The Resource Generator: social capital quantification with concrete items

Abstract

In research on the social capital of individuals, there has been little standardisation of measurement instruments, and more emphasis on measuring social relationships than on social resources. In this paper, we propose two innovations. First, a new measurement method: the Resource Generator; an instrument with concretely worded items covering ‘general’ social capital in a population, which combines advantages of earlier techniques. Construction, use, and first empirical findings are discussed for a representative sample ($N=1004$) of the Dutch population in 1999–2000. Second, we propose to investigate social capital by latent trait analysis, and we identify four separately accessed portions of social capital: prestige and education related social capital, political and financial skills social capital, personal skills social capital, and personal support social capital. This underlines that social capital measurement needs multiple measures, and cannot be reduced to one total measure of indirectly ‘owned’ resources. Constructing a theory-based Resource Generator is a challenge for different contexts of use, but also retrieves meaningful information when investigating the productivity and goal specificity of social capital.

Introduction

Many scholars have come to a definition of individual social capital that regards it as the collection of resources owned by the members of an individual’s personal social network, which may become available to the individual as a result of the history of these relationships (Van Der Gaag and Snijders, 2004). However, one of the problems that has been hampering research and theory development in the field of individual social capital research is the lack of comparable measurements (see Flap, 1999, Lin, 2001a, Lin, 2001b). Many measures seem to have been constructed from data that happened to be available, but were not specifically designed for the purpose of making comparisons between populations or sociodemographic subgroups. Furthermore, only some dimensions of social capital have been measured thoroughly. Much emphasis has been put on social networks and their sizes, but much less on the resources that could be accessed through the network ties, and how these may become available to the individual (Flap, 1999). Finally, measures mostly have been designed for a specific life domain only, and not with an aim to investigate the ‘general’ social capital of a general population. In summary, the information we have on the distribution and productivity of social capital is quite fragmented (Van Der Gaag and Snijders, 2004).

Progress in the field of social capital requires valid, reliable, and preferably parsimonious measurement instruments that can be applied in the investigation of three main issues. First, to give a good overview of the distribution of social capital over the general population, which is as yet still lacking (Flap, 1999). Second, for use in—preferably prospective—studies of the basic idea behind social capital, namely its productivity: how it helps individuals to attain their goals in addition to personal resource collections. Third, to investigate to what extent social capital is goal- and context-specific in the production of individual returns (see Flap, 1999, Lin, 2001a): which part of the social capital is responsible for which effects, and under which conditions?

By trying to capture social capital in a single measure we may lose a lot of information, and make it nearly impossible to investigate its goal specificity (Van Der Gaag and Snijders, 2004). In this paper, we will develop multiple individual social capital measures, each referring to separate parts of social capital, for use in cross-sectional, prospective research. Within this perspective, we propose a new measurement instrument, and a new method of analysing questionnaire items that can lead to the construction of such measures.

When we wish to develop social capital measures that could fulfill the tasks mentioned above, several questions are encountered. First, a decision should be made on what we mean by ‘social capital’. Lin, 2001a, Lin, 2001b made a distinction between the access and the use of social capital: ‘access’ to social capital refers to an individual’s collection of potentially mobilisable social resources; the ‘use’ of social capital refers to actions, and mobilisation of the resources in order to create returns. To develop cross-sectional, ‘yardstick’-like measures of social capital for prospective research, it is more useful and more straightforward to focus on measuring the potential ‘access’ of individuals to social capital. The ‘use’-perspective offers good opportunities for the retrospective study of social capital mobilisation and effectivity in specific contexts, but in prospective application involves many additional phenomena that influence measurement, such as personal preferences, the personal need for help, and the availability of institutional solutions to goal attainment (Van Der Gaag and Snijders, 2004). Here, we concentrate on measuring social capital within the ‘access’ perspective, and define social capital as the collection of all potentially available network members’ resources. How to quantify this is treated later on.

A second measurement development question refers to the composition of social capital. If we wish to measure the access to ‘general’ social capital in a population, we must first establish which life domains are potentially important for goal attainment, and which resources should be measured within these domains. In this paper, we use the term ‘general’ social capital referring to social resources in a wide set of life domains that covers the needs of an ‘average person in modern, industrial society’. This comprises a potentially enormous, varied collection of possibly useful resources: access to advice, love, practical assistance, attention, influence, physical strength, knowledge, expertise, status, money, food, health care, etc. Therefore, the construction of measures for ‘general’ social capital should begin with clear theoretical classifications; we return to this issue in a later section.

A third issue is that the social capital available to individuals is not only a function of alters who own various kinds of resources—but also of these alters’ willingness to give access to their resources (Flap, 1999). If we assume that every measured social resource is equally available, this could lead to overestimation of social capital. Therefore, indicators for the availability of resources should be included in social capital measurements.

The measurement of social capital with a focus on individuals’ ‘access’ to social capital, considering a diversity of measured resources, and including resource availability indicators, has as yet been pursued following two methodological paths. The oldest method is the ‘Name Generator/Interpreter’ approach (McCallister and Fischer, 1978). This method maps the ego-centered social network as a starting point for a subsequent social resource inventory, which—dependent on the inclusion of name interpretation questions—can result in very detailed and informative social capital descriptions. The single ‘core’-network identifying Name Generator “With whom do you talk about personal matters?” stems from this approach, and has been widely used ever since (e.g. in the American General Social Survey; see Marsden, 1987). Nevertheless, as a social capital measurement the Name Generator method can be considered unsatisfactory. Most important is that the collection of such data is a heavy burden on both interviewer and interviewee; especially, when larger networks are found. Furthermore, because of differences in focus, the grounds for inclusion of name generating and name interpreting social resource questions have led to many different studies with incomparable findings (Lin, 2001b, p. 16). Third, much of the data collected with the Name Generator/Interpreter is theoretically redundant for the expression into social capital measures: many alters will give access to the same resources, and although similar resources available from several alters could be seen as a form of help ‘insurance’, usually one alter suffices to solve a certain problem. For cases where multiple alters are useful in providing resources, diminishing marginal returns can be expected from additional alters. It is therefore more critical to assess whether at least one alter is available to provide some given form of resources, than the total amount of alters doing so Snijders, 1999, Van Der Gaag and Snijders, 2004. Fourth, there has been no consistency in the way Name Generator data have been aggregated into social capital measures. Various indicators have been designed as network size and network range indicators (see, e.g. Campbell and Lee, 1991), but these have not led to standardised measures for social capital. Finally, most of these measures have referred to (structures of) social relationships only, and not to the resources that may become available through them, which makes them doubtful as indicators of access to resources.

A second measurement instrument that has been used to collect access-type social capital data is the ‘Position Generator’ Lin and Dumin, 1986, Lin et al., 2001; this method measures access through network members to occupations, seen as representing social resource collections based on job prestige in an hierarchically modelled society, following Lin’s theories of social capital Lin, 1982, Lin, 2001a. The administration of this instrument is easy and economical, and the questionnaire can be systematically adjusted for different populations. Its data is also straightforwardly modelled into social capital measures that have a clear theoretical basis (range of accessed prestige, highest accessed prestige, and number of different positions accessed). However, these measures also have their disadvantages. They contain little specific information about social resources and the diversity of this collection. Also, their interpretation hinges on the theoretical importance of job prestige or other position-related dimensions, which may not be dominant for all social capital issues. For the investigation of the goal- and context-specificity of social capital, multiple measures are needed that each refer to separate portion of accessible social resources (Van Der Gaag and Snijders, 2004); for this purpose, Position Generator measures have limited use.

To overcome these disadvantages, Snijders (1999) proposed to combine the positive aspects of the Position Generator (economy, internal validity) and Name Generator/Interpreter (detailed resource information) by more clear referral to specific resources, and omitting name identification from Name Generator questions. The resulting instrument, the ‘Resource Generator’, asks about access to a fixed list of resources, each representing a vivid, concrete subcollection of social capital, together covering several domains of life. It has the same basic questionnaire structure as the Position Generator: the availability of each of these resources is checked by measuring the tie strength through which the resources are accessed, indicated by the role of these ties (family members, friends, or acquaintances). This instrument can be administered quickly, and can result in valid and easily interpretable representations of social capital, with possibilities for use in goal specificity research of social capital.

Incomparability problems can occur with this measurement instrument also, because the list of specific resource items to be included may vary over populations. The composition of the Resource Generator should therefore result from systematical, theoretical considerations about which social resources represent the ‘general’ social capital of individuals. Several theoretical classifications can be considered useful.

At a very basic level, we can argue that social capital measurements should refer to all different personal resource collections of network members that are generally distinguished within sociology: human, cultural, financial, political, and physical capital. More in accordance with social resources and social capital theory, we can argue that the universally valued resources power, wealth, and status should be referred to Lin, 1982, Lin, 2001a, Lin, 2001b. Some more concrete guidance is offered by social production function theory (SPF) Lindenberg, 1986, Ormel et al., 1997, that orders goals universally pursued by individuals. An empirical reconstruction of SPF for the contemporary Netherlands showed that individuals generally distinguish six cognitive domains in goal attainment: (1) private productive activities, (2) personal relationships, (3) private discretional or recreational activities, (4) public productive activities, (5) public relationships, and (6) public non-institutionalised interactions, involving everyday contacts with unknown individuals (Van Bruggen, 2001). The last of these domains does not refer to individual social capital, because by definition there is no shared past with unknown people.¹ Together, the other five domains can be used to inspire measurement items that represent potentially productive social resources. On the basis of considerations of personal resource collections, universally valued resources, and domains in individual goals, a set of Resource Generator items was constructed that comes close to measuring ‘general’ social capital (see Section 2).

Once answers to a list of questionnaire items on social capital are available, a next question is how to aggregate these into a measure that indicates access to social capital. Earlier researchers have suggested several principles to construct measures. First, an emphasis on volume, suggesting simply that access to bigger, larger, or more social resources is beneficial Bourdieu, 1980, Flap and Boxman, 2001, Burt, 1992; this could be expressed as a measure of the total of social resources present in the network. Second, diversity, indicating that the more differentiation is present in social resources, the better social capital it represents Lin and Dumin, 1986, Erickson, 1996, Lin, 2001a. Third, a high upward reach in social resources, indicated by hierarchical evaluation of accessed resources Lin and Dumin, 1986, Lin, 2001a—this principle, implying a beneficial effect of the best resources available, can only be applied to data that include some ordinal characteristic (such as job prestige). The most straightforward way to operationalise social capital is to calculate one single volume or diversity measure, counted as the total number of different items that is accessed. However, such a measure leaves a lot of interesting information unused, because it will yield the same numerical values for very different collections of social capital.

To compose multiple measures for social capital, we need an argued basis to aggregate information; which subcollections of items should lead to separate measures? One method is to start from a theoretical basis, and group items by the effects they could have within a certain life domain: social resources that are additive in helping to attain the same goal (Snijders, 1999). In this way, we could, for example, construct a measure for each of the domains distinguished by Van Bruggen (2001) mentioned earlier. However, the knowledge we have on the productivity and goal specificity of social capital is currently too fragmented and incomplete for this purpose. Therefore, we group items not based on their effects on the attainment of specific goals, but based on their correlational structure on a population level (Snijders, 1999). To explain how such empirically independent social capital domains can be distinguished, we must reconsider the basis of social capital creation: the relationship.

In explaining relationship formation and maintenance, three determinants are generally discussed. First, an ‘opportunity structure’ is needed to get into contact with persons and keep the contact going, defined by, e.g. locations of the home and the work place, and other people who figure in these surroundings (Van De Bunt, 1999). Second, the choice of others within this opportunity structure. An important explanation here is homophily: investment in relationships with persons who are similar with respect to demography, education, and lifestyle (Homans, 1950, Lazarsfeld and Merton, 1954; Lin, 2001a, pp. 38–40; review in McPherson et al., 2001). In relation to social capital, we could argue more specifically that relationships are formed with those others from whom greater returns are expected, who may or may not be similar to ego (Flap, personal communication). Third, personality characteristics have recently begun to be considered as determinants in relationship formation, suggesting that some of the generally distinguished components of personality—the ‘Big Five’ extraversion, agreeableness, conscientiousness, emotional stability, and intellect (Digman, 1990)—have considerable impact on personal network formation (e.g. Vodosek, 2003, Negrón and McCarthy, 2003). In addition, relationship formation is constrained by time and resource budgets. Social capital is created and maintained given these and perhaps other determinants and constraints, and may result from deliberate, goal-oriented investments in relationships, and as a by-product from on-going activities and relationships.

For each individual this process results in access to a unique, personal collection of social capital. Because not all individuals will access the same subcollections of social capital, observation on the population level of access patterns may lead to the distinction of meaningful social capital domains. Positive correlations between resource items in some group of items indicate that individuals who access one of these items also have a higher probability of accessing other items from that group. Such a group of items can thus be considered to represent a social capital domain, in which no specialisation takes place in terms of concentrating on some of the resources at the expense of others. Items from each group can therefore be aggregated into a domain-specific social capital measure. Thus, identified domains for social capital are population-specific, and we expect that for most populations there are several of these roughly independent, empirically distinct domains of social capital.