Civic Engagement and Social Cohesion: Measuring Dimensions of Social Capital to Inform Policy (2014)

2

What Should Be Measured?

A range of factors having to do with the social capital characteristics of communities, and, more generally, of society, have been linked to outcomes in population health, economic performance, social functioning, and general well-being. To support analytic work that advances understanding of these linkages, high-quality data are necessary.

Exactly what kinds of data to collect, what methods to use, and who is best positioned to carry out the task, however, are largely unanswered questions. In the first part of this chapter, we consider the definitions that have been offered for key terms that appear in the study charge. In the second part, we consider which of the measureable subcomponents of social capital are most promising in terms of policy relevance, measurement feasibility, descriptive content, and evidence tying them to important social, economic, and health outcomes.

2.1. DEFINITIONS AND KEY MEASUREMENT CONCEPTS

The statement of task to the panel (see Chapter 1) refers to three constructs: social capital, civic engagement, and social cohesion. However, there is little agreement on the definitions of these constructs, which is a major roadblock to quantifying them.

The granular, tangible measures listed in Table 2-1 are possible to track over time and can be combined in ways that are appropriate for addressing various research and policy questions.

This idea—that social capital is not a construct that can be sensibly measured as a formulaic, catchall aggregate of a predetermined set of parts and that a more policy and context specific approach that breaks down the concept into better defined components is needed—has been made by many researchers. Grootaert and Van Bastelaer (2002, p. 5) wrote that a “concept that encompasses too much is at risk of explaining nothing” and that “the challenge for research…is to give meaningful and pragmatic content to the rich notion of social capital in each context and to define and measure suitable indicators.”

Similarly, Stones and Hughes (2002, p. 40) wrote

We agree, but, it is still useful to consider the meaning of the top-level measurement constructs. Ultimately, many fundamental national statistics, such as worker or multifactor productivity, involve separate measurement and aggregation stages.

Civic Engagement

Civic engagement is a cluster of individual efforts and activities oriented toward making “a difference in the civic life of…communities and developing the combination of knowledge, skills, values and motivation to make that difference. It means promoting the quality of life in a community, through both political and non-political processes” (Ehrlich, 2000, p. vi).

Civic engagement may arise in response to problems—a local crime wave, deteriorating schools, ineffective trash collection, or oppressive leadership—whose very existence can be the result of failures of citizens to collaborate on effective solutions, police themselves, or hold public leaders accountable. Civic engagement may also take place habitually (as may sometimes be the case with voting) or because someone is asked to participate (as may sometimes be the case with volunteering) rather than as a reaction to a particular event. The efficacy of engagement is at least partially a function of citizens’ socialization, mastery of civic skills (e.g., running or chairing meetings, organizing petition drives), and knowledge of how to become involved. These skills are often learned in voluntary associations, political campaigns, and religious institutions.

Although political interest and action are primary components of civic engagement, they are not the only ways that citizens become civically active. People engage in a number of ways though their social networks. When friends and acquaintances are recruited to participate, the process is likely faster and more successful when embedded in a base of trust, reciprocity, and a sense of being a stakeholder in outcomes that affect one’s community. Moreover, when the trust in these networks extends beyond friendship circles to include interactions with others (e.g., strangers, non-alike groups), trust and reciprocity are especially valuable in achieving collaborative action. Civic engagement is about much more than voting behavior and volunteerism, though these are certainly key elements.

The United States has a long tradition of rhetoric and action to foster voting, facilitating volunteerism to address community needs, and engaging citizens in various forms of social and political activity. As early as the 1830s, the French observer Alexis de Tocqueville commented on the vitality and significance of voluntary behavior in shaping American democracy. Beginning with the New Deal, there have been periodic federal government initiatives to provide formal opportunities for civic engagement. The Civilian Conservation Corps, launched in 1933, and the Volunteers in Service to America Program (VISTA), initiated in 1965, are examples. The Voting Rights Act of 1965 sought to remove barriers to voting. The 2009 Serve America Act, which reauthorized and expanded the AmeriCorps Program initially established in 1993, is a more recent example. Although these actions were organized at the national level

TABLE 2-1 Broad Categories and Measurable Elements of Social Capital

(though they have strong community-level missions—for example, the Summer of Service and Youth Engagement Zones), there are also many civic programs organized locally by schools, clubs, churches, and other organizations.

Social Cohesion

Social cohesion refers to the extent to which groups and communities cooperate, communicate to foster understanding, participate in activities and organizations, and collaborate to respond to challenges (e.g., a natural disaster or disease outbreak). Because actions and attitudes may integrate people or separate them, research on social cohesion also considers social cleavage between opposing groups that are each cohesive around their positions (e.g., advocates of gun rights versus advocates of gun control).

Civic engagement, as noted above, customarily involves taking action, while social cohesion is more about the conditions that may initiate and facilitate actions or are consequences of them. Though the primary focus is often on groups, the relevant unit of analysis in studies of connectedness and social cohesion—individuals, families, neighborhoods, nations, etc.—depends on the research (or policy) question of interest. It is imperative to identify the level of aggregation. For example, during a civil war, there are high levels of cohesion within factions, such as the Confederacy or the Union, but obviously not for the country as a whole.

Forrest and Kearns (2001, p. 2128) characterized social cohesion as reflecting “the need for a shared sense of morality and common purpose; aspects of social control and social order; the threat to social solidarity of income and wealth inequalities between people, groups and places; the level of social interaction within communities or families; and a sense of belonging to place.” They added (pp. 2128-2129):

Although conceding that there is no single way of defining it, Jensen (1998) identified five dimensions of social cohesion: (1) belonging versus isolation, (2) inclusion versus exclusion, (3) participation versus noninvolvement, (4) recognition versus rejection, and (5) legitimacy versus illegitimacy. Chan et al. (2006, p. 290) defined social cohesion as “a state of affairs concerning both the vertical and the horizontal interactions among members of society as characterized by a set of attitudes and norms that includes trust, sense of belongingness and the willingness to participate and help as well as their behavioral manifestations.” Differing levels of trust within and across groups may play a role in how social ties are formed and in how social cohesion can be fostered, but it can also lead to polarization. Political tolerance and willingness to compromise are other characteristics that affect the social cohesion of groups and populations.

Social Capital

Social capital is a term that has been used to portray many of the elements of civic engagement and social cohesion described above as well as others having to do with the connectedness of people to others. Although the research literature on social capital has produced numerous insights into the functioning of society, it has not produced a scholarly consensus about what the term includes.¹ One of the early scholars to use the term “social capital” was Hanifan (1916, pp. 130-131), who wrote about social cohesion and personal investment in the community:

__________________

¹In a review of 13 articles, Dasgupta and Serageldin (2001) found that 9 of them contained “extended discussion of what social capital means…authors recognize that if they are going to use the term, then they must define how they will use it” (cited in Sobel, 2002, p. 144).

Jacobs (1961) used the term when discussing how neighborliness contributed to more effective functioning of communities. His book, The Death and Life of American Cities, examined how the vitality of neighborhoods depends on social connectedness among its citizens and includes the now often cited example of the Greenwich village delicatessen owner who served as the “eyes of the neighborhood,” even providing a service as custodian of apartment keys for local residents. From there, the literature flourished: Pierre Bordieu (1979) used data from the 1960s and 1970s to examine boundaries between classes in France; James Coleman and colleagues (1982) analyzed how the performance of Catholic schools benefited from a network of social relations characterized by trust; and Robert Putnam (2000) presented hypotheses about why American society was, in his view, unraveling in certain respects at the end of the 20th century.

Putnam (1993, 2000) argued that social capital is built most effectively through encouraging voluntary associations as a way to address social inequality and lack of cohesive social trust associated with ethnic diversity. He expects that increased voluntary associations between people will lead them to transcend differences and “come together” as a cohesive citizenry. As noted in Chapter 1, he introduced two types of social capital: bridging and bonding. Bridging social capital is exemplified by voluntary associations and horizontal ties based on common interests that transcend differences of ethnicity, religion, and socioeconomic status in communities. Bonding social capital refers to social ties that people build around group homogeneity, usually determined along ethnic, racial, or socioeconomic lines (Putnam, 2003).

Putnam considered bridging social capital more essential for the kind of social cohesion that allows minority ethnic groups to integrate beyond their immediate community and into wider society. He found that in diverse, mixed neighborhoods, citizens were overall less trusting of others relative to homogenous communities. This model associated immigration with ethnic diversity, which may result in social fractures of values and obligations in a community. Other studies (e.g., Laurence, 2011) have found that exposure to diversity strengthens some forms of social capital by facilitating the bridging of social gaps between ethnicities and improving perceptions and tolerance toward groups other than one’s own. In all of the above studies, social capital building, through informal or formal

mechanisms, is then posited as a mechanism for alleviating disruption resulting from increased diversity resulting from immigration or other sources. This line of reasoning suggests a number of policy and practical actions: for example, English language and citizenship courses for immigrant groups may be useful for promoting the creation of bridging social capital.

Bourdieu (1986, p. 248), distinguishing between economic capital, cultural capital, and social capital, defined the latter as “the aggregate of the actual or potential resources which are linked to possession of a durable network of more or less institutionalized relationships of mutual acquaintance and recognition.” Emphasizing the connectedness component, he continued: “the volume of social capital possessed by a given agent…depends on the size of the network of connections that he can effectively mobilize” (p. 249). Unlike economic capital, social capital is not depleted by use, but in fact depleted by nonuse.² In this respect, it is similar to human capital. Portes’s (1998, p. 1) critical assessment of social capital research—which he argues too simplistically extends the concept “from an individual asset to a feature of communities and even nations”—is based in no small part on problems created by definitional ambiguity. Paxton (1999, p. 90) noted: “[T]he lack of an obvious link between theory and measurement has, in some cases, led to the use of questionable indicators of social capital. For example, voting should be considered an outcome of social capital rather than a part of social capital itself.”

Francis Fukuyama (2002, p. 27) described social capital as “shared norms or values that promote social cooperation, instantiated in actual social relationships.” He emphasized the role of certain subjective states and attitudes, such as trust, which: “…acts like a lubricant that makes any group or organization run more efficiently” (1999, p. 16). Bowles and Gintis (2002, p.1) stated: “Social capital generally refers to trust, concern for one’s associates, a willingness to live by the norms of one’s community and to punish those who do not.” This relative agreement that trust is an important component of social capital³ is reflected by the trend among statistical agencies and others to include trust questions in surveys—for

__________________

²Arrow (1999) and Solow (1999) also pointed out disconnects in the analogy between physical capital and social capital—missing analogs to rate of return and depreciation; that social capital is mainly a public good and does not belong to any one individual or firm; and that social capital is produced by societal investment but not in as direct a manner as human and physical capital.

³Knack and Keefer (1997) showed that a 1.0 increase in the standard deviation for a measure of country-level trust is associated with economic growth levels greater than 0.5 of a standard deviation. La Porta et al. (1997) found that, across countries, an increase in the standard deviation of 1.0 in the same measure of trust is associated with greater judicial efficiency (0.7 of a standard deviation) and lower government corruption (0.3 of a standard deviation).

example, do members of society trust that their votes count? Do people trust their neighbors so that they are comfortable leaving their houses to go to work? Do they trust that they and their neighbors will be treated equitably by those in authority? (Below we review recent research that has attempted to test the extent to which these kinds of specific questions track with actual levels of trust in experimental contexts.)

Stiglitz et al. (2009) highlighted subjective states and attitudes, defining social capital as “social networks and the associated norms of reciprocity and trustworthiness.” They added (pp. 182-183):

Portes (1998, p. 7) emphasized the capacity of personal and group connections and other support resources to affect “the ability of actors to secure benefits by virtue of their membership in social networks or other social structures.” Lin (2001) emphasized social relationships—investments, connections, and access to resources—associated with expected returns in the marketplace. As discussed below, comparatively strong evidence exists on the association between social connectedness—or, the opposite, social isolation—and health.⁴

There are many candidate indicators for representing the extent and nature of an individual’s connections and networks: examples include memberships in organizations, numbers and diversity of friends, frequency of contact with friends and family, and mode of contact (face to face or virtual and remote). Granovetter (1973) made an important distinction between strong and weak ties. Strong ties are typically thought of as including immediate family and close friends who provide emotional support and often share resources. Weak ties typically extend to a much broader circle of people beyond immediate family and friends and therefore include more diverse connections. In the context of job search, for example, one person may find employment directly through a family connection (going to work in the family firm); another may take advantage of weaker ties to find out about job opportunities through what amounts to an informal employment referral system.

Proliferation of Internet and email use and, more recently, social media has enabled individuals to maintain increasingly large numbers

__________________

⁴For meta-analyses of the links between social relationships and mortality risk, see Berkman and Syme (1979), Cohen (2004), and Holt-Lunstad et al. (2010).

of weak ties (some, such as LinkedIn, are organized around a specific life domain—in this case, career). Focusing primarily on email and using data from the Pew Internet & American Life Survey, Rainie et al. (2006) addressed the question of what impact the Internet is having on Americans’ relationships:

In the context of Putnam’s analysis (1993, 2000), interactions facilitated by technology and social media would seem to have the potential to generate bridging social capital—that is, networking across socially heterogeneous groups. Weak ties facilitated by technology are more likely to include people from different social, ethnic, and occupational backgrounds. This contact with a diverse range of individuals creates access to a variety of knowledge sources and social opportunities, and has been shown to lead to more socially tolerant attitudes and openness to new ideas (Boase and Wellman, 2006).

The rapid pace of change in information exchange and communication technologies are also revolutionizing the ability, effectiveness, and nature of the way in which people take collective action. A decade or more ago, Putnam emphasized face-to-face interaction as being crucial to tapping the benefits of social capital. But, since then, texting, tweeting, Facebook, Instagram, and other tools have come into play not only for basic communication, but also to organize community rallies, group events, and even political actions. It is a research question whether and to what extent the use of new technologies has begun to repair (or added to degradation of) some of the perceived deterioration of connectedness and civic engagement that has taken place over the past few decades.

Recently, research has focused on computer-mediated communication and social ties created by social media—and whether the Internet increases, decreases, or supplements social capital (Wellman et al., 2001). Wellman et al. (2003) investigated the changes that the Internet has had on community life, found that it “is adding on to other forms of communication, rather than replacing them.” They concluded that this has important implications for civic engagement (and, by extension for its measurement). The rapid saturation of social media in communication networks and interest in its impact on personal and social life (Das and Sahoo, 2011) had only added to the relevance of this research area, a

trend documented in a review of that scholarship by Boyd and Ellison (2007).

2.2. INDICATORS FOR MEASURING SOCIAL CAPITAL

As with assessments of the overall economic health of a nation, state, city, or community—which involves measuring such factors as unemployment, inflation, income distribution, and potentially many others—there are measureable pieces of social capital that provide evidence about the social and civic health of a nation, state, city, or community. The importance of a given indicator will vary by place and time and by the questions being asked. Putnam (2000) addressed the structural question, reporting on the collection of data for 14 key indicators in the areas of community or organizational life, engagement in public affairs, community volunteerism, informal sociability, and social trust (see Appendix A). America’s Civic Health Index 2009, produced by the National Conference on Citizenship (NCoC), included 28 indicators organized into 10 areas: connecting to civic and religious groups, trusting other people, connecting to others through family and friends, citizen-centered engagement, giving and volunteering, staying informed, understanding civics and politics, participating in politics, trusting and feeling connected to major institutions, and expressing political views. The Civic Engagement Supplement to the Current Population Survey (CPS) typically includes 15-20 questions that have varied from year to year.

To go from a long list of questions, such as those in the Social Capital Community Benchmark Survey developed by the Saguaro Seminar⁵ (and from which Civic Engagement Supplement questions were originally distilled) to a much smaller set of questions requires prioritization. There are some narrow topics for which one question can be revealing—for example, whether a person voted in the last presidential election. Others—for example, whether a person has adequate social networks to operate effectively in society—require many.

There is no consensus about what an optimal number of indicators might be for the purpose of assessing civic health or about what content is most valuable to nations, states, or cities. What is clear is that multidimensional, multimode data collection efforts facilitate far greater analytic flexibility for researchers than can a single indicator or even information

__________________

⁵The survey embodies a detailed conceptualization of social capital that includes more than 100 items, administered to both a national sample and to representative samples in 41 communities across the United States. The items cover 11 dimensions in the domains of trust, informal networks, formal networks, political involvement, and equality of civic engagement across the community (constructed measure across race, income, and education levels).

from a single module in a national-level survey. As articulated by Paxton (1999, p. 90):

By contrast, multiple indicators allow for a fuller conceptual representation and make it possible to tailor a measure to specific applications. Drawing from Coleman (1988), Bourdieu (1983), and others, Paxton (1999, p. 93) suggested a two-component definition of social capital that distinguishes between more objective and more subjective aspects of resources that inhere in social relationships:

• objective indicators: for example, network structures that link individuals (such as voluntary association memberships), access to resources that can be tapped
• subjective indicators: for example, trust in others, norms of reciprocity (obligations created by exchanges of benefits or favors) obtaining among individuals in a community, extent of positive and negative feelings toward others (for example, levels of morale in a neighborhood)

This two-component classification—while not without its limitations⁶—reflects the traditional division in social theory between quantitative and qualitative dimensions, described by Simmel (1971), and could reasonably be extended to organize the content of civic engagement:

• objective indicators: for example, political engagement (voting, discussing politics, contacting politicians, participating in cam

__________________

⁶In each category, one can marshal counterarguments: to what extent can social isolation really be measured objectively? Why are exchange relationships (reciprocity) less objective? And so on. Simmel’s “form/content” distinction provides an alternative categorization and gets at some of these subtleties. He associated “content” with the purpose or motive behind a social phenomenon or interaction and “form” with the mode of the interaction.

• paigns); volunteering and giving; association memberships; frequency of interaction with neighbors

• subjective indicators: for example, attitudes about efficacy (do individuals believe they can make a difference in the community, help solve problems in the community?); civic values related to citizenship and to living in a community; civic culture

For social cohesion, objective and subjective elements could reflect the capacity of diverse members of a community or cohorts across disparate communities to collaborate on behalf of a shared sense of the greater good:

• objective indicators: for example, diversity of connections, extent and nature of network ties and of voluntary associations; network “embeddedness” of particular organizations; fractionalization—political and otherwise
• subjective indicators: for example, trust within and across groups (who is a citizen?); attitudes toward having people from “non-like” groups as neighbors, family members, or church members

The above distinctions are suggestive of how the broad concepts (social capital, civic engagement, and social cohesion) could be represented in a more granular and more tangible measurement and data collection framework. The content of Table 2-1 is illustrative of data elements that have been used to define or characterize social capital and highlights the heterogeneity of the data used in studies of social capital. That heterogeneity includes the variation in the unit of measurement or analysis, measurement strategies (e.g., survey or nonsurvey) and the distinctions between subjective and objective aspects and among feelings, experiences, and behaviors.

Table 2-1 does not map the universe of social capital—it is admittedly incomplete.⁷ Community engagement, for example, might include activities like participating in a parade or charity run, buying Girl Scout cookies from neighborhood kids, engaging in a community or neighborhood listserv or message board. Any of these activities can happen without membership in an organization. And a survey respondent may be informed about a community without reliance on traditional news. Likewise, it is not clear how the boycott variable in the CPS supplement fits with existing notions of civic engagement, but it clearly representative of the kinds of topics that need further study. Thus, there is a need for broad measures of community engagement.

__________________

⁷Selected taxonomies used in research and in survey modules are in Appendix A.

In order to construct a comprehensive taxonomy or relevant variables, a clear conceptual, definitional, and analytic objective is required, which in turn depends on the research or policy question of interest. One set of questions may be essential on a crime survey, another on a health survey, and yet another on a survey of social mobility among immigrants. For example, trust and interaction among neighbors may affect crime in a neighborhood (and crime may in turn affect trust), while connectedness with one’s children and friends may be more important in explaining differences in health and longevity.⁸

The value in going through a list of candidate data elements organized along different dimensions is not to create a universally comprehensive list (which may not be possible), but to indicate how characteristics of social capital suggest types of analyses and alternative data collection modes. Ideally, as described in the next chapter, an empirical justification for data collection should be established using a case-by-case assessment of the strength of research evidence linking measures to social, economic, health, and political outcomes. However, for the immediate future, some data collection is needed for exploring if and where such linkages exist. It is encouraging that the evidence base shedding light on the relationships between components of social capital and important social outcomes is accelerating. Ever since Putnam (1993, 2000), interest in social capital has expanded rapidly in research and policy communities (see, e.g., Forsman, 2005; Widén-Wulff, 2007).

The broad categories in Table 2-1—political engagement, social cohesion, and trust—are not directly measureable, but they serve to group specific elements—voting, frequency of contact with people, attitudes toward neighbors—that often are. Depending on the context and the questions asked, different elements are linked to the mechanisms that produce change. For example, reducing social isolation or improving trust in a neighborhood may be tools to improve health and reduce crime, or they may be the policy objectives in and of themselves.

One prominent distinction among the variables listed in Table 2-1 is the relevant unit of observation (which may refer to either the unit on which measurements are taken or the unit used in analysis). Some ele-

__________________

⁸Many of these factors have appeared on various surveys including the American National Election Study and the General Social Survey.

ments more naturally emphasize the individual; others focus on groups, ranging from families, to neighborhoods, to communities, to regions, to nations.⁹ While important elements of social capital are possessed by individuals—such as connections that allow people to be more effective and efficacious in the world—many upstream precedents and facilitators of these individual capacities are community characteristics. Examples include the concentration or density of proximate individuals who have social capital and use it to assist others, and the institutions such as schools, churches, clubs, and local markets that facilitate making connections. The presence of individuals possessing social capital and access to facilitating institutions create a positive feedback loop that can reinforce and grow social capital in a community. Additionally, although in some cases the construct of interest is an aggregation of individual properties (e.g., unemployment rate for a state), it may nonetheless reflect effects that take place at other levels, such as for neighborhoods. But other properties of social, political, or economic entities exist only at the specified level; for example, unemployment insurance benefits are a property of a governmental unit within which the individual is located. These aggregations, and the way individual and group level concepts interrelate carry implications for statistical analysis and modeling (e.g., which unit of analysis has what property, how units of analysis are nested within each other).

The literature, at least since Coleman’s landmark works (1988, 1990), has largely portrayed social capital as a community-level attribute, suggesting a need for place-based measurement and a data collection strategy that can provide estimates at the neighborhood, city, and state as well as national level. An increasingly massive and complex challenge for researchers is the fact that “communities” are becoming less and less defined by geography. A person in town A may volunteer in town B, go to church in town C, gives money to national offices of several organizations, and use Skype to talk with family members around the country or the world. Following each item in a survey with questions about where a contact lives or where an activity occurred would continue to exacerbate survey burden problems. This means that other (probably nonsurvey) data approaches will need to be implemented to analyze these complexities (see Chapter 5 for a more detailed discussion of alternative data sources).

Coleman (1988) and Putnam (1995) both conceptualized social capital as a property of groups rather than of individuals. Along these lines, the

__________________

⁹Indeed, something like the Gini index of income inequality is by definition distributional and does not describe any individual. But individual measures of income are the micro-level unit used to get to the structural indicator. The literature sometimes makes distinctions between compositional, structural, and global indicators (e.g., mean income, inequality, and proximity to wealth as respective examples).

former argued that, “unlike other forms of capital, social capital inheres in the structure of relations between actors and among actors. It is not lodged either in the actors themselves or in physical implements of production” (p. 98). The basic argument that social capital cannot exist without social relationships between at least two people is sensible. Individual members of a group can benefit from the social relations of others. A child, for example, may benefit from such “spillover effects” if his or her parents are socially well-connected with others who possess high levels of social capital characteristics such as trustworthiness and strong networks. And some group phenomena that interact with dimensions of social capital—for example, inequality—clearly take place at aggregations above the family or the community. Concerns about the top one percent or, at the other end of the ideological spectrum, about overemphasizing class conflict pertain to a loss of social cohesion that is not a local phenomenon.

Going the other direction on the spatial dimension scale, Coleman (1988) made explicit links between an individual’s or a family’s human capital and social capital. And Glaeser et al. (2002) analyzed the formation of social capital using a model of optimal individual investment decisions. In this economic approach, the emphasis is shifted from “institutions, norms, conventions, social preferences, and aggregate/group outcomes” to the individual’s “social characteristics—including social skills, charisma, and the size of his Rolodex—which enables him to reap market and non-market returns from interactions with others” (p. 438). Likewise, Portes (1998) emphasized the individualist perspective. He noted the logical danger of models based on aggregate-level characteristics, such as crime rates, which could be interpreted simultaneously as affecting levels of social capital or as an outcome resulting from it. Portes illustrated this problem by observing that an indicator of social capital, such as the average number of neighbors known, would be much stronger for making causal claims if it could observed longitudinally both before and after a change in the crime rate.

The relevant unit of observation can also be suggestive of how to collect data appropriate to the analytic goals. Information about many actions or attitudes is collected through surveys of individuals, after which indicators of interest may be aggregated to various geographic levels. Surveys ask respondents if they voted or if they trust their neighbors, yet the ultimate interest may be in national level voter turnout trends or community levels of trust. If all one is interested in is total voter turnout, newspaper circulation in a media market, or total membership in associations, there are administrative and other kinds of data sources. But if one is interested in the attributes of individuals engaged in various behaviors or with specific attitudes, microdata obtained from individual respondents are essential. As the field moves forward, it is likely that

nonsurvey, digital data will increasingly be combined to link records and build profiles of individuals (discussed in detail below).

Another level of differentiation shown in Table 2-1 is among behavior, feelings, and experience:

• Peoples’ behaviors or actions, which are frequently quantifiable. Examples include participating in specific activities (political or nonpolitical, organized or nonorganized); interacting with family; and volunteering time and contributing money.
• Peoples’ feelings, perceptions and attitudes, which often involve subjective assessments. Examples are trust in others and in institutions and support and sense of belonging or not belonging.
• People’s experiences, which are generally measurable. They include such elements as social, geographic, or economic mobility; discrimination; and political polarization.

Data across these categories are typically gathered at the individual level, but the question of interest often involves reference groups: for example, what is the role of family support for health of elderly people or the education of children? What is the level of trust in government among Republicans in comparison with that of Democrats? What is the level of income mobility among immigrants originating from one country compared with those from another? This is just one dimension in what might be thought of as “nested” indicators, and data on these can be aggregated to create reference levels of engagement and cohesion at household, neighborhood, municipal, state, or national scales. These “nature of phenomena” distinctions do not by themselves establish a clear demarcation of what to cover and what not to cover, but they are important considerations in developing a data collection strategy.

Clearly defined activities or behaviors such as voting or volunteering can often be reported in a comparatively straightforward way with a few questions on a population survey.¹⁰ Data on other observable actions, such as interacting using social media or donating money to charity, which can be asked about on surveys, may be obtainable using nonsurvey

__________________

¹⁰This distinction can be overdrawn. Sometimes “objective” phenomena are also difficult to measure. For instance it may seem that volunteering is easily measured by the single CPS question, “Since September 1st of last year, have [you/NAME] done any volunteer activities through or for an organization?” But asking people to remember what they did over an entire year can be fraught with error, not to mention that people may have dramatically different understandings of what volunteer activities mean. See Turner and Martin (1984) for an excellent treatment of this and other methodological and measurement issues related to surveying subjective phenomena.

sources.¹¹ Other dimensions, such as self-assessments of trust in others or of loneliness require a subjective assessment of feelings or attitudes and may only be measurable by asking questions directly to people in the population of interest. However, word correlation mining tools applied to social media data or from records of communication patterns (e.g., telephone, texting) are now used as evidence even about these phenomena.¹² The frequency of such activities as interacting with friends and family, and even of political discussion, can be scraped from Twitter and other online forums, but knowing the relationship between discussants online would often be more ambiguous than would be possible with surveys.

Like feelings and emotions, the “experience” variables (mobility, discrimination) are also complex and difficult to measure. Experience variables often serve as contextual data in studies—things that need to be looked at alongside the central inquiry. For example, neighborhood crime, discrimination, social mobility, or changing family structures could all factor into levels of reported trust, and trust or lack thereof may in turn have an impact on these conditions. The last three categories in the table—fairness, political polarization, and social integration—are examples of characteristics of the social environment that relate to social capital but that are major topics in their own right, each with deep research literatures. Certainly, this is the case for research on the causes and effects of political polarization (see, e.g., Glaeser and Ward 2006; McCarty et al. 2006; Prior, 2013). The “fairness” variables relate to the social, legal, and economic environment, but they are not often identified as social capital, though they may be reflective of the elements that are.

Similarly, intergenerational mobility may not be considered an element of social capital, yet a lack of it may in turn undermine trust and community cooperation and cohesion. Studies of the transmission of social capital (e.g., Borjas, 1991) have shown that social ties developed by parents have a significant impact on children’s economic and social mobility. Weiss (2012, p. 212) uses data from the National Longitudinal Study of Adolescent Health on this point:

__________________

¹¹In these cases, webscraping and administrative data from the Internal Revenue Service would be principal options.

¹²For example, happiness indexes have been compiled by tracking individual positive and negative words used in Twitter tweets or Facebook posts, which show day-to-day variation along with factors underlying happiness (Christmas) or sadness (a mass shooting): see http://www.hedonometer.org/index.html [February 2014].

Economic and social mobility and the other social environment variables listed in Table 2-1 are important (and related) measurement topics; for some analyses, they are key covariates and possibly a reflection of civic engagement and social cohesion. However, they are beyond the scope of this report. We leave it to others to decide whether or not, for example, the CPS should include a question on parents’ education or occupation or whether it is better asked on other surveys or instruments. However, we note that our recommendations on the timing and frequency of the CPS supplement questions (see Chapter 5) have an impact on whether space may be available for such additions.

As with mobility, income inequality—while not typically treated in the research literature as a dimension of social capital per se—is a particularly closely related issue (it is also one that is too expansive to deal adequately in this report; fortunately the topic has been the focus of a number of careful studies). Kawachi et al. (1997) is the most frequently cited paper on the relationship between social capital, income inequality, and health. While the authors did not establish a causal linkage between income inequality and “increased mortality via disinvestment in social capital” (p. 1491)—they do lend support to the hypothesis by demonstrating that income inequality is correlated with social capital¹³ and, in turn, that social trust and group membership are associated with total mortality and deaths attributable to coronary heart disease, malignant neoplasms, and infant mortality. An interesting aspect of Kawachi et al. is that their hypothesized linkages posit social capital both as effect (i.e., higher levels of inequality reduce social capital) and as a cause (i.e., affecting health and mortality). Along these lines, Kennedy et al. (1998) and Lin (2001) explored the interactive relationship between income inequality and such phenomena as social capital formation, firearm violence, and health.

A number of the themes from Kawachi et al. were picked up and given high visibility by Wilkinson and Pickett (2009), who explored the effects of inequality on mental and physical health and educational outcomes. Using data from a wide range of sources (including the United Nations, the World Bank, the World Health Organization and the U.S.

__________________

¹³The indictors of social capital used in the paper, derived from the General Social Survey, are social trust (measured by responses to the question “Do you think most people would try to take advantage of you if they got the chance, or would they try to be fair?”; and level of agreement with the statements “You can’t be too careful in dealing with people” and “People mostly look out for themselves”). Per capita density of group membership (measured by the per capita number of groups and associations to which residents in each state belonged) was also included in the analysis.

census), they presented the case that, among developed countries, societies characterized by more equal income distribution tended to be happier and healthier than those with greater income disparities. Deaton (2001) explored the connection between income inequality and health in both poor and rich countries. He cites work by both Wilkinson and Kawachi to acknowledge the possibility that “equal societies have more social cohesion, more solidarity, and less stress, they offer their citizens more social support and more social capital, and they satisfy humans’ evolved preference for fairness” (p. 113). Elsewhere, Bartels’ (2005) examination of economic inequality and political representation called into question whether Robert Dahl’s (1971, p. 1) observation that “a key characteristic of a democracy is the continued responsiveness of the government to the preferences of its citizens, considered as political equals” still applies in the United States. Examining voting records on divisive issues (such as abortion and the minimum wage), Bartels found that legislators’ votes do not equally reflect the views of people in the bottom, middle, and top thirds of the income distribution. Specifically, “senators appear to be considerably more responsive to the opinions of affluent constituents than to the opinions of middle-class constituents, while the opinions of constituents in the bottom third of the income distribution have no apparent statistical effect on their senators’ roll call votes” (p. 1). The research cited above, and many other studies, examined how inequality interacts with civic engagement, social cohesion, and other dimensions of social capital: this work is suggestive of the kinds of data needed to advance understanding of these relationships.

A wide range of additional factors—beyond income inequality, including access to education opportunities, outdoor space, clean water and air—also can be linked at some level with trends and variation in social capital. Even phenomena such as home ownership—postulated to reduce geographic mobility and incentivize investment in neighborhood-specific social capital (DiPasquale and Glaeser, 1999)—create channels whereby the health of neighborhoods and society at large are affected. Depending on the type of analysis, some of these background or environmental factors may themselves be outcome measures—related to crime and safety, effectiveness of government and other institutions, community resilience and efficacy—which suggests a strong feedback loop between the outcomes associated with social capital and the individual measurable pieces of it.¹⁴

One could continue to make reasonable links into domains even fur-

__________________

¹⁴For discussions of embedded unfairness, links to arrests and opportunity, and similar factors and on the link between income inequality and health and social problems, see Stiglitz (2013).

ther afield from the core areas of social capital to include such interacting factors as education, immigration, technology, the rapidly changing global economic structure, and—even very reasonably—to climate change. For example, on the latter, Sampson (2013, p. 1) has noted:

These kinds of characteristics are fascinating and no doubt should be measured because they affect social, economic, and health outcomes in profound ways. However, it is not possible to consider all possible factors that are connected with the concept of social capital in one report; happily, these are ongoing major areas of investigation.

This page intentionally left blank.