Social Trust and Political Participation

24
Social trust and Political Participation – A multilevel analysis of 25 European democracies Maria Bäck 1 and Henrik S. Christensen 2 1 st DRAFT. PLEASE DO NOT QUOTE. 1 Dpt. of political science, Åbo Akademi University, Fänriksgatan 3A, FIN-20500 Åbo. E-mail: [email protected] 2 Dpt. of political science, Åbo Akademi University, Fänriksgatan 3A, FIN-20500 Åbo. E-mail: [email protected]

description

confianza social y participación política

Transcript of Social Trust and Political Participation

  • Social trust and Political Participation A

    multilevel analysis of 25 European democracies

    Maria Bck1 and Henrik S. Christensen2

    1st DRAFT. PLEASE DO NOT QUOTE.

    1 Dpt. of political science, bo Akademi University, Fnriksgatan 3A, FIN-20500 bo. E-mail: [email protected] 2 Dpt. of political science, bo Akademi University, Fnriksgatan 3A, FIN-20500 bo. E-mail: [email protected]

  • Abstract. The link between social trust and political participation has often been found to be

    either weak or non-existent. However, these studies have neglected the interaction between

    different levels of analysis and have instead examined them at either the aggregated level or

    the individual level. These studies hereby overlook that there are good reasons to assume

    that the effect of social trust on individual behaviour is context dependent. With the help of

    multilevel logistic regressions, we examine how four contextual elements affect the impact

    of social trust on three forms of political participation: voting, institutionalised, and non-

    institutionalised participation. The data for the study comes from the fourth round of the

    ESS from 2008 and include a total of 25 European democracies and 47489 respondents at the

    individual level. The results suggest that the context does indeed play a significant role in

    shaping the effect of social trust on political participation.

  • 1

    INTRODUCTION

    To see social capital is an important predictor for political participation has virtually become

    a truism within the study of political participation. Following the seminal work of Robert

    Putnam (1993, 2000), several studies have examined the link with political participation.

    Social capital is traditionally defined as having two central components: social networks,

    and interpersonal trust (Putnam, 2000: 19). When it comes to political participation, most

    studies find that the network component is the central element for explaining how active

    citizens are (Krishna 2002; Teorell 2003; Howard and Gilbert, 2008; Maloney and van Deth,

    2010; Bck, 2011). Putnam also grants primacy to voluntary associations that instil habits of

    cooperation and public-spiritedness in their members and function as schools of democracy

    (Putnam 2000: 338-339). All social networks do, however, not produce social trust or make

    democracy work. There are organizations that do the exact opposite, creating unsocial

    capital instead. Rothstein argues that it is logically contradicting if social capital creates

    social traps instead of dissolving them (Rothstein 2003: 7094). Previous research has

    indicated that social trust may be the very fabric of social life. While associations may

    vitalise social life, social trust is the moral base, a resource that makes it possible to co-

    operate even if we are not certain how others are going to act (Uslaner, 1999: 122; Ilmonen,

    2000: 28). Nevertheless, when it comes to generalized social trust and the link to political

    participation, most studies find, only a weak or even non-significant relationship both at the

    individual (Rothstein 2003; Bck 2011) and at the aggregated level (Uslaner and Brown,

    2005).

    However, most of these studies have only paid limited attention to how the context can

    mediate the effect of social trust on political participation. This is unfortunate since it has

    been noted that the impact may well be dependent on the context, thus constituting a

    missing link that can help explain some of the ambiguity in past research reports. Social

    capital can be both an individual asset as well as a public good, and individuals whose

    personal social capital is low may take advantage of the potentially high contextual social

    capital (Putnam 2000: 20). In a similar vein, generalized social trust can, at the societal level,

    be seen as an indicator of social cohesion, allowing the citizens to come together and join

    forces in social and political networks. Social trust may facilitate life in diverse societies,

    promoting both tolerance and acceptance (Stolle, 2002; Uslaner, 2002). There are therefore

    good reasons to believe that generalized social trust matters more under certain

    circumstances that in others.

    For this reason, this study sets out to examine how the context influences the effect of social

    trust on the propensity to become active in three forms of political participation: voting,

    institutionalised participation, and non-institutionalised participation. We examine the

    impact of four central contextual elements at the country level: state legitimacy and

    contextual social trust are both central cultural elements, whereas the effective electoral

  • 2

    threshold and fiscal decentralisation captures the extent of power sharing embedded in the

    institutional structure of the state. Specifically, we examine the following research questions:

    1. The direct effect of social trust on political participation (individual-level).

    2. Contextual effects (an examination of the direct contextual effect on participation).

    3. Interaction effects (an examination of whether the context mediates the effect of

    social trust on participation).

    The data for examining these research questions come from the fourth round of the

    European Social Survey from 2008 (ESS, 2008) and in all we include 25 democratic countries

    and a total of 47489 respondents. The results suggest that there are indeed significant

    contextual effects that moderate the impact of social trust on political participation.

    THEORETICAL CONSIDERATIONS ON THE LINK BETWEEN SOCIAL TRUST AND

    POLITICAL PARTICIPATION

    Following the work of Putnam (1993, 2000), the link between social capital and political

    engagement has been debated within political science. Most research focuses on the link

    between being active in voluntary associations and being politically active (Krishna 2002;

    Teorell 2003; Howard and Gilbert, 2008; Maloney and van Deth, 2010; Bck, 2011). The link

    between social trust and political participation has also been the topic of interest, but there is

    a certain ambivalence in the findings. While some studies find a positive relationship

    between social trust and political participation (Fennema & Tillie, 2001; Kaase, 1999;

    Putnam, 1993; 2000) others find the connection to be either weak, non-existent or even

    negative (Bck, 2011; van Deth, 2001; Milner, 2000; Muhlberger, 2003; Pattie et al., 2003).

    Hence, there is need for a more detailed study on how and why social trust may affect

    political participation.

    With social trust we mean generalized interpersonal trust as opposed to particularized

    interpersonal trust. While the former entails the general feeling that most people can be

    trusted, the latter is confined to people we already know, for example people in our own

    social or demographic group. While particularized trust does not benefit the society as a

    whole, but may rather harm it by creating a sense of us and them among the citizens,

    generalized trust enhances the feeling of a common moral foundation, identity and norms. It

    may also reduce the risk of free riding (Oskarsson, 2003; Uslaner, 2002).

    When studying the link between social trust and political participation, it is necessary to

    note that political participation has evolved to become a heterogeneous phenomenon that

    can be conceptualized as involving everything from voting and party membership to

    political consumerism and taking part in demonstrations (Norris, 2002; van Deth et al., 2007;

    Christensen, 2011).Acknowledging that political participation is a multifaceted concept, we

  • 3

    use the definition offered by Brady (1999: 737): action by ordinary citizens directed towards

    influencing some political outcomes. To capture the richness of political participation, we here

    focus on three central forms of political participation in the representative democracy:

    voting, institutionalized participation in close vicinity to the political system, and non-

    institutionalised participation outside the formal political system (Kaase, 1999; Marien et al.,

    2010).

    A large battery of variables has been proposed to account for the different forms of

    participation (Barnes, Kaase, et al, 1979; Norris, 2002; Dalton, 2006; van Deth et al., 2007;

    Christensen, 2011). For social trust, the intuitive expectation is that it has a positive effect on

    all forms of political participation. The more trusting an individual is, the more political

    active she is. The literature on the topic gives some suggestions on how the mechanism

    between social trust and political participation works. According to Putnam (1993: 99120) a

    good society is made up of an interested and trustful citizenry and the emphasis is on civic

    engagement, political equality, solidarity and tolerance. These principles are reflected in

    what is presumed to be a good citizenship and what norms are perceived to characterize

    such a citizenship. Community oriented citizens do not only have the right to participate in

    common matters, but they are also expected to be active, both for their own sake and the

    sake of the community. Among others, Oskarsson argues that interpersonal trust moderates

    the free passenger-problem. What he means by this is that individuals who act rationally

    tend to free ride and take advantage of the cooperation of others without participating

    themselves. The idea is that we pitch in, for example by recycling, paying our taxes etc. as

    long as we trust that other people do the same. Interpersonal trust is an important

    prerequisite for this to happen (Oskarsson, 2003: 70; Rothstein, 2003). This leads us to

    propose the following hypotheses on the link between social trust and the three forms of

    political participation:

    Hypothesis1a (H1a): Social trust has a positive effect on the propensity to vote.

    Hypothesis1b (H1b): Social trust has a positive effect on the propensity to be active in

    institutionalised participation.

    Hypothesis1c (H1c): Social trust has a positive effect on the propensity to be active in non-

    institutionalised participation.

    Although the theoretical arguments are convincing, the empirical evidence is as noted far

    from conclusive. There may be several reasons for this. Social trust may work as a double-

    edged sword (Pattie et al., 2003: 458). Trusting people may become politically active because

    they expect that they will be successful in whatever the purpose of the activity is, but they

    can also remain inactive because they believe that other people can be trusted to get the job

    done. Untrusting individuals may also become politically active to ensure that the

    unreliable government officials will be held accountable. Finally people can remain

    disengaged if they feel that the officials cannot be controlled by the public and social trust

  • 4

    may therefore, according to the authors, work in an opposite or self-cancelling way when it

    comes to political engagement for certain people.

    This shows that uncertainties still linger when it comes to the link between trust and

    participation. A pertinent question concerns how the link between social trust and

    participation is affected by the context in which the political actions occur. Previous research

    indicates that social trust is dependent on the context in which it is embedded (Stolle, 2002).

    We therefore examine the role of contextual factors. We focus on four central aspects of the

    context.

    The first two aspects concern cultural dimensions of the environment. In their original work

    on political culture, Almond and Verba (1963: 13) state that [W]hen we speak of the

    political culture of society, we refer to the political system as internalized in the cognitions,

    feelings, and evaluations or its population (Almond and Verba 1963: 13). Some of the key

    aspects of political culture are support in government, trust in the political system and

    political efficacy. These aspects are of central meaning for both social trust (Rothstein 2003,

    2011; Bck 2011) and political participation (e.g. Booth and Seligson, 2009). Two cultural

    aspects are of particular relevance for the present purposes. The first aspect concerns the

    legitimacy of the state. Gilley (2006: 499) asserts that political legitimacy is a major

    determinant of both the structure and operation of states and that it profoundly affects the

    states behaviour toward the citizens and others. According to Gilley, legitimacy is an

    endorsement of the state by citizens at a moral and normative level. Following David

    Beethams (1991) theory of the legitimation of power, he makes a distinction between three

    constitutive sub-types of legitimacy that together define the idea of rightfulness as it is

    believed by citizens: views of legality, views of justification and acts of consent (Gilley, 2006:

    502). While the latter in Gilleys measurement approach is solely focused on behaviour, we

    will concentrate on the attitudinal indicators of the two former attributes.

    The second cultural aspect concerns the community level of social trust. The positive link

    between social trust and participation would be expected to apply at the aggregate level as

    well, that is to say, a trusting community displays higher levels of political activity than

    communities with scarce levels of social trust. Parallels can be drawn to Putnams study of

    the Italian regions and the assumption that democracy will work better in an environment of

    high social capital: Stocks of social capital, such as trust, norms and networks, tend to be

    self-reinforcing and cumulative. Virtuous circles result in social equilibrium with high levels

    of cooperation, trust, reciprocity, civic engagement and well-being, as well as honesty and

    law-abidingness (Putnam 1993: 111, 177). There are many, albeit disputed (see e.g. Jackman

    and Miller 1998), attempts to treat social capital as culture. Inglehart (1997: 188) defines

    social capital as a culture of trust and tolerance and Almond and Verba (1963) identify

    interpersonal trust as an important trait conducive to democracy when they find that nations

  • 5

    with longer tradition of stable democracy demonstrate higher levels of trust and that trust

    also correlates with political activism. According to Putnam (1993: 99120) a good society is

    made up of an interested and trustful citizenry and the emphasis is on civic engagement,

    political equality, solidarity and tolerance.

    In addition to the two cultural factors, we also examine the effects of two institutional

    factors. The context of political institutions has been argued to either directly or indirectly

    influence both patterns of political engagement (Powell, 2000; Kittilson and Schwindt-Bayer,

    2010; Christensen, 2011; Marien and Christensen, 2012) and has also been argued to be affect

    levels of social trust (Freitag and Bhlmann, 2009). Following Lijphart (1999), a basic

    distinction is frequently drawn between power sharing and power concentrating democratic

    systems (Kittilson and Schwindt-Bayer, 2010: 991). The power sharing democracies value

    inclusion and the representativeness of the political majorities and therefore aim to generate

    governments that are representative of a wide array of interests. Power concentrating

    democracies prioritise accountability and the rule of the majority and therefore aim to

    produce efficient majority governments with clear responsibility for the decision-making.

    The extent of powers sharing or institutional openness plays a central role in determining

    how easy it is for citizens to influence the political decision-making between elections

    (Christensen, 2011: 5961).

    We focus on two central institutional aspects. The effective electoral threshold captures how

    easy it is for political parties to win representation in parliament (Powell, 2000). A low

    effective electoral threshold increases the number of effective parties competing in elections

    and thereby extends the extent of power sharing within the system. Fiscal decentralisation is a

    measure for the extent of vertical decentralisation of the political system and denotes the

    proportion of local and regional spending in relation to the total spending (Morales, 2009:

    170). Increased decentralisation adds to the extent of power sharing by dividing decision-

    making powers between several layers of government. Both of these have been used in

    previous research examining the institutional effect on political behaviour (Powell, 2000;

    Kittilson and Schwindt-Bayer, 2010; Christensen, 2011; Marien and Christensen, 2012).

    These four contextual factors may affect political participation directly. Previous literature

    has found that cultural factors affect the likelihood of being active in political activities

    (Almond and Verba, 1963; Inglehart, 1997; Denk, 2009). The level of political participation is

    generally higher in countries with high contextual social capital (Bck, 2011). In a similar

    vein, studies have found that institutional factors affect participation in political

    organisations (Morales, 2009) and protest (Dalton et al., 2010), although other studies finds

    the direct effect of the institutional context on individual behaviour to be limited when

    controlling for individual level factors (Marien and Christensen, 2012). Nevertheless, this

    leads us to examine the following hypotheses for the direct effects of the contextual factors:

  • 6

    Hypothesis2a (H2a): The context directly affects the propensity to vote.

    Hypothesis2b (H2b): The context directly affects the propensity to be active in

    institutionalised participation.

    Hypothesis2c (H2c): The context directly affects the propensity to be active in non-

    institutionalised participation.

    In addition to a direct effect of the contextual factors, the context may also determine the

    impact of social trust on political participation. To our knowledge, no previous studies have

    examined this question for social trust.

    Nevertheless, there are good reasons to expect the cultural factors to have a mediating effect

    on how social trust affect political participation. Communities have a certain level of

    (contextual) social trust, and within these contexts there are individuals with varying stocks

    of individual social trust. Putnam (2000: 20) contends that there may be spillover-effects

    from the societal level to individuals in the way that poorly connected individuals benefit

    from a well-connected society. According to Uslaner (1999: 132) social capital will spread

    more easily in contexts where the general level of social capital is already high. He further

    claims that when individual social capital decreases, there will be less contextual social

    capital to go around, which in turn will make its effects weaker (see also Field, 2003: 4041).

    For the institutional factors, Freitag and Bhlmann (2009) find that power sharing

    institutions promote social trust among citizens. Previous studies also suggest that the

    institutional context may interact with individual level characteristics such as political trust

    (Marien and Christensen, 2012) and political efficacy (Karp and Banducci, 2007) and thereby

    indirectly shape political behaviour. Since power sharing institutions are expected to have a

    positive influence on both political behaviour and levels of social trust, it is likely that there

    is also a positive interaction between the two, leading social trust to have a stronger effect on

    participation in power sharing democracies. In sum, we examine the following hypotheses

    for the mediating effect of the context on the link between social trust and participation:

    Hypothesis3a (H3a): The context mediates the effect of social trust on the propensity to

    vote.

    Hypothesis3b (H3b): The context mediates the effect of social trust on the propensity to be

    active in institutionalised participation.

    Hypothesis3c (H3c): The context mediates the effect of social trust on the propensity to be

    active in non-institutionalised participation.

    The following section outlines how we examine the hypotheses derived from the theory.

  • 7

    RESEARCH DESIGN

    The data for studying the nine hypotheses come from the fourth round of the European

    Social Survey from 2008 (ESS Round 4, 2008). In addition to the restriction to a European

    context, the study is restricted to democratic countries in order to avoid conceptual

    travelling (Sartori, 1970). For this reason, Turkey, Russia and Ukraine are excluded from the

    study, since they are not rated as democracies. Israel has also been excluded since it is

    geographically located outside of the European context. This leaves 25 countries1 and a total

    of 47489 respondents, although some respondents are excluded from the analyses due to

    missing data.

    Political participation is measured by three variables which gauge the involvement in three

    forms of political participation that have all been the topic for previous research2:

    Voting: voted or not last national election.3

    Institutionalised participation: Contacting politicians, Working for a political party,

    and being member of a political party.

    Non-institutionalised participation: Sign petitions, take part in public

    demonstrations and boycott products.4

    All three dependent variables are coded dichotomously. This means the study examines the

    choice to become active rather than the extent of involvement. Since the questions used only

    indicate if the respondent has performed each activity within the last 12 months and not

    how often, any attempt to probe the extent of involvement is tentative at best (cf. Topf, 1995:

    68).

    The central independent variable is social trust, which is measured with the help of a

    standard index based on three questions concerning the perceived trustworthiness of fellow

    citizens: 1)Most people can be trusted or you can't be too careful; 2) Most people would try

    to take advantage of you if they got the chance; 3) People try to be helpful (Cronbachs

    alpha = 0.80). Although the measuring of a complex concept such as social trust is inherently

    difficult in survey research (cf. Nannestad, 2008: 417), this measure is in line with the

    standard approach and presents the best available opportunity for operationalizing the

    concept. The index is coded to range between 0 and 1 with 1 indicating the highest level of

    social trust.

    The context is operationalized with measures that capture the central aspects of the cultural

    and the institutional dimensions of the environment. The two variables used to capture the

    cultural dimension are operationalised as follows:

    State legitimacy (Gilley, 2006): Since we examine forms of political behaviour we use a

    purely attitudinal measure that consists of views on legality and views of justification.

  • 8

    The variable has been coded to vary between 0 and 1 with 1 indicating the highest

    extent of state legitimacy for a country.

    Contextual social trust: This is the social trust measure at the individual level

    aggregated to the country level. The measure has been coded to vary between 0 and 1

    with 1 indicating the highest extent of contextual social trust for a country.

    For the institutional dimension, the two aspects are operationalised as follows:

    The effective electoral threshold: The effective threshold gauges how easy it is for

    political parties to win representation in parliament (Powell, 2000). This is calculated

    as 75 %/( M + 1) (M = number of seats allocated) and ranges between 0 and 37.5 as the

    highest theoretical value. Values taken from Lundell and Karvonen (2008). Recoded

    to vary between 0 and 1 with 1 indicating lowest effective threshold.

    Fiscal decentralisation denotes the proportion of local and regional spending in

    relation to the total spending (Morales, 2009: 170). The percentages come from IMF

    Government finance statistics. Ranges between 0 (no decentralisation) and 100

    (totally decentralised). The values are since recoded to vary between 0 and 1 with 1

    denoting the highest extent of decentralisation (Switzerland =0.44) since the notion of

    a totally decentralised state is unrealistic.

    All four contextual variables are coded to vary between 0 and 1, with 1 denoting high

    legitimacy/contextual social trust/power sharing.

    The control variables are all found at the individual level and they include age, gender and

    education, which are three socio-demographic aspects known to influence political

    behaviour (Marien et al., 2010). The analyses also control for political interest and left-right

    ideology, which has been known to affect the likelihood of engagement in different political

    activities (Barnes, Kaase et al., 1979; Norris, 2002). These control variables are also coded to

    vary between 0 and 1.

    The appropriate method of analysis is multilevel logistic analysis since the data is nested in

    countries and the dependent variables are dichotomous (Snijders and Bosker, 1999: 215216).

    Since all variables are coded to vary from 0-1, the estimated coefficient for an explanatory

    variable indicates increase or decrease in the logit of the odds ratio of the dependent variable

    (participate or not) when the independent variable changes from its minimum to its

    maximum value. As is customary, the explanatory variables are centred around the grand

    mean to achieve stable models (Hox, 2002: 62). For ease of interpretation, central estimates

    are translated into odds ratios by taking the exponential of the estimates.

    ANALYSIS

  • 9

    Before examining the hypotheses, we scrutinize the distribution of the key variables among

    the countries included in the study.

    Table 1 Country level distribution of central variables, ESS 2008.

    Political participation Cultural context Institutional context

    Voted last general election

    Institutional participation

    Non-institutional participation

    Social trust State

    legitimacy

    Effective electoral threshold

    Fiscal decen-tralisation

    Percentage have done Mean scores

    Belgium 92.1 19.2 33.3 0.53 0.57 0.76 0.33

    Bulgaria 72.8 9.8 9.8 0.36 0.29 0.89 0.20

    Croatia 78.9 16.7 30.0 0.42 0.34 0.85 0.23

    Cyprus 93.4 24.5 11.1 0.45 - 0.90 0.07

    Czech Republic

    58.0 17.7 19.0 0.48 0.37 0.87 0.35

    Denmark 94.2 24.0 47.4 0.68 0.82 0.95 0.79

    Estonia 64.7 15.2 12.3 0.54 0.39 0.79 0.32

    Finland 83.2 25.1 47.7 0.63 0.74 0.86 0.58

    France 77.6 17.5 48.4 0.49 0.57 0.00 0.32

    Germany 83.7 19.2 47.7 0.53 0.72 0.87 0.76

    Greece 87.9 15.4 18.3 0.36 0.41 0.68 0.02

    Hungary 80.2 9.1 11.2 0.43 0.46 0.47 0.32

    Ireland 79.4 26.2 33.1 0.59 0.78 0.60 0.21

    Latvia 62.9 12.0 13.4 0.48 0.41 0.87 0.41

    Netherlands 86.0 18.8 28.8 0.59 0.73 0.98 0.26

    Norway 85.7 26.4 47.6 0.65 0.81 0.89 0.35

    Poland 72.8 8.4 10.5 0.42 0.49 0.87 0.40

    Portugal 73.7 8.5 8.3 0.42 0.65 0.83 0.19

    Romania 67.8 15.2 7.9 0.36 0.24 0.87 0.46

    Slovakia 77.7 8.5 25.6 0.43 0.33 0.87 0.18

    Slovenia 72.6 14.7 12.6 0.47 0.39 0.80 0.23

    Spain 81.9 11.7 26.8 0.49 0.63 0.74 0.58

    Sweden 91.1 19.2 61.0 0.64 0.70 0.89 0.80

    Switzerland 64.5 17.1 47.8 0.59 0.62 0.77 1.00

    UK 70.3 18.6 46.5 0.55 0.64 0.00 0.20

    Total 77.8 16.5 28.9 0.50 0.55 0.74 0.39

    Note: The entries for the participatory activities indicate the percentage of the populations who indicate having been active in voting, institutionalised participation or non-institutionalised participation. The entries for the contextual factors indicate mean scores across countries. Data weighted with design weight.

    As can be seen, there are important differences in the extent of political involvement in the

    countries under scrutiny. When it comes to electoral participation, the majority indicate

    having voted in the last general election in all countries. Although self-reporting does not

    perfectly measure the actual extent of voter turnout, there is little doubt that this is the most

    popular political activity in Europe despite the fears raised over falling levels of electoral

    turnout. For the institutionalised activities, only 8.5 per cent of the population in Slovakia

    have done at least one of these activities within the last 12 months, whereas the highest

    active proportion is found in Norway, where 26.4 per cent have done one of these activities

    within the last 12 months. Hence, even the highest level of participation is considerably

  • 10

    lower than the lowest electoral turnout. The participation rates are more diverse for non-

    institutionalised participation, where only 7.9 per cent performed one of these activities in

    Romania within the last 12 months compared to a high of 61 per cent in Sweden.

    For the four contextual factors, it should be noted that there is limited variation for two

    factors, namely contextual social trust and the effective electoral threshold. Whereas the lack

    of variation in the effective electoral threshold is due to the limited variation in this regard in

    a European context, the limited variation for contextual social trust was to be expected. The

    values tend to cluster around the mean since it is unlikely that any society would consist of

    individuals who completely trust or distrust other citizens. Hence, the results for these

    factors should be interpreted with some caution.

    Moving on to the central research question of the role of social trust at the individual level

    and the interplay with the context in explaining political participation, the first three

    hypotheses concern the effect of social trust on the three forms of political participation.

    Table 2 displays information on the analyses concerning these hypotheses. For each form of

    participation, three models are displayed. Model 0 is an empty model that decomposes the

    variance into group and individual level. Model 1 only includes social trust as an

    explanatory variable to examine the relationship, whereas model 2 includes the control

    variables to ensure that the relationship is robust.

    Table 2 Multilevel logistic analyses of individual level effects on Voting, Institutionalised participation, and Non-Institutionalised participation.

    Voting Institutionalised Non-institutionalised

    Model 0 Model 1 Model 2 Model 0 Model 1 Model 2 Model 0 Model 1 Model 2

    Fixed Part

    Constant 1.27

    (0.12)*** 1.29

    (0.11)*** 1.58

    (0.12)*** -1.60

    (0.08)*** -1.61

    (0.08)*** -1.77

    (0.08)*** -0.93

    (0.16)*** -0.95

    (0.15)*** -0.93

    (0.16)***

    Social trust

    0.96 (0.20)***

    0.62 (0.16)***

    0.25 (0.11)*

    -0.12 (0.12)

    0.65 (0.09)***

    0.30 (0.11)**

    Age

    2.80 (0.26)***

    0.62 (0.17)***

    -1.15 (0.18)***

    Gender

    -0.13 (0.02)***

    0.19 (0.04)***

    -0.18 (0.05)***

    Education

    0.77 (0.12)***

    0.64 (0.07)***

    0.92 (0.07)***

    Political interest

    1.64 (0.11)***

    1.83 (0.09)***

    1.33 (0.08)***

    Ideology

    0.41 (0.14)***

    -0.13 (0.11)

    -0.54 (0.14)***

    Random Part

    Country level

    Between country error variance

    0.34 (0.07)***

    0.33 (0.07)***

    0.41 (0.09)***

    0.16 (0.03)***

    0.15 (0.03)***

    0.12 (0.03)***

    0.68 (0.11)***

    0.61 (0.10)***

    0.54 (0.10)***

    Estimated intra class correlation (ICC)

    0.09 0.09 0.11 0.05 0.04 0.03 0.17 0.16 0.14

    Respondents (countries)

    43497 (25)

    43002 (25)

    37285 (25)

    47143 (25)

    46583 (25)

    39956 (25)

    47113 (25)

    46546 (25)

    39886 (25)

    Note: The entries are estimates of multilevel logistic regression with standard errors in parenthesis. Since all variables are coded 0-1, the coefficients indicate

    increase or decrease in the logit of the odds ratio of the dependent variable when the independent variable moves from 0 to 1. Data weighted with design weight. Data weighted with design weight. ICC estimated as p = var (uj)/(var (uj) + 2/3) (Snijders & Bosker 1999: 224). *p

  • 11

    The ICC scores in the null models show the extent of variation at the group level. As can be

    seen, this is highest for non-institutionalised participation, where about 17 per cent of the

    variance is at the country level, whereas it is relatively low for institutionalised participation,

    where only 5 per cent is at the country level. Nonetheless, simulation studies suggest that

    even modest levels of ICC can make tests of significance in traditional linear models too

    liberal (Barcikowski, 1981).

    As expected, the results for voting show that social trust has a significant positive impact on

    the propensity to vote. The estimate of 0.62 in the model 2 corresponds to an odds ratio of

    1,87, meaning those with the highest level of social trust are almost two times as likely to be

    active compared to those with the lowest level of social trust when holding other factors

    constant. Hypothesis 1a is therefore confirmed by the results. For institutionalised

    participation, a weak propensity for social trust to enhance the likelihood of being active is

    found in model 1, but the effect disappears in model 2 when controlling for other factors.

    Hence, it is not possible to conclude that there is any effect of social trust on this form of

    participation and H1b has to be rejected. Nevertheless, when it comes to non-

    institutionalised participation, we find, in line with the expectations, a positive effect on the

    propensity to become active. The initial estimate of 0.65 found in model 1 drops to 0.30 when

    controlling for other factors. Although not particularly strong, this estimate equals an odds

    ratio of 1.35, meaning those with the highest level of social trust are about one third more

    likely to be active compared to those with the lowest level of social trust. This means that the

    results validate H1c.

    Moving on to the question of contextual effects for political participation and H2a- H2c,

    table 3 displays the results for each of the four contextual factors for the three forms of

    participation.5

  • 12

    Table 3 Multilevel logistic regression of individual and contextual effects on Voting, Institutionalised participation, and Non-Institutionalised participation.

    Voting Institutionalised Non-institutionalised

    M3 M4 M5 M6 M3 M4 M5 M6 M3 M4 M5 M6

    Fixed Part

    Constant 1.54

    (0.12)*** 1.58

    (0.13)*** 1.58

    (0.13)*** 1.57

    (0.13)*** -1.80

    (0.07)*** -1.78

    (0.07)*** -1.78

    (0.08)*** -1.78

    (0.08)*** -0.93

    (0.12)*** -1.00

    (0.12)*** -0.93

    (0.15)*** -0.94

    (0.13)***

    Social trust 0.64

    (0.08)*** 0.62

    (0.08)*** 0.63

    (0.08)*** 0.63

    (0.08)*** -0.11 (0.13)

    -0.14 (0.12)

    -0.12 (0.12)

    -0.12 (0.12)

    0.33 (0.07)***

    0.31 (0.07)***

    0.30 (0.07)***

    0.31 (0.07)***

    Age 2.81

    (0.09)*** 2.80

    (0.09)*** 2.81

    (0.09)*** 2.80

    (0.09)*** 0.60

    (0.17)*** 0.62

    (0.17)*** 0.62

    (0.17)*** 0.62

    (0.17)*** -1.24

    (0.07)*** -1.22

    (0.07)*** -1.17

    (0.07)*** -1.19

    (0.07)***

    Gender -0.13

    (0.03)*** -0.13

    (0.03)*** -0.13

    (0.03)*** -0.13

    (0.03)*** 0.18

    (0.04)*** 0.19

    (0.04)*** 0.19

    (0.04)*** 0.19

    (0.04)*** -0.20

    (0.02)*** -0.19

    (0.02)*** -0.18

    (0.02)*** -0.19

    (0.02)***

    Education 0.78

    (0.05)*** 0.78

    (0.05)*** 0.78

    (0.05)*** 0.77

    (0.05)*** 0.65

    (0.07)*** 0.64

    (0.07)*** 0.64

    (0.07)*** 0.64

    (0.07)*** 0.97

    (0.04)*** 0.97

    (0.04)*** 0.93

    (0.04)*** 0.95

    (0.04)***

    Political interest 1.67

    (0.06)*** 1.64

    (0.05)*** 1.65

    (0.05)*** 1.65

    (0.05)*** 1.86

    (0.08)*** 1.83

    (0.08)*** 1.83

    (0.09)*** 1.83

    (0.09)*** 1.43

    (0.05)*** 1.40

    (0.05)*** 1.35

    (0.04)*** 1.38

    (0.05)***

    Ideology 0.41

    (0.07)*** 0.41

    (0.07)*** 0.41

    (0.07)*** 0.41

    (0.07)*** -0.12 (0.12)

    -0.13 (0.11)

    -0.13 (0.11)

    -0.13 (0.11)

    -0.58 (0.06)***

    -0.57 (0.05)***

    -0.55 (0.05)***

    -0.56 (0.05)***

    State legitimacy 0.89

    (0.70) 0.61

    (0.39) 2.87

    (0.66)***

    Contextual social trust

    -0.33 (1.43)

    1.29 (0.67)^

    5.66 (1.24)***

    Effective threshold

    0.52 (0.53)

    0.11 (0.16)

    -0.72 (0.60)

    Fiscal decentralisation

    -0.50 (0.54)

    -0.11 (0.21)

    1.65 (0.54)***

    Random Part

    Between country error variance

    0.36 (0.10)***

    0.43 (0.12)***

    0.42 (0.12)***

    0.41 (0.12)***

    0.10 (0.02)***

    0.11 (0.03)***

    0.12 (0.03)***

    0.12 (0.03)***

    0.32 (0.09)***

    0.32 (0.09)***

    0.55 (0.16)***

    0.41 (0.12)***

    ICC 0.10 0.12 0.11 0.11 0.03 0.03 0.03 0.03 0.09 0.09 0.14 0.11

    Respondents (countries)

    36331 (24)

    37285 (25)

    37285 (25)

    37285 (25)

    38960 (24)

    39956 (25)

    39956 (25)

    39956 (25)

    38882 (24)

    39886 (25)

    39886 (25)

    39886 (25)

    Note: The entries are estimates of multilevel logistic regression with standard errors in parenthesis. Since all variables are coded 0-1, the coefficients indicate increase or decrease in the logit of the odds ratio of the dependent variable when the independent variable moves from 0 to 1. Data weighted with design weight. ICC estimated as p = var (uj)/(var (uj) +

    2/3) (Snijders & Bosker 1999: 224). ^p

  • 13

    decentralised country is 5.2 times more likely to participate compared to an average citizen

    in a totally centralised country. These findings indicate how important contextual factors are

    for this form of participation and H2c is therefore verified.

    However, the context does not only matter through direct effects, it can also affect the link

    between social trust and political participation. To examine H3aH3c we include cross level

    interaction effects for each of the contextual effects and social trust. The models are random

    slope models that allow the coefficient of social trust to vary across country. For reasons of

    space, we only display the significant results. These show how the propensity to be active

    develops in the extreme contextual cases at different levels of social trust when holding all

    other factors constant. Table 4 displays the significant interaction effects found for each form

    of participation.

    Table 4 Multilevel logistic regression of individual, contextual and cross-level interaction effects on Voting, Institutionalised participation, and Non-Institutionalised participation.

    Voting Institutionalised Non-institutionalised

    M 7 M 8 M 9 M10 M 7 M8 M7 M8 M9

    Fixed Part

    Constant 1.55

    (0.13)*** 1.51

    (0.12)*** 1.54

    (0.13)*** 1.54

    (0.13)*** -1.76

    (0.08)*** -1.76

    (0.08)*** -0.91

    (0.15)*** -0.91

    (0.12)*** -0.97

    (0.12)***

    Social trust 0.71

    (0.16)*** 0.79

    (0.14)*** 0.73

    (0.13)*** 0.71

    (0.15)*** -0.17 (0.12)

    -0.17 (0.11)

    0.26 (0.12)*

    0.34 (0.11)**

    0.33 (0.11)**

    Age 2.77

    (0.09)*** 2.78

    (0.09)*** 2.77

    (0.09)*** 2.78

    (0.09)*** 0.63

    (0.17)*** 0.63

    (0.17)*** -1.14

    (0.07)*** -1.23

    (0.07)*** -1.20

    (0.07)***

    Gender -0.13

    (0.03)*** -0.13

    (0.03)*** -0.13

    (0.03)*** -0.13

    (0.03)*** 0.19

    (0.04)*** 0.19

    (0.04)*** -0.18

    (0.02)*** -0.21

    (0.02)*** -0.20

    (0.02)***

    Education 0.76

    (0.05)*** 0.77

    (0.05)*** 0.76

    (0.05)*** 0.76

    (0.05)*** 0.64

    (0.07)*** 0.64

    (0.07)*** 0.92

    (0.04)*** 0.97

    (0.04)*** 0.97

    (0.04)***

    Political interest 1.63

    (0.05)*** 1.66

    (0.05)*** 1.64

    (0.05)*** 1.64

    (0.05)*** 1.83

    (0.08)*** 1.83

    (0.08)*** 1.34

    (0.04)*** 1.44

    (0.05)*** 1.41

    (0.05)***

    Ideology 0.41

    (0.06)*** 0.41

    (0.07)*** 0.41

    (0.07)*** 0.42

    (0.07)*** -0.13 (0.11)

    -0.13 (0.11)

    -0.54 (0.05)***

    -0.58 (0.06)***

    -0.57 (0.06)***

    State legitimacy 0.84

    (0.69)

    2.90 (0.67)***

    Contextual social trust

    -0.41 (1.39)

    5.72 (1.25)***

    Effective threshold

    0.10

    (0.16)

    Fiscal decentralisation

    -0.52 (0.53)

    Social trust * state legitimacy

    2.07

    (0.81)*

    -1.93 (0.61)**

    Social trust * Contextual social trust

    5.25 (1.47)***

    -3.33 (1.19)**

    Social trust * Effective threshold

    0.77

    (0.30)*

    Social trust * Fiscal decentralisation

    1.68 (0.61)**

    Random Part: Country level

    Country level variance 0.40

    (0.11)*** 0.34

    (0.10)*** 0.41

    (0.12)*** 0.39

    (0.11)*** 0.12

    (0.03)*** 0.12

    (0.03)*** 0.56

    (0.16)*** 0.33

    (0.10)*** 0.33

    (0.09)***

    Covariance -0.15

    (0.10)* -0.16

    (0.09)^ -0.11 (0.09)

    -0.10 (0.09)

    -0.13 (0.03)*

    -0.14 (0.03)***

    -0.15 (0.09)*

    -0.03 (0.06)

    0.00 (0.06)

    Social trust slope variance 0.45

    (0.17)* 0.32

    (0.14)* 0.28

    (0.12)* 0.36

    (0.15)* 0.24

    (0.09)* 0.21

    (0.09)* 0.25

    (0.10)* 0.15

    (0.08)^ 0.17

    (0.08)*

    ICC 0.11 0.09 0.11 0.11 0.03 0.03 0.14 0.09 0.09

    Respondents (country) 37285 (25) 36331 (24) 37285 (25) 37285 (25) 39956 (25) 39956 (25) 39886 (25) 38882 (24) 39886 (25)

    Note: The entries are estimates of multilevel logistic regression with standard errors in parenthesis. Since all variables are coded 0-1, the coefficients indicate increase or decrease in the logit of the odds ratio of the dependent variable when the independent variable moves from 0 to 1. Data weighted with design weight. ICC estimated as p = var (uj)/(var (uj) +

    2/3) (Snijders & Bosker 1999: 224). ^p

  • 14

    Model 7 for each of the three forms of participation introduces a random slope for social

    trust. The results here suggest that the effects of social trust vary across countries for all

    three forms of participation, since a joint test of the two new parameters covariance and

    slope variance is significant according to an approximate Wald test with 2 degrees of

    freedom.

    For voting, models M8-M10 show that there are three significant interaction effects,

    indicating that there are differences in the effect of social trust dependent on the context

    despite the lack of direct contextual effects. The first significant interaction effect is between

    state legitimacy and social trust, where the interaction effect of 2.07 implies that social trust

    has a stronger effect on the propensity to vote when the extent of state legitimacy is high.

    However, even though the differences are significant, the practical implications are limited

    since the differences are small, especially considering that the extreme values are unlikely to

    be encountered in real life situations. In a similar vein, there is a significant interaction effect

    for contextual social trust. In this case the significant estimate of 5.25 indicates that the effect

    of social trust is weaker in a situation of high contextual social trust compared to a situation

    of low social trust. The differences are more pronounced than for state legitimacy, but again,

    the practical implications may be less notable given that the extreme values are again

    hypothetical. There is a significant interaction effect for one of the institutional variables,

    since fiscal decentralisation has a significant interaction with social trust of 1.68. This

    estimate implies that the effect of social trust is significantly weaker when there is a high

    extent of fiscal decentralisation. Although the practical significance of these estimates may

    be limited due to the overall high levels of voter turnout and the hypothetical nature of

    some of the extreme values, the results nonetheless indicate the potential importance of

    considering the context when estimating the effect of social trust on voting behaviour.

    Hence, these findings suggest that the context mediates the effect of social trust on voting

    and H3a is therefore approved.

    For institutionalised participation, we find one significant interaction effects between social

    trust and the effective electoral threshold shown in model 8, even if the estimate for social

    trust is not significant. This result indicates that the effect is stronger when the effective

    electoral threshold is at its maximum value. In other words, power sharing increases the

    impact of social trust on the propensity to be active in institutionalised activities and people

    with high social trust are more likely to become active within the political system when

    more parties are competing for electoral victories. Although the estimates for the cultural

    contextual effects are not significant, this nonetheless indicates that the context can mediate

    the effect of social trust on participation in institutionalised activities and H3b is therefore

    supported by the evidence.

    Finally for the non-institutionalised activities, Models 8 and 9 show the significant

    interaction effects for the two cultural indicators and social trust. For state legitimacy, there

    is a negative coefficient of -1.93 for the interaction with social trust. A similar result is found

  • 15

    for contextual social trust, where the negative interaction coefficient is -3.33. Since it is only

    for non-institutionalised participation that the contextual factors had both significant direct

    effects and interaction effects, we use these last two results to visualise the importance of the

    interaction effects.

    Figure 1 Interaction effects for non-institutionalised participation

    Social trust * State legitimacy Social trust * Contextual social trust

    Note: The full lines represent the effect of social trust on non-institutionalised participation when the contextual factors are at their maximum value. The dotted

    lines represent the effect of social trust on non-institutionalised participation when the contextual factors are at their minimum value. All other factors are held

    constant at their mean value. See table 4 for more specific information.

    As can be seen, the negative coefficient of -1.93 entails that the effect of social trust on the

    propensity to be active in non-institutionalised activities depends on the legitimacy of the

    state. The likelihood of being active in non-institutionalised participation is much higher

    when the state is legitimate in the eyes of citizens, as indicated by the full line. Furthermore,

    the effect of social trust is stronger in this situation, since the slope of the line is steeper than

    the dotted line representing a situation where the state is completely illegitimate, meaning

    the difference between the two lines increases as social trust rises. This entails that the

    probability of participation increases from 0.53 to 0.60 as social trust moves from 01 when

    state legitimacy is 1 compared to a limited increase from 0.07 to 0.09 when state legitimacy is

    0. The same is true for contextual social trust, where the coefficient of -3.33 means that the

    effect of social trust is much stronger when there is a high extent of social trust among

    citizens. Here the probability of participation increases from 0.80 to 0.84 when the contextual

    social trust is highest compared to an increase of 0.02 to 0.03 when contextual social trust is

    at its lowest value. Hence these findings entail that H3c is approved.

    0.00

    0.20

    0.40

    0.60

    0.80

    1.00

    0.00 0.20 0.40 0.60 0.80 1.00

    State legitimacy = 0 State legitimacy = 1

    0.00

    0.20

    0.40

    0.60

    0.80

    1.00

    0.00 0.20 0.40 0.60 0.80 1.00

    Contextual social trust = 0 Contextual social trust = 1

  • 16

    DISCUSSION OF THE RESULTS

    The connection between social capital and political participation has been controversial in

    the literature ever since the seminal work of Putnam (1993, 2000). The bulk of literature has

    focused on the influence of the associational network component of the concept, whereas the

    social trust component has been somewhat neglected since much research has indicated that

    this dimension was less relevant for predicting involvement in different forms of

    participation (Bck, 2011; van Deth, 2001; Milner, 2000; Muhlberger, 2003; Pattie et al., 2003).

    In line with this, we generally found that social trust was linked positively to voting,

    institutionalised participation, and non-institutionalised participation, but the link was

    weakened considerably when controlling for other factors, and in the case of

    institutionalised participation, even grew insignificant. This finding thus confirms previous

    studies suggesting that social trust may have a positive effect on social outcomes such as

    political participation, but the effect appears to be limited compared to other factors at the

    individual level (cf. Nannestad, 2008: 428430).

    However, this account neglects the possibility that the effect of social trust is dependent on

    contextual factors. For this reason, we examined the interplay between social trust, political

    participation, and the context understood as central cultural and institutional aspects. The

    findings here suggested that the cultural and institutional contextual aspects were less

    relevant for predicting voting and institutionalised participation, but it did promote

    participation in non-institutionalised activities outside the formal political sphere. The

    cultural aspects of state legitimacy and contextual social trust in particular had strong

    positive impacts on being active, but power sharing institutions in the form of fiscal

    decentralisation was also a significant predictor. This finding thus confirms previous results

    that the context can have a significant direct impact on the propensity to be active at the

    individual level (Almond and Verba, 1963; Morales, 2009; Dalton et al., 2010). Nevertheless,

    the effect in particular of political institutions is often fragmentary and depends on the

    form of participation under consideration and what specific institutional aspect is seen as

    important (Christensen, 2011).

    The most remarkable finding concerned the interaction between the contextual elements and

    social trust in shaping political participation. Here the results suggested that the effect of

    social trust on participation is linked to the context, but the specific effect depends both on

    the form of participation and the contextual aspect under consideration. The cultural aspects

    appeared to be of particular relevance, since state legitimacy and contextual social trust had

    significant mediating effects for both voting and non-institutionalised participation.

    However, the institutional aspects also mediated the effect of social trust, since there were

    significant effects between social trust and fiscal decentralisation for voting and between the

    effective electoral threshold and social trust for institutionalised participation. Although the

    contextual effects on participation depend on both form of participation and contextual

    aspect under scrutiny, the cultural and institutional factors both increase the effect of social

  • 17

    trust on the propensity to be active in the political activities under consideration. These

    results thus show that the impact of social trust on participation to a large extent depends on

    the context where the activities take place. In this sense, this study helps explain the

    divergent findings in previous literature when it comes to the impact of social trust on

    political participation.

  • 18

    Appendix: Coding of variables and descriptive statistics

    Variable Questions and coding N Min Max Mean Std VIF

    Dependent variables

    Voting

    Question: Some people don't vote nowadays for one reason or another. Did you vote in the last [country] national election in [month/year]? Coded 0/1 with 1 indicating having voted.

    43001 0.00 1.00 0.78 0.42

    Institutionalised participation

    Index based on three answers to questions concerning the following activities: 1) Contacted a politician, government or local government official.(See non-institutionalised activities for phrasing) 2) Worked in a political party or action group. (See non-institutionalised activities for phrasing) 3) Are you a member of any political party? Coded 0/1 with 1 indicating having performed at least one of the activities.

    47143 0.00 1.00 0.17 0.37

    Non-institutionalised participation

    Index based on three answers to question: There are different ways of trying to improve things in [country] or help prevent things from going wrong. During the last 12 months, have you done any of the following?: 1) Signed a petition. 2) Taken part in a lawful public demonstration. 3) Boycotted certain products. Coded 0/1 with 1 indicating having performed at least one of the activities.

    47106 0.00 1.00 0.29 0.45

    Independent variables, contextual variables and control variables

    Social trust

    Composite index based on 3 questions: 1)Most people can be trusted or you can't be too careful; 2) Most people would try to take advantage of you if they got the chance; 3) People try to be helpful. All scored on scale 0-10; Index coded 0-1 with 1 highest level of

    social trust (Cronbachs alpha = 0.80)

    46918 0.00 1.00 0.50 0.20 1.278

    State legitimacy Index of attitudes towards state legitimacy from Gilley (2006). Recoded to vary between 0-1 with 1 indicating high state legitimacy.

    46274 0.24 0.82 0.55 0.17 3.293

    Contextual social trust

    Mean country scores for social trust, coded to vary between 0-1 with 1

    indicating highest extent of contextual social trust. 47489 0.36 0.68 0.50 0.09 3.961

    Effective

    electoral

    threshold

    Calculated as 75 %/( M + 1) (M = number of seats allocated). Ranges between 0 and 37.5 as the highest theoretical value (Lundell and

    Karvonen, 2008) . Recoded to vary between 0 and 1 with 1 indicating lowest effective threshold.

    47489 0.00 0.98 0.74 0.26 1.135

    Fiscal

    decentralisation

    Values from IMF Government finance statistics. Ranges between 0

    (no decentralisation) and 100 (totally decentralised). Since the notion

    of a totally decentralised state is unrealistic, the max value is subsequently coded with the highest extent of decentralisation

    (Switzerland =0.44) as the maximum value 1.

    47489 0.02 1.00 0.39 0.24 1.574

    Age Age in years divided by 100. 47335 0.15 1.23 0.47 0.18 1.066

    Gender Gender of respondent; Dichotomous 0/1, 1=male. 47467 0.00 1.00 0.47 0.50 1.023

    Education Highest level of education achieved? Coded to vary between 0-1 with 1 highest level of education.

    47384 0.00 1.00 0.52 0.33 1.147

    Left/right-ideology

    Left/right ideology where 0 left and 10 right. Coded to vary between 0-1, 1 = right.

    40683 0.00 1.00 0.51 0.22 1.008

    Political Interest How interested in politics? Recoded to vary between 0-1, 0= Not at all interested, 1 = Very

    interested.

    47355 0.00 1.00 0.45 0.30 1.165

  • 19

    REFERENCES

    Almond, G. & Verba, S. (1963). The Civic Culture: Political Attitudes and Democracy in Five

    Nations. Princeton: Princeton University Press.

    Armingeon, K; D. Weisstanner; S. Engler; Panajotis Potolidis; M. Gerber and P. Leimgruber

    (2011). Comparative Political Data Set 1960-2009, Institute of Political Science, University

    of Berne.

    Barcikowski, R.S. (1981). Statistical power with group mean as the unit of analysis, Journal of

    Educational Statistics, 6(3), 267285.

    Barnes, S., Kaase, M. et al. (1979). Political Action: Mass Participation in Five Western

    Democracies, Beverly Hills, CA: Sage Publications.

    Beetham, D. (1991). The Legitimation of Power. London: Macmillan.

    Booth, J. & Seligson, M. (2009). The Legitimacy Puzzle: Democracy and Political Support in Eight

    Latin American Nations. Cambridge: Cambridge University Press.

    Bck, M. (2011). Socialt kapital och politiskt deltagande i Europa. bo: bo Akademis frlag.

    Christensen, H. S. (2011). Political Participation Beyond the Vote. How the Institutional Context

    Shapes Patterns of Political Participation in 18 Western European Democracies. bo: bo

    Akademi University Press.

    Dalton, R. J. (2006). Citizen Politics: Public Opinion and Political Parties in Advanced Industrial

    Democracies, 4th edition. Washington D.C.: CQ Press.

    Dalton, R.J., Van Sickle, A. and Weldon, S. (2010). The individualinstitutional nexus of

    protest behaviour. British Journal of Political Science, (40): 5173.

    Denk, T. (2009) Politisk kultur. Malm: Liber.

    ESS Round 4 (2008). Data file edition 4.0. Norwegian Social Science Data Services, Norway

    Data Archive and distributor of ESS data.

    Field, J. (2003). Social capital. Key Ideas. London: Routledge.

    Fennema, M. & Tillie, J. (2001). Political Participation and Political Trust in Amsterdam.

    Civic Communities and Ethnic Networks Journal of Ethnic and Migration Studies 25 (4):

    703726.

    Freitag, M. and Bhlmann, M. (2009). Crafting trust: The role of political institutions in a

    comparative perspective. Comparative Political Studies, (42): 15371566.

    Gilley, B. (2006).The Meaning and Measure of State Legitimacy: Results for 72 countries.

    European Journal of Political Research (45): 499525.

    Hox, J. (2002). Multilevel analysis Techniques and applications. Mahwah: Lawrence Erlbaum

    Associates.

    Howard, M. M. and Gilbert, L. (2008) A Cross-National Comparison of the Internal Effects

    of Participation in Voluntary Organizations, Political Studies, 56:1, 12-32.

    Ilmonen, K.(2000). Sosiaalinen poma ja luottamus. SoPhi 42. Jyvskyl: Jyvskyln

    yliopistopaino.

  • 20

    Inglehart, R. (1997). Modernization and Postmodernization: Cultural, Economic, and Political

    Change in 43 Societies. Princeton, NJ: Princeton Univ. Press

    Kaase, M. (1999). Interpersonal Trust, Political Trust and Non-institutionalised Political

    Participation in Western Europe. West European Politics, 22(3): 121.

    Karp, J.A. and Banducci, S.A. (2007). Political efficacy and participation in twenty-seven

    democracies: How electoral systems shape political behaviour. British Journal of Political

    Science, (37): 124.

    Kittilson, M.C. and Schwindt-Bayer, L. (2010). Engaging citizens: The role of power-sharing

    institutions. Journal of Politics, 72(4): 9901002.

    Krishna, A. (2002) Enhancing political participation in democracies: What is the role of social

    capital?, Comparative Political Studies, 35, 437-460.

    Lijphart, A. (1999) Patterns of Democracy Government Forms and Performance in Thirty-Six

    Countries. New Haven: Yale University Press.

    Lundell, K. & Karvonen, L. (2008). A Comparative Data Set on Political Institutions. bo

    Akademi University: Department of Political Science.

    Maloney, W.A. and Jan van Deth (eds.) (2010) Civil Society and activism in Europe

    Contextualizing engagement and political orientations, London: Routledge.

    Marien, S. & Christensen, H.S. (2012) Trust and Openness: Prerequisites for Democratic

    Engagement?, in Demetriou, Kyriakos N. (Ed.) Democracy in Transition - Political

    Participation in the European Union, Springer.

    Marien, S., Hooghe, M. and Quintelier, E. (2010). Inequalities in non-institutionalised forms

    of political participation: A multilevel analysis of 25 countries. Political Studies, 58(2):

    187213.

    Milner, H. (2000). Social Capital and Political Participation: What Can We Learn from

    Comparative Data on Local Elections. Paper for the Political Studies Association UK

    50th Annual Conference, April 2000. London.

    Morales, L. (2009). Joining Political Organisations Institutions, mobilisation and participation in

    Western democracies. Colchester: ECPR Press.

    Muhlberger, P. (2003). Political Trust vs. Generalized Trust in Political Participation. Paper

    for the 2003 Annual Meeting of the American Political Science Association, August

    2003.

    Nannestad, P. (2008). What have we learned about generalized trust, if anything. Annual

    Review of Political Science, (11): 413436.

    Norris, P. (2002). Democratic Phoenix Reinventing political activism. Cambridge: Cambridge

    University Press.

    Oskarsson, S. (2003). Valdeltagandets mekanismer. In Esaiasson, P. & Westholm, A. (eds.).

    Deltagandets mekanismer. Det politiska engagemangets orsaker och konsekvenser. Malm:

    Liber.

    Pattie, C., Sayd, P. & Whiteley, P.(2003). Citizenship and Civic Engagement: Attitudes and

    Behaviour in Britain. Political Studies, 51(3): 443468.

  • 21

    Powell, G.B., Jr. (2000). Elections as instruments of democracy Majoritarian and proportional

    visions. New Haven: Yale University Press.

    Putnam, R. D. (1993). Making Democracy Work: civic traditions in modern Italy. With R.

    Leonardi and R. Nanetti. Princeton: Princeton University Press.

    Putnam, R.D. (2000). Bowling Alone. The Collapse and Revival of American Community. New

    York: Simon and Schuster.

    Rothstein, B. (2003). Sociala fllor och tillitens problem. Stockholm: SNS frlag.

    Rothstein, B. (2011). The Quality of Government. Corruption, Social Trust and Inequality in

    International Perspective. Chicago and London: The University of Chicago Press.

    Sartori G. (1970). Concept Misformation in Comparative Politics. American Political Science

    Review, 64(4): 10331053.

    Snijders, T. and Bosker, R. (1999). Multilevel analysis An introduction to basic and advanced

    multilevel modelling. London: Sage.

    Stolle, D. (2002). Trusting Strangers The Concept of Generalized Trust in Perspective.

    sterreichische Zeitschrift fr Politikwissenschaft, 2002 (4): 397412.

    Teorell, J. (2003) Linking social capital to political participation: Voluntary associations and

    networks of recruitment in Sweden, Scandinavian Political Studies, 26:1, 49-66.

    Topf, R. (1995). Beyond electoral participation. In H.D. Klingemann and D. Fuchs (eds.)

    Citizens and the State (52-91). Oxford: Oxford University Press.

    Uslaner, E. (1999).Democracy and Social Capital. In Warren, M.E. (ed.). Democracy and

    Trust. Cambridge: Cambridge University Press.

    Uslaner, E. (2002). The Moral Foundations of Trust. Cambridge: Cambridge University Press.

    Uslaner, E. and Brown, M. (2005) Inequality, Trust and Civic Engagement American Politics

    Research, 33, 868-894.

    van Deth, J. (2001). The Proof of the Pudding: Social Capital, Democracy and Citizenship

    Paper for the EURESCO Conference, September 2001, Exeter, UK.

    van Deth, J.W.; Montero, J.R., and Westholm, A. (eds.) (2007) Citizenship and Involvement in

    European Democracies A Comparative Analysis. London: Routledge.

  • 22

    ENDNOTES

    1 Lithuania is excluded since the data set does not include appropriate weights. 2 More information on coding and descriptive statistics for all variables is available in the appendix. 3 Although self-reporting is not a perfect measure of actual voting (see Bck, 2011: 78), it is the best available

    for the present purposes. 4 An exploratory factor analysis for the 6 items included in institutionalised and non-institutionalised

    participation show that they load onto two separate dimensions in accordance with this distinction. Two items

    (work other organisation and wear badge) were excluded since the phrasing of the question does not make it

    possible whether the activity was aimed at the formal political system or not. 5 Since the VIF scores indicate that there may be issues with multicollinearity among the contextual factors (see

    appendix) and there are a limited number of countries included in the study, we refrain from examining the

    contextual factors simultaneously. 6 Although not shown for reasons of space, simple linear regression models suggest that more contextual

    estimates have significant effects, but the effects evaporate once controlling for the individual variables.