Issues that mobilize Europe. The role of key policy issues for voter turnout in the 2019 European Parliament election

In light of the unexpectedly high turnout in the 2019 European Parliament election, we explore how major transnational policy issues mobilize voters in European electoral contests. Based on the analysis of two data sets, the Eurobarometer post-election survey and the RECONNECT panel survey, we make three important observations. First, European citizens show a higher tendency to participate in European Parliament elections when they attribute greater importance to the issues ‘climate change and environment’, ‘economy and growth’, and ‘immigration’. Second, having a more extreme opinion on the issue of ‘European integration’ increases people's likelihood to vote in European elections. Third, the mobilizing effect of personal issue importance is enhanced by the systemic salience that the respective policy issue has during the election campaign. These findings show the relevance of issue mobilization in European Parliament elections as well as its context-dependent nature.


Introduction
After a decade during which the European Union (EU) has been riddled by a multifaceted crisis and increased politicization, the 2019 election to the European Parliament (EP) was accompanied by the dictum that the future of Europe is at stake (Braun, 2021;Treib, 2021). While similar debates in the past, such as around the introduction of the Spitzenkandidaten system in 2014, had only limited effects on voter mobilization (Braun and Schwarzbözl, 2019;Gattermann et al., 2016), fears about the future of Europe seem to have incited a strong rise in EU-wide electoral participation in 2019. Voter turnout increased from 42.6% (in 2014) to 50.7%the highest turnout in the past 20 years and the first time that overall participation in EP elections did not decrease but increased. Although electoral contests at the European level can still be considered 'second-order elections' Plescia et al., 2020a;Reif and Schmitt, 1980;Schäfer, 2021;Schmitt et al., 2020), the 2019 election might have been different. The presumption that more Europeans went to the ballot box to shape the future path of the EU is reinforced by some immediate analyses of the 2019 EP election that ascribe the increase in electoral participation, at least partly, to a greater 'Europeanness' (Braun, 2021) and a higher politicization of EP elections (De Sio et al., 2019: 64).
The suggestion that a more intense debate over European issues in the run-up to the elections positively impacted individual participation in the 2019 EP election, is reinforced by a simple observation: despite the rise in turnout across Europe, there is still a considerable variance in the absolute participation levels between the EU member states (see Figure 1(a)). Importantly, such differences are also visible when we take a look at the most important mobilizing issues during the 2019 EP election (see Figure 1(b)). When being asked about the issues that 'made you vote in the recent European Parliament elections', the majority of people in many countries reported that the economy was key for their participation. However, in other countries the strongest mobilizing issues were climate change, immigration, or European integration. Observing such a variance might seem surprising, but it indicates the persisting second-order nature of EP election campaigns, where issues are discussed by national actors in national public spaces (Boomgaarden and De Vreese, 2016). The intriguing question is how relevant these issues were for voter turnout in the 2019 EP election and which issues exerted the strongest mobilization power. To answer these questions, we need to shift the focus from the macro-level to the citizen-level and delve into the micro-foundations of the relationship between policy issues and electoral behavior.
This article explores whether and through which mechanisms policy issues affect electoral participation. In particular, we investigate the impact of the four major transnational policy issues 'economy', 'immigration', 'environment/climate change' and 'European integration' on individual electoral participation. We argue, first, that EU citizens showed a higher tendency to participate in the 2019 European election when they attributed a greater relevance to one or more of these issues, meaning that they found an issue to be salient or held a rather extreme opinion on a policy issue. Second, we argue that people were more likely to vote when the policy issues they subjectively found most relevant were also systemically salient in their country. Our empirical analyses are based on two different data sources: the large-scale post-election study of the Eurobarometer (EB 91.5) comprising all 28 EU member states (at the time of the election) as well as the original panel survey data for seven selected EU countries taken from the H2020 project RECONNECT. The findings reveal the context-dependent nature of issue mobilization during EP elections and bear important implications for the legitimizing role of EP elections and the future of European integration.

Policy issues and European elections
Although we know a lot about the factors fueling and hindering citizens' participation in elections, much less is known about the peculiar link between policy issues and the decision to vote. Do citizens go to the ballot box because specific issues matter to them or their country? In other words, do political issues mobilize citizens? And does such a relationship also hold for EP elections?

Policy issues and the individual decision to vote
We know from Down's (1957) famous insights on rational voting behavior that voters generally have rather low incentives to become deeply informed about specific policy issues. However, elections are contests that 'are fought over policies and issues that voters, parties, and leaders consider to be important and relevant at the time of the Notes: (a) is based on official turnout data from the European Parliament (for over-time development for country groups, see the Online appendix). The data source for (b) is the Eurobarometer 91.5 post-election study. It displays the most frequently mentioned item per country to the question 'What are the issues which made you vote in the recent European Parliament elections?' (QG7) that was posed to respondents who reported having participated in the 2019 election. election' (Aardal and van Wijnen, 2005: 192). Therefore, it is intuitively plausible that central policy issues and voter turnout should be linked to some degree. One could even make a stronger claim: failing to account for policy issues in the research on electoral participation would mean taking 'politics out of elections' (Aaardal and van Wijnen, 2005: 192).
Accordingly, the scholarly literature presents empirical evidence indicating a relationship between policy issues and turnout. For example, countries with high electoral salience and fierce competition over political issues report higher turnout rates (Franklin, 1996;Pacheco, 2008). For the US, we see that citizens' probability of turning out in elections is higher if they perceive differences in issue positions between candidates (Adams and Merrill, 2003;Adams et al., 2006;Leighley and Nagler, 2014;Zipp, 1985). Equally impressive are more recent studies on 'policy representation' illustrating that the congruence between voters and political parties over particular, salient political issues affects electoral participation (Dinas et al., 2014;Lefkofridi et al., 2014;Reher, 2014). Finally, Kiousis and McDevitt (2008) shed light on the process behind the link connecting issues and turnout. They show that greater news attention during election campaigns increases individual 'issue importance' and 'opinion extremity', which eventually furthers the chance of electoral participation.
Although we lack similar studies regarding European elections, investigating EP elections can significantly contribute to our understanding of the relationship between policy issues and electoral participation. First, EP elections create an opportunity to investigate a simultaneously conducted set of elections in various countries and electoral contexts (Söderlund et al., 2011;Van der Eijk et al., 1996). This feature implies that EP elections are well suited to test the effects of different issue contexts. Second, turnout in EP elections is traditionally rather low, which is mostly attributed to the perception that there is 'less at stake' compared to national first-order elections (Franklin and Hobolt, 2011;Reif and Schmitt, 1980;Schmitt and Mannheimer, 1991;Van der Eijk and Schmitt, 2009). However, the rising turnout in the most recent EP elections might indicate that there is now 'more at stake' and that key policy issues mobilize people across the continent. Empirical findings from previous EP elections support this argument. Although Clark (2014) shows that the lower salience of issues under the jurisdiction of the EU can explain the reduced turnout in European elections, many scholars from the 'Europe matters' camp find that the issue of European integration indeed plays a role in present-day European elections. Besides the well-known 'EU issue voting' argument that is imperative for the decision to vote for a party (De Vries, 2010;De Vries and Hobolt, 2016), EU attitudes are relevant for turnout. While pro-EU attitudes bolster the chance of electoral participation (Braun, 2021), non-voting in EP elections is motivated by Eurosceptic attitudes (Blondel et al., 1998;Clark, 2014;Hobolt and Spoon, 2012;Hobolt and Wittrock, 2011;Schäfer, 2021;Wessels and Franklin, 2009). In a similar vein, Schäfer and Debus (2018) demonstrate that issue congruence between voters and political parties matters for participation in EP elections and that this relationship has strengthened during the Euro crisis. Against this backdrop, we assume that turnout rose in the 2019 EP election because citizens perceived that there was more than the usual at stake in these elections and because several policy issues at the EU level were highly politicized during the election campaign.

The key mobilizing issues in the 2019 European Parliament election
The 2019 EP election was held at a time when a limited number of key issues mattered to voters across EU member statesyet to a varying degree in each country (see Figure 1(b)). Besides this inductive approach of identifying the most relevant policy issues from survey data, we also provide theoretical arguments for our interest in the mobilizing role of four main issues: the economy, immigration, environment/climate change, and European integration. In the following, we argue why these four issues were highly politicized among the European publics during the 2019 EP election campaigns and why they should be related to the increase in turnout.
It is commonplace to say: 'when you think elections, think economics' (Tufte, 1978: 65). But apart from the central role of the economy in elections more generally, economic issues were indeed still at the top of the public agenda during the 2019 EP election. This was particularly the case in numerous countries that experienced economic hardship following the sovereign debt crisis and austerity politics in Europe. The 'Euro crisis' yielded severe political repercussions in elections all over Europe (Hernández and Kriesi, 2016) and already affected the previous European elections in 2014 (Braun and Tausendpfund, 2020;Hobolt and De Vries, 2016). The fact that the economy is a primary concern for many voters is certainly not a new insight, as the economic voting literature illustrates (Lewis-Beck and Stegmaier, 2000). However, the economy's role in electoral participation is disputed. Research shows that economic conditions can be both mobilizing and demobilizing and that the effects often depend on other moderating factors (Gomez and Hansford, 2014;Rosenstone, 1982;Weschle, 2014). Nevertheless, it is still true that economic issues are crucial during elections, especially when people experience economic hardship.
Second, immigration issues have surely been among the most salient themes in Europe during recent years, especially with the so-called 'migration crisis' or 'refugee crisis' of 2015 (Grande et al., 2018;Green-Pedersen and Otjes, 2019;Van der Brug et al., 2015). Increasing immigration towards Europe attracted a great deal of media coverage and public debates, but it also led to a rise in the populist radical right in Europe, thereby transforming electoral competition in many EU member states (Abou-Chadi et al., 2020;Dennison and Geddes, 2019). Even more consequential, the salience of the immigration issue strongly contributed to the decision by an EU member state, for the first time ever, to leave the Union (Goodwin and Milazzo, 2017;Hobolt, 2016). Although the number of immigrants coming to Europe notably decreased directly before the 2019 EP election, the struggle to find a common migration and asylum policy remained a major and ongoing source of political conflict on the European level, particularly between some of the Eastern and the North-Western member states.
Third, concerns over climate change have brought environmental issues to the top of the public agenda in 2019. In the months leading up to the election, school strikes and the 'Fridays for Future' protest movement demanded political action and dominated public debate in many EU member states. The protests culminated in an address to the EP by the leading figure of this movement, Greta Thunberg, in April 2019 (Pearson and Rüdig, 2020). Not surprisingly, the subsequent EP elections resulted in a 'green wave' in many EU member states, especially in the Northern and Western countries. 1 Consequently, the electoral successes of Green parties and the increased salience of environmental issues and climate change later pressed the new President of the European Commission, Ursula von der Leyen, to announce a 'European Green Deal' as a top priority of her presidency (Von der Leyen, 2019).
Lastly, the European integration issue itself certainly vitalized the EP election campaigns in 2019, with many observers styling the electoral contest as 'a battle over Europe's future' (Treib, 2021: 174). While previous EP elections had been predominantly marked by Eurosceptic challenger parties from the left and right ends of the political spectrum, many mainstream-center parties now started to engage in pro-European mobilization efforts . Eventually, explicitly pro-integration parties, such as Emmanuel Macron's newly established 'La République en Marche!', successfully stopped further inroads by Eurosceptic parties. Moreover, grassroots movements like 'Pulse of Europe' and pro-European niche parties like 'Volt!' and 'DiEM25' mobilized citizens in favor of European integration. All in all, the former 'sleeping giant' (Van der Eijk and Franklin, 2004) seems to have finally been awakened and the 'political conflict over Europe' (Schäfer et al., 2021a) appears to be salient for voters in contemporary European elections. 2

The link between policy issues and electoral participation
Based on previous research regarding the relation between policy issues and electoral participation, we argue that the link between the key policy issues and the individual's decision to turn out during European elections functions through two direct and one indirect channel. First, EU citizens should be more inclined to vote when they perceive a major transnational policy issue as highly important. Second, people should feel more incentivized to participate in an EP election when they hold an extreme opinion on a key policy issue. These two direct channels imply that both meta-attitudinal cognitions (such as issue importance) as well as more operative forms (such as attitude extremity) affect political behavior (Bassili, 1996). Third, the public relevance of certain policy issues should enhance the mobilization effect of personal issue importance among the citizens of this country.
The relevance that a person attaches to a policy issue can be called personal 'issue salience' or 'issue importance' (Fournier et al., 2003). Some scholars understand personal issue salience as 'a political attitude like any other' (Dennison, 2019: 443), exerting direct and indirect effects on many political behaviors. Other authors see issue importance as a form of 'attitude strength' but also find that it affects voting behavior (Bassili, 1996: 637-8). On a more general note, we can assume that personal issue importance is related to political awareness and political interest (Zaller, 1992). From previous research, we know that people who are politically interested and informed have a stronger tendency to participate in elections (Delli Carpini and Keeter, 1996;Lassen, 2005;Smets and van Ham, 2013)which is also true for EP elections (Bhatti, 2010;Braun and Tausendpfund, 2019;Hogh and Larsen, 2016;Schäfer, 2019).
Accordingly, the more people become interested and competent in political matters, the more likely they are to form political attitudes and attach importance to different policy issues. This assumption implies that, the individual perception of relevance of certain policy issueswhether for herself (ego-centric motivation) or for the country she lives in (socio-tropic motivation)is consequential for politics. We then expect that this person has an increased tendency to follow political events, receive political information and participate in political decision-making processes. Consequently, in our first hypothesis we argue that the act of electoral participation can be understood as a mean to put personally relevant policy issues on the agenda of the entity or institution a person votes forin our case, the EP or the EU more generally.
H1: If a person perceives a major policy issue at the time of the EP elections (economic, environmental, immigration, and European integration issues) as important, she is more likely to cast a vote in the 2019 EP election.
However, insights from social psychology suggest that salience effects do not represent the only impact that policy issues have on electoral participation. Building on the notion of 'opinion extremity' (Kiousis and McDevitt, 2008), we assume that people make their electoral participation dependent on whether they have a strong view on a particular policy issue relevant in the elections. More specifically, we argue that voters feel incentivized to participate when they want to express a strong (i.e. extreme) preference, whereby extreme is defined as 'the extent to which the attitude deviates from neutrality' (Krosnick and Petty, 1995: 6). Strong opinions on policy questions signal that people perceive an issue to be 'at stake', especially during an election. From this perspective, electoral participation can be seen as an attempt to shape future policies in line with preferred policy positions by supporting a political party representing these positions. In contrast, people with more ambivalent, neutral, or indifferent opinions on policy issues should have lesser incentives to participate in an election because there clearly is 'less at stake' for them.
H2: If a person holds a more extreme opinion on a major policy issue (economic, environmental, immigration, and European integration issues), she is more likely to cast a vote in the 2019 EP election.
Lastly, it is not only individual issue importance that appears crucial for political behavior. From communication studies, we know that people have a good sense of the general 'opinion climate' (Matthes et al., 2018) in a society. They follow the issues on the public agenda and adjust their own (issue) priorities and behavior, as we know from the literature on agenda-setting (Dearing and Rogers, 1996) and the 'spiral of silence' (Noelle-Neumann, 1974). This is where the 'systemic salience' of policy issues comes into play: individuals become more confident when perceiving their personal salience attributions reflected in public opinion. In contrast, when they notice that their own issue priorities are not shared by society, they will feel socially isolated and disincentivized to express their views.
Taken together, we argue that the combination of high personal issue importance and high systemic issue salience should have an additional effect on electoral participation.
Compared to the two individual-level mechanisms described before, we put forward at this stage a systemic argument of contextual moderation: People should feel incentivized to vote when they see other people around them sharing their own concerns. As soon as people perceive a certain issue to be relevant in their country, their personallevel salienceon the very same issueshould have a greater mobilizing effect. Hence, our third hypothesis considers the interaction between the individual-level mechanisms outlined above and the contextual-level issue salience.
H3: The more a major policy issue is systemically salient, the stronger is the mobilization effect of individual issue importance. Figure 2 visualizes the three theoretical hypotheses within our explanatory model. As laid out, we conceptualize the relevance of policy issues for electoral participation via three different theoretical arguments. First, we draw on the concept of personal issue importance and presume that the more relevant a person considers a policy issue, the more she will be motivated to make this issue visible by turning out and voting in an election (H1). Second, electoral participation is also enhanced when a person has a strong view on a particular policy issue. This idea is conceptualized via opinion extremity regarding the policy issue in question (H2). Third, adding to this direct individual-level link between issue salience and participation, we posit that the effect of personal issue importance is conditioned by the systemic salience of the policy issues under consideration (H3).

Research design
To test our theoretical assumptions, we make use of two different surveys with varying scope and scale conducted during the 2019 EP election: the Eurobarometer (EB, 91.5) study and the 'RECONNECT Europe' (RE) panel survey. Each dataset has its own advantages. The EB integrates a set of items regarding people's issue salience attributions and guarantees a representative sample from the whole EU population. In contrast, the RE data contain valuable information concerning people's issue positions and allows better identification of causal relationships due to its panel structure. Moreover, in contrast to the EB, the RE survey enables us to study the European integration issue.

Datasets
In a first step, we examine hypotheses H1 and H3 with the large-scale EB post-election study that contains respondents from all 28 EU member states and guarantees high representativeness due to a multi-stage random sampling. It was carried out by KANTAR in June 2019 and covered the population of all EU member states aged 15 years and over (European Commission and European Parliament, 2019;Zalc et al., 2019). In total, the dataset includes 27,464 respondents, i.e. roughly 1000 people per country. The sample was drawn in a multi-stage random (probability) process, and households were selected by a standard 'random route' procedure, starting from an initial random address. Concerning the data capture, computer-assisted personal interviews (CAPI) were conducted in the countries where this technique was available.
In the second step, we test H2 with original survey data from the RE project. The panel study includes a pre-electoral and a post-electoral online survey in seven different EU member states: Denmark, France, Germany, Hungary, Italy, Poland, and Spain (Plescia et al., 2020b). While the pre-electoral wave was carried out in April 2019, the postelectoral interviews were led between May 27 and June 26, 2019. Respondents were sampled from pre-existing online panels provided by IPSOS. To participate in the study, respondents had to be at least 18 years old and eligible to vote in the 2019 EP election. National quota sampling was employed based upon the key demographics age, gender, education, and region. In total, 8780 respondents took part in both panel waves while the number per country ranges between 908 (Denmark) and 1393 (Spain) respondents.

Operationalization
Across the two datasets, the dependent variable in our analyses is reported electoral participation in the 2019 EP election. In both surveys, respondents were asked whether they had participated (1) or not (0) in the previous European election. Comparing the reported turnout on the aggregate level with the official numbers shows significant over-reporting in both surveys. However, while the average over-reporting is rather small in the EB (5 points), it is much larger in the RE survey (26 points). This difference possibly stems from the fact that the EB survey was carried out as a proper random sample from all member state populations, whereas the RE survey took a stratified sample from an online access panel (see the Online appendix).
The three main independent variablespersonal 'issue importance', 'systemic issue salience', and 'opinion extremity'are measured with slightly different yet similar indicators. Both personal issue importance and systemic issue salience are operationalized via the 'most important issue' (MII) question typically asked in voter studies. Although this indicator is not without problems (Jennings and Wlezien, 2011;Johns, 2010), it appears to be an adequate item in the post-election surveys at hand. Moreover, 'aggregate MII appears to broadly reflect what actually matters to the typical voter' (Bartle and Laycock, 2012: 687). Therefore, we use the MII questions both for the voter level (as individual issue importance) as well asin their aggregated formfor the country level (as systemic issue salience). 3 To measure issue importance, we create a sum index out of three different EB items. Respondents were asked to select three MII (at most) for three levels: personal issues, country issues, and European/EU issues (with similar response options for all three items). We create an additive index that ranges from 0 (issue is not among the MIIs at any level) to 1 (issue is among the MIIs at all three levels). Importantly, we group single issues into the areas 'economy' (economic growth, unemployment, inflation, taxation, finances), 'environment' (environment, climate change), and 'immigration' (immigration, asylum). A country's systemic issue salience is obtained by aggregating this MII index. For each of the major policy issues (economy, immigration, environment/climate change), the variable ranges from 0 to 1 (see also the Online appendix for more detailed information on the operationalization).
Individual opinion extremity is operationalized in two steps using the RE dataset. First, we extract a respondent's view on the main issue areas of interest, as they were confronted with two opposing policy statements and asked to position themselves on a scale from 0 to 10, where the two extremes represent the two opposing statements. Regarding the economy issue, we used an item asking for their view on redistribution of wealth from the rich to the poor. Opinion extremity regarding immigration is measured via a person's view on whether immigrants take jobs away from fellow nationals. For the issue of environment, people are asked whether they prefer environmental protection at the cost of economic growth or demand economic growth even at the cost of damage to the environment. Lastly, for the EU integration issue we relied on a question asking respondents whether European unification has already gone too far or should be implemented further. Second, in line with earlier scholarly work (Krosnick and Petty, 1995) we compute the absolute difference between a respondent's preferred policy position and the respective country's mean on the same item. 4 Apart from our main variables discussed above, we also included a series of control variables in our regression models. These comprise attitudes towards the EU and other institutions (EU support, trust in the EP, support for the national government) as well as socio-psychological dispositions (political interest, EU-specific knowledge, sociotropic economic evaluations) and socio-demographic information (age, gender, education, social class/income). These variables account for the relevance of EU attitudes in EP elections (Blondel et al., 1998;Stockemer, 2012), but also belong to the standard determinants of electoral participation (Smets and van Ham, 2013). Moreover, the mere positions on each of the policy issues are also included as control variables in the regression models where we test opinion extremity. The reason is that the effect of opinion extremity could otherwise easily be confounded by the direction of the issue opinion, i.e., effects could be driven by only one camp of an issue dimension. Finally, to avoid omitted variable bias on the country level, we include country dummies (in fixed effects models) and the official turnout figure of the 2019 EP election (in multi-level models). 5

Empirical analysis
We estimate logistic regression models on both datasets due to the binary nature of the dependent variable (see Online appendix for the full regression results). With the EB dataset, we first compute country-fixed effects models to test the direct impact of issue importance (H1). After that, we estimate hierarchical multilevel models to identify the moderating effects of systemic issue salience (H3). The RE dataset is analyzed by employing logistic, fixed effects regression models to test the effects of opinion extremity (H2).

The Eurobarometer dataset
First, we inspect the results of the fixed effects models, which control for the influence of confounding factors on the country level. Figure 3 displays the predicted probabilities for electoral participation in function of individual issue importance across three policy areas (see the Online appendix for the full results). As H1 suggests, issue importance employs significant direct mobilizing effects for all policy issues. The effect is particularly pronounced for environmental and climate change issues (see Figure 1(c)), and less pronounced for economic issues (see Figure 1(a)). People who attribute greater relevance to these issues had a remarkably higher likelihood of participating in the 2019 EP election, even when controlling for a series of other individual-level factors. As illustrated in Figure 3, the predicted probability of casting a vote increases by 13 percentage points for citizens who attach high salience to the environment and climate change than for people who do not attribute any importance to these issues. Although we can observe similar mobilizing effects for the other two issues, the increase in the participation likelihood is stronger for environmental issues than economic and immigration issues (plus 3 and 5 percentage points). Nevertheless, our first theoretical hypothesis receives broad empirical support across all policy issues.
In a second step, we inspect the cross-level interactions specified in the multi-level models, expressed as conditional marginal effects of individual issue importance dependent on the systemic salience of the same policy issue (see Figure 4 or the Online appendix for the full results). In accordance with H3, all three interaction effects are positive and statistically significant. 6 This finding implies that high salience contexts mobilize people who perceive the same policy issues as important, whereas low salience contexts are rather demobilizing. However, the three issues under observation display notably diverging pictures concerning their mobilization potential.
The story is straightforward regarding immigration (see Figure 4(b)): a higher personal issue importance does not mobilize people in low salience contexts. However, it reveals a strong mobilization effect in high salience contexts (10 percentage points increase). A different picture emerges for economy (see Figure 4(a)): in contexts where economic issues are not systemically salient, higher individual importance for economic issues leads to a decreasing probability to participate in the election (10 percentage points). This finding implies that the economic issue carries demobilizing effects overall. Even in high salience contexts, there is no real mobilization effect of the economy (but also no demobilization effect). The very opposite is true for environment/climate change (see Figure 4(c)): even in low salience contexts, higher individual issue importance contributes to a slightly higher likelihood to participate (5 percentage points). And in high salience contexts, the mobilization potential of environmental issues is the greatest across all issues (15 percentage points).

The RECONNECT Europe dataset
Analyzing the RE dataset allows us to test our opinion extremity hypothesis (H2) and has two additional benefits. First, it includes the European integration issuea key policy issue in present-day European elections. Second, it has a panel structure, meaning that we can measure the independent variables temporally prior to the dependent variable. We again estimate binary logistic regression models for all policy issues together and separately (see the Online appendix). Figure 5 illustrates the effects of opinion extremity across all four major issues. It shows that individual opinion extremity is not a significant predictor of participation in EP elections for all but one policy area. Having a strong view on the economy, the environment, and immigration does not increase the probability to vote in European elections when controlling for standard determinants of electoral participation ( Figure 5(a) to (c)).
In contrast, being strongly in favor of or in opposition to further European integration has a positive effect on electoral participation ( Figure 5(d)). Therefore, H2 receives only qualified empirical support, as it only applies to a single issue. However, this finding is remarkable and represents a significant contribution to the EP election literature. Both Europhile and Eurosceptic voters apparently feel more incentivized to express their views at the ballot box during electoral contests on the EU level than ambivalent citizens. The broader implication of this finding is that 'Europe matters' for EU electorates and voter turnout in contemporary European elections.

Conclusion
The surprising increase in EU-wide voter turnout in the 2019 EP election has been largely attributed to increased political conflicts over key European issues and a greater degree of politicization in these elections. Against this backdrop, we investigated whether European citizens were indeed mobilized through four major policy issues in the 2019 EP election: the economy, environment/climate change, immigration, and European integration itself. Using survey data from two different sources, we assessed our research questions from a cross-national perspective. Overall, we conclude that the four issues under investigation all played a role in mobilizing European citizens to cast a vote. More specifically, we found empirical evidence for three different effects of issue relevance.
First, the primary pathway of issue mobilization is a direct effect exerted by personal issue importance. The more relevant a person perceives key policy issues (economy, environment, and immigration), the more likely she feels incentivized to participate in European elections. On the one side, we believe that this effect speaks for the electoral impact of citizens' political awareness more generally. The more people care about political issues, the more they strive to participate in politics. On the other side, the particularly pronounced effects of environmental and climate change issues reveal that issue mobilization is not homogeneous across all political issues. We believe that the differential effects might be a consequence of the great political urgency felt by people who worry about climate change (see also Schäfer et al., 2021b). This sense of urgency was also expressed in the 'Fridays for Future' movement and, subsequently, in the 'green wave' during the 2019 EP election in many.
Second, we observed that the impact of personal issue importance was enhanced by the systemic salience that the respective policy issues had during the election campaign. It confirmed our expectations derived from the literature on agenda-setting and the spiral of silence. People observe the 'opinion climate' around them and feel additional incentives to participate in EP elections when the issue they perceive as most relevant is also publicly salient in their country. In contrast, if people feel socially isolated with their personal salience attributions, they tend to stay 'silent' (i.e. at home) during electoral contests on the EU level. Although this finding is an important contribution to the EP election literature, it is in line with the traditional second-order model of European elections. National contexts are still highly relevant when it comes to issue mobilization during European elections.
Third, having a particularly strong or 'extreme' opinion on one of the three issues mentioned above (economy, climate change/environment, and immigration) has virtually no mobilizing effect in European elections. However, we have found such an effect for the EU integration issue. This observation indicates that citizens use European elections to express their views on the European integration process, which is in line with the 'Europe matters' model of EP elections, claiming that the consequences of European integration 'become politically more important and more contested' (Van der Brug and Van der Eijk, 2007: 226) in the electoral arena. Since EU issues seem to play a role in electoral participation (Braun, 2021), future studies should ultimately consider the multi-faceted nature of European issues and, for instance, distinguish between so-called EU policy and EU polity issues (Braun et al., 2016). It seems very plausible that different types of European issues have diverging electoral consequences in the EU member states. 7 What are the implications of our study for the future of EP elections and empirical research? Although we do not contest that European elections can still be considered second-order national elections, we also found evidence for their increasing politicization and the conclusion that 'Europe matters' in present-day EP elections (Schäfer, 2021). Moreover, our results show that the debate and electoral campaigns over crucial transnational policy issues bear the potential to increase turnout in European elections further. It might be through this process that EP elections can finally lose their second-order character and establish 'the missing link' between voters and parties on the European level. However, it is important to note that key political issues do not necessarily need to be explicitly related to European integration itself but can be of European-wide concernsuch as environmental or economic issues. Against this backdrop, future research could investigate political actors' potential to increase turnout via an intensified debate over transnational policy issues. Linked to this, it would be worthwhile to analyze whether actual (and perceived) competencies in specific policy areas ascribed to the EU or the national level of governance have different implications for electoral participation. Insights on these two aspects will further strengthen our knowledge on the multi-level logic of European politics and the future of European integration more generally.

Declaration of conflicting interests
The authors declared no potential conflicts of interest with respect to the research, authorship, and/ or publication of this article.

Funding
Daniela Braun has conducted this study on behalf of the project 'Conflict structuring in European Elections' funded by the German Research Foundation (DFG) between 2016 and 2020. The project is directed by Daniela Braun and Edgar Grande at the LMU Munich and the WZB Berlin. Constantin Schäfer has received financial support from the European Union's Horizon 2020 research and innovation programme under Grant Agreement no 770142. RECONNECT is a four-year multidisciplinary research project on 'Reconciling Europe with its Citizens through Democracy and the Rule of Law'.

Supplemental material
Supplemental material for this article is available online.

Notes
1. See also The Guardian (2019): A quiet revolution sweeps Europe as Greens become a political force. Available at: https://www.theguardian.com/politics/2019/jun/02/european-parliamentelection-green-parties-success (accessed 22 June 2021). Overall, Green parties achieved doubledigit shares of the vote in 11 countries and the Green-EFA political group in the EP increased its seat share from 6.9% to 9.5%. 2. It is important to note, however, that the European integration issue is a multifaceted one (see also introduction of the SI), which is cross-cutting numerous other (politicized) policy issuesespecially those at the socio-economic and a socio-cultural dimension (Schäfer et al., 2021a). Therefore, to avoid intercorrelation with other issues, we focus mainly on the EU polity and neglect the dimension of EU policy (Braun et al., 2016). 3. Although the systemic salience of a policy issue could also be measured with elite-level, manifesto or media data, we opted for aggregated survey data for at least two reasons. First, we assume that aggregated survey data roughly reflects public opinion and, thus, the societal-issue agenda in an appropriate way. Second, neither manifesto nor media data has yet been made available for the 2019 EP election for all EU member states. 4. The new scales, in theory, take values from 0 to 10. However, as country means do not take extreme values, the observed differences between a respondent and her country mean range between 0-6.5 (economy), 0-6.4 (migration), 0-7.4 (environment), and 0-6.8 (EU integration). 5. The choice to include the official 2019 EP election turnout as a control variable is a means to avoid including multiple contextual variables that affect participation in European elections. This is done because the multi-level models contain only a small number of higher-level units (28 countries) and thus exhibit a very limited number of degrees of freedom.
6. Please note that the cross-level interaction effect of economic issues was insignificant in a model specification that integrated all policy issues simultaneously (see the Online appendix). 7. This is supported by the insights of two papers of this SI (Katsanidou et al., 2022;Pellegata and Visconti, 2022) showing that the concept of transnational solidarity requires a differentiation between different types of European integration. Findings from a preliminary analysis of the EEC online survey (Braun and Grande, 2021), illustrating that EU budgetary questions have an effect on electoral participation in Austria, whereas in Germany this is the case for EU immigration, and in France and Sweden for EU taxation (results not shown), underline this idea.