The year 2019 saw not only new protest waves in the wake of climate change (Taylor et al. Reference Taylor, Watts and Bartlett2019; Wahlström et al. Reference Wahlström, Kocyba, De Vydt and de Moor2019), but the European Parliament elections, too, were described as a showdown, namely between the ‘Green wave versus right-wing populism’ (Waldholz Reference Waldholz2019). Attached to such descriptive accounts and translated to a large body of academic research is the idea that (successful) mobilization is recurrent and dependent on contextual factors – that is, political opportunity structures and (discursive) windows of opportunity (Chiaramonte and Emanuele Reference Chiaramonte and Emanuele2017; Koopmans and Muis Reference Koopmans and Muis2009; Spoon and Klüver Reference Spoon and Klüver2019). Moreover, the outcomes of political campaigns and elections have become more and more difficult to predict because of increasing electoral instability, turnout fluctuations and party switching, which are frequent patterns throughout industrialized democracies (Dalton et al. Reference Dalton, McAllister, Wattenberg, Dalton and Wattenberg2000; De Vries and Hobolt Reference De Vries and Hobolt2020; Emanuele et al. Reference Emanuele, Chiaramonte and Soare2020; Mair Reference Mair2008). Focusing on individual-level characteristics, in this article we study to what extent short-term changes in political concerns can explain electoral volatility. Do rapid changes in the most salient issues for the respondents translate into changing prospects for turnout and party switching? And, since young people tend to be less supportive of traditional party politics (Franklin Reference Franklin2004; Mair and van Biezen Reference Mair and van Biezen2001), what role does age play in accounting for turnout and party switching?
First, as the bulk of the literature suggests (e.g. Franklin Reference Franklin2004; van Biezen et al. Reference van Biezen, Mair and Poguntke2012), our results with individual-level panel survey data show that young people are less willing to cast a vote but are at the same time more likely to switch their party preferences. Second, we shed light on issue salience's asymmetric impact on volatility. On the one hand, we explore whether being concerned about material/economic issues does translate into increased electoral volatility. One would expect that material concerns would be linked to grievances, bearing a reactive component and possibly inviting electoral turnout and party switching. However, empirical support for this argument is mixed: increased material concerns are not associated with electoral turnout, and the effects of concerns about inflation and unemployment on party switching are contradictory. Although concerns over rising prices and the fall in the purchasing value of money push people to switch their electoral preferences, widespread apathy, despair and anxieties about meagre economic prospects seem to counterbalance the positive effect of economic concerns on volatility. On the other hand, and similar to unemployment, embracing so-called postmaterialist aspects such as immigration – and, to a qualified extent, terrorism – as the most salient issues increases electoral turnout but decreases party switching. While the politicization of these issues activates partisan constituencies, it does not lead to a change in party preferences but largely reaffirms already existing political divides. However, climate concern works differently, as it invites change in electoral preferences.
Notwithstanding increased electoral volatility – and notwithstanding some asymmetric effects of issue concerns on electoral volatility – we believe that visible and important political divisions are still in place and tend to be reinforced by recurring changes in issue salience. If established political actors ignore such cleavages, the result could be a further undermining of party attachment, instead of working against the trend of volatile electorates for any party. In fact, political parties are becoming more responsive to the mean voter, which, in turn leads to a further erosion of their core electorate (Dassonneville Reference Dassonneville2018) given the diverse pluralist interests in society not covered by the mean voter. Moreover, on climate concerns, young people may be less prone to react along traditional party ideological lines than over more postmaterialistic and/or pressing political concerns.
We investigate hypotheses derived from the literature through original panel survey data. While accounting for longitudinal variation at the individual level, these data also allow us to control for different political and cultural contexts, as they include information from nine European countries, namely France, Germany, Greece, Italy, Poland, Spain, Sweden, Switzerland and the United Kingdom. The article is structured as follows. In the next section, we review relevant literature and present our theoretical framework on the interplay between issue salience and electoral volatility. After that we introduce our data and methodological design. In the subsequent section, we discuss our main results. In the concluding part we summarize the main contributions of the article, reflect on the broader implications for political engagement and policymaking, and signal some avenues for further inquiry.
Young people, volatility and issue salience
Volatility refers to change in voting behaviour between elections. However, this change may relate to different aspects for different scholars. In their seminal study on campaign effects, Paul Lazarsfeld et al. (Reference Lazarsfeld, Berelson and Gaudet1948) distinguished between three different processes linked to volatility: conversion, where people start out with a voting intention but end up voting for another party; crystallization, namely acquiring a vote choice during the election campaign; and reinforcement, the strengthening of the preference for the original voting decision. Beyond their obvious differences, all these processes represent, in the context of an electoral campaign, a change that affects electoral turnout and/or electoral preferences.
Broadly speaking, we focus on two of the main mechanisms that may produce electoral volatility: differential turnout and party switching.Footnote 1 Differential turnout refers to the electoral involvement of some voters and, conversely, the deactivation of others – as already noted, ‘the potential effect of abstention on election outcomes is quite high, even in countries with high voting rates’ (Boyd Reference Boyd1985: 521). By switching parties, we refer to ‘an active behaviour on the side of voters who decide to choose different parties in consecutive elections’ (Gómez Reference Gómez2018: 174).Footnote 2 Much empirical research to date has addressed volatility at the level of parties and elections, looking at the overall shifts in party system and party support in society (e.g. Gómez Reference Gómez2018; Mair Reference Mair2008; Pedersen Reference Pedersen1979; Tavits Reference Tavits2008). In fact, ‘the choices of elites may be more responsible for instability in the early stages of party system development than the erratic behaviour of voters’ (Tavits Reference Tavits2008: 537). This point is supported by recent studies of (large) parties trying to ‘own’ the issues of challenger parties (Abou-Chadi et al. Reference Abou-Chadi, Green-Pedersen and Mortensen2020) and on issue-ownership attack, where competing parties try to reframe and blame an issue ‘owned’ by another party (Seeberg Reference Seeberg2020).
However, at odds with the emphasis on the aggregate and meso levels of analysis, in this article we build on the individualization thesis – that is, that individuals are increasingly required to construct their own lives because of social changes in late modernity (Beck Reference Beck2001). In other words, we argue that individual interests and preferences can also be key drivers of volatility. Along these lines, Peter Mair (Reference Mair2008) understands that cleavages offer several constraints, tying voters down to various social and organizational identities. In his study of the Dutch case, he shows that ‘depillarization, secularization and individualization have obviously left … voters with scarcely anything with which to anchor themselves into place in terms of cleavages and other social identities’ (Mair Reference Mair2008: 240). As cleavages and partisanship no longer account for the voting behaviour of a cognitively mobilized electorate, turnout and voting choices are more likely to depend on short-term factors (Dalton et al. Reference Dalton, McAllister, Wattenberg, Dalton and Wattenberg2000; Dassonneville Reference Dassonneville2013).
In short, increasing volatility is leading to far-reaching consequences and changes in political environments across advanced industrial democracies. Voters seem to be increasingly prone to switching parties and breaking traditional allegiances (De Vries and Hobolt Reference De Vries and Hobolt2020: 71–80). This is related to the dealignment thesis in a context of individualization – that is, a large portion of the electorate in advanced democracies abandons its previous partisan membership and identification, and thus does not show stable voting patterns but chooses who to vote for on the issues of the day.
With this article we make a few moves that are uncommon in the literature on volatility. First, to avoid dangers related to ecological fallacy we firmly believe that examining individual-level determinants of volatility is a critical task. The literature on the determinants of volatility with micro-data exploring the characteristics and attitudes of volatile voters, such as political sophistication and satisfaction with democracy, has mostly reached mixed, inconclusive results. Conflicting arguments have been put forward for a positive, negative or even curvilinear relationship between political sophistication and volatility (Dassonneville and Dejaeghere Reference Dassonneville and Dejaeghere2014; Dassonneville and Stiers Reference Dassonneville and Stiers2018; Lachat Reference Lachat2007). Similarly, some evidence suggests that frustration with politics and dissatisfaction with the political system will increase volatility – although this has been disputed as well (Dassonneville and Stiers Reference Dassonneville and Stiers2018; Voogd and Dassonneville Reference Voogd and Dassonneville2020; Zelle Reference Zelle1995). Second, we use (two-wave) panel data with interviewees responding to the same set of questions, so we can track actual longitudinal change in attitudes – including issue salience at the personal level – and voting information. To be sure, studies on the determinants of volatility have used panel data before. For example, Sabine Geers and Jesper Strömback (Reference Geers and Strömbäck2019) find that political knowledge influenced patterns of electoral volatility in the 2014 Swedish national election – not for stored knowledge in the first wave but only for acquired political knowledge across subsequent waves. Third, in contrast to existing case studies, we engage in a comparative study of nine European countries with panel data on the same individuals. Fourth, with this type of empirical evidence, we give age variables a central place in our account and explore how generational aspects play out in volatility.
As young people are a dynamic subset of the population, they are supposed to be particularly affected by the above-mentioned developments. At the same time, however, young people have often been depicted as one of the least politically engaged groups in the population, with little interest in political issues on average, often falling into apathy and disaffection towards politics (Blais et al. Reference Blais, Gidengil and Nevitte2004). Indeed, relative to older cohorts, young people tend to move away from more formal, institutional political activity like voting and working through political parties (Franklin Reference Franklin2004; Grasso Reference Grasso2016; Mair and van Biezen Reference Mair and van Biezen2001). While a body of literature has forcefully shown that young people are keener to engage in activities like political consumerism, volunteering, protesting and digital activism than older people, the former still turn out to vote at lower rates (Dalton Reference Dalton2009; Earl et al. Reference Earl, Maher and Elliott2017; Zukin et al. Reference Zukin, Keeter, Andolina, Jenkins and Delli Carpini2006).
In contrast, in terms of party switching, there is evidence and reason to assume that young people might be more volatile. Political socialization literature suggests that young people might be more open to changes in political identity, party affiliation and electoral loyalty (Niemi and Sobieszek Reference Niemi and Sobieszek1977: 223); moreover, their political attitudes (and the perceived importance of issues) are less stable (Alwin and Krosnick Reference Alwin and Krosnick1991; Niemi and Klingler Reference Niemi and Klingler2012). While young people are influenced by their parents, and form their disposition relatively early, this subset of the population can only start voting at the age of 18 when their ‘preferences may gradually become associated with their adult correlates such as social characteristics and issue attitudes’ (Rekker et al. Reference Rekker, Keijsers, Branje and Meeus2019: 49). In general terms, young people, who are often still in education and only starting to move out of the parental home, are also exposed to peers, the media and (life and exogenous) events that might lead to a change in their political attitudes and behaviour (see Pew Research Center 2020). Indeed, net of time-related (period and cohort) effects, a significant negative effect of age on party switching has been reported for the Netherlands between 1971 and 2010 (Dassonneville Reference Dassonneville2013). Thus, we will test whether:
Hypothesis 1A: Young people turn out to vote to a lesser extent than older people
and
Hypothesis 1B: Young people are a more volatile subset than older people in terms of party switching.
We will first check if we really observe a greater probability among young people to change their main issues of concern as compared to older age groups. This could help understand why young people are more volatile in terms of turnout and party switching. Apart from this reasoning connected to our first set of hypotheses, we propose to zoom further into the explanatory role of changing issue salience. Salience refers to ‘the degree to which a person is passionately concerned about and personally invested in an attitude’ (Krosnick Reference Krosnick1990: 60). In other words, salience designates the importance of issues, particularly for voters (Wlezien Reference Wlezien2005). As James Dennison notes, ‘in spite of the increased use of issue salience in recent years, with impressive explanatory results, the concept of issue salience remains underspecified and, at times, contradictory and … its antecedents remain relatively unknown’ (Reference Dennison2019: 436). Salient issues are important drivers of the decision to vote and the specific vote orientation, but issue salience is also important because large proportions of people can accurately perceive the differences between the parties on the issues that were salient to them (RePass Reference RePass1971). Moreover, as Catherine De Vries and Sara Hobolt argue, voters choose challenger parties ‘on the basis of high appropriability issues and motivated by antiestablishment considerations’ (Reference De Vries and Hobolt2020: 183). According to the issue-ownership theory of voting (Budge and Farlie Reference Budge and Farlie1983; Lefevere et al. Reference Lefevere, Tresch and Walgrave2015; Petrocik Reference Petrocik1996), voters will cast a ballot for the party that is the most credible proponent of a particular issue (Bélanger and Meguid Reference Bélanger and Meguid2008; Neundorf and Adams Reference Neundorf and Adams2018). By way of example, through individual-level analyses of vote choices in the 1997 and 2000 Canadian federal elections, the effect of issue ownership on vote choice was found to be ultimately conditioned by the perceived salience of the issue in question (Bélanger and Meguid Reference Bélanger and Meguid2008). In sum, we focus on issue-specific mechanisms because, depending on the issue at stake – however it was made salient, we only capture if issue saliency has changed, not why – and the part of the electorate concerned with that particular issue, the impact of increased concern on volatility might differ.
Looking at the change in the political space in its economic and cultural dimensions, a general distinction has been made between materialist and postmaterialist values and issues (Inglehart Reference Inglehart1977), which arguably conflate left–right and libertarian–authoritarian values (Grasso and Giugni Reference Grasso and Giugni2019). Materialist people tend to be preoccupied with satisfying immediate physiological needs (Inglehart Reference Inglehart1977; Inglehart and Catterberg Reference Inglehart and Catterberg2002). Indeed, concern for materialist issues such as the state of the economy, unemployment, inequality, inflation or government debt is likely to be connected to (perceived) personal affectedness (Singer Reference Singer2011). While the economy is often the most important issue in an election, it is even more likely to dominate other issues of concern in contexts of high volatility (Singer Reference Singer2011). In line with recent research on political protest (Grasso and Giugni Reference Grasso and Giugni2016; Kurer et al. Reference Kurer, Häusermann, Wüest and Enggist2019; Portos Reference Portos2021), one would expect citizens who perceive a deterioration in their economic prospects and who experience feelings of relative deprivation to become more worried about material issues. These citizens will seek to voice socioeconomic grievances and anxieties by increasing their political participation and turning out to vote. As the economic voting literature argues, voters hold the government accountable for the state of the economy; indeed, ‘the citizen votes for the government if the economy is doing all right; otherwise, the vote is against’ (Lewis-Beck and Stegmaier Reference Lewis-Beck and Stegmaier2000: 183). Not only do sociotropic retrospective economic evaluations influence vote choice (Lewis-Beck and Stegmaier Reference Lewis-Beck and Stegmaier2000, Reference Lewis-Beck, Stegmaier, Congleton, Grofman and Voigt2019), but recent evidence suggests that voters may turn against parties that have introduced austerity measures and similar policies (Bojar et al. Reference Bojar, Bremer, Kriesi and Wang2022). In sum, a higher salience of economic issues should lead to higher volatility. Thus, we will test whether:
Hypothesis 2A: An increase in the salience of materialist issues increases electoral turnout
and
Hypothesis 2B: An increase in the salience of materialist issues increases volatility in terms of party choice.
While traditional social movements focused on issues of labour, ‘new social movements’ emerged in the 1960s around concerns such as women's liberation and environmental protection (Della Porta and Diani Reference Della Porta and Diani2020). As these movements declined in the 1970s, increasing support for environmental issues did not, which eventually led to the emergence of Green parties (Müller-Rommel Reference Müller-Rommel1998). Changes in postmaterialist values lie behind these shifts in Western party systems, electoral behaviour and broader political conflict structures, with the emergence of Green parties in (Western) Europe in the 1980s as well as the recent climate protests being a case in point (de Moor et al. Reference de Moor, Uba, Wahlström, Wennerhag and de Vydt2020; Zamponi et al. Reference Zamponi, Baukloh, Bertuzzi, Chironi, Della Porta and Portos2022). Connected to the politicization of issues such as crime, law and order, immigration and terrorism, another issue is the resurgence of the radical right in many countries. For the case of Israel, Claude Berrebi and Esteban Klor (Reference Berrebi and Klor2006, Reference Berrebi and Klor2008) have shown that terror attacks and, consequently, the issue salience of terrorism resulted in electoral success for the right. Similar effects have been shown for Turkey (Aytaç and Çarkoğlu Reference Aytaç and Çarkoğlu2021). Based on Eurobarometer survey data, another study confirms that terror attacks also pushed voters towards right-wing preferences in 12 European Union countries (Economou and Kollias Reference Economou and Kollias2015).
Likewise, the salience of immigration as a contested topic in public debates contributes to explaining the success of right-wing parties (Dennison and Geddes Reference Dennison and Geddes2019). Dennison (Reference Dennison2020: 398) argues that the salience of immigration as an issue plays into ‘economic competition, cultural backlash, and political demography’ – factors that are commonly used to explain the success of right-wing populist milieus. It is important to emphasize that the reason for this electoral success is not an increase in anti-immigrant sentiment or even (increasing) immigration itself, but rather the effect of the salience of the topic (Dennison and Geddes Reference Dennison and Geddes2019). The far-right Alternative for Germany, for example, capitalized on this issue particularly among right-wing non-voters but also even among non-voters from the whole spectrum, potentially including supporters of other parties (De Vries and Hobolt Reference De Vries and Hobolt2020: 219f.). In sum, increasing salience of these issues may appeal to additional voters (e.g. from climate strikes) and open up opportunities for small parties to capitalize on. Therefore, we will test whether:
Hypothesis 3A: An increase in postmaterial issue salience – that is, concern for climate, immigration and terrorism issues – increases electoral turnout
and
Hypothesis 3B: An increase in postmaterial issue salience – that is, concern for climate, immigration and terrorism issues – increases volatility in terms of party choice.
Data and operationalization
In the framework of a large collaborative research project, EURYKA, we built a representative panel survey that includes nine European countries, namely France, Germany, Greece, Italy, Poland, Spain, Sweden, Switzerland and the United Kingdom.Footnote 3 At the European level, these nine countries cover a lot of variation in terms of regional scope (east/west, north/south), economic performance (especially after the Great Recession), political culture and developments. These data have two unique features. First, this is a sample that is representative of the general population in each country: a specialized polling agency collected the data ad hoc through administered online panels using balanced country quotas in terms of sex, age, region and education level to match national population statistics (EURYKA 2018).Footnote 4 According to Kevin Kiley and Stephen Vaisey (Reference Kiley and Vaisey2020), attitude change – and forecasted vote – with panel data is more likely to occur for younger people, variables related to public opinion, and for exceptionally salient issues. The EURYKA panel survey data are likely to account for this high figure for electoral volatility, also providing an excellent resource with which to closely examine to what extent these flows are determined by shifting issue salience. Second, it is a panel study with the first wave conducted during 2018 and the second wave approximately 12 months later (Figures A1–A2 in the Online Appendix). In total, 7,240 respondents have been interviewed in the two panel waves (for the country distribution of respondents, see Figure A3, Online Appendix).
To build the main dependent variable, we use prospective voting information as a key indicator of electoral volatility and turnout. First, we ask, ‘If there were a general election in your country tomorrow, would you vote?’ (‘I would probably vote’ = 1; ‘I would probably not vote’ = 0; ‘I am not eligible to vote’ is recoded as missing). On average, 86% of respondents in Wave 1 declared they would turn out to vote, and 87% did so in Wave 2. Prospective turnout change varied across countries, decreasing by 2% in the United Kingdom and increasing by 4% in Spain and Greece (see Figure 1, top). Second, in order to measure party switching, we ask ‘for which party?’ Overall, we provide information on eight parties with the highest vote intention for each of the nine countries, plus the options ‘other’ and ‘don't know’. Since we are only interested in overall change, not in the specific shift of votes, we create a dummy variable depending on whether the person has changed their vote intention across the two waves between the 10 categories. We observe an average party switching of 57% between the two waves, with the UK (46%) and Greece (76%) reporting the minimum and maximum values, respectively (see Figure 1, bottom). In other words, 4,155 of 7,240 interviewed persons in the nine European countries covered throughout declared that they had changed their party preferences between 2018 and 2019. To mitigate for the limitations of our measurement – namely that prospective information is not a good indicator of volatility, national elections take place at different times in different countries, and our indicators refer to national elections – we replicate our results with retrospective (for Wave 1)/prospective (for Wave 2) electoral turnout and party-switching information at the former/forthcoming European Parliament elections, as well as with party attachment (see Figure A13, Online Appendix).Footnote 5
Figure 1. Individual Average Change in Prospective Electoral Turnout across Waves by Country (top) and Mean Value of Party Switching by Country (bottom)
Our key predictor is constructed from the question, ‘Personally, what are the two most important issues for you at the moment?’ The phrasing deliberately focuses on the individual level – thus overcoming the problem of confusing the importance of issues and the degree to which issues are a problem at the national level (Wlezien Reference Wlezien2005). We consider concern for six issues (out of 17 possible items),Footnote 6 namely unemployment, inflation, the economic situation, immigration, climate and terrorism. These items were picked on the basis of two criteria: on the theoretical level, they are representative of the material–economic and postmaterial–cultural issue dimensions, respectively; moreover, empirically we found statistically significant variation in these items between the panel waves (see Figures A4–A5, Online Appendix) but almost negligible variation in the other items. Overall, we observe fluctuations in issue concerns, with three of the issues in Wave 1 (unemployment, terrorism and immigration) decreasing their salience in Wave 2; conversely, concern over the economic situation, inflation and climate increased between panel waves. Note that these trends are uneven across countries, with concerns over the economy increasing, particularly in Italy and France, concerns over immigration increasing (instead of decreasing) in Greece and Spain, and concerns over terrorism decreasing, especially in the United Kingdom (see Figure 2).
Figure 2. Individual Average Change in Choosing Unemployment, Inflation and Economic Situation as Main Concerns by Country (top); Individual Average Change in Choosing Climate, Immigration and Terrorism as Main Concerns by Country (bottom)
Next to change in main issues of concern, we are interested in the age group to which the respondents belongs. We employ this as an ordinal variable (1 = 18–24, 2 = 25–34, 3 = 35–49, 4 = 50–64, 5 = 65+ years old) and keep it invariant across time. Obviously, age varies across time, but not at a different rate between individuals.Footnote 7 Regarding variation across age groups, Figure 3 shows that, overall, the younger cohorts do not change their main concerns more frequently – for instance, average change of concern is −0.003 among 18–24 year-olds and −0.010 among the oldest age group. While unemployment (climate) concern dramatically decreased (increased) among the youngest adults, terrorism and immigration concerns strongly decreased their salience among the +65 year-old subset. We will study below whether age and changing concerns interact in accounting for volatility.
Figure 3. Individual Average Change in Choosing Unemployment, Inflation, Economic Situation, Climate, Immigration and Terrorism as Main Concerns by Age Group
In order to strengthen our arguments against alternative explanations, we use standard controls both in individual propensity to vote and party-switching bodies of literature. First, we know that different sociodemographic characteristics explain variations in turnout and party preferences (Blais Reference Blais2000; Dassonneville Reference Dassonneville2013; Wolfinger and Rosenstone Reference Wolfinger and Rosenstone1980), such as level of education, sex and urban/rural residence. We use a three-point ordinal scale for education – primary, secondary and higher education. A dummy variable measures the respondent's sex assigned at birth (1 = female; 0 = male).Footnote 8 Sex is, like age group, treated as a time-invariant indicator in the panel regression models reported throughout. Since urban citizens might be more likely to turn out to vote and switch party preferences (Gimpel et al. Reference Gimpel, Lovin, Moy and Reeves2020; McKee Reference McKee2008), we also control for the urban–rural cleavage through a 1–5 scale that best describes the area in which the respondent lives (1 = ‘big city’; 2 = ‘suburbs or outskirts of a big city’; 3 = ‘town or small city’; 4 = ‘country village’; 5 = ‘farm or home in the countryside’). The experience of economic strain is captured through perceived relative deprivation, specifically if the respondent acknowledges that they have ‘experienced real financial difficulties (e.g. could not afford food, rent, electricity) in the past 12 months’ or not.
Besides sociodemographic factors, we also consider change in political values and social capital, which are usual determinants in research on electoral turnout/abstention and party switching (see, e.g. Dassonneville and Dejaeghere Reference Dassonneville and Dejaeghere2014; Dassonneville and Stiers Reference Dassonneville and Stiers2018; Pattie and Johnston Reference Pattie and Johnston1998; Portos et al. Reference Portos, Bosi and Zamponi2020). Left–right ideology is measured through a self-placement scale that ranges from 0 (= left) to 10 (= right). As an indicator of political sophistication, we also measure how often the respondent gets political information through various media sources.Footnote 9 To account for social capital, we include information on how often the respondent has ‘met socially with friends not living in [his/her] household’ during the past month (1 = almost every day; 2 = every week; 3 = once or twice a month; 4 = less than once a month) and how many social media accounts the respondent has.Footnote 10 Descriptive statistics of the variables used throughout are reported in Table 1.
Table 1. Summary Statistics of Individual-Level Predictors and Controls in Wave 1 and 2
Results and discussion
To test our hypotheses, we fit several logit panel regression models with random effects so that we can keep time-invariant age (and sex). While in Model 1 we study to what extent shifts in our material–economic and postmaterial–cultural concern variables account for variation in electoral turnout (both time-variant and time-invariant), sociodemographic controls are included in Model 2 (Table 2). In Model 3 (Table 2) we add controls related to political values and social capital. Models 4, 5 and 6 replicate specifications 1, 2 and 3, replacing change in electoral turnout with party switching as the dependent variable (Table 2). We also include country fixed effects in all models to account for the specific (institutional, historical and cultural) characteristics of each country that may correlate with changes in prospective turnout and volatility in terms of party preferences. Table A1 in the Online Appendix reports a correlation matrix of predictors and control variables. We replicate Table 2 with country fixed effects to test for within-individual variation, removing the time-invariant controls (Table A2, Online Appendix); we also replace the dependent variables with retrospective/prospective turnout and party switching in the European Parliament elections and changing party attachment in Table A3 in the Online Appendix.
Table 2. Panel Logit Regression Models with Random Effects
Notes: Dependent variable: electoral turnout (Models 1, 2 and 3) and party switching (Models 4, 5 and 6). ° p < 0.10, * p < 0.05, ** p < 0.01, *** p < 0.001. Coefficients are log odds.
Overall, empirical support for our issue-salience hypotheses is mixed (Table 2; Figure 4; Tables A2–A3 and Figure A14, Online Appendix). While concern over unemployment has a negative effect on party switching, concern over inflation undermines constant party preferences – these effects are not large in substantive terms (Figures A7, A9, A14 and A16, Online Appendix). Moreover, salience of other postmaterial issues like immigration and terrorism decreases party switching, but the positive effect reported for climate is strong, significant and robust (Figures A7, A9, A16 and A17, Online Appendix). In contrast, increased salience of economic and materialist issues among respondents does not increase electoral turnout (Figure 4; Figures A6, A8, A14 and A15, Online Appendix). In sum, we observe asymmetric effects of issue concerns on volatility: whereas H2A must be rejected, H3A is (partially) confirmed as we find some support for the effect of postmaterialist issue salience on electoral turnout. Empirical evidence is mixed for H2B and H3B: while it is confirmed that inflation and climate concerns are positively associated with party switching, the opposite holds for other issues like unemployment, immigration and terrorism.
Figure 4. Plot of Coefficients from a Logit Panel Regression with Random Effects (with Age Group, Sex and Country Dummies as Time-Invariant Variables)
Notes: From Models 3 and 6, Table 2. Country dummies are not plotted but are included in the model specification. Dependent variable: electoral turnout/party switching.
In terms of age, our results confirm our expectations: young people are less likely to vote but they are more likely to change their electoral choices (H1A and H1B, see Table 2; Figure 4; Table A2 and Figure A14, Online Appendix). We find not only that young people engage to a low extent through formal mechanisms of participation (Blais et al. Reference Blais, Gidengil and Nevitte2004; Dalton Reference Dalton2009) but also that party loyalties and attachments are less stable among the younger cohorts. It has been noted that young people were overrepresented among protesters against precarious work and racism (Cini et al. Reference Cini, Maccarrone and Tassinari2022; Della Porta Reference Della Porta2019). A recent study found that parties with strong ties to civil society organizations have a more stable voter base (Martin et al. Reference Martin, de Lange and van der Brug2022). Our findings, too, imply that politicization of the above-mentioned issues does trigger electoral turnout, but it also crystallizes party preferences: at least to a qualified extent, citizens participate based upon previously existing issue alignments. Respondents concerned with immigration or unemployment will feel urged to turn out to vote for parties ‘owning’ these issues.
Indeed, research on the supply side of electoral participation could help us interpret our results: we know that organizational settings and strategies tend to be rather slow to change because parties often choose to maintain their reputational advantage and focus on their historically strong issues (Aragonès et al. Reference Aragonès, Castanheira and Giani2015). The stronger and more established the party's reputation on key (material) issues is, the less competitive the election becomes and therefore the less likely it is that the party will try to ‘steal’ the issue from its opponents (Aragonès et al. Reference Aragonès, Castanheira and Giani2015). Hence, if voters are more concerned about materialist topics, it does not seem to have a clear impact on volatility. This could be explained by the fact that fears and anxieties over material conditions and accompanying emotions (stress, social isolation and feelings of alienation) often trigger not political participation but apathy (Schaub Reference Schaub2021). Citizens who become increasingly concerned over material issues such as unemployment opt for resignation, embracing risk-averse options and staying with parties that ‘own’ relevant issues, thus reinforcing previously existing preferences and allegiances (economic stability on the right and social justice on the left). This would mean that we see less party switching. The salience of inflation seems to be the exception here, as concerns over generalized rises in prices seem to foster indignation, further undermining traditional party allegiances.
As already suggested, the activation of new groups of voters around postmaterialist concerns could stand behind the mobilizing potential of these issues. For instance, the climate emergency and the subsequent campaigns of the Fridays for Future and Extinction Rebellion movements or the hundreds of deaths in the Mediterranean in recent years following the 2015 long summer of migration have led to mass movements across European countries (BBC News 2020). The politicization of issues such as climate or immigration could push people to cast a vote, but these effects are not robust to model specifications with fixed effects. However, two different patterns can follow pre-established party choices, which can be explained by the two above-mentioned mechanisms. On the one hand, the increased salience of immigration and terrorism activates fears and anxieties, rarely leading people to the risky choice of changing party preferences, a factor that might in turn help mobilize the electoral potential of existing postmaterialist cleavages. But this mechanism does not operate in the same way for all groups of the population: among those who are concerned about immigration, younger cohorts are much more likely to switch party preferences than older people in order to give voice to this pressing concern (Table A4 and Figure A12, Online Appendix).Footnote 11 Similarly, people worried about the climate are more likely to switch party preferences, as indignation with the way this issue was handled by the previous electoral choice prevails.
Besides the hypotheses related to issue salience and age, we also find significant asymmetric effects of other control variables on volatility (Table 2; Figure 4). In line with the extant literature, increasing one's level of education, meeting more often with friends, social media presence and following political news are positively associated with electoral turnout. Conversely, these controls have a negative impact on the propensity to change voters' partisan choices. However, these effects are not always robust (Tables A2, A3 and A4 and Figure A14, Online Appendix). In addition, being a woman is associated with increased willingness to change party preferences. Crucially, experiencing real financial difficulties hinders electoral turnout but seems to facilitate party switching (see Giugni and Grasso Reference Giugni and Grasso2019). Hence, socioeconomic grievances depress participation but, among those who turn out to vote, experiences of deprivation facilitate the break with party loyalties to give a political voice to personal financial hardship and meagre circumstances – again this effect does not meet all robustness checks (Tables A2, A3, A4 and Figure A14, Online Appendix).
Conclusion
In this article we set out to investigate specific individual-level drivers of electoral volatility across nine European countries with panel survey data. We tested whether changing issue salience among voters leads to altered party choice and change of turnout. Recent economic crises, the climate emergency as well as migration and subsequent anti-racist protest against official migration policy could be driving volatility, especially among young voters. Our findings corroborate that issue salience is relevant, although the hypotheses derived from the literature receive only partial support. In this respect, we provide important insights into the mechanisms of volatility.
We can conclude that increased salience of materialist issues does not lead to higher turnout. While there was reason to believe that grievances and frustration with economic policies increase turnout, an experience of deprivation supresses voting. Moreover, increasing postmaterialist issue salience increases electoral turnout, as we can observe from the Fridays for Future demonstrations and the success of Green parties among young voters. However, an increase of salience of postmaterialist issues has a mixed impact on volatility in terms of party switching. Parties that own these issues tend to benefit from increasing salience, leading to higher turnout and, in turn, we observe a decrease in volatility. We can find examples for this development on the political right, where extremist parties are successfully ‘owning’ migration issues, and on the left where social-democratic parties are able to stabilize their voter base against the general likelihood of changing party preferences. This is in line with prior evidence showing how unemployed and labour market outsiders prefer parties that give priority to social protection and welfare, thus voting ‘with their wallet, not with their guts’ (Negri Reference Negri2019). Lastly, we confirm that young people generally turn out to vote to a lesser extent than older people and are also more volatile in terms of party switching compared to older people. Whether this is a finding connected to generational or age effects, we cannot conclusively determine here, but we can rule out interpretations based on young people changing issues of concern more frequently than older cohorts in light of pressing societal challenges.
Our study has a few limitations: first, while our study takes a big step by using actual panel data, the time span covered is short and consists of two waves only. Second, we do not have actual voting data at hand. Such cross-national panel data – if it were available in the future – would make the study of actual voting behaviour possible, going beyond party attachment, voting recalls and intentions. Third, as some studies have demonstrated, party identification shapes policy preferences (Franklin Reference Franklin1984). Accordingly, a change in party priorities could make people more aware about certain (materialistic or postmaterialistic) issues, and thus reverse causality is a possibility we cannot conclusively rule out. Moreover, we cannot determine why the change of issue salience occurs. Certainly, however, issue salience emerges and is transmitted through the media. The existing literature (e.g. Statham and Trenz Reference Statham and Trenz2013) gives us reason to believe that combining such a panel study with the analysis of issue salience in the public sphere would further increase the complexity and explanatory power of the mechanisms investigated here.
We leave it to future studies, moreover, to investigate how diverging news-consumption patterns between the young and the old, the well-off and the precarious, feed into patterns of issue salience and how this could explain polarization and volatility in contemporary societies. While we may assume based on socialization theory that young people will develop more stable voting preferences as they grow older, it is also possible that we will see further increasing volatility in the future based on the pluralization and diversification of modern societies.
What our findings suggest, therefore, is that parties fare better if they stick to ‘their’ issues and benefit from increasing salience of those issues they own. This corroborates the findings in the literature, according to which it is not so much the voters that are responsible for the instability of contemporary party systems, but the fact that parties ignore visible and important cleavages that are still in place. Indeed, for some time political parties have been found to be more and more oriented towards the mean voter, leading to further erosion of their core electorates (Dassonneville Reference Dassonneville2018). While the pluralization of lifestyles may have already made the electorate more volatile, ignoring this pluralization along with the reinforcement as well as diversification of existing cleavages – visible in changing issue salience – is likely to further undermine party attachment. Having said this, parties cannot put aside the pressing issues of the day, such as skyrocketing inflation or record temperatures and fires associated with global warming, which invite party change.Footnote 12 Moreover, young people should not be regarded as ideologically arbitrary and as changing their main issues of concern much more frequently. Rather, young people follow existing divides, are as moved by increased issue salience as older cohorts, and tend to follow and stick to those parties they believe are fit to deliver on these issues. This also opens the potential future of more stable party attachments – if parties choose to focus on those issues that their electorates find most important.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/gov.2022.49.
Acknowledgements
Earlier versions of this article have been presented at the EURYKA Final Conference (Brussels, 2020) and the 27th and 28th International Conference of Europeanists (online, 2021; Lisbon, 2022). We are grateful to participants in these events, and especially Björn Bremer, James Dennison and Piotr Zagorski for helpful feedback. We also thank the editors of Government and Opposition and three anonymous reviewers for their comments and suggestions. Finally, we are also grateful to Lorenzo Bosi, Marco Giugni, Maria Grasso and Christian Lahusen for their help and support throughout.
Financial support
This work has received funding from the European Union's Horizon 2020 research and innovation programme under Grant Agreement No. 727025. Martín Portos also acknowledges support from the Conex-Plus Marie Curie framework (G.A. No. 801538).
Open access
