Examining political participation in Lithuania: The direct and mediated effects of social trust.

This article investigates the effects of social trust, both direct and mediated – via internal and external efficacy – on different forms of political participation in post-communist Lithuania. The relationship between social trust and participation features prominently in the social capital and civic culture literature, but little empirical evidence exists that supports it, especially in post-communist democracies. We use the Lithuanian National Elections Study 2012 to test our hypotheses and replicate our analysis with the European Social Survey waves of 2014 and 2016. Our results show that social trust increases turnout, because it is related to a sense of external efficacy, which in turn enhances the likelihood that people vote. There is, however, no association between social trust and being involved in other institutionalised politics, namely, working for a political party. Interestingly, we find a positive indirect effect for non-institutionalised political participation: social trust increases external efficacy, which in turn enhances protest behaviour. Overall, however, social trust does not lead to more protesting, because the former is at the same time positively related to political trust, which seems to decrease, rather than increase non-institutionalised participation. In sum, our findings demonstrate that explanations for political participation based on the core element of social capital – social trust – work out differently for different forms of political participation.


Introduction
Almost two decades ago, the ground-breaking study of Howard (2003) underlined the conventional wisdom that citizens in Central and Eastern European (hereafter -CEE) democracies are less likely to engage in civil society practices than people from other post-authoritarian countries, and especially, mature Western democracies. More recent studies suggest that there is a considerable revival of civil society in the post-communist region which is demonstrated, for instance, by re-emerging mass protests in Hungary, Romania, Ukraine, Slovakia and Poland (Ekiert and Kubik, 2017;Foa and Ekiert, 2017;Ekman et al., 2016;Jacobsson and Saxonberg, 2013). Yet, despite the positive signs of recovery of post-communist societies, participation in both electoral and non-electoral realms still remains relatively low in most CEE countries as noted in various recent studies (e.g. Kostelka, 2014;Huber and Montag, 2020;Hooghe and Quintelier, 2014;Bäck and Christensen, 2016;Vráblíková, 2013;Karp and Milazzo, 2015). Why, after thirty years of political freedom, would that be the case?
Using the Lithuanian case as an example, we argue that political participation stems, at least partly, from trusting social relationships that, in turn, stimulate political efficacy. Lithuania is a typical post-communist case, in the sense that despite the substantial institutional and economic development since the collapse of communism, political participation levels are still considerably low. In fact, as we will show later, the levels are even relatively low compared to other postcommunist-countries (Gaidytė, 2015). Moreover, post-communist countries, including Lithuania, are characterised as societies that inherited a huge social trust deficit, which is more pronounced if we compare these countries with other post-authoritarian countries, let alone mature democracies (Horne, 2014). This observation is our starting point why it is necessary to analyse the relation between social trust and political participation. 1 Do social trusters in Lithuania participate in politics more actively than others? If so, why do they do so? And those who trust people less, do they participate less and why so? Put differently, what is the underlying causal chain between social trust and political participation? We particularly focus on the mediating effects between social trust and political participation, namely, via internal and external efficacy, as they are arguably the most important motivators to participate in politics (Verba et al., 1995). Our research focus is placed within a broader context of the post-communist developments. Addressing the relationship between social trust and political participation in Lithuania, we seek to provide a better understanding of civil society and political activism in the CEE countries. More specifically, we contribute to our knowledge about the determinants of individual choices to participate in politics in two ways.
First, the necessity of our study lies in the observation that there are conflicting findings on how social trust affects political participation, especially in post-communist societies. Are the Western-driven theories of social capital also applicable in Eastern Europe? A few empirical studies provide evidence that social trust and some forms of political participation in the postcommunist context are positively related (Bădescu et al., 2004;van Deth et al. 2002;Armingeon 2007;Sedláčková and Šafr, 2009;Letki, 2003). However, to the best of our knowledge, the existing findings are limited to the direct effect of social trust on political participation. What is missing in such studies, and what this article addresses, is that we are not only interested in the direct effect (if any) of social trust on political participation, but also in the possible mediation of this relationship by political efficacy. To support our arguments, we rely on the political psychology literature about the effects of social motives on cooperation (Kerr and Harris 1996;Kerr 1996). We scrutinise these mechanisms of mediation, because while it is commonly assumed and empirically supported that cooperative motives determine a sense of efficacy in Western societies 2007; Putnam 1993). It encourages tolerance for pluralism and a variety of lifestyles, which is inevitable for the implementation of fundamental human rights and freedoms in democratic regimes (Misztal, 1996;Sztompka, 1999;Newton, 2007;Uslaner, 2002;Mishler and Rose, 2005). Moreover, social trust allows for peaceful conflict resolution, compromise and consensus, because when people trust each other, they are committed to the same democratic values and principles (Žiliukaitė, 2005, 87). This insight is key when talking about the consolidation of democracy in post-communist countries.
The perceived collective obligations instilled in social trust connect it to political participation, a sine qua non of democratic political systems (Armingeon, 2007). Different traditions of democratic theory emphasize the value of political participation (Pateman, 1970;Barber, 2003). Barber (2003, 145) suggests that only through participation people can become citizens who can enjoy freedom, equality, and social justice, because these are common goods which can only be generated through politics. Through participation, politicians can be hold accountable for their decisions, and will thus be more responsive towards their electorate.
To sum up the social capital and civic culture literature, the conventional perception is that both social trust and political participation are desirable for democracy, but scholars debate whether there is a direct empirical relationship between these two concepts. Basically, two different claims can be distinguished. On the one hand, the advocates of social capital theories assume that citizens need a basic level of social trust before they will embark on various forms of political participation. The underlying reason of this argument is that social trust facilitates coordinated actions and cooperation, which eventually "spills over" to the political realm (Gambetta, 1988;Putnam, 1993Putnam, , 2000Norris, 2002b;Van Deth et al., 2002). The more trustful an individual, the more s/he tends to cooperate and is exposed by civic norms, including norms about participation. When citizens participate in small-scale civic associations, they are taught habits of cooperation and thus are socialised into larger political involvement -the process is conceptualised as "social spiral" with social trust and political action at both ends (Putnam, 1993(Putnam, , 2000Newton, 2007;Van der Meer and Van Ingen, 2009). These assumptions are first of all applicable to voting and other forms of institutionalised participation, such as party membership, campaigning, contacting politicians, and doing voluntary work for a political party.
Moreover, some authors argue that non-institutionalised participation, which includes a range of less formalised ways citizens make their voice heard, like signing petitions, demonstrating, striking and boycotting, relies even to a greater extent on social trust than institutionalised participation. Social trust is particularly important for facilitating mobilization, as demonstrating and other forms of elite-challenging participation are usually more risky. As Benson and Rochon (2004) point out, taking part in a demonstration may sometimes entail arrest and prosecution, depending on the political context and the atmosphere (for instance, whether some of the protesters become violent, how the police reacts, etc.). Social trust reduces the expected risks of participation. It leads to optimistic estimates about the likelihood of success, because it allows to form expectations about the actions of others (Dasgupta, 1988). People engage in collective actions because they trust others will do so as well. Klandermans (1984) argues that the odds of the success of a protest action are directly related to how many people participate.
On the other hand, some authors claim that social trust means that citizens will withdraw from some forms of political participation. This alternative approach posits that social trust may work as "double-edged sword", meaning that trusting people can remain passive, because they believe that others can be trusted to participate (and "do the job") for them (Pattie et al., 2003, 458). The scholarly literature distinguishes different associations between social trust and the different modes of political participation, for instance, empirically supporting the negative link between social trust and working for political parties/campaigns (Hooghe and Marien, 2013;Hooghe and Quintelier, 2014;Van der Meer and Van Ingen, 2009). This could be explained by the competitive political nature and strategic interests an individual possess (Imbrasaitė, 2008). However, drawing on the social capital literature, we expect that the relationship between social trust and political participation, irrespective of its mode, follows the same logic as with civic participation. Accordingly, we formulate the following three hypotheses: H 1-3: Social trust will be positively associated with voting (H1), participation in other forms of institutionalised politics (H2), and participation in non-institutionalised politics (H3).
It is evident that the impact of social trust on participation cannot be investigated in an isolated manner (cf. Kriesi and Westholm, 2007). In order to understand how trust impacts participation, we need to delve into the underlying mechanism of this relationship. We find the Civic Voluntarism Model, developed by Verba, Schlozman and Brady (1995), helpful here. This Model emphasises the importance of different resources for political participation. A major resource that determines individual's willingness to participate in politics is political efficacy, defined as 'the feeling that individual political action does have or can have an impact upon the political process' (Campbell et al., 1954, 187). Political efficacy has been studied extensively since the 1950s and its positive effect on political activity is deep-rooted in empirical research (Abramson, 1983;Rosenstone andHansen, 1993, Verba et al., 1995;Morrell, 2005;Van der Meer and Van Ingen, 2009). In line with Lane (1959), we distinguish between internal and external political efficacy. Internal efficacy indicates a person's competence and his perceived capabilities to take part in the political process, whereas external efficacy indicates a person's perceived responsiveness of the political system (Niemi et al., 1991). The reason why this distinction is important, is that it might contribute to different types of political participation differently. Internal efficacy is assumed to foster participation in both institutionalised (including voting) and non-institutionalised politics, because it refers to individual's awareness of political issues and points towards the ways how to influence politics (Gamson, 1968, 48;Sheerin, 2007). Individuals with higher levels of external efficacy are more inclined to embark on (only) traditional political activities, because they feel that their voice is heard by the political system (Hooghe and Marien, 2013).
The scholarship dealing with how social trust and political efficacy both directly affect political participation is mainly confound within the fields of political science and sociology, and does not provide more in-depth explanations how these concepts could be related. Borrowing from the field of political and social psychology, we elaborate on the psychological mechanisms of the mediation effects of efficacy. Kerr (1996, 210-211) proposes that "a variable that can be established as an antecedent of cooperation might, in fact, also be an antecedent of feelings of efficacy and, could either partially or wholly exert its effect on cooperation indirectly". This argument stems from the so-called transformation hypothesis stating that those with more cooperative social motives attach greater weight to others' outcomes (Kerr and Harris, 1996). In other words, social trust as a cooperative social motif would translate into a sense of efficacy, and efficacy in turn enhances cooperation. The connection between social trust and internal efficacy is most straightforward. While trusting, people are more inclined to socialize with each other, thus they more often acquire political knowledge and information in general about the subjects to be acted on and the methods to employ such actions. We do agree that distrustful persons could also gain political information, but trust in others widens the perspectives of political community (Anderson, 2010;Milner 2002;Kornberg and Clarke, 1992).
The relationship between social trust and external political efficacy is more intuitive and less theoretically developed, but there are several hints in the social/political psychology literature. Anderson (2010: 61) claims that individuals who are more trusting, are also more successful in building relationships with others and in influencing the opinions of fellow citizens. Following this logic, such individuals might be encouraged to believe that s/he can also be influential in the political arena. A closely related explanation is that a trusting person feels that s/he can have a control, in the sense that others would act in a predictable (not harmful) way (Kornberg and Clarke, 1992). Trust suppresses fear and thus enables for an action, including political one (Sztompka, 1999). Hence, trusting individuals have good reasons to believe that his/her action might be potentially effective. We formulate the following hypotheses: If these hypotheses hold true, the findings would add to the theoretical arguments about the consequences of social trust and social capital in general, or, paraphrasing Putnam, it would help to explain the linkages within the "social spiral" of trust and participation. The mediation effect would demonstrate that social resources, at the one end of social spiral, could be transformed into individual political resources, at the other end of the spiral. Having formulated our hypotheses, in the following section we shortly overview the results of previous studies that relate to our hypotheses.

Prior results on the relation between social trust and participation
The currently available empirical findings on how social trust affects political participation is quite ambiguous. Almost two decades ago, van Deth (2001) found no correlation between social trust and voter turnout in Western European societies. Consequently, he concluded that social trust as a collective good does not seem to be relevant for the decision to spend a few minutes at a ballot box once every few years. But more recent research finds a positive relationship. For instance, both Hooghe and Marien (2013), using European Social Survey (ESS) 2006 data, and Armingeon (2007), using Citizenship, Involvement, Democracy (CID) data gathered in 12 European democ-racies, reported a weak, although positive correlation between social trust and voter turnout. Similarly, using the ESS 2008 data, Bäck and Christensen (2016) find a positive significant effect of generalised trust on voting.
The relationship of social trust with other institutionalised forms of participation (than voting) in available studies is, in contrast to what we have hypothesised, mostly negative (Hooghe and Marien, 2013;Hooghe and Quintelier, 2014;Armingeon, 2007;van der Meer and van Ingen, 2009;Bäck and Christensen, 2016). The authors notice that social trust especially correlates negatively with contacting politicians, being a member of a political party, and working for political parties and campaigns. In a similar vein, Uslaner and Brown (2005) claim that generalised trusters are less likely to attend political meetings or rallies, but rather choose more communal or charity-related participation. Drawing on this evidence, it makes even more sense to keep voting and other forms of institutionalised participation separate in our analyses.
When turning to non-institutionalised forms of political participation, Kaase (1999) finds a positive relation between social trust and participation in boycotts and street demonstrations in nine Western European countries, using the 1981-1996 European/World Values Survey. Similarly, using the ESS data, a positive effect of social trust on protesting and other non-institutionalised forms of participation is found in the analyses of Hooghe and Marien (2013) and Bäck and Christensen (2016). And the aforementioned study of Uslaner and Brown (2005) observes a positive effect of social trust on signing petitions (on the aggregated level). In Asian countries, Kim (2014) found that generalised trust is mainly positively associated with non-institutionalised participation: political boycotting and petitioning.
The communist legacy might have a profound effect on the linkage between social trust and political participation. Three main reasons are singled out in the literature (Markova, 2004;Howard, 2003;Kornai et al., 2004). First, post-communist societies inherited distrust in organisations and people who join organisations. Second, people in this region are often disappointed in the transformation and in democracy, which deters them from participating, regardless of the social resources they possess (citizens rather choose the freedom "to not participate"). Third, the societal structure of a post-communist society relies on family and friendship-based networks that replace second-level associations that, according to Putnam, breed particularised and, eventually, generalised trust.
In post-communist societies, the relationship between social trust and political participation is yet empirically underresearched, apart from a few notable exceptions (Bernhagen and Marsh 2007;Letki 2003;Uslaner 2004;Armingeon 2007;Sedláčková, and Šafr 2009). The most extensive studies, comparison-wise, are of Armigeon (2007) and Bernhagen and Marsh (2007), who find no significant differences in terms of the relationship between social trust and different forms of political participation in Eastern and Western European societies. The latter study observes that social trust has a positive impact on protest participation, both in Eastern and Western European countries (Bernhagen and Marsh, 2007, 63). Armigeon (2007, 379) finds that social trust is positively related to voting, but negatively to contacting (other relations are not significant). Letki (2003, 20) concludes that social trust significantly influences partisanship and voting in Eastern Europe.

Case selection
As we already pointed out, our study focuses on Lithuania. Different aspects of political participation and social capital in Lithuania have been discussed and analysed by many scholars (Ramonaitė, 2006(Ramonaitė, , 2007Žiliukaitė et al., 2006;Matonytė, 2004;Matonytė et al., 2017;Imbrasaitė, 2004;Bartuškaitė ir Žilys, 2011;Genys, 2018;Riekašius, 2011;Kavolis, 1997). The scholarship widely acknowledges that Lithuanians have less social resources and are less politically engaged than citizens in Western democracies (cf. Genys, 2018). According to the European Values Study, participation in demonstrations in Lithuania dropped dramatically from 34.0 in 1990 to 7.7 percent in 2008. The European Social Survey indicates that in the waves between 2012 and 2016 only around 2 percent of the respondents participated in lawful demonstrations, 6.6 percent signed a petition (2016), while boycotting slightly increased from 3.7 percent in 2014 to 7.2 percent in 2016. In the most recent Lithuanian parliamentary elections (in 2016), the voter turnout of the parliamentary elections (first round) was as low as 51 percent (53% in 2012). This is one the lowest scores in the region (and represents one of the strongest declines since 1990), and these alarming figures are only comparable with the relatively low levels in Poland, Romania and Bulgaria (data from the European Election Database).
Two factors are in common between Lithuania and other young democracies in the CEE region and make Lithuania an exemplary case. On the one hand, civil society is characterised by the negative experience of a totalitarian communist regime (Aidukaitė, 2016;Sztompka, 1999;Paldam and Svendsen, 2001). Due to the restrictions of civic and political rights, Lithuanians used to invest their time mostly in private circles, rather than embarking on civic or political actions. Whereas in Western democracies people participate in political life because of their needs and desires, people in Lithuania, as in many other post-communist societies, tend to assume that their social needs are already fulfilled in family-based circles (Imbrasaitė, 2004).
On the other hand, there are hardly any political-institutional barriers to participate anymore. From an institutional perspective, most post-communist countries are consolidated democracies. The transition that followed by the rapid globalisation and Europeanisation has opened up opportunities to be exposed by wider social circles and, in turn, be involved in various types of political participation.

Data and measurements
The data for studying the hypotheses come from the Lithuanian National Elections Study (hereafter: LNES), which is a post-election survey carried out in 2012, November -December. 2 Additionally, we also replicate our analysis and test the hypotheses using ESS data, merging its waves 7 (2014) and 8 (2016). The findings of this replication are shown and discussed in the appendix. 3 For the sake of readability, if not indicated otherwise, the reported operationalisation and results here in the main text are from the LNES.
The LNES offers a rare opportunity to examine the relationship between social trust and participation in depth: it provides extensive information on political activities, social and political attitudes, as well as political skills (not all these measures were included in the ESS). In total 1500 residents in Lithuania of 18 years and older were interviewed in 161 localities. The sample was selected in a stratified random way. Each district was represented proportionally to its residents' share to the Lithuanian population. The sample includes residents regardless of their citizenship, nationality, religion, language or legal status. Face-to-face interviews at the respondents' homes were performed. The response rate was 61%. 4

Dependent variables
Voting is gauged by the question: Did you vote on the national parliamentary elections day, 13th of September, 2012? Answers: yes=1; no=0.
Involvement in other forms of institutionalised political participation is tapped by the following question: Did you work in a political party or participated in its activity during the last 12 months? (Yes/No). Unfortunately, we did not have other items that would measure this type of participation. For instance, party membership is not asked in both the LNES and the ESS.
Participation in non-institutionalised politics is measured by 5 indicators: 1) Participated in a civic action/movement, which was not related with charity in the last 12 months; 2) Signed a petition (not via internet) in the last 12 months; 3) Took part in a demonstration or picket in the last 12 months; 4) Participated in a strike in the last 12 months; 5) Bought or boycotted certain products because of moral or political reasons in the last 12 months; Answer categories: 1 = yes, 0 = no. The distribution of the summed variable is highly skewed: the majority of the respondents did not participate in any movement action. Therefore, we decided to apply a dichotomised scoring: We give a score of 1 to participants who participated at least in one movement politics activity, and a score of 0 to the non-participants. The ESS only includes 4 out of these 5 indicators of non-institutionalised participation, thus our replication slightly differs in this respect (see more details in the appendix).

Independent variables
Social trust is measured with a standard question "Generally speaking, would you say that most people can be trusted or you can't be too careful?" Measurement: 0 = you can't be too careful; 10 = most people can be trusted. In the ESS, this variable is operationalised with two additional questions (cf. Zmerli and Newton, 2008).
We measure external political efficacy as perceived government's responsiveness, which is captured with the question "Do average citizens have an influence on the important decisions in the country?" Measurement: 0 = No influence, 10 = Very big influence.
Internal political efficacy is tapped by four items, measuring agreement or disagreement with the following statements: 1) "I consider myself to be well qualified to participate in politics"; 2) "I feel I have a good understanding of the most important political issues facing our country" 3) "I feel that I could do as good job in public office as most other people"; 4) "I think that I am better informed about politics and governing than most people". Answers ranged from Disagree strongly (1) to Strongly agree (5). Including all respondents with at least one non-missing value on the items, the general score for each individual is a mean of the items (Cronbach's α = 0.91). 5 The question wording and answer categories of all other variables (controls) are presented in Table A1 in the appendix. We included several control variables because links between variables can, of course, be spurious -social trust, political efficacy, and political participation might be just correlated due to common source variables, for instance, socio-economic resources (Bernhagen and Marsh, 2007). In addition to social status, financial situation, and level of education, we also control for religious attendance, trade union participation and engagement in what Putnam calls 'secondary organizations' (we call this social embeddedness). We also control for the two sociodemographic characteristics, age and gender, since these factors influence political participation (Verba et al., 1995).
Finally, we control for political trust, political knowledge (political information), and political interest. It is important to isolate the impact of social trust and political efficacy by controlling for these attitudinal factors. The existing literature suggest that efficacy and political knowledge are two fundamentally different concepts that need to be disentangled, but one might hypothesize that individuals who lack political knowledge and interest are less likely to experience (internal) political efficacy (Reichert, 2016). Therefore, we control for these potential confouders. Political trust is included as control variable, since it might emerge as possible alternative explanation for the link between social trust and political engagement. The effect of social trust needs to be empirically isolated from the impact of political trust on participation.

Descriptives
Before examining the hypotheses, we scrutinise the descriptives of the key variables. Table 1 shows that 65% of the respondents reported to have voted in the 2012 elections (in ESS 2014-2016 this is 57% 6 ). The levels of participation in both institutionalised and non-institutionalised forms of politics are considerably lower. Only six percent of the Lithuanians were active in party politics activities. Overall, at least one of the movement actions were taken only by 15% (in ESS 2014-2016 this is 4% and 13%, respectively). Thus, except for voting, a vast majority participate in neither institutionalised nor non-institutionalised politics at all.
Social trust is generally rated below the midpoint (the mean is 4.13, on a 0-10 scale), suggesting that most respondents think that they cannot be "too careful" with other people. In ESS 2014-2016, the mean of this item is 4.90). When it comes to external and internal political efficacy, the descriptives show a big gap between these two variables: perceived external efficacy of the respondents is considerably lower (2.46 on a 0-10 scale) than internal efficacy (2.49 on a 1-5 scale). Put differently, the belief that the government is responsive to citizens' demands is more than twice as low as the confidence in one's own ability to participate in politics. To assess to what extent the variables are related to each other, we consider the Pearson's correlation coefficients ( Table 2). The results show that voting and participation in institutionalised politics positively correlates with most key variables, thus supporting the traditional paradigm of explaining conventional participation: civic values of a good citizen endorses his/her activity in institutional politics. Yet, the correlation between social trust and voting and working for a political party is rather weak (Pearson's r = 0.10 and 0.03), which is likewise the case in ESS 2014-2016 (Pearson's r = 0.07 and 0.08). Non-institutionalised political participation is most significantly correlated with social embeddedness (0.29), internal efficacy (0.22) and political interest (0.22). As for social trust, it most firmly goes along with external efficacy (0.30) and political trust (0.27), but, surprisingly, only weakly correlates with embeddedness in formal social networks (0.09).

Results
We performed three logistic regressions for each form of political participation. 7 Table 3 displays the results for voting. Model 1 only includes socio-demographic variables and social trust. It demonstrates that social trust increases the chance that citizens take part in the ballot. The tentative answer to our first research question is thus that there is a positive relationship between social trust and political participation. We have reported the odds ratios and the changes in probabilities attributable to changes in the independent variable. OR reports the change in the odds ratio associated with one unit increase in the predictor variable. Hence, our results show that one unit increase in the amount of social trust (on a scale from 0 to 10) increases the odds of voting by a factor of 1.11 (which hardly differs from what the ESS 2014-2016 shows: 1.08). The depicted predicted probabilities are more easily interpretable: citizens with the lowest level of social trust have a much lower propensity (60%) to go to the polls compared with citizens who have the highest level of trust (81%), holding all other variables constant. This is a considerable effect. If we could raise the average level of social trust in the Lithuanian electorate by just one unit (from the actual figure of 4.13 to 5.13), we predict that the turnout, which is currently 65.0 per cent, would increase to 67.3 per cent. 8 To conclude, hypothesis 1 (Social trust has a positive direct effect on voting) is tentatively confirmed by the results. Note: * p < 0.10, ** p<0.05, *** p<0.01, **** p<0.001 (two-tailed tests). Source: LNES 2012.
Model 2 adds external and internal political efficacy, and additional control variables, to examine why social trusters tend to vote more often. We find that external efficacy significantly increases the chance to go voting. This (predicted) chance is considerably boosted from 65% to 92% if we compare people with the least and most external efficacy. As conventional theories on political participation would assert, our results show furthermore that political interest and political information are positively related with voting. The impact of social embeddedness is also significant; for people who are strongly involved in both a civic association and the local community, the probability to go voting is no less than 90%, which is almost 20 percentage points higher than those who score zero on social embeddedness (71%). Taken all these variables into account, the influence of social trust becomes insignificant.
We also tested the mediation effects of external and internal efficacy in a more sophisticated way by using the PROCESS macro for SPSS developed by Hayes (2013). 9 The results (depicted in Table 4) show that hypothesis 4A (External political efficacy plays a mediating role between social trust and voting) is empirically supported. At the same time, however, there is no empirical support for hypothesis 4B (Internal political efficacy plays a mediating role between social trust and voting). Internal political efficacy has no significant effect on the propensity to vote. In sum, our results demonstrate that social trust enhances electoral participation, because social trusters perceive the political system as more responsive, which, in turn, stimulates voting. Internal efficacy, however, does not mediate the relation between social trust and voting. The latter could be explained by the fact that we control for both political information and political interest; particularly the latter correlates strongly with internal efficacy. A closer inspection indeed reveals that if we would not control for these two variables, hypothesis 4B can be confirmed. In that case, the effect of internal efficacy (OR = 1.634; SE(B) = 0.080) on voting becomes substantial and highly significant. However, if one strictly considers political interest and political information as different concepts than internal efficacy that need to be controlled for, we indeed have to reject hypothesis 4B. Finally, we note that replication with ESS data largely confirms our conclusion (see appendix B for more details).
Concerning participation in institutionalised politics (in this case -working for a party), Table  5 indicates that there is no direct effect of social trust. This refutes hypothesis H2 and instead confirms the argument that the competitive and conflict-based atmosphere of institutionalised politics is not related to cooperative attitudes in Lithuania (cf. Imbrasaitė, 2008). We obviously neither find support for the mediation effect of efficacy (H5A and H5B). Nevertheless, internal efficacy significantly increase participation in party politics activities. In sum, our findings suggest that involvement in institutionalised politics is driven by particular interests, rather than social trust and altruistic behaviour. Note: * p < 0.10, ** p<0.05, *** p<0.01, **** p<0.001 (two-tailed tests). Source: LNES 2012.
Our analysis replicated with ESS data (see appendix B) yields exactly the same conclusion in terms of which of our key variables have positive significant effects on working for a party. Likewise, internal efficacy enhances participation, whereas external efficacy does not. We find an interesting difference, however, because the ESS analysis supports hypothesis 4B (Internal political efficacy plays a mediating role between social trust and participation in party politics). The reason for this difference is not that the effect of internal political efficacy on the propensity to be active for a party is smaller (the effect in the LNES (Exp(B)=2.20) is considerably stronger than in the ESS (Exp(B)=1.22), but can be explained by the fact that social trust is more strongly related with internal efficacy in the ESS sample than in the LNES sample (LNES: r=0.13; ESS: r=0.20).
Finally, the analysis of participation in non-institutionalised (Table 6) indicates that there is no effect of social trust on this type of political activities (the total effect is insignificant). Hence, we have to reject H3 (Social trust has a positive effect on participation in non-institutionalised politics). This finding leads us to consider that people are recruited to protest on the basis of particularised trust (strong ties between people we know), rather than generalised social trust (week ties between strangers). This was also observed at the beginning of the independence movement in 1987 (Ramonaitė, 2011). We will elaborate more on this point in the discussion section. When we include other variables in our analysis (Model 2), the results show that both internal and external efficacy have a positive effect on engagement in protest behaviour. They both spur social movement activities: the predicted percentage of people who engage in non-institutionalised politics increases from 8% to 21%, if we compare people who score lowest and highest on internal efficacy. Similarly, people who perceive the least external efficacy have a much lower chance (9%) to be politically active than people who perceive the most external efficacy (23%). Contrary to the results of voting, internal efficacy has an independent (significant) effect on protests actions, even if we control for political information and political interest. As one might expect, we find that political interest and information are positively associated with the likelihood to engage in non-institutionalised activities.
The effect of social trust remains insignificant in Model 2. The mediation analysis using the above-mentioned macro of Hayes (Table 5) reveals that there is an indirect effect of social trust on non-institutionalised participation, via external efficacy. This indirect path results from the fact that social trust is a strong predictor of external efficacy, and, in its turn, external efficacy fosters participation. H6A is therefore corroborated.
It is worth asking why then, overall, social trust does not stimulate protesting: this is due to the fact that it significantly increases both external efficacy and political trust. The significant positive effect of external efficacy on non-institutionalised participation supports the assumption that disappointment in government's responsiveness diminishes protests actions. In contrast, at the same time we find that political trust has a negative influence (although the effect is insignificant), indicating that the less individuals trust political institutions, the more they tend to protest (cf. Marchenko, 2016). Our findings importantly indicate that the conviction that politicians are responsive works out quite differently than the conviction that they are trustful: the former leads to activism, the latter might yield apathy.
Finally, our replication based on the ESS data -we refer again to the appendix B for more details -reveals that the total effect of social trust on protesting hardly differs between the LNES (Exp(B)=1.05) and the ESS (Exp(B)=1.06), and this effect is significant in the ESS (because this sample is larger). Moreover, we observe a similar intriguing pattern as we just have described above: the indirect effects show that social trust both increases and decreases participation at the same time, because it stimulates political trust and external efficacy, which have contradicting effects.

Conclusion and discussion
This paper investigated the direct and mediating effects of social trust on political participation. In the literature, competing claims have been put forward with regard to the relation between social trust and various forms of political participation. While some authors claim that social trust enhances all forms of political participation, others assume that social trust has no direct association with political action. In this article, we conduct a comprehensive analysis of this relation, based on arguments about the consequences of social trust and social capital in general. Paraphrasing Putnam, we theorised on the linkages within the "social spiral" and argued that social resources (i.e. social trust), at the one end of social spiral, could be transformed into individual political resources (i.e. politial efficacy), at the other end of the spiral. We used the 2012 Lithuanian National Elections Study to test our hypotheses and we replicated the analysis with the ESS 2014-2016 data.
Our results showed a positive association between social trust and voter turnout. Moreover, we found that social trust translates into a sense of external political efficacy, which, in turn, boosts electoral participation. In this regard we can conclude that the lack of social trust by a majority of the Lithuanian population leads to lower levels of electoral participation, since it diminishes one's perceived responsiveness of the political system. Interestingly enough, however, we found that internal efficacy has no significant additional effect on the propensity to vote. This could be due to the fact that we also controlled for two factors that are conceptually distinct, yet empirically related, namely political interest and political knowledge.
For other forms of institutionalised participation we studied, namely, working for a political party or participating in party activities, no relation with social trust could be observed. That is perhaps not surprising, keeping in mind that the existing empirical research done in other European countries found similar results (Hooghe and Marien, 2013;Hooghe and Quintelier. 2014;Armingeon, 2007;van der Meer and van Ingen, 2009;Bäck and Christensen, 2016). In line with these studies, our findings suggest that those who are engaged in party activities rely on other individual resources than social trust in fellow citizens, and in this regard this form of participation is different from voting.
Furthermore, we found no association between social trust and non-institutionalised forms of political participation. Yet, the same as with voting, we find that social trust indeed increases a sense of external efficacy, which in turn leads to participation in non-institutionalised politics. That we however find no overall association between protesting and social trust, is due to the fact that the latter is not only positively associated with external efficacy, but also with political trust. And our results suggest that people who are efficacious engage more in protest behaviour, whereas people who have more trust in politics engage less. Taken everything into account, social trust does not stimulate non-institutionalised participation, at least not in this particular dataset of 2012. At the same time, this finding underlines the value of our mediation approach in order to provide the causal mechanism between social trust and political participation.
A shortcoming of our study is that we only investigated one particular national context, namely Lithuania, at one specific point in time. Although our analysis replicated with other, newer data confirmed our main conclusions, it revealed also a few differences. More importantly, it is difficult to say how well our findings represent similar patterns in other post-communist democracies. Given the fact that there are similarities between these countries, our findings may be generalisable beyond the case at hand. Our study could provide general lessons for the whole post-communist region for two reasons.
First, to our surprise, we shed doubt on the claim that social trust is a resource that brings people to demonstrations, unlike in many Western European countries (for instance, Kaase, 1999;Bäck and Christensen, 2016). There are a few arguments to explain this. The first argument rests on the observation that in less democratic countries, social trust functions as a guarantee for social order (Rossteutscher, 2010). It means that in such societies, socially trusting people are also politically trusting (and vice versa), and thus, are not willing to embark on elite-challenging political activities. In more mature, Western democracies, social trust and political trust are less strongly connected (Newton, 2007). The second argument refers to the type of protests citizens embark on. The literature observes that social trust, as a resource to seek for common good, leads to participation in demonstrations that deal with post-materialist (universalistic) issues, in contrast to the so-called "bread and butter" (particularistic) demonstrations (Verhulst and Walgrave, 2009). Presumably, the former ones are more prevalent in older democracies, and the latter more in newer democracies. Unfortunately, our questionnaire did not specify the issues of the non-institutionalised political actions. The third argument stresses the role of civic organisa-tions/associations in involvement in protest behaviour. Apparently, people who belong to such secondary organisations are not more often generalised trusters, which could be because of the hierarchical, competitive, non-democratic nature (cf. Letki 2004) of these organisations or, because of a self-selection effect. Regarding the latter, as Karakoc (2013) observes, less economically privileged people prefer associations whose members are alike or choose not to participate at all. As a result, these people have less chances to increase interactions with different others, they are less likely to feel as a part of the community and, in turn, they tend to not cherish a feeling of generalised social trust (Uslaner, 2002).
The second broader lesson of our study is that the two concepts of political efficacy and political trust should be kept separately -the conviction that one can influence politics is different than the belief that it is necessary to do so. Gamson (1968) already stressed this a long time ago. We derive this from our finding that external efficacy is positively associated with non-institutionalised participation (i.e. protesting), while political trust is not. This observation is worth scrutinising further, because it seems that external efficacy and political trust work in contradicting directions. Future studies could investigate whether there are any interaction effects between the two concepts in explaining why people embark on a protest action or not.
To conclude, although formal institutions in the post-communist region are displaying attributes of consolidated democracies, citizens generally feel distant from political life and are unaware of means to influence politics. In addition to political-institutional structures, we have demonstrated the role of lacking cooperative social attitudes and political skills to explain why citizens are not willing to use their political rights for which they have struggled so hard. This paradox of political apathy will probably still be hunting the post-communist region in the years to come.

Appendix B: robustness check (cross-validation with the ESS data)
In order to assess the robustness of our findings, we tested our hypotheses also with different data. In their fourfold typology of replication, Freese and Peterson (2017) call this form 'repeatability' (doing the same analyses on new data), although our attempt also somewhat resembles 'generalisation' (doing different analyses on new data), because we could not rely on using exactly the same variables and the investigated time periods differ.
We relied on the European Social Survey (ESS) and merged rounds 7  To ensure comparability, we did not include this item -besides, it is debatable whether contacting politicians is actually a valid measurement of institutionalised political participation since it is sometimes associated more with clientelism. Value 1 if at least one of the items has the score 1, value 0 if none of the items has the score 1.
The results of the logistic regressions are reported in Tables B2-B4. Additionally, we also tested the mediation effects of external and internal efficacy by using the PROCESS macro developed by Andrew Hayes. These findings are depicted in Table B5. To accommodate the comparison, we have also reported the findings on the analyses based on the LNES in the same tables. Different outcomes in terms of the significance and/or direction of the effects are marked in bold in the results based on the ESS datasets.
Based on the LNES, we found that social trust enhances voting, because social trusters have more efficacy, which, in turn, stimulates voting. The ESS analysis largely confirms this conclusion: it shows that the total indirect effect is significant, which means that, taken together, both types of efficacy jointly mediate the relationship between social trust and voting (see Table B5). However, considering the effects of the two mediators separately, we find that the effect of external efficacy is not statistically significant, because the effect in the ESS (Exp(B)=1.10) is smaller than in the LNES (Exp(B)=1.19). The latter revealed that external efficacy has a positive significant effect on voting (see Table B2).
This difference could perhaps be due to different measurements. In the LNES, respondents' perceived government's responsiveness is captured with the question "Do average citizens have an influence on the important decisions in the country?" The ESS formulation is somewhat more abstract and asks respondents how much they would say that "the political system in Lithuania allows people to have an influence on politics".
Next, Table B3 concerning institutionalised political participation (in this case -working for a party) shows that the ESS analysis yields largely similar outcomes in terms of which of our key variables have positive significant effects: embeddedness, political interest, political trust, and internal efficacy all enhance participation, whereas external efficacy does not. Our conclusion based on the LNES was exactly the same.
We find an interesting difference, however, when we consider hypothesis 4B (Internal political efficacy plays a mediating role between social trust and participation in party politics). The ESS analysis supports this hypothesis, whereas the LNES analysis rejected it (see Table B5). The reason for this difference is certainly not that the effect of internal political efficacy on the propensity to be active for a party is smaller, because we find that the effect in the LNES (Exp(B)=2.20) is considerably stronger than in ESS (Exp(B)=1.22). A better explanation is that social trust is more strongly related with internal efficacy in the ESS sample than in the LNES sample. This is already evident from a simple glance at the correlation between the two variables (LNES: r=0.13; ESS: r=0.20). That also explains why the total effect of social trust on working for a party is significant in the ESS analysis (see model 1 in Table B3).
Finally, we consider participation in non-institutionalised politics. Overall, social trust significantly stimulates protesting -we did not find this effect in the LNES. Interestingly, the total effect of social trust on protesting hardly differs between the ESS (Exp(B)=1.06) and the LNES (Exp(B)=1.05) datasets (see model 1 in Table B4). However, this effect is statistically significant in the ESS data, whereas it is insignificant in the LNES.
Furthermore, we find that internal efficacy stimulates participation: this effect of internal efficacy on protesting is exactly the same in both datasets (LNES: Exp(B)=1.35; ESS: Exp(B)=1.35). The mediation analysis (see Table   B5) shows however that the indirect effect of social trust on non-institutionalised participation via internal efficacy is only statistically significant in the ESS sample, as this indirect effect in the LNES is much smaller. This is for the same reason that we have just explained above: social trust is more strongly related with internal efficacy in the ESS sample than in the LNES sample.
The mediation analysis of ESS data shows that H6A is rejected -in contrast, our main findings based on the LNES corroborated this hypothesis. This hypothesis claimed that social trust leads to participation in non-institutionalised politics, because social trusters perceive more external efficacy, which in turn leads to participation. In contrast, based on the ESS analysis, we find that those who perceive high responsiveness are actually less inclined to participate in protests. Interestingly, at the same time we find that political trust has a positive influence (and the effect is significant), indicating that the more individuals trust political institutions, the more they tend to protest.
Remarkably, we found exactly the opposite pattern in our main analysis based on the LNES data. The only intriguing similarity is that both variables have contradicting effects on protesting (see Table B4). It shows that concerning the question why people participate in non-institutionalised political activities, it is probably hard to disentangle the role of political trust and external efficacy. Based on both the ESS and the LNES we find a support for both the positive and negative effects of external efficacy ('protesting because it is effective' vs 'protesting because politicians don't listen to us'), as well as a support for both the positive and negative effects of political trust ('protesting because politicians will listen to us' vs 'protesting because it is necessary'). Note: *p < 0.10, **p<0.05, ***p<0.01, ****p<0.001 (two-tailed tests).  Note: * p < 0.10, ** p<0.05, *** p<0.01, **** p<0.001 (two-tailed tests).  Note: *p < 0.10, **p<0.05, ***p<0.01, ****p<0.001 (two-tailed tests).