Does Participation in Voluntary Associations Contribute to Social Capital? The Impact of Intensity, Scope, and Type
Dag Wollebaek Per Selle
Norwegian Research Center
in Organization and Management
Voluntary associations are often ascribed a fundamental role in the formation of social capital. However, scholars disagree on the extent to which face-to-face contact, that is, active participation, is necessary to create this resource. This article examines the impact of participation in associations on social capital using three dimensions: intensity (active vs. passive participation), scope (many vs. few affiliations) and type (nonpolitical vs. political purpose). Whereas those affiliated display higher levels of social capital than outsiders, the difference between active and passive members is absent or negligible. The only cumulative effect ofparticipation occurs when a member belongs to several associations simultaneously, preferably with different purposes. The article challenges the notion that active participation is necessary for the formation of social capital and suggests that more attention should be paid to the importance ofpassive and multiple affiliations within associations.
According to Putnam (1993,1995a, 1995b, 2000), it is through experiences of interaction face to face with people from different backgrounds that we learn to trust each other. The voluntary association represents one of the main arenas for interaction of this type. Associations create networks that allow social trust to spread throughout society. Moreover, associations are thought to generate civic engagement and to further the ability of their members to influence public affairs by being "schools in democracy." Consequently, Putnam regards associations as the prime sources for social trust, horizontal social networks, and civic engagement, that is, social capital.
This article addresses this claim empirically. The analyses will investigate five questions, which are all derived from the implications of Putnam's version of social capital:
1.  Does participation in associations contribute to social capital?
Nonprofit and Voluntary Sector Quarterly, vol. 31, no. 1, March 2002 32-61 © 2002 Sage Publications
32
Participation in Voluntary Associations
33
If yes,
2. Does active participation contribute more to the formation of social capital than passive membership? If yes, is active participation necessary for the formation of social capital?
3. Do several affiliations contribute more to social capital than one affiliation?
4. Is the level of activity more important than the number of affiliations?
5. Do members of nonpolitical associations display higher levels of social capital than members of political associations?
Following Putnam, we expect to give confirmatory answers to all of these questions, especially the first and the second. If this is the case, Putnam's main hypothesis—that the overlapping horizontal networks of face-to-face interaction created by associations contribute to the formation of social capital—is strengthened. If this is not the case, a discussion of the implications this may have for the theory is in order.
SOCIAL CAPITAL: TRUST, SOCIAL NETWORKS, AND CIVIC ENGAGEMENT
Putnam (1995b) defines social capital as "features of social life—networks, norms and trust—that enable participants to act together more effectively to pursue shared objectives" (p. 665). The basic idea of the concept is thus to specify the conditions under which collective action is facilitated. In this study, the concept is deconstructed into three major components: trust, social networks, and civic engagement.1
The argument for trust is based on the model of the rational individual. If collaborative action is to be achieved, individuals must trust that the people with whom they are about to cooperate will not act contrary to shared objectives. Thus, trust is seen as a way out of dilemmas of collective action (Putnam, 1993, pp. 163-164). As one of the major preconditions for cooperation, trust is an essential element of any conceptualization of social capital.
Trust exists in different forms. On one hand, a form of thick trust is often present in institutions with a high degree of closure (Coleman, 1988; Newton, 1997; Williams, 1988), such as tribes or social cliques (Scott, 1991). Thick trust between insiders may be related to distrust toward outsiders. Thus, if intercommunity integration is strong without extracommunity linkages, the type of trust generated through interaction can inhibit rather than facilitate collective action and can damage integration at the societal level (Newton, 1999; Woolcock, 1998). On the other hand, there may exist a thin, generalized social trust between individuals who do not know each other very well, if at all. Our concern here is primarily with the latter form of trust, which "presumes that
34
Wollebaek, Seile
the structure of the situation is more important than personal character" (Putnam, 1993, p. 243, footnote 45).
Social networks can be seen as individual resources that facilitate goal attainment—the goal being of either a personal or societal character. This instrumental interpretation of the value of networks is more prominent in Bourdieu's (1985) formulation of social capital than in Putnam's writings. Putnam tends to focus more on how networks generate other types of social capital, namely, trust and civic engagement, through direct interaction between the participating individuals. Networks also act as vehicles for trust, as information about the trustworthiness of other people "become[s] transitive and spread[s]: I trust you, because I trust her and she assures me that she trusts you" (Putnam, 1993, p. 169).
However, the structure of these networks is all important to their relevance for social capital. Their degree of horizontality versus verticality (or power symmetry vs. asymmetry) determines how effectively they generate or disperse social capital. According to Putnam, although lower ranking individuals in a vertical network may act in compliance with orders, there is no symmetrical reciprocity in the relationship: The client may not sanction the patron, and opportunism is likely to occur. Thus, vertical networks contribute little to solving collective dilemmas: "Avertical network, no matter how dense and no matter how important to its participants, cannot sustain trust and cooperation" (Putnam, 1993, p. 174). By contrast, Putnam's view of the typical voluntary association implies a greater degree of horizontality. The association creates secondary networks that foster interaction face to face, with low barriers toward outsiders and with participants concurrently involved in multiple associations. Although Putnam (1993, p. 173) concedes that "even bowling teams have captains," power and domination give way to solidarity and reciprocity as the centerpieces of this image.
To make democracy work in accordance with Putnam's ideals, trust and networks represent only necessary not sufficient preconditions. A society may abound with trust among its citizens, and they may be interconnected with any number of horizontal social networks. Yet, for these resources to be relevant for democracy, individuals also need to care about issues reaching beyond their private lives. They have to possess sufficient knowledge and skills to receive and perceive information and to be able to formulate responses. An active, interested, and informed citizenry is necessary for a participatory democracy to function. For this reason, a third dimension of social capital needs to be examined more closely, namely, civic engagement (Putnam, 1993,
p. 15).
If the civic community with an active and competent citizenry, on which participatory democracy heavily depends, is to be realized, no single component of social capital is sufficient. It is in the conjunction of trust, networks, and civic engagement that Putnam claims to have found the key to how democracy works.
Participation in Voluntary Associations
35
VOLUNTARY ASSOCIATIONS: EXTERNAL AND INTERNAL EFFECTS
How do voluntary associations contribute to the pool of these resources necessary for the functioning of democracy? Putnam distinguishes between associations' internal effects on their participants and their external effects on the political system. Externally, they are seen as an alternative channel of influence. Associations provide institutional links between the members and the political system and serve as intermediary institutions. The articulation and aggregation of interests and values is eased by the presence of a "dense network of secondary associations" (Putnam, 1993, p. 90), as arguments refined through discussion in an association take on a clearer and more precise shape. This improves social collaboration, on one hand, and democratic governance, on the other, thus making democracy more responsive and effective.
The internal effects are more closely related to the formation of social capital at the microlevel. According to Putnam (1993, pp. 89-90), "Associations instill in their members habits of cooperation, solidarity and public-spirited-ness." This is, on one hand, an effect of socialization, in which participants are taught the right "habits of the heart" (Newton, 1999). On the other hand, it is also an educational effect, because participants gain skills and competencies that are important for a democracy. This may concern knowledge about specific political issues, how political institutions function, how to manage an organization, or how to discuss matters in a civilized manner. Thus, associations are often viewed as "schools in democracy."
Which types of participation are most successful in creating such effects? This question is examined along three main dimensions, which will follow us through the empirical analyses: intensity, scope, and type.
INTENSITY OF INVOLVEMENT: THE IMPORTANCE OF FACE-TO-FACE INTERACTION
One of the most important changes in the voluntary sector over the past decades, which appears to be a near universal trend in Western countries, has been the growth of what Putnam calls "tertiary associations" at the expense of "classic secondary associations" (Maloney, 1999; Putnam, 1995b; Selle & Stromsnes, 2001; Skocpol, 1999a). Tertiary associations are characterized by centralized, paid-staff leadership; they tend to be nondemocratically structured; and the support of the members tends to be channeled through money rather than time. This implies a trend in which the intensity of associational involvement is declining.
Putnam (1995b) quickly dismisses the growth of tertiary associations as a potential countertrend to the decline of social capital in the United States. The main reason given is that, although the members may feel a common attachment to symbols and values, they do not interact face to face. In Putnam's
36
Wollebaek, Seile
view, social capital can only be formed and transmitted through direct interaction. Furthermore, tertiary groups consist of vertical rather than horizontal networks. Their structure is better characterized as client-patron relationships than as interaction among equals (Maloney, 1999, p. 109), and "two clients of the same patron, lacking direct contact, hold nothing hostage to each other" (Putnam, 1993, p. 175). Therefore, due to the low level of active participation of members and the vertical organizational structure, tertiary associations do not contribute to horizontal networks or social trust.
Putnam does admit that tertiary associations may have important external effects on the wider polity. Passive supporters play an important role in providing such associations with economic support and numerical strength, which gives more weight to their arguments. However, the internal effects on the participants are thought to be weak or absent. In sum, therefore, their contribution to social capital is not comparable to associations within which active members interact face to face.
This leads Putnam to dismiss passive support as a source of social capital, which may be problematic on several counts. First, Putnam does not directly address secondary associations within which many or even most members are passive. In many countries outside the United States, this is very common. The high amount of passive memberships is one of the main structural characteristics of the voluntary sector in Scandinavia and the Netherlands (Dekker & van den Broek, 1998). In Norway, passive members outnumber volunteers by three to one, and in many humanitarian, advocacy, or political organizations, passivity is the rule rather than the exception (Wollebaek, Selle, & Lorentzen, 2000b). Although tertiary organizations are gaining ground in Norway too, most associations are still of the classic secondary form, in the sense that they rely on the efforts of volunteers and are democratically structured internally.
Thus, passive members may belong to institutions within which large stocks of social capital are embedded. Although empirical evidence is scanty, some studies do suggest that passive members feel neither alienated nor disconnected from these social systems (Maloney, 1999; Selle & Stromsnes, 1998; Stromsnes, 2001; Wollebaek, 2000). If Putnam's emphasis on face-to-face contact is correct, the level of social capital should nevertheless be observably higher among the core of activists than among passive supporters, even within the same association.
Moreover, one of Putnam's main sources of inspiration, Almond and Verba's (1963) The Civic Culture, attributes an importance to passive memberships that exceeds their external effects. In Almond and Verba's classic study, passive members displayed a significantly higher level of civic competence than nonmembers across five countries. Passive supporters were also more supportive of democratic norms than outsiders, and they shared a higher sense of political efficacy.
This suggests that passive membership may indeed have internal effects on associations' participants, a notion that is corroborated by other studies. Godwin (1992, as cited in Maloney 1999, p. 113) emphasized that even the
Participation in Voluntary Associations
37
most passive forms of participation, financial support through check writing, "reduces political alienation, as contributors believe their contributions make a difference. This, in turn, reduces the support for aggressive political participation." This effect is related to the durability of involvement: The longer individuals remain in their passive roles, the more influence the affiliation has on their attitudes.
Furthermore, why should a sense of community or identification with a cause, which may be conducive to trust in compatriots, presuppose contact face to face? The theory of imagined communities provides a clue to how social capital of this kind can develop without personal interaction (Anderson, 1991; Newton, 1999; Whiteley, 1999). The imagined community is a group one feels a psychological affinity to, even though it is too large to allow face-to-face contact between all of its members. Therefore, imagined communities are based on an abstract form of trust rather than on thin or thick trust (Newton, 1999). In this vein, Whiteley (1999, p. 31) shows that those who most strongly identify with the imagined community (i.e., they express patriotism) are more likely to express a generalized sense of trust in other people than are individuals whose patriotism is weaker.
Even though in many cases the emotional bonds between passive members and associations will be weaker than those between a citizen and a nation, which is the typical example of an imagined community (Anderson, 1991, p. 6), the two types of relations clearly share important characteristics. Associations with passive members and nations are both social systems that are too large to allow face-to-face contact, but their members, or citizens, may still feel a common affinity to symbols or values or share a commitment to a cause. To the degree that individuals hold overlapping memberships in associations, their sense of identification and abstract trust may be transferred to several contexts, and possibly to society as a whole.
The suggestion that low-intensity participation may also have internal effects raises the question of whether the role of passive memberships in the formation of social capital may be understated. This necessitates a more detailed empirical investigation into the nature of the effect of participation.
SCOPE OF INVOLVEMENT: THE IMPORTANCE OF MULTIPLE AFFILIATIONS
It is possible to identify two ways in which multiple affiliations may contribute more to social capital than singular affiliations do. These can be labeled a moderating and a cumulative effect.
First, at the societal level, the more overlapping and interlocking networks that exist, the higher the probability that people from different backgrounds will meet. To the extent that associations create horizontal networks that span underlying cleavages, they may have a moderating effect on the level of conflict in societies. This occurs as a result of cross-pressures experienced through
38
Wollebaek, Seile
participation in multiple networks (Putnam, 1993, p. 90). When interacting with individuals of diverse backgrounds, goals, and preferences, each person is forced to moderate his own attitudes to create a lasting social contract in the different settings. Dense, overlapping, and interlocking networks thus contribute to compromise and negotiation where there might otherwise be warring of factions and centrifugal, escalating conflict. It is hypothesized that associations are particularly able to generate cross-cutting, multiple networks because the relationships between those involved are characterized by weak ties of relatively low intensity (Granovetter, 1973).
Second, at the individual level, multiple affiliations mean more and broader interaction. Consequently, it should have a cumulative effect on the level of trust and civic engagement and should extend the scope of networks, which eases their diffusion into society. Almond and Verba (1963, pp. 264-265) found the number of memberships held by an individual to affect civic competence cumulatively: "Membership in one organization increases an individual's sense of political competence, and membership in more than one organization leads to even greater competence." In fact, the number of multiple affiliations discriminated more clearly between the "civic" and "less civic" countries in their study than the proportion of the population holding memberships or the level of membership activity.
Intensity and scope do not always move in accord, and although they are related, they should be treated as two separate aspects of participation. On one hand, intense involvement in one association may foster an interest to take part in others. On the other hand, highly intense involvement may build barriers toward outsiders and, consequently, narrow the scope of the networks created.
Although Putnam emphasizes both scope and intensity of involvement, it follows from the weight attached to face-to-face interaction that intensity has first priority. Passive memberships in several associations are seen as less productive for building social capital than active membership in one association. In the analyses to come, this proposition will be put to the test.
DOES TYPE MATTER?
Until now, the discussion of the impact of participation in association and social capital has focused on form rather than content. However, given the immense variety of different purposes for which associations exist in Western countries, the question of which activities the participants are involved in also needs to be taken into consideration.
At a glance, Putnam's (1993, p. 90) contention seems to be that the type of association is virtually irrelevant for the extent to which social capital is developed, which is identical to Almond and Verba's (1963, p. 265) assertion that memberships even in nonpolitical associations lead to a more competent citizenry. However, if type means little in Putnam's work, structure means
Participation in Voluntary Associations
39
everything. The three demands of horizontality, face-to-face interaction, and ability to transcend subcultural barriers will be met more often by nonpoliti-cal, leisure-oriented associations than by associations with a manifest political purpose. Politically oriented associations are often products of cleavages in society, centralized in structure, with a large minority or even a majority of passive members.
Thus, in Putnam's understanding of how to make democracy work, a choral society or a bird-watching club plays a vital role, whereas the value of social movements, labor unions, and political parties is downplayed (Foley & Edwards, 1996). This somewhat counterintuitive notion stems directly from Putnam's (1995a, p. 71) distinction between external and internal effects, of which the latter appears to be given priority.
This view is, naturally, contested. With regard to the internal effects of participation, Quigley (1996, p. 3) claimed thatnonpolitical associations cannotto the same extent as organizations contesting state authority foster the civic skills necessary to promote social capital and strengthen democracy. Foley and Edwards (1996) and Rueschemeyer (1998) emphasized that externally, nonpolitical associations cannot play the role of a counterweight to the state like political associations or social movements do. Finally, Skocpol (1999b) and Selle and Stromsnes (2001) questioned the degree to which associations can act as intermediary institutions between the individual and the state, as long as the associations are nonpolitical in purpose and often purely local in structure.
If Putnam's contention is correct, we might expect that members of political associations display lower levels of social capital than do members of associations without an expressed political purpose. Therefore, in the analyses to come, the impact of the type of associations participants are affiliated with will be subjected to empirical examination.
OPERATIONALIZATIONS AND DATA
OPERATIONALIZATION OF PARTICIPATION IN ORGANIZATIONS
The independent variables are measurements of the scope, intensity, and type of association involvement. Scope is operationalized as the number of associations with which each individual is affiliated. The scope of the involvement includes both memberships and volunteering. If a person has reported membership and volunteering for the same association, the association is only counted once.2
The intensity of the involvement is operationalized as the combined amount of time spent participating and volunteering in associations over the past year. This provides a reasonably accurate measure of the extent of face-to-face contact to which the member or volunteer has been exposed. It should be
40
Wollebaek, Selle
observed that one additional hour of participation does not necessarily mean the same for highly active participants as for the more passively affiliated. Therefore, rather than introducing a linear measure of the number of hours spent participating, the variable aims at dividing the sample into groups of approximately equal size. In most of the following analyses, those affiliated with associations will be divided into four groups, each representing about one quarter of the members or volunteers. Those who are members of one or more associations but have not spent any time participating are classified as completely passive and given the lowest value. Those who have spent up to 50 hours in the past year are classified as relatively passive, whereas a level of 150 hours (3 per week) in the past year serves as the cutoff point between the relatively and highly active.
To compare the impact of intensity with the impact of scope, a composite variable representing four different modes of participation is introduced, based on the number of affiliations and time spent participating. Both variables are dichotomized: Intensity is dichotomized between active (1 hour per week or more) and passive (less than 1 hour per week), and scope is dichotomized between singular (one) and multiple (two or more) affiliations. If intensity is more important than scope, a higher level of social capital should be observable among those who are active in one association than among those passively affiliated with several.
Type is operationalized as the main purpose of the association or associations with which a person is affiliated. The degree of political involvement will be the main dimension in differentiating between associations of different purposes. On one hand, we have the pure nonpolitical associations, among which sports associations assume a dominant position but among which are also included music groups (e.g., choirs, bands), hobby associations (e.g., bird-watching, model airplanes), and fraternity groups (e.g., Masons' lodges). Religious associations are rarely involved in politics in Norwegian society. However, they are sufficiently different from leisure-oriented associations, both in purpose and structure, to deserve a separate subcategory of the nonpo-litical type. At the other end of the scale we find the organizations whose main purposes are political, namely, political parties and unions. In between, there are a number of associations with a more or less manifest political purpose: for example, local community associations, humanitarian associations, environmental associations, or advocacy groups for disabled persons. These are given the somewhat unsatisfactory "semipolitical" label. They will tend to be more involved in politics than, for example, a choir, but their interactions with decision makers will in most cases tend to be less frequent than is the case with political parties or unions.
As many people are affiliated with more than one type of association, two additional categories are in order: Those who are affiliated with both a nonpo-litical and a semipolitical association are classified in the first combinatory type, whereas those affiliated with nonpolitical or semipolitical associations and a political association are grouped in the second type.3 In the regression
Participation in Voluntary Associations
41
analyses, the distinction between political and nonpolitical purpose is captured by means of a dichotomized variable distinguishing those persons who are affiliated with a nonpolitical leisure association from those who are not.
Through a combination of the three measures presented above, we hope to cover the three aspects of the participatory network of each individual that, following Putnam's theory, are the most important to the formation of social capital: scope, intensity, and type. The next step is to operationalize the components of the dependent variable: social capital.
OPERATIONALIZATION OF TRUST, SOCIAL NETWORKS, AND CIVIC ENGAGEMENT
Social trust is measured by a single question: "In general, would you say that most people can be trusted or that you can't be careful enough in dealing with other people?"4 This common operationalization of social trust has been widely used in empirical studies, including Putnam's (1995a, 1995b) own.
Social networks are operationalized as groups outside the sphere of family and relations that are considered to be important parts of one's social circle. In the questionnaire, respondents were asked to check a box for each of five groups they considered to be part of their social network.5 From this, two measures of social networks were computed. First, the breadth of networks was operationalized as the number of different groups that a respondent considered to be part of his or her network. Second, a dummy variable representing the presence or absence of friends obtained in the current situation, that is, either current colleagues, fellow students, or neighbors, was computed. This indicator intends to measure the degree to which involvement in associations contributes to the construction of new social networks.
Finally, civic engagement is operationalized as the level of expressed civic and political interest. To measure this, three items are used, both independently and in a composite index: voting behavior, readership of news material in daily newspapers, and political interest. The second and third indicators both measure knowledge and resources that may be activated to act collectively in a democracy. In this respect, the proportion of respondents voting at every parliamentary election can be said to be in a different category, because voting is a political act, not a resource. In this context, however, voting behavior is used as a measure of a minimum level of political interest, which is mainly used to distinguish between those who are moderately engaged in civic matters and those who are more oblivious to their political surroundings. From these three indicators, a composite index with seven values was cre-ated.6
In summary, the indicators chosen to measure the subdimensions of social capital are either identical to the ones that Putnam uses (social trust, voting behavior, and newspaper readership) or directly derived from his theoretical work (social networks). It therefore seems sound to conclude that the definitory validity of the indicators used is high.
42
Wollebaek, Selle
CONTROL VARIABLES
The control variables include standard sociodemographic variables (age, education, gender, marital status), and factors that may represent alternative sources of social capital (full-time employment, time living at current place of residence, and population density).
THE DATA
The analyses are based on a nationwide survey carried out by the Norwegian Centre of Research in Organization and Management (LOS Centre) in 1998, as part of the Johns Hopkins Comparative Nonprofit Sector Project (Wollebaek, Selle, & Lorentzen, 1998). A survey on giving and volunteering (SGV) was administered by means of mailed questionnaires to 4,000 randomly selected Norwegians aged 16 to 85. The respondents were contacted four times: once by postcard 1 week before the questionnaire was mailed out, once by the mailing of the questionnaire itself, and twice by follow-ups. The last follow-up was carried out by telephone. In all, we received 1,695 valid responses, which equals a response rate of 45% (adjusted gross sample). This response rate is somewhat lower than is the case for most mailed surveys in Norway. The probable reason for this is the combination of a relatively long questionnaire with the relatively low salience of the topic.
A low response rate increases the probability of systematic sample biases. In this case, as in many other postal surveys, middle-aged individuals with higher education are overrepresented. There are no biases with regard to gender or county of residence. Weighting by age and education produced almost identical results to the unweighted results reported in this article, thus increasing our confidence that our findings have not been decisively influenced by biases in the sample.7 Another potentially problematic aspect of a survey such as ours is that the most active participants in associations may tend to respond more frequently than the less active. There is no observable tendency in this direction. The overall number of volunteers and members in most categories of associations corresponds well with other comparable sources (Wollebaek,
2000, p. 48).
EMPIRICAL ANALYSIS
The propositions presented above are now put to the test, and the importance of participation in associations for the formation of social capital is analyzed. First, for each indicator, a graph indicating the importance of scope and intensity is presented. The relative importance of these two dimensions of participation is also compared by using the composite index presented above. Second, we analyze whether it matters which type of association one is affiliated with. This is done by simple cross-tabulations or comparisons of means. Third, by means of regression analysis, the impact of participation is contrasted
Participation in Voluntary Associations
43
with and adjusted for the potential impact of other sources. To begin with, we examine whether participation has an effect at all, using the entire sample. Thereafter, the respondents who are affiliated with associations are selected to examine which types of participation are most conducive to social capital.
TRUST
The relationship between social trust and participation in organizations is illustrated in Figure 1. The results indicate that there is, in accordance with expectations, a clear relationship between participation and social trust. Although only slightly more than half of those not affiliated with any association answer that "most people can be trusted," the corresponding ratio among members and volunteers is almost three out of four. The percentage of "trusters" increases slightly with level of activity, with 79% among the highly active compared to 70% among the completely passive. The curve representing the scope of the participation rises sharply until three affiliations, where it flattens out.
However, the difference between joiners and nonjoiners is much larger than that between active and passive members. Even the completely passive, that is, those who have not spent a single hour on associational activities over the past year and therefore have not been exposed to any face-to-face contact, are much more trusting than the outsiders.
Furthermore, contrary to expectations, the scope of the participation appears to be a more powerful predictor of the level of trust than its intensity (see table below Figure 1). Passive members with multiple affiliations are more trusting than active members with only one affiliation, and they trust other people to the same extent as active members with more than one affiliation. The difference between active participants and passive supporters is not significant when number of affiliations is held constant. Scope, on the other hand, does matter. When activity level is held constant, number of affiliations still has a significant impact.
Another central tenet of Putnam's formulation of social capital is the importance attributed to nonpolitical associations, which tend to be more horizontally structured and to involve more face-to-face contact than associations with a manifest political purpose. Thus, we expect that those affiliated with associations engaged in nonpolitical activities, that is, culture and recreation, display the highest level of generalized social trust.
Table 1 does reveal significant variations between supporters of different types of associations but not in the direction predicted by Putnam. Members of all types of associations are more trusting than nonmembers, but the proportion of trusters in associations active in fields of culture and recreation is, surprisingly, slightly lower than in the population as a whole.
Again, the scope of the participation appears to be the most important factor. The highest level of trust is found among those not only affiliated with more than one association but with several associations of different types. Four out of five members or volunteers in the two composite categories trust
44
Wollebaek, Seile
85 T 80 75 70
65 -60 -
55 50
45 40
Highly active (N=341)
# Completely
passive (N=258)
Not a
member/ volunteer (N=368)
85 T 80
70 65
CD -q
3   60 4
_0) q.
o
CD
o.
55 50
45 -40
2 affiliations (N=307)
3 affiliations (N=302) (N=208)_
1 affiliation (N=469)
♦  Not a
member/ volunteer
(N=368)
CD
3
b
CD
0)
q.
o
q.
(J) O
o
Intensity/scope-index and social trust (per cent)
	"Outsider"	"Passive"	"Active"	%-diff.
	(No affiliations)	(0-1 hr./week)	(more than 1 hr./week)	(sig.)
	54			
1 affiliation		67	70	-3
		(N=310)	(N=159)	(.700)
2 or more affiliations		79	80	-1
		(N=299)	(N=518)	(.143)
	%-diff.	-12	-10	
	(sig.)	(.004)	(.022)	
Figure 1.   Participation in Associations and Social Trust
other people, compared to two thirds of the supporters of nonpolitical associations and slightly more than half of those not affiliated with any associations.
Participation in associations does appear to have an impact on social trust. However, this relationship may be spurious, and the effect of participation needs to be compared with the impact of other possible sources of social capital. In the regression analysis shown in Table 2, the effect of participation is adjusted for the impact of the selected control variables. Model 1 includes the entire sample, whereas Models 2 through 6 include only those affiliated with associations, in order to examine variations between different forms of participation.
The results in Model 1 confirm that participation has an impact on social trust even when controlled for sociodemographic characteristics. In fact, when the entire sample is included, affiliation with associations ranks above education as the most important independent variable. Models 2 and 3 show that both level of activity and number of affiliations have an impact on trust. However, when both are included in the same equation (Model 5), the effect of activity level disappears. The dummy variable representing whether the respondent is affiliated with a nonpolitical association does not have an effect
Participation in Voluntary Associations
45
Table 1.   Percentage of Respondents Who Say Most People Can Be Trusted, by Type of Association
Type of Association					%	D
Not affiliated					54	368
Nonpolitical, culture and recreation					68	346
Nonpolitical, religious					73	52
Semipolitical organizations					66	113
Political organizations (parties and unions)					77	137
Both nonpolitical and semipolitical					81	195
Both non- or semipolitical and political					82	435
Total members/volunteers					76	1,278
Total sample					71	1,654
Note: %2 = 35.017; significance = .000 (only members and volunteers included).						
Table 2.   Regression Analysis of Social Trust (ordinary least squares)						
	Entire					
	Sample		Members/Volunteers Only			
Model 1		Model 2	Model 3	Model 4	Model 5	Model 6
Women (1)	.002	.04	.03	.04	.04	.03
Age (years)	.06*	.05	.03	.05	.04	.04
Education (1-3)	.16**	.18**	.17**	.19**	.17**	.17**
Population density (1-3)	.02	.03	.04	.03	.04	.04
Residential stability (1-5)	.001	.03	.04	.04	.04	.04
Working full-time (1)	.03	.01	.01	.01	.01	-.01
Married/cohabitant (1)	.08**	.08*	.08*	.08*	.08*	.08*
Affiliated with association(s) (1)	.16**					
Level of activity (1-4)		.07*			.05	.05
Number of affiliations (1-4)			.09**		.07*	.07*
Nonpolitical association (1)				.03		-.01
R2 (adjusted)	.076	.046	.048	.042	.049	.048
Note: N = 1,452 for Model 1; n = 1,135 for Models 2 through 6, listwise deletion of cases. Dependent variable = social trust, coded as follows: 2 = most people canbe trusted; 1 = Don'tknow; 0 = You cant be careful enough when dealing with other people. *p < .05. **p < .01.
on social trust, either alone or together with other participation variables (Models 4 and 6).
The insignificant effect of age in Models 2 through 6 above deserves a more detailed treatment, partly because of the weight attached to generational differences in social capital in Putnam's recent work. Table 3 gives a more detailed account of the relationship between age and social trust in the Norwegian setting.
The above results clarify, first, that there is no linear association between age and social trust and that there are few indications of a "long civic generation" in the Norwegian setting. The youngest age group is the least trusting,
46
Wollebaek, Seile
Table 3.   Percentage of Sample Responding That Most People
Can Be Trusted, by Age Group and Intensity of Participation
Age Group	Outsider	Passive	Active	All	n
16-24 years	38	54	70	55	211
25-49 years	60	74	80	73	813
50-66 years	55	81	80	76	344
67 years and older	55	71	79	69	172
followed by the oldest, whereas the highest level of trust is found in the middle age groups. Second, participation in associations has a particularly strong effect among young people. Only 38% of youths not involved in any associations say that most people can be trusted, compared to more than two thirds among young active members. There are virtually no differences between various age groups among the activists. Third, active participation, compared to passive support, plays a more important role in generating social trust among young people than in any other age group. Among those aged 50 to 66, the passive supporters are as trusting as the active members.
In summary, the results substantiate Putnam's notion that participation in associations contributes to social trust. The results do not support, however, his emphasis on face-to-face contact. Passive affiliations appear to be of slightly less importance to the development of social trust than active ones, but multiple affiliations more than compensate for the lack of face-to-face contact. The youngest participants represent a notable exception to this general trend. Furthermore, participants in nonpolitical associations do not display a higher level of social trust than the population as a whole. Generally speaking, bowling in organized leagues has an impact on social trust only when accompanied by affiliation with associations with other, preferably more politicized, purposes. The analyses above show, first and foremost, that belonging to several associations, regardless of degree of face-to-face contact or the association's purpose, appears to be the most productive source of social trust.
SOCIAL NETWORKS
Two indicators of social networks are used: first, the breadth of such networks and, second, the presence of neighbors, colleagues, or students from the respondents' current local community, workplace, or educational institution in the respondents' set of connections. In the following, the second type will be referred to by the slightly inaccurate term new friends for reasons of linguistic convenience.
Figure 2 shows that participation in associations is related to both the breadth of social networks and the presence of new friends. Even those who hold completely passive memberships have broader networks and are more often connected with new friends than those who are not affiliated at all.
Participation in Voluntary Associations
47
a) Breadth of social networks (average number of groups)
q.
CO O
.
E
o
-c
1,8 4
1,6
1,4
1,2
Relatively passive (N=358)
♦
-Netn-
member/ volunteer (N=384)
Completely passive (N=264)
Highly active (N=348)
E
cd
2 I
q.
=s O
1,8
1,6
1,4
1,2 4
4 er mere
(N=308)
(N=310)
(N=212)
J.
♦  Net a
1 affiliatien (N=479)
member/ velunteer
(N=384)
b) New friends important parts of social network (per cent)
'S
q.
75 4
I *
b o
=3 ä
E?
CD I
cd
a) cd
Highly active (N=348)
Relatively Relatively
^assive-acive—
Completely passive (N=264)
(N=358)
(N=339)
Not a member/ volunteer (N=384)
Cl
cc
I *
t o
= ä
E 1?
■§
cd
'I
cd o)
■2
cd
cd
TT
4 er mere (N=308)
3 affiliatiens (N=212)
(N=310)
(N=479)
♦  Net a
member/ volunteer (N=384)
Ö
i
cd
tz
2
2
CD O
CD O
o
cc
.
o
3 affiliations
2 affiliations
g
o
cd
n E □ tz
o
cd
85 -
CD O
o
75 -
2 affiliations
65 -
65
55
45
45
Figure 2.   Participation in Association and Social Networks
The relationship between networks and scope is more clear-cut than between networks and intensity. With regard to the latter, only the highly active (those who spend 3 hours per week or more on association activities) differ from those with less time-intensive involvements. The relationship between scope of involvement and social networks, on the other hand, is both stronger and more linear. For each new affiliation (except for the second and third), the social networks are extended, and the probability of the presence of new friends inthesetofconnections increases sharplyupuntilthethirdaffiliation.
Table 4 confirms that scope is more important than intensity in the creation of social networks. When number of affiliations is held constant, activity level
48
Wollebaek, Selle
Table 4.   Intensity/Scope-Index and Social Networks
Outsider (No Affiliations)
Passive (0-1 Hour/Week)
Active (> 1 Hour/Week)
Difference (Significance)
Breadth of social networks (average number of groups) 1.25
1 affiliation
2 or more affiliations
Difference
(Significance)
New friends that are
important parts of
social network (%) 56
1 affiliation
2 or more affiliations
Difference (Significance)
1.34 (n = 319)
1.65 (n = 303)
-.31 (.000)
63
(n = 319)
72
(n = 303) -9 (.011)
1.38 (n = 160)
1.60
(n = 527)
-.22
(.015)
63
(n = 160)
74
(n = 527)
-11 (.005)
-.04 (.635) .05 (.443)
0
(1.000) -2
(.215)
has absolutely no impact on either the breadth of social networks or the presence of new friends. On the other hand, the differences between members with multiple, as compared to singular, affiliations are evident among passive and active participants alike. The social networks of individuals active in only one association are not much broader than those of outsiders. Apparently, face-to-face contact in one association does not contribute much to the extension of networks.
With regard to type, another central tenet of Putnam's thesis is the role attributed to nonpolitical associations in the creation of horizontal networks. Although the possibility that the networks of participants in political associations are more vertical than horizontal in type cannot be entirely ruled out, Table 5 does not render much support to this supposition.
The members and volunteers who only take part in nonpolitical associations do not possess broader networks than the population at large, and new friends are not more frequently found among their social connections. The members only of religious associations distinguish themselves from participants in other associations by an even narrower social network than those not involved at all. At the other end of the scale, members of political associations and of different types of associations possess the broadest networks and the highest frequency of new friends in their set of connections.
The above analyses illustrate that although participation in itself broadens social networks, the time spent participating has virtually no bearing on the
Participation in Voluntary Associations
49
Table 5.   Average Number of Groups and "New Friends" That Are Important Part of Social Network (%), by Type of Organization
	Average Number		
Type of Organization	of Groups	% New Friends	n
Not affiliated	1.25	56	384
Nonpolitical, culture and recreation	1.40	65	352
Nonpolitical, religious	1.11	52	54
Semipolitical organizations	1.36	62	117
Political organizations (parties and			
unions)	1.59	73	138
Both nonpolitical and semipolitical	1.47	73	199
Both non- and semipolitical and			
political	1.71	75	441
Total members/volunteers	1.52	70	1,301
Total entire sample	1.46 F = 6.730	66 X2 = 22.098	1,693
Significance = .000    Significance = .001
Note: Only members and volunteers are included in significance tests.
breadth of social networks, and it is certainly subordinate to the impact of the number of affiliations. Furthermore, participants in nonpolitical associations have narrower networks than those involved in political activities. These results run to some extent counter to what might be expected from Putnam's theory. The findings therefore actualize the question of whether the relationships described above may be spurious and whether involvement in associations in fact is relatively unimportant to the development of networks.
Table 6 confirms that factors other than participation in associations explain more of the variation in breadth of social networks. Notably, young people, those who are better educated, and women have broader sets of connections than those who are older, have less education, or are male. Nevertheless, affiliation with associations does have an independent, positive effect. As indicated by the analyses above, there is no difference between the active and passive supporters with regard to social networks. When only those affiliated are included, the only participation variable with an impact on the breadth of social networks is number of affiliations. Neither activity level nor type of association influences social networks significantly.
The impact of participation on the presence of new friends was identical to that of breadth of social network (not shown). Among volunteers and members, the only participation variable with an impact is the number of affiliations. Level of activity and affiliation with nonpolitical associations are inconsequential for the formation of social networks.8
In summary, even when adjusted for the impact of other potential sources of social networks, the scope of participation has an effect on certain types of social networks, as well as the breadth of connections in general. Activity
50
Wollebaek, Selle
Table 6.   Regression Analysis of Breadth of Social Network (ordinary least squares)
Entire
	Sample		Members/Volunteers Only			
	Model 1	Model 2	Model 3	Model 4	Model 5	Model 6
Women (1)	11**	.13**	.12**	.12**	.12**	.11**
Age (years)	-.20**	-.22**	-.24**	-.22**	-.24**	-.25**
Education (1-3)	.18**	.16**	.14**	.16**	.14**	.14**
Population density (1-3)	.02	.03	.04	.03	.04	.04
Residential stability (1-5)	-.007	.04	.05	.04	.05	.05
Working full-time (1)	-.07*	-.05	-.06	-.05	-.06	-.06
Married/cohabitant (1)	.001	-.006	-.005	-.006	-.005	-.004
Affiliated with association(s) (1)	.08**					
Level of activity (1—4)		.02			-.03	-.02
Number of affiliations (1-4)			.12**		.13**	.14**
Nonpolitical association (1)				.006		-.04
R2 (adjusted)	.100	.086	.098	.085	.098	.098
Note: N = 1,470 for Model 1; n = 1,148 for Models 2 through 6, listwise deletion of cases. Dependent variable = number of groups that are important parts of social network, coded 0 to 5. *p < .05. **p < .01.
level, which is an approximate measure of the degree of face-to-face contact within associations, does not. This finding runs counter to the notion that it is primarily direct interaction between individuals in secondary associations that generates horizontal networks.
CIVIC ENGAGEMENT
The final component of social capital identified above, which provides the crucial link between social connectedness and participatory democracy, is civic engagement. In this context, it is interpreted as an individual resource that may be activated to cooperate with other people for civic or political ends. Civic engagement is measured using four indicators: political interest, daily news readership, regular voting at parliamentary elections, and a composite index constructed on basis of the three variables.
If active participation "inculcates skills of cooperation as well as a sense of shared responsibility for collective endeavors" (Putnam, 1993, p. 90), we would expect that members of voluntary associations are more civically engaged than nonmembers and that active members are even more engaged than passive supporters. Furthermore, if Putnam's (p. 90) assertion that "these effects ...donot require that the manifest purpose of the association be political" is correct, there should be no difference in civic engagement between participants in nonpolitical and political associations.
The impact of intensity and scope was similar for two out of three indicators in our civic engagement index (political interest and regular voting behavior), with a consistent positive correlation between scope of participation and level of civic engagement, and an absent correlation between engagement
Participation in Voluntary Associations
51
and intensity. For the third indicator, news material readership, neither scope nor intensity had any effect. On all three indicators, however, members differed from nonmembers. Thus, when all three measures are combined in Figure 3, a clear pattern emerges.
Civic engagement is moderately to strongly related to being a member of an association. However, there is nothing to suggest that active members are more engaged in civic matters than passive members are. Number of involvements, conversely, distinguishes clearly between the more and less civically engaged, among passive and active participants alike. These findings support the results presented by Almond and Verba (1963), which suggested that multiple affiliations have a cumulative effect on civic engagement, regardless of activity level.
Almond and Verba also asserted, as does Putnam, that this was true regardless of which type of association the member was affiliated with. Although the differences between various types of affiliations are significant for all indicators except news material readership, the results presented in Table 7 do render some support to this notion. Members of all types of associations, with the exception of those only affiliated with religious groups, display a somewhat higher level of civic engagement than those who are not affiliated at all. However, with the exception of voting behavior, members of political associations are, perhaps not surprisingly, more engaged than those affiliated with leisure associations.
Again, the respondents in the combinatory category, comprising those with membership in a non- or semipolitical association in addition to a political affiliation, score highest on the indicators. As was the case for trust and social networks, those affiliated with several associations with different purposes display the highest levels of social capital.
As the results in Table 8 show, however, association involvement is only one among several factors influencing civic engagement. Age has a remarkably strong effect when analyzing both the entire sample (Model 1) and only those affiliated with associations (Models 2 through 6). Urbanity and, not surprisingly, education also emerge as relatively strong predictors of civic engagement; married men and women are more engaged than singles; and men are slightly more engaged than women.
Nevertheless, the impact of affiliations with associations, both active and passive, is not cancelled out by these factors. Among those who are members or volunteers, only number of affiliations matters. Neither level of activity nor the dummy variable representing affiliation with nonpolitical associations influences the level of civic engagement when controlled for background variables. Although secondary to age, and possibly to education and urbanity, participation in several associations does appear to contribute to engagement in civic matters.
In summary, those who are affiliated with voluntary associations are more civically engaged than those who are not. Therefore, Putnam's view of associations as networks and generators of civic engagement is corroborated.
52
Wollebaek, Selle
cd o
cd
(J)
Highly active -(N-347)
Completely  Relatively Rela^vely active
paSSive pasSive (n=338) (N-263)      (N-356) (N-338)
Not a
member/ volunteer (N=372)
4 or more (N=307)
1 affiliation 2 affiliations (N=476) (N=309)
♦  Not a
member/
volunteer (N-372)
80 -
80 -
cd
o
cd
75
75
x -a
x
a
TD
70
70
E
E a)
65 -
65 -
60
60
cd
55
cd o) co ci
<
55
50
50
Intensity/scope-index and civic engagement index (average scores 0-100)
	"Outsider"	"Passive"	"Active"	Diff.
	(No affiliations)	(0-1 hr./week)	(more than 1 hr./week)	
	57.4			
1 affiliation		64.2	62.0	2.2
		(N=317)	(N=159)	(.298)
2 or more affiliations		67.5	68.5	1.0
		(N=302)	(N=526)	(.510)
	diff.	-3.3	-6.5	
		(.045)	(.001)	
Figure 3.   Participation in Associations and Civic Engagement Index (average score on scale of 0 to 100)
SUMMARY OF FINDINGS
Table 9 summarizes the findings from the analyses above with reference to the research questions specified at the outset of this article. With regard to the first question in Table 9, whether participation in associations contributes to social capital, confirmatory answers are given for all indicators. Furthermore, even though other factors, notably age and education, proved to be at least as important predictors of social capital as participation, the relationship is still present when adjusted for the potential effects of sociodemographic and contextual variables. Thus, the weight attached to the role of voluntary associations in the formation of social capital in Putnam's theory is corroborated by empirical results at the individual level.
Our expectations were also confirmed with regard to the third question: whether multiple affiliations have a cumulative effect on social capital. In fact, the scope of the involvement proved to be a more powerful predictor of trust, networks, and civic engagement than was intensity of involvement. When
Participation in Voluntary Associations
53
Table 7.   Civic Engagement Index, Political Interest, Daily News Material Readership, and Voting at Every Parliamentary Election, by Type of Association
Civic News Votes at Every
Engagement Political Material Parliament
Index Interest Readership Election
	Mean		Mean		% Who		% With	
	Score		Score		Reads		the Right	
Type of Organization	(0-100)	n	(0-100)	n	Daily	n	to Vote	n
Not affiliated	57.4	372	45.1	379	79	374	79	333
Nonpolitical, culture and								
recreation	62.5	351	51.2	349	86	346	83	320
Nonpolitical, religious	58.5	53	47.8	53	78	51	77	48
Semipolitical organizations	67.8	116	56.8	114	93	115	85	111
Political organizations								
(parties and unions)	68.0	138	57.9	137	91	136	82	134
Both nonpolitical and								
semipolitical	64.4	199	52.8	195	88	193	88	184
Both non- and semipolitical								
and political	70.9	439	60.6	438	90	433	91	431
Total members/volunteers	66.5	1,296	55.7	1,286	88	1,274	87	1,228
Total entire sample	64.5	1,676	53.3	1,665	86	1,648	85	1,561
	F = 8.878		F=8.939		x2 =	12.03	X2 = 17.30	
Significance		.000		.000		.034		.004
Note: Only members and volunteers are included in significance tests.
Table 8.   Regression Analysis of Civic Engagement Index (ordinary least squares)
Entire
	Sample		Members/Volunteers Only			
	Model 1	Model 2	Model 3	Model 4	Model 5	Model 6
Women (1)	-.06*	-.07*	-.07**	-.06*	-.07*	-.07**
Age (years)	.34**	.32**	.30**	.32**	.30**	.30**
Education (1-3)	.15**	.15**	.13**	.15**	.13**	.13**
Population density (1-3)	.11**	.11**	.12**	.10**	12**	.11**
Residential stability (1-5)	-.04	-.02	-.02	-.02	-.02	-.02
Working full-time (1)	.03	.02	.02	.02	.02	.02
Married/cohabitant (1)	.09**	.08**	.08**	.07*	.08**	.08**
Affiliated with association(s) (1)	.10**					
Level of activity (1-4)		.04			-.003	-.001
Number of affiliations (1-4)			.10**		.10**	.11**
Nonpolitical association (1)				.03		-.01
R2 (adjusted)	.177	.133	.142	.133	.141	.140
Note: N = 1,465 for Model 1; n = 1,147for Models 2 through 6, listwise deletion of cases. Dependent variable = civic engagement index (0-100). *p < .05. **p < .01.
activity level was held constant, multiple affiliations contributed more to forms of social capital than singular affiliations, whereas activity level failed to
54
Wollebaek, Seile
Table 9.   Participation in Associations and the Formation of Social Capital: Summary of Empirical Findings
Indicator
Civic Engagement
	Trust	Social Networks		News		
	Social		New	Material	Political	Regular
Research Question	Trust	Breadth	Friends	Readership	Interest	Voting
1. Does participation in						
associations contribute	Yes	Yes	Yes	Yes	Yes	Yes
to social capital?						
2a. Does active participation	Only	No	No	No	No	No
contribute more to the	among					
formation of social capital	young					
than passive memberships?	people					
2b. Is active participation	No	No	No	No	No	No
necessary for the formation						
of social capital?						
3. Do several affiliations	Yes	Yes	Yes	No	Yes	Yes
contribute more to social						
capital than one affiliation?						
4. Is level of activity more	No	No	No	No	No	No
important than number of						
affiliations?						
5. Do members of nonpolitical	No	No	No	No	No	No
associations display higher						
levels of social capital than						
members of political associations?						
distinguish between members with the same number of affiliations. Indeed, the consistent cumulative effect of multiple affiliations emerges as one of the principal findings of this study.
The assertion that participation in nonpolitical associations is more conducive to the formation of social capital than participation in associations with more expressed political purposes was not strengthened by the results. On the contrary, the analyses showed that participation in leisure or cultural associations needed to be accompanied by an affiliation in semipolitical or political organizations to have an impact. The members of only leisure-oriented associations failed to distinguish themselves from the population at large in terms of trust, networks, and civic engagement. This serves as a specification of the argument above: The most productive form of participation with regard to the formation of social capital seems to be not only participation in several associations but multiple affiliations in associations with different purposes.
Finally, and most important, the question of whether time-intensive, active participation is more conducive to social capital than passive support is, contrary to expectations, given a negative answer. There is nothing in the data to
Participation in Voluntary Associations
55
suggest that active participation, compared to passive, broadens social networks or strengthens civic engagement, and the relationship between trust and intensity of involvement is tenuous at best. Furthermore, even passive memberships had a positive influence on all of the indicators presented above. Thus, at least based on the present data, a preliminary negative answer must be given to the question of whether active participation, that is, face-to-face interaction, is necessary for the formation of social capital.
IMPLICATIONS FOR SOCIAL CAPITAL RESEARCH
In sum, our findings reveal a pattern at odds with Putnam's understanding of how social capital is developed. Naturally, using population surveys in broad investigations such as this one is not problem free. However, the results do not appear to be entirely anomalous. A study built on a survey of Norwegian environmentalism points in the same direction (Selle & Stromsnes, 1998; Stromsnes, 2001). One of the few other studies that have investigated the relationship between passive memberships and social capital, by Dekker and van den Broek (1998, pp. 33-35), also indicates that some of the findings here may be valid in other contexts than the Norwegian.
What are the theoretical implications of the findings? Three possible interpretations of the results presented above will be discussed. First, that the role of voluntary associations in the formation of social capital is overstated in the work of Putnam and other social capital theorists; second, that passive affiliations may have internal effects on those loosely connected to associations; and third, that the main contribution of associations to social capital may be more accurately located if a more institutionalist perspective is employed.
A first interpretation of the absent effect of face-to-face contact is that Putnam exaggerates the role of voluntary associations in the formation of social capital. If active participation does not have an impact on trust, networks, and civic engagement over passive support, this sufficiently proves that Putnam is mistaken in placing voluntary associations at the center of his social capital thesis. In this interpretation, the observed relationship between passive affiliations and social capital is seen as spurious. The differences between joiners and nonjoiners stem from members having resources of some kind that outsiders do not possess.
There is some empirical support for this claim. The impact of participation was rather modest in statistical terms. Furthermore, the indicators used to measure other sources of social capital, such as education and employment, may be too crude to claim that the impact of status or integration in the workforce has been controlled for. Regarding the relationship between participation and social capital as spurious would not be contrary to common sense. After all, association involvement is for most people, compared to other activities we take part in on a more frequent basis, of rather low intensity measured both in time and emotional commitment. At the very least, the relationship
56
Wollebaek, Selle
needs to be qualified. Specifically, Putnam's view of participation in birdwatchers' associations or bowling leagues as the hallmark of civic engagement and a vibrant democracy is not given much support by the data.
However, considering the numerous studies that, like the present one, have found participation in associations as a whole to be related to social capital and civic engagement—even when controlling for a wide range of factors—it does not seem plausible to dismiss the entire relationship as spurious.9
Furthermore, such a dismissal would rest on the a priori assumption that being passively connected to associations cannot affect the supporter. This is not an uncontestable truth but an empirical question. Elsewhere, we have shown that passive members are connected to and care about their organizations to a greater extent than what is generally assumed (Wollebaek, 2000). Therefore, rather than rejecting the relationship between participation and social capital, a closer examination of whether the nature of the relationship is correctly spelled out by Putnam is in order. In particular, the presumption that social capital can only come from face-to-face interaction should be subjected to critical examination.
A second interpretation, therefore, is that the results show that voluntary associations contribute to social capital but not necessarily by means of direct interaction between members. This implies that even passive affiliations may have internal effects on those participating.
How might this occur? It is possible to distinguish between four understandings of the relationship between the passive supporter and his or her association: the associations as social systems, imagined communities, information systems, and networks of political influence.
When regarding the passive member as part of a social system, it is implied that socialization may take place even if the passive member does not interact with other members within the context of the association. Many passive members are likely to socialize with activists but in other social settings than the organization. Although most members in tertiary associations, at least in the United States, are recruited through direct marketing techniques (Maloney, 1999), many, especially in associations combining active and passive membership, are recruited through already existing social networks (Stromsnes, 2001). The passive members keep in touch with the association by way of their networks of contacts with activists. Although this will certainly not always be the case, and the socialization is of a less intensive character than is the case for active participants, it is a possibility that should not be ruled out a priori. The finding that the number of affiliations completely cancels out the effect of activity level on breadth of social networks and the presence of friends in the current situation suggests that this interpretation may not be entirely off the mark.
Second, the affiliation may contribute to a sense of identification with and commitment to a cause. As discussed above, associations relying on passive support resemble Anderson's (1991) idea of an imagined community, a concept referring to all social systems too large to allow face-to-face contact,
Participation in Voluntary Associations
57
wherein members nonetheless share emotional ties to a community. Passive affiliations may foster a sense of affinity to a cause that the individual knows is not only important to himself or herself but also to others. If the association is successful, the membership, regardless of activity level, conveys a sense of the value of cooperation for common purposes, of political efficacy, and of a shared belonging to something important. Clearly, these virtues are all conducive to forming social capital.
Third, associations may function as information systems. Norwegian nationwide voluntary associations publish more than five journals and newsletters per person in the adult population annually (Hallenstvedt & Trollvik, 1993). Furthermore, the information networks in the new tertiary associations founded in the past couple of decades, which rely almost entirely on passive support, are at least as comprehensive as in traditional voluntary associations. It is not unlikely that this function will become increasingly important with recent technological developments, such as the Internet. Norwegian tertiary associations are at the forefront in making use of these new opportunities of information dissemination. This implies that the passive member is not necessarily out of touch with the goings-on of the association; perhaps he or she will be even less so in the future. The extensive networks of information disseminate knowledge about current issues and how the association relates to them. As such, these networks may serve as "schools in democracy" and promoters of civic engagement—even though the members do not interact personally.
Finally, associations might serve as networks of political influence, even for those not actively involved. As demonstrated above, passive members are not entirely marginalized with regard to internal decisions in the associations. They have the power to withdraw their support—to exit—which is clearly not inconsequential to the associations. If passive members hold multiple affiliations, as many do, they have the opportunity to exert influence on several arenas simultaneously. This participation by proxy exercised by many passive supporters may be of no less significance for democracy than active participation.
The notion that the passive members' affiliations with extensive information networks, imagined communities, and networks of political influence leave them unaffected in terms of trust, social networks, and civic engagement is an assumption, which so far has not been supported by empirical results. Pending evidence to the contrary, the postulation that only face-to-face contact within the realms of voluntary associations has internal effects on those affiliated should not be too readily accepted.
The third interpretation of the absent effect of face-to-face contact is that associations contribute to social capital but primarily as institutions in which norms and resources are embedded. The logic of this argument may be clarified by means of a contrafactual thought experiment. What would the level of social capital be like if associations were absent? Regardless of whether those joining associations possess more social capital than nonjoiners to begin with, associations contribute to the sustenance and transformation of values and norms because they are an important part of the political, social, and cultural
58
Wollebaek, Selle
infrastructure of a society. The existence of a multitude of visible voluntary associations is in itself evidence of the value and rationality of collaborative efforts, even for individuals who do not actively take part themselves. Thus, in this perspective, the role of associations as generators or catalysts of trust, networks, or civic engagement is subordinated to their role as institutions expressing and sustaining the same values and resources in the society at large.
This implies a move from the internal effects of the participation to the external effects of associational life on society at large. These external effects can be divided into two broad categories: integration and democracy. Their contribution to integration lies in the ability to create multiple and overlapping networks that reduce the conflict potential in society. This occurs because the loosely knit networks may span existing cleavages and patterns of loyalty. In this perspective, the "broad" voluntary sector, exemplified by the Scandinavian countries, may be particularly well suited to institutionalizing social capital, because it creates weaker ties to multiple institutions instead of or in addition to strong ties to few. As Dekker, Koopmans, and van den Broek (1997) showed, levels of social trust are much higher in the Scandinavian countries and the Netherlands, which are used to exemplifying the broad model of a voluntary sector, than in the rest of Europe and North America, where the "parochial" or "active" models are found.
If associations contribute to the sustenance of values of moderation and generalized trust among citizens, they also contribute to a stable democracy. Their democratic relevance, however, is also of a more direct character. A diverse range of associations is necessary for democratic pluralism, which can be seen as a value in itself. Associations' involvement in the public discourse can convey a sense of political efficacy among members and nonmembers alike—a feeling that participation is not futile. The role of the passive supporter is not unimportant in this respect. A broad membership base, more or less regardless of activity level, is an essential source of legitimacy for many associations with an outward-oriented, political purpose.
Regarding associations as institutionalizations of trust, networks, and civic engagement, rather than generators, catalysts, or vehicles, entails a different perspective from that of Putnam. Rather than focusing on internal effects of the participation on the active member (i.e., socialization), this other perspective draws attention to associations' external effects on the wider polity or society (i.e., pluralism and cross-cutting cleavages). Less significance is attributed to the intensity of the involvement (i.e., degree of face-to-face contact) as the main indicator of vibrancy of associational life than to the scope of participation (i.e., multiple, overlapping memberships).
Finally, this new perspective does not discard associations with a political purpose simply because they tend to be more vertical in structure than leisure-oriented associations. On the contrary, purely horizontal networks (if such power-neutral networks exist at all) (Mouzelis, 1995) are insufficient if democratic pluralism as well as political and social cohesion is to be ensured
Participation in Voluntary Associations
59
(Berman, 1997; Rueschemeyer, 1998). To contribute to these ends, associations need to provide institutional links reaching beyond local communities and into the political system. Thus, by definition, the networks need to be vertical in one meaning of the word. Purely local voluntary groups, the ideal in a bottom-up neo-Tocquevillian model, do not suffice if associations are valued for more than by-products of membership activity, that is, as important social, cultural, and democratic institutions in their own right (Skocpol, 1999b).
This does not exclude or undervalue horizontal networks at the local level, which are still found in most outward-oriented and politically relevant secondary associations in Norway. Rather, it stresses the fact that the desire for social connectedness through active participation in local, nonpolitical asso-ciational life should not lead us to neglect the role of associations as intermediary institutions—between the citizen and individual on one hand and the political system and society at large on the other. This function may indeed be of more consequence to democracy than bowling in any number of organized leagues.
Notes
1. See Wollebaek (2000) for a more detailed theoretical and conceptual discussion.
2. In the questionnaire, respondents were asked to check a box for each category of associations they belonged to and, thereafter, were asked to provide the names of these organizations. This accommodates an important methodical objection to Putnam's empirical analyses raised by Skocpol and Fiorina (1999, p. 8), namely, that multiple within-category memberships are only counted once in the General Social Survey data Putnam uses.
3. If a person is affiliated with both a religious and another nonpolitical association, he or she is classified in the first combinatory category.
4. When used in regression analyses, the responses are recoded into a 3-point scale, where those who say that "most people can be trusted" are given the highest value, the "don't knows" are given a middle value, and those responding that "you can't be careful enough" are given the lowest value.
5. The alternatives included "neighbors and local community where you live now," "current colleagues or fellow students," "former colleagues or fellow students," "friends from where you grew up," and "others."
6. At the lowest level we find those who usually do not vote at elections and who read news material less frequently than once a week. The second lowest value is given to those who either vote sometimes or read newspapers at least once a week. Those who do both are given the third lowest value. The fourth value is given to those who either vote at every election or read news material daily. The fifth value is given to those who do both. The sixth value is given to those who, in addition to reading and voting, are somewhat interested in politics, whereas the seventh and highest value is given to the small minority who say they are very interested in politics regardless of their voting behavior or newspaper readership. This allows for the possibility that abstention from voting may be an act of protest and, hence, a profound act of civic engagement. Those who are too young to vote are given values on the index corresponding with how often they read news material in papers and how interested they are in politics.
7. Weighting the results reduced overall levels of social capital very slightly on most indicators. For example, the percentage of respondents agreeing that "most people can be trusted"
60
Wollebaek, Selle
decreased from 71% to 69% due to the overrepresentation of higher educated respondents in the unweighted sample. There were no substantial changes in any of the relationships discussed in the article that would affect our interpretations. No relationships changed direction or level of significance (above or below 95%).
8. In this analysis, gender and education were rendered insignificant. Age was negatively related to the presence of new friends among members, or volunteers, and nonmembers alike, whereas residential stability and being married had minor but significant positive effects.
9. See Dekker and van den Broek (1998), Whiteley (1999), Torcal and Montero (1999), and Stolle and Rochon (1999) for studies corroborating the positive relationship between association membership and social capital. Van Deth (1997) provides an overview of studies of the relationship between participation in associations and political involvement, among which the majority report positive correlations.
References
Almond, G., & Verba, S. (1963). The civic culture: Political attitudes and democracy in five nations.
Princeton, NJ: Princeton University Press. Anderson, B. (1991). Imagined communities: Reflections on the origins and spread of nationalism (Rev.
ed.). London: Verso.
Berman, S. (1997). Civil society and the collapse of the Weimar Republic. World Politics, 49,
401-429.
Bourdieu, P. (1985). The forms of capital. In J. G. Richardson (Ed.), Handbook of theory and research
for the sociology of education (pp. 241-258). New York: Greenwood. Coleman, J. S. (1988). Social capital in the creation of human capital. American Journal of Sociology,
94, S95-S121.
Dekker, P., Koopmans R., & van den Broek, A. (1997). Voluntary associations, social movements
and individual political behavior in Western Europe. In J. W. van Deth (Ed.), Private groups and
public life (pp. 220-241). London: Routledge. Dekker, P., & van den Broek, A. (1998). Civil society in comparative perspective: Involvement in
voluntary associations in North America and Western Europe. Voluntas, 9, 11-38. Foley, M. W., & Edwards, B. (1996). The paradox of civil society. Journal of Democracy, 7, 38-52. Granovetter, M. (1973). The strength of weak ties. American Journal of Sociology, 78, 1360-1380. Hallenstvedt, A., & Trollvik, J. (1993). Norske organisasjoner. Oslo: Fabritius Forlag. Maloney, W. A. (1999). Contracting out the participation function. Social capital and cheque-book
participation. In J. W. van Deth, M. Maraffi, K. Newton, & P. F. Whiteley (Eds.), Social capital and
European democracy (pp. 3-24). London: Routledge. Mouzelis, N. P. (1995). Sociological theory: What went wrong. Diagnosis and remedies. London:
Routledge.
Newton, K. (1997). Social capital and democracy. American Behavioral Scientist, 400, 575-586. Newton, K. (1999). Social capital and democracy in modern Europe. In J. W. van Deth, M. Maraffi,
K. Newton, & P. F. Whiteley (Eds.), Social capital and European democracy (pp. 3-24). London:
Routledge.
Portes, A. (1998). Social capital: Its origins and applications in modern sociology. Annual Review of Sociology, 24, 1-24.
Putnam, R. D. (1993). Making democracy work: Civic traditions in modern Italy. Princeton, NJ: Princeton University Press.
Putnam, R. D. (1995a, January). Bowling alone: America's declining social capital. Journal of Democracy, 1, 65-78.
Putnam, R. D. (1995b). Tuning in, tuning out: The strange disappearance of social capital in America. Political Science and Politics, 28, 664-683.
Putnam, R. D. (2000). Bowling alone: The collapse and revival ofAmerican community. New York: Simon & Schuster.
Participation in Voluntary Associations
61
Quigley, K.F.F. (1996). Human bonds and social capital. Orbis, 40, 333-343.
Rueschemeyer, D. (1998). The self-organization of society and democratic rule: Specifying the relationship. In D. Rueschemeyer, M. Rueschemeyer, & B. Wittrock (Eds.), Participation and democracy: East and west. Comparisons and interpretations. Armonk, NY: M. E. Sharpe.
Scott, J. (1991). Social network analysis. Newbury Park, CA: Sage.
Selle, P., & Stromsnes, K. (1998). Organized environmentalists: Democracy as a key value? Voluntas, 9, 319-343.
Selle, P., & Stromsnes, K. (2001). Membership and democracy: Should we take passive support seriously. In P. Dekker & E. M. Uslaner (Eds.), Social capital and politics in everyday life. London: Routledge.
Skocpol, T. (1999a). Advocates without members: The recent transformation of American civic life. In T. Skocpol & M. P. Fiorina (Eds.), Civic engagement in American democracy (pp. 461-509). Washington, DC: Brookings/Russell Sage Foundation.
Skocpol, T. (1999b, March-April). Unraveling from above. The American Prospect, 25.
Skocpol, T., & Fiorina, M. P. (1999). Making sense of the civic engagement debate. In T. Skocpol & M. P. Fiorina (Eds.), Civic engagement in American democracy (pp. 1-23). Washington, DC: Brookings/Russell Sage Foundation.
Stolle, D., & Rochon, T. R. (1999). The myth of American exceptionalism: A three-nation comparison of associational membership and social capital. In J. W. van Deth, M. Maraffi, K. Newton, & P. F. Whiteley (Eds.), Social capital and European democracy (pp. 3-24). London: Routledge.
Stromsnes, K. (2001). Demokrati i bevegelse. Unpublished Ph.D. thesis, University of Bergen, Norway, Department of Comparative Politics.
Tocqueville, A. (1954). Democracy in America. New York: Random House. (Original work published 1835)
Torcal, M., & Montero, J. R. (1999). Facets of social capital in new democracies: The formation and consequences of social capital in Spain. In J. W. van Deth, M. Maraffi, K. Newton, & P. F. Whiteley (Eds.), Social capital and European democracy (pp. 167-191). London: Routledge.
Torpe, L. (1998). Lokal organisasjonsdeltakelseog demokrati. Mellem individ ogfaellesskab samtborgerog styre. Bergen, Norway: LOS-senteret. 9816.
van Deth, J. W. (1997). Introduction: social involvement and democratic politics. In J. W. van Deth (Ed.), Private groups and public life. London: Routledge.
Whiteley, P. F. (1999). The origins of social capital. In J. W. van Deth, M. Maraffi, K. Newton, & P. F. Whiteley (Eds.), Social capital and European democracy (pp. 3-24). London: Routledge.
Williams, B. (1988). Formal structures and social reality. In D. Gambetta (Ed.), Trust: Making and breaking cooperative relations. Oxford, UK: Basil Blackwell.
Wollebaek, D. (2000). Participation in voluntary associations and the formation of social capital. Unpublished doctoral dissertation, University of Bergen, Norway, Department of Comparative Politics.
Wollebaek, D., Selle, P., & Lorentzen, H. (1998). Undersokelse om frivillig innsats.
Dokumentasjonsrapport. Bergen, LOS-senteret. 9834. Wollebaek, D., Selle, P., & Lorentzen, H. (2000a). Frivillig innsats. Oslo: Fagbokforlaget. Wollebaek, D., Selle, P., & Lorentzen, H. (2000b). Giving and volunteering in Norway. A brief summary
of results from a national survey. Bergen, Norway: LOS-Centre/Institute for Social Research. Woolcock, M. (1998). Social capital and economic development: Toward a theoretical synthesis
and policy framework. Theory and Society, 22, 151-208.
Dag Wollebaek is a graduate of the University of Bergen with a degree in comparative politics. He has been doing research on the nonprofit sector and voluntary associations since 1997 as part of the Johns Hopkins Comparative Nonprofit Sector Project.
Per Selle is professor of comparative politics at the University of Bergen and senior researcher at the Norwegian Centre for Research in Organization and Management. His research interests include voluntary organizations, political parties, political culture, and environmental policy.