Abstract
Despite the growing literature on polarization, students of comparative politics have not yet been able to reach much assured understanding of how party polarization influences voter turnout in multiparty settings, which often put on offer both centrist, and divergent mainstream and niche party policies. I evaluate how politically sophisticated and unsophisticated citizens with different ideological preferences respond to high and increasing party polarization by employing individual- and party system-level data from 17 European multiparty democracies. I hypothesize that high levels of actual and perceived party polarization increase voter turnout, and policy seeking, sophisticated citizens are more likely to turn out when polarization in party policy offerings in the short run increases their utility from voting. The empirical analyses show that high party polarization increases both politically sophisticated and unsophisticated citizens’ propensities to turn out. However, such positive effect for the most part comes from the between- and within-party systems differences in actual party polarization, rather than how individual citizens perceive that. The implications of these findings with respect to strategic position taking incentives of political parties and the effects of the knowledge gap between sophisticated and unsophisticated citizens on political participation and democratic representation are discussed in the concluding section.
Similar content being viewed by others
Avoid common mistakes on your manuscript.
Introduction
Students of political science have recently started to acknowledge party polarization as a defining feature of party systems (Budge and McDonald 2006; Ezrow 2008; Dalton 2008; Dow 2011). Most comparative research to date focuses on the impacts of electoral institutions on party polarization and the differences in the extent of polarization across party systems. Although it has long been suggested that diversity and distinctiveness of party policy offerings have important consequences in terms of political participation and voting behavior (Committee on Political Parties 1950), scholarly interest in such behavioral consequences of political polarization has been scarce.
The literature on the effect of political polarization on voter turnout is limited to a handful of aggregate level comparative studies (Crepaz 1990; Siaroff and Merer 2002; Dalton 2008; Steiner and Martin 2012), and few others employing individual-level data (Abramowitz and Stone 2006; Hetherington 2008, 2009; Abramowitz 2010; Fiorina et al. 2011; Rogowski 2014). All individual-level studies to date indeed focus on the United States. This study departs from previous literature by employing individual-level data to assess the effect of party polarization on voter turnout in European democracies with proportional rules. European multiparty systems provide political parties with ‘centrifugal incentives’ (Cox 1990) and put on offer not only more numerous, but also more distinct party policies.
Individual-level studies investigating the effect of polarization on voter turnout in the U.S. have contradictory theoretical expectations and empirical findings. While some find that constantly increasing political polarization in recent decades has increased citizen participation in politics, others conclude that increasing polarization disengages citizens. The former group suggests that polarization increases voter turnout by providing citizens with clear partisan cues and policies that better represent their views. The main argument of the latter group of studies is that citizens with moderate policy or ideological demands and with no tolerance for conflict are discouraged by the increasing conflict among political elites (Fiorina et al. 2011; Rogowski 2014). Although either view may well be the case, both propositions are indeed conditional on a number of party system- and individual-level characteristics: the number and the distinctiveness of party policy offerings in a party system, and individuals’ expectations from, knowledge about, and perceptions of the distinctiveness of party policies.
This article is the first study investigating the effects of actual and perceived party polarization on voter turnout in European democracies. It seeks to answer whether constantly high and increasing party polarization in European multiparty systems, and citizens’ perceptions of such polarization affect their decisions to turn out. Because citizens’ policy demands as well as perceptions of and responses to party polarization are hardly uniform, another question I ask is whether sophisticated and unsophisticated citizens have heterogeneous responses to party polarization.
This article differs from previous studies on voter turnout and political polarization in a number of regards: First, by focusing on other established democracies than the U.S., it aims to examine the effect of party polarization on voter turnout from a comparative perspective. Second, by employing longitudinal data on party polarization, it aims to assess whether ‘increasing’ party polarization over time mobilizes citizens. Third, while accounting for the actual party polarization at the aggregate level, this article also investigates whether individuals’ perceptions of polarization in party policy offerings affect their propensities to turn out. Lastly, by modeling the effect of perceived polarization on voter turnout as conditional on political sophistication, it explores whether citizens with varying levels of engagement in politics and with varying policy demands are heterogeneously influenced by party polarization.
I employ an interactive logistic regression applied to party system- and individual-level data on 23 elections in 17 European multiparty democracies between 2006 and 2011. This article concludes that while both sophisticated and unsophisticated citizens are more likely to turn out in response to constantly high party polarization as a feature pertinent to some party systems that provide their electorates with both centrist, niche, and divergent party policies, increasing party polarization in the short run does not uniformly influence the turnout decisions of politically sophisticates and unsophisticates with differing knowledge about and engagement in politics, perceptions of party polarization, and policy demands.
Following a brief review of previous literature in the next section, the third section presents the theoretical framework and hypotheses, and the fourth section covers the research design. I conclude by discussing the normative and policy implications of the empirical findings presented in the fifth section with respect to the strategic position taking of political parties, and the knowledge gap between politically sophisticated and unsophisticated citizens.
Previous Research on Polarization and Turnout
The Downsian spatial theory suggests that citizens’ utility from voting decreases proportionally to the increasing distance (i.e., dissimilarity) between their preferred policies and parties’ policy offerings (Downs 1957). Along with the assumption of single-peaked preferences, it predicts that vote support of a party or candidate in a two-party system reaches its peak at the median of voter distribution along a uni-dimensional ideological (or policy) continuum. Rabinowitz and Macdonald’s (1989) directional theory, on the other hand, divides each ideological (or policy) scale into its two components—i.e., direction and intensity. It then predicts that a party maximizes its vote support at around the extremes (i.e., region of acceptability). Despite those different expectations, both spatial theories of voting predict that citizens with non-moderate ideological preferences have higher utilities from voting for parties with more intense or polarized ideological stands.
Further work on the spatial theory suggests that citizens get alienated from politics and abstain when no candidate stands sufficiently close to their preferred policies (Adams and Merrill 2003; Adams et al. 2006b; Callander and Wilson 2007). Despite the common conclusion in spatial studies that parties and candidates are better off by converging to the center of a two-party system, the literature focusing on the elite polarization in the U.S. shows that political elites have become increasingly polarized in their legislative behavior and ideological stands in recent decades (Poole and Rosenthal 1997; McCarty et al. 2006).
Previous studies investigating whether such divergence in party and candidate policies, at least in particular issue domains, affects political participation show contradictory theoretical expectations and empirical findings. On the one hand, policy shifts of parties away from the center decrease the alienation of citizens with distinct policy demands. Nevertheless, divergence in party policies alienates others with no such demands. Moreover, party policy shifts towards the center make citizens with moderate views indifferent to similar party policy offerings and depress their turnout (Plane and Gershtenson 2004; Adams et al. 2006b; Callander and Wilson 2007; Fiorina et al. 2011; Rogowski 2014). On the other hand, while increasing polarization provides citizens with clearer cues about parties’ ideological and policy stands (Knutsen and Kumlin 2005; Lachat 2008; Garner and Palmer 2011; Thornton 2013) as well as a higher “importance of policy stakes” (Franklin 2004, p. 49), polarizing rhetoric of elites may still disengage those who are intolerant of ideological conflict (Ulbig and Funk 1999; Fiorina et al. 2011; Rogowski 2014) or those whose views are not represented by government policies that diverge from the preferences of the median voter (Ezrow and Xezonakis 2011, 2016).
A closer look to previous works providing empirical evidence for the ‘mobilization hypothesis’ (Rogowski 2014) that elite polarization increases citizen participation shows that all introduce some measure of ‘perceived’ (dis)similarity between candidates (Abramowitz and Stone 2006; Abramowitz and Saunders 2008; Abramowitz 2010; Hetherington 2008, 2009). While some take into account feeling thermometers for candidates (Abramowitz and Stone 2006; Abramowitz and Saunders 2008), others employ perceived distances between candidates on the liberal-conservative or distinct policy scales (Hetherington 2008). Studies finding a negative effect of elite polarization on voter turnout, on the other hand, introduce more ‘objective’ measures of polarization, which take into account ideological differences between candidates at party system, state or district levels (Rogowski 2014).
Regardless of how they measure polarization, all individual-level studies to date focus on the U.S. as perhaps the most prominent and long studied example of high polarization (Poole and Rosenthal 1997; Miller and Schofield 2003; McCarty et al. 2006). However, although some scholars highlight the increasing trend in polarization over time, they only investigate its behavioral consequences by employing separate samples for consecutive elections (Abramowitz and Stone 2006; Abramowitz and Saunders 2008) or pooled samples without accounting for the temporal change in polarization (Rogowski 2014). Even though the theoretical expectation is often about the effect of a positive trend in polarization on political participation, no individual-level study to date investigates whether ‘increasing’ polarization over time affects voter turnout in one direction or another.
Another point worthy of note is about the generalizability of the empirical findings in the previous studies on behavioral consequences of polarization to other party systems. To the extent that either one or both parties move away from the center, polarization in the U.S. without a centrist third party alternative may alienate citizens at around the center of the ideological space, and therefore lead to a decrease in turnout due to the alienation of those with moderate ideological views.
Previous literature shows that the negative effect due to the alienation of citizens with moderate views is substantially higher than: the negative effect of decreasing polarization on voter turnout—due to the indifference of citizens with moderate views to increasingly similar party policy offerings—and its positive effect—due to the decreasing alienation of policy-seeking citizens (Plane and Gershtenson 2004; Adams et al. 2006a). That is, the American political parties face a tradeoff between alienating policy-seeking citizens who stand away from and those with moderate views stading at around the center. In fact, such tradeoff is conditional on both the peculiarities of the two-party system in the U.S., and citizens’ policy demands, engagement in politics, and perceptions of polarization.
The major differences between the multiparty systems in Europe that this article focuses on and the bipartisan system in the U.S. can be summarized as follows: First, all party systems examined here have proportional electoral rules, which provide parties with ‘centrifugal incentives’ to take ideological and policy stands away from the centers of such party systems (Cox 1990). More permissive electoral rules also allow parties to gain considerable electoral support by representing distinct cleavage groups. Second, all European party systems examined in this article put on offer at least one party policy close to the party system center along with others closer to the end points of the left-right continuum, though their electoral viabilities vary. Third, the literature has long suggested that the European voter has somewhat distinct policy expectations than ‘the American voter’ (Campbell et al. 1960; Lewis-Beck et al. 2008), which have been increasingly represented by niche parties (Inglehart 1977, 2008).
In the first comparative study on the effect of polarization on political participation, Crepaz (1990) examines whether the dispersion in party policies affects overall turnout in advanced democracies, including but not limited to European multiparty democracies. The author suggests that the presence of post-materialist (i.e., niche) parties in a party system enhances the representation of distinct policy demands of those who are otherwise underrepresented. In a similar vein, Siaroff and Merer (2002) find that a distance more than 3.75 points between two relevant most extreme parties on the traditional left-right continuum increases the turnout in European democracies by about 2.6% points.
Dalton (2008) also argues that citizens in polarized party systems are more likely to find party policies that represent their views, and therefore high party polarization would decrease abstention due to the alienation of citizens with distinct ideological views. Dalton, however, finds no significant relationship between turnout and party polarization unlike other studies.
In a more recent study Steiner and Martin (2012) investigate the effect of increasing similarity in macroeconomic policy offerings of political parties on voter turnout. The authors show that increasing dispersion in macroeconomic policy offerings positively affects overall turnout. Similarly to previous comparative studies though, this study does not take into account how party polarization influences turnout decisions of individual citizens with or without distinct policy demands and with differing perceptions of party polarization.
Theoretical Framework
The number of party policy alternatives put on offer has long been considered an important determinant of voter turnout (Downs 1957; Powell 1982; Norris 2004) and a remedy to ‘abstention due to alienation’ (Adams and Merrill 2003; Adams et al. 2006b). Given the common distributional assumption in spatial studies and the empirical reality that the center of a policy space is more populated than its end points, political parties facing the tradeoff between alienating sophisticated, policy-seeking citizens and those with moderate views are incentivized to strategically converge toward the center when electoral rules restrict third party entry. Even in the U.S. with only two viable party policies, therefore where policy shifts of parties are likely to have relatively more significant impacts on voter turnout, parties distinguish themselves from their opponents and provide citizens with increasingly polarized policies (Poole and Rosenthal 1997; McCarty et al. 2006).
Scholars have long been interested in explaining such suboptimal outcome according to the spatial theory. Previous works on spatial competition in the U.S. suggest that potential costs and benefits of policy shifts, and relatedly the extent of polarization in party policy offerings differ from one election to another due to diverse goals of parties such as mobilizing different cleavage lines (Schattschneider 1960), and appealing to party activists or non-activist citizens (Aldrich 1983; Aldrich and McGinnis 1989; Miller and Schofield 2003; Schofield and Miller 2007). In these respects, polarization might be considered a strategic decision of party elites with election-specific needs or longer-term goals. Party elites in need of resources that party activists provide may shift their policies to appeal to policy-seeking activists at the expense of alienating non-activists (Aldrich 1983; Aldrich and McGinnis 1989; Miller and Schofield 2003; Schofield and Miller 2007). Party elites may also craft policies in particular domains that can mobilize the electorate in the longer run (Schattschneider 1960; Riker 1986).
Potential costs and benefits of policy shifts away from the center not only differ from one election to another, but also across party systems due to differences in electoral rules, the distribution of citizens with distinct policy demands, as well as the number and distinctiveness of party policy offerings. Focusing on the multiparty democracies in Europe and modeling the effect of perceived party polarization on voter turnout as conditional on distinct policy demands of sophisticated and unsophisticated citizens, this article suggests that the relationship between party policy offerings in a party system and voter turnout is mainly driven by the quality, rather than the quantity, of party policy offerings.
This is not a novel argument. Scholars highlighting the normatively desirable function of distinct party policies for representative democracy and responsible party governments have long suggested that “whether a real choice is presented” (Committee on Political Parties 1950, p. 90) matters more to citizens than the mere count of parties (Dalton 2008; Steiner and Martin 2012; Thornton 2013). Nevertheless, to the best of my knowledge, no study to date has investigated whether constantly high and increasing party polarization in multiparty democracies influence individual citizens’ decision to turn out.
The number of party policy offerings and the extent of polarization in them are not necessarily interrelated. A number of studies examining the effect of electoral rules on the diversity in party policies show that the latter is not a function of the simple or effective count of political parties in a party system (Dalton 2008; Ezrow 2008; Dow 2011). There is substantial variation in how the diversity of party policy offerings varies across party systems that provide their electorates with similarly high numbers of party policy alternatives (McDonald and Moral 2015).
Both the number and distinctiveness of party policies in the European democracies examined here are higher than those in the U.S. This article suggests that a high level of party polarization in the proportional party systems of Europe does not alienate citizens with moderate views due to the existence of party policy alternatives that are sufficiently close to their preferred ideological stands. Hence, my expectation with respect to the effect of party polarization on voter turnout in European party systems is different from that in the previous literature focusing on the U.S. with only two viable party policies (Fiorina et al. 2011; Rogowski 2014).
High polarization in party policy offerings in some multiparty systems that provides citizens with more numerous of and more distinct party policy alternatives, including both centrist, and divergent mainstream and niche parties, is expected to lead to an increase in voter turnout. While politically sophisticated citizens who are more engaged in politics and who have distinct policy demands are mobilized by high levels of party polarization that answer their policy demands, unsophisticated citizens are not alienated as long as they are provided with centrist party policy alternatives.
The relationship between polarization and citizen participation in politics is not only about the extent of party polarization in a party system, but also about how individuals perceive that, and how such perception is translated into their utilities from voting. At the end, “to the electorate, their perceptions are reality” (Dalton 2008, p. 909). Some previous studies suggest that engaged, sophisticated citizens and political unsophisticates have differing perceptions about the polarization in a party system (Abramowitz and Saunders 2008; Rogowski 2014). A common theoretical explanation in previous literature, in reference to the vast literature on political knowledge and sophistication, is that political sophistication conditions citizens’ polarization perceptions (Lachat 2008; Rogowski 2014) and their responses to institutional structures (Pardos-Prado et al. 2014).
Such conditional effect is of theoretical importance because some previous studies suggest that polarization disengages unsophisticated citizens (Ezrow and Xezonakis 2011; Fiorina et al. 2011; Rogowski 2014), while others argue that it might encourage political sophisticates to turn out as a result of divergent party policy offerings answering their policy demands (Crepaz 1990; Franklin 2004; Dalton 2008). Indeed, some individual-level studies provide empirical evidence for such conditional effect. Abramowitz and Saunders (2008) show that well educated citizens have greater ideological polarization than the less educated, while Rogowski (2014) concludes that elite polarization heterogeneously affects those with low and high levels of education and political information.
I divide individuals’ perception of party polarization into its two components: The first one is the actual party polarization in a party system. To the extent that political parties offer sufficiently distinct policies to citizens, voter turnout in party systems with high levels of party polarization is expected to be higher. The second component is about the individual variation in perceived party polarization as a result of the heterogeneity in citizens’ demands from politics as well as their engagement in and knowledge about politics. To the extent that an individual perceives polarization in party policy offerings as higher than the actual level, her propensity to turn out is expected to be higher. The first part of my argument takes form in the following hypotheses:
H1a
High actual party polarization increases voter turnout in multiparty systems.
H1b
High perceived party polarization increases voter turnout in multiparty systems.
In addition to between-party system variation in actual party polarization and how citizens perceive it, this study aims to assess the extent to which individuals’ turnout decisions are affected by the change in actual party polarization within particular party systems. Taking the potential effects of party polarization on electoral polarization into account, and relaxing the assumptions that individual preferences are normally distributed along the traditional left-right continuum and that such distribution is constant over time lead to different theoretical expectations with respect to the effect of the change in party polarization on the propensity to turn out.
As explained in the previous section in detail, the theoretical expectations and the empirical evidence in the previous literature regarding the effect of the change in party polarization on voter turnout are inconsistent. On the one hand, being offered party policies that are closer to their preferred policies, sophisticated, policy-seeking citizens with higher engagement in politics are more likely to turn out when party policies become more polarized, particularly in policy domains they find salient. Hence, sophisticated citizens are expected to have a higher utility from voting according to the spatial theories of voting (Downs 1957; Rabinowitz and Macdonald 1989) when actual party polarization increases over time. Moreover, strategic position taking of political parties provides citizens with clearer heuristic and partisan cues (Knutsen and Kumlin 2005; Lachat 2008; Garner and Palmer 2011; Thornton 2013). To the extent that political parties take increasingly polarized stands and it echoes through information channels, increasing party polarization decreases the complexity of politics, especially to politically unsophisticates. In addition, increasing party polarization may lead to polarization of citizens’ policy preferences (i.e., electoral polarization) and strengthen their partisan attachments over time (Hetherington 2001.
On the other hand, increasing party polarization, which is expected to mobilize sophisticated citizens, may not simultaneously influence unsophisticates’ turnout decisions. Some unsophisticated citizens may start to behave like the sophisticated to the extent that their policy preferences and partisan attachments become more intense and increasing party polarization provides them with clearer cues about party policies. Some others with no tolerance for ideological conflict (Ulbig and Funk 1999; Fiorina et al. 2011; Rogowski 2014) or whose policy demands are not represented by polarized government policies (Ezrow and Xezonakis 2011, 2016) may however get alienated from politics. While the former is expectedly less likely in the multiparty democracies this article focuses on, the extant literature on the behavioral consequences of polarization does not provide us with conclusive evidence regarding the short and long-term attitudinal consequences of party and electoral polarization.
To summarize, we have different expectations from the constantly high party polarization as a feature pertinent to some multiparty systems and from the increasing polarization over time. Although one might suggest that increasing polarization should decrease the knowledge gap between politically sophisticated and unsophisticated segments of the society by reducing the costs of information and voting through clearer heuristic cues, by increasing the strength of their partisan attachments, and by changing the saliency of particular issue domains, these effects are likely to operate in the long run due to constantly high or constantly increasing polarization. Due the lack of cross-national data on the change in actual party polarization extending to more than two consecutive elections, and, perhaps more importantly, the lack of panel data on citizens’ changing perceptions of polarization, policy preferences, and partisan attachments over time, this study is limited to the increasing spatial utilities of sophisticated citizens from increasingly polarized party policies between consecutive elections. Hence, I hypothesize:
H2
Increasing party polarization in multiparty systems between two consecutive elections increases the propensity of politically sophisticated citizens to turn out.
Data and Research Design
Along with measurement related concerns and strong assumptions about citizens’ competence and polarization perceptions, previous comparative studies on the effect of party polarization on voter turnout draw inferences about individual behavior using aggregate-level data. Differently from previous literature, this article employs individual-level survey data. Among few cross-national studies asking respondents to place political parties on the traditional left-right continuum, the comparative study of electoral systems (CSES) data are perhaps the most comprehensive. CSES data cover a large number of European party systems and provides several party system, district, and individual level variables, which are introduced in the empirical analyses as explanatory and control variables.
The two criterion of case selection are data availabilityFootnote 1 and electoral rule.Footnote 2 The data employed in Models 1 and 2 cover 23 post-election surveys conducted in 17 European parliamentary democracies.Footnote 3 The party systems without available data for previous parliamentary elections in the second and third modules of the CSES (2007, 2013) were dropped from the sample in Model 3, which aims to assess the effect of the change in actual party polarization on voter turnout.Footnote 4
The dependent variable in all empirical analyses, voter turnout, is binary and scores 1 if a respondent reported to have voted.Footnote 5 In coding the main party-system level independent variable (i.e., actual party polarization) the following formula is used to calculate the ideological distance (displacement) of left- and right-wing parties from the center of a party system (McDonald and Moral 2015):
where \(\overline{P_{R}}\) is the mean perceived ideological stand of the parties to the right, and \(\overline{P_{L}}\) is the mean ideological stand of the parties to the left of the center in the election for which the CSES data were collected.Footnote 6
While the perceived party polarization measure in Model 1 is based on individual respondents’ placement of political parties, ‘perceived party polarization (difference)’ measure in Models 2 and 3 subtracts the actual party polarization in a party system from the perceived party polarization by each respondent and divides it by the standard deviation of this measure in a given post-election survey for standardization across party systems and elections. This operationalization rests on the assumption that perceived polarization is a function of the actual polarization in a party system (Hetherington 2001).
As indicated above, the lack of panel data from European multiparty democracies does not allow us to account for changes in respondents’ partisan attachments, political information, and perceptions of polarization. Since the samples for post-election surveys in the CSES data are randomly drawn, however, we can compute the change in actual polarization in the policy offerings of the same parties between two consecutive elections.Footnote 7 Our third measure of party polarization, entitled “Change in Actual Party Polarization,” is equal to the difference in the actual party polarization in a given party system between two consecutive elections. With the exception of this variable coded from Module 2 (2007) for some party systems (i.e., Denmark, Ireland, Poland, Portugal, Sweden and Switzerland), all variables are coded from Module 3 of the CSES data (2013).Footnote 8
Although empirical findings presented below are robust to introducing other commonly employed measures of party polarization in the literature,Footnote 9 the measure above is preferred for the following reasons: First, it provides an opportunity to compare party systems under examination regardless of the variation in number, distinctiveness, and dispersion of party policies put on offer to citizens. Second, unlike some previous studies (Crepaz 1990; Siaroff and Merer 2002), I assume that party polarization is not a function of the distance between the rightmost and leftmost parties, but the displacement of all left- and right-wing parties from the center of a party system. In this regard, the measure employed in the empirical analyses in this article can also be considered a dispersion-based measure unlike the solely distance-based measures in some previous studies.Footnote 10 Moreover, because most extreme parties are often of niche character, thus less likely to be considered viable and correctly placed by a majority of respondents, especially by those with lower levels of information about politics, employing a measure based on the distance between the two most extreme party policies decreases the sample size to a considerable extent. Third, although some other measures weight the policy stands of parties by their respective vote shares, in addition to the endogeneity problem it would create in our analyses taking voter turnout as the dependent variable, most citizens are not able to make such calculation. Lastly, although some of those measures also divide overall party polarization by standard deviations of citizen or party stands, this information is unavailable to citizens (Knutsen and Kumlin 2005). Perhaps more importantly for our purposes, we cannot account for how party polarization is perceived by individual respondents if we employ a measure using a variance-like formula.
Despite the relatively less demanding measure employed in this article, we still had to drop some respondents from the sample since they are not sufficiently knowledgeable about party policy offerings. While some place left-wing parties to the right of right-wing parties, some others are unable to locate most parties on the traditional left-right continuum. In these regards, the sample is limited to those who could correctly place left-wing parties to left of right-wing parties and who could locate more than two of the largest four parties in a party system.Footnote 11
In addition to election- and party system-level differences that are accounted for by introducing aggregate-level controls and election fixed effects, another source of variation in voter turnout is related to heterogeneous citizen characteristics (see, for instance: Abramowitz and Saunders 2008; Pardos-Prado et al. 2014; Rogowski 2014). The CSES data provide three factual questions for each post-election survey. I penalized wrong answers, ‘Don’t Know’s and ‘No Response’s to create an additive index. Because the difficulty of political information questions vary across surveys,Footnote 12 this index is then standardized. The standardized variable, with a mean of 0 and a standard deviation of 1, is intended to capture the level of political information a respondent possesses, relative to the mean level of political information in each survey.
In line with our theoretical expectations and previous studies (Abramowitz and Saunders 2008; Rogowski 2014), the model equations include an interaction term between respondents’ perception of party polarization and their political information.Footnote 13 The interaction term allows us not only to assess the effect of perceived party polarization on voter turnout conditional on political information, but also to account for any systematic pattern that may bias our perceived polarization measures.Footnote 14 , Footnote 15
Lastly, the empirical analyses in the next section also control for a number of aggregate- and individual-level determinants of voter turnout. Most comparative studies on voter turnout find that the effective number of electoral or legislative parties increases voter turnout [For excellent surveys of the literature on voter turnout, see Geys (2006) and Blais (2006)]. Even though I do not find appropriate its use to approximate the diversity in party policy offerings, it is nevertheless a good proxy for the competitiveness of elections, nationalization of political parties, and distribution of different groups of voters across electoral districts. To this end, effective number of electoral parties at the district level is introduced in all models to account for any district-level variation in voter turnout. Another control variable that is worthy of note is ideological extremity. This variable is equal to the distance of each respondent from the respective party system center and is intended to account for the intensity of respondents’ ideological views, which is expected to increase their utility from voting for divergent party policies according to the spatial theories of voting. In addition, I also control for the robust socio-demographic determinants of voter turnout in previous literature such as income, education, age, gender, union membership, and strong partisan attachment. Lastly, all models include election fixed effects to account for unobserved election-specific factors. The operationalization of the dependent and independent variables is explained in Table 2 in Appendix and descriptive statistics are presented in Table 3 in Appendix.
Empirical Findings and Discussion
The logistic regression estimates on the full sample and on the subsample composed of party systems with available longitudinal data on the change in actual party polarization between two consecutive elections are reported in Table 1.Footnote 16 The extent to which the proportional parliamentary systems in Europe examined here put on offer divergent party policy alternatives to their electorates varies to a considerable extent. This variation in actual party polarization may manifest itself in how citizens in those countries perceive party policies as distinct from each other. Because taking perceived polarization as the only explanatory variable can thus lead to Type I and II errors in a cross-national sample with considerable variation in actual party polarization across space and over time, I comment below on the estimates from Models 2 and 3 that take the difference between perceived and actual party polarization in a given party system as the primary explanatory variable.Footnote 17
The marginal effect of perceived party polarization (hereinafter perceived polarization) on turnout is statistically significant for the range of the political information variable in our sample and it is, ceteris paribus, about 1.1% points.Footnote 18 It is also conditional on and decreases by increasing political information using which we approximate political sophistication.
This marginal effect translates into an estimated difference of 4.4% points in the turnout propensities of two hypothetical respondents with average political information, one perceiving party polarization as two standard deviations lower and another perceiving it as two standard deviations higher than the actual party polarization. Exploiting the inherent interactivity in the logistic regression, the difference in predicted propensities of these hypothetical respondents for varying levels of party polarization can also be computed. The estimated differences for the 10th (2.9), 50th (3.3), and 90th percentiles (4.5) of actual party polarization (hereinafter actual polarization) are in turn 5.5, 4.8, and 3.3% points. These decreasing predicted probabilities show that the actual polarization in a party system conditions the effect of respondents’ polarization perceptions on their decisions to turn out.
Since the marginal effect of a constitutive term in interactive models is conditional on the other constitutive term(s), coefficient estimates, especially those of continuous variables, cannot be straightforwardly interpreted (Brambor et al. 2006). In Fig. 1, the predicted probabilities for the 10th, 50th, and 90th percentiles of the actual polarization variable are plotted for three hypothetical respondents with varying levels of political information (10th (−1.2), 50th (.2), and 90th (1.3) percentiles) within the range of the perceived polarization (difference) variable.Footnote 19
Comparing three lines representing those hypothetical respondents with varying levels of political information from the leftmost to the rightmost plot in Fig. 1 shows the substantive (conditional) effect of actual polarization in a party system, while comparing the lines representing politically informed and uninformed respondents within each plot shows the substantive effect of perceived polarization on turnout conditional on political information.
The substantive effects of perceived and actual party polarization on voter turnout \(\mid\) political information (Model 2)
When predicted probabilities in the leftmost and the rightmost plots in Fig. 1 are compared, except politically uninformed respondents who perceive party polarization as 2.5 standard deviations lower or 2 standard deviations higher, and politically informed respondents who perceive it as 1.75 standard deviations lower or 1.25 standard deviations higher than the actual polarization, the confidence intervals around the predicted probabilities at 95% level in the leftmost and rightmost plots are distinguishable from each other. That is, when we take into account both the multiplicative interaction and the compression effects, the effect of perceived polarization (difference) on voter turnout (conditional on political information and actual polarization) is statistically significant for 86% of all respondents in our sample, except those with very high or very low polarization perceptions.
When predicted probabilities of turnout for politically uninformed and informed respondents within each plot are compared, predicted probabilities are distinguishable from each other in all plots. The predicted probability of turnout for a politically uninformed respondent perceiving party polarization two standard deviations lower than the actual polarization in the party systems respectively with low and high actual polarization is about 9.3% points lower in the former. This difference is about 5% points for a politically informed respondent. When perceived polarization is set to two standard deviations higher than the actual polarization in a party system, the difference becomes 6.3% points for a political unsophisticate and 2.9% points for a sophisticated respondent. Hence, high party polarization influences the turnout decisions of politically sophisticated citizens and, to a smaller extent, those who already perceive polarization as higher than the actual polarization.
Predicted probabilities of turnout of politically uninformed and informed respondents increase in turn by 16.2 and 9.5% points within the range of the perceived polarization (−3.25 to 4.25) when actual polarization is low (2.9). When the actual polarization in a party system is set to a high level (4.5), the differences become 10.9 and 5.8. Such increases in predicted probabilities show that both the actual polarization in a party system and how citizens perceive that substantially affect the propensity to turn out. In addition, the major source of variation in voter turnout due to party polarization comes from the between-party system differences in actual polarization, rather than how individuals perceive that.
To summarize our main findings, first it should be reiterated that the marginal effect of perceived polarization on voter turnout is statistically significant for the whole range of the political information variable. In terms of the effect of party polarization, it should also be noted that constantly high party polarization significantly increases voter turnout in European multiparty democracies. Hence, we fail to reject hypotheses 1a and 1b given the statistical and substantive significance of the perceived and actual polarization variables in Model 2.
In order to conclude that ‘increasing’ party polarization in the short run increases voter turnout, however, we need to introduce another variable in the model to account for the change in actual party polarization (hereinafter change in actual polarization). Model 3 investigates whether increasing actual polarization between two consecutive elections in 12 European party systems affects politically sophisticated and unsophisticated citizens’ turnout decisions.
To start with the marginal effect of perceived polarization (difference) on voter turnout conditional on political information in Model 3, we find that it is not statistically significant for respondents with a political information level 1.07 standard deviations lower than the mean political information. This range corresponds to the least sophisticated group of respondents in each party system, constituting about 10.2% of the sample.
When we account for the change in actual polarization, we also find that the substantive effect (i.e., first difference) of perceived polarization (difference) on the probability of turnout decreases to about 0.7% points. When the change in actual polarization varies, the effect of perceived polarization increases to 1.4% points in a party system with a substantial decrease in actual party polarization between two consecutive elections (−.61 as in the case of Portugal in 2009), 0.7% points for a relatively low decrease (−.04 equal to the median change corresponding to that in Norway in 2009), and 0.4% points for a substantial increase in actual party polarization (0.49 as in the case of Germany in 2009). Hence, the substantive effect of perceived polarization on voter turnout decreases conditionally on the increasing actual polarization in the short run.
The substantive effect of the change in actual polarization on the turnout decisions of politically uninformed and informed respondents is plotted in Fig. 2. To start with the differences in the propensity to turn out in party systems with low and high change in party polarization in the short run, the predicted probabilities for both politically informed and uninformed citizens are distinguishable from each other in the left- and right-most plots. Hence, the effect of the change in actual polarization on voter turnout in the short run is substantively significant for the whole range of perceived polarization (difference) and for both politically sophisticated and unsophisticated citizens. In other words, increasing actual polarization in the short run mobilizes both politically sophisticated and unsophisticated citizens regardless of how they perceive polarization.
The substantive effects of perceived and \(\Delta\) actual party polarization on voter turnout \(\mid\) political information (Model 3)
When we take a closer look at the predicted probabilities for politically sophisticated and unsophisticated citizens, the confidence intervals around predicted probabilities are only distinguishable from each other within our hypothetical party systems with a decrease in party polarization except very low levels of perceived polarization (less than 3 standard deviations). When the change in actual polarization is very high, however, politically sophisticated and unsophisticated citizens do not differ in their propensities to turn out. Put differently, when we model the change in actual polarization in the short run, except those who perceive polarization as quite lower than the actual polarization and except when actual polarization increases to a large extent, political sophistication conditions the short term effect of the change in actual polarization.
The predicted probabilities of turnout for a politically uninformed respondent perceiving party polarization two standard deviations lower than the actual party polarization in two party systems respectively with low and high change in actual polarization is about 22.2% points lower in the former. Such difference is about 12.4% points for a politically informed respondent. When the respondents perceive party polarization as two standard deviations higher than the actual polarization, the differences are 18.3% points for a politically unsophisticated and 7.5 for a sophisticated respondent. Hence, the substantive effect of the change in actual polarization on the propensities of turnout of those who already have a high polarization perception and who are politically sophisticated is lower.
To summarize, Model 3 that models the change in actual polarization in the short run shows that increasing party polarization between two consecutive parliamentary elections substantively increase the turnout propensities of both politically sophisticated and unsophisticated citizens. Given the substantial increase in sophisticated citizens’ propensity to turn out in response to increasing party polarization, we fail to reject the second hypothesis.
In light of these substantial changes in predicted probabilities of turnout, we should revisit our conclusion and add that the short-term change in actual polarization plays a crucial role in explaining voter turnout in European multiparty democracies. When all three components of party polarization, namely the actual party polarization in a party system, how it is perceived by respondents, and how it changes over time, are modeled, all three components influence voter turnout in multiparty settings, while the change in actual polarization has the largest substantive effect on one’s propensity to turn out.
This finding is noteworthy in light of the previous studies on the effect of polarization on voter turnout that employ either individuals’ polarization perceptions, or mean party or candidate placements as more objective measures. When both are accounted for in the same model, the major source of variation is the actual party polarization in a party system. Furthermore, not only systemic differences in party polarization help to explain why voter turnout is higher in some party systems, it is its increase or decrease that explains the significant changes in turnout within particular party systems between consecutive elections.
Party system- and individual-level controls also perform as expected in all models reported in Table 1. While high levels of income, education, age, ideological extremity and strong partisan attachment have positive effects in line with previous literature, gender and union membership do not have any effect on voter turnout in European democracies.
In light of our expectation that the distinctiveness of party policy offerings matters more to citizens than the mere count of political parties, a coefficient estimate that merits more discussion is that of the effective number of electoral parties at the district level. When the average (marginal) effects of the actual polarization and effective number of electoral parties on voter turnout are compared, the former is substantively more significant than the latter in Models 2 and 3. An increase from the 25th to 75th percentiles of actual party polarization increases the probability of turnout by about 2.9% points in Model 2 and 2.8% points in Model 3. An increase of the same magnitude in the effective number of electoral parties at the district level increases the predicted probabilities by about 1.4 in Model 2 and 1.7% points in Model 3. Consequently, such differences provide empirical evidence for that both the distinctiveness and the number of party policy offerings matter to citizens, while the former is a more important determinant of individual-level variation in voter turnout.
To compare our estimates with those in the previous literature on the effect of elite polarization on voter turnout in the U.S., we revisit two notable studies that respectively employ perceived and actual elite polarization as their main explanatory variables. Abramowitz and Saunders (2008) examine the effect of the feeling intensity toward George W. Bush on voter turnout in the U.S., and find that an increase from the 25th percentile to the 75th percentile in feeling intensity increases the propensity to turn out by 6% points. An increase from the 25th to the 75th percentile of the perceived polarization variable in Models 2 and 3, ceteris paribus, translates into a difference in predicted probabilities of turnout by about 1.7 and 1.3% points. Hence, Models 2 and 3 taking into account other components of party polarization—i.e., actual party polarization and its change between consecutive elections—reveal a substantively less significant effect in the Europan multiparty democracies compared to the Abramowitz and Saunders’ (2008) findings for the U.S.
Rogowski (2014) introduces absolute ideological distances between individual candidates in the House and Senate elections in the U.S. as his main explanatory variable, and finds that an increase in the ideological divergence of candidates from the 25th to the 75th percentile decreases the propensity to turn out by about 5% points for both types of elections. Compared to Rogowski’s findings, we find an inverse relationship between actual party polarization in multiparty settings and voter turnout. Moreover, the substantive significance of our actual party polarization measure is lower than what Rogowski reports. That is, an increase from the 25th to the 75th percentile in the party polarization variable (3.2–4.1), ceteris paribus, translates into an increase of 2.9% points in Model 2, and 2.8% points in Model 3. Although the measures of polarization and model specifications employed in previous studies and in this article are different, we can suggest that party system-level differences between the U.S. and the European multiparty democracies this article focuses on constitute the major reason for such differences.Footnote 20
Conclusion
As a normatively desirable function of representative democracies and responsible party governments, distinctiveness of party policy offerings has important consequences for citizen participation in politics. However, while all individual level studies on the effect of political polarization on voter turnout to date focus on the U.S., comparative research on European multiparty democracies is limited to few studies examining voter turnout at the aggregate level that disregard the variation in citizens’ policy demands, engagement in politics, perceptions of polarization in party policy offerings, and how such variation translates into their individual utilities from turnout.
To the best of my knowledge, this article presents the first individual-level investigation of whether actual and perceived party polarization influence voter turnout in European multiparty democracies with proportional rules. I employ party system and individual level data on 23 elections in 17 European democracies to assess the effects of both the perceived party polarization and the differences in actual party polarization between party systems and between consecutive elections within particular party systems on the turnout decisions of politically sophisticated and unsophisticated citizens.
Putting on offer more numerous and more distinct party policy alternatives to their electorates, the expectations with regard to the effect of party polarization on voter turnout in multiparty European democracies are different from what the literature on the U.S. has long suggested. On the one hand, due to party system-level differences, especially in regard to more proportional electoral rules that lead to not only a higher number of political parties but also centrifugal incentives to political parties, I expect high and increasing party polarization in multiparty European democracies to increase voter turnout. High party polarization is expected to increase political sophisticates’ propensity to turn out by offering policies that respond to their policy demands. In fact, high party polarization in multiparty settings with centrist party policies also works to not alienate political unsophisticates, and allows them to more easily distinguish party policy offerings due to the heuristic and partisan cues it provides. Our theoretical expectation about the effect of the change in party polarization on voter turnout, on the other hand, is in line with the previous literature on spatial theories of voting that increasingly intense and divergent party policies increase the utilities of sophisticated, policy-seeking citizens from voting.
The empirical analyses provide support for our hypotheses. While high levels of perceived and actual party polarization increase all citizens’ propensity to vote in multiparty democracies, the marginal effect of perceived party polarization conditional on political information loses its significance for the least sophisticates when the temporal change in party polarization is also assessed. Political sophisticates who can follow the shifts in party policy offerings and find increasing party polarization answering their policy demands are more likely to turn out in response to increasing party polarization in the short run.
I report a number of robustness and sensitivity checks for both models on all post-election surveys in the full sample and on the subsample composed of party systems for which longitudinal data on the change in actual party polarization between two consecutive elections are available. The conclusions are not sensitive to the changes in model specification, estimator, sample size, and how we measure perceived and actual party polarization and approximate political sophistication.
Despite the robustness of the empirical findings, some important limitations of this study should be noted: First, the lack of cross-national panel data on changing perceptions of party polarization, political information, partisan attachment, and policy demands of citizens does not allow us to account for whether high and increasing party polarization produce some behavioral effects, especially in the long run, which are presumed here following the previous literature focusing on the U.S. Second, by only taking into account the changes in party polarization between two consecutive elections in 12 European democracies, this article disregards an important feature pertinent to some party systems—i.e., constantly increasing polarization. Without longitudinal cross-national data extending to a longer time frame, the scope of this study is thus limited. Third, by assuming that the mapping of particular policies onto the traditional left-right policy continuum is uniform across the party systems examined, the multidimensionality of political competition in some European party systems is also disregarded. Lastly, only focusing on parliamentary elections in multiparty systems, our findings may not be generalizable to party systems with less permissive electoral rules, different regime types, and other types of elections.
Nonetheless, this article has important implications. To the extent that distinct party policy offerings are put on offer to the electorate, abstention from alienation is not necessarily a concern in multiparty democracies. Citizens with moderate views to whom mainstream political parties in two-party systems are motivated to appeal are not negatively affected by high party polarization in multiparty systems putting on offer centrist party policies. As long as a party system provides citizens with sufficiently distinct party policy alternatives, the effect of the knowledge gap between sophisticated and unsophisticated citizens on their participation in the democratic process is thus not an important concern in multiparty settings.
The literature has long suggested that the distinctiveness of party policy offerings is a normatively desirable function of representative democracies and responsible party governments. The empirical analyses in this article lead to a common conclusion in the previous literature on the relationship between party competition and political participation, albeit with a slight modification: not only more numerous, but also more distinct party policies are needed to encourage and sustain higher citizen participation in politics. In this regard, I believe further research should focus more on how the distinctiveness, rather than the mere count, of party policy offerings affects political participation and electoral behavior, while taking into account the limitations and contributions of this study noted above.
Notes
Some post-election surveys do not provide information about political information (Slovenia 2008), union membership (Spain 2008), income (Latvia 2010), and vote shares of parties at the district level (Romania 2009).
France (2007) is not included in the sample due to the semi-presidential system and two-round elections. For the effects of electoral rules on political polarization, see Ezrow (2008), Dow (2011), Best and Dow (2015), McDonald and Moral (2015). Polish election in 2005 is also excluded due to the second round of the presidential election that took place shortly before the parliamentary election, which is likely to produce a coattail effect (Cox 1997; Clark and Golder 2006), thus to influence respondents’ polarization perceptions.
Post-election surveys examined in Models 1 and 2 are: Austria (2008), Croatia (2007), Czech Republic (2006, 2010), Denmark (2007), Estonia (2010), Finland (2007, 2011), Germany (2005, 2009), Greece (2009), Iceland (2007, 2009), Ireland (2007), Netherlands (2006, 2010), Norway (2005, 2009), Poland (2007), Portugal (2009), Slovakia (2010), Sweden (2006), Switzerland (2007).
The sample employed in Model 3 consists of the latter post-election surveys in Czech Republic, Denmark, Finland, Germany, Iceland, Ireland, Netherlands, Norway, Portugal, Sweden, and Switzerland.
CSES data provide validated voting records for only three post-election surveys in Norway (2005, 2009) and Sweden (2006). In such instances, the dependent variable is accordingly recoded, and the mean turnout rates in our data are closer to the observed turnout percentages. Overreporting of turnout with a sense of citizen duty may be a possible explanation for such observed differences between turnout rates (Gerber et al. 2008). Another possibility is that listwise deletion of observations with missing data might lead to higher observed turnout rates in our sample. Additional sensitivity checks for all models in Table 1 do not show a systematic relationship between perceived and actual party polarization and overreporting of turnout: Although predicted probabilities of turnout are expectedly lower to a small extent, the coefficients estimates do not change significantly when self-reported turnout data are employed.
Descriptive statistics in Table 3 in Appendix show that 23 post-election surveys in our sample have a mean actual party polarization of 3.59 with a standard deviation of 0.61.
While some European countries have constantly high or low party polarization, others reveal increasing or decreasing trends between consecutive elections. The mean change in actual party polarization is −0.11. The most substantial decrease in actual party polarization is about 0.8 points in Poland between 2005 and 2007, while party polarization reveals a positive trend between the consecutive elections in Denmark, Germany, Iceland, and Netherlands. Table 4 in Appendix shows the trends in all post-election surveys examined in this article.
An example might better illustrate the coding of these three measures of party polarization. For the German post-election surveys in 2005 and 2009, the respondents were asked to place six major parties on the traditional left-right continuum: Christian Democratic Union (6.4), Social Democratic Party (3.7), Free Democratic Party (5.8), Left Party (1.1), Alliance ’90/Greens (3.3), and Christian Social Union (7). Using the mean placements in parentheses, I computed the distance between the party groups to the left (i.e, Left, Alliance 90 and SPD), and to the right (i.e., FDP, CDU and CSU) of the party system center (in turn 2.7 and 6.4) to create the “actual party polarization” variable (3.7). The “change in actual party polarization” between 2009 and 2005 (Mean placements in 2005 were respectively 6.1, 3.6, 5.2, 1.2, 3.3, 6.5) is thus a substantial increase by 0.5 point (from 3.2 to 3.7). To reiterate, the measures entitled “perceived party polarization” in Model 1 and “perceived party polarization (difference)” in Models 2 and 3 take individual respondents placements of parties into account. For instance, if a respondent in the 2009 survey placed those parties respectively at 7, 3, 6, 1, 2 and 8, his perceived polarization variable score would be 5. I then calculated the difference between perceived and actual party polarization (1.3) and divided it by the standard deviation of this variable for the German post-election survey in 2009 (1.55) for standardization. Hence, the hypothetical respondent perceives party polarization as .8 standard deviations higher than the actual party polarization and has a “perceived party polarization (difference)” score of 0.8.
For a comparison of the measure of polarization employed here with other measures of polarization in previous literature and the sensitivity checks taking those measures as the primary party system-level independent variable, see Appendices C and E.
Some commonly employed measures in previous literature approximate the extremity, dispersion, or diversity of party policy offerings, rather than their polarization. See Best and Dow (2015) and McDonald and Moral (2015) for useful discussions about those interrelated but distinct concepts often used interchangeably in previous literature.
Factual knowledge items in the CSES surveys are usually not very cognitively demanding such as those about the length of electoral terms and the names of prominent elected officials. In surveys with lower mean scores for the unstandardized index, respondents are asked more demanding questions such as the relative share of public expenses across distinct policy domains (e.g., Denmark (2007), where the mean is −.4) or a series of knowledge items in a row (e.g., Norway (2005), where the mean is −1.3). However, survey means of the unstandardized variable can be as high as 1.3—e.g., Poland (2007).
Models A5 and A6 in Table 5 in Appendix relax this assumption and introduce an interaction term with education as another proxy for political sophistication.
Some citizens may rationalize ideological and policy positions of candidates (Campbell et al. 1960; Page and Brody 1972; Abramowitz 1978; Markus and Converse 1979). To my knowledge, there is no study to date with the exception of Jacoby and his colleagues’ (2010) that examines whether such perceptual bias is a function of political knowledge. Although the authors do not find a systematic difference between politically informed and uninformed respondents, because such pattern would constitute an important problem for the empirical analyses in this article, introducing an interaction term between political information and perceived polarization, rather than solely relying on compression effects in the logistic regression, is preferred.
In addition to political knowledge, the literature on projection bias suggests that citizens’ partisan attachments and ideological extremity condition their perceptions about parties and candidates’ policy or ideological stands. The correlations between perceived polarization (difference) and ideological extremity and strong partisan attachment are, however, 0.12 and 0.27 in turn.
Replication data and code are available online at https://dataverse.harvard.edu/dataverse/mertmoral.
Model 1 in Table 1, which takes perceived party polarization, rather than its difference from the actual party polarization, as the primary independent variable is reported for comparison purposes. Model 1 also provides empirical evidence supporting our major theoretical expectation regarding the positive effect of perceived party polarization on voter turnout in multiparty settings.
The marginal effect of perceived polarization is calculated by setting perceived polarization first to 0 and then to 1 to compute the (first) difference in predicted probabilities. 10,000 estimates are drawn from a multivariate normal distribution with the means of estimated coefficients and variances specified by the variance-covariance matrix from Model 2.
Four sets of robustness checks are presented in Table 5 in Appendix. Since comparative studies on political polarization suggest that electoral rules might affect the diversity and polarization in party policy offerings (Ezrow 2008; Dow 2011; McDonald and Moral 2015), two mixed party systems - Germany and Greece- were dropped from the samples employed in Models A1 and A2. The respondents who failed to place less than three of four largest parties, and those who misplaced left-wing parties to the right of right-wing parties are included in the samples employed in Models A3 and A4. In Models A5 and A6, perceived polarization is interacted with education as another proxy for political sophistication. Lastly, multilevel mixed-effects logistic regressions in Models A7 and A8 drop election fixed effects from model equations, and introduce random intercepts for each post-election survey and random coefficients for the actual party polarization and change in actual party polarization variables. Our conclusions that high perceived polarization and that high and increasing actual party polarization increase voter turnout are robust to such changes. In addition, four sensitivity checks reported in Table 6 in Appendix replace the measure of polarization employed in the empirical analyses in this article with other measures in previous literature—i.e., individual-level perceived polarization and aggregate-level party polarization in Crepaz (1990), weighted party system extremism in Dow (2011), and the measures of party polarization in Lachat (2008) and in Esteban and Ray (1994). Changing how we approximate party polarization does not alter the conclusions.
References
Abramowitz, A. I. (1978). The impact of a presidential debate on voter rationality. American Journal of Political Science, 22(3), 680–690.
Abramowitz, A. I. (2010). The disappearing center: Engaged citizens, polarization, and American democracy. New Haven: Yale University Press.
Abramowitz, A. I., & Saunders, K. L. (2008). Is polarization a myth? Journal of Politics, 70(2), 542–555.
Abramowitz, A. I., & Stone, W. J. (2006). The bush effect: Polarization, turnout, and activism in the 2004 presidential election. Presidential Studies Quarterly, 36(2), 141–154.
Adams, J., & Merrill, S, I. I. I. (2003). Voter turnout and candidate strategies in American elections. Journal of Politics, 65(1), 161–189.
Adams, J., Clark, M., Ezrow, L., & Glasgow, G. (2006a). Are niche parties fundamentally different from mainstream parties? The causes and the electoral consequences of western European parties’ policy shifts, 1976–1998. American Journal of Political Science, 50(3), 513–529.
Adams, J., Dow, J. K., & Merrill, S, I. I. I. (2006b). The political consequences of alienation-based and indifference-based voter abstention: Applications to presidential elections. Political Behavior, 28(1), 65–86.
Aldrich, J. H. (1983). A downsian spatial model with party activism. American Political Science Review, 77(4), 974–990.
Aldrich, J. H., & McGinnis, M. D. (1989). A model of party constraints on optimal candidate positions. Mathematical and Computer Modelling, 12(4/5), 437–450.
Best, R. E., & Dow, J. K. (2015). Simplicity is a virtue: Measuring party system polarization. Annual Meeting of the American Political Science Association, San Francisco, CA.
Blais, A. (2006). What affects voter turnout? Annual Review of Political Science, 9, 111–125.
Brambor, T., Clark, W. R., & Golder, M. (2006). Understanding interaction models: Improving empirical analyses. Political Analysis, 14(1), 63–82.
Budge, I., & McDonald, M. (2006). Choices parties define. Party Politics, 12(4), 451–466.
Callander, S., & Wilson, C. H. (2007). Turnout, polarization, and duverger’s law. Journal of Politics, 69(4), 1047–1056.
Campbell, A., Converse, P. E., Miller, W. E., & Stokes, D. E. (1960). The American voter. New York: Wiley.
Clark, W. R., & Golder, M. (2006). Rehabilitating duverger’s theory: Testing the mechanical and strategic modifying effects of electoral laws. Comparative Political Studies, 39(6), 679–708.
Committee on Political Parties APSA. (1950). Toward a more responsible two-party system: A report of the committee on political parties. American Political Science Review, 44(Supplement), 1–96.
Cox, G. W. (1990). Centripetal and centrifugal incentives in electoral systems. American Journal of Political Science, 34(4), 903–935.
Cox, G. W. (1997). Making votes count: Strategic coordination in the world’s electoral systems. Cambridge: Cambridge University Press.
Crepaz, M. M. (1990). The impact of party polarization and postmaterialism on voter turnout: A comparative study of 16 industrial democracies. European Journal of Political Research, 18, 183–205.
CSES (2007) CSES Module 2 full release. The comparative study of electoral systems. Retrieved June 12, 2014, from http://www.cses.org.
CSES (2013) CSES Module 3 full release. The comparative study of electoral systems. Retrieved June 12, 2014, from http://www.cses.org.
Dalton, R. J. (2008). The quantity and the quality of party systems: Party system polarization, its measurement, and its consequences. Comparative Political Studies, 41(7), 899–920.
Dow, J. K. (2011). Party-system extremism in majoritarian and proportional electoral systems. British Journal of Political Science, 41(2), 341–361.
Downs, A. (1957). An economic theory of democracy. New York: Addison-Wesley Publishing.
Esteban, J. M., & Ray, D. (1994). On the measurement of polarization. Econometrica, 62(4), 819–851.
Ezrow, L. (2008). On the inverse relationship between votes and proximity for niche parties. European Journal of Political Research, 47, 206–220.
Ezrow, L., & Xezonakis, G. (2011). Citizen satisfaction with democracy and parties’ policy offerings. Comparative Political Studies, 44(9), 1152–1178.
Ezrow, L., & Xezonakis, G. (2016). Satisfaction with democracy and voter turnout: A temporal perspective. Party Politics, 22(1), 3–14.
Fiorina, M. P., Abrams, S. J., & Pope, J. C. (2011). Culture war? The myth of a polarized America (3rd ed.). New York: Pearson-Longman.
Franklin, M. N. (2004). Voter turnout and the dynamics of electoral competition in established democracies since 1945. New York: Cambridge University Press.
Garner, A., & Palmer, H. (2011). Polarization and issue consistency over time. Political Behavior, 33(2), 225–246.
Gerber, A. S., Green, D. P., & Larimer, C. W. (2008). Social pressure and vote turnout: Evidence from a large-scale field experiment. American Political Science Review, 102(1), 33–48.
Geys, B. (2006). Explaining voter turnout: A review of aggregate-level research. Electoral Studies, 25(4), 637–663.
Hetherington, M. J. (2001). Resurgent mass partisanship: The role of elite polarization. American Political Science Review, 95(3), 619–631.
Hetherington, M. J. (2008). Turned off or turned on? How polarization affects political engagement. In P. S. Nivola & D. W. Brady (Eds.), Red and blue nation? Consequences and correction of America’s polarized politics. Harrisonburg, Virginia: Brookings Institution Press.
Hetherington, M. J. (2009). Putting polarization in perspective. British Journal of Political Science, 39(2), 413–448.
Inglehart, R. (1977). The silent revolution : Changing values and political styles among western publics. Princeton, NJ: Princeton University Press.
Inglehart, R. F. (2008). Changing values among western publics from 1970 to 2006. West European Politics, 31(1–2), 130–146.
Jacoby, W. G., Ciuk, D., & Pyle, K. (2010). Citizen perceptions of candidate ideology in the 2008 election. Annual meeting of the American political science association, Wasington, DC.
Knutsen, O., & Kumlin, S. (2005). Value orientations and party choice. In J. Thomassen (Ed.), The European voter: A comparative study of modern democracies (pp. 125–166). Oxford: Oxford University Press.
Lachat, R. (2008). The impact of party polarization on ideological voting. Electoral Studies, 27(4), 687–698.
Lewis-Beck, M. S., Jacoby, W. G., Norpoth, H., & Weisberg, H. F. (2008). The American voter revisited. Ann Arbor: The University of Michigan Press.
Markus, G. B., & Converse, P. E. (1979). A dynamic simultaneous equation model of electoral choice. American Political Science Review, 73(4), 1055–1070.
McCarty, N., Poole, K. T., & Rosenthal, H. (2006). Polarized America: The dance of ideology and unequal riches. Cambridge, MA: MIT Press.
McDonald, M. D., & Moral, M. (2015). On observing the variety of party policy offerings in national left-right policy space. Annual meeting of the American political science association, San Francisco, CA.
Miller, G., & Schofield, N. (2003). Activists and partisan realignment in the united states. American Political Science Review, 97(2), 245–260.
Norris, P. (2004). Electoral engineering: Voting rules and political behavior. New York: Cambridge University Press.
Page, B. I., & Brody, R. A. (1972). Policy voting and the electoral process: The vietnam war issue. American Political Science Review, 66(3), 979–995.
Pardos-Prado, S., Galais, C., & Munoz, J. (2014). The dark side of proportionality: Conditional effects of proportional features on turnout. Electoral Studies, 35, 253–264.
Plane, D. L., & Gershtenson, J. (2004). Candidates’ ideological locations, abstention, and turnout in U.S. midterm senate elections. Political Behavior, 26(1), 69–93.
Poole, K. T., & Rosenthal, H. (1997). Congress: A political-economic history of roll call voting. New York: Oxford University Press.
Powell, G. B. (1982). Contemporary democracies: Participation, stability and violence. Cambridge, MA: Harvard University Press.
Rabinowitz, G., & Macdonald, S. E. (1989). A directional theory of issue voting. American Political Science Review, 83(1), 93–121.
Riker, W. H. (1986). The art of political manipulation. New Haven: Yale University Press.
Rogowski, J. C. (2014). Electoral choice, ideological conflict, and political participation. American Journal of Political Science, 58(2), 479–494.
Schattschneider, E. E. (1960). The semisovereign people: A realist’s view of democracy in America. New York: Harcourt Brace Jovanovich College Publishers.
Schofield, N., & Miller, G. (2007). Elections and activist coalitions in the united states. American Journal of Political Science, 51(3), 518–531.
Siaroff, A., & Merer, J. W. A. (2002). Parliamentary election turnout in europe since 1990. Political Studies, 50, 916–927.
Steiner, N. D., & Martin, C. W. (2012). Economic integration, party polarisation and electoral turnout. West European Politics, 35(2), 238–265.
Thornton, J. R. (2013). The impact of elite polarization on partisan ambivalence and indifference. Political Behavior, 35, 409–428.
Ulbig, S. G., & Funk, C. L. (1999). Conflict avoidance and political participation. Political Behavior, 21, 265–282.
Acknowledgements
Earlier versions of this paper were presented at the 111th Annual Meeting of the American Political Science Association and at the Comparative-American Politics Workshop in Binghamton University. I thank to the participants, Michael D. McDonald, Olga V. Shvetsova, Robin E. Best, three anonymous reviewers, and the Editor for helpful suggestions. Any remaining errors are, of course, my own.
Author information
Authors and Affiliations
Corresponding author
Appendix
Appendix
Rights and permissions
About this article
Cite this article
Moral, M. The Bipolar Voter: On the Effects of Actual and Perceived Party Polarization on Voter Turnout in European Multiparty Democracies. Polit Behav 39, 935–965 (2017). https://doi.org/10.1007/s11109-016-9386-0
Published:
Issue Date:
DOI: https://doi.org/10.1007/s11109-016-9386-0