Political confidence in new and established democracies in Europe: Individual and contextual bases of confidence in parliamentary and regulatory institutions Paper presented at the ECPR Joint Sessions in Münster, Germany 22 27 March 2010 Workshop 19: The Interrelationship between Institutional Performance and Political Support in Europe: Discussing Causes and Consequences Workshop Directors Sonja Zmerli Goethe-Universität Frankfurt am Main Marc Hooghe KU Leuven (Belgium) Daniela Braun (M.A.) University of Stuttgart PhD candidate (German-Franco Graduate School Comparing democratic societies in Europe ) Mannheim Centre for European Research Research assistant
Abstract The main focus of the paper is to investigate individual-level and contextual-level sources of political confidence in new democracies in Europe: Central and Eastern European countries as well as Southern European countries. Since the literature emphasizes two major types of institutions, we analyze citizens confidence in regulatory and representative institutions using data from the forth wave of the World Values Survey (WVS). The paper supports the results of previous research on that topic showing that arguments derived from institutionalism explain the major part of political confidence while cultural factors have only little impact. Since the level of political confidence differs between the countries under consideration, the paper seeks to identify, additionally, mechanisms related to their specific democratization experiences. Using linear hierarchical modelling (HLM) in order to simultaneously estimate individual and contextual bases of political confidence various indicators derived from democratic consolidation theories are tested. The final results show that political confidence can be only explained to a small degree by macro-indicators. Nevertheless, the tested democratization experiences of the countries under consideration have effects on political confidence especially on confidence in the parliament while confidence in regulatory institutions is, if anything, influenced by the economic development of the country. Content 1. Introduction p. 3 2. Theoretical Framework p. 4 3. Confidence in regulatory and representative institutions p. 7 4. Individual-level sources of political confidence p. 9 5. Testing competing theories: Culture or performance? p. 12 6. Individual political confidence in a multi-level-perspective p. 15 7. Conclusion p. 20 8. Literature p. 22 Appendix 1: Figures and Tables p. 26 Appendix 2: Individual-level und aggregate-level Indicators p. 32
1. Introduction More than a half of the current 27 member states of the European Union (EU) are shaped by non-democratic traditions 1. Thus, the EU consists of more member states with nondemocratic background than of countries with pure democratic traditions. With regard to future enlargements of the EU, countries with a non-democratic past will be more strongly represented: Croatia and Turkey as recognised candidates for membership, but also the Western Balkans states are in the front line for becoming full member states of the EU. Though not in the near future, one could expect the application of the former Soviet Union republics, for example Belarus, the Ukraine, or Moldavia. Hence, an interesting question is in what way these non-democratic traditions affect the stability of the new democratic systems and therefore the future development of the EU. Scholars of historical institutionalism would argue that the autocratic past can have a strong impact on democratic institution-building. However, in addition to political institutions, democracy requires a distinctive set of political values and orientations, or as the authors of the first political culture study put it: What must be learned about democracy is a matter of attitude and feeling, and this is hard to learn (Almond/Verba 1965: 4). In this study, we are primarily interested in attitudes of people who are living in new democracies, more specifically the confidence 2 they put in their political institutions. After changes from autocratic to democratic regimes, people have to adapt their lives to the changed political sphere and learn democratic rules learn to trust their recently established democratic institutions as Mishler/Rose (2002: 5) argue: [ ] citizens in new democracies have spent most of their lives under an undemocratic regime; therefore, the new regime challenges people to relearn political attitudes and behaviour. Since the purpose of this paper is to analyze political confidence against the background of different democratization experiences, countries under consideration are new democracies with and without EU membership. 3 The low levels of political confidence in new European democracies easily reveal the difficulty of building trust in both the parliament (cf. Figure 3) and regulatory institutions (cf. Figure 4): While confidence in most established democracies 4 1 In the early 20 th century the austro-fascist regime in Austria, the Nazi regime in Germany, Mussolini s Italy, and also Vichy in France were prevalent. After WWII Europe had to struggle with two different non-democratic system types: on the one hand authoritarian regimes in Southern Europe, on the other hand totalitarian (or in the later period post-totalitarian communist) regimes in Central and Eastern European countries. 2 The use of confidence and trust throughout the trust/confidence-research is not consistent. Though there is a tendency of using trust generally in order to describe attitudes towards political actors and confidence towards institutions we will use the two terms as synonyms. 3 Southern Europe: Greece, Portugal, Spain (two samples: 1999 and 2000). CEE (EU members): Bulgaria, Czech Republic, Estonia, Germany East, Hungary, Latvia, Lithuania, Poland, Romania, Slovakia, Slovenia. Eastern European Countries (non-eu members): Belarus, Croatia, Ukraine, Russian Federation. 4 Levels of political confidence are presented additionally for established democracies in order to highlight the major differences. 3
is higher, the value in new democracies is clearly lower than the European mean 5. In the most recently established democracies in Central- and Eastern Europe (CEE) as well as in the failed states or weak democracies Russia and Belarus especially low levels of confidence are prevalent. In contrast, Southern European countries, where regime changes took place in the 1970s, have different levels of confidence: While Spain and Portugal are more similar to established Western European established democracies Greeks have only little confidence in their respective political institutions. Having identified different levels of political confidence in European countries, this paper seeks to investigate its sources with a special focus on the mentioned democratization issues. The paper will be divided into three different sections: Firstly, we will give some insights into conceptual and theoretical aspects of political confidence and link it to prior empirical work. We will especially draw on two competing sources of political confidence: culture based vs. institution based explanations. In the following section the individual bases of political confidence will be tested empirically using data from the forth wave of the World Values Survey (WVS). Finally, we will focus on the macro level: Since the level of political confidence differs strongly between the different countries under consideration, the contextual level sources of confidence will be analysed. In contrast to conservative macrolevel factors as for example cultural (Huntington 1991) or political indicators (Criado/Herreros 2007; Fuchs/Klingemann 2006; Magalhaẽs 2006), we will focus on different democratization experiences. 2. Theoretical Framework Since political confidence is clearly linked to the concept of trust, we will first define the general concept and differentiate one from another. In a broader sense trust is, compared to behaviour, conceived as a type of attitude. It can be defined as positive orientation of people towards objects; it is based on specific standards and expectations (Gabriel et al. 2002: 52). According to Levi (2001) three different types of trust can be distinguished: 1. Personal trust as a matter of relationship between two or more but definitly concrete persons. 2. Interpersonal trust, also called social trust or generalized trust, defined as trust towards a broader group of people, e.g. towards neighbours, friends, or strangers and respectivly towards large groups like people of the same nationality. Social trust as an important feature of social capital (Putnam 1993; 2000; 2001) is a kind of social relationship based on prior and future experiences as well as the respect of values and norms. 5 European mean confidence in the national parliament is at 1,80 in regulatory institutions at 1,58 (using a scale from 0 high confidence to 3 no confidence at all). 4
3. Political trust or confidence, or rather trust or confidence towards institutions or organizations in general. However, the differentiation between social trust and political confidence is less clear than between personal trust and political confidence. According to Gabriel (2002) political confidence can even be defined as a special case of social trust. The only difference is the orientation towards institutions (and respectively the general idea the institution represents) in the case of political confidence, while social trust is orientated towards people (Gabriel 1999: 202). Thus, politcal confidence does not apply to horizontal relationships between people, but vertical relationships between citizens of a state and political authorities or institutions and is therefore based on systems, rules, and procedures (Newton 2008). Furthermore political confidence can be orientated towards either political authorities or genuine political institutions: Confidence in authorities means that they are perceived as the group s agents, that the group members identify with them [ ]. Confidence in the political institutions means that the group believes either that these institutions produce authorities who are its agents or else produce favorable decisions regardless of the particular incumbents (Gamson 1986: 54). Since confidence in political authorities is rather affected by specific political events or personal scandals we will focus on confidence in institutions. 6 In addition to the above explained broader differentiation between confidence towards political authorities or political institutions, the latter can be further subdivided. Political institutions have different roles or functions within a state or society: Police, for instance, maintains order in the society through controlling if people stick to the rules established in a society while justice penalizes the violation of these rules. On the contrary, institutions e.g. political parties, the parliament or the government are involved solely in political issues. Since one could assume that confidence in institutions is driven by different mechanisms, scholars distinguish mainly between these regulatory institutions on the one hand and institutions of the liberal democracy on the other (cf. e.g. Denters et al. 2007). One can assume that citizens distrust the government, the parliament or parties because they are involved in daily political affairs whereas they place more confidence in the police an institution that is more in touch with the needs of citizens. 7 Despite its importance for democracy and the pronounced research interest, a theory for political confidence suitable for empirical work has not yet been developed (Hibbing/Theiss- Morse 1995; Bernstein 2001; Gabriel 2002). So far, most studies on confidence in political institutions referred to general approaches relative to the legitimacy of political systems. These extensive approaches end frequently in confounding analyses of political trust 6 Recall for example the Lewinsky scandal : People placed less trust in the president because of his personal behavior and not because of his role as president of the United States. 7 The issue of different objects of confidence will be treated more extensively in the following chapter. 5
(Owen/Dennis 2001: 209). The most commonly used approach for studies on political confidence is Easton s theory of support for political systems (1965; 1975), which has been developed primarily in order to understand macro phenomena, but was later also used to answer individual level research questions. Critical examination of Easton s approach, done for example by Fuchs (1989), Westle (1989), or Norris (1999), sharpened the concept, however for the limited purpose of studying political confidence the political support concept seems too widespread. Since an appropriate theory is not available, we will first emphasize the effective relevance of political confidence for democratic systems. The widespread argument for its importance is the commonplace view that without commitments by citizens, government cannot gain obedience from citizens (Hardin 1989: 10). According to this, one can infer that political confidence ensures the functioning of democratic institutions (Chanley et al. 2001), reduces transaction costs (Braithwaite/Levi 1998), and the necessity to justify every decision made (Tyler 1998). Furthermore, institutions like the parliament or parties represent one of the most important links between citizens and governors (Norris 1999). Especially in new democracies political confidence is eminently necessary, but at the same time one of the critical issues. Most citizens in CEE are generally shaped by a climate of distrust (Diamond 2002; Marková 2004). Since they lived their entire lives under nondemocratic regimes they have a good reason to distrust political and social institutions (Mishler/Rose 1997: 419), and have to learn trusting after a system change. In addition, trust in political institutions is based on the fundamental knowledge about democracy, but citizens have no experience in democracy and its procedures. Therefore, political confidence in new democracies is exclusively orientated towards the future (e.g. hope for a better life in the future democracy), and that makes it unstable and risky. One can be easily aware of possible constraints for the system stability during the process of consolidation. More recent studies on political confidence deal with two dominating approaches (cf. Braithwaite 1998; Bianco 1998; Hardin 1998; Levi 1998; Tyler 1998; Uslaner 2002): On the one hand, the rational or cognitive approach, on the other hand the affective, emotional, thus non-cognitive perspective. Indeed, trust or confidence consists of two components: rationality and emotion. Rational or cognitive trust is based on knowledge about preceding performance; trust is responsive to beliefs about the trusted and to the likely outcomes of a trusting relationship. According to Valerie Braithwaite (1998) we characterize the two dimensions as exchange trust and communal trust. The latter is based on emotions towards the object likely to be trusted. These emotions emerge from common shared values and norms, demonstrated also by social connectedness, group loyalties and common identities. While exchange trust is frequently described by the pattern A trusts B to do X, A 6
trusts B (more generally) is true for communal trust. Both trust components refer also to different levels of stability. Since linked to the specific experience with the object of trust, exchange trust is not that stable, whereas the broader communal trust is enduring (cf. Braithwaite 1998; Tyler 1998 and 2001; Uslaner 2002). However, it should be mentioned that the two dimensions of trust are not exclusive but necessary for balancing eventual mutual shortcomings (cf. Braithwaite 1998; see also: Tyler 1998). Moreover, they can be linked to the sources of trust which have been identified so far in empirical research. 8 3. Confidence in regulatory and representative institutions Original research on political confidence (mostly in the U.S.) was primarily interested in trust in government (cf. Citrin 1974; Miller 1974). Later, together with an European approach to that topic, the study focus was on institutions of the representative party democracy (parliament, parties), but also on political institutions characterized by their regulatory tasks, such as the justice or the police (e.g. Denters et al. 2007; Gabriel 2008; Zmerli 2004; Zmerli/Newton 2008). In a democratic state, both types of institutions, the regulatory and the party democracy institutions, are relevant. Although the latter ones are important for the political procedures, citizens of the state could even assign more importance to the regulatory institutions (cf. Gabriel et al. 2002: 192). Beside the mentioned political institutions, scholars tend to include in their analyses, institutions of the civil society, such as churches, labour unions, as well as the press, and international actors (e.g. Mishler/Rose 1997; 2001; 2005). The question most authors at present deal with is whether people are really able to discern those institutions from one another. Numerous analyses, mostly based on data in established democracies, showed, that people differentiate primarily between regulatory institutions on the one hand, and institutions of the representative party democracy on the other (Gabriel 1999; Gabriel/Kunz 2002). Theoretical arguments and empirical evidence pointed even to the necessity of distinguishing between three objects of political confidence (cf. Denters et al. 2007): Actors in a representative party democracy (political parties and politicians), institutions in a liberal democracy (parliament and cabinet), and the institutions of the Rechtsstaat. However, some empirical analyses showed that the theoretically expected differentiation between regulatory institutions and institutions of the representative party democracy is not given (Listhaug/Wiberg 1995; Mishler/Rose 1997; 2001; 2005; Rohrschneider/Schmitt-Beck 2003; Zmerli 2004.; Zmerli/Newton 2008). Mishler/Rose (1997), for instance, analysed political confidence in recently established democracies in CEE and found that people tend to evaluate institutions along what is fundamentally a single continuum (433). Following Mishler/Rose (1979; 2001, 8 This issue will be treated more extensively in Chapter 4 (Individual-level sources of political confidence). 7
2005) we would expect thus that especially citizens of recently established democracies are not able to differentiate between the major institutions of the state, with which they have so little familiarity or experience. In a nutshell, the direction seems to be blurry: In the context of cross-national analyses in established democracies but also in the framework of comparisons between East and West Germany, evidence for the need of differentiation was provided, whereas particularly research in CEE pointed to the reverse. Since neither theoretical argumentation, nor empirical evidence indicated non-ambiguous results on that topic we will run principal component analyses using data from the forth wave (1999-2004) of the WVS. 9 It disposes of six (relevant) indicators of confidence in national institutions: confidence in the parliament, the civil services, the police, the justice system, labour unions, and the press. According to the discussion of theoretical arguments and empirical evidence one could expect the extraction of the following three factors: 1. Institutions of the representative party democracy (parliament, civil services) 10, 2. Regulatory institutions (police, justice system, civil services 11 ), and 3. Institutions of the civil society (labour unions, press). 12 Table 1 presents the results of a principal component analysis 13 of pooled data as well as grouped data because of major differences between the country groups. Since citizens in new democracies have difficulty making fine-grained distinctions about institutions with which they have so little familiarity or experience (Mishler/Rose 2001: 43), we distinguish between three country groups according to their differing democratization experiences: 1. Southern European Countries: Greece, Portugal, and Spain. 2. Central and Eastern European Countries (EU members): Bulgaria, Czech Republic, Estonia, Germany East, Hungary, Latvia, Lithuania, Poland, Romania, Slovakia, and Slovenia. 9 For all pooled data analyses, data has been sampled (N=1000), while preserving the internal proportions as originally provided by participants (cf. http://www.worldvaluessurvey.org). 10 Unfortunately in the forth wave of the WVS confidence in the government and in political parties two nonambiguous institutions of the liberal party democracy have not been asked. These indicators are only available in the third wave of the WVS which has not been used due to the small number of (relevant) countries in particular for the multi-level-approach of this paper a high number of cases is desirable. 11 Neither theoretical arguments, nor empirical evidence show unequivocally if civil services belong to the institutions of the representative party democracy or to the group of regulatory institutions. However, most scholars agree that it is affiliated rather to the latter group of institutions. 12 Although institutions of the civil society won t be analyzed in this paper, they have been included in the principal component analysis in order to illustrate that citizens clearly distinguish between different types of institutions. 13 Principle component analysis has been run first without presetting a specific number of factors, resulting one single factor for each country group. However, since theoretical arguments are available, the number of factors was preset to three. 8
3. Eastern European Countries (non-eu members): Belarus, Croatia, Ukraine, and the Russian Federation. While the results with pooled data clearly indicate the predominance of one factor for purely political institutions, analyses per country group reveal a different pattern. With the exception of Eastern Europe s unstable democracies citizens of all country groups seem to differentiate between institutions of the representative party democracy, regulatory institutions, and institutions of the civil society. 14 Solely citizens living in unstable democracies do not distinguish between institutions of the representative party democracy and regulatory institutions. The latter result is less surprising because at least in Russia, Belarus and the Ukraine, where democratization has not taken place in the same way as in CEE, institutions are indeed strongly tied to each of the more or less authoritarian leaders. To sum up, the results show that the citizens of most countries assess their institutions quite realistically, even in the consolidating periods of democratization. Although Mishler/Rose (1997; 2001; 2005) argue that citizens of new democracies wouldn t distinguish between institutions, they mention that [o]ver time, as citizens acquire more experience with the institutions of state and society, the weak and inchoate distinctions they currently draw between civil and political institutions may crystallize and emerge with greater clarity and strength (Mishler/Rose 1997: 433). The presented results seem to confirm this assumption: After having spent more than ten years in democratic countries, citizens in CEE have finally learned to distinguish between their political institutions. Therefore, we deal with two 15 different dependent variables 16 in this paper. Confidence in the parliament and confidence in regulatory institutions (police, justice system, civil services). 17 4. Individual-level sources of political confidence Generally two competing sources of political confidence can be identified: The political culture tradition competes with institutional arguments. However, most recent publications tend to link the two types of explanations (Rohrschneider 1999; Mishler/Rose 2001; Pollack et al. 2003; Campbell 2004; Torcal/Montero 2006; Denters et al. 2007; Zmerli et al. 2007). 14 Results in Southern European countries are not that clear: Police is rather perceived as pertaining to factor 1 (together with the parliament and civil services). However, analysis using data from the third wave of the WVS show that confidence in the government, the parliament, and in parties are clearly loading on a single factor, while confidence in civil services, the justice system, and the police are perceived as a separate one. Therefore, it can be argued that people in Southern Europe, generally, differentiate between the two indicators. 15 Since we are primarily interested in the basic political institutions we will abstain from analyzing confidence in institutions of the civil society. 16 The question has been asked as follows: I am going to name a number of organisations. For each one, could you tell me how much confidence you have in them: is it a great deal of confidence, quite a lot of confidence, not very much confidence or none at all?. [1 to 4 (1 'A great deal'; 2 'Quite a lot'; 3 'Not very much'; 4 'None at all')]. For better interpretation, the variable has been recoded to 0 a great deal to 3 none at all. 17 An additive index scaling from 0 a great deal to 3 none at all has been constructed. If one out of three variables holds missing values, the mean value of the two remaining variables was taken. 9
Cultural argumentations follow the seminal Civic Culture Study (Almond/Verba 1965), the redefinitions by Inglehart s studies on the value shift in modern societies (Inglehart 1971; 1977; 1990; 1997; Welzel/Inglehart 2005), and most recently by Putnam s theory of social capital (Putnam 1993; 2000; 2001). Hence, advocates of cultural arguments are most basically interested in political culture, i.e. in general socialization issues, values and norms, as well as social trust. Primary as well as political socialisation in democratic regimes is considered to have positive effects on political confidence (Mishler/Rose 1999). The same applies to sharing basic democratic values and norms, as well as social trust, which can be acquired within social networks (Putnam 1993; 2000; 2001). Since post-materialist values are generally strongly correlated with critical attitudes towards politics, the argument is that post-materialist value orientations have a negative impact on political confidence (Gabriel 1986). On the other hand, explanations of political confidence are strongly related to political economy, thus, inspired by an economic point of view towards social relationships (Coleman 1995; Esser 1999). In the framework of this argumentation people trust their institutions following the principles of rational choice by calculating their cost and benefits. Put simply, this line of argument conceives that institutional trust is the positive feedback that results from citizens perceptions of and expectations about institutional performance (Campbell 2004). Research on that topic showed that multiple indicators have strong significant effects on political confidence: satisfaction with political and economic performance, political responsiveness boosts political confidence as well as the fact that the preferred party (the so called home-team) is represented in the government (Campbell 2004; Catterberg/Moreno 2005; Denters et al. 2007; Miller/Listhaug 1999; Mishler/Rose 2001; 2005; Newton 2008; Rohrschneider/Schmitt-Beck 2003). To sum up, a lot of the mechanisms of political confidence are known: confidence is primarily explained by institutional or rational indicators and only to a smaller degree by cultural explanations. The question this paper seeks to answer, however, is whether these explanations show similar effects at different stages of democratization. Therefore three country groups, each one at a different step of democratization are under examination: 1. Southern European countries with the regime change in the 1970s scholars agree that these countries can be considered as fully consolidated at the time of the survey (Linz/Stepan 1996). 2. Central and Eastern European countries with EU-membership whose regime change took part in 1989-91. Although, from a minimalist point of view they can be considered as consolidated regimes, maximalistic advocates argue that the attitudinal 10
dimension needs to be fully developed before consolidation is completed (Diamond 2002). 3. Eastern European countries without EU-membership cannot be considered as consolidated at the time of the survey, but rather as unstable democracies (Croatia) or defective democracies (Russia, Belarus, Ukraine). Figure 1: Individual level research design Since shortly after a system change honeymoon effects higher levels of trust have been identified (Catterberg/Moreno 2006; Inglehart/Catterberg 2002), one can assume that cultural explanations prevail in the first period of democratization. 18 Only after this first honeymoon period citizens will realize that political as well as economic changes will take more time and efforts than expected and consequently direct the attention to the system s performance (cf. Gabriel/Zmerli 2006). While during this period rational explanations will have more predictive power on political confidence than before, the subsequent time span will be characterized again by cultural explanations. Surprisingly, however, the theoretical conceptualization of trust described above, has been rarely connected to the sources of trust the only exceptions are Gabriel/Zmerli (2006) Gabriel/Walter-Rogg (2008). The two dimensions of trust, community-based trust and exchange-based trust (Braithwaite 1998), can be linked to the two competing explanations of trust. Institutional argumentations are driven by top-down-procedures explaining trust through theories of exchange. In contrast, cultural theories seek to explain trust by shared values, norms, and social relationships taking somehow a bottom-up-perspective which is meant by 18 Positive attitudes towards the new regime originate through rejection of the non-democratic regime and the idealized democracy s image. 11
community-based trust (cf. also Campbell 2004). Figure 1 illustrates the top-down and bottom-up mechanisms of individual-level explanations of political confidence in combination with other considerations the following hypothesis on the individual-level can thus be derived. Hypothesis 1: After a successfully accomplished consolidation of democracy, community-based trust will be rather prevalent. In Southern European countries (where the system change has been taken place in 1974/75) community-based trust will be stronger developed than in CEE (EU and non-eu countries). Thus, the predictive power of the cultural model is higher in Southern European countries, whereas the institutional model will be stronger developed CEE countries (where the system change has been taken place in 1990/91). 5. Testing competing theories: Culture or performance? In order to test the hypothesis adequately all individual indicators need be operationalized 19 first. On the one hand, top-down mechanisms will be tested, i.e. the impact of the satisfaction with political 20 or economic performance 21 on political confidence as well as the assumption of home-team 22 effects. On the other hand, bottom-up mechanisms, derived from cultural theories will be analysed, i.e. the impact of socialization (using age as indicator), social capital (social trust 23, norms of reciprocity 24, and community-based values 25 ), and postmaterialist values (using the four-item index provided by WVS) on political confidence. Furthermore prior research showed that socio-demographic indicators as well as political 19 For all complete review of all independent variables on the individual-level see Table 6 (Appendix 2). 20 Question asked in WVS: People have different views about the system for governing this country. Here is a scale for rating how well things are going: 1 means very bad; 10 means very good. Where on this scale would you put the political system as it is today? 21 Since the WVS does not dispose of any other indicator, the following proxy variable has been used: All things considered, how satisfied are you with your life as a whole these days? (1) Dissatisfied (10) Satisfied. 22 The variable has been computed using the question Which party would you vote? 23 Question asked in WVS: Generally speaking, would you say that most people can be trusted or that you need to be very careful in dealing with people? Most people can be trusted (1), can t be too careful (2). 24 Question asked in WVS: Please tell me for each of the following statements whether you think it can always be justified, never be justified, or something in between: Claiming government benefits to which you are not entitled; cheating on taxes if you have a chance; someone accepting a bribe in the course of their duties. 25 Unfortunately, due to a high proportion of missing values in many countries community-based values have been operationalized merely inadequately using only one indicator. WVS disposes of a list of variables concerning major education objectives feeling of responsibility; tolerance and respect for other people und unselfishness (cf. also Gabriel et al. 2002: 71). The following indicator has been used: Here is a list of qualities that children can be encouraged to learn at home. Which, if any, do you consider to be especially important? Please choose up to five. 12
attitudes can have a (minor) impact on political confidence. These variables will be included as control variables 26 into the model. Two different regression models are computed: one for confidence in the parliament as dependent variable, another for confidence in regulatory institutions. The results are presented for the three different country groups described above. First of all, we will have a look on the explained variance of the overall models: Having confidence in the parliament as dependent variable it ranges from 15% in CEE democracies to 12% in Southern European countries. In unstable eastern European democracies the model can be explained by 13%. Eye-catching are the differences between the model fits of the different blocks: The control model has by far the smallest predictive power while the cultural model explains only little, the performance-based model explains the bulk of political confidence. These results are not surprising as they replicate prior work on that topic (Campbell 2004; Catterberg/Moreno 2005; Denters et al. 2007; Miller/Listhaug 1999; Mishler/Rose 2001; 2005; Newton 2008; Rohrschneider/Schmitt-Beck 2003) the only shortcoming is the relatively little explained variance, which can be due to the WVS data. 27 According to the assumption made above we would expect that community-based trust is stronger developed in Southern Europe than in CEE. This hypothesis can be confirmed for the dependent variable confidence in parliament : Though the predictive power of the cultural model is not pretty high in all three country groups, it is definitely higher in Southern European countries (Blocks A+B: R 2 =0,05) than in CEE (Blocks A+B: R 2 =0,01). All indicators that have been introduced as part of the cultural model have significant effects in Southern European countries. Social capital at least norms of reciprocity as well as social trust has a positive impact on confidence in the parliament. Also in line with the assumptions, postmaterialist value holders (compared to materialist orientated people) are less likely to trust the parliament. With regard to the age, older Southern Europeans have more confidence in the parliament than younger ones, what is contradicting to the socialization hypothesis and rather pointing to possible generation effects. Surprisingly, the explained variance of the cultural model in Eastern European Countries (non-eu-members) is little higher than in CEE (with EU-membership). However, the only effects found in the model are due to a very small and slightly significant extent to social trust which has a positive impact on confidence in the 26 While sex, the education level, and religion (self-evaluation as religious person) can be included in the analyses, political attitudes (right-left-self-evaluation, political interest) are not analyzed because questions haven t been asked in various countries. 27 The same models have been computed using data from the European Social Survey (ESS) disposing of better measures for certain indicators (e.g. political confidence is measured by an 11-point rating-scale instead of a 4- point rating-scale; social trust is measured by three different items).using ESS data the same models explain about 30% of the total variance, cultural explanations have a bit more predictive power on political confidence. However, the general proportion between both the cultural and the rational model holds: the latter has most predictive power while cultural argumentations explain only little of the variance. 13
national parliament, and to age. Compared to the youngest cohort, older people (older than 65 years) trust their parliament. 28 Furthermore, we assumed that the institutional model (or respectively the exchange-based trust dimension) will be stronger developed in CEE countries because in the earlier periods of consolidating democracies political confidence is more characterized by top-downmechanisms, i.e. good performance fosters trust. This hypothesis can be confirmed as well: Compared to Southern European countries the predictive power of the institutional model is much higher in CEE (with and without EU-membership). In CEE (with EU-membership) all institutional-based indicators have a significant effect on confidence in the parliament however, in all country groups, respondents who are satisfied with the political performance in their country, put more confidence in the parliament. The assumptions have been tested in the same way for the dependent variable confidence in regulatory institutions : In general, the effects are comparable to the above described model. The predictive power of the cultural model is higher in Southern European countries (Blocks A+B: R 2 =0,09) than in CEE (EU) (Blocks A+B: R 2 =0,02) and CEE (not EU) (Blocks A+B: R 2 =0,04). Only some effects in the cultural models are different: While in the analyses with confidence in the parliament as dependent variable, social capital as well as post-materialism had only minor or insignificant effects in CEE (EU and non-eu membership), norms, social trust, and post-materialist values have a stronger impact on confidence in regulatory institutions. Possibly, cultural indicators are more likely to influence confidence in regulatory institutions than in the parliament. Moreover, control variables, as sex, education, and religion show more significant effects in the regulatory model: Apparently, women, less educated, and religious people (especially in EU-member countries) are more likely to trust regulatory institutions, while education and religion have only significant effects on confidence in the parliament in CEE. Two unexpected effects can be identified in the two models: Firstly, contradicting the hypothesis, values of the society have a small, but negative impact on the confidence in the parliament. With respect to confidence in regulatory institutions the effects would point in the expected direction, but the results have no statistical significance. However, these effects can be due to the partly inappropriate operationalization of the indicator. Secondly, the home-team-hypothesis has not been proved in Southern European countries, but only in CEE. One explanation could be 28 This can be explained by two different arguments: Since these respondents were born in 1935 or earlier, they experienced their socialization during the most pronounced period of totalitarism under Stalin. Therefore the results would confirm general cultural assumptions: Socialization as striking factor for political attitudes has an impact on political confidence. However, we should not forget that the countries under consideration are not fully consolidated, but rather defective democracies with untrustworthy parliaments: Older people placing less confidence in their parliament would be thus right in doing so. In that case, the cultural hypothesis would not be confirmed. 14
that the elections in Spain, Greece, and Portugal have been hold closer at the time of the survey and respondents were less in favour of the preferred party afore. To sum up, the results showed that confidence in the parliament, as well as in regulatory institutions are explained rather by institutional factors than by cultural ones. Accordingly, in general, community-based trust is less developed than exchange-based trust. However, it has been showed that the sources and respectively the dimensions of political confidence in new democracies depend on the different stages of democratization. While just after the transition communal trust prevails, the post-honeymoon-disillusion period (Catterberg/Moreno 2006; Inglehart/Catterberg 2002) is characterized by exchange-based trust. Only after this period trust dimensions are well-balanced, even though exchange-based trust will be always more present. In the next chapter, we will now focus on the implications of different democratization paths on political confidence (cf. also Torcal 2006: 186ff.). 6. Individual political confidence in a multi-level-perspective In addition to individual-level sources, political confidence can be explained by context-level factors. Mishler/Rose (2002) describe these mechanisms for performance-based theories as follows: Macro-measures of regime performance [ ] imply that all citizens will evaluate regime performance similarly. Political support will thus tend to be uniform within a society while varying between societies as political performance varies. By contrast, micro-theories emphasize that individuals will differ in the way the regime s performance affects them (Mishler/Rose 2002: 8). In order to highlight these macro-measures the authors use two different strategies: On the one hand, they compare different regression models one for CEE countries, the other for post-soviet-regimes on the other they introduce two macro-measures ( Increased freedom since 1989 and Mean level of support for return to communist rule ) in each regression model. Most scholars conducting cross-cultural research on political confidence use this comparative strategy (and compare countries or country groups, i.e. samples) in order to identify country-specific effects of political confidence (cf. e.g. Newton 2006; Magalhaẽs 2006; Zmerli 2004; Zmerli et al. 2007). Torcal (2006) investigates the sources of institutional confidence within the framework of the concept of political disaffection in new and established democracies all over the world using in a first step aggregate data. In a second step he estimates OLS-regression models including individual-level indicators as well as macro-level indicators (years of democracy, social trust and education in new democracies, new vs. established democracy, ideology, average number of strikes). Although, most macro-level show significant results, it has to be mentioned that results can be overestimated since the multi-level character of the data has been ignored. 15
In contrast, Gabriel/Walter-Rogg (2008) analyse both individual-based and context-based sources of political confidence throughout Europe using a multi-level-approach. Despite the fact that a bulk of variables on the macro-level has been tested (e.g. years of democracy, corruption, HDI, GDP, cultural fragmentation) some consolidation-specific indicators are still missing. Probably, this could be the reason why [a]lmost all the differences in the average level of political confidence among the nations [ ] can be attributed to compositional effects, but not to truly contextual variables (Gabriel/Walter-Rogg 2008: 238ff.). Criado/Herreros (2007) are primarily interested in the effects of majoritarian democracies compared to proportional democracies on political confidence. Their results demonstrate that [t]hose who did not vote for the incumbent tend to be more supportive [ ] in proportional democratic systems than in majoritarian ones (1528). Though the authors applied the appropriate multilevel-modelling for this kind of research question, the approach should be extended theoretically. Particularly in new democracies, it seems to be useful to include factors related to the consolidation of democracies into the analyses, or as Torcal (2006) puts it: The observed cross-national differences are due to distinct democratization experiences. In new democracies, these democratization experiences are more frequently characterized by decades of convulsive processes of political exclusion for an important part of the population and the presence of very erratic and distinctive mobilization episodes. These problematic democratization processes have left an enduring imprint in many citizens making them very critical of the democratic process, of politics, and of the mechanisms of democratic representation (Torcal 2006: 185). Hence, we are primarily interested in indicators related to different democratization experiences as for example economic or social development after the occurred regime changes. While the breakdown of autocratic regimes and the subsequent first transition period towards democracy are important steps in order to finish autocratic leadership, there are still many tasks that need to be accomplished [ ] before democracy could be considered consolidated (Linz/Stepan 1996: 5). Following general scholars of democratization (Huntington 1991; Linz/Stepan 1996; O Donnell et al. 1986), we will include in particular factors conducive to democratic consolidation: a. Experience with democracy (years of democracy, pre-autocratic democratic experience) b. Economic and social development (HDI) c. Type of pre-democratic system (totalitarian vs. authoritarian) d. International influence (democratic neighbourhood, membership or influence of international organizations) e. Regime change (type of change, continuity vs. discontinuity afterwards) f. Governments competency in solving problems 16
Figure 2: Micro-and macro-level research design Though these indicators can be cited independently, they are strongly related to traditional macro-indicators, such as performance or political variables but also cultural ones. A country s performance the level of corruption and unemployment, GDP, and income equality is subjected to the system transformation and consolidation. Therefore, macroeconomic factors can be used as indicators for the progress during the consolidation period. The same applies for cultural indicators such as a country s cultural heritage or the existence of political, ethnic, or religious minorities, and subsequently a higher level of ethnic fragmentation in a country. 29 Figure 2 illustrates the implications of macro-indicators on political confidence: Democratization experiences will have direct as well as indirect effects on political confidence, summarized by the following assumptions: Hypothesis 2: The more countries are democratized, the higher is the level of trust in political institutions 30 Hypothesis 3: In socio-economically higher developed countries, confidence in political institutions is more widespread: Less corruption, unemployment, income inequality, or high levels of GDP create higher levels of trust. Hypothesis 4: Political confidence will be more widespread in less ethnic and religious fractionalized societies. 29 In non-democratic regimes ethnic fractionalization is less problematic since the dictatorship generally has the means to oppress ethnic conflicts. Mostly, after the regime change ethnic conflicts arise (cf. e.g. Spain and the Basque conflict or ex-yugoslavian countries). 30 People living in less experienced democracies and having spent their lives in more restrictive regimes are less trustful. Countries with democratic neighbourhood or proximity to international (democratic) organizations and with governments having competency in solving problems have higher levels of trust. 17
Since the combination of micro- and macro indicators within one model causes statistical problems when the two-level-structure is ignored (cf. Steenbergen/Jones 2002), we will estimate the effects through a multi-level-approach controlling simultaneously for the effects of macro- and micro-level variables. 31 In our case there are two levels of analysis: individuals nested in states with different democratization paths. While the general aim of multilevel analysis is to account for variance in a dependent variable that is measured at the lowest level of analysis by considering information from all levels of analysis (Steenbergen/Jones 2002: 219), the motivation for this analysis is twofold: Firstly, we seek to answer the question if there is a significant variation in political confidence at the country level. Secondly, the effects of democratization experiences on political confidence for new democracies in Europe shall be identified. We apply a step-by-step-procedure (cf. Table 4 and Table 5), estimating first the empty model in order to see if there is a significant variation in political confidence between the new democracies under consideration. This model will be followed by the estimation of random intercepts and random slopes of the individual level variables 32 ; only then two-level effects are. In order to control for other effects than democratization issues, additional indicators have been taken as control variables. 33 The empty models (Model 1) show that for both dependent variables confidence in the parliament as well as in regulatory institutions estimates for the grand mean and for the variance components are statistically significant. Hence, there is significant variance in political confidence between new democracies in Europe: In the case of confidence in the parliament 6% of variation can be attributed to the country level, in the case of confidence in regulatory institutions 5% 34 ; this is the evidence that the multilevel character of political confidence should not be ignored: One can ignore this aspect of the data only at the peril of drawing incorrect statistical inferences (Steenbergen/Jones 2002: 231). In Model 2 individual level effects, taking into account the multi-level character of the data, are computed. Firstly, we can see that all individual-level indicators have significant effects 31 For the estimated models presented in Table 5 and 6 Hierarchical Linear and Nonlinear Modeling (HLM 6.04) has been used. 32 Only relevant individual-level variables have been included: Since socio-demographic indicators which have been introduced as control variables in the OLS-regression models did not show significant changes, they have been omitted for multi-level-analyses. The indicator for values of the society has been also excluded, because of insignificant effects in most countries (moreover, the indicator hasn t been asked in Croatia). 33 Since social capital is perceived as a systemic characteristic (cf. Gabriel/Walter-Rogg 2008: 230), the mean level of social trust per country will be also tested on the aggregate level. Political indicators such as the fractionalization of the party-system, voter turnout, the amount of social security transfers, or the question if the country has a majoritarian or proportional/consensus system or respectively a parliamentary or presidential government have been tested. However, none of the cited indicators showed a significant effect on political confidence. 34 These values denoted intra-class correlation have been computed using the variance components of the individual as well as the contextual level. 18