Social Capital and Democracy: An Interdependent Relationship

by Pamela Paxton
Social Capital and Democracy: An Interdependent Relationship
Pamela Paxton
American Sociological Review
Start Page: 
End Page: 
Select license: 
Select License




The Ohio State University

Current democratic theory and recent international policy initiatives reveal an in- tense interest in the relationship between social capital and democracy. This interest is the most recent variant of a long theoretical tradition positing that a vigorous associational life is beneficial for the creation and maintenance of democracy. De- spite the popularity of this view, little quantitative empirical evidence exists to sup- port the relationship. Here, the relationship between social capital and democracy is tested using data from a large, quantitative, cross-national study. Two additional tests are introduced. First, the plausible reciprocal effect-from democracy to social capital-is included in models. Second, the potentially negative impact of some associations on democracy is considered. Using data from the World Values Survey and the Union of International Associations in a cross-lagged panel design, results show that social capital affects democracy and that democracy affects social capital. Additional tests demonstrate that associations that are connected to the larger com- munity have a positive effect on democracy, while isolated associations have a nega-

tive effect. Theory relating social capital to democracy is drawn from the literature on civil society, political culture, and social movements.

THEORISTS HAVE LONG argued that when a country has a vigorous associa- tional life it is better able to create and main- tain a democracy. When citizens interact of- ten, join groups, and trust each other, their relationships aid democratization by crystal- lizing and organizing opposition to a non- democratic regime. Once a democracy is es- tablished, these relationships expand citizen access to information and political ideas, which increases governmental accountabil- ity. Furthermore, voluntary associations pro- vide a training ground for new political lead- ers, help members practice compromise and

Direct all correspondence to Pamela Paxton, Department of Sociology, 300 Bricker Hall, 190 North Oval Mall, The Ohio State University, Co- lumbus, OH 43210-1353 ( I thank Ken Bollen, Ed Crenshaw, Jennifer Glanville, David Jacobs, Vinnie Roscigno, Nella van Dyke, and the ASR reviewers for helpful comments on an early draft of this paper. I gratefully acknowledge the support of the National Science Foundation (GER-9554569).

learn tolerance, and stimulate individual par- ticipation in politics.

Theories concerning democracy's depen- dence on associations are long-standing. Tocqueville ([1835, 18401 1990) is often credited for first noting the relationship in the United States. The relationship is linked to a rich historical tradition on civil society, how- ever, through such thinkers as Ferguson ([I7671 1995) and Montesquieu ([I7481 1989). Although the definition of what con- stitutes "vibrant" or "associational" varies, the basic theory has appeared under diverse names, including "civil society" (Habermas 1989; Calhoun 1993), "social capital" (Putnam 1993), "pluralism" (Lipset, Trow, and Coleman 1956; Truman 195 I), "mass so- ciety" (Arendt 1948; Horkheimer 1947), and "civic culture" (Almond and Verba 1963).

The hypothesized relationship between as- sociations and democracy has influenced public policy in Eastern Europe and the de- veloping world. Current development efforts concentrate on the formation of nongovern- mental organizations (NGOs) in the belief


that, once created, such institutions will help foster and maintain stable democracies. Based on these assumptions, funding to NGOs by governments, foundations, and in- ternational agencies has increased dramati- cally over the past decades (Nelson 1995). The proportion of total aid from members of OECD countries that is funneled through NGOs rose from .7 percent in 1975 to 3.6 percent in 1985 to 5.0 percent in 1994 (Edwards and Hulme 1996).

Despite the longtime theoretical interest in the relationship between social capital and democracy, and its obvious policy implica- tions, little quantitative empirical evidence exists to support the idea that associations affect democracy. Qualitative case studies of civil society provide little concrete evidence for the relationship, as they remain mainly at the theoretical or descriptive level (Kubik 1998: 132). Some quantitative evidence ex- ists, but it focuses largely on concepts like civic culture, which are based on aggregate attitudes rather than on direct measurement of associations (Almond and Verba 1963; Muller and Seligson 1994). The only study to measure associations is Putnam's (1993) research on social capital in Italy. But Putnam considers governmental performance rather than democracy, and his study is re- stricted to Italy. Other quantitative research related to the question also tends to be geo- graphically limited (Inglehart 1990, 1997).'

Two additional questions have not been addressed in previous research. First, most empirical work does not incorporate the likely reciprocal effect of democracy on as- sociations. A reciprocal effect is quite plau- sible because democratic institutions permit the formation of voluntary associations to a greater extent than do nondemocratic insti-

Although there has been no direct cross-na- tional test of the civil society or social capital hypotheses, there are a few related studies. Fol- lowing in the tradition of Almond and Verba (1963), three studies-Inglehart (1990, 1997) and Muller and Seligson (1994)-use "civic cul- ture," or the attitudes of a population, to explain democracy. These studies do not consider asso- ciations among individuals directly, however. Two studies consider the effects of trust or asso- ciation memberships, but on governmental per- formance rather than democracy (Knack and Keefer 1997; Putnam 1993).

tutions. While most theories stress the asso- ciations-to-democracy causal path, various authors have noted the potential for a recip- rocal relationship (Fox 1996; Levi 1996:49- 51; Offe 1999:73-74; Rahn, Brehm, and Carlson 1999; Warren 1999).

Second, different associations need not have equivalent effects on democracy (M. Olson 1982; Putnam 2000). Previous theory and empirical research tends to treat all as- sociations the same. That is, a bird-watching group is as likely to promote a healthy, ef- fective democracy as the AARP. It is highly likely, however, that certain types of asso- ciations will do better in promoting democ- racy. And some types of associations may actually be detrimental to democracy. For example, nationalist groups are likely to ex- acerbate societal cleavages and interfere with democratic consolidation. Such asso- ciations could also reduce levels of toler- ance, thereby undermining the overall demo- cratic political culture. Thus, a significant question about the relationship between de- mocracy and associations remains unanswered: Do different types of associations differ in their effects on democracy?

I test the relationship between democracy and associations (operationalized as social capital) using a large, quantitative, cross-na- tional, panel study. To my knowledge, this is the first cross-national empirical test of this long-standing thesis. For my measure of so- cial capital, I draw on two very different sources of data: the World Values Survey and the Union of International Associations. With sample sizes of 48 and more than 100, respectively, these sources include much greater developmental and regional variation than was available in past research. My measure of democracy, taken from Bollen (1998), is also superior to other measures on a number of criteria (see Bollen and Paxton 2000). Because my two data sources contain measures at multiple points in time, I use a panel design to assess a reciprocal causal re- lationship. I also distinguish associations that are connected to other associations from those that are isolated in order to test for dif- ferences in their impact on democracy.

To operationalize a "vibrant associational life," I make use of the newest theoretical concept to emerge from the participatory democratic literature-social capital. I de- fine social capital with two dimensions (Paxton 1999): First, social capital requires an objective network of ties among individu- als. Second, it requires that the ties among individuals be trusting, reciprocal, and emo- tionally positive. Like civil society, social capital represents voluntary association memberships, but it captures other important features of association, such as subjective trust.


The argument linking a vibrant associational life to democracy is found in a variety of lit- eratures under the rubric of pluralism, civil society, and civic culture. Unfortunately, the link is often presented incompletely, in part because the operationalization of a vibrant associational life has not been clearly or completely specified. What is "vibrant"? What is "associational"? The first step in theoretically linking a vibrant associational life to democracy, therefore, is to define terms.

For my definition, I use the relatively new concept of social capital, which was popu- larized by Bourdieu (1983) and Coleman (1988, 1990). In brief, social capital is the notion that social relations can facilitate the production of economic or noneconomic goods. Social capital is explicitly social; ac- cording to Coleman (1988:S98), social capi- tal resides not in individuals but in the rela- tions between individuals. For Coleman, so- cial capital takes many forms, including ob- ligations within a group, trust, intergen- erational closure, norms, and sanctions. Bourdieu (1983:249) stressed that the rela- tionships between individuals must be du- rable and subjectively felt.

Following my earlier work (Paxton 1999), which builds on these two classic defini- tions, I define social capital with two dimen- sions. Social capital requires (1) objective associations among individuals, and (2) as- sociations of a particular type-reciprocal, trusting, and involving positive emotion. When present, social capital can facilitate the production of individual or collective goods. For example, social capital in the form of network ties and trust among neigh- bors can be seen either as a benefit for indi- viduals, who can freely walk the streets, or as a benefit to the community as a whole in the form of reduced crime rates. With these two dimensions, this definition of social capital captures both relations between indi- viduals (Coleman 1988) and the fact that these relationships must be subjective, trust- ing, and positive (Bourdieu 1983).

Although social capital is only one of many possible operationalizations of a vi- brant associational life, it is a useful operationalization for this analysis. With two distinct dimensions, social capital can be more precisely related to democracy. Be- cause social capital is not constrained to at- titudes (like civic culture is), it captures re- lationships among citizens. Yet, emphasizing associations alone is also limited in that there is no way to distinguish economic ties from true community. To distinguish vibrant associations, we should therefore think about the content of associations, and at a minimum this would involve assessing trust.

Although social capital was theorized by Coleman (1988) and Bourdieu (1983) as a feature of groups, Putnam (1993) brought the concept into macrosociological theory by claiming that social capital could be aggre- gated and influence effective government.

1 My definition is general: Social capital can

I be measured at multiple levels and produce goods at multiple levels (i.e., at the indi- vidual, group, and community levels). My

1 definition therefore divorces social capital from its potential consequences, which, ac- cording to Portes (1998), is a problem with other definitions of social capital. I focus on social capital as an aggregate feature of na- tions and consider its effect on the creation and maintenance of healthy democratic in- stitution~.~

This aggregate formulation has been cri- tiqued by Portes (1998, 2000), who argues that social capital should operate only at the indi- vidual level of analysis. However, I see social capital as a single concept-trusting ties among individuals-that can vary in its outcomes. The strategy of defining the term apart from the level of analysis is also found in human capital re- search (e.g., Becker 1964:23-24; Sen 1999:292- 97). Portes (1998) gives four criteria for analy- ses of social capital, of which the first three are relevant to my analysis. An analyst must proceed by, "first, separating the definition of the con-


Social capital can affect democracy in two ways. First, social capital can help to create democracy in a country that is not demo- cratic. Alternatively, it can help to maintain or improve an already existing democracy. In the first case, strong associations reduce the ability of the state to directly oppress citi- zens and provide a space for growth in orga- nized opposition to a nondemocratic regime. In the second case, these associations teach tolerance, promote compromise, stimulate political participation, and train leaders-all of which contribute to a healthy democracy.

There are two main ways that social capital can aid in a transition to democracy: (1) it provides a space for the creation and dis- semination of discourse critical of the present government, and (2) it provides a way for active opposition to the regime to grow. Initially, non state-sponsored relation- ships between individuals spark conversa- tions. Such conversations are a potential source of opinions that may differ from pre- vailing state ideology. Once these opinions have been formed, rather than remaining iso- lated, individuals can use their associations and trust to share ideas and speak about their dismay with the current regime. Ultimately, associations help disseminate information about protest activity and aid in the growth of opposition social movements.

SOCIALCAPITAL AS A SPACE FOR DIS- COURSE. High levels of social capital can help create and disseminate antigovernment discourse. This antigovernment discourse can be intellectual or popular, open or clan- destine. Artists, novelists, and playwrights can ridicule the regime and keep alternative

cept, theoretically and empirically, from its al- leged effects; second, establishing some controls for directionality so that the presence of social capital is demonstrably prior to the outcomes that it is expected to produce; third, controlling for the presence of other factors that can account for both social capital and its alleged effects" (p. 20). My analyses meet these criteria.

values alive (O'Donnell and Schmitter 1986:49). Or the intelligentsia may find ways to subvert regime policies (Goldfarb 1978; Michta 1997). At the same time, popu- lar protest (e.g., stories, songs, complaints) can continue under even the worst oppres- sion (Fatton 1995; Scott 1985, 1990).

Research on the creation and maintenance of dissenting opinions supports the role of trusting relationships in the origin of antiregime discourse. Group ties shield indi- viduals from outside sanctions and allow un- orthodox or dissenting views. For example, Finifter (1974) finds that friendship groups provide a protective space in workplaces for dissident political opinions. Milgram's (1965) classic research shows that individu- als are more likely to defy established au- thority in the presence of confederates.

Informal social ties then become impor- tant in spreading information about griev- ances (Oberschall 1993:24; Useem 1980). For example, opposition pamphlets may be distributed through trusted informal net- works. Opp and Gem (1993:662) find that having friends critical of the East German regime was an important determinant of an individual's tendency to protest during the revolution of 1989. Formal, organized asso- ciations play a role as well. In Eastern Eu- rope (Michta 1997) and in Latin America (Smith 1991), the church both created an ideological alternative to the regime and helped to broadcast its alternative view.

SOCIALCAPITAL AND COLLECTIVE AC- TION. In the last stage of resistance to an oppressive regime, high levels of social capital provide resources for the organiza- tion of opposition movements and large- scale collective action. To begin, trusting associations can both form the site of early opposition and support a nascent movement (Morris 1981). Meetings can be planned, events can be organized, and opposition leaders can be trained when individuals meet in associations. In Brazil, Sandoval (1998) explains that in the early years of the democratization movement (1977 to 1980), organized opposition "first appeared in the relatively guarded environment of meetings of professional associations or . . . the church" (p. 176). For example, "the Brazil- ian Society for the Progress of Science, the country's oldest scientific association, faced government obstruction of its thirty-seventh annual meeting as the government sought to avoid anti-government criticism . . . " (p. 176). In a very different time period, Markoff (1996a) explains how Sunday Mass brought the rural community together during the decades preceding the French Revolution. "After Mass, communal issues could be discussed, grievances aired, anger focused, plans hatched, and actions taken"

(p. 308).

Social ties then aid in the growth of more large-scale, sustained opposition movements. Trusting associations provide a place for individuals to indicate their intent to counteract the state, thereby activating the critical mass necessary for larger-scale col- lective action (Marwell and Oliver 1993). Social movements actively opposing a re- gime can draw on preexisting association ties to mobilize for collective action (Mor- ris 1984; Snow, Zurche, and Ekland-Olson 1980; Curtis and Zurcher 1973; Useem 1980). Oberschall (1973) explains how es- tablished networks act as sites for "bloc re- cruitment" to social movement^.^

There is evidence for the importance of social capital in the creation of democracy in Eastern Europe. Autonomous organiza- tions persisted there despite years of totali- tarian rule (Pacek and Kanet 1991:195; Tong 1994), quietly resisting (Hankiss 1989; Ekiert 1997), and helping to organize antistate activity and promote democratiza- tion (Bernhard 1993; Opp and Gern 1993).

Social movement theory and civil societylso- cia1 capital theory complement each other in the explanation of democratization. Social move- ments are traditionally defined by explicit, in- tended goals, like democratization, while social capital theory stresses the possibility of unin- tended consequences toward democratization through association. Therefore, as an unintended consequence, social capital may provide the "cooptable associational landscape" (McCarthy 1997:250) that is a resource for opposition groups in creating and organizing sustained col- lective action (Morris 1984; Coleman 1988: S108). Vibrant associations can also work along- side established social movements in the process of democratization. For example, case studies of the 1989 mass protests in Eastern Europe indi- cate that the protests far outstripped the estab- lished opposition movement's organizing capa- bility. Karklins and Peterson (1993:600) explain

Once a democracy is established, it must be consolidated so that democracy becomes the "only game in town" (Linz and Stepan 1996). And even consolidated democracies need to remain healthy. The social ties and trust of social capital help maintain democracy by affecting both the quantity and quality of political participation by citizens.

First, formal memberships in voluntary as- sociations help increase the amount of po- litical participation-its quantity. Tocqueville ([I8401 1990) held that an individual's participation in "civil" associations would create feelings of duty, increase a sense of interdependence with others, and produce a habit of participation. These theoretical as- sertions are upheld by empirical research in a variety of countries linking membership in voluntary associations to increased political participation (Barnes and Kaase 1979; Verba, Nie, and Kim 1978; Wolfinger and Rosenstone 1980; for a more comprehensive list see Leighley 1995: 183). A related line of research argues that informal ties to the com- munity increase individual political partici- pation (Guest and Orpesa 1986).

The bonds of social capital also affect the nature of participation-its quality. Quality of participation is important to theories of civil society that advocate the creation of the public sphere, which is a space outside es- tablished authorities for informed, reasoned, rational-critical discourse (Calhoun 1993; Habermas 1989). This idealized space should strengthen democratic virtues such as open-mindedness, tolerance, and respect for opposing viewpoints, while also creating an informed and reasoned public opinion.

that opposition groups played only a limited role in organizing protests, taking a leadership role only in the last stages when negotiation with the regime was needed. Instead, protest resulted from communication through informal associations about obvious times, dates, and places for pro- test (e.g., the 71St anniversary of the founding of Czechoslovakia, in the central city square). Of course, much movement research focuses on spe- cific policy changes rather than democratization per se. McCarthy (1996:142-45) provides an ex- tended discussion on how informal ties and asso- ciations act as mobilizing structures for social movements of all types.

Because relationships increase communi- cation and the flow of information, they in- crease the exposure of individuals to politi- cal ideas and debates. New ideas and opin- ions are more quickly disseminated through- out the population, yet extremist ideas are more easily challenged, as they have less chance of remaining isolated. Finally, the bonds of community help socialize the next generations, which ensures the future stabil- ity of democracy.

To summarize: Through participation in trusting associations, individuals may expe- rience changes in their values, preferences, and capacity to act. They should participate more in the democratic process, and the quality of their participation should increase. At the same time, it is the presence of asso- ciations, not simply changes in individuals, that provides further resources to collec- tively mobilize and pursue specific goals in a large society (Lipset et al. 1956). The ar- gument for the relationship between social capital and democracy therefore occurs at both the individual and association levels.


Previous theory and empirical research tends to treat all associations as equivalent in their effects on democracy (Stolle and Rochon 1998). Yet the associations of social capital need not always produce positive benefits for democracy (M. Olson 1982; Putnam 2000; Warren 2001). Social capital within a single group should be positive for that group, but it could be positively or negatively related to social capital at the commu- nity level. For example, a militia or ethnic separatist group might have high internal so- cial capital but exacerbate societal cleavages in the larger community. The internal social capital of other highly member-oriented as- sociations could help mobilize members in favor of policy innovations that do not ben- efit the community as a whole (Berman


The distinction mirrors Pollock's (1982) separation of "intentional" and "unintentional" mobilization, or Warren's (2001) differentiation between the developmental and institutional ef- fects of associations.

The important distinction is between groups that are tied to the wider community through associations and trust, and those that are not. Negative effects on democracy would be expected when there is high within-group trust and networks but low be- tween-group trust and networks (Paxton 1999). This distinguishes the ties and trust present in Sweden from those in Yugoslavia prior to its breakdown.

One traditional way to distinguish such groups is to consider whether they cross-cut social boundaries (Blau 1977; Blau and Schwartz 1984). Recently Putnam (2000:22) introduced a similar distinction: "bridging" versus "bonding" associations. Cross-cut- ting, or bridging, associations are connected to other associations and to the larger com- munity. These associations should traverse social boundaries, increase members' toler- ance through contact with diverse others (Allport 1954), and prevent the creation of pockets of isolated trust and networks. In contrast, isolated associations could inten- sify inward-focused behavior, reduce expo- sure to new ideas, and exacerbate existing social cleavages. For these reasons, associa- tions that are connected to the larger com- munity should be more beneficial to democ- racy than associations that remain isolated.


Although most theoretical perspectives claim causation runs from social capital to democ- racy, the reverse could be true as well (Fox 1996; Levi 1996:49-5 1; Offe 1999:73-74; Rahn et al. 1999).5 Put simply, more associa- tions would be expected to exist when gov- ernments allow them to exist. Because non- democratic regimes often actively oppose the formation of associations, political or other- wise, liberalization of a nondemocratic re- gime should provide an opportunity for asso- ciating that did not previously exist. Associ- ating outside of political parties could be- come legal, or if still illegal, might not be actively sanctioned. In addition, democrati-

Conn (1973) makes a reciprocal argument with regard to pluralism. Relevant arguments can also be found in the social movements literature (Eisinger 1973; Tarrow 1998, chap. 5).

zation should reduce regime surveillance and "disappearances," while restoring a variety of civil liberties. Together, these reforms should provide a climate that increases inter- personal trust and encourages informal rela- tionships among individuals.

Although a reciprocal relationship has never been quantitatively tested, case stud- ies of civil society and transitions to de- mocracy provide some evidence for such a reciprocal effect. In their summary of tran- sitions from authoritarian regimes in South- ern Europe and Latin America, O'Donnell and Schmitter (1988) contend that the first step in democratization is actually elite po- litical conflict rather than the independent growth of civil society, for "no transition can be forced purely by opponents against a regime which maintains the cohesion, ca- pacity, and disposition to apply repression"

(p. 21). When a crack in the system ap- pears, then opposition has the opportunity to form (Fox 1996; Tilly 1978).

The Eastern European transitions provide additional evidence. Although scholars have searched for evidence of civil society in the predemocratic Soviet bloc, in some coun- tries civil society was effectively stifled be- fore the totalitarian regimes began to thaw.6 For example, Bernhard (1993) shows that in four Eastern European countries, only in Po- land, with Solidarity and its umbrella groups acting to unite many citizens, did an active civil society lead to democratization. In East Germany and Czechoslovakia, civil society was barely present before governmental re- forms (D. Olson 1997:154). Friedheim (1993) claims that East Germany and Czechoslovakia had low levels of civil soci- ety organization, possibly because of a high degree of regime homogeneity. Tong (1994) states the issue succinctly: "Given the totali- tarian tendencies of state socialist systems, an autonomous civil society rarely emerges in a bottom-up fashion, except when the re- gime is in serious crisis. Instead, its emer- gence is often the result of top-down efforts, that is, through tolerance, encouragement, or sponsorship by state policies" (p. 334).

The literature on Eastern Europe uses the ter- minology of "civil society" almost exclusively. As discussed above, civil society is just another way to discuss a vibrant associational life-or

To summarize, theory and previous re- search point to a likely positive impact of social capital on democracy. However, iso- lated associations might have a negative im- pact on democracy. In addition, I expect the relationship between social capital and de- mocracy to be reciprocal and mutually rein- forcing.


To empirically test the link between social capital and democracy, I measure social capital and place it in a model of democracy with relevant controls. I introduce two mea- sures of social capital, each with different strengths and weaknesses. I then define the dependent measure, democracy, and addi- tional explanatory variables.

Cross-national data on social capital are sparse, but at least two sources of data are available and each has distinct strengths. The first source, the World Values Survey (WVS, World Values Study Group 1994), provides fairly precise measures of the two dimensions of social capital-associations and trust. This dataset has individual-level information on trust and voluntary associa- tion memberships in 48 countrie~.~

The WVS asks questions about individu- als' number and type of group memberships, which can be used to estimate the general level of associations in the nation as a whole. To measure associations I sum two vari- ables: the mean number of voluntary association memberships of individuals in a country, and the mean number of voluntary association memberships for which the

social capital.

'Forty-two countries from the WVS were supplemented with six from Muller and Seligson (1994). However, the Muller and Seligson data were collected in a survey that was designed to be representative of the urban population. The measured levels of trust in the Central American sample are higher than those collected for other countries in Latin America. Because these Cen- tral American countries have low levels of de- mocracy, if these levels of trust are artificially high, the test of the social capital/democracy re- lationship will be conservative.

member did unpaid voluntary work in the past year (the sum therefore provides a mea- sure of depth as well as breadth of associa- tion member~hip).~

By aggregating group memberships, I estimate the density of asso- ciations in each nation.

To measure trust, I take the percentage of individuals in each country who believe that others can be trusted (from the following question: "Would you say that most people can be trusted or that you can't be too care- ful in dealing with others?"). Here the focus is on an individual's estimate of the trust- worthiness of generalized others. Trust is important in democratic systems because in- dividuals must be willing to place political power in the hands of "the people." With low levels of observed trust, individual citi- zens would be unwilling to relinquish politi- cal power to those with opposing view- points, even for a short time, thereby pre- venting successful turnovers of power. Trust in one's particular group, or in only a seg- ment of the population, is inadequate. In di- verse societies, individuals must display generalized trust because, at any time, any group could obtain power (see Uslander 1999 for further justification).

The WVS therefore provides fairly direct measures of social capital, but has a small cross-national ample.^ Although there is re-

g One issue in aggregating variables across countries is whether the questions were inter- preted in the same way across all countries. If this were not true, then differences among coun- tries might occur simply because of an interpre- tation difference rather than from an actual dif- ference in aggregate trust or association member- ships. Recent work provides evidence that indi- viduals in the different countries did interpret the questions from the WVS in similar ways (Paxton 1998). This work used new statistical techniques (i.e., multiple group analysis with a multiple in- dicator model) to test reliability across social sys- tems as proposed by older work on methods for comparative research (Przeworski and Teune 1970). Comparability of the measures of social capital across countries is therefore not a prob- lem for the analysis.

Although most countries in the WVS have information about trust, fewer countries have in- formation about voluntary associations. The countries in which questions about memberships were not asked are: South Africa, Poland, Nige- ria, Belarus, India, Czechoslovakia, and Turkey.


gional variation in the data, the sample is weighted toward developed nations.1° There are two time periods of the WVS, 1980 and 1990, which allows for the estimation of re- ciprocal effects. "

The second source of data on social capital comes from the International Yearbook of Organizations (Union of International Asso- ciations 1990-1991). This yearbook provides information on the number of international nongovernmental organizations (INGOs) present in a much larger sample of countries. The number of INGO memberships is a sub- stantially different measure of social capital than the one taken from the WVS. To begin with, rather than a measure of individual as- sociation memberships based in a probabil- ity sample, it is an actual count of organiza- tions. The different nature of the measure is not, in itself, problematic. In fact, a count is preferable in some ways. Yet INGOs repre- sent only a specialized subset of all the asso- ciations present in a country, so this measure cannot be considered an exhaustive source of association information.

INGOs are associations. Like other volun- tary associations, INGOs require voluntary action by individuals for their continuance (Boli and Thomas 1997: 180). INGOs are also international. INGOs include the World Health Organization and the Asian Women's Human Rights Council, as well as Alcohol- ics Anonymous and the Caribbean Badmin- ton Confederation. Some of these associa-

Respondents in Switzerland were asked about only a portion of the associations.

lo The countries included in the analysis are Argentina, Australia, Austria, Belarus, Belgium, Brazil, Britain, Bulgaria, Canada, Chile, Costa Rica, Czechoslovakia, Denmark, East Germany, El Salvador, Estonia, Finland, France, Guate- mala, Honduras, Hungary, Iceland, India, Ire- land, Italy, Japan, Latvia, Lithuania, Mexico, Netherlands, Nicaragua, Norway, Panama, Po- land, Portugal, Romania, Russia, South Africa, South Korea, Slovenia, Spain, Sweden, Switzer- land, Turkey, United States, and West Germany. Some of these countries are missing data on vari- ables in one of the two time periods (N = 46).

The third wave of the World Values Survey, containing data for 1995, was recently released. Unfortunately, the wording of the question on as- sociations was changed substantially, making across-time comparison impossible.

tions, such as Rotary Clubs International, are indicative of a national network of associa- tions in a particular country.12 Others, how- ever, are less clearly indicative of national- level social capital (e.g., the Association for the Promotion of African Community Initia- tives has 30 individual members from 14 countries). Therefore, only some INGOs ex- press voluntary associations at the local level; the others present a measurement er- ror problem.

There are some benefits to considering INGOs as a measure of social capital. They are an "idealized" version of all associa- tions-those that would be expected to posi- tively affect democracy. Contrary to many other types of associations, INGOs have "resolutely democratic formal structures" and "habitually invoke the common good of humanity as a goal" (Boli and Thomas 1997: 180-8 1). INGOs represent nonlocal ties among associations within a country as well as links to the larger global community (Chatfield 1997:21). Although INGOs by design do not pressure governments for democratic reform, like transnational social movement organizations (TSMOs) do, they do promote democratic norms within their own organizations. Therefore, there is less likelihood of negative effects on democracy when using INGOs as a social capital mea- sure. Cohen (1999:244), argues that many of the most important civil society organiza- tions are now global in nature.

Although INGOs may represent the most positive segment of associations in a coun- try, when used as a measure of social capital they clearly contain measurement error. For this reason, I explicitly include measurement error in my model of INGOs. Accounting for measurement error allows an assessment of the true relationship between social capital and democracy without the potential bias from inaccurate measurement. INGO mem- berships are available for four time periods from the Yearbook of International Organi-

Nelson (1995) argues that nongovernmental organizations may be viewed as a pyramid, with many grassroots organizations at the bottom, some intermediate national organizations above them, and fewer INGOs at the top. From this per- spective, more INGOs in a country would indi- cate more NGOs as well.

zations: 1960, 1966, 1977, and 1990. An equivalent change in the number of INGOs should have a greater impact in a country with few INGO memberships than a country with more memberships. A logarithmic trans- formation of the number of INGOs can cap- ture these diminishing effects while also re- ducing skewness and the potential for outli- ers. Together, these two different measures of social capital provide a fuller picture of the relationship between social capital and democracy than either can alone.I3

Unlike social capital, there is a long tradi- tion of research on the measurement of de- mocracy. Bollen's (1998) measure of liberal democracy is the most carefully constructed measure of democracy available and ranges from 0 to 100. More information about this measure can be found in Bollen's (1998) data archive, which contains scores for 1972 through 1988. Using the components of the measure, I created the scores for 1991.

To assess the impact of social capital on de- mocracy, social capital must be inserted in a reasonable baseline model of democracy. To begin, quantitative research has consistently

l3 A number of validity checks were performed (Lin 1976: 172-74). In terms of convergent valid- ity, trust correlated highly with average associa- tion memberships (r = .69, p < .0001) and mod- erately with the number of INGOs (r = .36, p < .01). The measures of social capital also display discriminant validity, being uncorrelated (or less correlated) with a number of other cross-national concepts they are not intended to measure, in- cluding fertility and child mortality (Singleton et al. 1988). Because trust, connected associations, and INGOs are all indicators of positive social capital, they should reflect a single underlying latent variable. An auxiliary confirmatory factor analysis demonstrates that these three indicators significantly load on a single factor for the sample of countries with overlapping data. And their R-squares, as a measure of reliability (Bollen 1989b:221), were within acceptable range. As would be expected, the INGO measure had the lowest reliability. Goodness-of-fit statis- tics were not available because the model was exactly identified.

demonstrated a strong, positive relationship between economic development and democ- racy (Bollen 1983:469; Burkhart and Lewis- Beck 1994; Lipset 1960; for a review of other studies, see Rueschemeyer, Stephens, and Stephens 1992). In response to the eco- nomic development hypothesis, world sys- tem theorists argue that current developing countries face a different set of opportuni- ties than capitalist countries experienced in the past, and are, in fact, hindered in their democratic progress (Chirot 1977; O'Donnell 1973). This alternative hypoth- esis has received some support in empirical research (Bollen 1983; Bollen and Jackman 1985; Gonick and Rosh 1988). I employ the most commonly used measure of industrial- ization-the log of energy consumption per capita (United Nations, various years).14 For world system position I distinguish between countries in the "core" of the world system and those that are not in the core (taken from Burkhart and Lewis-Beck 1994). In view of the small sample size in the WVS, I did not further distinguish between semi-peripheral and peripheral countries.

Models of democracy should include some measures of the social forces that can affect democracy. I include variables with theoreti- cal merit that have been empirically justified in at least a few previous studies. Protestant traditions and whether a country was a former British colony are often used to re- flect cultural tradition and diffusion (Bollen 1979:575; Bollen and Jackman 1985; Schumpeter 1942). Other important social forces are indicated by school enrollment ra- tios (Crenshaw 1995) and ethnic homogene- ity. I include a dummy variable for countries that are predominately Protestant and one for countries that were colonized by the British and became independent after World War 11. School enrollment ratios came from the 1997 Statistical Yearbook (United Nations, various years). The percentage of the popu- lation of the most populous ethnic group is my measure of ethnic homogeneity (Sullivan 1991). These determinants of democracy are

l4 The United Nations did not provide separate information for the countries of the former

U.S.S.R. (Lithuania, Latvia, Estonia, Belarus, and Russia), so I assigned all of them the value reported for the "former USSR."

not exhaustive. Instead, because of the cross-national nature of this study, they nec- essarily capture broad, systematic forces, rather than process-oriented or path depen- dent explanations that may hold during cer- tain times or in certain regions.15

Table 1 provides descriptive information for democracy and the independent variables for both datasets. The differences between the WVS data and the larger INGO dataset highlight why using both sources in the analysis is beneficial. The countries in the smaller dataset are generally more demo- cratic, more industrialized, more Protestant, and more central to the world system. They also have higher school enrollments and greater ethnic homogeneity than countries in the larger sample. The standard deviations and ranges indicate that the smaller dataset also has less variation in democracy, indus- trialization, and ethnic homogeneity. In es- sence, with the specialized questions in the WVS comes a more restricted set of coun- tries for analysis. Also, fewer cases are available in the WVS with data on associa- tions (31 instead of 46).


The analysis is divided into several parts. First, the WVS and INGO panel analyses provide slightly different perspectives on the social capital-democracy relationship. Be- cause of its smaller sample size, the panel design for the WVS includes only those ad- ditional explanatory variables that were sig- nificant in an auxiliary cross-sectional analy-

l5 Additional variables were included in the WVS models to test for spurious relationships. First, a measure of state strength (total govern- ment expenditures as a percentage of GNP) did not significantly influence democracy and did not affect any other relationships in the model. Sec- ond, a measure of political culture-support for gradual reform (Inglehart 1997; Muller and Seligson 1994)-was included. Political culture had a positive significant effect on democracy, but no other relationships in the model changed. If political culture is viewed as aggregate demo- cratic values, then it may mediate the relation- ship between social capital and democracy. Fu- ture research could attempt to elucidate the causal relationships among these variables.

Full Information, 1990



World Values Survey

Democracy Percent who trust most people Mean number of association memberships Industrialization (energy per capita) (log) Noncore member of world system Protestant country School enrollment ratio Ethnic homogeneity

Mean number of connected association memberships

Mean number of isolated association memberships

International NGOs

Democracy score, 1991

Number of international NGOs, 1990 (unlogged) Industrialization (energy per capita) (log) Noncore member of world system Protestant country Former British colony School enrollment ratio Ethnic homogeneity Range

Number of Standard Countries Mean Deviation Minimum Maximum

sis. After presenting these models, I investigate whether cross-cutting associations have a different impact on democracy than do non- cross-cutting associations. This analysis is possible using the WVS, which breaks out associations by type.



I begin with the WVS and analyze the model displayed in Figure 1. At the heart of this structural equation model are the cross- lagged relationships between social capital and democracy. The two components of so- cial capital, mean number of associations and trust in 1980, affect democracy in 1991. Likewise, a measure of democracy in 1982 affects social capital in 1990. Stability pa- rameters are also estimated (social capital in 1980 has effects on social capital in 1990, and democracy in 1982 affects democracy in 1991). The errors in the equations of social capital and democracy are correlated to re- flect possible covariation between social capital and democracy that is not captured by the cross-lagged, stability, or control ef- fects of the model.

The main concern when estimating this model was the small sample size of the WVS. In addition, comparatively more countries are available for the 1990 wave, creating an "unbalanced" panel design (Finkel 1995). There were 23 countries in common across the two waves. The smaller number of cases for 1980 demands that I account for missing data in some way in or- der to be able to use the full 46 cases. A full information maximum-likelihood missing value routine implemented in AMOS (Arbuckle 1995) estimates the model with

Figure 1. Cross-Lagged Panel Model of Social Capital and Democracy: World Values Survey

the maximum amount of information avail- able for each case and allows me to main- tain a sample size of 46.16Another issue in the WVS panel was that two outliers were identified in a cross-sectional analysis- China and Nigeria. I report the results with those two outliers omitted. Although the es- sential pattern of results does not change with the inclusion or exclusion of those cases, there is a significant negative corre- lation between two of the errors of the

l6 Maximum-likelihood missing value estirna- tion is not an imputation procedure. Instead, the likelihood for the entire sample is created by summing the likelihoods for each case, using whatever information each case has available. This means that the 46 cases are contributing the maximum amount of information possible to the estimation of the model. The estimates are con- sistent and efficient under the condition that the data are missing at random (Arbuckle 1996).This is an easier condition to meet than the assump- tion that the data are missing completely at ran- dom, which some other missing value procedures require. See Allison (2002) for a review of tech- niques for dealing with missing data.

equations with China and Nigeria in the model. Negative correlations between errors can be an indication of poor fit in simi- lar synchronous simultaneous equations models (Gillespie and Fox 1980). When these two influential cases are removed, the negative correlations are no longer signifi- cant.

The most important result in this part of the analysis is that democracy significantly affects the two components of social capi- tal-associations and trust-while the re- verse effect is not found. This finding stands in stark contrast to the ubiquitous theory in the area. Table 2 shows that democracy in 1982 influences both associations and trust. Although the effect of democracy on asso- ciations is substantively marginal, a 10-unit increase in a counuy's democracy score in- creases the percentage of individuals who say that others can be trusted by approxi- mately 1 percentage point. We see these re- sults despite high stability in trust and asso- ciations over time.

As for the other explanatory variables, there is a positive effect of industrialization

Table 2. Coefficients from Panel Model of Social Capital and Democracy: World Values Survey,

1980 and 1990
Independent Variable 1991
Democracy, 1982 .18***
Associations, 1980  
Percent who trust most people, 1980 .13
Industrialization, 1980  
Noncore position in world system .5 1
Ethnic homogeneity  
R2 .63
Mean Number of Associations, 1990 Percent Who Trust, 1990
.01* (.(lo) .13*** (.03)
-.01 -1.15
(.36) (2.77)
.64 .9 1

Note: Numbers in parentheses are standard errors; N = 46. Goodness-of-fit statistics: x2 = 4.7, d.f. = 2, p = .09; IF1 = .99; NFI = .98; RMSEA = .17.

*p< .05 **p< .O1 ***p< ,001 (two-tailed tests)

+p < .05 (one-tailed tests)

on democracy, while noncore position in the world system does not have a signifi- cant effect. We see a positive effect of eth- nic homogeneity on democracy as well. The three control variables do not strongly in- fluence social capital in the later time pe- riod. The only exception is industrializa- tion, which has a negative effect on trust. Although disappointing, this finding is not entirely unexpected by theories that argue that industrialization breaks down trust (Fukuyama 1999). The R2s for the three equations in the model are high, including an R2 of .91 for the trust equation, reflect- ing its strong stability effect.

Structural equation models like the one in Figure 1 can be evaluated for their goodness of fit to the data (Bollen 1989b:256-89). Fit statistics for the model appear in the note to Table 2 and indicate that the model fits the data well. Consider first the chi-square test statistic-we do not reject the null hypoth- esis that the model fits the data perfectly (al- though the p-value is close to conventional levels of significance). Other fit statistics agree with the chi-square: The incremental fit index (IFI) (Bollen 1989a) and the normed fit index (NFI) (Bentler and Bonett 1980) both indicate an excellent fit. The only indication of a poor fit is the Root Mean Square Error of Approximation (RMSEA).17 Overall, the picture provided by the fit sta- tistics is of a well-fitting model.


Sample size and missing data are not a con- cern for the INGO analyses. Instead, I must account for the likelihood of measurement error. I begin by estimating a model without measurement error and then demonstrate its robustness to various levels of measurement error.

Figure 2 shows a four-wave cross-lagged panel model for the INGO data. As in the WVS analysis, the heart of this model (indi- cated by lines in bold) includes the cross- lagged and stability paths between social capital and democracy. In 1966, 1977, and 1990, INGOs and democracy are determined by their own values in the previous period,

"The closer the IF1 and NFI are to 1.0, the better the fit of a model. The closer the RMSEA is to 0, the better the fit of the model. Generally, values below .05 are considered to be an excel- lent fit, between .05 and .lo, good, and values above .I0 indicate some problem with the fit of the model (Browne and Cudeck 1993).

Figure 2. Four-Wave Cross-Lagged Panel Model of Social Capital and Democracy: International Nongovernmental Organizations

*Errors in the variable equations are intercorrelated at every time point.

the lagged value of the other variable, some exogenous variables, and an error term. All of the explanatory variables except former British colony are hypothesized to influence both democracy and INGOs in every time period. Former British colony is not hypoth- esized to influence INGOs. The errors in the equations for INGOs and democracy are cor- related in each time period to reflect possible covariation not captured by the rest of the model.

Table 3 presents the results of the INGO panel model. In contrast to the WVS panel, the INGO analysis reveals significant cross- lagged effects for both democracy and social capital. The effects are not simultaneous but instead alternate by year-democracy has an effect on INGOs in the first time period (1960 to 1965), while INGOs affect democ- racy in the later time periods (1965 to 1977 and 1977 to 1991). Overall, the effect of the number of INGOs on democracy increases over the time period, while the effect of de- mocracy on the number of INGOs decreases over the time period. Because the number of INGOs is logged, the effect sizes of the re- ciprocal relationship are not immediately obvious from the coefficients. In 199 1, for a 10-percent increase in INGO memberships, a country's democracy score is expected to increase by 1. For example, if a country's INGOs increase by 10 percent-from 50 to 55, or from 500 to 550-its democracy score, holding other variables constant, would be expected to increase by 1. Note that for a country with a small number of INGOs, a small number of additions can make a substantial difference whereas for a country with a large number of INGOs, a proportionately larger number of additions is needed for the same impact on democracy. For the reciprocal effect, a 10-unit increase in democracy would result in a 6-percent in- crease in INGOs. This analysis provides more support for the traditionally hypoth- esized direction of causality, with social capital significantly affecting democracy in two of the three time periods. We continue

Table 3. Coefficients from a Four-Wave Cross-Lagged Model of Social Capital and Democracy: International Nongovernmental Organizations, 1966,1977, and 1990

Democracy Number of INGOs (In)
Independent Variable 1965 1977 1991 1966 1977 1990
1960 .838*** - - .006*** - -
  (.047)     ,002    
1965 - .45 1 *** - - .OOOl -
    (.loo)     (.0007)  
1977 - - .358*** - - .0006
      (.092)     (.OOl)
Number of INGOs (In)    
1960 1.060 - - .393*** - -
  (.999)     (.034)    
1966 - 8.672* - - .867*** -
    (3.792)     (.027)  
1977 - - 10.813** - - .749***
      (4.059)     (.045)

Noncore position in world system -4.471 -23.750*

(4.220) (8.684)

Protestant country 2.500 7.486

(3.281) (6.558)

Former British colony 5.266+ 2.605

(3.151) (6.367)

School enrollments .065 ,020 (.093) (.18 1)

Ethnic homogeneity .133** .266* (.051) (.103)

R2 .87 .69 .67 .89 .97 .92

Note: Numbers in parentheses are standard errors; N = 101 countries. Goodness-of-fit statistics: x2=47.7,

d.f. = 27, p = ,008; IF1 = 1.0; NFI = .99; RMSEA = .09 *p< .05 **p< .01 ***p < ,001 (two-tailed tests) +p = .05 (one-tailed tests)

to see evidence for the reciprocal effect of has a positive, significant effect on democ- democracy on social capital, however. l8 racy. Its effect in the second time period

The dummy variable for former British (1977) is not significantly different from colony displays a fascinating effect over zero, and in the final period (1991) it has a time. In the earliest time period (1965), it

the model fits the data well. The IF1 and NFI are

l8 The fit of the model is good. Although the 1.0 and .99 respectively while the RMSEA is .09 chi-square test statistic is significant (47.7, d.f. = (values below .10 indicate a good fit). Regarding 27, p = .008), the other fit statistics indicate that the significant chi-square value, I found evidence

significant negative effect. Previous research has found an unequivocal positive effect on democracy for British colonies (e.g., Bollen and Jackman 1985). But most of those stud- ies were conducted using data from the 1960s and 1970s. This longer-term analysis indicates that while the transition to self- rule, as facilitated by the British, enhanced democracy early in a country's indepen- dence, that effect diminished over time. Eventually, those countries experienced democratic reversals. We again see very high R2s, ranging from .67 to .97, for the equa- tions in the model.

Because some measurement error is likely in the INGO variables, I also explicitly in- clude measurement error to test the robust- ness of the results in the face of mild, mod- erate, and severe levels of measurement er- ror. That is, I model the INGO variables with measurement error and fix the variance of that measurement error to predetermined values. The values were chosen to represent reliabilities of .9, .8, and .7 (mild, moderate, and severe measurement error). For more in- formation on this procedure, see Bollen (1989b: 168-71).19 Only under the condition of severe measurement error do the results change substantially.

With reliabilities of .9 and .8 for INGOs, there are some changes in the coefficients of both the cross-lagged portion of the model and in the paths from the control variables (minor at a reliability of .9 but with increases in magnitude at a reliability of .8). The largest changes occur in the INGO equations, as expected, and center around the stability effects for INGOs, the coefficients for noncore position in the world system, and the coefficients for school enrollment. There are no changes in significance, however, until a reliability of .7 is used. There, in the cross-lagged por- tion of the model, the path from democracy

for some multivariate nonnormality in the data. So I bootstrapped the chi-square (Bollen and Stine 1993) and obtained a p-value of .79-a nonsignificant value indicating an excellent fit to the data.

l9 Single indicators are less able than multiple indicators to explore random or nonrandom mea- surement error. However, this procedure tests the sensitivity of the results to varying levels of ran- dom error and to nonrandom specific error.

in 1960 to number of INGOs in 1966 be- comes nonsignificant, the path from number of INGOs in 1977 to democracy in 1991 be- comes nonsignificant (with a severe change in the coefficient as well), and the path from democracy in 1977 to INGOs in 1990 becomes significant. Measurement error in- fluences our assessment of the democracy- social capital relationship, but only when severe, and then in an ambiguous direction. Changes in significance for some of the control variables (noncore position in the world system and school enrollment) are seen as well. The fit of the model declines substantially as the reliability of the vari- ables is lowered. In addition, first one, then two negative error variances appear in the model as measurement error is increased. Overall, this analysis indicates that the find- ings for the INGO analysis are robust to mild and moderate levels of measurement error.

What can we conclude from the panel analysis of the WVS and INGOs? While the INGO analysis supports the hypothesized ef- fect of social capital on democracy, it also supports a reciprocal effect from democracy to social capital. The WVS analysis shows only the effect from democracy to social capital. These effects hold through the inclu- sion of controls for other relevant variables and the introduction of measurement error to the INGO variables. At a minimum, there- fore, these results indicate that we can no longer ignore a reciprocal path from democ- racy to social capital.


Still to be addressed are the potential differ- ences in impact by type of association. To reiterate my earlier arguments, although as- sociations that are connected to the wider community are expected to positively influ- ence democracy, isolated associations could be detrimental to democracy. At a minimum, membership in an association with ties to other associations would keep individuals from being isolated. I assess the connected- ness of associations by looking at the mul- tiple memberships of their members. Asso- ciations whose members have many ties to

1   1 1   1 111 1   1 11
r   I I   I       I     I
.5   .6 .7   .8       .9     1.O
    Proportion of Members with Other Memberships, by Association      
    Average Number of Memberships, by Association      

Figure 3. Differentiating Associations

Note: Standardized scores are in parentheses.

other types of associations are more likely to cross-cut social boundaries and promote contact with diverse others than associations with fewer such ties. Figure 3 presents, for each type of association in the WVS, the proportion of members who have at least one other membership and the mean number of other memberships of members, respec- tively. Three associations stand out as less connected than others: trade unions, sports associations, and religious associations. In addition to their obvious separation from the other associations, these are the only asso- ciations with standardized scores of less

than -1.0.~~

I code these three types of as- sociations as isolated and the others as con- nected. This produces two variables to mea- sure associations in each country: the mean number of connected voluntary association memberships of individuals in a country,

20 The isolated nature of these associations is confirmed in another dataset. The American Citi- zen Participation Survey (Verba, Schlozman, Brady, and Nie 1995) contains detailed informa- tion on the memberships of respondents. Using a variety of measures of connectedness produces low scores for trade unions, sports associations, and religious groups.

Table 4. Coefficients from the Model of Social Capital and Democracy: World Values Survey, 1980 and 1990

Democracy, Connected Isolated Percent Who
Inde~endent Variable 1991 Associations. 1990 Associations. 1990 Trust. 1990
Democracy, 1982 .22*** .Ole* .01* .14***
  (.05) (.OO) (.OO) (.02)
Connected associations, 34.23' .63*    
1980 (20.16) (.38)    
Isolated associations, -40.75* -    
1980 (22.01) -    
Percent who trust most .23* -    
others, 1980 (.I31      
Industrialization, 1980 1.16 .07 .06 -2.26**
  (3.48)   (.07) (34)
Noncore position in 2.32      
world system (4.78)      
Ethnic homogeneity .25** -    
R~ .66 .54 .73 .9 1

Note: Numbers in parentheses are standard errors; N =46. Goodness-of-fit statistics: x2 = 13.17, d.f. = 12; p = .36; IF1 = 1.0; NFI = .95; RMSEA = ,047.

*p< .05 **p < .01 ***p < ,001 (one-tailed tests)

and the mean number of isolated associa-   ciations (which has a range of .23 to 2.1) is
tion~.~'   expected to increase a country's democracy
Table 4 presents the WVS model with av-   score by 34 units. A one-unit increase in iso-
erage association memberships split into two   lated associations (range of .18 to 1.6) de-
types.22 The results indicate that connected   creases democracy by 41 units.
associations have a strong positive influence   The relatively similar size of these coeffi-
on democracy, while isolated associations   cients helps to explain why the overall ef-
have a strong negative influence on democ-   fect of associations on democracy in the
racy. A one-unit increase in connected asso-   WVS analysis (Table 2) was not signifi-
    cantly different from zero. This analysis also
21 The results presented below are robust to a variety of modifications. These include breaking out associations by the memberships in which in- dividuals do unpaid work (a conservative esti- mate of membership) and by both membership and unpaid work. Including the associations with   helps account for the divergence in the ef- fect of social capital on democracy across the original WVS analysis and the INGO analysis (Tables 2 and 3). Recall that INGOs are more likely to be nonlocal, internally democratic, and have links to global demo-
the next lowest standardized scores as non-cross-   cratic norms.23 Associations had a positive
cutting does not change the results either. If as-   impact on democracy in the INGO analysis
sociations are divided along political lines (using information from the American Citizen Partici- pation Survey), neither politicized nor non-politicized associations had an effect on democ- , because of this specialized nature. In this model, trust has a positive effect on democracy, lending further support to
22 The model is also trimmed, with the nonsig-   23 Indeed, a simple correlation analysis shows
nificant coefficients from Table 2 removed. This   that number of INGOs correlates positively and
strategy conserves the ratio of parameters to   significantly with connected associations but is
cases. The results do not change if all paths are   not significantly correlated with isolated associa-
included.   tions.

the hypothesis that social capital affects de- mocracy. Yet the return effect of democracy on trust and associations persists, indicating that the relationship is reciprocal. Democ- racy increases both cross-cutting and non- cross-cutting associations, which corresponds to theory-if liberalization enhances the climate for associations, this should oc- cur for all associations.

The fit of the model with both measures of associations is substantially better than the fit of the model with the single measure of associations. The chi-square test statistic is now solidly nonsignificant, indicating that the model fits the data perfectly. More strik- ing is the substantial decrease in the RMSEA, from .17 to .047. The RMSEA now falls within the range for good-fitting models (Browne and Cudeck 1993: 144).

In a parallel analysis, I tested a model with an interaction between trust and associa- tions. The interaction effect was positive and significant (2.12, p < .001) while the main effects for trust and associations were nega- tive (-1.98,~ < .001, and-100.9,~ < .001, respectively). This means the impact of as- sociations on democracy depends on the level of trust present in society. At low lev- els of trust, increases in association member- ships have a negative impact on democracy. For example, when only 24 percent of the population expresses trust, the effect of a one-unit change in associations on democ- racy is -50 (-100.9 + [2.12 x 241). Only when approximately 50 percent of the popu- lation is trusting do increases in associations lead to increases in democracy. These results support the findings in Table 4-associa- tions in a climate of distrust are potentially harmful to democracy (Paxton 1999). Over- all, these results provide evidence that the type of associations present in a country has implications for its level of democracy.24

24 Although this analysis focused on the aggre- gate level, a note on individual-level relation- ships is relevant. If trusting associations promote tolerance and political participation, then indi- vidual-level relationships should demonstrate this effect. Using individual-level responses from the WVS, I tested whether connected versus iso- lated associations had different effects on intol- erance toward different groups, measured as the number of different groups the respondent would not like as neighbors (Muslims, Jews, immi-


Current democratic theory and recent inter- national policy initiatives reveal an intense interest in the relationship between social capital and democracy. I began by outlining the theoretical justification for the long- standing thesis that vibrant associations help create and maintain democracy. I then pre- sented an equally plausible alternative hy- pothesis-that democracy can increase social capital through a reciprocal effect. Quantita- tive, cross-national tests of both the conven- tional hypothesis and the alternative hypoth- esis were introduced. Social capital was operationalized as associations and trust us- ing two sources of data-the World Values Survey (World Values Study Group 1994) and international nongovernmental organiza- tions (Union of International Associations 1990-1991). I also tested the hypothesis that connected associations are more beneficial to democracy than are isolated associations.

The findings have shown that the relation- ship between social capital and democracy is reciprocal. In the panel analyses, social capital was found to promote democracy while a return effect from democracy to so- cial capital was also established. The analy- sis also confirmed that certain types of asso- ciations do better in promoting democracy. When associations were broken into two types using the WVS, connected associa- tions had a strong positive influence on de- mocracy, while isolated associations had a strong negative influence on democracy.

grants, etc.), and political efficacy, measured as the reverse-coded response to the question, "If an unjust law were passed by the government, I could do nothing at all about it." Interestingly, connected associations had a significant negative effect on intolerance, while isolated associations had no significant effect. In predicting "If an un- just law were passed, I could do nothing," both connected and isolated associations had a nega- tive effect. However, the coefficient for connected associations was three times the size of the coefficient for isolated associations. In both cases, the R-square for the model with indi- vidual-level indicators of social capital was con- siderably higher than a model with basic demo- graphic variables. These analyses are intriguing and should be investigated further in future re- search.

Overall, these findings demonstrate that to fully understand how democracy and social capital are related, both their reciprocal na- ture and the possibility of negative effects must be recognized.

There are some important limitations of the present analysis. Essentially, the analy- sis sailed between the Scylla of sample size and the Charybdis of precise measurement. The WVS provided better measures of social capital than did the Union of International Associations, but it had a smaller sample. The smaller sample meant that the develop- mental variation of the WVS sample was somewhat limited. Most of the countries come from western or Eastern Europe, with a few countries from South America. Africa, the Middle East, and Asia remain under- represented in the sample. Most of the find- ings of the WVS were replicated in the INGO analysis, with its larger sample size and greater developmental variation giving more confidence in the overall findings.

Further, because this analysis remained largely at the aggregate level, it cannot adju- dicate whether the effect of social capital on democracy is due to the different values vi- brant associations may impart to their mem- bers or to specific activities undertaken by the associations themselves (intentional ver- sus unintentional mobilization). Disentan- gling the two explanations could be difficult, but future research using multilevel models, comparative or case-study techniques, or data with greater detail about the character- istics of associations may bear fruit. In addi- tion, future research should consider the pos- sible correlation of cases resulting from as- sociation ties across national boundaries (Hironaka 2002), the invigoration of protests through demonstration effects (Karklins and Petersen 1993:601-602), or mimicry by gov- ernment officials (see Markoff 1996b: 26-34 for a discussion of the transnational context of democratization).

With larger and larger proportions of their aid being funneled to NGOs (Edwards and Hulme 1996), international funding agencies are clearly committed to the view that social capital will positively increase democracy. The present analysis has provided some jus- tification for this position. Yet it also pro- vides some cautionary evidence: (1) Some types of social capital may be detrimental to democracy, and (2) social capital can in turn be affected by democracy. Funding agencies should consider these issues in funding NGOs to foster democracy.

First, if certain types of associations have substantially greater benefits for democracy, or if certain types of associations are actu- ally harmful to democracy, then agencies that fund NGOs or nonprofit organizations should consider a differentiated funding policy. With evidence that certain types of social capital are more beneficial to democracies, the cre- ation of cross-cutting associations and trust among individuals should be the focus of at- tempts to increase social capital. Ultimately, a more precisely focused donor policy could help the national and international commu- nity foster more stable, healthy, and effective democracies.

In addition, donor policy should recognize that a singular focus on funding NGOs and grassroots organizations may not, by itself, produce democratization. Instead, the insti- tutional environment in which NGOs are embedded is important for their growth and expansion. Other strategies to encourage de- mocratization, such as promoting the rule of law, assuring respect for human rights, or encouraging transparent governmental pro- cesses, may result in increases in social capi- tal or civil society indirectly.

Pamela Paxton is Assistant Professor of Sociol- ogy at The Ohio State University. She has inter- secting interests in political sociology, stratifica- tion, and methodology. Current projects include the political consequences of prejudice against immigrants and ideological determinants of women's national legislative participation. Some of her recent articles include "Is Social Capital Declining in the United States? A Multiple Indi- cator Assessment" (American Journal of Sociol- ogy, 1999, vol. 105, pp. 88-127), and "Women's Suffrage in the Measurement of Democracy: Problems of Operationalization" (Studies in Comparative International Development, 2000, vol. 35, pp. 92-1 11).


Allison, Paul. 2002. Missing Data. Thousand Oaks, CA: Sage. Allport, Gordon W. 1954. The Nature of Preju- dice. Cambridge, MA: Addison-Wesley. Almond, Gabriel A. and Sidney Verba. 1963. The Civic Culture: Political Attitudes and Democ-

racy in Five Nations. Boston, MA: Little, Brown, and Co.

Arbuckle, James L. 1995. AMOS 3.5 for Windows. Chicago, IL: SmallWaters Corp. . 1996. "Full Information Estimation in the Presence of Incomplete Data." Pp. 243-77 in Advanced Structural Equation Modeling: Issues and Techniques, edited by G. A. Marcoulides and R. E. Schumaker. Mahwah, NJ: Lawrence Erlbaum Associates.

Arendt, Hannah. 1948. The Origins of Totalitari- anism. New York: Harcourt and Brace.

Barnes, Samuel H. and Max Kaase. 1979. Political Action: Mass Participation in Five West- ern Democracies. Beverly Hills, CA: Sage.

Becker, Gary S. 1964. Human Capital: A Theo- retical and Empirical Analysis, With Special Reference to Education. New York: National Bureau of Economic Research.

Bentler, Peter M. and Douglas G. Bonett. 1980. "Significance Tests and Goodness of Fits in the Analysis of Covariance Structures." Psychological Bulletin 88:588-606.

Berman, Sheri. 1997. "Civil Society and the Col- lapse of the Weimar Republic." World Politics 49:401-29.

Bernhard, Michael. 1993. "Civil Society and Democratic Transition in East Central Eu- rope." Political Science Quarterly 108:307


Blau, Peter M. 1977. Inequality and Heterogene- ity: A Primitive Theory of Social Structure. New York: Free Press. Blau, Peter M. and Joel E. Schwartz. 1984. Crosscutting Social Circles: Testing a Macro- structural Theory of Intergroup Relations. Orlando, FL: Academic. Boli, John and George M. Thomas. 1997. "World Culture in the World Polity: A Century of In- ternational Non-Governmental Organization." American Sociological Review 62: 17 1-90. Bollen, Kenneth A. 1979. "Political Democracy and the Timing of Development." American Sociological Review 44:572-87. . 1983. "World-System Position, Depen- dency, and Democracy: The Cross-National Evidence." American Sociological Review 48:468-79. 1989a. "A New Incremental Fit Index for General Structural Equation Models." Sociological Methods and Research 17:303-16. . 1989b. Structural Equations With Latent Variables. New York: Wiley. . 1998. Cross National Indicators of Lib- eral Democracy, 1950 to 1990. Codebook. University of North Carolina, Chapel Hill, NC.

Bollen, Kenneth A. and Robert W. Jackman. 1985. "Economic and Noneconomic Determi- nants of Political Democracy in the 1960s." Research in Political Sociology 1:27-48.

Bollen, Kenneth A. and Pamela Paxton. 2000. "Subjective Measures of Liberal Democracy." Comparative Political Studies 33:58-86.

Bollen, Kenneth A. and Robert A. Stine. 1993. "Bootstrapping Goodness of Fit Measures in Structural Equation Models." Pp. 11 1-35 in Testing Structural Equation Models, edited by

K. A. Bollen and J. S. Long. Newberry Park, CA: Sage.

Bourdieu, Pierre. 1983. "Forms of Capital." Pp. 241-58 in Handbook of Theory and Research for the Sociology of Education, edited by J. G. Richardson. New York: Greenwood.

Browne, Michael and Robert Cudeck. 1993. "Al- ternative Ways of Assessing Model Fit." Pp. 136-62 in Testing Structural Equation Models, edited by K. A. Bollen and J. S. Long. Newbury Park, CA: Sage.

Burkhart, Ross E. and Michael S. Lewis-Beck. 1994. "Comparative Democracy: The Eco- nomic Development Thesis." American Politi- cal Science Review 88:903-10.

Calhoun, Craig. 1993. "Civil Society and the Public Sphere." Public Culture 52:67-80.

Chatfield, Charles. 1997. "Intergovernmental and Nongovernmental Associations to 1945." Pp. 19-41 in Transnational Social Movements and Global Politics: Solidarity Beyond the State, edited by J. Smith, C. Chatfield, and R. Pagnucco. Syracuse, NY: Syracuse University Press.

Chirot, Daniel. 1977. Social Change in the 20th- Century. New York: Harcourt, Brace, and Javanovich.

Cohen, Jean. 1999. "Trust, Voluntary Associa- tions, and Workable Democracy: The Contem- porary American Discourse of Civil Society." Pp. 20848 in Democracy and Trust, edited by

M. E. Warren. Cambridge, England: Cam- bridge University Press.

Coleman, James S. 1988. "Social Capital in the Creation of Human Capital." American Jour- nal of Sociology 94(supp.):S95-S120. . 1990. Foundations of Social Theory. Cambridge, MA: Harvard University Press.

Conn, Paul H. 1973. "Social Pluralism and De- mocracy." American Journal of Political Sci- ence 17:237-54.

Crenshaw, Edward M. 1995. "Democracy and Demographic Inheritance: The Influence of Modernity and Proto-Modernity on Political and Civil Rights, 1965 to 1980." American So- ciological Review 60:702-18.

Curtis, Russell L. and Louis A. Zurcher. 1973. "Stable Resources of Protest Movements: The Multi-Organizational Field." Social Forces 52:53-61.

Edwards, Michael and David Hulme. 1996. "Too Close for Comfort? The Impact of Official Aid on Nongovernmental Organizations." World

Development 24:961-73.

Eisinger, Peter K. 1973. "The Conditions of Pro- test Behavior in American Cities." American Political Science Review 67: 11-28.

Ekiert, Grzegorz. 1997. "Rebellious Poles: Politi- cal Crises and Popular Protest Under State So- cialism, 1945-89." East European Politics and Societies 11:299-339.

Fatton, Robert, Jr. 1995. "Africa in the Age of Democratization: The Civic Limitations of Civil Society." African Studies Review 38:67


Ferguson, Adam. [I7671 1995. An Essay on the History of Civil Society. Edited by F. Oz- Salzberger. Reprint, Cambridge, England: Cambridge University Press.

Finifter, Ada. 1974. "The Friendship Group as a Protective Environment for Political Devi- ants." American Political Science Review 68: 607-25.

Finkel, Stephen E. 1995. Causal Analysis with Panel Data. Thousand Oakes, CA: Sage.

Fox, Jonathan. 1996. "How Does Civil Society Thicken? The Political Construction of Social Capital in Rural Mexico." World Development 24:1089-103.

Friedheim, Daniel V. 1993. "Bringing Society Back into Democratic Transition Theory after 1989: Pact Making and Regime Collapse." East European Politics and Societies 74:82


Fukuyama, Francis. 1999. The Great Disruption: Human Nature and the Reconstitution of So- cial Order. New York: Free Press.

Gillespie, Michael and John Fox. 1980. "Specifi- cation Error and Negatively Correlated Distur- bances in 'Parallel' Simultaneous Equation Models." Sociological Methods and Research 8:273-308.

Goldfarb, Jeffrey C. 1978. "Social Bases of In- dependent Public Expression in Communist Societies." American Journal of Sociology 83:920-39.

Gonick, Lev S. and Robert M. Rosh. 1988. "The Structural Constraints of the World-Economy on National Political Development." Comparative Political Studies 2 1 :17 1-99.

Guest, Avery M. and R. S. Orpesa. 1986. "Infor- mal Social Ties and Political Activity in the Metropolis." Urban Affairs Quarterly 21:550


Habermas, Jorgen. 1989. The Structural Trans- formation of the Public Sphere: An Inquiry into a Category of Bourgeois Society. Cambridge, MA: MIT Press.

Hankiss, Elemer. 1989. "Demobilization, Self- Mobilization, and Quasi-Mobilization in Hun- gary, 1948-1987." Eastern European Politics and Societies 3: 105-5 1.

Hironaka, Ann. 2002. "World Polity Influences on Political Democratization." Working paper, Department of Sociology, University of Min- nesota, Minneapolis, MN.

Horkheimer, Max. 1947. Eclipse of Reason. New York: Oxford University Press.

Inglehart, Ronald. 1990. Culture Shift in Ad- vanced Industrial Society. Princeton, NJ: Princeton University Press.

. 1997. Modernization and Postmodern- ization: Cultural, Economic, and Political Change in 43 Societies. Princeton, NJ: Princeton University Press.

Karklins, Rasma and Roger Petersen. 1993. "De- cision Calculus of Protesters and Regimes: Eastern Europe 1989." The Journal of Politics 55:588-614.

Knack, Stephen and Philip Keefer. 1997. "Does Social Capital Have an Economic Payoff? A Cross-Country Investigation." The Quarterly Journal of Economics 62: 1251-88.

Kubik, Jan. 1998. "Institutionalization of Protest during Democratic Consolidation in Central Europe." Pp. 131-52 in The Social Movement Society: Contentious Politics for a New Cen- tury, edited by D. S. Meyer and S. Tarrow. New York: Rowman and Littlefield.

Leighley, Jan E. 1995. "Attitudes, Opportunities, and Incentives: A Field Essay on Political Par- ticipation." Political Research Quarterly

48: 181-209.

Levi, Margaret. 1996. "Social and Unsocial Capi- tal: A Review Essay of Robert Putnam's Mak- ing Democracy Work." Politics and Society 24~45-55.

Lin, Nan. 1976. Foundations of Social Research. New York: McGraw-Hill.

Linz, Juan J. and Alfred Stepan. 1996. Problems of Democratic Transition and Consolidation: Southern Europe, South America, and Post- Communist Europe. Baltimore, MD: Johns Hopkins University Press.

Lipset, Seymour M. 1960. Political Man: The Social Bases of Politics. Garden City, NJ: Doubleday.

Lipset, Seymour M., Martin A. Trow, and James

S. Coleman. 1956. Union Democracy: The In- ternal Politics of the International Typographical Union. Glencoe, IL: Free Press.

Markoff, John. 1996a. The Abolition of Feudal- ism: Peasants, Lords, and Legislators in the French Revolution. University Park, PA: Penn- sylvania State University Press. . 1996b. Waves of Democracy: Social Movements and Political Change. Thousand Oaks, CA: Pine Forge.

Marwell, Gerald and Pamela Oliver. 1993. The Critical Mass in Collective Action: A Micro- Social Theory. New York: Cambridge Univer- sity Press.

McCarthy, John D. 1996. "Constraints and Op- portunities in Adopting, Adapting, and Invent- ing." Comparative Perspectives on Social Movements, edited by D. McAdam, J. D. McCarthy, and M. N. Zald. Cambridge, En- gland: Cambridge University Press.

. 1997. "The Globalization of Social Movement Theory." Pp. 243-59 in Transnational Social Movements and Global Poli- tics: Solidarity beyond the State, edited by J. Smith, C. Chatfield, and R. Pagnucco. Syra- cuse, NY: Syracuse University Press.

Michta, Andrew A. 1997. "Democratic Consoli- dation in Poland after 1989." Pp. 66-108 in The Consolidation of Democracy in East-Cen- tral Europe, edited by K. Dawisha and B. Parrott. Cambridge, England: Cambridge Uni- versity Press.

Milgram, Stanley. 1965. "Liberating Effects of Group Pressure." Journal of Personality and Social Psychology 1:127-34.

Montesquieu, Charles de Secondat. [I7481 1989. The Spirit of the Laws. Translated and edited by A. M. Cohler, B. C. Millerm, and H. S. Stone. Cambridge, England: Cambridge Uni- versity Press.

Morris, Aldon D. 1981. "Black Southern Student Sit-In Movement: An Analysis of Internal Or- ganization." American Sociological Review 46:744-67. . 1984. The Origins of the Civil Rights Movement: Black Communities Organizing for Change. New York: Free Press.

Muller, Edward N. and Mitchell A. Seligson. 1994. "Civic Culture and Democracy: The Question of Causal Relationships." American Political Science Review 88:635-52.

Nelson, Paul. 1995. The World Bank and Non- governmental Organizations: The Limits of Apolitical Development. New York: St. Martin's Press. Oberschall, Anthony. 1973. Social Conflict and Social Movements. Englewood Cliffs, NJ: Prentice-Hall. . 1993. Social Movements: Ideologies, In- terests, and Identities. New Brunswick, NJ: Transaction.

O'Donnell, Guillermo A. 1973. Modernization and Bureaucratic-Authoritarianism: Studies in Latin American Politics. Berkeley, CA: Insti- tute of International Studies, University of California.

O'Donnell, Guillermo A. and Philippe C. Schmitter. 1986. Transitions from Authoritar- ian Rule: Tentative Conclusions about Uncer- tain Democracies. Baltimore, MD: Johns Hopkins University Press.

Offe, Claus. 1999. "How Can We Trust Our Fel- low Citizens?" Pp. 42-87 in Democracy and Trust, edited by M. E. Warren. Cambridge, England: Cambridge University Press.

Olson, David M. 1997. "Democratization and Political Participation: The Experience of the Czech Republic." Pp. 150-96 in The Consoli- dation of Democracy in East-Central Europe, edited by K. Dawisha and B. Parrott. Cam- bridge, England: Cambridge University Press.

Olson, Mancur. 1982. The Rise and Decline of Nations: Economic Growth, StagfZation, and Social Rigidities. New Haven, CT: Yale Uni- versity Press.

Opp, Karl-Dieter and Christiane Gern. 1993. "Dissident Groups, Personal Networks, and Spontaneous Cooperation: The East German Revolution of 1989." American Sociological Review 58:659-80.

Pacek, Alexander C. and Roger E. Kanet. 1991. "Revolutionary Change in Eastern Europe: The Societal Basis of Political Reform." Pp. 187- 217 in Reform and Transformation in Commu- nist Systems: Comparative Perspectives, edited by I. J. Kim and J. S. Zacek. New York: Paragon.

Paxton, Pamela. 1998. "Capitalizing on Commu- nity: Social Capital and the Democratic Soci- ety." Ph.D. dissertaion, Department of Sociol- ogy, University of North Carolina, Chapel Hill, NC. . 1999. "Is Social Capital Declining in the United States? A Multiple Indicator Assess- ment." American Journal of Sociology 105:88


Pollock, Philip H. 1982. "Organizations as Agents of Mobilization: How Does Group Ac- tivity Affect Political Participation?" American Journal of Political Science 26:485-503.

Portes, Alejandro. 1998. "Social Capital: Its Ori- gins and Applications in Modern Sociology." Annual Review of Sociology 22: 1-24.

. 2000. "The Two Meanings of Social Capital." Sociological Forum 15: 1-1 1.

Przeworski, Adam and Henry Teune. 1970. The Logic of Comparative Social Inquiry. Malabar, FL: Krieger.

Putnam, Robert D. 1993. Making Democracy Work: Civic Traditions in Modern Italy.

Princeton, NJ: Princeton University Press.

. 2000. Bowling Alone: The Collapse and Revival of American Community. New York: Simon and Schuster.

Rahn, Wendy M., John Brehm, and Neil Carlson. 1999. "National Elections as Institutions for Generating Social Capital." Pp. 11 1-60 in

Civic Engagement in American Democracy,

edited by T. Skocpol and M. Fiorina. Wash- ington, DC: Brookings Institution.

Rueschemeyer, Dietrich, Evelyne Huber Stephens, and John D. Stephens. 1992. Capitalist Development and Democracy. Chicago, IL: University of Chicago Press.

Sandoval, Salvador A. M. 1998. "Social Move- ments and Democratization: The Case of Bra- zil and the Latin Countries." Pp. 169-201 in From Contention to Democracy, edited by M.

G. Giugni, D. McAdam, and C. Tilly. Lanham, MD: Rowman and Littlefield. Schumpeter, Joseph. 1942. Capitalism, Socialism and Democracy. New York: Harper and Row.

Scott, James C. 1985. Weapons of the Weak: Ev- eryday Forms of Peasant Resistance. New Ha- ven, CT: Yale University Press.

. 1990. Domination and the Arts of Resis- tance. New Haven, CT: Yale University Press. Sen, Amartya. 1999. Development as Freedom.

New York: Knopf. Singleton, Royce, Jr., Bruce C. Straits, Margaret

M. Straits, and Ronald McAllister. 1988. Approaches to Social Research. Oxford, England: Oxford University Press.

Smith, Christian. 1991. The Emergence of Lib- eration Theology: Radical Religion and Social Movement Theory. Chicago, IL: University of Chicago Press.

Snow, David A., Louis A. Zurche, and Sheldon Ekland-Olson. 1980. "Social Networks and Social Movements: A Microstructural Ap- proach to Differential Recruitment." American Sociological Review 46:787-801.

Stolle, Dietland and Thomas Rochon. 1998. "Are All Associations Alike? Member Diversity, Associational Type, and the Creation of Social Capital." American Behavioral Scientist 42:47-65.

Sullivan, Michael J. 1991. Measuring Global Values: The Ranking of 162 Countries. New York: Greenwood.

Tarrow, Sidney. 1998. Power in Movement: So- cial Movements and Contentious Politics. 2d ed. Cambridge, England: Cambridge Univer- sity Press.

Tilly, Charles. 1978. From Mobilization to Revo- lution. Reading, MA: Addison-Wesley. Tocqueville, Alexis de. [1835, 18401 1990. De

mocracy in America. Reprint, New York: Vin- tage.

Tong, Yanqi. 1994. "State, Society, and Political Change in China and Hungary." Comparative Politics 26:333-53.

Truman, David B. 1951. The Governmental Pro- cess: Political Interests and Public Opinion. New York: Knopf.

Union of International Associations. 1990-199 1. Yearbook of International Organizations. New York: K. G. Saur.

United Nations. Various years. Statistical Year- book. New York: United Nations Statistical Office.

Useem, Bert. 1980. "Solitary Model, Breakdown Model, and the Boston Anti-Busing Move- ment." American Sociological Review 45:357


Uslander, Eric M. 1999. "Democracy and Social Capital." Pp. 121-50 in Democracy and Trust, edited by M. E. Warren. Cambridge, England: Cambridge University Press.

Verba, Sidney, Norman H. Nie, and Jae-On Kim. 1978. Participation and Political Equality: A Seven-Nation Comparison. Cambridge, En- gland: Cambridge University Press.

Verba, Sidney, Kay L. Schlozman, Henry E. Brady, and Norman Nie. 1995. American Citi- zen Participation Study, 1990. Ann Arbor, MI: ICPSR.

Warren, Mark E. 1999. "Democratic Theory and Trust." Pp. 310-45 in Democracy and Trust, edited by M. E. Warren. Cambridge, England: Cambridge University Press. . 2001. Democracy and Association. Princeton, NJ: Princeton University Press.

Wolfinger, Raymond and Steven Rosenstone. 1980. Who Votes? New Haven, CT: Yale Uni- versity Press.

World Values Study Group. 1994. World Values Survey, 1981-1984 and 1990-1993. Ann Ar- bor, MI: ICPSR.

  • Recommend Us