MASARYKOVA U N I V E R Z I T A F A K U L T A S O C I Á L N Í C H S T U D I Í Voter Turnout in the 2019 European Parliament Elections: Do Differences in Electoral Systems Matter? Diplomová práce HANA BARTUŠKOVÁ Vedoucí práce: Zuzana Ringlerová, PhD. Katedra mezinárodních vztahů a evropských studií obor Evropská studia Brno 2 0 2 0 MUNI FSS VOTER TURNOUT IN THE 2 0 1 9 EUROPEAN PARLIAMENT ELECTIONS: DO DIFFERENCES IN ELECTORAL SYSTEMS MATTER? Bibliografický záznam Autor: Název práce: Studijní program: Studijní obor: Vedoucí práce: Rok: Počet stran: Klíčová slova: Hana Bartušková Fakulta sociálních studií Masarykova univerzita Katedra mezinárodních vztahů a evropských studií Voter Turnout in the 2019 European Parliament Elections: Do Differences in Electoral Systems Matter? Mezinárodní teritoriální studia Evropská studia Zuzana Ringlerová, PhD. 2020 76 volební účast, volby do Evropského parlamentu, Evropská unie, evropské volby 2019, volební chování 2 VOTER TURNOUT IN THE 2 0 1 9 EUROPEAN PARLIAMENT ELECTIONS: DO DIFFERENCES IN ELECTORAL SYSTEMS MATTER? Bibliographic record Author: Hana Bartušková Faculty of Social Studies Masaryk University Department of International Relations and European Studies Title of Thesis: Voter Turnout in the 2019 European Parliament Elections: Do Differences in Electoral Systems Matter? Degree Programme: International Territorial Studies Field of Study: European Studies Supervisor: Zuzana Ringlerová, PhD. Year: 2020 Number of Pages: 76 Keywords: voter turnout, European Parliament election, European Union, European elections 2019, electoral behaviour 3 VOTER TURNOUT IN THE 2 0 1 9 EUROPEAN PARLIAMENT ELECTIONS: DO DIFFERENCES IN ELECTORAL SYSTEMS MATTER? Abstrakt Volební systémy v členských zemích EU při volbách do Evropského parlamentu jsou odlišné, na nadnárodní úrovni jsou dány pouze rámcové podmínky pro konání voleb. Stejně tak se liší ekonomické podmínky v jednotlivých členských zemích. Hlavním cílem práce je posoudit, zda to, že voliči volí dle jiných podmínek a v jiném ekonomickém prostředí, má vliv na volební účast voličů v těchto volbách. Text posuzuje chování závislé proměnné volební účast v závislosti na řadě nezávislých proměnných individuálních, volebního systému i ekonomických. 4 VOTER TURNOUT IN THE 2 0 1 9 EUROPEAN PARLIAMENT ELECTIONS: DO DIFFERENCES IN ELECTORAL SYSTEMS MATTER? Abstract The electoral systems in the EU Member States in the elections to the European Parliament are different, at the supranational level only the framework conditions for holding elections are given. Economic conditions also differ from one Member State to another. The goal of this thesis is to assess whether the fact that voters vote according to different conditions and in a different economic environment, has an impact on voter turnout. The text assesses the behaviour of the dependent variable turnout on several independent variables: individual, in the electoral system and economic. 5 VOTER TURNOUT IN THE 2 0 1 9 EUROPEAN PARLIAMENT ELECTIONS: DO DIFFERENCES IN ELECTORAL SYSTEMS MATTER? Čestné prohlášení Prohlašuji, že jsem diplomovou práci na téma Voter Turnout in the 2019 European Parliament Elections: Do Differences in Electoral Systems Matter? zpracovala sama. Veškeré prameny a zdroje informací, které jsem použila k sepsání této práce, byly citovány v textu a jsou uvedeny v seznamu použitých pramenů a literatury. V Brně 6. září 2020 Hana Bartušková 7 VOTER TURNOUT IN THE 2 0 1 9 EUROPEAN PARLIAMENT ELECTIONS: DO DIFFERENCES IN ELECTORAL SYSTEMS MATTER? Poděkování Na tomto místě bych chtěla poděkovat své vedoucí, která mi nejen poskytla mnohé rady a cennou zpětnou vazbu, ale především mě jemně dotlačila do složitějšího výzkumu, než jsem měla původně na mysli. Poděkovat chci také univerzitě, fakultě i katedře za to, jak se postavily k organizaci semestru během koropandemie, protože během jarních pandemických měsíců jsem se bohužel k psaní práce vůbec neměla šanci do­ stat. Šablona DP 3.0.7-FSS (2020-04-14) © 2014, 2016, 2018, 2019 Masarykova univerzita 9 OBSAH Obsah List of Figures 13 List of Tables 14 List of Abbreviations 15 Introduction 16 1 Literature review 19 1.1 Voting systems for the European Parliament elections in the EU Member States 20 1.2 Electoral participation in the European Parliament elections .27 1.3 Literature review 27 1.4 The hypothesis and the research question of this thesis 32 2 Methodology of the model and its data 36 2.1 Methodology and variables 36 3 Model and its findings 43 3.1 Hypotheses and the model findings 58 3.2 Result and discussion 59 4 Conclusion 61 Literature 63 Příloha A Voting systems in the EU Member States for the EP elections 69 Příloha B Data 70 Příloha C R codes 71 Rejstřík 74 11 INTRODUCTION List of Figures Figure 1: Odds ratios for independent variables in model 2 48 Figure 2: Predictions of selected variables' impact in model 2 (Age, Education, Satisfaction with the EU, Support for the EU mem.) 49 Figure 3: Predictions of selected variables' impact in model 2 (National elections, Economic growth, Inflation, Unemployment) 51 Figure 4: Impact of the variable Compulsory voting and the variable EU mem. length on the probability of voting 52 Figure 5: Predictions of the model 3a 56 Figure 6: Predictions of the model 3b 57 Figure 7: Predictions for country-level variables, model 3a and 3b 58 13 INTRODUCTION List of Tables Table 1: Turnout in the EP elections in the EU Member States 1979 - 1999 (in %) 25 Table 2: Turnout in the EP elections in the EU Member States 2004 - 2019 (in %) 26 Table 3: Summary of the dependent variable in the model 39 Table 4: The independent variables (individual-level) in the model 40 Table 5: The independent variables (country-level) in the model 42 Table 6: Summary of the model 1 44 Table 7: Summary of the model 2, odds ratios for the independent variables 45 Table 8: Summary of the model 2, country-level variations for the independent country-level variables 46 Table 9: Summary of the model 3a 53 Table 10: Summary of the model 3b 54 14 INTRODUCTION List of Abbreviations 2019 EP election EP election EU EU mem. EU mem. length TEU TFEU - elections to the European Parliament in 2019 - elections to the European Parliament - European Union - membership of a state in the European Union - length of a state's membership in the EU - Treaty on European Union - Treaty on the Functioning of the European Union 15 INTRODUCTION Introduction Turnout in elections is one of the basic characteristics of political behaviour. Turnout can depend on many factors, from the democratic and electoral systems in a given country to the individual characteristics of individual voters. Turnout thus differs from system to system, each individual elections, even in geographical areas. The European Union is a specific entity in this, which is characterized by its unique appearance. It is not a clear political integration (confederation or federations), it is not a unified state, nor is it an international organization in the sense that we usually understand international organizations. Likewise, the European Union is not just a grouping of states for which we evaluate electoral systems. The system of direct elections to the European Parliament is thus a hybrid of the national electoral system with an European harmonisation. Basic common principles are given. And the details of the electoral system of national representatives in Parliament are left to the individual EU Member States, which adapt this European electoral system to their national customs and the national traditional way of doing elections. This thesis focuses on country-level determinants of voter turnout in the European Parliament elections. The main question of this thesis is whether the differences between the electoral systems in each EU Member State and differences between states' economic conditions affect voter turnout. In other words, the question is whether differently set rules for elections to the European Parliament and different economic environment in the Member States affect the participation of voters in these elections. And whether these systematic factors are more important than individual characteristics of voters. The main research question of this theses is to evaluate which factors and circumstances impact turnout in the European Parliament elections. The main emphasis of the research question will be on the specific nature of the EU parliamentary voting system. Mainly on the fact that the voting systems in the Member States differ and how these differences can influence turnout in these elections. The basic premise of this work is the fact that the existing literature deals mainly with national elections and circumstances that affect turnout at the national level. However, the European Union is a relatively new, independent unit with its own electoral system, for which the same 16 INTRODUCTION does not apply as for the national electoral system. The European one is made up of two levels. First the national one and the national electoral system. The second layer is the individual voters. The system of elections to the European Parliament is thus not just the sum of the electoral systems of individual Member States. This is qualitatively influenced by common rules and the overall state of European economic integration. But also by individual voters, for whom the European elections represent something other than the national ones. Most of the current literature summarizes the main characteristics of the European elections under the notion of second-order elections. And then it compares the differences between European elections and national first-order elections (parliamentary, presidential or other). However, the main question of this work is whether differences between the EU Member States (in electoral systems and other features) may have any effect on turnout in these European elections. Thus, in contrast to the approach taken by the second-order election theory, the thesis does not compare national elections with European elections in a given country but focuses on the differences between the EU Member States in elections to the European Parliament. In assessing this research question, many variables are taken in account within the econometric model, which implement independent variables from both the voters themselves and the electoral systems in the EU Member States, as well as other economic characteristics of both voters and national economies. The elections to the European Parliament represent a unique opportunity to compare the influence of the electoral system because the elections are essentially uniform. A single parliament is elected, but under slightly different conditions, which just define the different electoral systems in the EU Member States. The main emphasis of the thesis is on the differences between the Member States, the thesis does not compare the European elections with the national ones, as is the focus of most of the existing literature. On the contrary, the main goal is to compare turnout within one elections but using different electoral systems, using the latest data from 2019. The research question will be discussed on data from the European Parliament 2019 Elections (2019 EP elections) which was held in the 28 EU Member States in May 2019. The research question will examine potential factors influencing voter turnout at two levels. First, the theses 17 INTRODUCTION will evaluate factors at the individual-level, as gender, social status, economic status and attitude of the respondent to the EU, politics and other. Second, the thesis will look at factors at the level of countries. The second level will consist of the elections system in every Member State. The analyses on the second level will focus on the mandatory voting, diversity in voting days or economic conditions in every Member State. These factors are common for large group of respondents and the main goal of this research is to assess how factors at the country-level can influence an individual decision to vote or not to vote. Based on the assessment of several econometric models, the main conclusion is that the differences between member electoral systems are not decisive for the differences that arise in the context of voter turnout in the Member States. Influences concerning the individual setting of individual voters are proving to be more important factors. The diploma thesis proceeds as follows. First, the next chapter summarizes the differences between the electoral systems in the Member States and also the influences that the relevant literature lists as significant for influencing turnout in different electoral systems. Based on the literature review of these influences, the individual hypotheses of the work are determined. In this chapter it is also possible to find a summary of voter turnout in Member States in the European elections. The second chapter summarizes the methodology of the model, explains the individual variables used. The model itself can then be found in the third chapter, together with an evaluation of the hypothesis. The fourth chapter is then devoted to the conclusions of the work. 18 LITERATURE REVIEW 1 Literature review Voter turnout is one of the basic characteristics of elections in advanced democracies. However, since the 1990s, turnout as such has been on a declining trend (Solijonov, 2016). And citizens are going to the polls less and less. What exactly is turnout? It is a manifestation of the participation of citizens in the democracy of their state. Turnout is then measured as the percentage of votes cast (valid and invalid) in the total number of eligible voters. However, it is possible to count eligible voters either on the basis of official state data on registered voters or as a proportion of the population that is entitled to vote on the basis of their age (Solijonov, 2016, p. 17). There are countries where the data on both indicators are almost identical, in other countries the number of registered voters and the number of voters differ significantly. Regarding trends over the last few decades, it must be said that studies mostly examine turnout since the 1980s, including data from the 1950s and later on. In the last seventy years, there have been several significant changes in turnout in the global comparison (Solijonov, 2016, pp. 23 - 29): • There has been an increase in democracies where regular free elections are held. • The number of eligible voters has also increased, many social groups have won the right to vote, and the global electorate population is growing. • In Europe, turnout has been declining since the 1990s; in the US and Asia, turnout has traditionally been lower but more stable. • In the European countries, there are significant differences between advanced democracies and post-communist countries, with turnout declining mainly in post-communist democracies. IDEA (International Institute for Democracy and Electoral Assistance) distinguishes the following factors influencing turnout in countries around the world (Solijonov, 2016, p. 35): • institutional: electoral system, obligation to vote, frequency of elections, obligation to vote, election conditions (election days, time, use of technology...), • socio-economic: population of the country, population growth vs. demographic decline, economic development, 19 LITERATURE REVIEW • political: proximity of elections and elected participants, perception of political issues, election campaign expenditures, political frag- mentation, • individual: age, education, personal political participation, civil ser- vice. The thesis will focus mostly on the institutional and socio-economic factors. And the specified model is focus mainly on accessing the institutional differences between the EU Member States. 1.1 Voting systems for the European Parliament elections in the EU Member States The procedures for electing the European Parliament are governed both by European legislation defining rules common to all Member States and by specific national provisions, which varyfrom one state to another (The European Parliament: electoral procedures, 2020). According to the European legislation only framework rules are set for the electoral procedure for the elections to the European Parliament. This framework is to be modified by the Member States and the final rules are set by the Member States. The common framework is set in Article 14 of the Treaty on European Union (TEU); Articles 20, 22 and 223 of the Treaty on the Functioning of the European Union (TFEU) and Article 39 of the Charter of Fundamental Rights. 1.1.1 Common rules Originally the members of the European Parliament were appointed by the national parliaments. The direct elections of the members of the European Parliament were implemented in 1976 and first direct elections were held in 1979. The direct universal suffrage was the first common principle for the election. This brought a more democratic parliament. According to the amended 1976 Electoral Act, the voting system is proportional and either the list system or the single transferable vote system has to be used (The European Parliament: electoral procedures, 2020). The common procedure was brought by the Maastricht Treaty. It first suggested the uniform procedure but as this was not passed thought 20 LITERATURE REVIEW the Council only common principles were adopted. Only few uniform principles were adopted: proportional representation a few incompatibilities between national and European mandates for the member of the European Parliament. The latest amendments in 2018 (Council Decisions, 2018) brought more variety to the methods of voting. It is possible to vote remotely using the internet and online tools, or postal voting. Variety was also brought on thresholds, on possibility of the voting in the third countries and on other issues concerning for example the data protection or double voting penalization. According to common rules all citizens of the European Union have the rights to vote and to stand as a candidate even in other country than their original. As long as they fulfil the conditions for voting or to be voted for in the country of the residence. The election date for the EP elections is set by the Council and approved by the Parliament before the end of its 5-year mandate. This timeframe for the elections is set by the Council. Final election date in the Member States is set in every Member State within the timeframe given by the Council. Common principles for the voting systems are: • proportional representation, • some kind of threshold between 2% and 5% for voting more than 35 seats in a country (first mandatory in the 2024 elec- tions), • direct voting, • right of a non-national to vote and to stand as a candidate as the citizen of the EU, • frame dates for the election days. These principles do not differ in the EU Member States and they are not part of the independent variables in the model. This thesis is going to focus on other parts of the election systems and the differences between countries that are specified in detail in the following sub-chapters. 1.1.2 Rules stated by the EU Member States Rules for the election process are different for every Member State. As long as the proportional system and list or single transferable vote system is used. The Member States do have right to set their own election days, conditions for who can vote and who can be a candidate. 21 LITERATURE REVIEW The differences between the Member States are due to the different traditions for national elections. And the form of the national elections usually set the conditions for the EP elections. What are the differences between the Member States? Electoral systems There are a lot of differences in electoral systems between all the Member States and their national traditions. But all the Member States have to use a proportional representation system for the EP election. The electoral threshold for the system is not obligatory till 2024 but can be set as high as 5%. Currently the following countries implemented some kind of threshold for the elections to the European Parliament (The European Parliament: electoral procedures, 2020): • Belgium, Croatia, the Czech Republic, France, Hungary, Latvia, Lithuania, Poland, Romania and Slovakia apply 5%. • Austria, Italy and Sweden apply 4%. • Greece applies 3%. • And Cyprus applies 1.8%. • Threshold is not applied in Bulgaria, Denmark, Germany, Estonia, Finland, Ireland, Luxembourg, Malta, Netherland, Portugal, Slovenia, Spain and the UK. Most of the EU countries vote in a single constituency, only Belgium, Ireland, Italy and Poland have more than one constituency in the national territory. Lists of candidates are for most countries open and voters can put their preferential votes to change the order in the list and to get the preferred candidate bigger chance to be elected. No preferential vote is allowed in France, Germany, Hungary, Portugal, Romania, Spain. In Malta and Ireland single transferable vote system is used. Compulsory voting Some of the Member States does have a duty to vote. Which is in fact usually stated as a main impact on the voters' participation (see next subchapter) and will be tested in this study. In the EU, voting is obligatory in Belgium, Bulgaria, Cyprus, Greece, and Luxembourg. The form of the duty and the form of the fine for not voting is different in each of these states. The obligation to vote does apply to the citizens of these states and also to other EU countries' citizens registered for elections in these states. 22 LITERATURE REVIEW In other EU countries voting is facultative and there are no fines and consequences for not voting. The election participation is a main right for the citizens but not duty of the citizen. Threshold for the voting age Age requirement for the right to vote to the EP elections is mainly 18 years old for almost all the countries in the EU. Only three countries have traditionally lower age, you can vote at 16 year old in Austria and Malta and you can vote at 17 in Greece (The European Parliament: electoral procedures, 2020). Election days Dates of the elections are set by the European Parliament, but the Parliament sets only the framework and the days can be modified by the national rules for the elections. Election days were set by the Articles 10 and 11 of the 1976 Electoral Act. The elections to the European Parliament are to be held within the same period starting Thursday morning ending Sunday evening (The European Parliament: electoral procedures, 2020). The Member States have right to set their own election days within this timeframe. Election days for the EU have to be set at least one year before the end of the five-year term of the Parliament. The dates are set by the Council and have to be approved by the Parliament. After setting the dates for the EU, the EU Member States have right to set the national dates within the given timeframe. In 2019 dates for the elections were set as from May 23rd to May 26th. National dates were as follows (How to vote, 2018): • European elections started on Thursday May 23rd in the Netherlands and the UK. • On Friday May 24th elections were held in Ireland. • Elections continued on Saturday May 25th in Slovakia and Latvia. • The only country in the EU that has two election days is the Czech Republic. European elections were held on Friday 24th and Saturday 25th. • Most of the countries did vote on Sunday 26th, that goes for: Belgium, Bulgaria, Denmark, Estonia, Finland, France, Croatia, Italy, Lithuania, Luxembourg, Cyprus, Hungary, Malta, Germany, Poland, Portugal, Austria, Romania, Greek, Slovenia, Spain, Sweden. 23 LITERATURE REVIEW How the vote is casted Voters can usually participate in the elections personally, by post, by proxy voting or online. How the elections are organized and how the votes are casted is set by the Member States. In most countries voters have to vote personally by coming to the election rooms and casting their votes. But there are some differences in the Member States (How to vote, 2018). • Only personal voting is allowed in Belgium, Bulgaria, Croatia, Cyprus, the Czech Republic, France, Greece, Hungary, Ireland, Italy, Latvia, Luxembourg, Malta, Poland, Portugal, Romania, Slovenia, Slovakia and Sweden. • In Finland and Lithuania voters can vote in person but not only on the election days but also in advance under given circumstances. • In Netherlands it is possible to vote also in proxy. • Voters can vote in person and also by post in Austria, Denmark, Germany, Spain. • Voters can vote in person, by post and also in proxy in the UK. • Estonia allows e-voting. There are also some differences in whether citizens of the Member State can cast their votes in other country, in other EU country or in the third country. Some EU Member States allow voting from foreign countries, some of the EU Member States do not allow that (How to vote, 2018). • You can vote only within the national borders of the EU Member State in Cyprus, the Czech Republic, Ireland, Malta and Slovakia. • And you can vote also in other countries for your national candidates if you are citizen of Austria, Belgium, Bulgaria, Croatia, Denmark, Germany, Estonia, Finland, France, Greece, Hungary, Italy, Latvia, Lithuania, Luxembourg, Netherlands, Poland, Portugal, Romania, Slovenia, Spain, Sweden and the UK. Ways of voting from foreign countries differ, usually voters cast their votes in the embassy or by post. Some kind of registration is often required for voting from foreign countries. 24 LITERATURE REVIEW Table 1: Turnout in the EP elections in the EU Member States 1979 -1999 (in %) year 1979 1981 1984 1987 1989 1994 1995 1996 1999 Belgium 91.36 92.09 90.73 90.66 91.05 Denmark 47.82 52.38 46.17 52.92 50.46 Germany 65.73 56.76 62.28 60.02 45.19 Ireland 63.61 47.56 68.28 43.98 50.21 France 60.71 56.72 48.80 52.71 46.76 Italy 85.65 82.47 81.07 73.60 69.76 Luxembourg 88.91 88.79 87.39 88.55 87.27 Netherlands 58.12 50.88 47.48 35.69 30.02 United Kingdom 32.35 32.57 36.37 36.43 24.00 Greece 81.48 80.59 80.03 73.18 70.25 Spain 68.52 54.71 59.14 63.05 Portugal 72.42 51.10 35.54 39.93 Sweden 41.63 38.84 Austria 67.73 49.40 Finland 57.60 30.14 Czechia Estonia Cyprus Lithuania Latvia Hungary Malta Poland Slovenia Slovakia Bulgaria Romania Croatia Total EU 61.99 58.98 58.41 56.67 49.51 2019 European elections results (2019) 25 LITERATURE REVIEW Table 2: Turnout in the EP elections in the EU Member States 2004 - 2019 (in %) year 2004 2007 2009 2013 2014 2019 Belgium 90.81 90.39 89.64 88.47 Denmark 47.89 59.54 56.32 66.08 Germany 43.00 43.27 48.10 61.38 Ireland 58.58 58.64 52.44 49.70 France 42.76 40.63 42.43 50.12 Italy 71.72 66.47 57.22 54.50 Luxembourg 91.35 90.76 85.55 84.24 Netherlands 39.26 36.75 37.32 41.93 United Kingdom 38.52 34.70 35.60 37.18 Greece 63.22 52.54 59.97 58.69 Spain 45.14 44.87 43.81 60.73 Portugal 38.60 36.77 33.67 30.75 Sweden 37.85 45.53 51.07 55.27 Austria 42.43 45.97 45.39 59.80 Finland 39.43 38.60 39.10 40.80 Czechia 28.30 28.22 18.20 28.72 Estonia 26.83 43.90 36.52 37.60 Cyprus 72.50 59.40 43.97 44.99 Lithuania 48.38 20.98 47.35 53.48 Latvia 41.34 53.70 30.24 33.53 Hungary 38.50 36.31 28.97 43.36 Malta 82.39 78.79 74.80 72.70 Poland 20.87 24.53 23.83 45.68 Slovenia 28.35 28.37 24.55 28.89 Slovakia 16.97 19.64 13.5.20 22.74 Bulgaria 29.22 38.99 35.84 32.64 Romania 29.47 27.67 32.44 51.20 Croatia 20.84 25.24 29.85 Total EU 45.47 42.97 42.61 50.66 2019 European elections results (2019) 26 LITERATURE REVIEW 1.2 Electoral participation in the European Parliament elections Voter turnout in the European Parliament elections is usually lower than the national elections' voter turnout during national parliamentary elections. The turnout is influenced by the so called second-order elections (Reif a Schmitt, 1980, Hix a Marsh, 2011) and the fact that these elections are usually too distanced from the voter (Malkopoulou, 2009). Turnout has been steadily declining since the first elections in 1979. The first elections had around 60% average turnout and this has been steadily falling down to 42% in 2014. Turnout rose up again in 2019 but it is too early to say if this means a new trend or if this was just a coinci- dence. There are also some differences between countries in the turnout. Traditionally the turnout is higher in the old Member States and lower in the new Member States. These differences were still visible at the 2019 elections and can be seen in Table 1 and Table 2 in the previous sub- chapter. 1.3 Literature review Voter turnout is a highly researched topic. Academic studies look at the topic of turnout in various contexts. Many studies have been conducted in the US environment, both in local and presidential elections. Similarly, many studies have been conducted comparing multiple countries. In his book Comparative Democracy, Powell (1982) described citizens' participation in democratic elections as a fundamental feature of democracy. He then examined (Powell, 1986) how it is possible that Americans are more politically active than citizens of other countries, but turnout in the United States is traditionally lower than in other countries. As part of the study of this paradox, he focused on the cultural and institutional environment and on the personal characteristics of voters. The study was conducted on participation in the 60's and 70's, identifying the factor of voting registration in the USA as essential. A stable environment without civil unrest also contributes to high turnout. Theoretically, it is possible to distinguish the examined influences on turnout into two basic groups (Blais, 1998). The first is the system setting in that country. Influences such as the political and electoral 27 LITERATURE REVIEW system, the institutional setting of democracy and others fall into this category. The second category is the socio-economic and demographic characteristics. These can be examined at the individual-level, i.e. what are the living conditions of a given voter, as well as at the aggregate level as a country-level factors, for example what is the overall economic development of a given country. As a third category, it is possible to add specific EU factors in the case of elections to the European Parliament, such as the length of the EU membership, trust in the EU institutions, the role of second-order elections and more. 1.3.1 Electoral system as a factor influencing turnout One of the fundamental issues addressed by turnout studies is how political institutions and the political and electoral system in general affect turnout. Among the factors that affect it is possible to count: the electoral system, the form of the formed government, bicameral vs. unicameral parliament, federalism, separation of government power and other positions of government institutions (Herrera, Morelli and Nunnari, 2016). In addition, turnout is traditionally higher in electoral systems where there is some form of voting obligation. Turnout then increases in proportion as the cost to the voter increases when he or she cannot vote and must pay a fine (Singh, 2011). Jackman (1987) conducted a study of turnout in the 1970s in 19 countries, and the main institutional influences were: national constituencies, disproportionate election results, multiple political parties in the system, unicameral parliament, and mandatory turnout. Laws restricting the right to vote to selected groups of the population, e.g. in South Africa or Switzerland, have a significant negative effect on turnout (Jackman, 1987, p. 419). Further studies confirm that turnout increases with the size of constituencies, the highest in the national constituency, and also with the proportionality of election results (Blais & Carty 1990, Jackman & Miller 1995, Kostadinova 2003). One of the influences is also the day of the elections. Some studies state that the largest turnout is associated with Sunday as election day (Franklin's, 1996), but others argue (Norris, 2002) that the turnout is not affected by any of the following factors: number of election days, election 28 LITERATURE REVIEW day on a weekend or election day on a working day or alternative voting options, such as postal voting. Blais et al. (2003), on the other hand, state that the possibility of correspondence voting or delegation of voting rights has a positive effect on turnout. Another factor influencing turnout is the so-called electoral context, so how political parties performed in the last elections period and whether there is a change to win (Powell GB, Whitten GD. 1993, p. 397). It is often the case that the ruling party loses in the next elections following the parliamentary national elections, which may be due to lower turnout. 1.3.2 Socio-economic and demographic factors According to MacKuen, Erikson, and Stimson (1992, p. 597), economics moves politics. And the economic results of the national economy affect the elections results. Likewise, economic indicators influence the electoral behavior of the population. In terms of economic indicators, studies most often examine the impact of inflation, unemployment and economic growth on both election results and turnout (Powell GB, Whitten GD. 1993, p. 395). And studies often draw different conclusions. According to some studies, for example, the effect of inflation is negligible and unemployment and industrial production have no effect on turnout (Strom and Lipset, 1984), while other authors acknowledge some effect of inflation, but economic growth has no effect on turnout (Lewis-Beck and Mitchell, 1990). Paldam (1991) more or less denies the effect of inflation on turnout. Most of these studies were conducted on several countries, examining the aggregate level of the turnout. However, it is often the case that if turnout studies try to prove the same in only one country in a series of national elections, there are often different results (Powell GB, Whitten GD. 1993, p. 392). The overall economic development of the country also has an effect on turnout (Blais & Dobrzynska 1998). At the same time, it can be observed that high turnout is associated with the size of the country. The smaller and the more economically advanced democracies, the higher the turnout (Blais, 2008). In addition to economic characteristics, important factors include socio-demographic indicators related to specific voters. Certainly, the most important are age and education (Wolfinger & Rosenstone 1980, 29 LITERATURE REVIEW Blais 2000). The basic personal indicators that influence voter turnout include the age of voters. Turnout increases with the age of the voter (Wolfinger & Rosenstone 1980). Overall turnout thus varies depending on whether the voting age threshold is 18 or 21. Blais & Dobrzynska (1998) even state that raising the threshold for exercising the right to vote to 21 leads to a reduction in turnout by 5 percentage points. Recent studies also look at the impact of turnout itself (in the past) on the future turnout. The influence of the latest generation of voters, who have not yet adopted a specific voting habit (to vote vs. not to vote), is considered important. These studies thus focus on the current electoral environment in interaction with newly entering groups of voters (Franklin, 2004). 1.3.3 Factors influencing the European elections According to the Eurobarometer Survey 91.5 of the European Parliament (2019), one of the key factors influencing the European elections is the attitude of voters towards the EU as such. One of the main factors is whether the voter believes that his/her vote really has its weight in the European Union. This is a factor that can also be traced in studies dealing with national elections (see Blais & Carty 1990, Jackman & Miller 1995, Kostadova 2003), where turnout is growing along with citizens' confidence that the vote will have weight in the post-election distribution of power. However, in the context of the European elections, the Eurobarometer Survey 91.5 of the European Parliament (2019) also lists factors such as satisfaction with the EU itself, whether the EP elections have become a permanent part of political life, and the length of time the country is a member of the EU. Participation in the EP elections often suffers from so-called secondorder elections, where elections to a more distant institution (the European Parliament) are perceived as inferior to elections to national institutions (Reif and Schmitt, 1980, Hix and Marsh, 2011). In the case of the EP elections, the time distance of the EP elections themselves to the next national elections is also an important factor. According to Malkopoulou (2009), European Parliament elections are mainly influenced by factors associated with the electoral system and, above all, by the fact that Brussels is too far from ordinary voters. Turnout then increases when EP elections are linked to other elections. 30 LITERATURE REVIEW And turnout in the European elections is declining as the European Parliament has little confidence in the population (Clark, 2014). In addition, the last European elections in 2019 significantly affected the turnout of young voters, which in the overall assessment of turnout also increased the percentage of voters who went to the polls. 1.3.4 Country-level factors To look on the two-level logic of the influences: we can distinguish influences on the country-level and on the individual-level of the voters. If we look and evaluate the factors influencing the EP elections on the countrylevel, the most important factors are the voting systems and probably also economic conditions in the EU Member states. However, authors can agree that economic has some effect on the turnout (Blais & Dobrzynska 1998), probably GDP and economic growth and mostly good economic situation could have some effect on the turnout (Powell GB, Whitten GD. 1993, p. 395), however, it is with minor impact. The voting systems are what matter more. These factors include for example voting age (Blais & Dobrzynska, 1998), compulsory voting (Singh, 2011), election days (Franklin's, 1996), possibility to vote by post (Blais et al., 2003) and others. These could be the most influential factors for the turnout. However, some authors argue that the characteristics of the voting systems have no impact on the turnout at all (Norris, 2002). Factors connected with the EP elections are usually connected to the second-order election characteristic and also to the distance of the EU administration. The specific country-level factor connected with the EU is usually the country's length of membership, the longer the country is in the EU, the more accustomed to vote voters are, the higher the turnout (the Eurobarometer Survey 91.5 of the European Parliament, 2019). 1.3.5 Individual-level factors Among the most important factors that influence voting behavior at the individual-level are mainly age and education (Wolfinger & Rosenstone 1980, Blais 2000.) And with increasing age and education, the probability that the voter will go to the poles increases. Education increases the cognitive abilities of the voter, enables him/her to better interpret political events and stimulates voters to participate more in political life. Fowler (2013) examines the question of the effects on turnout in five case studies, ranking among the individual factors the age, education, 31 LITERATURE REVIEW economic status of the voter and whether the voter is a regular voter. Voter turnout in elections is also affected by whether another social event that interests voters takes place within reach, thus reducing voters' determination to go to the polls. A political interest and civil duty exercised by the electorate are also among the major individual factors as found out by Danny and Doyle (2008) . The voter's social participation in political and social life in the system increases the likelihood that the voter will go to the polls. And voter turnout is also affected by a voter's personality, with aggressive personality and comprehension skills increasing the likelihood that a voter will go to the polls. The main factor at the individual-level then includes whether the voter is a regular voter. This factor is mentioned by many authors (Fowler, 2013, Danny and Doyle, 2009). According to Danny and Doyle (2009), participation in one election increases the probability that a voter will go to the polls by another 13% while maintaining the same cir- cumstances. Other factors that positively influence voter turnout include the following (Kavanagh and Durkan, 2020): marriage, home or apartment ownership, stable permanent residence, agricultural employment, social well-being, geographical proximity of the voter to polling stations, political awareness and policy monitoring. 1.4 The hypothesis and the research question of this thesis The research goal was set as followed: to evaluate which factors had an impact on the turnout in the 2019 EP elections. And the main purpose is to answer the question how the different voting systems in the EU Member States are important for the turnout and how these differences can influence the turnout of the voters in the EP elections. The hypotheses do not address the most significant feature of the EP elections, namely the fact that these are the so-called second-order elections. This factor is cited by many authors as the most important for influencing turnout in these elections. However, this says nothing about the differences between the Member States and their electoral systems. The hypotheses of this thesis instead focus on individual differences between electoral systems. Since turnout is tested on individual answers 32 LITERATURE REVIEW in a questionnaire, the hypotheses are based on testing two different factors. The individual ones that many authors cite as important for turnout, such as age, education, interest in politics and more. However, the voter lives in the Member State and in the living conditions provided by that Member State. This may further influence his decision to vote. Therefore, factors that deal with electoral systems and overall conditions in the Member States are further tested. Such as the obligation to vote, national election days, economic conditions in a given state and others. While the thesis is going to build on the existing literature and the main known factors for voters' turnout, the main emphasis is going to be on the two-levels model (individual and national factors). And the research question is to answer whether the voting system's circumstances are or are not the main influence on the turnout in the elections. The thesis should bring an answer if the differences between the EU Member States' voting systems are causing the differences in the turnout. Other differences between the Member States as the economic conditions or position in the economic integration will be also taken into account. The same goes for individuals' preferences to vote or not vote caused by the socio-economic conditions of the individual. These questions can be summarized in the following hypothesis. Most of the relevant literature lists basic socio-economic characteristics as important for a voter's decision whether to vote or not to vote. These include, in particular, the age of the voter and his education (Wolfinger & Rosenstone 1980, Blais 2000). And it is these individual factors that should most significantly influence turnout. Which brings the first hypothesis of this thesis. Hla Individuals with higher age were more likely to turn out in the 2019 EP elections than individuals with lower age. Hlb Individuals with higher education were more likely to turn out in the 2019 EP elections than individuals with lower education. Other than the individual characteristics the electoral system was cited as one of the main influences. Turnout is associated with scheduled election days (Franklin's, 1996) or with the ability to vote not only in person but also by correspondence or proxy (Blais et al., 2003). The hypothesis thus tests the fundamental differences between electoral 33 LITERATURE REVIEW systems and assesses whether these factors are relevant to turnout. Factors such as election days, voting methods or the minimum age for passive participation in elections will be tested (Wolfinger & Rosenstone 1980, Blais & Dobrzynska, 1998). And are part of these two hypotheses. H2 Characteristics of the voting systems did have impact on the probability of voters' turnout in the 2019 EP elections. H3 Voters in the Member States with compulsory voting were more likely to turn out in the 2019 EP elections than individual in countries where there is not compulsory voting. In addition to the electoral systems, the voter is also affected by the overall living conditions in his country, as overall economic conditions can have an effect on turnout according to Blais & Dobrzynska (1998). This hypothesis will further assess the importance of Member States' economic performance, such as GDP growth or unemployment. H4a Voters living in an economic system characterized by higher economic growth were more likely to turn out in the 2019 EP elections. H4b Voters living in an economic system characterized by low inflation were more likely to turn out in the 2019 EP elec- tions. H4c Voters living in an economic system characterized by low unemployment rate were more likely to turn out in the 2019 EP elections. The effect of the so-called second-order elections is one of the main specifically EU factors. It will not be tested, because the data comes only from the EP elections. However, other factors related to the specific shape of the EU will be tested. Such as voter satisfaction with the European Union, the length of a country's membership in the EU (Eurobarometer Survey 91.5 of the European Parliament, 2019) or voter confidence in the democracy of the European Union (Clark, 2014). 34 LITERATURE REVIEW H5 The longer period a country is a member of the EU the more likely voters from this country did vote in the 2019 EP elections. H6a Voters with higher confidence in the EU were more likely to vote that voters with lower confidence in the EU. H6b Voters with higher satisfaction with the EU were more likely to vote that voters with lower with the EU. It is expected alongside with all hypotheses that in general individual-level factors will have stronger influence on the voter turnout that the differences between voting systems and economic situation. According to the literature: the age can be expected as the most influential variable. Other individual characteristics are not to have the same influence as the age. The compulsory voting can be expected as the most prominent variable on the aggregated level and the effect of the economic variables is expected to be marginal. From the EU specific variables, the strongest influence is to be seen connected with number of years a country is a member of the EU. Next chapter is going to go through the methodology of the model and chapter three focuses on the real data from the 2019 EP elections and is going to test the hypothesis through an econometric model (specifically the Multilevel Generalized Linear Model for Dichotomous Outcome Variable Random Intercept Logistic Regression). 35 METHODOLOGY OF THE MODEL AND ITS DATA 2 Methodology of the model and its data Data for the analyses come from various sources. Main part of the data and all data for individuals are from survey: the European Parliament Election Study 2019 (Schmitt, Hermann, Hobolt, Sara B, van der Brug, Wouter and Popa, Sebastian A, 2019). This survey, held after the elections, collects people's political attitudes that may be related to their behavior in the European elections. It also collects various data on key sociodemographic variables. The survey is traditionally executed after every EP election. Data needed for specification of the voting systems originate mainly in official EU information about the elections (How to vote, 2018). Data for economic variables are taken from the Eurostat (2020a, b, c). Independent variables were coded as stated below and all data were also centered with grand-mean centering.1 2.1 Methodology and variables 2.1.1 Dependent variable The dependent variable of the model is defined as voted or not voted, it shows weather an individual did or did not voted in the 2019 EP elections. It is a binary variable, where 0 means not voted and 1 stands for voted in the elections to the European Parliament. Data for this variable are taken from the European Parliament Election Study 2019 (Schmitt, Hermann, Hobolt, Sara B, van der Brug, Wouter and Popa, Sebastian A, 2019). The question wording is as follows: "The European Parliament elections were held on the [insert correct date for each country]. For one reason or another, some people in the UK did not vote in these elections. Did you vote in the recent European Parliament elections? yes, voted/ no, did not vote / dk." Entries for dk (I don't know) was omitted from the analysis (408 entries in total). 1 Centering the variables simply means substracting a mean of the variable's values from every value of the variable. It shifts the scale of the variable without impact on the variable. It allows better comparison between all variables in the model. 36 METHODOLOGY OF THE MODEL AND ITS DATA 2.1.2 Independent variables of the model There are two sets of independent variables: 1) Variables at the individual-level of analysis, 2) variables at the country-level of analysis. Data on the individual-level were drawn from the European Parliament Election Study. Data on the aggregate country-level are mainly from Eurostat and official information of the EU (How to vote, 2018). Data for all independent variables were centered, grand mean centering was used for binary, categorical and continuous variables. The indicators from the questionnaire The following variables are for individual-level of the voters and are drawn from the European Parliament Election Study. These variables are: • Gender. Variable goes as 1 for male, 2 for female and 3 for other. • Age. States age of the person reached in 2019 in full year. • Education. This indicates highest received education. It goes from 1 for elementary school, 2 for high school, 3 for higher education or still studying. • Interest in elections. Variable goes from 0 (interested not at all) to 10 (interested a lot). • Interest in politics. This indicator varies between 0 as not at all interested to 3 as interested very much. • Right/left ideology. This indicator shows what political orientation has the voter. The variable goes from 0 as for left to 10 as for right. The higher value the more right a person identifies him/her- self. • Economic situation. The indicator for the economic situation goes from 0 as for lot worse to 4 as for a lot better, compared to the situation before one year. • Employment. This indicator is a categorical variable and shows what is the employment of the person. Variable goes from 1 till 7: 1 self-employed, 2 employed, 3 in school, 4 working in the household, 5 retired, 6 unemployed, 7 others. • Satisfaction with the EU as whole. The variable goes from 0 as not satisfied at all to 3 as very satisfied. The higher value the more satisfaction with the EU. • National election. Showing the participation in the national election. This is a binary variable that goes from 0 not voted to 1 voted in the national elections. 37 METHODOLOGY OF THE MODEL AND ITS DATA • Support for EU membership. The variable goes from 0 for the EU as a bad thing to 2 as EU as a good thing. The higher value the better attitude to the EU. Indicators at the country-level Variables at the country-level are mainly related to the national election rules. The voting system is proportional for the EU as whole and there are only some minor differences between countries. But some parts differ. For example, countries vary in the number of and the character of election days, the possibilities how to vote, or the age limit for executing the right to vote. (See chapter 2 for more details.] There are also differences in attitudes of the country to the EU and in economic conditions in the country. All these are common for large part of the voters as they are part of the country. The variables were collected for 28 countries (as the UK was part of the 2019 election) and are part of the random effect of the model. • Compulsory voting. Variable is binary and goes from 0 for facultative and 1 for mandatory voting in the national system. These data are taken from the official site of the EU about the elections in each and every EU Member State (How to vote, 2018). • Voting days. This variable is a series of categorical variables indicating how voting days are organized. There are four voting days and five different voting-day setups. The Czech Republic is the only one with two voting days (Friday and Saturday). Most of the EU states do vote on Sunday. Data origin from official web site about the 2019 elections (How to vote, 2018) and can be found in table 5. • Distant voting option. Most countries do have only personal voting, but it is possible to vote even online (Estonia) or by post in some countries. Data comes from the official site about the elections (How to vote, 2018). This variable is binary and goes 0 for voting only in person and 1 for voting in person or in another way. • Expat voting. The EU Member States limit possibility to vote for national candidates from other foreign country (the EU or a third country). Some states allow to vote in its embassy or by post. Data comes from official EU site about the 2019 elections (How to vote, 2018). This is a binary variable, 0 for not possible voting from other country, 1 for possibility to vote abroad. • Early voting age. Most countries do have an eligible age set on 18. But some of them have lower age limit. Data came from official 38 METHODOLOGY OF THE MODEL AND ITS DATA information from the European Parliament (The European Parliament: electoral procedures, 2020). It is a binary variable, 0 for voting only at the age of 18 and more, 1 for voting possible at lower age (16 or 17). • Population. Number of eligible voters in the country. Data came from European Elections Study 2019 Technical Report (2019). • Economic variables: employment, economic growth, inflation. Data came from Eurostat, in actual values. • Economic growth as change from the previous year for first quarter of 2019, current price (Eurostat, 2020a), • Inflation as change for one year, data from May 2019 (Eurostat, 2020c). • Unemployment as % of active population, data from May 2019 (Eurostat 2020b), • EU membership length. The variable is in actual years of time as a member, that's going to be finished or have been finished in 2019 (Countries: The 27 member countries of the EU, 2020). 2.1.3 Summary statistics for the indicators Detailed descriptions of variables as well as their descriptive statistics for all variables can be found in the following tables. Table 3 shows the dependent variable, Table 4 the independent individual variables, and Table 5 the independent variables on the country-level. Table 3: Summary of the dependent variable in the model DEPENDENT VARIABLE Percent Q6: The European Parliament elections n o t v o t e d 23,71% voted/not were held on the [insert correct date for voted each country]. Did you vote in the recent European Parliament elections? voted 76,29% Own calculation, original data: Schmitt Hermann, Hobolt, Sara B, van der Brug, Wouter and Popa, Sebastian A, 2019 39 METHODOLOGY OF THE MODEL AND ITS DATA Table 4: The independent variables (individual-level) in the model INDEPENDENT VARIABLES: individual-level Code Question Values Percent 1 (male) 52,37% Gender (SEX) D3: Are you... 2 (female) 3 (other) 47,55% 0,08% recoded as how Age (AGE) D4_l. What year were you born? old a voter is in 2019 in full years 100,00% 1 primary educa- tion 1,67% Education (EDU) D2_l: How old were you when you stopped full-time education? 2 secondary edu- cation 40,12% 3 higher education 58,21% or still studying 58,21% 0 not at all 8,00% 1 3,92% 2 5,61% 3 7,51% Interest in Q8: How closely did you follow the campaign ahead of the European Parliament 4 c 6,94% 14,64% 11,80% 14,77% 12,40% 5,96% 8,46% election (CAM) elections in the media or on social me- dia? o 6 7 8 9 10 very closely 6,94% 14,64% 11,80% 14,77% 12,40% 5,96% 8,46% 0 not at all 7,17% Interest in po- Q21: To what extent would you say you 1 a little 28,51% litics (INT) are interested in politics? 2 somewhat 3 very 45,87% 18,45% 0 a lot worse 9,75% Economic situation (ECO) Q19: What do you think about the economy? Compared to 12 months ago, do you think that the general economic situation in the country? 1 a little worse 2 stayed the same 3 a little better 4 a lot better 19,14% 38,10% 28,13% 4,88% 40 METHODOLOGY OF THE MODEL AND ITS DATA INDEPENDENT VARIABLES: individual-level Code Question Values Percent Right/Left ideo logy(LER) Q l l : In political matters people talk of "the left" and "the right". What is your position? Please indicate your views using any number on an 11-point-scale. On this scale, where 0 means "left" and 10 means "right," which number best describes your position 0 for left 1 2 3 4 5 6 7 8 9 10 right 6,15% 3,11% 5,78% 8,16% 8,08% 28,55% 9,70% 10,54% 8,24% 3,51% 8,17% 1 self-employed 2 employed 3 in school Employment D6: What is your current work situation? 4 working in the (WOR) Are you: [as stated] household 5 retired 6 unemployed 7 other. 8,00% 49,70% 6,76% 3,61% 23,96% 5,74% 2,21% Satisfaction with the EU {SAT) Q4: All in all again, are you very satisfied, fairly satisfied, not very satisfied or not at all satisfied with the way democracy works in the European Union? 0 not at all satisfied 16,25% 1 not very satisfied 37,73% 2 fairly satisfied 41,24% 3 very satisfied 4,78% National election (NAT) Q9: Which party did you vote for at the [General Election] of [Year of Last General Election]? 0 didn't vote 1 voted 11,59% 88,41% 0 a bad thing 14,52% , , Q22: Generally speaking, do you think support tor the [ c o u n t r y ] membership of the Euro- 1 neither 21,77% EU mem. fATT) L T T . . _ H pean Union is..? 2 a good thing 63,72% Own calculation, original data: Schmitt, Hermann, Hobolt, Sara B, van der Brug, Wouter and Popa, Sebastian A, 2019 41 METHODOLOGY OF THE MODEL AND ITS DATA Table 5: The independent variables (country-level) in the model INDEPENDENT VARIABLES: country-level Code Measurement Values Percent Compulsory vo- Is voting obligatory in the national voting 0 not obligatory vo- ting 84,36% ting (OBL) system? 1 obligatory voting 15,64% 1 weekend (Saturday May 25th or 87,01% Sunday May 26th), 2 two voting days Voting days T A n , , (DAY) What are the voting days? (Friday May 24th and Saturday May 25th), 3 weekday 1,83% (Thursday May 23rd 11,16% or Friday May 24th) 0 only in person 66,34% Distant voting „ , . . n . r . * r ? How can be a vote casted? option (WAY) 1 in person + other (mail, in advance, proxy, e-voting) 33,66% Expact voting Can the vote be casted from other coun- 0 not possible 14,63% (FOR) tries? 1 yes 85,37% Early^votinga £ e ^ j i a t a g e a v o t e r j s eligible to vote? 0 only at 18 1 lower age 90,35% 9,65% ^ 0 ( P O ^ ) 0 n ^ o w W S is the population of the country? number of eligible voters in mil. 100,00% Economic How big is the GDP growth in the counin %,annual change 100,00% growth(GDP) try? in %,annual change 100,00% . n f I . , m How big is the inflation rate in the counInflation fINF) „ & try? in %,annual change 100,00% Unemloyment How big is the unemployment rate in the (UNE) country? %, unemployed people to total working force 100,00% EU mem. lenghtHow long was country a member of the (MEM) EU? in whole years 100,00% Own calculation, original data: How to vote (2018), The European Parliament: electoral procedures (2020), Eurostat (2020a, b, c) 42 MODEL AND ITS FINDINGS 3 Model and its findings The hypotheses of this paper are to be tested by logistic regression, twolevel generalized linear model for dichotomous outcome variable, with random intercept. The choice of model corresponds to independent variables and established hypotheses, which take into account both individual influences that directly concern the voter and factors, which are based on the specific electoral systems of the EU Member States. The model has a dependent variable in binary form: voted - did not vote. Therefore, a logit model with a dichotomous outcome variable is chosen. The structure of the data - with some variables at the individuallevel and some variables at the country-level - suggests that a multilevel model will be needed. Data were first tested with a simple pooled logit model and the Hosmer-Lemeshow Goodness of Fit test showed that the model with one level of data is not a good fit as the national level is omitted from such an analysis. Simple logit model also does not allow to evaluate importance of the voting systems in the Member States. The twolevel hierarchical model allows to evaluate the individual factors and also the factors of the election. Given the binary nature of the dependent variable, the following multilevel binary model was used: Multilevel Generalized Linear Model for Dichotomous Outcome Variable (Random Intercept Logistic Regression.) Two levels include an individuals' level 1 and countries of the EU as level 2. All variants of the model were estimated in R and programmed in R studio. The codes for the models are part of the attachment of this thesis. 3 models were specified in the analyses, with the third one having two modification. Model 1 deals with evaluating the hypotheses from the view of variation on the individual and country-level factors. This model is a simple glmer model with intercept only, that evaluates the influences of the Member States. This model only distinguishes that there are differences between countries but cannot specify the factors causing the differences. And this model is actually a quite good fit. However, this model showed that only 8 % of differences in voters' participation can be explained by the country-level variables defined in the model and differences between the EU Member States. It points to the fact that differences at the individual-level and differences between voters are more important than the voting systems 43 MODEL AND ITS FINDINGS and other countries' specific variables. It shows that the differences between the voting systems and economic situation of the EU Member States do cause some differences in the voter turnout, but these are marginal influences and individual characteristics of a voter are able to explain much more of the variation in the voter's turnout. However, this support the hypothesis H2 and partly H3, that the electoral systems and economic conditions do have some influence on the turnout. Summary for the model 1 can be found in Table 6. Table 6: Summary of the model 1 VOTED Predictors Odds Ratios CI p (Intercept) 3,47 2,82-4,27 <0,001 Random Effects o2 3,29 TOO hCountry 0,31 ICC 0,08 N hCountry 28 Observations 20785 Marginal R2 / Conditional R2 0,000 / 0,085 Own calculation Finding out which variables explain differences in the turnout is the next task in evaluating the hypotheses. Model 2 therefore adds variables at the country-level and variables at the individual-level. Therefore, this model is a two-level random intercept model, glmer model in R. Variances in the model are in the individual characteristics of the voters and in the voting systems, the economic differences and the differences in integration process in the EU Member States. Summary for model 2 can be found in Table 7 and Table 8. Model 2 includes all the variables identified by the methodology of this thesis. This model also has actually worst fit than the previous model and is singular. Which points to the fact that there is too much variables that do not offer a large variability of the outcome and that a simpler model with less variables could be a better fit for the data. 44 MODEL AND ITS FINDINGS Table 7: Summary of the model 2, odds ratios for the independent variables VOTED Predictors Odds Ratios CI P (Intercept) 4,65 4,12--5,26 <0,001 Gender 0,96 0,89--1,04 0,287 Age 1,01 1,01--1,01 <0,001 Education 1,27 1,18--1,37 <0,001 Interest in election 1,22 1,20--1,23 <0,001 Interest in politics 1,27 1,21--1,34 <0,001 Right/Left ideology 1,02 1,00--1,04 0,01 Economic situation 1,03 0,99--1,07 0,213 Employment 1,01 0,98--1,03 0,64 Satisfaction with the EU 1,10 1,04--1,17 <0,001 National election 9,36 8,40--10,42 <0,001 Support for the EU mem. 1,26 1,19--1,33 <0,001 Compulsory voting 1,80 1,11--2,90 0,016 Voting days 1,57 0,68--3,64 0,291 Distant voting option 0,63 0,51--0,78 <0,001 Expact voting 1,20 0,94--1,52 0,143 Early voting age 1,28 0,77--2,13 0,333 Population 1,00 1,00--1,01 0,547 Economic growth 1,14 1,04--1,25 0,004 Inflation 1,13 1,00--1,29 0,057 Unemloyment 1,04 1,01--1,08 0,005 EU mem. lenght 1,02 1,01--1,03 <0,001 Own calculation 45 MODEL AND ITS FINDINGS Table 8: Summary of the model 2, country-level variations for the independent country-level variables Random Effects a2 3,29 TOO hCountry 0,00 T11 hCountry.compulsory voting 0,33 T11 hCountry.voting days 0,61 T11 hCountry.distant voting option 0,03 T11 hCountry.expact voting 0,05 T11 hCountry.early voting age 0,23 T11 hCountry.population 0,00 T11 hCountry.economic growth 0,01 T11 hCountry.inflation 0,01 T11 hCountry.unemployment 0,00 T11 hCountry.EU mem.lenght 0,00 P01 0,09 0,02 0,10 0,17 -0,06 0,01 -0,06 -0,11 -0,11 -0,12 N hCountry 28 Observations 20785 Marginal R2 / Conditional R2 0,350 /NA Own calculation 46 MODEL AND ITS FINDINGS The poor fit of Model 2 suggests that there are not enough differences between the election systems for the EP elections in the EU Member States. That further suggests that even with given differences in the national voting system the effects of the voting systems for the European Parliament are, in fact, mainly uniform in all Member States. In other words, cross-country differences in voting system in the 2019 EP elections did not have a significant systematic influence on voter turnout. Differences between voting systems and the differences in economic situation across countries are so small that only a small part of the variance can be explained by them. This is probably due to only small deviations between the EU Member States concerning the voting systems for EP elections. For example, only 5 of the 28 states have mandatory participation in elections, only 3 states have a reduced age threshold for exercising the right to vote, only 5 states are not allowed to vote from abroad and only 9 states allow an alternative voting option other than personal. The effect of this variances is scattered in the data as the deviation from average are not in the same states. It shows that hypothesis H2, stating that the voting systems have impact on the turnout, is supported by the data. However, the effect of voting systems on the turnout is minimal and not essential. The same is essentially true for economic conditions, where the EU Member States are currently so economically interconnected that differences between GDP growth, unemployment and inflation are minimal and are essentially limited to differences between the southern states (especially Greece, Italy, Spain) and the rest of the EU. On Figure 1: Odds ratios for independent variables in model 2 we can find the odds ratios for all variables in model 2. Odds ratio shows how much the dependent variable is going to change depending on the independent variable. As the dependent variable is binary (did vote or did not vote) the odds ratio says how much the independent variable increase or decrease the probably that a voter will go to the polls. Two variables are independent with odds ratio at 1, the odds ration larger than one indicates positive correlation and the odds ration smaller than one indicates negative correlation. The larger the deviation od the odds ration from 1, the larger the dependency between the variables. 47 MODEL AND ITS FINDINGS Figure 1: Odds ratios for independent variables in model 2 V O T E D GenderA ~ m * • Age • • • 1 1 | Education • • • • 1 1 | Interest in elections Interest in politics Right/Left ideology Economic situation Employment - Satisfaction with the EU National elections Support for the EU mem." Compulsory voting - • • • 1 1 | Interest in elections Interest in politics Right/Left ideology Economic situation Employment - Satisfaction with the EU National elections Support for the EU mem." Compulsory voting - 4 • • • 1 1 | Interest in elections Interest in politics Right/Left ideology Economic situation Employment - Satisfaction with the EU National elections Support for the EU mem." Compulsory voting - 1 t • • • 1 1 | Interest in elections Interest in politics Right/Left ideology Economic situation Employment - Satisfaction with the EU National elections Support for the EU mem." Compulsory voting - < • • • 1 1 | Interest in elections Interest in politics Right/Left ideology Economic situation Employment - Satisfaction with the EU National elections Support for the EU mem." Compulsory voting - P • Interest in elections Interest in politics Right/Left ideology Economic situation Employment - Satisfaction with the EU National elections Support for the EU mem." Compulsory voting - P • Interest in elections Interest in politics Right/Left ideology Economic situation Employment - Satisfaction with the EU National elections Support for the EU mem." Compulsory voting - • Interest in elections Interest in politics Right/Left ideology Economic situation Employment - Satisfaction with the EU National elections Support for the EU mem." Compulsory voting - • Interest in elections Interest in politics Right/Left ideology Economic situation Employment - Satisfaction with the EU National elections Support for the EU mem." Compulsory voting Voting days Distant voting option Expact voting Early voting age Voting days Distant voting option Expact voting Early voting age Voting days Distant voting option Expact voting Early voting age Voting days Distant voting option Expact voting Early voting age - | Population • Economic growth Inflation - \ V • • I Population • Economic growth Inflation - V • • I Population • Economic growth Inflation - V • • I Unemployment • { I V • • It u mem. lengnt \ V • • I 0.1 0.5 1 1 1 1 s 10 50 Odds Ratios Own calculation You can see from the Figure 1: Odds ratios for independent variables in model 2 that most of the variables do not have a significant influence on the turnout. Only few variables have bigger impact on the turnout. The most important one being if the voter is a regular voter. Other significant variables are: compulsory voting, voting days, distant voting options, education, interest in elections, interest in politics, support for the EU mem. and early voting age. There are all variables from model 2 in the figure, but as you can see from summary of the model (Table 7 and Table 8] some of the variables are statistically not significant. Summary of the selected variables from the model can be found below. It was stated in the first hypothesis H1 that age (H la) and education (Hlb) influence the voter turnout. The higher the age and the education, the bigger the turnout. Model 2 shows that the hypotheses are valid, odds ratio for voting goes as high as 1,27 for education, meanwhile stays at 48 MODEL AND ITS FINDINGS 1,01 for age. Predictions for these variables and their influence on the turnout can be found in Figure 2. Figure 2: Predictions of selected variables' impact in model 2 (Age, Education, Satisfaction with the EU, Support for the EU mem.) Impact of Age on probability of voting Impact of Education on probability of voting 74%- 1 -1.5 -1 0 -0.5 0.0 05 Support for the EU mem. Figure 2 shows how the probability of voting increase if there is a change in the stated variable (Age, Education, Satisfaction with the EU and Support for the EU mem.), while all other variables stay fixed. The odds ratio for education is higher than for the age of the voter, but the age of the voter is on a much bigger scale so the significance of the rising probability can be really noticeable after few decades. Satisfaction with the EU Own calculation 49 MODEL AND ITS FINDINGS Other individual variables can be found in the hypothesis H6. Hypothesis H6a says that the more positive attitude to the EU and the bigger the satisfaction with the E (H6a), the more probably a voter was going to vote in 2019 EP elections. Both of these variables in model 2 confirm this hypothesis. Visualization of these two variables can be found also in Figure 2. If we look on all the variables on the individual-level in Model 2 we could see that the most important variable influencing the odds of voting is whether the person did or did not voted in the last national elections (variable NAT). The odds for probability of voting in 2019 EP elections are more that 9. Meaning that person regularly going to vote is more likely to vote also in the EP elections. Compared to regular turnout, other variables at the individual-level are almost negligible. Visualization of this variable can be found in Figure 3. A quick look at the country-level variables seems to point out that the most important variables are the obligation to vote, voting days and method of voting. This shows that differences between electoral systems also partially affect turnout, which confirms hypothesis H2 (that electoral systems have an impact on turnout). But the impact is not highly significant as discussed above. Other country-level variables are rather negligible due to their influence. Economic variables (GDP growth, unemployment, inflation) have a rather negligible effect, the biggest one can be found with GDP growth. These variables and their predictions of impact on the voters' turnout can be also found in Figure 3. 50 MODEL AND ITS FINDINGS Figure 3: Predictions of selected variables' impact in model 2 (National elections, Economic growth, Inflation, Unemployment) Impact of National elections on probability of voting Impact of Economic growth on probability of voting 75*-2 -1 0 1 2 -5 • 5 10 Inflation Unemployment Own calculation Figure 4 shows the predictions for two country-level variables. Compulsory voting does have some effect on the turnout, but it is rather neglible. As is the difference between the length of the membership of the Member States in a short period, more significant influence can be found in a longer period of time and between old and new EU Member States. That confirms the hypothesis H5, that longer time of membership, the higher voter turnout in that EU Member State. It also confirms the hypothesis H3 that the obligatory voting in the voting systems do have positive impact on the turnout. 51 MODEL AND ITS FINDINGS Figure 4: Impact of the variable Compulsory voting and the variable EU mem. length on the probability of voting Impact of Compulsory voting on probability of voting Impact of EU mem. lenght on probability of voting 0 0 0.3 0 6 -20 0 20 Compulsory voting EU mem. lenght Own calculation As the model 2 is singular and the variances in the model shows that the model is nearly unidentifiable let's simplify the model. Model 3 does contain only most important variables, as stated by the relevant literature. This model is specified as the previous model but does include only variables that the relevant literature stated as the most possible to influence the voter's turnout. These variables are the following: education level, interested in politics (measured by the interest in campaign), regular voter (measured by voting in the last national elections), attitude to the EU, obligation to vote and how long the country is a member of the EU. This model 3 then includes six independent variables and one nesting on the country-level. Four independent variables are part of the individual-level and two variables are part of the second level of the model. Two-level logistic regression with random intercept and fixed effect of the model is used. To better estimate the effects of the country-level variables model 3 does have two modification, model 3a deals with obligation to vote as a country-level indicator (hypothesis H3), model 3b deals with length of the EU membership as a country-level indicator (hypothesis H5). Summary of both variations can be found in the following tables. Table 9: Summary of the model 3 a shows the variables and summary of the model 3a. 52 MODEL AND ITS FINDINGS Table 10 shows the variables and summary of the model 3b. Table 9: Summary of the model 3a VOTED Predictors Odds Ratios CI P (Intercept) 4,36 3,63--5,24 <0,001 Education 1,22 1,13--1,31 <0,001 Interest in election 1,26 1,25--1,28 <0,001 National election 10,32 9,28--11,47 <0,001 Support for the EU mem. 1,34 1,27--1,41 <0,001 Compulsory voting 1,57 0,80--3,09 0,194 Random Effects a2 3,29 TOO hCountry 0,17 T11 hCountry.compulsory voting 0,56 P01 hCountry -0,04 ICC 0,07 N hCountry Observations 28,00 20785,00 Marginal R2 / Conditional R2 0,268/0,318 Own calculation 53 MODEL AND ITS FINDINGS Table 10: Summary of the model 3b VOTED Predictors Odds Ratios CI P (Intercept) 4,42 3,72--5,24 <0,001 Education 1,22 1,13--1,31 <0,001 Interest in election 1,26 1,25--1,28 <0,001 National election 10,31 9,28--11,47 <0,001 Support for the EU mem. 1,34 1,27--1,41 <0,001 EU mem. lenght 1,01 1,00--1,02 0,004 Random Effects a 2 3,29 TOO hCountry 0,17 T11 hCountry.EU mem. lenght 0,00 P01 hCountry -0,12 ICC 0,06 N hCountry Observations 28 20785 Marginal R2 / Conditional R2 0,284/ 0,325 Own calculation 54 MODEL AND ITS FINDINGS Summary of the model 3a and model 3b points out to the fact that the most important factor is still whether the individual is or is not a regular voter. Other variables also have some effect, mostly the attitude to the EU and obligation to vote. Compulsory voting increases the odds of voting which support the hypothesis H3. On the other hand, the length of membership in the EU as such has smaller effect on voting (increase in probability of voting is only 1%). The significant effect can be seen only in a longer period of time and shows the difference between the old and new Member States (similar effect as in the previous model 2). However the effect is not significant in explaining for example differences between only old or between only new EU Member States. The value of ICC in both models is rather small (0,07 and 0,06) and that still shows that variables between the states and between their voting systems can explain only small part of the variation in the voters' turnout in the EP elections. Variables between individuals on the individual-level are more important and do have more effect on the turnout. Comparing the visualization of model 3a and 3b can be found on Figure 5 and Figure 6, predictions for the country-level variable in both model are part of Figure 7. 55 MODEL AND ITS FINDINGS Figure 5: Predictions of the model 3a Impact of Education on probability of voting Impact of Interest in elections on probability of voting National elections Support for the EU mern. Own calculation 56 MODEL AND ITS FINDINGS Figure 6: Predictions of the model 3b Impact of Education on probability of voting Impact of Interest in elections on probability of voting National elections Support for the EU mem. Own calculation 57 MODEL AND ITS FINDINGS Figure 7: Predictions for country-level variables, model 3a and 3b Impact of Compulsory voting on probability of voting Impact of EU mem. lenght on probability of voting Compulsory voting EU mem. lenght Own calculation 3.1 Hypotheses and the model findings As the summary of individual models showed, the data support some hypotheses while other hypotheses were not supported by the data. The research question of this thesis examined whether differences between countries' electoral systems affected voter turnout in the 2019 EP elections. This main question was mostly expressed in hypothesis H2 and it was supported by data. However, the analysis of the data showed that the influence of electoral systems is rather small and it is overshadowed by the importance of factors at the individual-level, especially the fact that the voter is a regular voter. As for the differences between the electoral systems, the obligation to vote proved to be one of the main influences, as the membership length in a long period. Hypothesis H3 was thus supported, in contrast, hypothesis H5 was partly supported by the data. The length of the membership can explain for example differences between the turnout in old and in new EU Member States but does not have a significant effect in short period of time. As for the economic conditions in the Member States, its impact on turnout was really minimal. Model 2 shows that better economic development in the Member State may contribute to a higher probability of a voter going to the polls, but only to a limited extent. Likewise, whether a 58 MODEL AND ITS FINDINGS voter evaluates their economic situation as good or bad has only a minimal effect on whether they go to the polls. As for individual characteristics and individual-level variables, the fact that it is a regular voter is one of the most important. In contrast, the age of the voter had only a small effect on increasing the probability that the voter went to the European elections. However, bigger effect can be seen between really young and really old voters. The education of the voter had a more significant effect, the data confirmed that the higher the education, the greater the probability that the voter voted. Hypothesis HI can thus be accepted as true. Among the individual-level influences that are related to the European Union and at the same time had an impact on turnout, we can mention in particular whether the voter supports his country's membership in the EU. The better the voter's attitude towards the European Union, the more likely the voter is to go to the polls. Likewise, the more satisfied the voter is with the European Union as a whole, the more likely he or she will go to the polls. The impact on turnout is rather smaller in this case. However, hypothesis H6 is also supported. 3.2 Result and discussion The data and the models pointed to the following conclusions. Firstly, the results show that the differences between the individual electoral systems in the Member States do not have a systematic and significant influence on voter turnout. Although countries have the power to adapt the electoral system to their national customs, these variations are not substantively important in explaining cross-country differences in the turnout. This applies, for example, to the method of election, voting from abroad or the age limit for exercising the right to vote. Some of these can have some impact but it is usually not substantively significant. A large amount of literature indicates compulsory voting as a major impact of the voting system on turnout. Mandatory voting has some impact on the voters' turnout and increase probability of voting as 2019 EP election's data showed. However, the obligation to vote proved to be a statistically insignificant in the models for 2019 EP election. Of the countries with compulsory voting (Belgium, Bulgaria, Cyprus, Greece, and Luxembourg), only Belgium, Luxembourg and, to a lesser extent, Greece showed above-average participation. This result is probably 59 MODEL AND ITS FINDINGS linked to the fact that fines for non-compliance are not very high in the EU and the impact of this variable on turnout is minimal. As far as economic variables are concerned, those at the individuallevel (economic situation - i.e.. how the voter perceives his current economic situation, or employment - the employment of the voter) were not systematically related to the turnout in the given models. Economic variables at the national level (GDP growth, unemployment, inflation) had some impact on increasing turnout, but it was also not significant according to a model 2. The period of the EU membership (EU membership length) and the voter's attitude towards the European Union both proved to be statistically significant. For both variables, the longer a country has been in the EU and the more positive the voter's attitude towards the EU, the more likely it is that the voter will go to the polls and that the turnout will be higher in that country. This finding is consistent with the existing literature. However, data showed that the dependency is really minor in a short period of membership and these variables influence the probability of voting only by few percent. Significance of this variable grows in time and this explain differences in turnout between the old and the new EU Member States. The fact that a voter is used to going to the polls has a fundamental influence on turnout. If a voter voted in national elections, then it is almost 9 times or even 10 times in some models more likely that he/she did vote in the European elections than a voter who had not participated in the last national elections. The presented findings require some caveats, though. First the models included only 2019 elections and do not say more about the changes in time and do not test the presumptions in time and on more data. There are also some variables missing in the model. These variables are sometimes mentioned in the literature as having impact on the voters' turnout. These are for example time from last election, the turnout in the last elections, number of political parties, overall political culture, established voter behavior and others. These variables are usually stated as others and there are not supported by a strong research on their influence on the voters' turnout and usually are marked as less significant and with less influence. Also collecting these variables goes beyond the scope of this thesis. However, due to their small impact suggested by the existing literature, it is not expected that omitting these variables significantly influenced the results presented in this thesis. 60 CONCLUSION 4 Conclusion This diploma thesis focused on the electoral systems of the EU Member States for elections to the European Parliament and their effects on turnout in these countries. The work examined a number of factors that can affect turnout at two levels. We can find factors such as compulsory voting, election days, the economic situation and more at the Member State level. And then factors at the level of the individual variables such as age, employment, interest in politics and elections and more. Based on the econometric model, the thesis evaluated the effects of country-level and individual variables on the individual decision of the voter, whether or not to go to the polls. This econometric model verified the assumptions on data from 2019 EP elections. This thesis shows that variations in the electoral system do not exert a substantively significant influence on voter turnout. Some features of the electoral system have a minor influence on turnout, others have no systematic effect. The obligation to vote has positive effects on the probability of voting, as well as the determination of election days or the ways in which the voter can cast his/her vote. However, these systemic differences between the Member States can only explain a very small part of the variance in turnout. Factors at the individual-level proved to be more important. At the individual-level, the fact whether the voter is a regular voter or not is the key factor influencing the probability of turning out. Education and also support for the country's membership in the European Union were the other substantively relevant individual factors that influenced participation in these elections. The thesis provides an original contribution to the literature on the study of the European Parliament elections as well as to the literature on the effects of electoral systems. By looking at the European Parliament elections in a single year, this study is able to eliminate possible confounding factors that stem from different times or circumstances and the context of different elections. Thanks to this, the presented analysis eliminates further influences on voter turnout and make it possible to isolate only individual differences between individuals or systemic differences between member countries. Therefore, in addition to providing insights into the determinants of voter turnout in the most recent (and thus not so well researched) European Parliament elections, this study offers a 61 CONCLUSION original way of studying the effects of variations of electoral systems on voter turnout. Future research can extend the analysis presented in this thesis by focusing on time comparisons, or on other systemic differences between member countries, such as differences in the systems of political parties. These research goals go beyond the scope of this thesis but certainly offers interesting suggestions for further research into the future. 62 LITERATURE Literature 2019 European election results. European Parliament [online], [cit. 2020-07-29]. Available: https://europarl.europa.eu/election-results- 2019/en/turnout/ Blais A, Carty K. 1990. Does proportional representation foster voter turnout? European Journal of Political Research. 18:167-81 Blais, A. and Dobrzynska, A. 'Turnout in electoral democracies', European Journal of Political Research, 33 (1998), pp. 239-61 Blais A. 2000. To Vote Or Not To Vote? The Merits and Limits of Rational Choice. Pittsburgh: University Pittsburgh Press Blais, A., Massicotte, L. and Dobrzynska, A. (2003) 'Why is Turnout Higher in Some Countries than in Others?' Blais, A. (2008). What Affects Voter Turnout?. Annual Review of Political Science. 9.10.1146/annurev.polisci.9.070204.105121. Clark, N. (2014). Explaining Low Turnout in European Elections: The Role of Issue Salience and Institutional Perceptions in Elections to the European Parliament. Council Decision (EU, Euratom) 2018/994 of 13 July 2018 amending the Act concerning the election of the members of the European Parliament by direct universal suffrage, annexed to Council Decision 76/787/ECSC, EEC, Euratom of 20 September 1976. \n:ST/9425/2018/INIT. 2018. Available: https://eur-lex.eu- ropa.eu/eli/dec/2018/994/oj Countries: The 27 member countries of the EU. European Union [online], [cit. 2020-08-24]. Available: https://europa.eu/european- union/about-eu/countries_en#the-27-member-countries-of-the-eu 63 LITERATURE Denny, K. and 0. Doyle. Political Interest, Cognitive Ability and Personality: Determinants of Voter Turnout in Britain. British Journal of Political Science Vol. 38, No. 2 (Apr., 2008), pp. 291-310 Denny, K. and 0. Doyle . Does Voting History Matter? Analysing Persistence in Turnout. American Journal of Political Science. Vol. 53, No. 1 (Jan., 2009), pp. 17-35 Eurobarometer Survey 91.5 of the European Parliament. 2019. The 2019 Post-Electoral Survey. Have European Election Entered a New Dimension? [online]. 2019, , 118 [cit. 2020-01-06]. Available: https://www.europarl.europa.eu/at-your-service/files/be-heard/euro- barometer/2019/post-election-survey-2019-complete-results/re- port/en-post-election-survey-2019-report.pdf European Elections Study 2019 Technical Report. European Election Studies [online]. 2019,, 40 [cit. 2020-09-03]. Available: http://europea- nelectionstudies.net/wp-content/uploads/2019/ll/Technical-Report- Final.pdf Eurostat. GDP and main components (output, expenditure and income). Eurostat [online]. 2020a [cit. 2020-02-20]. Available: https://appsso.eurostat.ec.europa.eu/nui/show.do7da- taset=namq_10_gdp&lang=en Eurostat. Unemployment by sex and age. Eurostat [online]. 2020b [cit. 2020-02-20]. Available: https://appsso.eurostat.ec.eu- ropa.eu/nui/show.do?dataset=une_rt_m&lang=en Eurostat. HICP (2015 = 100) - monthly data (annual rate of change). Eurostat [online]. 2020c [cit. 2020-02-20]. Available: https://appsso.eurostat.ec.europa.eu/nui/show.do7da- taset=prc_hicp_manr&lang=en Fowler, G. (2013). Five studies on the causes and consequence of voter turnout. Harvard: Harvard Doctoral Dissertation 64 LITERATURE Franklin M. 1996. Electoral participation. In Comparing Democracies: Elections and Voting in Global Perspective, ed. L LeDuc, RG Niemi, P Norris, pp. 216-35. Beverly Hills, CA: Sage Franklin M. 2004. Voter Turnout and the Dynamics of Electoral Competition in Established Democracies Since 1945. Cambridge, UK: Cambridge University Press Herrera, H. a Morelli, M. a Nunnari, S., 2016. Turnout Across Democracies. American Journal of Political Science, Vol. 60, No. 3 (JULY 2016), pp. 607-624 Hix, S. a Marsh, M, 2011. Second-Order Effects Plus Pan-European Political Swings: An Analysis of European Parliament Elections Across Time, Electoral Studies, 30,1, 2011, 4-15 International Institute for Democracy and Electoral Assistance a International Institute for Democracy and Electoral Assistance. Voter Turnout in Western Europe. Stockholm, 2004 Jackman, R. W., 1987. Political Institutions and Voter Turnout in the Industrial Democracies. The American Political Science Review, Vol. 81, No. 2 (Jun., 1987), pp. 405-424 (19 pages) Jackman RW, Miller RA. 1995. Voter turnout in the industrial democracies during the 1980s. Comparative Political Studies 27(4):467-92 Kavanagh, A. and W. Durkan. The Brainstorm Long Read: there are many factors to explain why hundreds of thousands of registered voters do not vote at Irish elections. Maynooth University [online], [cit. 2020-08- 23]. Available: https://www.maynoothuniversity.ie/research/spotlight- research/getting-out-vote-what-influences-voter-turnout Kostadinova T. 2003. Voter turnout dynamics in post-Communist Europe. European Journal of Political Research. 42(6):741-59 Lewis-Beck, M, and Mitchell, G. 1990. "Transnational Models of Economic Voting: Tests from a Western European Pool." Revista del Instituto de Estudios Economicos 4:65-81 65 LITERATURE MacKuen, M. B., Erikson, R. S. and Stimson, J. A., 1992. "Peasants or Bankers?: The American Electorate and the U.S. Economy." American Political Science Review86: 597-611. Norris P. 2002. Electoral Engineering: Voting Rules and Political Behavior. New York: Cambridge University Press Malkopoulou, A. 2009. "Lost Voters: Participation in EUElections and the Casefor Compulsory Voting." Reif K. a Schmitt, H. (1980). Nine second-order national elections: A conceptual framework for the analysis of European elections results, European Journal of Political Research 8: 3-44. Paldam, M. . 1991. "How Robust Is the Vote Function? A Study of Seventeen Nations over Four Decades." In Economics and Politics: The Calculus of Support, ed. Helmut Norpoth, Michael Lewis-Beck, and JeanDominique Lafay. Ann Arbor: University of Michigan Press. Powell GB. 1982. Contemporary Democracies: Participation, Stability and Violence. Cambridge, MA: Harvard Univ. Press Powell GB Jr. 1986. American voter turnout in comparative perspective. American Political Science Review 80(1): 17-43 Powell GB, Whitten GD. 1993. A cross-national analysis of economic voting: taking account of the political context. American Political Science Review 37:391-414 Singh, S., 2011. How Compelling is Compulsory Voting? A Multilevel Analysis of Turnout, Political Behavior, Vol. 33, No. 1 (March 2011), pp. 95-111 Schmitt, H., Hobolt, S. B., van der Brug, W. and Popa, S. A (2019) European Parliament Election Study 2019, Voter Study. Solijonov, A., 2016. Voter Turnout Trends around the World, International Institute for Democracy and Electoral Assistance 66 LITERATURE Strom, K, and Lipset, SM. 1984. "Macroeconomics and Macropolitics: The Electoral Performance of Democratic Governments." Presented at the annual meeting of the American Political Science Association, Washington, DC. Sundstrom, A, Stockemer, D. (2015). Regional Variation in Voter Turnout in Europe: The Impact of Corruption Perceptions. Electoral Studies, 40,158-169 The Treaty on European Union (TEU). In: Official Journal C 326 , 26/10/2012 P. 0001 - 0390. 2012. Available: https://eur-lex.eu- ropa.eu/legal-content/EN/TXT/?uri=CELEX%3A12012M%2FTXT The Treaty on the Functioning of the European Union (TFEU). In: Official Journal C 326 , 26/10/2012 P. 0001 - 0390. 2012. Available: https://eur-lex.europa.eu/legal-con- tent/EN/TXT/?uri=CELEX%3A12012E%2FTXT Wolfinger RE, Rosenstone SJ. 1980. Who Votes? New Haven, CT: Yale Univ. Press How to vote [online]. AEGEE, With the support of the Erasmus+ Programme of the European Union, 2018 [cit. 2020-06-08]. Available: https://www.howtovote.eu The European Parliament: electoral procedures. European Parliament [online]. 2020 [cit. 2020-06-08]. Available: https://www.euro- parl.europa.eu/factsheets/en/sheet/21/the-european-parliament-elec- toral-procedures 67 VOTING SYSTEMS IN THE E U MEMBER STATES FOR THE E P ELECTIONS Pnloha A Voting systems in the EU Member States for the EP elections Compulsory Voting days Distant voting options Expact voting Voting age Austria no Sun by post yes 16 Belgium yes Sun only in person yes 18 Bulgaria yes Sun only in person yes 18 Croatia no Sun only in person yes 18 Cyprus yes Sun only in person no 18 Czech Republic no Fri, Sat only in person no 18 Denmark no Sun by post yes 18 Germany no Sun by post yes 18 Estonia no Sun e-voting yes 18 Finland no Sun in advance yes 18 France no Sun only in person yes 18 Greece yes Sun only in person yes 17 Hungary no Sun only in person yes 18 Ireland no Fri only in person no 18 Italy no Sun only in person yes 18 Latvia no Sat only in person yes 18 Lithuania no Sun in advance yes 18 Luxembourg yes Sun only in person yes 18 Malta no Sun only in person no 16 Netherlands no Thür in proxy yes 18 Poland no Sun only in person yes 18 Portugal no Sun only in person yes 18 Romania no Sun only in person yes 18 Slovenia no Sun only in person yes 18 Slovakia no Sat only in person no 18 Spain no Sun by post yes 18 Sweden no Sun only in person yes 18 UK no Thür in proxy, by post yes 18 Sources: How to vote (2018), The European Parliament: electoral procedures (2020), European Elections Study 2019 Technical Report (2019) 69 DATA Příloha B Data Data for the models can be found in the IS system as individual file. 70 R CODES Příloha C R codes modell<-glmer(VOTED~l+(l|hCountry), family=binomial("logit"), data=dt] parameters::p_value(modell,p.kr=TRUE) performance::icc(modell] tab_model(modell] summary(modell] model2<- glmer(VOTED~SEX+AGE+EDU+CAM+INT+LER+ECO+WOR+SAT+NAT+ ATT+OBL+DAY+WAY+FOR+VAG+POP+GDP+INF+UNE+MEM+(OBL+DA Y+WAY+FOR+VAG+POP+GDP+INF+UNE+MEM|hCountry), family=binomial("logit"), data=dt, glmerControl(optimizer = "bobyqa", optCtrl = listfmaxfun = 100000))) tab_model(model2) summary(model2) performance::icc(model2) model3a<-glmer(V0TED~EDU+CAM+NAT+ATT +0BL + (OBL|hCountry), family=binomial("logit"), data=dt) tab_model(model3a) summary(model3a) model3b<-glmer(V0TED~EDU+CAM+NAT+ATT +MEM + (MEM|hCountry), family=binomial("logit"), data=dt) tab_model(model3b) summary(model3b) plot_model(model2, axis.labels = c("EU mem. lenght","Unemployment","Inflation",''Economic growth","Population","Early voting age","Expact voting","Distant voting option","Voting days'7'Compulsory voting","Support for the EU mem.","National elections","Satisfaction with 71 R CODES the EU'',''Employment'',''Economic situation","Right/Left ideology","Interest in politics","Interest in elections","Education","Age","Gender")] plot_model(modelllglmer.2, type = "pred", term="SEX",title="Impact of Gender on probability of voting"] plot_model(model2, type = "pred", term="AGE",title="Impact of Age on probability of voting"] plot_model(model2, type = "pred", term="EDU",title="Impact of Education on probability of voting"] plot_model(model2, type = "pred", term="CAM",title="Impact of Interest in elections on probability of voting"] plot_model(model2, type = "pred", term="INT",title="Impact of Interest in politics on probability of voting"] plot_model(model2, type = "pred", term="LER",title="Impact of Right/Left ideology on probability of voting"] plot_model(model2, type = "pred", term="ECO",title="Impact of Economic situation on probability of voting"] plot_model(model2, type = "pred", term="WOR",title="Impact of Employment on probability of voting"] plot_model(model2, type = "pred", term="SAT",title="Impact of Satisfaction with the EU on probability of voting"] plot_model(model2, type = "pred", term="NAT",title="Impact of National elections on probability of voting"] plot_model(model2, type = "pred", term="ATT",title="Impact of Support for the EU mem. on probability of voting"] plot_model(model2, type = "pred", term="OBL",title="Impactof Compulsory voting on probability of voting"] plot_model(model2, type = "pred", term="DAY",title="Impact of Voting days on probability of voting"] plot_model(model2, type = "pred", term="WAY",title="Impact of Distant voting option on probability of voting"] plot_model(model2, type = "pred", term="FOR",title="Impact of Expact voting on probability of voting"] plot_model(model2, type = "pred", term="VAG",title="Impact of Early voting age on probability of voting"] plot_model(model2, type = "pred", term="POP",title="Impact of Population on probability of voting"] plot_model(model2, type = "pred", term="GDP",title="Impact of Economic growth on probability of voting"] 72 R CODES plot_model(model2, type = "pred", term="INF",title="Impact of Inlfation on probability of voting") plot_model(model2, type = "pred", term="UNE",title="Impact of Unemployment on probability of voting") plot_model(model2, type = "pred", term="MEM",title="Impact of EU mem. lenght on probability of voting") plot_model(model3a, type = "pred", term="EDU",title="Impact of Education on probability of voting") plot_model(model3a, type = "pred", term="CAM",title="Impact of Interest in elections on probability of voting") plot_model(model3a, type = "pred", term="NAT",title="Impact of National elections on probability of voting") plot_model(model3a, type = "pred", term="ATT",title="Impact of Support for the EU mem. on probability of voting") plot_model(model3a, type = "pred", term="OBL",title="Impact of Compulsory voting on probability of voting") plot_model(model3b, type = "pred", term="EDU",title="Impact of Education on probability of voting") plot_model(model3b, type = "pred", term="CAM",title="Impact of Interest in elections on probability of voting") plot_model(model3b, type = "pred", term="NAT",title="Impact of National elections on probability of voting") plot_model(model3b, type = "pred", term="ATT",title="Impact of Support for the EU mem. on probability of voting") plot_model(model3b, type = "pred", term="MEM",title="Impact of EU mem. lenght on probability of voting") 73 REJSTŘÍK Rejstřík A age, 29, 31, 33, 37, 49 attitude of voters towards the EU, 30 c common framework, 20 common principles, 21 compulsory voting, 22, 31, 34, 38, 52, 55, 58 confidence in the EU, 35 country-level factors, 31, 38, 42 D direct elections, 20 E economic growth, 29, 31, 34, 39, 51 education, 29, 31, 33, 37,49, 55, 56, 57 election days, 23, 28, 31, 38 election study 2019 survey, 36 electoral participation, 27 electoral system, 28, 33 EU electoral systems, 21 EU level factors, 28, 30 e-voting, 24, 38 expat voting, 24, 38 F factors influencing turnout, 19 I individual-level factors, 31, 37, 40 inflation, 29, 34, 39, 51 L legislation for the EP elections, 20 length of the EU membership, 31, 52 length of the EU membership, 55 length of the EU membership, 58 R regular voter, 30, 32, 37, 51, 55, 56, 57 S satisfaction with the EU, 30, 35, 37, 49 second-order elections, 28, 30, 31, 34 socio-economic factors, 28, 29 support for the EU membership, 38, 49, 55, 56,57 T threshold, 22 turnout in the EP elections, 25 turnout trends, 19 u unemployment, 29, 34, 39, 51 V voter turnout, 19 voting age, 23, 31, 38 voting by post, 24, 31, 38 voting in proxy, 24 voting in proxy, 38 voting systems, 20, 27, 31, 42, 69 74