (.018) .575 *** (.037) .549 *** (.028)
Trust in National Parliament x Lower secondary
-.045 ** (.019) -.134 *** (.041) -.009 (.031)
Trust in National Parliament x Upper and postsecondary -.034 * (.019) -.084 ** (.04) -.007 (.031) Trust in National Parliament x BA and Higher -.049 ** (.02) -.062 (.044) -.029 (.032)
Education (Reference: less than lower secondary)
Lower secondary -.038 (.094) .653 *** (.182) .151 (.167)
Upper and postsecondary -.082 (.095) .317 * (.18) -.082 (.169)
BA and Higher .208 ** (.103) .56 *** (.199) .181 (.182) Interest in politics .06 *** (.015) .093 *** (.035) .093 *** (.027) News consumption 0 (0) 0 * (0) 0 (0) Corruption .022 *** (.001) -.01 * (.005) -.112 *** (.011) Individual-level controls Female .251 *** (.025) .085 (.057) .431 *** (.042) Social trust .096 *** (.006) .11 *** (.012) .095 *** (.011) Age -.014 *** (.001) -.003 * (.002) -.023 *** (.001)
Satisfaction with the
economy .078 *** (.006) .064 *** (.015) .089 *** (.011)
Constant .604 *** (.116) 1.659 *** (.343) 3.019 *** (.297)
Mean dependent variable 4.453 4.306 4.595
R-squared 0.368 0.316 0.466
Standard deviation
dependent variable: 2.530 2.677 2.430
Number of observations 27941 6277 7436
Parliament decreases with 0.048 points. This means that the proxy effect - which is the effect of trust in the national parliament on trust in the European parliament – is less intense for people who are more interested in politics. These people are better able to make up their opinion on the European Parliament, independent from their opinion in the national parliament.
If we compare the 5 most corrupt countries with the 5 least corrupt countries, we see that the proxy effect itself is more intense for countries in model 2, but that these countries also have a more intense correctional effect of the moderator interest in politics. Together with earlier results in previous tables, this shows that the effects of CM differ between countries these two groups of countries.
Table 4: Interaction for trust in the National Parliament with interest in politics
Independent variable Model 1 (Std. error) Model 2 (Std. error) Model 3 (Std. error) All 16
countries corrupt 5 most corrupt 5 least
Main variables of interest
Trust in National Parliament .641 *** (.013) .673 *** (.028) .596 *** (.028)
Interest in politics .267 *** (.026) .434 *** (.056) .196 *** (.052)
Trust in National Parliament
x Interest in politics -.048 *** (.005) -.09 *** (.012) -.022 ** (.009)
Education (Reference: less than lower secondary)
Lower secondary -.231*** (.052) .156 (.114) .107 (.082)
Upper and postsecondary -.224 *** (.053) .013 (.115) -.111 (.082)
BA and Higher .008 (.055) .343 *** (.125) .034 (.084) News consumption 0 (0) 0 * (0) 0 (0) Corruption .022 *** (.001) -.012 ** (.005) -.11 *** (.011) Individual-level controls Female .24 *** (.025) .073 (.057) .428 *** (.042) Social trust .095 *** (.006) .112 *** (.012) .094 *** (.011) Age -.014 *** (.001) -.003 * (.002) -.023 *** (.001)
Satisfaction with the economy
.078 *** (.006) .066 *** (.015) .089 *** (.011)
Constant .312 *** (.105) 1.403 *** (.33) 2.783 *** (.299)
Mean dependent variable 4.453 4.306 4.595
R-squared 0.370 0.321 0.466
Standard deviation
dependent variable: 2.530 2.677 2.430
Number of observations 27941 6277 7436
Cognitive mobilization: news consumption
At last, we present the table 5 with results for the regression analysis with an interaction between trust in the national parliament and consumption of news on politics and current affairs. As expected from the descriptive statistics and results in table 2, the estimate for news consumption and the interaction with trust in the national parliament is negligible and in model 1 and 2 even insignificant.
Table 5: Interaction for trust in the national parliament with news consumption
Independent variable Model 1 (Std. error) Model 2 (Std. error) Model 3 (Std. error) All 16 countries 5 most corrupt 5 least corrupt Main variables of interest
Trust in National Parliament .54 *** (.006) .49 *** (.015) .543 *** (.012)
News consumption .001 *** (0) 0 (0) .001 (0)
Trust in National parliament x News consumption
0 *** (0) 0 (0) 0 (0)
Education (Reference: less than lower secondary)
Lower secondary -.221 *** (.052) .175 (.114) .111 (.082)
Upper and postsecondary -.216 *** (.053) .037 (.116) -.108 (.082)
BA and Higher .002 (.055) .363 *** (.125) .03 (.084) Interest in politics .06 *** (.015) .092 *** (.035) .093 *** (.027) Corruption .022 *** (.001) -.011 ** (.005) -.111 *** (.011) Individual-level controls Female .25 *** (.025) .082 (.057) .43 *** (.042) Social trust .096 *** (.006) .111 *** (.012) .095 *** (.011) Age -.014 *** (.001) -.003 * (.002) -.023 *** (.001)
Satisfaction with the economy
.078 *** (.006) .064 *** (.015) .089 *** (.011)
Constant .701 *** (.096) 1.981 *** (.324) 3.033 *** (.274)
Mean dependent variable 4.453 4.306 4.595
R-squared 0.368 0.315 0.466 Standard deviation dependent variable: 2.530 2.677 2.430 Number of observations 27941.000 6277.000 7436.000 *** p<.01, ** p<.05, * p<.1
This implies that news consumption is not a good indicator for the underlying concept of CM: people who read, watch or listen more news on politics are not better able to distinguish
Chapter 6: Conclusion
This research contributes to a better understanding of the causal mechanisms that define trust in the European Parliament. The purpose is to investigate if political support for the European Parliament differs between different levels of Cognitive Mobilization (CM). This section provides a discussion of the main findings, as well as potential limitations and recommendations for future research. Let’s start by repeating the main research question one more time:
“How does political support work in multi-level settings: is the proxy effect of trust in the national parliament on trust in the European parliament less present for people with higher
levels of Cognitive Mobilization (CM)?”
The tested theory is based on the following propositions: first, we assume a rational evaluative framework for trust in institutions. If political institutions perform poorly, people withhold trust or vice versa. To this, we acknowledge bounded rationality. Second, we assume that people with high levels of CM are better able to make up their opinion on the European Parliament and their national parliament without making use of cues and proxies. This brings us to the theory that the effect of trust in the national parliament on trust in the European parliament – which is considered the most strong predictor by scholars – is less intense for the higher educated, the political interested and the one who follows the news on politics and current affairs.
Based on our descriptive statistics, we would first conclude that national parliaments are not on average more trusted than the European Parliament. Our whole sample of 16 member states shows exactly a 50/50 split. This is contrary to the idea that the position of an institution within the overall structure, ceteris paribus, affects citizens’ trust. However, when we look closely, we also see that this split is almost exactly a split between relative corrupt and relative “clean” countries. This indicates that in the face of uncertainty about the vulnerability of actions of the national parliament, citizens turn towards the European Parliament as if it were some kind of “lifebuoy”. Put differently: it is likely that that compensation effect is more
more trust in the European Parliament (hypothesis 6). We do not spot real differences for trust in the national parliament between low and high educated in relative corrupt societies (hypothesis 7). Citizens living in relative “clean” countries relatively put more trust in their national parliament. Therefore, it is likely that the congruence effect is more prevalent in these countries.
From our whole sample (16 member states), we can conclude that trust in the national parliament is indeed the most important predictor for trust in the European Parliament (hypothesis 1). Empirical findings show that this relation is as expected, moderated by individual levels of CM. Not education, but interest in politics itself show to be the most important moderating variable. People proxy effect is less present for people that are interested in politics (hypothesis 4). This follows a reasonable logic: being high educated does not per definition mean that you know a lot about politics and therefore make less use of cues and proxies in your value judgement. Consumption of news show to have no correctional effect at all (hypothesis 5).
The moderating effect of education only shows small differences between educational levels for the whole sample, but its correctional effect is most intense for the highest educated people in the whole sample (with all 16 member states). This is in line with our theory that higher educated are better able to distinct between the national and European Parliament (hypothesis 3).
From our sample with 16 member states of the European Union, we made subsamples for the five most corrupt countries (Hungary, Bulgaria, Italy, Cyprus and Czechia) and the five least corrupt countries (Finland, the Netherlands, the United Kingdom, Germany and Austria). The direct effect of education on trust in the European Parliament is heterogeneous: it is positively associated for the sample with the five most corrupt countries, whilst education has no association with trust in the European Parliament for the sample with the five least corrupt countries (hypothesis 2). This contrasts to the findings of Arnold et al., (2012) who found an overall negative association for education on trust in the European Parliament. Arnold et al., (2012) suggested that it could be that education leads to more awareness of the shortcomings
outweigh possible shortcomings. We do not spot the same moderating effect for the “least corrupt” countries.
Consumption of news on politics and current affairs has no effect at all on trust in the European Parliament. This might be due to the fact that we are not able to adjust for the type of media source. It might be interesting to adjust for this in future research, knowing that people who are more exposed to new media such as Twitter and Facebook are more likely to take a more critical stance towards the EU (Conti & Memoli 2017, p. 41). Interest in politics turns out to have a significant positive effect for both the whole sample and the grouped countries.
This research found that the proxy effect of trust in the national parliament on trust in the European Parliament is indeed less present for people with higher levels of Cognitive Mobilization. Self-reported interest in politics is the best reflection for levels of CM in this research. Next to this, this research showcases that it is crucial to account for national characteristics. Therefore, it is important to fetch that perceived corruption is not the only difference between the five ‘most corrupt’ and the five ‘most clean’ countries. This is a limitation of this research. As mentioned in the reliability and validity subsection, other differences between the two group of countries are East / Western Europe, net contributor / net receiver and new / older member states.
An important message from this study to politicians and policy advisors working at national and European intuitions would be that, in order maintain democratic representation – which can be partly reflected in levels of trust according to Habermas – it is important to acknowledge that the modern-day multi-level structure of the European Union can be hard to comprehend for citizens. Especially for those that are less interested in politics and lower educated. At the same time, the European Union has a growing impact on citizens’ everyday lives. It is therefore crucial to listen, inform and involve them in the European project in order to prevent mistrust from spreading into cynicism. As found before by Janssen (1991, p. 467): The process of European Integration is experienced less threatening for people who better understand it.
