Is It Time to Disinvest in Social Capital?
by
Michael W. Foley
Department of Politics
The Catholic University of America
Washington, D.C. 20064
Phone: 202.319.5999 (w)
Fax: 202.319.6267
E-mail: Foley@cua.edu
Bob Edwards
Department of Sociology
East Carolina University
Greenville, NC 27858
Phone: 252.328.4863 (w)
Fax: 252.328.4837
E-mail: EdwardsR@mail.ecu.edu
Paper prepared for presentation at the 94th Annual Meeting of the American Political Science
Association, Boston, MA, September 3-6, 1998. Panel 15-1T (co-sponsored by 37-12): Social
Capital in Comparative Perspective.
Abstract
The paper examines 30 recent articles reporting empirical research employing the concept of
"social capital." We find that the literature is roughly equally divided between those who treat
social capital as an independent variable and those who consider it as a dependent variable, and
between those who operationalize the concept principally in terms of norms, values and attitudes
and those who choose a more social structural operationalization, invoking social networks,
organizations and linkages. Work on social capital as a mainly normative variable is dominated
by political scientists and economists, while sociologists and a wide range of applied social
scientists utilize more social structural understandings of the term. We find little to recommend
in the use of "social capital" to represent the norms, value and attitudes of the civic culture
argument. We present empirical, methodological and theoretical arguments for the irrelevance
of "generalized social trust," in particular, as a significant factor in the health of democracies or
economic development. Social structural interpretations of social capital, on the other hand,
have demonstrated considerable capacity to draw attention to, and illuminate, the many ways in
which social resources are made available to individuals and groups for individual or group
benefit, which we take to be the prime focus and central attraction of the social capital concept.
Their success is owing, in part, to their attention to the context-dependent character of social
capital and its embeddedness in concrete social situations and structures.
Key words: social capital, trust, democracy, economic development, networks.
Is It Time to Disinvest in Social Capital?
In the three years since Robert Putnam's "Bowling Alone" popularized the concept of "social capital," a welter of scholarship has appeared employing the concept, roughly equally divided between reports on empirical research and argumentative pieces directed towards readers of journals of opinion. The empirical research has been dominated by political scientists and sociologists, but a number of economists and other social scientists have taken up the notion in studies that range from renewed attempts to document or deny a "decline in social capital" in the United States to analyses of the impact of varying levels or kinds of social capital on economic performance, public health, or delinquency among teenagers. Other analysts have sought to explain variations in the kinds and value of social capital by examining cross-national differences in the predominance of politicized rather than "civic" groups operating in a society(1)
or by historical changes in the structure of local economies. Despite the evident fruitfulness of this body of work, we continue to have concerns about the scope of the claims made for social capital as an independent variable and about its operationalization.
The most influential formulation of the concept of social capital is that of sociologist James Coleman. Analysts have interpreted Coleman variously, but all claim to be following his lead with operational definitions ranging from specified durable attitudes (trust, civic norms) borne by individuals, to number of voluntary association memberships, to network location or organizational linkages. This conceptual and operational "stone soup" stems from at least two factors in Coleman's own work on social capital. First, though Coleman's discussions of social capital were conceptually consistent and generally clear, their degree of operational specificity befitted a "grand" theorist nearing the close of a long career and not that of an empirical researcher. His treatments were illustrative and suggestive more than explanatory. Second, Coleman explicitly mentions norms and social organization among the key forms of social capital and leaves the precise nature of the relationship between them underspecified and largely implicit in the nature of his illustrations.(2)
As we have argued elsewhere (Edwards and Foley 1997, 1998) the key to understanding what social capital is and how it facilitates individual and collective actions lies in seeing that its "use value" and "liquidity" do not exist independent of specific social contexts.
The context-dependent nature of social capital suggests why neither attitudes (norms, trust), nor infrastructures (networks, organization), per se can be understood as social capital in isolation from the issue of access.(3)
And access, as we have argued elsewhere, is neither brokered equitably nor distributed evenly. The access required to convert social resources (the "raw materials" of social capital), whether norms or infrastructures, into social capital has two distinct, but necessary, components the perception that a specific resource exists and some form of social relationship that brokers individual or group access to those particular social resources. That brokerage can be socially organized at the level of dyads, informal networks (Burt 1997; Heying 1997), voluntary associations (Eastis 1998), religious institutions (Wood 1997), communities (Bebbington 1997; Schulman 1998), cities (Portney and Berry 1997), or national (Minkoff 1997) and transnational (Smith 1998) social movements. Thus specific social contexts in which social capital is embedded not only influence its "use value"; they also shape the means by which access to specific social resources is distributed and managed.
The context dependency of social capital poses conceptual and methodological difficulties for analysts using it to explain the kinds of macro-social, political and economic outcomes of interest to many political scientists. In such analyses a perverse trade-off exists. The more bluntly one measures social capital and the higher the level of social organization characteristic of the process being explained, the more the model must posit that all social capital is of equal value and that all brokering relationships provide equal access. Seeking to explain social capital as a dependent variable, on the other hand, has the advantage that it focuses analytical attention toward, rather than away from, the role of context. It does so by explicitly or implicitly framing a two-fold question related to the production of and access to social resources (which, again, constitute social capital only in conjunction with one another).
For social scientists to speak meaningfully (or usefully) of "social capital," social resources must be both present and accessible. For example, the fact that one neighbor acts in accordance with the norm that adults will watch out for unattended children is of no "use value" to me unless I have access to it. My conscious access to this resource is what converts my neighbor's possession of a norm into an expectation which allows me to plan my own actions accordingly. The social capital in this specific case is manifested as an expectation that facilitates my efforts to take care of my children.
This point merits some elaboration because it clarifies some of the measurement issues
discussed below. Since both resources and access must be present for there to be social capital,
each has served as an indicator of social capital in empirical research. The problem, in our
estimation, occurs when a specific operationalization of either the resource (individual norms,
for example) or a means of access (organizational membership)is implicitly or explicitly treated
as a direct measure of social capital. For example, when an individual who reports that she or he
thinks that most people can be trusted is deemed to both "have" social capital and to provide it
to others, or when the close proximity of individuals and organizations or social networks are
taken to indicate direct access to social capital independent of more nuanced consideration of
either concrete social relations or the ways in which social structures exclude some and include
others.
Recent empirical research
We reviewed thirty recent papers from a variety of disciplines reporting research in which "social capital" played a major role.(4)
In the majority of cases, social capital is treated as an independent variable affecting such outcomes as national-level economic growth (Knack & Keefer, 1997; Fukuyama, 1995), volunteering (Wilson & Musick, 1997), dropping out of school (Teachman, Paasch, & Carver, 1996, 1997), mortality rates (Kawachi, Kennedy, & Lochner, 1997), returns to human capital (Burt, 1997; Friedman & Krackhardt, 1997), fertility (Schoen, Kim, Nathanson, Fields & Astone, 1997), local economic development (Flora, Sharp, Flora & Newlon, 1997; Bebbington, 1997), government-community relations (Brown & Ashman, 1997), and juvenile delinquency (Rubio, 1997).
In other studies, social capital is treated as a dependent or intervening variable, with particular emphasis on the sorts of voluntary organizations said to produce it (Banks, 1997; Minkoff, 1997; Wood, 1997; Portney & Berry, 1997; Stolle & Rochon 1998; Eastis 1998; Smith, 1998), but including work on the roles of schools and youth programs (Youniss, McLellan & Yates, 1997), political context (Booth & Richard 1998a, 1998b), government policy (Nowland-Foreman 1998), economic restructuring (Heying, 1997; Schulman & Anderson, 1997), and even national elections (Rahn & Brehm, 1997) in producing and/or shaping it. One study tests a structural model of social capital which posits tight reciprocal relations among social trust, trust in government and civic engagement (Brehm and Rahn, 1997); another surveys research on social trust and trust in government across nations (and finds little such relation; see below) (Newton, 1998); still another looks at variation over time in "misanthropy" as measured by the GSS social trust items (Smith, 1997).
Even a cursory review of recent empirical literature utilizing the concept of social capital reveals striking differences across disciplines in both conceptualization and operationalization of the term of interest. Among political scientists and the handful of economists who have busied themselves with the concept, social capital refers mainly to attitudes, as measured by survey responses to items on social trust, norms of reciprocity and tolerance, and, occasionally, trust in institutions. Though "social organization" of all sorts is considered a form of social capital in Coleman's work, political scientists and economists tend to see "associational membership" more as a source of social capital than as an indicator of its presence. "Civic engagement," measured as voting, contacting public officials or, occasionally, membership in groups, is most often seen as an outcome of high levels of social capital, though sometimes these variables, too, are used as indicators of social capital. Most of this literature rests on, and sometimes tests, the hypothesis that social capital or associational density have a direct impact on economic performance and the health of democracy, following Putnam's lead (1993).
Sociologists, including applied sociologists working in international development or
community development in the United States, tend by contrast to conceptualize social capital as
primarily a social structural variable, operationalizing it as social networks, organizations or
linkages between individuals and/or organizations. Even where it is conceived partly in
subjective terms (i.e., as norms and values), it may be operationalized as social structures, as
when Flora, Sharp, Flora and Newlon identify their "entrepreneurial social infrastructure" as a
"format for the mobilization of social capital" which effectively creates and directs energies for
rural community economic development (1997). In such studies, comparisons among nations
are difficult if not impossible, because the level at which social capital is presumed to operate is
tied to mainly local social structures.
Social trust and social capital
Because political scientists, in their work on social capital, rely so heavily on survey research, the variable is usually conceived of as something that inheres in individuals (norms and attitudes such as trust). In other words social capital is conceived of as an individual attribute that constitutes a fully portable resource, the value of which does not fluctuate as the individual moves in and out of numerous social contexts. When treated this way social capital generally flies beneath the radar of the kinds of localized social contexts invoked by Coleman. In cases when social capital is measured at the national level by aggregating survey responses into a "grand mean," it cruises at an altitude from which differences among such contexts are indistinguishable. John Brehm and Wendy Rahn's sophisticated effort to rebuild the internal scaffolding of the civic culture argument under the social capital label may be taken as a good illustration of the limitations of this strategy.
Brehm and Rahn's model attempts to demonstrate the "tight reciprocal relationship between levels of civic engagement and interpersonal trust" (1997: 1001). Trust generates participation in the community, and participation generates trust. The two together, moreover, are said to enjoy a similar reciprocal relationship with trust in government, establishing the "virtuous circle" that Putnam found threatened by reduced levels of associational membership and social trust in the United States (1995). Brehm and Rahn's model thus remains wholly at the individual level, hence does not measure "social capital" in Coleman's sense or test its consequences directly. The central argument of Coleman's treatment, that the presence of such attitudes and behaviors will provide resources for others who depend upon them, is never directly engaged. Rather, the model attempts to strengthen our confidence in the social capital thesis by showing that higher levels of interpersonal trust in the individuals who make up a polity will naturally be associated with higher levels of civic engagement and trust in government on the part of those same individuals.
It is hard to carry Brehm and Rahn's findings very far, as they are based solely on U.S. data and avoid systematic comparison over time. In fact, a recent survey of cross-national research on social trust, civic engagement and trust in government by Kenneth Newton shows that Brehm and Rahn's results are idiosyncratic to the United States (Newton, 1999). Not only are levels of social trust, political trust and their variation over time significantly different even among developed Western nations, but there is no reliable correlation between levels of social trust and trust in government across nations. In general, moreover, both social trust and political trust seem to be only weakly related to membership in social organizations. Most important, Newton notes that "patterns of social and political trust vary from one social group to another in different countries in a way which makes it difficult to generalize about trust as a general concept" (1998). While social trust "tends to be high among those who hold a central position," and higher within richer nations than poorer, "political trust is often rather randomly distributed throughout social groups and types" (1998). Political distrust tends to be associated with high partisanship, particularly where one's preferred party is out of office for a long time. In general, as Newton puts it, "Those who are satisfied with life are trusting, and they are satisfied with life because their income, education, status, ...social position" and, we would add, political fortunes "give them good cause to be so" (Newton 1998: PP).
On theoretical grounds, too, there is cause to question political scientists' efforts to resurrect the civic culture argument under the rubric of social capital. As Jackman and Miller point out, the norms and values of the political culture theorists have always been treated as exogenous variables, whereas Coleman explicitly casts social capital as endogenous to particular social structural contexts (Jackman and Miller, 1998). Putnam, they note, traces the origins of the patterns he found in northern and southern Italy back to the late middle ages, and is markedly pessimistic about the chances for creating social capital in the absence of such deeply rooted traditions (1993). More in keeping with the expectations generated by Coleman's argument, Brehm and Rahn show a stronger relationship running from trust in institutions to interpersonal trust than the other way around, suggesting that more trustworthy governmental institutions make for greater social trust in a society (1997; see also Booth & Richard, 1998). More profoundly, however, the sorts of effects posited for "generalized social trust," or even the "civic norms" of some of this research, are likely to be highly mediated by much more "local" social structures (Tarrow 1996).
An example drawn from Coleman's exposition might help make clear what we mean. The presence of norms ensuring that "unattended children will be looked after by adults in the vicinity" constituted social capital for a Jewish acquaintance of his who noted the difference between her old home in the United States and Jerusalem (Coleman 1990, p. 303). Such norms are no doubt borne by individuals. But even their wide distribution may be irrelevant to the peace of mind of parents unless they are active for their children (e.g., Arab children as well as Jewish children, Yemenite and Ethiopian Jews as well as those of obviously European origins) in their neighborhood (East Jerusalem, as well as West; Arab neighborhoods, as well as settler communities). The wider the distribution of such norms, no doubt, the more reason I may have to trust my children to themselves (so to speak); but my knowledge that one individual possessed of such a norm is at my small neighborhood park today may be enough to allow me to permit my child to wander over there and hence may constitute social capital for me.
Even more than the distribution of certain norms, the context which makes it possible (or not) for me to rely upon other people's adherence to them determines their significance as social capital. A given neighborhood may have many individuals predisposed to such norms, moreover, without the neighbors having a sense that theirs is a safe place for children. People's perception that it is so may depend upon such factors as the extent to which people are out on the streets, the external reputation of the area, or a dramatic incident "demonstrating" the trustworthiness of neighbors. Similarly, high levels of "generalized social trust," absent information about who has access to such trust under what conditions, can tell us little about a polity or a community. Context counts, as we have observed before, and counts crucially.
The theoretical difficulties inherent in linking attitudinal data with macro-economic and political outcomes is compounded, in much of the political science literature, by the opaque character of the aggregate data employed. Cross-national research relying on mean scores at the national level on variables such as social trust, civic norms, or trust in government ignore the significance of varying distributions that may lie behind identical statistical profiles. In this respect, work like that of Ronald Inglehart on "political culture" and Putnam's on the "civic culture" in Italy is vulnerable to the same criticism that has been applied in the development literature against the use of per capita GNP as a measure of "development". In both cases, the underlying distributions may reflect wildly varying national or regional patterns.
The elementary statistical point bears reiterating: the same mean may reflect a normal
distribution or a bimodal one; the same mean and standard deviation can capture bi-modal or
relatively flat distributions (see Figure 1). Surely it matters whether support for democratic
values is normally distributed in a population; or represented by a core of activists, organized
labor and peasant groups, but counterbalanced by an anti-democratic upper middle class and
military; or evenly distributed along a continuum ranging from radical democrats to radically
anti-democratic elements. Given the fact that most research on these topics ignores such
possible differences, it is little wonder that Newton found little correlation among the key
variables of Putnam's thesis.(5)
At some lower level the aggregation of individual characteristics could indicate
empirically accessible variations in social context capable of capturing the sorts of expectations
and access to social resources Coleman had in mind, and their impact on people's behavior. To
return to the earlier illustration, if I know that my neighbor will look out for my children at the
park, then his presence is social capital for me. But this is strictly a function of our relationship
and of his availability, and not one that is necessarily generalizable to the whole neighborhood.
If we are going to talk about social capital at the level of the neighborhood, the norm that
neighbors should look after unattended children probably has to be adhered to by some "critical
mass" of residents. More important, that fact would need to be widely known and accepted as a
characteristic of the neighborhood before it could generate the expectations that convert those
norms to social capital. Aggregate scores on such norms and perceptions at the level of the
social network or neighborhood might well describe significant differences among
neighborhoods, though even here we would advocate a closer look at how such differences play
out in concrete cases. Aggregating at this level might be empirically difficult or expensive but
clearly makes sense theoretically. At the level of the city or state or nation, a similar aggregate
score may well be useless, because it may mask extreme and systematic differences between
neighborhoods.
Figure 1. Actual and Hypothetical Distributions
of Political Views Around a Common Mean
Similarly, interpersonal trust is certainly important at the level of the firm, organization or neighborhood. But the sort of aggregate national measures employed in cross-national studies of political culture since Almond and Verba's The Civic Culture mask very real differences within societies between the affluent and the poor, white and black, dominant ethnic groups and the marginalized, and political winners and losers, as Newton's survey of the literature shows. Tom Smith's analysis of the data on "misanthropy," operationalized as negative scores on the "social trust" items of the GSS, also should give pause to those who have adopted Putnam's reading of the evidence. Smith notes that responses have been shown to be sensitive to both wording and context in the overall questionaire, and both shift over the years in which the GSS has been administered. A more careful look at the data shows no clear overall trend, but reveals that misanthropy ("negative social trust") "is shaped by socioeconomic and minority status, noneconomic life events, religion, and age-cohort." It is higher among "the less educated, those with lower incomes, and those with recent financial reversals," "among subgroups toward the social periphery," among victims of crime and those in poor health, among non-church goers and fundamentalists, and among younger adults (1997:172-180 and 191). Clearly, "social trust" itself depends upon a larger social context than that captured by the usual measures of associational membership. By the same token, its "use value" to individuals will vary systematically in ways scarcely considered by macro-analysis of the sort preferred by most of the political scientists in this literature.
We might sum up these observations with the following axiom: The more the "use value" of social capital is shaped by systematic variations between and within groups, the more distorted the connection between the "grand mean" of self-reported attitudes and beliefs among survey respondents and the varying social locations and contexts within which social resources are capitalized and made accessible to people.
Both theoretical and methodological considerations, then, point to the difficulty of sustaining the sorts of claims that have been made about the relationships between social capital, conceived in fundamentally normative terms, and macro-economic or political outcomes. Similarly, recent research demonstrates wide variations in the association between membership in voluntary organizations and the sorts of normative outcomes posited in Putnam's argument (Stolle and Rochon 1998; Eastis 1998; Booth and Richard 1998a). Moreover, work outside the social capital framework on the determinants of political participation in American politics points to a complex interplay of factors including the skills attendant on participation in churches, higher levels of education, and networks of mobilization and recruitment with variable effects for different sorts of participation (Verba, Schlozman, & Brady, 1995; Rosenstone & Hansen, 1993). Finally, comparative research on the bases for economic growth in higher income, industrial nations suggests that, while cooperation is important, it is won more through conflict, the threat of sanction, and institutions than as a result of exogenously generated trust or norms (Kenworthy 1995, 1997).
These observations underline the importance of returning social capital to its original
conception and paying closer attention to the context in which it is said to operate. In the next
section, we take a look at some of the recent sociological literature in an exploration of how
social capital has been deployed as a structural rather than a cultural construct.
Social capital and social structure
Political science research that has examined the relationship between associations and social capital generally identifies social capital with norms and attitudes and looks for effects of associational life (formal and informal) on individual attitudes and behavior. The presumption is that associations facilitate economic growth or democratic performance through their impact on individual norms and attitudes, which in turn have an impact on society through individual behavior (Stolle & Rochon, 1998; Eastis, 1998). Coleman's conception, however, is both more expansive and less indirect. Social capital includes social organization of all sorts, including "appropriable social organization" and "intentional organization" (1990: 311-313). Associations are created for specific purposes but can also be turned to other uses; like social networks, they may be appropriated or intentionally created as "investments in social capital." In either case, they are "social capital" for those who are able to use them, i.e., for those who have access to them. The same may be said of informal relations between people.
The norms and attitudes which may become social capital for an individual, in fact, do so only by virtue of the institutions or social networks in which they are embedded. The presence in my neighborhood of one person -- unconnected to me in any relational or communication network -- who is disposed to look out after unattended children is valuable to me only in the unusual circumstance that my children are in need of help, and he is there to help. His norms are more valuable to me when we have a relationship that allows me to count upon him. When these norms are generalized to our small network of parents, they begin to have still more the character of social capital. When I and others know that we can depend upon him and one another, and where that knowledge generates both expectations and reciprocating behavior that reinforces these norms in him, in ourselves and in others in our neighborhood, then such norms have become a wide social resource.
Trust, at each level of social organization in this case, is not the source of our behavior, but a product of our relationships and of our experience with one another, a point Jackman and Miller make when they insist that trust might most fruitfully be considered endogenous to the particular situation at hand. In Coleman's examples, they note, trust is "cast as endogenous to the structure of the situation..., not as evidence of durable social norms absorbed by individuals" (1998: 54). Or, we might add with a deeper bow to the culturalist presumption, social relationships which generate trust also make it possible for people to make use (or abuse) of the norms and values that others bear.
Moreover, the social capital available in my neighborhood stems not only from the subjective attributes carried around by the individuals who live there, but more profoundly from emergent and existing social infrastructures which facilitate individual and collective actions of many kinds. For example, the PTA, neighborhood churches, the volunteer fire department, local realtors, the area newspaper, and the community policing program might all be employed to reinforce the notion that this is the kind of neighborhood where neighbors look out for one another's children and where unattended children are safe on the streets; these and other organizations may contribute to the contrary perception. As instances of "appropriable social organization" they may be mobilized directly in the service of social goals (good or bad), whether or not they are particularly good at generating attitudes of trust, norms of reciprocity, or civic engagement in their own endeavors. To expand upon our hypothetical example, when a neighborhood child is struck by a drunken driver, his mother, a member of the network of neighborhood parents who gather in the park for picnic dinners every Friday night, mobilizes other parents in the group to form a chapter of Mothers Against Drunk Driving. Her brother enlists the members of the volunteer fire department, up to now better known for misusing public funds for their clubhouse. Other parents join with the police department to lobby the mayor's office to seek a block grant to beef up the town's community policing program. And the mayor, building on the town's growing reputation as an urban "multicultural Mayberry" boosts his own clout with county officials by hosting a waffle breakfast at the local coffee shop for the Democratic candidates for governor and assembly, drawing on the business association, realtors, and the new Crime Watch group for funding and ticket sales.
As these examples suggest, the focus of most political science research on the generation of attitudes and a narrow range of behaviors may miss the real meaning and scope of social organization for individual and group efforts. Social organization, both formal and informal, provides multiple resources to individuals and communities. But sheer "associational density" is not enough, just as the mere existence of other-regarding norms among scattered individuals is not enough to make such norms "social capital". Again, aggregate statistics at the national level hardly capture the sort of dynamic sketched above, which depends upon a social context in which social networks, existing organizations, and enterprising individuals conspire to create a cumulative effect which may be absent in other settings, whatever the average level of membership in associations either locally or across the nation. The difficulty of achieving lasting peace in Belfast and Beirut should warn us against assuming that "associational density" has the same meaning in all social and political contexts.
Social structures must be "appropriable" by individuals and groups to really be "social capital"; their use value as social capital will be multiplied to the extent that they enable multiple linkages across communities and beyond them (Bebbington, 1997; Flora, Sharp, Flora, & Newlon, 1997). Moreover, as the mention of Northern Ireland and Lebanon suggests, not all examples are so happy as the ones we have sketched here. Social networks and institutions may limit members' connections with the wider community; they may include some and exclude others; they may serve selfish and/or anti-social as well as "public" ends (Portes & Landolt, 1996); and they may battle one another furiously over the nature of the "public good".
Sociologists who have treated social capital as an independent variable have been much more attentive to the role of informal social networks and formal social organization in providing both individuals and their communities with vehicles for the advancement of their goals than most of the economists and political scientists. Because sociologists have tended to conceptualize social capital more as a structural variable than an attitudinal one, even their use of survey data to measure the concept has differed markedly from that of most political scientists. In developing an analytical model of volunteering, for example, Wilson and Musick (1997) first carefully distinguish between the human, cultural and social capital revealed in the panel data from the Americans' Changing Lives data set. Their measures of social capital are quite parsimonious -- the number of children living in the household and self-reported informal social interactions; but they are clearly tied theoretically to the extent to which respondents enjoy access to resources embedded in a wide range of social networks. Wilson and Muskick find that number of children in the household and informal interactions are positively related to volunteering. In addition religious ties predict higher levels of volunteering, presumably through providing opportunities and means to do so, as well as through normative processes. Formal volunteering appears to encourage informal helping, but being part of an informal network of family and neighbors helping one another does not necessarily lead to higher levels of volunteering. These results suggest both the complexity of the relations among forms of social interaction and civic engagement and the way in which the social structural dimensions of social capital might be unraveled.
Teachman, Paasch and Carver (1996, 1997) revisit some of Coleman's earlier work to examine the connection between social and human capital and the decision to drop out of school. They analyze a variety of measures of social capital drawn from National Educational Longitudinal Survey data, centering on family structure and parental involvement in the child's development, but including sibling example, school type and frequency of changes of schools. Catholic school attendance is thought to reflect density of social ties between families and the school, while frequency of school transfers reflects the strength of students' social ties to their current school. Children with access to higher levels of social capital in all these forms are significantly less likely to drop out of school.
Other analysts have pursued explicitly network analysis and treated social capital as an intervening variable whose "use value" for career advancement is affected by network structure (Burt 1997) and ethnicity (Friedman and Krackhardt 1997). Burt collected network, performance and background data on a probability sample of 170 men from among the 2,500 who occupied one of three positions just below the rank of vice president in a large U.S. electronics firm. The presence of many "structural holes" in a network and one's position in the network provide varying access to social capital in the form of "information and control benefits" which can be utilized to facilitate career advancement. Burt finds that managers with fewer peers doing the same work and whose positions bridge structural holes in the network are more likely to be promoted faster and receive larger bonuses. Friedman and Krackhardt collected similar network data on five work team in the computer services division of a major bank and found that social capital, operationalized as "advice and feedback centrality" was a key intervening factor in explaining why Asian employees were less successful translating human capital into positive career outcomes than were white employees of European descent.
Like these sociologists, a few political scientists have utilized data beyond the General Social Survey, National Election Studies, and the World Values Survey datasets to uncover social structural components of new models of citizen participation, as we noted above. Verba, Schlozman, and Brady (1995) utilize their own dataset to get at the sources of both the motivation and the capacity of Americans to take part in political life in family, school, and the institutional contexts of their adult lives. They draw attention to the role of churches relatively conservative and egalitarian institutions in developing the skills and motivations relevant for politics. They also point to the significant role of personal connections among friends, family, and acquaintances "often mediated through mutual institutional affiliations" in recruitment for political activities (p 17). Though they do not mention "social capital" in their treatment, the key variables here are "skills" ("human capital" in Coleman's terms) and contacts ("social capital"). Significantly, neither "trust" nor "civic culture" appear in their index. Similarly, Rosenstone and Hansen (1993) put to use a variety of relatively untapped data sources to demonstrate the considerable weight of formal and informal mobilization (above all, through personal contact) in generating political participation.
These studies suggest the potential of a social structural interpretation of social capital for explaining individual variations on a wide variety of outcomes. But social structural operationalizations of social capital are also important to explaining more "meso-level" social, political or economic outcomes. Bebbington (1997), for example, undertakes a comparative case analysis of six communities in the Ecuadorian and Bolivian Andes to explain their varying development trajectories. His structural conceptualization of social capital focuses directly on the presence (or not) in each community of base and federated organizations and of specific individuals a university professor in one case, European development volunteers in another, and priests "with a pragmatic view of development" in the others who linked communities to extra-local institutions and resources. The "organizations first helped create the pre-conditions for intensification, and then helped catalyze processes of intensification" (1997:194). The key individuals played crucial roles in renegotiating existing relationships between state, civic and market spheres, i.e., in establishing linkages between local organizations and extra-local agencies.
Flora, Sharp, Flora and Newlon (1997) surveyed elected or appointed officials in a random sample of 1,099 nonmetropolitan communities to test the relationship between local variations in "entrepreneurial social infrastructure" and its impact on their having undertaken an economic development project. Entrepreneurial social infrastructure conceptualized as a format for mobilizing social capital is comprised of three components: the community's capacity to accept controversy ("legitimacy of alternatives"); mobilize resources from diverse sources; and variations in the structure of community networks. They operationalize the "legitimacy of alternatives" in community deliberations from items asking informants to assess their community's acceptance of controversy, its depersonalization of local politics, and the openness of local governance processes. Mobilization of diverse resources and the development of broad-based networks with permeable boundaries are also measured with multiple community level indicators. They find that an unbiased local newspaper (legitimacy of alternatives), contributions to community projects from several types of financial institutions (resource mobilization), and more extensive network linkages to other communities were significant predictors of economic development activity.
Both of these studies suggest three important lessons for empirical investigations using
social capital as an independent variable. First, they demonstrate the potential of structural
conceptions of social capital to explain meso level social, political and economic outcomes.
Second, both support Coleman's emphasis on the importance of constructed forms of social
organization (e.g., networks, local organizations, linkages) rather than primordial ones (e.g.,
kinship, geographical proximity). As Flora, et al. note, this provides an important and optimistic
contrast to Putnam's account of Italy, where regional differences in social capital are said to date
to the Middle Ages and so appear rather impervious to change. Finally, both studies demonstrate
fruitful and relatively parsimonious models for gathering reliable data on social capital at the
same level of social organization as the political or economic outcome of interest. In so doing
both avoid entirely the methodological difficulties inherent in aggregating individual attitudes or
actions into political culture with effects posited at the level of the polity or region.
Explaining Social Capital
Another group of analysts have taken up the task of explaining social capital as a dependent, rather than independent, variable in empirical analyses. By seeking to explain empirically patterns in the production, presence and usage of social capital, they must, as noted above, wrestle directly or indirectly with the ways specific social contexts shape both the "use value" of social capital and broker access to it. The several studies discussed below point to the importance of context from more proximate organizational contexts (Eastis 1998) to emerging transnational political ones (Smith 1998).
Heying (1997) and Schulman and Anderson (1998) both examine how economic restructuring during the 20th century has affected the production of and access to social capital for the current residents of Kannapolis, NC and Atlanta. Heying's (1995) examination of elite networks in Atlanta from 1931 to 1991 finds that Atlanta's core civic leadership has been dominated by the highest ranking executives from locally owned firms. Through 1961 this core group constituted a cohesive network able to exert effective civic leadership. After 1961 the cohesiveness of this network declined, its capacity to exert leadership eroded, and executives from the corporations currently dominating the local economy became comparatively disengaged from civic affairs. Heying (1997) credits the recent trend of mergers, acquisitions and globalization and a resulting "delocalization" of the economy with causing the civic disengagement of elites. The "social capital" residing in this social network, in other words, declined in direct relation to shifts in the global economy and the corresponding changes in corporate outlook and operating procedures. Schulman and Anderson (1998) examine the firm of Cannon Mills and social relations in its associated textile community (Cannonpolis, N.C.) to demonstrate how the strong "paternalist" social capital rooted in firm-based hierarchical relations has been eroded by economic restructuring and contemporary market relations on the one hand, and partially replaced by a form of worker-based horizontal social capital produced by various union organizing campaigns. The case illustrates the context dependency of social capital and the social processes by which social capital can be both destroyed and reproduced.
Portney and Berry (1997) and Booth and Richard (1998b) make clear that local and national political contexts exert substantial influence on the kind and degree of mobilization of social capital. In a study of neighborhood organization across U.S. cities, Portney and Berry found that neighborhood associations chartered by city governments and given substantial local decision-making powers succeeded where other, less formal organizational structures failed to increase citizen satisfaction with city services and sense of ownership of their community, empowering racial and ethnic minorities along the way. Booth and Richard's analysis of survey data from Central America showed that high levels of repression affected both the kinds of organizations that citizens' tended to belong to and the degree to which democratic norms and attitudes prevailed. Like Stolle and Rochon (1998) and Eastis (1998) -- see discussion below -- Booth and Richard also find that different sorts of groups are associated with different sorts of attitudes, with poor people's organizations (unions and community groups) associated with "leftist ideological extremism, coup justification, and support for such confrontational tactics as civil disobedience, revolution, and political violence," while those representing wealthier constituencies were characterized by higher scores on right-wing extremism, opposition to coups, revolutionary change or reform, and negative attitudes on civil liberties (Booth and Richard, 1998a). These outcomes themselves undoubtedly reflect the highly polarized political situation in several Central American countries at the end of the civil wars of the 1980s.
Minkoff (1995, 1997) and Smith (1997, 1998) both use longitudinal data derived from editions of the Encyclopedia of Associations dating back to the middle 1950s to track the emergence of national and transnational social movement organizations and their production of an array of social resources and means of making them available to dispersed participants independent of face-to-face interactions. Specifically, Minkoff (1997) examines the role of national and transnational social movement organizations in producing social resources and making them available as social capital to wide-ranging constituencies through their organizational infrastructures, communication networks and mediating collective identities. Minkoff shows that national social movement organizations build upon and extend the social resources inherent in community-based institutions and organizations. By developing movement infrastructures and collective identities, national social movements knit together a dispersed membership and facilitate collective and individual participation in the public sphere. Smith extends Minkoff's analysis to demonstrate how the recent growth of international political and economic institutions has facilitated patterns of cooperation among transnationally organized social movements. Furthermore, she argues that transnationally organized social movements produce social capital that is pertinent to the global political arena by providing an infrastructure that facilitates communication and action, cultivates transnational identities and develops public discourse globally. Together, these studies demonstrate that the production of social capital, far from being restricted, as Putnam claims, to arenas where face-to-face interactions predominate, can be found wherever social resources are made available to individuals and groups. Social movements, just like more staid sorts of social organization, provide people with incentives and resources for action, and wide access to both, in local, national and international policy arenas.
Even when analysts focus on the 'neo-Tocquevillian', that is, normative, interpretation of social capital, context proves to be important once we ask under what circumstances what sorts of social capital are produced. Stolle and Rochon (1998) utilize cross-sectional survey data from the United States, Germany and Sweden to examine the relationship between membership in different types of voluntary associations and multiple indicators of what they term "public social capital," i.e., attitudes and behaviors thought to be associated with "civic engagement." They find important differences between type of group and levels of generalized trust, community reciprocity and tolerance among members. For example, memberships in most types of groups were correlated with generalized trust and community reciprocity, but not with tolerance. Members of political associations were both more politically active and less likely to evidence generalized trust or political trust. These findings support Jackman and Miller's conclusion that the traditional civic culture variables do not constitute a particularly coherent complex of cultural traits (1996: 646). In findings broadly similar to those of Tom Smith discussed above, and more congruent with the usual civic culture argument, Stolle and Rochon report that members of internally homogeneous groups were less likely to have high levels of generalized trust and community reciprocity.
Carla Eastis (1998) undertakes an ethnographic analysis of social capital in two choral
groups with some overlapping membership operating in the same city. She found that the
community chorus afforded members greater opportunities to acquire organizational skills and
gain access to social networks that might subsequently prove useful in civic or political settings
beyond the choir than did the more technically demanding, university supported group.
Furthermore, differences in the type of music performed --popular show tunes rather than pre-classical music-- promoted member diversity in the former and homogeneity in the latter thus
(following a "neo-Tocquevillian" logic) instilling norms of reciprocity and trust more effectively
among the members of the more diverse community chorus. While Stolle and Rochon (1998)
emphasize that different types of voluntary associations provide participants with differential
access to social capital, Eastis' work makes clear the analytical risks inherent in treating two
ostensibly similar groups as equals with respect to the social capital (however conceived).
Conclusions
Is it time to divest our stock in the social capital concept? Our answer is a qualified "No." If, on the one hand, social capital is conceived as little more than a stand-in for the old political culture variables, we can see very little use in applying a new label to the traditional stock of terms. Moreover, we judge the use of "generalized social trust" (as measured by one or more of the three GSS items commonly used for these purposes) as the primary focus of attention in political scientists' work on social capital a dead-end. While there is no doubt that trust of some sort is crucial to many social relations, there is little evidence that greater or lesser proportions of a population expressing themselves trustful of "people in general" has any bearing on the health of democracy or the prospects for economic achievement in a given country. On the contrary, such expressions appear to reflect the peculiar social, economic and political positions of the respondents: social trust is the result of a social, economic or political system that works well for some, if not others, not the cause of such felicity. Trust, moreover, is not the universal lubricant that oils the wheels of cooperation wherever it is applied. Rather, cooperation is achieved through a variety of mechanisms, not the least important of which is effective government regulation (Kenworthy, 1995). Where cooperation succeeds, trust may be presumed to follow.
Part of our discomfort with the use of "political culture" variables (norms and values) of all sorts in research inspired by the concept of social capital, is that such research tends to divorce such subjective attributes of individuals from the social context in which (and only in which) they can be understood as social capital. As we have argued above, the mere existence of any such norm "between the ears" of some number of individuals only becomes "social capital" to the extent that others in the community may draw upon those normative dispositions in formulating expectations, making plans and carrying out their activities. The social context including both how such norms, values and attitudes are distributed among a population and how and to whom they are "available" makes all the difference in the world. Hence our dispute with the common use of aggregate statistics at the level of the nation or state to measure something called "political culture." In the dispute between quantitative cross-national researchers and proponents of "small-N" comparative research, accordingly, our critique of recent uses of social capital in political science and economics would seem to put us squarely on the side of small-N comparison. While it is easy to imagine more adequate uses of the rich data in the World Values Survey, for example, than the sorts of comparisons we and others have criticized employing cluster analysis, for instance, to uncover significant "political cultural" cleavages in given societies it is harder to see how meaningful comparison might be achieved using such approaches. Given the likelihood that the link between social trust or any other norm or attitude and civic engagement is highly mediated by social, political and economic context, we find little reason to expect that further efforts to resurrect the civic culture argument under the guise of "social capital" are likely to bear much fruit.(6)
This said, it seems apparent that the notion of social capital provides a useful heuristic for capturing the ways in which social resources are created and made available to individuals and groups (Edwards & Foley, 1998). And properly operationalized, it is more than a heuristic, as we have seen. Recent research has shown how interpersonal relations and institutional context may affect outcomes as diverse as individual exploitation of their own human capital, juvenile delinquency, and the success of communities in attracting resources for economic development. At the same time, other work has shown to what extent economic and political context may shape the level and kind of social capital a community may enjoy.
Political scientists need to think more seriously about the ways in which the elements of social structure facilitate or constrain individual political participation or collective action and look for sources of data which can capture such phenomena. Verba, Schlozman and Brady (1995) and Rosenstone and Hansen (1993) have pointed the way in devising new datasets and creatively exploiting older ones to get at such variables. Mario Diani (1995), Debra Minkoff (1995) and Jackie Smith (1997) have each built extensive datasets focusing on regional, national and transnational organizations respectively, each of which has provided considerable leverage for thinking about how the connections and structures embodied in them work.
Finally, the notion of social capital has already proven useful for political scientists and
others concerned with understanding local processes of resource mobilization, economic
development, and political recruitment and mobilization. We find much that is promising in this
work. At the same time, social capital research in political science and political sociology need
not confine itself to local politics, as the work of Minkoff and Smith cited above demonstrates.
What these efforts have in common is attention to the social structures in which social resources
are embedded and to the kind and degree of access they provide individuals and groups to such
resources. Understood in this way, the concept of social capital provides useful leverage for
uncovering the ways in which individuals, groups and societies generate, broker and put to use
the non-economic resources that are crucial to our maneuvering the diverse social settings of
which the economists' "market-place" is only a minor subset.
References
Banks, Erik. 1997. "The Social Capital of Self-Help Mutual Aid Groups," Social Policy
(Fall):30-38.
Bebbington, Anthony. 1997. "Social Capital and Rural +Intensification: Local Organizations and
Islands of Sustainability in the Rural Andes." The Geographical Journal 163,2:189-97
(July 1997).
Booth, John A. and Patricia Bayer Richard. 1998a. "Civil Society in Central America: The Dark
Side?" American Journal of Political Science, forthcoming.
Booth, John A. and Patricia Bayer Richard. 1998b. "Civil Society and Political Context in
Central America." American Behavioral Scientist 42,1 (Sept. 1998).
Brehm, John and Wendy Rahn. 1997. "Individual-Level Evidence for the Causes and
Consequences of Social Capital," American Journal of Political Science 41,3
(July):999-1023.
Brown, L. David and Darcy Ashman. 1996. "Participation, Social Capital, and Intersectoral
Problem Solving: African and Asian Cases," World Development 24,9:1467-1479.
Burt, Ronald S. 1997. "The Contingent Value of Social Capital." Administrative Science
Quarterly 42:339-365.
Diani, Mario. 1995. Green Networks: A Structural Analysis of the Italian Environmental
Movement. Edinburgh: Edinburgh University Press.
Eastis, Carla. 1998. "Organizational Diversity and the Production of Social Capital: One of
These Groups is Not Like the Other American Behavioral Scientist 42,1 (Sept. 1998).
Edwards, Bob and Michael W. Foley, Eds. 1997. "Social Capital, Civil Society, and
Contemporary Democracy." American Behavioral Scientist, 40,6: 547-678.
Edwards, Bob and Michael W. Foley. 1997. "The Political Economy of Our Discontent."
American Behavioral Scientist 40,6: 669-678.
Edwards, Bob and Michael W. Foley. 1998. "Civil Society and Social Capital Beyond Putnam."
American Behavioral Scientist 42,1: 124-139.
Edwards, Bob and Michael W. Foley, Eds. 1998. "Beyond Tocqueville: Civil Society and Social
Capital in Comparative Perspective." American Behavioral Scientist 42,1: 1-139.
Flora, Jan L., Jeff Sharp, Cornelia Flora, and Bonnie Newlon. 1997. "Entrepreneurial Social
Infrastructure and Locally Initiated Economic Development in the Non-metropolitan
United States," Sociological Quarterly 38,4:623-645.
Foley, Michael W. and Bob Edwards. 1996. "The Paradox of Civil Society." Journal of
Democracy (Summer), 7,3: 38-52.
Foley, Michael W. and Bob Edwards. 1997. "Escape from Politics? Social Theory and the
Social Capital Debate." American Behavioral Scientist 40,6: 550-561.
Friedman, Raymond A. and David Krackhardt. 1997. "Social Capital and Career Mobility: A
Structural Theory of Lower Returns on Education for Asian Employees." Journal of
Applied Behavioral Science 33,3:316-317.
Fukuyama, Francis. 1995. "Social Capital and the Global Economy," Foreign Affairs
September/October: 89-103.
Granato, Jim, Ronald Inglehart, and David Leblang. 1996. "The Effect of Cultural Values on
Economic Development: Theory, Hypotheses, and Some Empirical Tests." American
Journal of Political Science 40,3 (August):607-31.
Heying, Charles. 1997. "Civic Elites and Corporate Delocalization: An Alternative Explanation
for Declining Civic Engagement." American Behavioral Scientist 40,5: 657-668.
Inglehart, Ronald. 1990. Culture Shift in Advanced Industrial Society. Princeton, N.J.:
Princeton University Press.
Inglehart, Ronald. 1997. Modernization and Postmodernization: Cultural, Economic, and
Political Change in 43 Societies. Princeton, NJ: Princeton University Press.
Jackman, Robert W. and Ross A. Miller. 1998. "Social Capital and Politics" Annual Review of
Political Science. 1:47-73.
Jackman, Robert W. and Ross A. Miller. 1996a. "A Renaissance of Political Culture?"
American Journal of Political Science 40,3 (August):632-59.
Jackman, Robert W. and Ross A. Miller. 1996b. "The Poverty of Political Culture," American Journal of Political Science 40,3 (August):697-716
Kawachi, Ichiro, Bruce P. Kennedy, and Kimberly Lochner. 1997. "Long Live Community:
Social Capital as Public Health," The American Prospect, November-December: 56-59.
Kenworthy, Lane. 1995. In Search of National Economic Success: Balancing Competition and
Cooperation. Thousand Oaks, California: Sage.
Kenworthy, Lane. 1997. "Civic Engagement, Social Capital, and Economic Cooperation."
American Behavioral Scientist 40,5: 645-656.
Knack, Stephen and Philip Keefer. 1997. "Does Social Capital Have an Economic Payoff? A
Cross-Country Investigation," Quarterly Journal of Economics November: 1251-1288.
Minkoff, Debra. 1995. Organizing for Equality: The Evolution of Women's and Race-Ethnic
Organizations in America, 1955-1985. New Brunswick, NJ: Rutgers.
Minkoff, Debra. 1997. "Producing Social Capital: National Social Movements and Civil
Society." American Behavioral Scientist 40,5: 606-619.
Newton, K. 1998. "Social and Political Trust." In Critical Citizens: Global Support for
Democratic Government. Ed. By Pippa Norris. Oxford: Oxford University Press,
Forthcoming, 1999).
Nowland-Foreman, Garth. 1998. "Purchase-of-Service Contracting, Voluntary Organizations
and Civil Society: Dissecting the Goose that Lays the Golden Eggs?" American
Behavioral Scientist 42,1 (Sept. 1998).
Portney, Kent and Jeffrey Berry. 1997. "Mobilizing Minority Communities: Social Capital and
Participation in Urban Neighborhoods." American Behavioral Scientist 40,5: 632-644.
Rahn, Wendy M., John Brehm, and Neil Carlson. 1997. "National Elections as Institutions for
Generating Social Capital." Paper prepared for presentation at the 1997 annual
meeting of the American Political Science Association, Washington, D.C., August 30-Sept. 1.
Rosenstone, Steven J. and John Mark Hansen. 1993. Mobilization, Participation and Democracy
in America. New York: MacMillan Publishing Company.
Rubio, Mauricio. 1997. "Perverse Social Capital -- Some Evidence from Colombia," Journal of
Economic Issues 31,3 (September):805-816.
Schoen, Robert, Young J. Kim, Constance A. Nathanson, Jason Fields, and Nan Marie Astone.
1997. "Why Do Americans Want Children?" Population and Development Review
23, 2 (June):333-358.
Schulman, Michael D. and Cynthia Anderson. (Forthcoming, 1998). "The Dark Side of the
Force: A Case Study of Restructuring and Social Capital." Rural Sociology.
Smith, Jackie. (1997). "Characteristics of the Modern Transnational Social Movement Sector.
In Jackie Smith, Charles Chatfield, & Ronald Pagnucco (Eds.) Transnational Social
Movements and World Politics: Solidarity Beyond the State. Syracuse, NY: Syracuse
University Press.
Smith, Jackie. 1998. "Global Civil Society? Transnational Social Movement Organizations and
Social Capital." American Behavioral Scientist 42,1 (Sept. 1998).
Smith, Tom W. 1997. "Factors Relating to Misanthropy in Contemporary American Society,"
Social Science Research 26:170-196.
Stolle, Dietlind and Thomas Rochon. (Forthcoming, 1998). "Are All Associations Alike?
Member Diversity, Associational Type and the Creation of Social Capital." American
Behavioral Scientist 42,1 (Sept. 1998).
Teachman, Jay D. Kathleen Paasch, and Karen Carver. 1996. "Social Capital and Dropping Out
of School Early," Journal of Marriage and the Family 58 (August):773-783.
Teachman, Jay D. Kathleen Paasch, and Karen Carver. 1997. "Social Capital and the Generation
of Human Capital." Social Forces 75,4:1343-59.
Verba, Sidney, Kay Lehman Schlozman, and Henry E. Brady. 1995. Voice and Equality: Civic
Voluntarism in American Politics. Cambridge, Massachusetts: Harvard University
Press.
Wilson, John and Marc Musick. 1997. "Who Cares? Toward an Integrated Theory of Volunteer
Work." American Sociological Review 62(October: 694-713).
Wood, Richard. 1997. "Social Capital and Political Culture: God Meets Politics in the Inner City." American Behavioral Scientist 40,5:595-605.
1. Notes
1.This distinction, which appears frequently in the literature, bespeaks the easy slide into normative concepts occasioned by the adoption, by Robert Putnam and others, of the "civic culture" argument of 1950s political science. Rarely theorized explicitly, its meaningfulness is even more rarely tested. One exception in the literature under review is Knack and Keefer (1997), who explore the differences between "Putnam-esque" and "Olsonian" groups in their effects on economic performance, with the latter being defined as traditional interest groups, while the former include religious and church organizations; education, arts, music, or cultural groups; and youth organizations. This in an attempt to test the conflicting claims that associational participation promotes trust and cooperative habits (Putnam) or harmful rent-seeking behavior (Olson). The authors find no empirical support for either position, as membership in the Olsonian groups appears to have no effect on investment or growth, while membership in Putnam groups appears ("perversely") to harm investment (1997:1273-4).
2. 2.The forms of social capital identified in Coleman's most extended treatment of the subject include "obligations and expectations" (here is where "trust" figures in Coleman's analysis not, it should be noted, as "generalized social trust" but as a feature of the specific context in which specified individuals or classes of individuals can be trusted), "information potential," "norms and effective sanctions" (grouped together because, as Coleman notes, norms are a "powerful, but sometimes fragile, form of social capital"), "authority relations," "appropriable social organization," and "intentional organization" (understood as "direct investment in social capital") (Coleman, 1990:306-313).
3. 3.As Coleman notes, the concept of social capital groups together some social processes and "blurs distinctions between types of social relations, distinctions that are important for other purposes. The value of the concept lies primarily in the fact that it identifies certain aspects of social structure by their function.... The function identified by the concept "social capital" is the value of those aspects of social structure to actors, as resources that can be used by the actors to realize their interests" (Coleman, 1990:305).
4. 4. Our "sample" is more opportunistic than representative, and we rely upon it accordingly for mainly illustrative purposes. It is based on two principal sources, the bibliographical databases Sociofile and Proquest which catalog articles from a variety of sources including Sociological Abstracts, Social Science Abstracts and Political Science Abstracts, and the papers gathered for publication in our two edited issues of American Behavioral Scientist (1997, 1998). We have restricted our focus to what we deemed to be empirical rather than those that we considered to be either strictly theoretical, social commentary, or opinion. Results from the bibliographical database search were further limited by our inability to obtain copies of all the articles of interest in a timely fashion. Our retrieval rate was roughly 75 percent.
5. 5. Jackman and Miller (1996a, 1996b) have found serious methodological faults on other grounds with the analyses employed by Putnam and Inglehart. They note that Putnam's single-factor solution in constructing his indicator of institutional performance in Making Democracy Work does not adequately represent the information in the underlying measures and that reanalysis using the original variables does not support his conclusions. Turning to Inglehart's Culture Shift in Advanced Industrial Society (1990), Jackman and Miller find only weak correlations among the key cluster of cultural variables and they note that each of these variables responds differently to changes in economic conditions. In general, they find that "the six 'components' of political culture do not form a coherent general structure" (1996a, p. 648). In a rejoinder to criticisms, they point out that the one variable that remains significant in Granato, Inglehart and Leblang's reanalysis (1996), McClellands' "need for achievement," is measured in 1990 to predict economic growth from 1960 to 1989 (1996b, p. 700).
6. 6. Nevertheless, we would not want to rule out further cross-national research on political culture. Such work might pay more attention to standard deviations than means on attitudinal variables and might incorporate such data as the Gini coefficient (see Knack & Keefer, 1997) in assessing the importance of economic variables in predicting outcomes. Any such analysis would have to be accompanied by more sensitive, case-oriented approaches, however.