Electoral Rules and Citizens Trust in Political Institutions

Transcription

1 Electoral Rules and Citizens Trust in Political Institutions Sofie Marien Centre for Political Research, University of Leuven, Belgium. Contact: Abstract There is an extensive literature on the effects electoral rules have on citizens attitudes towards the political system. While proportional rules stimulate inclusiveness and representation, disproportional rules foster government accountability and effectiveness. However, empirical studies yield mixed results showing proportional as well as disproportional election outcomes to increase political trust or finding no relation at all. In this paper, we will outline how political trust is affected by the proportionality of election outcomes using data from the European Social Survey ( ). Also the most common explanations that have been advanced for this relation will be tested empirically. These include the effect of the party-system on political trust and the effect of coalition governments on political trust. Also the effect of voting for the winning or losing party on citizens trust under different levels of proportionality will be investigated. Finally, the question whether the more diffuse nature of coalition governments has an effect on the evaluation of its performance will be tackled. Paper presented at the 38 th ECPR Joint Sessions Münster (Germany) March, 22-27, 2010 Workshop: Institutional Performance and Political Support in Europe Sonja Zmerli and Marc Hooghe

2 Introduction Various theories have been advanced to account for cross-national differences in levels of political trust (e.g. Nye et al. 1997; Norris 1999; Newton 2006). One line of research has mainly looked at government and regime performance, especially focusing on government s ability to handle the economy (Bok 1997; Kampen et al. 2006; Wagner et al. 2009). Another line of research has focused on cultural explanations such as the differences in levels of social capital and postmodern values (Inglehart 1999; Mansbridge 1997; Newton & Zmerli 2009). Another body of literature has stressed the role that institutions play in shaping citizens attitudes towards the political system. Studies on the importance of institutional structures for political trust have multiplied in recent years, stimulated by cases of divided societies and the emergence of new democracies (Lijphart 1999; Farrell & McAllister 2006). For its democratic establishment, a reservoir of public trust for the democratic structures needs to be built, minimizing the risk of reverting to authoritarianism (Norris 1999, 266; Almond & Verba 1963). In established democracies too, citizens widespread dissatisfaction with politics has been interpreted as the result of their dissatisfaction with the working of the main political institutions calling for institutional change (Dalton 2004,177-87). For instance, electoral reforms in New Zealand, Italy and Japan have been linked to citizens dissatisfaction (Shugart & Wattenberg 2001). Indeed, as institutions create the framework in which people interact, it is important to take institutional properties, both formal and informal, into account when explaining individual attitudes. Through institutions, especially the electoral system, citizens and elites are linked and political trust is often seen as an indication of the quality of this linkage (Aarts & Thomassen 2008). Citizens communicate their preferences to the elites and elites can be held accountable through elections. Electoral rules do not only determine election outcomes but also the strength of a party s victory or loss, the party system, party fragmentation, the occurrence of single-party governments or coalitions, the representation of political minorities, voter participation etc. (Lijphart 1994; Sartori 1994). While proportional electoral rules stress the inclusiveness of parliament, majoritarian electoral rules stress government accountability, hence, these electoral rules produce substantially different degrees of representativeness and accountability: the major difference between the majoritarian and proportional vision is their view on the essence of democratic government and consequently the function of elections (Aarts and Thomassen 2008, 6). In this line of reasoning, distrust can be interpreted as the result of successive exclusion from representation, a perceived lack of accountability or accumulated experiences with system 2

3 unresponsiveness. In general, political trust is not the result of a single election outcome, rather the accumulation of successive elections outcomes shapes citizens attitudes towards the political system. Despite, the substantial literature on the effects of institutional design on citizens attitudes towards the political system, empirical studies testing these propositions have been in much shorter supply. Moreover, evidence emerging from the existing studies is far from conclusive. Proportional as well as majoritarian systems are found to foster political trust or finding no relation at all (Lijphart 1999; Norris 1999; Van der Meer & Dekker 2009; Wagner et al 2009; Norris 2010). Farrell & McAllister (2006, 724) conclude: Whether and how electoral systems can affect levels of satisfaction with democracy is therefore unresolved; work is required. In this paper, we will disentangle this key research question empirically by providing a cross-national analysis on recent data of 23 countries. First, we provide an overview of the theoretical arguments and previous empirical research on the relation between electoral rules and political trust. Subsequently, we turn to comparative survey data to analyze whether there is a relation between political trust and proportionality of election outcomes and which features of proportionality are responsible for this. The main findings are summarized in the conclusion. Proportional and majoritarian electoral rules and political trust Proportional election outcomes are generally seen as more righteous than the winner take it all outcome of majoritarian electoral rules (Sartori 1994). By definition, seats are allocated to parties in proportion to the votes they receive at election and so voters determine to a great extent which parties obtain seats and only few votes are wasted. It is argued that these features of proportionality increase citizens sense of fairness and political efficacy (Karp & Banducci 2008). Further, it lowers the threshold for parliamentary representation encouraging more parties to enter the electoral contest resulting in the presentation of various alternatives to the electorate. In effect, in PR-systems political parties presenting themselves to the electorate are as diverse as socialists, neo-liberals, green parties, regionalists (leftleaning and right-leaning), conservative parties, communist parties, extreme-right parties, christian democrats, parties defending the interest of pensioners, pirate-parties, animal parties etc. Thereby citizens feeling of influence is strengthened as party choice can effectively be used to voice (specific) political preferences. New issues cherished by the public can easily be put on the agenda by new minor parties and discontent can be voiced by voting on minor 3

4 protest parties which can channel this discontent into the parliamentary arena (Miller & Listhaug 1990). Under proportional election rules, more parties are passing the threshold to parliamentary representation resulting in a more inclusive representation of the whole electorate and its political opinions. Political minorities are represented in parliament and can influence political decision-making (Lijphart 1994; Karp & Banducci 2008). Proportional electoral rules are an important element of power-sharing regimes which give multiple elites a stake in the decision-making process (Norris 2008), put another way, they produce more winners satisfying larger groups. Even if the party one has voted for does not become the party in office, often parliamentary representation is gained. Therefore, scholars argue that in proportional systems citizens voting for party which has lost the election will have higher levels of political trust than losers in disproportional systems (Anderson et al. 2005). By contrast, disproportional election rules are intended to fabricate clear majorities often leading to two-party systems (Sartori 1994). As it limits the choice offered to the electorate, it leaves many voters without a political party with which they can identify (Powell 1989). In effect, under majoritarian rules third parties obtain votes but hardly ever seats; and if they do, as in England, their parliamentary representation still remains too small to hurt (Sartori 1994, 39). As a result, a substantial amount of votes are wasted and political minorities are persistently excluded from the political system. It has been argued that these conditions erode political trust and are likely to result in alienation from the political system. However, political trust can also erode when many political parties gain representation as the aggregation of interests into broad ideological coalitions becomes difficult (Norris 1999, 225). Further, electoral rules determine to a great extent the number of parties in government: while coalition governments are the standing practice under proportional rules, disproportional rules most likely result in single-party governments. In effect, disproportional rules do not seek a parliament that reflects the voting distribution; they seek a clear winner. Their intent is not only to elect a parliament but at the same time elect (if only by implication) a government (Sartori 1994, 5). In majoritarian systems in general two (blocks of) parties are competing in elections, the winner of the election takes over the government and can implement a clear policy (change). The party in government does not have to compromise so much as a coalition government does. In this line of reasoning, citizens voting for the party which won the election have more to gain than winners in proportional systems, therefore, they are thought to have higher levels of trust. Given that election losers are theorized to have lower levels of trust in disproportional systems, the gap between winners and losers is larger 4

5 in countries with disproportional election outcomes. Coalitions on the other hand, have the advantage that a broad range of interests find their way to policy and this policy is found to be closer to the median voter than the policy of single-party governments (McDonald & Budge 2005). However, it can also be argued that voters incline to think that it does not matter who is in power leading to resignation and alienation. When a single party is in office, the policy responsibility is clear whereas responsibility is more diffuse in coalition governments: usually there is an overlap between the new and the old coalition, blurring the clarity of responsibility and making the sanction of elections as an instrument of accountability into a rather blunt weapon (Aarts & Thomaasen 2008, 7; see also Powell 1989). Disproportional election rules stress the rotation of parties in power more than proportional rules. In a PRsystem it is not unusual that a party losing the election returns to power, making it difficult to vote the rascals out (Norris 1999b; Powel 1989). Next to this clarity of responsibility, single-party governments are seen as more effective governments. In the debate on electoral reform in New-Zealand one of the main concerns when implementing more proportional electoral rules was a decrease in government s effectiveness and accountability (Vowles et al. 2006, ). Coalitions, especially with many parties, are argued to be more unstable and more often lead to policy impasse (Weil 1989). In sum, both proportional and disproportional election rules are theorized to foster political trust. While proportional rules stimulate inclusiveness and representation, disproportional rules foster government accountability and effectiveness. Empirical research Banducci et al. (1999) found that following the reform of the first-past-the-post to the more proportional MMP-system in New Zealand, citizens orientation towards the political system became significantly more positive, they felt more political efficacious and government was perceived as more responsive. Especially the attitudes of political minorities became more positive under the more proportional rules (Banducci et al. 1999). However, as only few countries changed their electoral system so profoundly in recent years (Farrell 2001), most research on the relation between electoral systems and political trust is based on cross-national comparisons. In this way, Lijphart s study (1999, p ) established that citizens satisfaction with democracy was 17 percentage points higher in countries with more proportional rules. i Political efficacy too was found to be higher under more proportional electoral rules, especially among supporters of small political parties (Karp & Banducci

6 based on CSES-data of 22 countries). However, Norris research on World values studies-data established higher levels of political trust in majoritarian systems than in proportional systems (Norris 1999, 25 countries were studied). Still other studies did not find any relation at all between citizens political trust and their electoral system. For instance Listhaug et al based on CSES-data in 28 countries, Norris (forthcoming) on more recent WVS-data in 42 countries, Wagner et al. (2009) on different waves of Eurobarometer data. Empirical studies looking at the effect of a country s party-system on citizens trust found higher trust levels in countries with more parties contesting elections: Citizens judge democracy less by what it gives them than by whether it presents them with real (but not polarized) alternatives and responds to their choices (Weil 1989, 699). Miller & Listhaug (1990) concluded from their empirical study on US, Sweden and Norway that the varying thresholds for parliamentary representation were crucial in determining the target, level and trends in political distrust. In Norway which had a rather low threshold for parliamentary representation, new parties won seats in elections and disaffected voters were represented, whereas in US and Sweden discontent was accumulated and resulted in higher levels of distrust: the institution of elections has failed, during the past twenty years, to act as a mechanism for reducing the accumulating dissatisfaction in either Sweden or the United States (Miller & Listhaug 1990, 383). However, some evidence suggests that party fragmentation lowers political trust (Norris 1999; Weil 1989). But in other studies fractionalization indices failed to reach conventional levels of statistical significance (Wagner et al 2009; Van der Meer & Dekker 2009). Empirical studies looking at the relation between the number of parties in office and citizens trust found lower levels of political efficacy in countries with more parties in office, also the likelihood to vote proved to be lower (Karp & Banducci 2008). Regarding political trust, the evidence points at a curvilinear relation: political trust is higher in countries with coalition governments, but too much coalition partners diminish political trust (Anderson et al. 2005). It is argued that in such countries elites rather than election outcomes determine governments composition. Weil s (1989) analyses of coalition size and political trust in six countries yielded mixed results with minimum-winning coalitions generating higher levels of trust than oversized or minority coalitions in the United States and Italy but lower levels in Spain. In Germany different indicators of trust even yielded different results. Cabinet instability, however, was clearly related to lower levels of political trust. Extensive research effort has been devoted to the effects of winning and losing elections on political trust in different institutional contexts. Anderson and Guillory (1997) 6

7 demonstrated that losers were more trusting in consensual systems than in majoritarian systems and winners were more trusting in majoritarian than consensual systems using the 1990 Eurobarometer-data of 12 countries. Hence, the gap in political trust is larger in majoritarian systems than consensual systems. This finding was confirmed in subsequent research (e.g. Criado and Herreros 2007 using ESS data of 17 countries). However, these studies make a distinction between two institutional systems based on elements such as the state structure, the number of legislative chambers, judicial reviews, parliamentary or presidential system etc. From these studies, it is unclear which features of these systems are decisive. Later research has sorted this out and proportionality was included as an independent variable leading scholars to conclude that proportional rules affected winners and losers trust (Anderson et al. 2005). However, this conclusion need to be qualified, while their analyses yielded a positive effect of proportional electoral rules on political trust, the differential effect of these rules on winners and losers was less convincing. ii Finally in the outline of the theoretical arguments, institutional design was also argued to influenced political trust through its effect on the clarity of responsibility. In empirical research, particular attention has been paid to the relation between economic evaluations and political trust in different institutional settings (Powell & Whitten 1993; Anderson 2000; Criado & Herreros 2007). The study of Anderson (2000) using the 1994 Eurobarometer data on 13 European democracies, established for example that the relation between voters evaluation of economic performance and political trust is stronger when mechanisms of accountability are simple i.e. when the institutional context clarifies who is in charge of policymaking (based on Powell and Whitten s index of clarity of responsibility), when the target of credit and blame is large (coalitions), and when citizens have fewer viable alternative choices (operationalized as parties in parliament). In sum, theoretical approaches underlie the argument that proportional rules foster political trust as well as disproportional rules. Empirical research yielded evidence supporting both claims or finding no relation between proportionality and political trust (Lijphart 1999, Norris 1999, Listhaug et al. 2009, Van der Meer & Dekker 2009). One possible explanation is the operationalization of proportionality. Often electoral systems are categorized in three or four systems which is a rather crude measure as it ignores quite some variance in proportionality within these systems. Some empirical studies simplify things even more using only two categories of systems majoritarian and consensual which are not only based on electoral rules but take also other system properties such as the state structure or the executive 7

8 into account. Notwithstanding, it is useful to take these features into account, it becomes difficult to disentangle which features affect political trust. In this paper, we will outline how political trust is affected by the proportionality of election outcomes i.e. the translating of votes into seats. Therefore, the main research question phrases: What is the relation between the proportionality of election outcomes in a country and citizens political trust? From the overview of the theory on institutional design and political trust, two contradicting hypothesis can be formulated. H1a: In countries with more proportional election outcomes, political trust is higher than in countries with less proportional election outcomes. H2b: In countries with more proportional election outcomes, political trust is lower than in countries with less proportional election outcomes. In the next part, we will test which hypothesis fits the data. However, we do not only want to investigate which electoral rules foster political trust. Also some of the explanations that have been advanced for this will be explored in the second part of the paper. Several explanations derived from the theoretical arguments on electoral rules and political trust are subject to scrutiny. Namely the effect of the party-system and coalition governments on political trust, the effect winning and losing elections has on citizens political trust under different levels of proportionality. Finally, the question whether the more diffuse nature of coalition governments has an effect on the evaluation of its performance will be tackled. These questions will be taken up in the second part but first we will investigate whether there is a relation between proportionality of election outcomes and political trust. Data and Methods The third and fourth wave of the European Social Survey ( ) will be used to assess the relation between proportionality and political trust. The European Social Survey (2006; 2008) provides reliable and up-to-date information on trust in political institutions in 23 European countries with a substantial range of variation in proportionality (Jowell et al. 2007; Jowell et al. 2009). iii This dataset is ideally suited for comparative analysis since particular attention has been paid to ensure an optimal comparability with regard to the operationalization and cross-cultural validity of concepts in the participating countries. Data 8

9 were collected between 2006 and 2009 by means of uniform face-to-face interviews among representative samples of the population of 23 European countries. iv Given the small timespan between both waves and the rather stable nature of trust in political institutions, wave 3 and wave 4 of the European social survey were merged into one dataset broadening the number of countries, and in particular electoral systems, included in the analyses. Response rates varied by country with a mean of 63 per cent and in most countries exceeding 60 per cent (Jowell et al. 2007; Jowell et al. 2009). Dependent variable The question on political trust in ESS 2006 and ESS 2008 was phrased: please tell me on a score of 0 to 10 how much you personally trust each of the institutions I read out. 0 means you do not trust an institution at all, and 10 means you have complete trust. The institutions include: the national parliament, politicians and political parties. Principal component analysis indicated that the three items loaded strongly on a single dimension, explaining per cent of the total variance with an Eigenvalue of 2.53 (Table 1). We therefore created an index as a simple additive summary of the three items which ranges from 0 to 30 with a mean of and a standard deviation of 6.65 (Cronbach s α: 0.90). Table 1. Political Trust Political trust Trust in country's parliament Trust in political parties Trust in politicians Eigen Value 2.53 Explained variance Extraction Method: Principal Component Analysis Independent variables In most empirical research, electoral systems are being categorized into three or four groups namely majoritarian systems, mixed (independent or dependent) systems and proportional systems, however, within these systems there is a substantial variation in proportionality depending on e.g. district size or legal thresholds. In order to take this variation into account, we will use Gallagher s disproportionality index (1991; 2008) which looks at the degree to which the allocation of seats matches the distribution of the national vote at the parliamentary elections. v In theory, this index can range from 0 to 100 with high values indicating disproportional election outcomes. Given that most authors argue that 9

10 accumulated experience with the electoral system (and proportional or disproportional election outcomes) influences citizens political trust, the Gallagher index of disproportionality was calculated for each country based on the election outcomes of the past 10 years. vi In our sample, the Netherlands have the most proportional election outcomes with a mean value of 0.99, while France has the most disproportional election outcomes resulting in a mean value of on the index. The mean value in our sample is 5.14 with a standard deviation of Without France and the United Kingdom, the two most disproportional systems in our sample, the mean value is 3.96 with a standard deviation of If we look only at the last election outcomes of the countries in our sample, the index amounts to 5.85 in 2006 and 4.53 in Table 2. Disproportionality in Europe N Mean SD Min Max Average level of disproportionality (last 10 years) All countries Index without UK and France Disproportionality in 2008 All countries Index without UK and France Disproportionality in 2006 All countries Index without UK and France Source: Gallagher (1991), author s calculations. A number of control variable were added on the individual level and country level (See Appendix for the descriptives of all variables). On the individual level, we included some general socio-demographic variables such as age, education level and gender which are assumed to affect citizens outlook to political institutions (Inglehart 1999; Schwindt-Bayer & Mishler 2005). The variable education level includes seven categories with a value 0 indicating not completed primary education and a value 6 indicating second stage of tertiary. Women received a code 0 and men received a code 1. Also satisfaction with the economy is included as it has been shown to be an important predictor of political trust (Bovens & Wille 2008). Furthermore, winning or losing elections was also found to influence political trust substantially (Anderson & Guillory 1997). A variable winner/loser was created: respondents that indicated to have voted during the last parliamentary elections for a political party which was in office at the moment of the fieldwork received value 1, 10

11 respondents that voted for another party or did not turn out received value 0. These data were gathered from various issues of the Annual Data Yearbook of the European Journal of Political Research. On the country level, three important control variables were added which have been demonstrated to influence political trust: democratic experience, performance and economic situation (Thomassen & Van der Kolk 2009; Tavits 2007; Uslaner 2008). Firstly, a dummy variable was created measuring a country s democratic experience based on the Polity IV dataset (Marshall & Jaggers 2009). Countries that did not experience a regime change within the last 20 years received a value 1. Secondly, a summary indicator for a country s performance was created by taking the mean of the five World Bank indicators of good governance for (Kaufmann et al. 2009). Finally, not only citizens subjective evaluation of the economy but also an objective indicator of a country s economic situation was added namely the average GDP/capita in US dollars for (WEO 2008). As these three control variables are also substantially correlated with each other, they will not be included in the same analysis to avoid issues of multicollinearity. In the literature, also other variables are found to influence political trust such as individuals social trust (Zmerli & Newton 2009), religious attitudes (Halman 2007) or the country s state structure (Elazar 1997). All analyses were performed including also these variables but given that the coefficients did not substantially change, and for reasons of parsimony, these variables were not included in the final models. Given that we are interested in the effect of country-level variables on individual attitudes, multilevel techniques will be used (Hox 2002). Description of the data Figure 1 visualizes political trust levels in the 23 countries included in the analyses revealing substantial cross-national differences to explain e.g. in Denmark the average political trust-score was 17.7 on a scale of 30, while the average political trust-score in Bulgaria was only 5.3 on a scale of 30. vii 11

12 Figure 1. Political trust in Europe ( ) Political trust DK NL FI CH NO SE CY BE IE ES AT SK DE UK EE FR SI IL RU PT HU PL BG Multilevel analysis A multilevel analysis was performed in order to explain these cross-national differences and to tackle our research question: What is the relation between the proportionality of election outcomes in a country and citizens political trust?. In table 3 the standardized coefficients of the regression are represented, the unstandardized results can be found in appendix. We start with a null-model, showing indeed a considerable amount of variance at the country level: per cent of all variance in political trust can be attributed to the country-level. In Model I, individual variables are included revealing a substantial effect of economic evaluations on political trust. Having voted for the party in government also proves to be an important determinant of political trust. The effects of the socio-demographic variables are rather modest. In a next step, we add the disproportionality-index to the model. The coefficient proved to be not significant, however, when modelling the relation in a curvilinear way, the coefficient reaches statistical significance and the effect is substantial providing evidence for both hypotheses: political trust seems to be high both in countries with very proportional election outcomes and in countries with very disproportional election outcomes. These two ideal-types of electoral rules are based on different views on the essence of democratic government and consequently produce substantially different degrees of accountability and representativeness. This curvilinear effect seems to suggest that institutions fostering inclusiveness or accountability can both increase levels of political trust. In this line of reasoning, it seems that taking a middle position (being somewhat inclusive and somewhat accountable) generates low levels of trust. Moreover, almost half of the remaining variance 12

13 between countries can be explained by adding the disproportionality-index to the model. This effect remains firmly in place when GDP/cap, democratic experience or good governance indicators are included as controls. Table 3. Political Trust and Disproportionality: A Multilevel Analysis Political trust Model 0 Model I Model II Model III Model IV Model V Individual level-variables Gender (male=1) *** *** *** *** *** Education 0.067*** 0.067*** 0.067*** 0.067*** 0.067*** Age 0.027*** 0.027*** 0.027*** 0.027*** 0.027*** Age *** 0.049*** 0.049*** 0.049*** 0.049*** Satisfaction with economy 0.387*** 0.387*** 0.387*** 0.387*** 0.387*** Winning/Losing 0.117*** 0.117*** 0.117*** 0.117*** 0.117*** Country level-variables Disproportionality (LT) *** ** ** ** Disproportionality 2 (LT) 0.410*** 0.326** 0.293* 0.300** Good governance Old regime (>20 years) GDP/capita (in US $) 0.104* σ²(e) σ²(u 0 ) 9.26*** 3.82*** 1.97*** 1.82*** 1.71*** 1.65 Intra-class correlation 21.13% 11.82% 6.46% 6.02% 5.68% 5.47% Number of cases 58,232 58,232 58,232 58,232 58,232 58,232 IGLS Deviance 371, , , , , , Source: European Social Survey ( ; wave 3-4.) Notes: Dependent variable is political trust. Entries are standardized parameter estimates of a multilevel regression. Sign.: p<0.001:***; p<0.01:**; p<.05:*, p<.1. What features of (dis)proportionality influence political trust? In this paper we do not merely want to investigate which electoral rules foster political trust, we also want to explore some of the explanations that have been advanced for this. Below, we briefly outline how these propositions will be tested. The explanations itself were outlined in the theoretical part. A first explanation focuses on the number of choices that are presented to the electorate at elections. We will explore this explanation by including a variable in the previous analysis measuring the number of effective parties contesting elections during the last election before the survey was administered based upon Laakso & Taagpera (1979). viii A second explanation outlined in the theoretical part of the paper, focuses on the inclusiveness of the legislature which will be operationalized by the number of effective parties gaining parliamentary representation as a result of the last election before the survey was administered ix. A third explanation relates to the effects of coalition governments on political trust. This proposition will be tested by including the number of parties in office at the moment of the fieldwork. x We also dichotomized this variable, in 13

14 order to see whether the effect is linear or is mostly due to the occurrence of coalition governments or single-party governments. A fourth explanation focuses on the difference in winning and losing elections under different electoral rules. While it has been found that winners have more trust in majoritarian systems than in consensual (Anderson &Guillory 1997), losers are found to have more trust in consensual systems than in majoritarian ones. It is however unclear whether this is resulting from the proportionality of the electoral rules or from other features of consensual or majoritarian systems. Therefore, a cross-level interaction between winning/losing and disproportionality will be included. A last explanation relates to the differences in clarity of responsibility under different electoral rules. We model this by adding a cross-level interaction between satisfaction with the economy and disproportionality and between satisfaction with the economy and coalition governments. All these explanations will be tested in the next part by including these additional variables in the previous analyses. That way we can gain more insight into the relation that was found between proportionality and political trust. Below these six explanations are summarized: H2: Political trust is higher in countries with more political parties participating in elections. H3: Political trust is higher in countries with more political parties in parliament. H4: Political trust is lower in countries with more parties in office. H5: In proportional systems losers political trust is higher than in disproportional systems. H6: In disproportional systems citizens satisfaction with the economy has a stronger effect on political trust than in more proportional systems. On average five effective parties contested elections in the 23 countries under study. In Hungary and Spain voters had the least choice with a mean of 2.9 effective parties. Israeli and Belgian voters, on the other hand, had the most choice with 9 effective parties contesting elections. On average, four parties managed to win seats in the sample. In France, UK, Hungary and Spain few parties gained parliamentary representation (about 2.5 effective parties). While Belgium and Israel had the most parties in parliament, more than 7 effective 14

15 political parties gained seats. While in France quite some parties contested election, only few succeeded to win seats (respectively 4.8 and 2.4 effective parties). During the fieldwork (2006-9), governments were constituted of one to five political parties. During the third wave (2006-7), four countries were governed by a single party while a coalition government was in office in 17 countries. During the fourth wave (2008-9), six countries were governed by a single party while a coalition government was in office in 15 countries. Disentangling the relation: Multilevel analysis We build on the previous models, including the same individual controls but adding some additional variables to the analysis. First, we look at the relation between political trust and the (effective) number of parties participating in elections. The evidence points at a curvilinear relation depicting that political trust is higher in countries in which citizens are offered more (party) choices, but there exists a cut-of point suggesting that party fragmentation decreases political trust. In the next model, the number of parties gaining parliamentary representation is added showing a similar effect: systems with more parties in parliament have higher political trust levels but again there is a cut-of point. In a last step, we look at the relation between political trust and the parties in government. Hypotheses H4 can not be confirmed, no relation was found between the number of parties in government and political trust nor between countries with a coalition government and countries with singleparty governments and political trust. In sum, political trust was found to be higher when several choices were presented to the electorate, but not too much, and when parliament was more inclusive but not too fragmented. The effect coalition governments have on political trust is less clear, no relation could be found. 15

16 Table 4. Disentangling the Relation Between Political Trust and Disproportionality Political trust Model VI Model VII Model VIII Model IX Individual level-variables Gender (male=1) *** *** *** *** Education 0.067*** 0.067*** 0.067*** 0.067*** Age 0.027*** 0.027*** 0.027*** 0.027*** Age *** 0.049*** 0.049*** 0.049*** Satisfaction with economy 0.387*** 0.387*** 0.387*** 0.387*** Winning/Losing 0.117*** 0.117*** 0.117*** 0.117*** Country level-variables Old regime (>20 years) 0.213*** 0.211*** 0.184** 0.188** Effective number of parties: elections Effective number of parties: elections * Effective number of parties: seats Effective number of parties: seats * Parties in government Coalition (1= yes) σ²(e) σ²(u 0 ) Intra-class correlation 6.12% 6.17% 7.72% 7.57% Number of cases 58,232 58,232 58,232 58,232 IGLS Deviance 360, , , , Source: European Social Survey ( ; wave 3-4.) Notes: Dependent variable is political trust. Entries are standardized parameter estimates of a multilevel regression. Sign.: p<0.001:***; p<0.01:**; p<.05:*, p<.1. In a last analysis, we will test the two other explanations which are frequently used to explain different levels of political trust under different electoral systems namely: (H5) the differential effect of winning and losing under different electoral rules and (H6) the differences in clarity of responsibility under different electoral rules. We model this by adding cross-level interactions between satisfaction with the economy and winning and losing, on the one hand, and disproportionality and coalition governments on the other hand. Only the coefficients of the interaction effects are reported below, the full models can be found in appendix. First, we tested whether the effect of winning and losing on political trust differed between the different countries. While the slope of winning/losing did vary significantly between the countries, the difference in electoral rules could not explain this variance. 16

17 Democratic experience seems to be more important than electoral rules to explain this crossnational variance: losers trust is higher in established democracies than in new democracies (row 4 of Table 5). Second, clarity of responsibility is often cited as an explanation for differences in political trust under different electoral rules. Indeed, the relation between satisfaction with economy and political trust is stronger when one party is in office than when different parties govern together and responsibility is more diffuse. Table 5. Disentangling the Relation Between Political Trust and Disproportionality Coefficient Slope IGLS P-value (S.E) winner/loser Deviance (0.06***) Disproportionality * winner (0.026) *** 360, Parties in parliament * winner (0.075) *** 360, Coalition * winner *** 360, Old democracy * winner (0.276) ** (0.197) Coefficient (S.E) *** 360, P-value Slope satisfaction economy (0.28***) IGLS Deviance 0.011ns *** 360, Disprop. * satisfaction eco. (0.010) * *** 360, Coalition * satisfaction eco. (0.106) Source: European Social Survey ( ; wave 3-4.) Notes: Dependent variable is political trust. Entries are parameter estimates and standard errors (between brackets) of a multilevel regression. Sign.: p<0.001:***; p<0.01:**; p<.05:*, p<.1. N= 58,

18 Conclusion In this paper, we have looked at institutional features to explain cross-national differences in levels of political trust. In particular, we looked at the proportionality of election outcomes. In his book Comparative Constitutional Design Sartori argued: Perhaps the advocates of plurality-pr hybrids believe that they are bringing together the best of two worlds; but they are likely to obtain, instead, a bastard-producing hybrid which combines their defects. (Sartori 1994, 75). The evidence in this case suggest Sartori s claim is correct: political trust is the highest in countries practicing one of both ideal-types of electoral rules. This finding could explain why empirical research has yielded such mixed and weak results on this relationship. Although proportionality is an age-old issue, there is still no consensus on its effects on citizens attitudes. Despite the clear theory, several studies could not find any relation between political trust and proportionality (Listhaug et al. 2009; Norris 2010; Wagner 2009). If citizens trust is highest within inclusive and within accountable electoral systems, modeling the relation between proportionality and political trust in a linear way dismisses its effects. Additional evidence needs to be marshaled in order to test whether this relation can also be found in other datasets. Next to the nature of the relation, also the use of appropriate measurements is crucial. When dividing electoral systems into three categories, the variation within these systems is completely ignored. Using indices of proportionality seems a better way to proceed. It is also quite striking that theories consistently argue that citizens accumulated experience with electoral outcomes matters but when looking at the relation between election outcomes and citizen attitudes only the last election outcome is taken into account. This paper has looked at the proportionality of a country s election outcomes over the last ten years which gives a better idea of citizens accumulated experience with election outcomes. Additional tests (not included in the paper) show that a country s average proportionality of election outcomes is indeed a better predictor of political trust than the proportionality of the last election outcome. Further, we tested several explanations that have been advanced to explain the effects of electoral rules on citizens attitudes. A first argument relates to the inclusive nature of proportional rules which encourages minor political parties to enter the electoral contest. Under proportional electoral rules, citizens can effectively use their party choice to voice their political preferences. Political discontent can also be voiced by voting on minor protest parties. In addition, these parties can effectively pass the threshold to parliamentary representation and defend the interests of their voters better than large catch-all parties. The evidence in this case supports these claims: citizens have higher trust levels in countries with 18

19 more parties contesting elections and more inclusive legislatures. However, party fragmentation does not foster trust as it becomes increasingly difficult to aggregate interests into ideological coalitions. In effect, moderation in all things! This does not need to contradict the first finding. Proportional electoral rules foster indeed multi-party systems but this relation is not perfectly linear: proportional rules do not automatically lead to party fragmentation. Other factors, e.g. democratic establishment are influencing a country s party system too. Surprisingly, there is no clear relation between political trust and coalition governments or single-party governments despite the strong arguments on single-governments effectiveness, coalition governments instability or coalitions inclusiveness. No relation was found between political trust and the number of parties in office nor between political trust and the dummy variable coalition government. However, other empirical studies also failed to find a relation between governments effectiveness and electoral system design (Armingeon 2002). Possibly, other important aspects are not captured in this variable. One aspect worth considering might be the switches in coalition composition. While a new coalition or frequent rotations of coalitions parties might foster trust, semi-permanent coalition might result in feelings of resignation and alienation. The same argument relates to the effects of electoral rules on winning and losing elections. While winners were significantly more trusting than losers and cross-national differences in this relation were found, electoral rules failed to explain this cross-national variation. Winning was operationalized by voting on one of the parties in office on the national level but in countries with federal state-structures there are additional governments which can have a different composition, and different winners. This argument can also be made for presidential systems or other kinds of checks and balances systems. Federal state structures, however, do not seem to influence political trust directly, nor as an intermediary variable between winning/losing and trust. Another explanation could be that electoral rules are not influencing winners and losers trust all that much. Losing elections is a normal part of the democratic process and it should not affect trust in political institutions. In newer democracies lacking repeated experiences with democratic experiences, however, losers are likely to be distrustful. Therefore, democratic experience can be argued to be more important than electoral rules. Lastly, we looked at the accountability argument finding support that responsibility is more clear under single-party governments. Evaluation of the economy determine citizens support more when only one party is in office, this can benefit parties but can of course also harm them. 19

20 REFERENCES Aarts, K. and Thomassen, J. (2008). Satisfaction With Democracy: Do Institutions Matter? Electoral Studies, 27(1), Almond, G. and S. Verba. (1963). The Civic Culture. Political Attitudes and Democracy in Five Nations. Princetown: Princetown University Press. Anderson, C., A. Blais, S. Bowler, T. Donovan, and O. Listhaug. (2005). Losers' Consent. Elections and Democratic Legitimacy. Oxford: Oxford University Press. Anderson, C. and Guillory, C. (1997). Political Institutions and Satisfaction with Democracy: A Cross-National Analysis of Consensus and Majoritarian Systems. American Political Science Review, 91(1), Anderson, C. J. (2000). Economic Voting and Political Context: a Comparative Perspective. Electoral Studies, 19(2-3), Armingeon, K. (2002). The Effects of Negotiation Democracy: a Comparative Analysis. European Journal of Political Research, 41(1), Banducci, S. A., Donovan, T., and Karp, J. A. (1999). Proportional Representation and Attitudes about Politics: Results from New Zealand. Electoral Studies, 18(4), Bok, D. (1997). Measuring the Performance of Government. In: Nye, Joseph S, Philip D Zelikow, and David C King (Eds.). Why People Don't Trust Government. Cambridge: Harvard University Press. Criado, H. and Herreros, F. (2007). Political Support Taking Into Account the Institutional Context. Comparative Political Studies, 40(12), Dalton, R. (2004). Democratic Challenges, Democratic Choices. The Erosion of Political Support in Advanced Industrial Democracies. Oxford: Oxford University Press. Elazar, D.J. (1997) Contrasting Unitary and Federal Systems. British Journal of Political Science, European Social Survey Round 3 Data (2006). Data file edition 2.0. Norwegian Social Science Data Services, Norway - Data Archive and distributor of ESS data. [ESS data: European Social Survey (2007). ESS Documentation Report. Edition 2.0. Bergen, European Social Survey Data Archive, Norwegian Social Science Data Services. European Social Survey Round 4 Data (2008). Data file edition 2.0. Norwegian Social Science Data Services, Norway - Data Archive and distributor of ESS data. [ESS data: European Social Survey (2009). ESS Documentation Report. Edition 2.0. Bergen, European Social Survey Data Archive, Norwegian Social Science Data Services. Farrell, D. (2001). Electoral Systems : A Comparative Introduction. Basingstoke: Palgrave. Farrell, D. M. and Mcallister, I. (2006). Voter Satisfaction and Electoral Systems: Does Preferential Voting in Candidate-Centred Systems Make a Difference? European Journal of Political Research, 45(5), Gallagher, M (1991) Proportionality, disproportionality and electoral systems, Electoral Studies, 10(1), pp Gallagher, M. and Mitchell, P. (eds, 2008), The Politics of Electoral Systems. Oxford and New York: Oxford University Press. Halman, L. (2007). Political values. In R. Dalton & H.-D. Klingemann (eds.), The Oxford handbook of political behaviour. Oxford: Oxford University Press, pp Hox, J. (2002). Multilevel Analysis. London: Lawrence Erlbaum Associates Publishers. Inglehart, R. (1999). Postmodernization Erodes Respect for Authority, but Increases Support for Democracy. In: Norris, Pippa (Eds.). Critical Citizens. Global Support for Democratic Government. Oxford: Oxford University Press. 20