DOES VOLUNTEERING CAUSE TRUST?: A comparison of the Czech Republic and Norway

A major controversy in research on social capital is whether volunteering for organizations generates generalized social trust and increased trust in political and administrative institutions. On one side, there is Putnam's seminal work Making Democracy Work, finding that social interaction in networks, generalized trust, and civic engagement can be mutually reinforcing (Putnam 1993: 177). Putnam assumes that generalized trust in other people develops through repeated face-to-face interaction. Organizations where people from different social groups and layers may interact on an equal footing, such as sports, cultural activities, and recreational organizations, are of particular importance. People experience the fact that cooperation not only can be meaningful and enjoyable, but also can enable the creation of collective goods that would not be possible by just focusing on individual gain (Putnam 2002). In this way, organizational engagement spills over into social trust and trust in institutions.

On the other side, there is the institutional approach represented by Rothstein and colleagues (Rothstein 2003; Rothstein and Stolle 2003, 2008). They point out that there is little evidence showing that individuals active in voluntary organizations have higher levels of social trust, especially considering that people recruited for active participation probably have higher levels of trust from the start. An indication of this self-selection effect is that passive membership can be just as important for social trust as volunteering (Wollebæk and Strømsnes 2008). The antecedents of social capital are found at the macro (institutional) level rather than at the micro (individual) level. For example, contacts with universal welfare-state institutions tend to increase social trust, whereas experiences with needs-testing social programs undermine it (Kumlin and Rothstein 2005). Trust in impartiality and in conflict-solving institutions (police and courts) strengthens the belief that opportunistic behavior will be sanctioned, and therefore, that people in general can be trusted. This explains why fair welfare-state arrangements in the Scandinavian countries promote social trust, and thereby, indirectly, can be beneficial for volunteering. Thus, according to Rothstein, the social capital literature tends to exaggerate the role of voluntary organizations as a cause for trust. The primary mechanism is that vertical trust, in particular trust in order institutions, such as the police, and the legal and defense systems, promotes horizontal trust, or general social trust (Rothstein and Stolle 2008). Since opportunistic behavior is less likely because of sanctioning institutions, it follows that it is also easier to get people to take part in the production of collective social goods.

This article argues that the effect of volunteering on social and institutional trust depends on (a) the institutional context and (b) the relationship between the organization that is the target for volunteering and the institutional context. The combination of these two factors affects whether volunteering has positive or negative consequences for social and institutional trust (Figure 1).

Examples can illustrate combinations of such organizational and institutional effects. In an institutional context with little corruption or partiality, and thereby, with a generally high level of institutional and social trust, volunteering does not add much to general social trust. The positive impact of an institutional context where people experience regular dealings with welfare services, police, legal institutions, and public administration is so substantial that the more limited positive experiences of face-to-face interactions through volunteering make little difference. Direct dealings with fair and impartial institutions have a much stronger impact than volunteering. Causal mechanisms similar to the institutional theory can be relevant for understanding how institutional, and thereby, social trust is generated in a high trust-context such as Scandinavia. In contrast, in a low-trust institutional context, where people regularly have to deal with corruption and clientelism, volunteering for organizations that define themselves as an alternative to the established social order may generate social trust because it brings people with similar oppositional views and purposes together. Causal mechanisms similar to the social interaction theory can be relevant for understanding how trust is generated by volunteering for certain types of organizations in a low-trust context such as the Czech Republic. Several other causal mechanisms that combine trust generating or disrupting effects of organizational and institutional contexts on volunteering are possible.

In short, the article examines three hypotheses: (1) the social interaction theory views volunteering as a cause for social and institutional trust, which are preconditions for a working democracy; (2) the institutional theory views fair institutions as causes for trust in institutions, and thereby, for social trust since opportunistic behavior can be expected to be sanctioned; and (3) the institutional/organizational theory presented here expects the effect of volunteering on social and institutional trust to depend on (a) whether the institutional context fosters expectations of fair treatment, or corruption and favoritism and (b) whether people volunteer for an organization associated with the existing institutional context or an organization that they see as disconnected from – or even in opposition to – this institutional context.

Previous research has focused on the large observable differences in levels of social and institutional trust between countries. Attempts to explain this on an individual level, by investigating volunteering as a cause, has for the most part been unsuccessful (Dekker and van den Broek 1998; Stolle 2001; Rothstein and Stolle 2008; Wollebæk and Strømsnes 2008). This has put the emphasis on contextual and institutional causes. However, the effect of volunteering for organizations that relate in different ways to the institutional context has not received sufficient attention. One of the reasons for this is that multinational survey-data on volunteering and social trust do not provide enough detail on the organization categories for which people volunteer. The conclusion that volunteering has a weak effect on social and institutional trust could be a net result of some volunteers choosing organizations with positive effects on trust and others selecting organizations with negative effects. It is possible, therefore, that the inability to differentiate between organizational categories has disguised such diverse effects in previous research. In addition, neither the institutional theory nor the interaction theory sufficiently specifies the effect of the interplay between the institutional and organizational contexts of volunteering, which makes the controversy between the theories seem overly clear-cut. This is why we need a third theory.

There is no need to consider fair institutions and social interactions as mutually exclusive causes of social trust. Trying to falsify one of these theories once and for all is futile since a quest for universal validity or social laws is bound to be futile in social science. A more fruitful ambition is to delimit empirically the areas of different social mechanisms’ validity (Boudon 1986; Hedström and Swedberg 1998; Elster 2007). Therefore, the article will show under what circumstances volunteering may affect social and institutional trust. It presents analyses of population surveys that allow for an unusually detailed differentiation between the types of organizations for which people volunteer. The surveys are from Norway and the Czech Republic. Both countries have relatively high levels of participation in volunteering for voluntary organizations, but Norway has very high levels of social and institutional trust while the Czech Republic has relatively low levels. Through comparisons of the country-cases, one can observe how institutional context affects the relationship between volunteering and social, and institutional, trust, while the effects of volunteering on trust can be observed for different types of organizations through analyzing survey-data from each country. In other words, this analytic design combining comparisons of cases with survey analysis enables testing of the institutional/organizational hypotheses.

The question of what produces social capital – face-to-face interactions or impartial and reliable institutions – is not just an academic exercise; it has far-reaching policy implications. Huge efforts currently go into stimulating civil society in developing, and former east-block, countries. For example, the World Bank, EU, and EEA countries, as well as humanitarian organizations and development agencies, support civil society on the presumption that this also promotes social trust, democracy, and better governance.¹ Do our research findings imply that this is money wasted? It would seem more reasonable to suggest that the ability to promote trust depends on supporting the ‘right’ voluntary organizations. But what organizations are they? This is the critical question.

Since institutional context plays an important role in the article's hypotheses, the first section outlines some important differences between Norway and the Czech Republic. The section thereafter analyses social and institutional trust and volunteering for different kinds of organizations, using questions taken from the two countries’ representative population surveys. The concluding section revisits the theoretical hypotheses about the effects of social interaction, institutions, and institutional/organizational contexts on trust, in light of the findings from the empirical analysis.

In the European Social Survey (2002/2003), the average social trust score for Norwegian adults was 6.53 on a scale of 0–10; other Nordic countries scored equally as high. The scores tend to be lower in Southern and Eastern Europe. The average score of Czech adults was 4.48, which was below adult scores from Western European countries but ahead of Portugal, Italy, Slovenia, Hungary, Poland, and Greece (Fridberg and Kangas 2008). In average scores of trust in five types of political and administrative institutions (parliament, political parties, politicians, legal system, police) from 24 countries, Norway is among the top seven countries, whereas the Czech Republic is among the bottom five (Listhaug and Ringdal 2008).

It seems that Czech people have good reasons not to trust institutions. According to Transparency International's Global Corruption Barometer (Table 1),² 14% of the population has paid bribes to one of nine different service providers during the last 12 months. Averages from Latvia, Poland, and Luxemburg are similar, but all are lower than Greece (18%), Hungary (24%), Romania (28%), and Lithuania (34%). Norway, however, is at the bottom with 1% of the population having paid bribes, together with the UK with 1%, and Denmark with 0%. One might conclude that countries with much corruption tend to have lower levels of social and institutional trust. This seems a simple causal relationship before bringing volunteering into the equation.

Both in Norway and the Czech Republic, large shares of the populations are members and volunteers in voluntary organizations. In 2009, 48% of Norwegian adults and 30% of Czech adults had volunteered for voluntary organizations at least once during the previous 12 months. Recent population surveys from Western industrialized countries put Norway on top, together with Sweden and followed by other Nordic countries and the UK. The Czech Republic scores on a level with Australia and New Zealand, but above the USA and Canada with 27%, and France with 25% (Wollebæk and Sivesind 2010: table 6).

Thus, the Czech Republic differs from most post-communist countries by having a relatively high percentage of both volunteering and membership in voluntary organizations. Hodgkinson explains this fact by referring to the historic tradition of citizen involvement and philanthropy (2003: 39), but others largely fail to reflect the differences within the ‘ex-communist cultural zone’ (such as Inglehart 2003: 60), even when the Czech case is identified as diverging. Voicu and Voicu show that all former communist countries have lower levels of volunteering compared with the Western European average, except for the Czech Republic, but they do not attempt to explain it (2003: 10). Even Hodgkinson's explanation does not address why the Czechs have a tradition of citizen involvement, while other post-communist countries in Central and Eastern Europe (CEE) do not. One important difference can account for this. In contrast to other CEE post-communist countries, the Czech Republic did not have its own national state during the nineteenth century, but at the same time, they had a strong national consciousness with roots in the famous medieval era of the Czech state. Czechs, as a part of the Austro-Hungarian Empire, substituted engagement in their national state with their engagement in civil society organizations, mainly in the fields of culture, sport, and recreation. This also contributed to preserving their national identity.

Both in Norway and the Czech Republic, the field of culture and recreation has the largest number of volunteer hours, with 44% in the Czech Republic and 54% in Norway, across all types of organizational volunteering. Thus, they are – together with the other Nordic countries, France, Germany and Italy – among the top seven countries out of 35 surveyed within the Johns Hopkins Comparative Nonprofit Sector Project that have over 40% of volunteering hours in the category culture and recreation (Salamon et al.2003: 58). Both in the Czech Republic, which was part of Czechoslovakia from 1918 to 1993, and in Norway, this dominance of culture and recreation volunteering stems from the traditionally lively associational life found in these countries in the nineteenth and twentieth centuries. Czech civil society organizations in the fields of culture and sports, including the two mass sports associations of Sokol and Orel, played an important role in nation-state building during the nineteenth and twentieth centuries. Unlike most other post-communist countries, there was a democratic regime in the Czech Republic (then Czechoslovakia) in the inter-war period (1918–1938) characterized by the existence of mass civic associations that were often patriotic, supported by and supportive of the state.

In Norway, the norms of volunteering grew strong in the traditional popular movements for farmers, fishermen, small-holders, temperance, lay Christians, and social and humanitarian associations in the nation-building period that started when the 400-year period of Danish rule ended in 1814. The institution of volunteering has gradually been transferred to the field of culture and recreation, as a modern leisure society emerged in the sixties (Sivesind and Selle 2010). Thus, both countries have a high level of volunteering, in particular in the sports, culture, and recreation, that is particularly beneficial for social trust, according to Putnam.

An important difference between the two countries, however, is that the communist regime disrupted the development of the voluntary sector in the Czech Republic. It seems probable that state socialism had a negative impact on both interpersonal and institutional trust. Post-communist countries show markedly lower levels of social and institutional trust than Western European countries do (Fridberg and Kangas 2008; Listhaug and Ringdal 2008; Howard 2003). There is a common assumption that state socialism uprooted civil society (Juknevičius and Savicka 2003). However, although associations were not independent from the state, associational life existed in the former Czechoslovakia (Skovajsa 2008), and in many respects, it retained important continuity after the end of the socialist regime and through the following political and economic transformations (Devaux 2005; Carmin and Jehlička 2010). However, there was a certain continuity of associational life throughout both historic transformations – in 1948 from democracy to communism, and in 1989, from communism to democracy. The activity-oriented mass organizations that survived from the inter-war democratic period of the Czech state also survived the era of communism, although the communist party controlled them within the compulsory institutional platform of the National Front. For many citizens, volunteering for these kinds of organizations at the community level, in the spheres of culture, sport, and recreation in the 1980s, provided an exit strategy from the state ideology. However, it may also be that activity in these organizations promoted particularized trust, not generalized social trust. Only members within certain closed groups could be trusted, while participation in the open public sphere was, for most people, dangerous or impossible because of the potential clash with the authoritarian regime.

After the fall of communism, ‘new’ civic organizations actively redefined the concept of volunteering and promoted democracy and the concept of civil society. In contrast, some of the old-interest organizations were more associated with the power structures and the clientelist networks of the former regime. Some sports, culture, and youth organizations dating back to the inter-war period still receive ‘per member’ funding at the national level through the old type of closed allotments from certain ministries; these organizations are not dependent on winning competitions for grants or project funding. Such closed support structures for mass organizations means that they have little interest in the recent civil society discourses. Consequently, one might assume a huge difference in the type and level of trust of people that choose to volunteer for ‘new’ or ‘old’ type of organizations.

Norway experienced no similar disruptive period. The suppression of the Nazi-occupation of civil society was quickly reversed. However, volunteering for different categories of organizations may still have different implications for trust. First, there are ‘self-oriented’ organizations where people volunteer to defend their own material interests. This includes economy, housing, and professional associations and trade unions. Second, there are more ‘other-oriented’ organizations, such as humanitarian, political, and environmental organizations. Third, there are ‘activity-oriented’ culture and recreation organizations, where companionship and entertainment are the main purposes (Janoski and Wilson 1995). Such different organization types may bring people into interaction with each other on different terms and in contact with the public administration and support arrangement in different ways. The nature of such interaction and contact may have very different consequences for social and institutional trust.

Norway and the Czech Republic share similarities: there are high levels of volunteering, and the culture and recreation field is dominant. However, the communist regime disrupted the development of the voluntary sector and democratic institutions in the Czech Republic. This has resulted in a particular role for ‘old’ voluntary organizations that is in line with a broader clientelist political culture of favoritism, observable as high levels of corruption and low levels of trust, as in other post-communist countries (Howard 2003). In contrast, Norway has high levels of trust and low levels of corruption. The two countries are well-matched cases for a comparative analysis, disentangling the effects of volunteering and institutional/organizational context on social and institutional trust.

This section compares individual survey data to see to what extent volunteering for voluntary associations in general, or for different types of associations in particular, has an impact on generalized social trust and trust in political and administrative institutions. It investigates whether volunteering may be associated with social and institutional trust, depending on institutional contexts (Norway vs. Czech Republic) and with types of organizations (culture and recreation, welfare, civic and advocacy, or interest organizations, in addition to ‘old’ and ‘new’ in Czech Republic).

The comparison data originate from two representative population surveys carried out in 2009 with identical questions about volunteering for different types of organizations, social and institutional trust, and social background. The first results and general descriptions of the surveys were published by Wollebæk and Sivesind (2010) and Frič and Pospíšilová (2010). The Norwegian survey analysis was based on telephone interviews Statistics Norway did with 1579 respondents aged 16–79. Added to the sample was a weighed-in sub-sample of 359 respondents from Africa and Asia (first and second generation) that had been living at least 5 years in Norway. The Czech survey analysis was based on face-to-face interviews with 3811 respondents 15 years and older. The analysis used respondents 79 years and below to match the Norwegian sample.

The main reason for using these data sets was the unusual detail of the questions about volunteering for different types of voluntary organizations. Without probing for several types of organizations, respondents would not recall all relevant instances of volunteering. The respondents were asked if they volunteered for organizations in a certain area at least once during the last 12 months; altogether the survey inquired about 15 areas in Norway and 18 areas in the Czech Republic, based on the ICNPO-categorization (International Classification of Non-Profit Organizations). Eleven of these organizational categories were similar in the Norwegian and Czech surveys. However, since the voluntary sectors were different, certain country-specific categories had to be included to capture the full specter of volunteering in each country (see organization categories in the Appendix). In addition, a question about ‘virtual volunteering’ – volunteering for an internet society or discussion group – during the previous 12 months was included in both surveys. For example, this could mean guiding new members, moderating discussions, or maintaining and developing web sites – in other words, facilitating other people's participation in the same way as many types of real world volunteering do. In Table 1, a dichotomous variable that includes volunteering for all kinds of organizations is the dependent variable. In Table 2, volunteering is divided into five main types of organizations. The Appendix shows the sub-categories of organizations included in each of the five main types.

Institutional trust and social trust alternated as dependent and independent variables through the different regression models. Institutional trust was measured by seven self-placement scales with five points each, ranging from very high trust to no trust about trust in municipal authorities, parliament, courts of law, government, police, public administration, and voluntary organizations. The variable used in the analysis is an additive index of these seven questions with theoretical scores from 0 to 10; the reliability was .82 in the Norwegian survey and .91 in the Czech survey (Cronbach's alpha).³

Generalized social trust was measured by three self-placement scales (0–10) consisting of contradictory statements about trust in other people. The variable used in the analysis was an additive index of these three questions with theoretical scores from 0 to10; the reliability was .65 in the Norwegian survey and .83 in the Czech survey (Cronbach's alpha). Although the reliability of social trust in the Norwegian data is a little weak, similar measures are components in much research on social capital. Because of the theoretical importance and the widespread use of general social trust variables, it is important to focus on this variable here as well.

Passive membership may have an impact on trust. In a classic study of the civic culture, Almond and Verba observed that any membership – passive or active – had an impact on political competence because pluralism was an important foundation of political democracy (Almond and Verba 1965 [1963]: 265). However, it is possible that, since passive membership does not entail spending time in organizational activities or face-to-face with other members, the effect could primarily be a result of self-selection. Those who decide to be members of an organization may do so because they already have political competence or institutional trust. Being a member without volunteering in any organization was added as a control variable.

In addition, the analysis controlled for gender, age (four levels), level of education (three levels), and individual income after tax (quintiles) to see if the effects of volunteering on trust could be spurious.

Table 2 first presents regressions of social trust on volunteering for voluntary organizations or on the internet (OLS, unstandardized coefficients). In step 2, the additive index for trust in institutions was added, to see if the effect of volunteering on social trust changed. According to the institutional approach, social trust and volunteering should primarily be effects of trust in institutions. If this were the case, one would expect to see a diminished effect of volunteering on social trust in step 2. We also checked for interaction between volunteering and institutional trust, but included the interaction variable only when it had a statistically significant effect.⁴ Step 3 added control variables: passive membership, gender, age, education, and individual income. The lower half of the table presents the same regression models, except that institutional trust is the dependent variable and social trust is the independent variable in steps 2 and 3.

The trust variables added in steps 2 and 3 were important because people with high levels of trust could probably more easily be persuaded to take part in volunteering. To be more trusting in general could interfere, therefore, when observing the effect of volunteering on social and institutional trust. However, one would expect that trusting people with high levels of institutional trust would also have high levels of social trust and vice versa. In addition, one would assume that trusting people more often would also include passive members of voluntary organizations. Furthermore, trust often relates to gender, age, high income, and high education. Thus, the plausibility of a recruitment effect was reduced, but not completely ruled out, by the control variables added in steps 2 and 3.

An important first finding was that the level of trust was very different in Norway and the Czech Republic. The first half of Table 2 shows regressions of social trust on volunteering. The constant in step 1 is 6.48 in Norway and just 4.54 in the Czech Republic. This shows the average level of social trust on a scale of 0–10 for people that do not volunteer. The difference is similar to the difference in average social trust scores in the European Social Survey (Fridberg and Kangas 2008). Even if volunteering has a much stronger effect on social trust in the Czech Republic (.92**) than in Norway (.21**), this is not enough to close the trust gap between the two countries. After adding trust in institutions in step 2, the effect of volunteering in Norway is much weaker (.12^a), significant only on 10% level. This is in line with expectations from the institutional approach. The effect of volunteering on social trust seems largely to be an indirect effect of trust in institutions. In the Czech Republic, however, the effect of volunteering on social trust in step 2 is much stronger (.73**) and still significant. In step 3, the direct effect of volunteering increases to 1.35** after adding control variables and interactions with institutional trust. However, even without the interaction variable, the effect of volunteering would have been .74**, which is similar to step 2. The effect of institutional trust is equally strong in steps 2 and 3 (.47** and .50**). The control variables in step 3 do not change this picture. Thus, the difference from step 2 seems to be due to a weak but statistically significant negative interaction between volunteering and institutional trust (−.13*). These results mean that in the Czech Republic, volunteering has a stronger positive effect on social trust for those who have little trust in institutions than for those who have high institutional trust. This could be because motivation for volunteering is opposition to the political culture for those who have little institutional trust. Through volunteering, they meet people with similar views and this increases their social trust. Some people may even try to use volunteering for certain organizations to build a more active and alert civil society.

Thus, volunteering has a strong positive effect on social trust in the Czech Republic, which is not in line with the institutional approach, since the experience of volunteering increases social trust, even if institutional trust is low. Nor is it in line with the common assumption that participation in voluntary organizations does not produce general social trust because it has deep roots in individual psychology and upbringing (Uslaner 2000). However, the social interaction-theory is a possible explanation: through volunteering people experience that cooperation about the production of collective goods is possible, and this enhances social trust. However, in contrast to Putnam's theory, instead of a positive reinforcing effect, there is negative interaction between volunteering and institutional trust. People who have high institutional trust and volunteer have lower social trust than those who have low institutional trust and volunteer. In other words, there is a negative spillover from volunteering and institutional trust on social trust not a positive spillover as would be expected. Thus, the social interaction effect of volunteering alone cannot explain social trust in the Czech case. Rather, one needs to take into account that in the Czech Republic, there are good reasons not to trust institutions, and that volunteering has very different meanings according to how it relates to the institutional context.

The findings concur with the study's expectations that experiences from collaboration in voluntary organizations could have a positive impact on social trust in a low-trust context such as the Czech Republic, where people have low institutional trust. In Norway, with a high trust context, however, there is no significant effect from volunteering. On its own, the Norwegian case can be accounted for by the institutional theory, which assumes that social trust is primarily an effect of institutional trust. However, neither the socialization perspective of Putnam nor the institutional approach of Rothstein suffices to explain the differences between the Czech Republic and Norway. One must take into account that institutional/organizational contexts may change the effect of volunteering on social trust, from one case to another.

The second half of Table 2 shows regressions of institutional trust on volunteering. The constant in step 1 is 6.30 in Norway and just 4.44 in the Czech Republic. The average level of institutional trust for non-volunteers is much higher in Norway, similar to Norway's level for social trust. However, volunteering has a very different effect. The positive effects of volunteering on trust in institutions are significant and on almost equal levels in Norway (.35**) and the Czech Republic (.41**) in step 1. In Norway, the effect is only slightly weakened and still highly significant after adding social trust in step 2 and control variables in step 3. In the Czech Republic, in contrast, the effect of volunteering disappears when social trust is added. There is a statistically significant effect of passive membership in step 3, but it does not alter the effect of social trust or the lack of effect from volunteering. There are no statistically significant interactions between volunteering and social trust in either country. The positive effect of volunteering on trust in institutions is independent of social trust in Norway but not in the Czech Republic – the opposite of the findings when analyzing social trust as a dependent variable. Again, it is difficult to see how either the institutions theory or the social interaction theory could explain the differences between the two countries without considering the institutional/organizational context.

Volunteering for all types of organizations seen together may have very different consequences for the levels of trust in Norway and the Czech Republic. The following section explores whether one can attribute the effects of volunteering to certain organization types, and how those different types are linked to the institutional contexts in Norway and the Czech Republic.

Table 3 presents the same regression models as Table 2, except that volunteering is divided into five main types of organizations. The first part of this typology is based on Janoski and Wilson's differentiation between ‘self-oriented organizations’ that promote the members’ self-interest, often in connection with material and economic conditions, and ‘other-oriented organizations’ that produce services and goods for externals, whether that means a particular social group, as disabled persons, or society in general. Janoski and Wilson also suggest that there may be a third category of organizations. In these organizations, companionship and entertainment is the main purpose (Janoski and Wilson 1995; Janoski 1998). In this study, these categories are called interest-, other-, and activity-oriented organizations. In addition, welfare organizations that comprise education and research, and health and social services, are a fourth category. Volunteering on the internet is the fifth category because it engages population-groups that increasingly do not volunteer for voluntary organizations: young, urban males, on welfare and with interest in politics (Wollebæk and Sivesind 2010). This may indicate that the reasons for volunteering on the internet, and hence the effect on trust, could be different from volunteering for voluntary organizations.

Organizations belonging to each of the five categories have a common orientation towards society and a common relationship to volunteers as a primary group of internal stakeholders. Stakeholder relations define an action space for the organization. The main categories of organizations have differently structured action spaces within which they may coordinate stakeholder expectations (Enjolras 2000). This means that the experience of volunteering for an organization is shaped by its room of possible actions, which depends on stakeholder relations. The state may be a contract partner for welfare organizations, a source of funding for activity-oriented organizations, a lobbying-target for interest-oriented organizations, or a partner in policymaking for other-oriented organizations. However, such stakeholder relations may vary from country to country, and this affects what implications volunteering for that organization-category could have for trust.

In the Czech Republic, it is important to recognize that some ‘old’ organizations are associated with power structures and clientelist networks that have survived from the former communist regime. In contrast, ‘new’ organizations may try to redefine volunteering and promote democracy and civil society. This aspect of the Czech voluntary sector is critical for the analysis of social and institutional trust. However, there are no similar categories of organizations in Norway, since there has been no communist disruption of the development of the voluntary sector. However, as the Appendix shows, there is a large degree of overlap between the ‘other-oriented’ organizations and the ‘new’ organizations in the Czech Republic, and between ‘interest-oriented’ organizations and the ‘old’ organizations in the Czech Republic. Table 3, therefore, compares the effects of these types of organizations on social and institutional trust, without assuming that they are identical in Norway and the Czech Republic. It is important to take into account that differences between these two countries imply that the voluntary organizations have very different characteristics and social roles. Indeed, that is the object of this study.

Table 2 showed that there is a positive effect of volunteering on social trust in the Czech Republic. Table 3 shows that there are strong positive effects on social trust of volunteering for activity-oriented, ‘new’, and welfare organizations. All these effects were statistically significant after adding institutional trust in step 2 and control variables in step 3. Volunteering on the internet also has a positive effect, but this was weak and no longer significant after adding control variables, which could be a result of a small number of respondents in this category. It is important to note, however, that there is no effect of volunteering for ‘old’ organizations in any step of the regression analysis.

Table 2 showed that there was only a very weak positive effect of volunteering on social trust in Norway. Table 3 shows that there is a positive effect of volunteering for other-oriented organizations, but it is somewhat weakened and no longer significant once the control variables are added. Volunteering on the internet has a strong negative effect on social trust, in contrast to the positive effect in the Czech data. However, it is somewhat weakened and only significant at the 10% level after adding control variables in step 3. This could be a channel for contention in Norway, but too few respondents volunteered in this category to find out more about this effect. It is important to note that such positive and negative effects of volunteering may cancel each other out when different types of organizations are grouped in one volunteering variable, as in Table 2. This may be one of the reasons that several surveys have found only weak or no relationships between volunteering and trust.

Looking at institutional trust as a dependent variable in the lower part of Table 3, one finds that in the Czech Republic, there are no significant effects of volunteering at all after controlling for social trust in step 2, which is similar to the findings in Table 2. In Norway, however, there is a positive effect of volunteering in Table 2. Concurring with this, Table 3 indicates a positive effect of volunteering for activity-oriented organizations and interest-oriented organizations after controlling for all other variables. There is also a positive effect of volunteering for other-oriented organizations, but it is weakened and no longer statistically significant in step 3.

Table 3 narrows down the positive effect of volunteering on institutional trust in Norway to volunteering for activity-oriented organizations, while the positive effect on social trust of volunteering in the Czech Republic can be the result of volunteering for activity-oriented, ‘new’, or welfare organizations. However, it is important to note that there is no effect of volunteering for ‘old’ organizations, more associated with the power structures and clientelist networks of the former regime. The concluding section discusses how these findings relate to the three theories about social interaction, institutions, and institutional/organizational context.

The social context has very different effects on social and institutional trust in Norway and the Czech Republic. In Norway, the universal welfare state and low corruption result in high levels of institutional trust that is a viable explanation for generally high levels of social trust. The institutional context has such a strong positive impact on social trust that volunteering makes little difference; this concurs with the institutional approach. However, surprisingly, volunteering does have a positive effect on an already high level of institutional trust. This is probably not an effect of recruitment, since the positive effect of volunteering on institutional trust persists after controlling for social trust, passive membership, and personal background. It seems more likely that through volunteering, people encounter well-functioning welfare institutions, and this experience increases their institutional trust. This assumption is supported by the finding that a positive effect of volunteering in Norway results primarily from activity-oriented organizations (Table 3). This is a field where the public sector and the voluntary organizations have common interests in encouraging participation and activity, whereas in other types of organizations, the relationship with the public sector may be more ambivalent or even conflicting. In the culture and recreation field, the state's income from gaming supports sports and cultural activities, in particular sports facilities, local and regional culture-houses, and meeting-places. In addition, three voluntary umbrella-organizations in the fields of music, amateur theatre, and children and youth organizations distribute the monetary public support for local activities. The state and the municipalities also give economic support to some types of voluntary organizations, in particular youth and children's organizations, based on predefined criteria, such as the number of members, geographic coverage, and the organization's statutes and main purpose. In other words, generous support is distributed by general, impartial criteria, or in some cases, by the voluntary sectors’ own umbrella organizations. People who volunteer in voluntary organizations in sports, culture, and recreation see how these support arrangements work in practice, and this is probably why they have a higher level of institutional trust. Consequently, the institutional/organizational context is important in determining the types of organizations that affect institutional trust and the direction in which volunteering takes institutional trust.

Seen in relation to the institutional theory of Rothstein, it is surprising to see a positive effect of volunteering on institutional trust at all, since that theory expects the role of volunteering to be secondary at best. However, the effect of volunteering for activity-oriented organizations can be explained in a similar manner as Kumlin and Rothstein explained social trust by welfare institutions: contacts with universal welfare-state institutions tend to increase social trust, whereas experiences with needs-testing social programs tend to undermine it (2005). Volunteering for sports, culture, and recreational organizations bring people in touch with public support arrangements, and in dialogue with public administration, which increases institutional trust.

In the Czech Republic, less universal welfare arrangements and more clientelism and corruption result in lower levels of both institutional and social trust. Volunteering has no effect on trust in institutions, but it has a strong positive effect on general social trust. The Czech Republic context is not conducive to social and institutional trust in general. Despite this, when people come together and see that volunteering works, this experience increases social trust, in line with Putnam's theory. This explanation gains support from the observation that a positive effect of volunteering on social trust can be observed in organization types less associated with power structures and clientelist networks of the former regime.

Furthermore, when there is a negative spillover effect from volunteering and social trust to institutional trust – in contrast to Putnam's theory about mutual reinforcement – it probably has to do with the reasons people have for volunteering. If people get involved in voluntary organizations because they look for something different from the present regime, its high levels of corruption and clientelism, little accountability and problems with enforcing the law and with justice, then higher levels of institutional trust among volunteers could hardly be expected. Furthermore, the challenges these organizations meet from the establishment, and the raised level of consciousness about social problems that participation in these organizations develop by working for change, may further reduce institutional trust.

Observations of trust related to volunteering for certain types of organizations after controlling for all other variables, indicates strongly that volunteering is a cause for trust and not the other way around. If the primary mechanism were that impartial institutions cause vertical trust, which causes horizontal trust, which can be beneficial for volunteering, as the institutional theory assumes, we would not see effects of volunteering limited to particular types of organizations.

Broad generalizations about a universal, mutually reinforcing relationship between social interaction in networks, generalized trust and civic engagement do not stand up to closer empirical scrutiny, nor do generalizations about the lack of causality between volunteering and social trust. To explain the different effects of volunteering on social and institutional trust within each country and the difference between Norway and the Czech Republic, neither the institutional approach nor the face-to-face effect of volunteering are sufficient. One has to consider the institutional/organizational context to understand how volunteering for certain organization types affects institutional or social trust. This is has important policy implications. If one wants to strengthen trust through volunteering, the types of voluntary organizations one chooses to support make a huge difference. In addition, it is also important to build reliable order institutions (police, legal system, and defense system) (Rothstein and Stolle 2008) and promote equality to avoid a social trap (Rothstein and Uslaner 2005), as demonstrated by previous research.

Weak correlations between volunteering and trust found in some surveys could be a result of the positive effects of volunteering for some types of organizations cancelling out the negative effects of volunteering for other types of organizations. Future research must garner detailed data on volunteering for different types of organizations to be able to observe the mechanisms that link social context, volunteering, and social and institutional trust together.

Karl Henrik Sivesind is Senior Researcher at Institute for Social Research in Oslo (since 2000). His main research focus is on civil society and the voluntary sector, nonprofit welfare services, management in different cultures, and comparative methods. He has been involved in several comparative, social scientific research projects and is currently leading a Scandinavian project on consequences of private and nonprofit service provision for active citizenship.

Tereza Pospíšilová is Assistant Professor at the Department of Civil Society Studies, Faculty of Humanities, Charles University in Prague (since 2004). Her main research focus is on civil society and the voluntary sector, philanthropy, and the processes of post-communist transformation. She has been involved both in academic research projects and in applied research projects answering the needs of Czech nonprofit organizations.

Pavol Frič graduated from the Department of Sociology at the Komenius University in Bratislava (Slovakia). At present he is professor at the Department of Public and Social Policy at the Institute of Sociological Studies, Charles University in Prague, Faculty of Social Science. Since the end of 1992 he has lectured on sociology of social problems and civil society. As of 2000 he is senior researcher at the Centre for Social and Economic Strategies also at the Charles University. Current research interests are the scope of civil society, volunteering, elites–public relation and corruption.