JCMC 8 (3) April 2003
Message Board
 |
NetStudy VEs VOs O-Journ HigherEd Conversation Cyberspace Web Commerce
Vol. 6 No. 1 Vol. 6 No. 2 Vol. 6 No. 3 Vol. 6 No. 4 Vol. 7 No. 1
Vol. 7 No. 2 Vol. 7 No. 3 Vol. 7 No. 4 Vol. 8 No. 1 Vol. 8 No. 2 Vol. 8 No. 3
Community Networking and Social Capital: Early Investigations
Christina Prell
Rensselaer Polytechnic Institute
- Abstract
- Introduction
- Community Networking and Social Capital
- Social Network Analysis and Social Capital
- Method
- Results
- Discussion
- Conclusions
- Footnotes
- References
- About the Author
Abstract
This paper draws upon an ongoing study pertaining to the early development of one component of a community network in the city of Troy, New York, USA. The component under study is that of a database to be distributed via a community network. Community networking literature posits a relationship between social capital and community networking, stating that community networking should positively affect levels of social capital in a community. This article begins exploring this relationship through reviewing the social capital concept as presented in the literature. Measures are developed from the field of social network analysis and applied to a group of community members involved in this database project. Results show high levels of in-degree centrality correlating with trustworthiness and resource exchange, and betweeness centrality correlating with trustworthiness. Although in-degree centrality proves to be the more useful measure for purposes of studying community networking and social capital, discussion is given to the surprising results found for betweeness centrality.
Introduction
Community networking refers to the use of computer-networking technology for purposes of strengthening a community. These networks are seen as local, community initiatives where owners and developers are community members, and the networks' content reflect community needs and interests. In keeping with this community focus, community networks are also perceived as tools for helping build and sustain democratic, civic cultures. In particular, through reflecting community needs and providing computer-based channels of communication for community members, community networks act as tools for promoting a participatory, civic culture. Thus, ideals of democracy and community act as both guiding principles for building community networks as well as hoped-for outcomes of a network's use.The present paper explores this relationship between community networks, community, and democracy by focusing on the development of one component of a community network: a youth services database. This database is being designed and developed by faculty members and students at Rensselaer Polytechnic Institute, in Troy, New York, and members of the Troy and Rensselaer County community. These participating community members consist of representatives of local not-for-profit youth agencies and employees from city and county government. In studying this particular group of participants and their involvement with the database project, questions regarding how these participants relate to and collaborate with one another are being explored. These questions emerge from the theoretical framework of social capital, a concept circulating in the community networking literature. In the following sections of this paper, an overview of social capital is given along with a review of the literature linking social capital to community networking. Before launching a discussion of social capital and community networking, however, a brief overview of the youth database project is necessary in order to give the reader a proper context for the following study.
Overview: The Need for a Youth Database
The youth database was conceptualized in 1999 in the midst of other ongoing technology projects. These projects were led by two faculty members from Rennselaer Polytechnic Institute in Troy, New York, USA. These faculty members, known as Teri and Jim, began a course in the spring of 1998 entitled "Web Design for Community Networking." The purpose of the course was to bring students from RPI in contact with local community members to build web systems such as web sites, online calendars and databases that reflected the needs of these community members. In this way, Teri and Jim began developing content for their community network. At the end of the semester, the community members were invited to RPI for a demonstration of the students' final projects. These projects were then brought together, housed on a common server, and shown through a common interface. The name given to this fledging community network was TroyNet (http://troynet.net).
In the spring of 1999, Teri and Jim taught Web Design for Community Networking a second time. This second iteration of the course provided further web content for TroyNet, and once again, community members enjoyed a demonstration of the students' work. These demonstrations, the constant involvement of community members, and tangential technological projects related to TroyNet, all provided Teri and Jim with opportunities for becoming acquainted with members of Troy's not-for-profit and city government community. One of these acquaintances was with a local city official named Will. Will was familiar with Teri and Jim's IT work with the community, and he had come to RPI's campus often to meet with the two professors and see students' work.
These encounters between Teri, Jim, and Will grew into a series of discussions on how to use IT for specific needs of Troy. In one such meeting, in the fall of 1999, Will spoke to Teri and Jim about the City mayor's desire to reinstate a youth bureau that would keep track of the various services and programs offered in Troy. Keeping track of these services was problematic, Will noted, as many of the not-for-profit agencies did not communicate with one another. This lack of communication translated into a duplication of services in Troy, where some agencies provided exactly the same services as their neighbors. Will hoped that once the youth bureau was re-instated, the youth bureau could reduce the duplication of services by improving the coordination and communication between not-for-profits. Attempting to improve communication, however, was a laborious, time-consuming task, and as Will expressed, the City was looking for ways in which to make this task simpler. Will therefore asked Teri and Jim if a computer system could help.
Will's question resonated with the professors' recent experiences with students in their course creating IT systems for various city and county uses. As Teri, Jim, and Will brainstormed the various ways IT might be used, the idea of a networked database emerged, one that would house and distribute information about the city's youth services, agencies, and events via the Internet. The meeting ended with the three agreeing to explore the possibilities of building such a database. This meeting marked the initial conceptualization of the youth-services database. Since its conception, this database has slowly undergone a number of transformations. For instance, community members, RPI students and faculty, and City Hall have all been involved in the database's design and development (see Harrison, Zappen, & Prell, 2001). In addition, Teri and Jim have written many grants to locate funding for this project. These grant writing activities, and the funding gained through awarded proposals, have shaped the database in particular ways. Through one grant proposal, the name "Connected Kids" was given the database project. Funding received from the National Science Foundation has allowed Teri and Jim to hire student programmers and interface designers for a period of three years. Funding has also helped Teri and Jim develop focus group meetings with community members and hold a series of design sessions with community members to test the different iterations of the database.
As of this writing, the Connected Kids database has yet to be completed. Even though Connected Kids is not yet completed, this study focuses on exploring the relationship between this community networking technology and democracy through focusing on the relationships among the participating community members. In doing so, the study focuses on one popular concept circulating within the community networking literature, that of social capital. Social capital is defined as the trust and resources found in social networks of relations within a community. Community networking is thus seen as improving social capital as a result of improving the relationships among users of the system.
A reader might very well wonder how a study might be be done on a community network's effects on social capital, when the technology in question is not yet completed. The author has eagerly watched and waited for the project to draw to a close in order to begin comparing the mediated relationships with their face-to-face counterparts. At the same time, defining a community network solely in terms of its hardware oversimplifies matters. Much of what makes a community network different from other communication technologies is the community's involvement and participation in the technology's design and development (Roberts, 1996; Schuler, 1996a, 1996b). Thus community networks are not just technological innovations, they are also social ones; their interventions take place on both a technological and social level. This study thus focuses its attention on the social aspects of the community network through concentrating on how the not-for-profit and community leaders communicate with one another, share resources, and locate trust in each other.
This study still holds relevance, however, for the technological enthusiast. As the community networking literature describes social capital as a potential outcome of this technology, a thorough review of social capital and development of measures relevant to a community networking scenario will assist future evaluators of these systems. Finally, one might also view this study as the basis upon which future research pertaining to the completion, adoption, and use of Connected Kids will take place.
The following sections thus offer a more thorough review of social capital as well as offer a discussion on how literature has linked social capital to community networking.
Community Networking and Social Capital
Community networking is often heralded as a tool for building a more democratic, civic community. In making such claims, many writers of community networking connect this technology to social capital, a concept that links features of a society such as social networks, trust, and reciprocity, to the functioning of democracy and communities (Putnam, 1993; Schuler, 1996). Proponents of community networks argue that levels of social capital should rise in those communities that adopt these networks (Friedland, 1996; Schuler, 1996a). In particular, community networks should increase the civic culture of the community (Friedland, 1996), the number of ties within that community, and should strengthen those ties that already exist. The simultaneous increase and strengthening of ties takes place as a result of the supplemental channels of communication provided by the community network. As community networks provide additional channels through which community members can meet one another, interact, communicate, and exchange information and resources, these increased interactions and connections among community members should also correspond with a rise in overall social capital (Blanchard & Horn, 1999).Although the literature discusses the relationship between community networking and social capital, no empirical studies (as of this writing) have yet to test this relationship fully. Part of the problem is the difficulty in measuring social capital (Foley & Edwards, 1999). Many social capital studies rely upon attitudinal measures on trust gathered from a random sample (e.g. Boxman, 1991; Knack & Keefer, 1997, Putnam, 1993). These trust measures are then compared with regional or national statistics on economic output, voting behavior, and average income. Thus trust, income, voting behavior, and so forth are not measured within networks based on social interaction. Putnam's (1993) study relied on such measures and consequently received a great deal of criticism for failure to link trust and civic behavior to actual social networks (Foley & Edwards, 1999). Although Putnam's later work (Putnam, 2001) addresses many criticisms put forth by the academic community, Putnam's (2001) findings are still frequently divorced from actual social networks. One analytical approach, however, has proven more successful in linking trust and resource exchange to social networks. This method, referred to as social network analysis (SNA), has yielded ample examples of studies focusing on social capital. The next section explores this method and sees how it relates to the topic of social capital.
Social Network Analysis and Social Capital
A social network is a set of individuals or groups who are connected to one another through socially meaningful relationships (Wellman & Berkowitz, 1988). Examples of such socially meaningful relationships include family, friends, or relations based on trust, giving advice, or sharing information. These relationships are then mapped out to see the patterns that emerge among individuals, groups, or organizations (Wellman & Gulia, 1999). In mapping out these patterns, the analyst looks at such issues as the quality of the relationships (Brass, 1992), the positions of actors within the network, and how both these aspects of the network affect the way information and resources flow (Wellman & Gulia, 1999). Social network analysis thus shifts the focus in social science research from the individual or group to the relationship between individuals or groups and how these relationships form an overall structure (Scott, 1991; Wasserman & Faust, 1994).
Common Networking Terminology
In addition to defining social networks, some common social networking terms need to be explained in order to understand the social network approach more fully. The following definitions are summarized from Wasserman and Faust (1999).
Actors
These are the nodes in the network. An actor can be an individual, a group, an organization, or even a nation-state. In general, social network analysts are concerned with actors of the same type, e.g. all individuals or groups.
Ties
These are the links between actors. These ties can be reciprocated, or unreciprocated, and they can be directed (e.g. a person giving another person money) or undirected (e.g. two people working at the same organization).
Relations
A relation is a specific type of tie between actors in a network. There are many different kinds of relations: communication or social interaction, friendship, reciprocity, trust, diplomacy, advice, and so forth.
Group
This is a bounded collection of actors on which ties are to be measured. One must be able to argue theoretically, empirically, or conceptually that the actors in this set are tied to one another and are more or less bounded. The actors belong together in a bounded set, one in which the number of actors is finite and the boundaries around this set of actors is clearly defined.
Social Network
Thus, returning to our original definition of social network, we can refine this definition even more: a social network is a finite set of actors who are connected to one another through relations. A social network can consist of groups and sub-groups of actors.
Social network analysts' view of social capital.
In terms of social capital, SNA scholars focus on the structure of social networks and see this structure as an indicator that social capital exists. Although all SNA scholars agree that structure is an indicator of social capital, some differences in perspective exist as to which structures act as the best indicators.
Borgatti, Jones, and Everett (1998) have grouped the different perspectives to measure social capital into two main groups: these are the "groupist" perspective and the "individualist" perspective. The "groupist" perspective sees social capital as existing within the whole network of relations. For example, a groupist approach would say that social capital is composed of cohesive networks (i.e. networks in which actors are interconnected with on another) in which trust and reciprocity reside. The more interconnected actors in a network are to one another, the more those actors trust one another, are able to exchange resources (tangible and intangible), and thus the group as a whole benefits.
Borgatti, et al. (1998) place Putnam (1993, 2001), Coleman (1990), and Bourdieu (1987) as lying within this "groupist" view of social capital. For instance, Putnam (1993, 2001) is interested in improving the well being of a society on many levels (e.g. politically, economically, and socially), and he sees well-interconnected social networks as one key component to allowing this to happen. Coleman (1990, 1993) looks at cohesive, bounded networks such as small villages and neighborhoods where actors frequently interact with one another. He then explores how such social structures and norms can be transposed to other areas of society through institutional design(s). Finally, Bourdieu (1987) describes cultural norms of generalized, symbolic reciprocity, where reciprocity both creates and emerges from the relations between actors within the network. Thus, all three look at the interconnectivity of the entire network, saying more interconnectivity coincides with more trust, reciprocity, and/or general well-being for the whole network. An individual may benefit from being a member of such an interconnected network, yet the emphasis remains on the quality of the whole network.
This groupist view of social capital cannot be fully explored in this paper, as thE study only looks at one network, and does not have other networks with which to compare how interconnectivity relates to trust and resource exchange. However, relations among individual actors within this network can be explored to see how these relationships correlate with such variables as trust and resource exchange. This attention to the structuring of relations surrounding each actor in a network reflects the "individualistic" perspective of social capital of Borgatti, et al.
The individualistic perspective focuses on actors' relationships with others in the network and seeks to measure who in the network has more social capital. For example, if an actor in a network has the most number of ties to others, then the individualist perspective would claim that this high number of ties is an indicator that this actor holds the most amount of social capital (Borgatti, et al., 1998). In social network terms, this person would be seen as holding a high level of degree centrality, i.e. a large number of social ties to other actors in the network.
Although Bourdieu (1987), Coleman (1990) and Putnam (1993) were earlier described in terms of the groupist view of social capital, they may also be seen as reflecting this SNA concept of degree centrality for individual actors. Bourdieu (1987) discusses individuals as benefiting from the social networks to which they belong, and he also mentions that individuals within a culture of reciprocal exchange benefit from having many ties within that cultural system, i.e. more ties within this larger cultural context corresponds to greater access to resources. Similarly, Coleman (1990) notes how individuals' social capital might be measured through a debit/credit system, where an individual holding the most amount of debits from others is the one with the most amount of social capital. In this case, the level of debits/credits acts as the indicator of the level of social capital. Finally, Putnam's (1993) work, especially his later work (Putnam, 2001), looks at how individuals with large social networks reap such benefits as finding jobs and having more healthy lifestyles.
Another type of centrality that acts as an indicator of social capital is that of betweeness centrality. Betweeness centrality refers to how many pairs of actors an actor stands between. If two actors in a network can only reach one another via a third actor, then this third actor is seen as holding more social capital as s/he can best manipulate the resources on either side (Burt, 1997; 2000).1
In terms of betweeness centrality, the work of Burt (1997, 2000) offers a good illustration of this alternative view of social capital. Burt's (1997, 2000) studies track individuals who occupy between positions within organizational settings. His research finds that these between actors receive earlier promotions than their peers, all other variables being held constant. Burt (1997, 2000) concludes from these findings that the unique position of being between two individuals offers that individual a strategic advantage.
In the context of community networking, one might ask why betweeness centrality is even being discussed as a possible measure for social capital. Community networking emphasizes building cooperation and community, not power and competition. Thus, one might argue, the only important measure to consider in the given context is degree centrality, and over time, the overall cohesion of the group. Both these measures focus more on building ties among actors rather than gaining advantage over competitors.
To counter this argument, one should consider the goal of building this community network, i.e. helping not-for-profits work together, collaborate, and better serve the youth population. These goals of collaboration and serving youth are intrinsically tied to notions of access to and exchange of resources. By comparing betweeness centrality to degree centrality, one might thus gain a clearer understanding of which kind of social structure correlates best with trust and resource exchange. Further, simply knowing whether the not-for-profits currently are structured along lines of betweeness or degree centrality is important information to have in attempts to understand more fully how not-for-profits currently relate to one another.
As can be seen, social network analysts measure social capital through structural features such as interconnectivity and centrality. These structural features have been associated with such benefits as higher income levels (Boxman, et al, 1991), time of promotion within an organization (Burt 1997), and reciprocity and trust among one's peers (Tsai 2000).
Given this discussion on the ways SNA scholars perceive and measure social capital, the nature of the setting of this study, and the wider discussion pertaining to social capital, the following research questions can be posed:
Research Question 1: How do actors' degree centralities for trust, social interaction, and resource exchange relate to one another?
Research Question 2: How do actors' betweeness centralities for trust, social interaction, and resource exchange relate to one another?
The next sections discuss the methods and measurements used for exploring these research questions.
Method
Thirty-three individuals, composed of representatives from various youth service not-for-profit agencies, faculty from RPI, city government employees, and administrators from the local school districts, were selected for this study. This selection was based on a list of individuals who had been contacted by Teri or Jim to be involved in the Connected Kids project. Of these 33 individuals, 30 were interviewed and given a questionnaire.Survey
A questionnaire containing Likert-scale and nominal items was administered one-on-one with each respondent during an interview. These questions were focused on the relationships among all 33 respondents and measured social capital concepts such as frequency of communication, trust, and exchange of resources.
All questions were asked out loud to respondents and respondents' answers were recorded on a data sheet. This was done for a particular reason. As each respondent was evaluating or offering information about the relationship s/he held with each of the other respondents, answers to these questions required not only thought pertaining to the questionnaire item, but also to evaluation of all other 32 respondents. I thus opted to have respondents focus their attention solely on a list of names of the other respondents, rather than focus attention on the questionnaire items. As respondents studied this list, I would ask a series of questions pertaining to each person on the list as well as each person's respective organizations. In doing so, I encouraged respondents to focus their attention on how they related to each person and how their organization related to the other organizations.
Measurements for Social Capital
The items used for measuring social capital are found below.
Centrality measures. Two centrality measures are being explored in this study across relations of trust, resource exchange, and social interaction.
Degree centrality. The degree centrality computes the centrality of each actor in the network and summarizes this result as a proportion of the maximum possible degree for all actors. The calculation for obtaining a person's degree centrality is quite simple. It involves simply counting the number of individuals immediately tied to that person. There are some options for how to count these ties: a) one might count how many nominations a person receives (referred to as a person's "in-degree"), b) one might count how many nominations a person gives (referred to as a person's "out-degree"), or c) one could symmetrize the data and calculate an overall degree centrality score.2
For this study, option "a" was chosen, i.e. calculating an actor's in-degree centrality. In-degree centrality was chosen as this calculation best reflects the notion of popularity: if one is interested in seeing how popular an individual is in a given network, one is most interested in how many nominations s/he receives, not in how many people s/he nominates. Thus, as in-degree centrality is the measure being used for analysis, the remainder of this paper will refer to in-degree centrality versus degree centrality.
Betweeness centrality. Betweeness calculates the number of times an actor stands between two other actors, with the two actors holding no connection with one another. Calculating betweeness centrality thus involves simply counting the number of times an actor holds this position.
Social Interaction. As noted earlier, social capital theorists look at how social interaction relates to trust and reciprocity (Bourdieu, 1987; Coleman, 1990; Putnam, 1993; 2001). To measure social interaction among this group of not-for-profit respondents, the following question was developed:
Which individuals on this list have you had the most frequent contact over the past 6 months? Please choose 3.Limiting the number of possible names to three was a decision based on looking at other social network measures (e.g. Burt, 1997; Coleman, Katz, & Menzel, 1966).
Resource Exchange. Resource exchanges for the current study were measured in terms of funding exchanges and joint programs.
Funding exchanges. Early initial interviews revealed that a primary way in which not-for-profit youth agencies exchange resources is through the funding and/or subsidizing of programs for youth. The following items were developed to reflect this type of resource exchange:
Within the past year, have any of these organizations funded your organization in any way?For these items, I approached respondents as representatives of their organizations. In this way, although only speaking with one representative from an organization, I gathered information on each individual's organization.
Vice versa?
I used the two items above as a means to validate whether an organization funded another organization. Thus, if one respondent told me, in response to the first item, that his/her organization received funding from another organization, I checked how the other organization responded to item two: if this second organization said that they gave funding to the first organization, I kept the original nomination. Nominations that did not pass this validity check were thrown out. After these validity checks were made on the data, I used only one dataset for analysis purposes. This dataset was the one based on the first item, i.e. "Within the past year, have any of these organizations funded your organization in any way?" The nominations respondents gave to this question told me from whom respondents received money. Conversely, the nominations a respondent received from others told me to whom that respondent gave money.
Funding was not the only means through which resource exchange took place. Another major activity that emerged from early interviews was that of program development. Program development consists of ongoing services and/or activities designed for a specific clientele, in this case youth. Often, programs involved two (and sometimes more) organizations, where one organization would provide a service to another organization's clientele. This would frequently happen when the two organizations offered radically different kinds of services, for instance art classes versus after-school sports. In such a scenario, the organizations would work together to coordinate how best to use the services of each for a specific group of youth. Thus, the following question was designed to capture programming activities between organizations:
Another way organizations collaborate is through program development. Within the past year, has your organization collaborated with this other organization via program development?With this question, an organization was reporting on other organizations with which they shared programs. Thus, directionality was not important as with the question pertaining to funding, where information was given on whether or not they received funding from that organization. Because the nature of this program development tie emphasized a shared link versus a directed one, I opted to symmetrize the data in UCINET. Symmetrizing data converts the data to only reciprocal nominations between actors. Thus, if one person nominated another person and that nomination was not reciprocated, then the data for that tie was thrown out. Otherwise, if nominations are reciprocated between two actors, then the data for that pair was kept within the dataset. Symmetrizing the data in this way thus helped ensure the validity of a shared link, as both actors nominated one another.
Trustworthiness. Tsai (2000)'s social network study of social capital developed questions that focus on trust within an intra-organizational setting. These questions from Tsai (2000) have been modified to reflect the nature and unique context of the current study. Thus, two questions were developed:
1. Suppose you were looking for partners and/or collaborators for a joint project. Which of the people on this list are you certain will do what you require (what you believe they should do) even without writing a contract to clearly specify their obligations?The nature of both these questions asked whom on the list a respondent would trust for either work on a project or information about youth. Thus, respondents were commenting on other actors' trustworthiness in providing something for that respondent. For data gathered from this question, knowing who in the network nominated whom was important information to maintain, thus data was not symmetrized.
2. Which of these people are you certain can provide you with reliable information regarding issues pertaining to your work with youth?
Analysis
The data gathered for this study was social networking data. As such, the data was first structured in matrices in UCINET. Data gathered from each item of the questionnaire were compiled in their own matrix. For exploring how in-degree centrality scores for social interaction, resource exchange, and trustworthiness relate to one another, the following calculations and analyses were performed:
- Program development data were symmetrized for better reliability regarding whether a joint program existed between two respondents. Funding data were not symmetrized as the data were directional. Data for the two trust items were likewise not symmetrized, as information on who nominated whom in the network as trustworthy was important information to maintain.
- The in-degree centrality scores were calculated in UCINET for social interaction, program development, funding, willingness to collaborate without a contract, and reliable information. This was done to keep the structural attributes of each actor consistent across all relations (trustworthiness, resource exchange, and social interaction) for testing purposes.
- The in-degree centrality scores were then correlated in SPSS using two-tailed Spearman Correlation Coefficient. Spearman Correlation was chosen over using Pearson Correlation as data were nonparametric.
Results
The descriptive statistics are offered in Table 1 and are then followed by the Spearman Correlation Coefficients for in-degree centrality and betweeness centrality, which can be found in Tables 2 and 3 respectively.| M | SD | ||
| Frequency of Contact | |||
| In-degree | 2.46 | 2.61 | |
| Betweeness | 29.72 | 47.646 | |
| Trustworthy Collaborator | |||
| In-degree | 8.03 | 5.58 | |
| Betweeness | 30.15 | 41.25 | |
| Trustworthy Information | |||
| In-degree | 8.91 | 4.55 | |
| Betweeness | 19.79 | 37.80 | |
| Nominations for funding | |||
| In-degree (incoming) | 2.76 | 4.87 | |
| Out-degree (outgoing) | 2.76 | 2.73 | |
| Betweeness | 4.79 | 14.69 | |
| Developed Program Together | |||
| In-degree | 1.88 | 2.01 | |
| Betweeness | 10.79 | 19.62 | |
The descriptives above show some important trends. First, one notices right away that the betweeness centrality scores are consistently larger than their in-degree counterparts. This reflects the difference in which the two centrality measures are calculated. In-degree centrality ignores nominations flowing away from a respondent and sums only those nominations an actor receives. Betweeness centrality, however, uses both incoming and outgoing nominations in making its calculations.
The second observation to be made by the above data is that all the standard deviation scores are consistently large and, in some cases, larger than the mean. Such large values indicate that actors vary a great deal in terms of in-degree and betweeness centrality. In fact, a close look at the data shows that a small minority of actors dominates the network in terms of both in-degree centrality and betweeness centrality. Figures 1 and 2 illustrate this uneven spread:
Figure 1. Graph showing betweeness centrality.
Figure 2. Graph showing in-degree centrality.
Figures 1 and 2 show many of the same actors dominating the network both in terms of in-degree and betweeness centrality. Thus, one can make a qualitative judgment in saying that this network is far from the Putnam (1993, 2001) ideal of a horizontally structured network, where actors are inter-connected and tied to one another.
I will return to these findings later on in the discussion. For now, Table 2 will be discussed in terms of the results for the in-degree centrality scores.
| Social interaction | ||
| Trustworthy information | . 81** | |
| Trustworthy collaborator | . 76** | |
| Program development | .15 | |
| Funding -outgoing nominations | .33 | |
| Funding – incoming nominations | .41* | |
| (n = 33) | * p < .05, ** p < .01 |
Table 2. Correlations among social interaction, trust, and resource exchange for in-degree centrality.
Table 2 shows the in-degree scores for social interaction having significant positive correlations with trust and resource exchange. These results translate into the following: an actor's popularity (i.e. number of nominations as someone with whom others in the network have frequent interaction) coincides with others' nominating that actor as a) one who gives funding and as b) one who is trustworthy. Popularity does not coincide with an actor's developing programs with others or with receiving money.
Discussion
In-degree CentralityThese results described above make sense given Coleman's (1990) notion of the debit/credit system, which states that people holding the most debits are those holding the most social capital. This interpretation, when viewed in the context of the current data, can be extended to say the following: if I am someone in the network to whom many owe debts, many will have frequent contact with me (perhaps in hopes of accruing more goods or resources, or perhaps to assure me that the debts will be repaid), and many will trust me, as I am obviously someone who can follow-through and provide resources in a time of need. In fact, a look at the data reveals that most of the popular actors are also employees of government agencies. These agencies circulate funding resources among the not-for-profits. Thus, funders in this network seem to hold a large amount of social capital.
Qualitative data gathered during the time of this study also help confirm this interpretation. As I learned from talking with my respondents, not-for-profits in Troy, and across the nation for that matter, have traditionally lacked both funds and staff. In lacking funds and staff, these not-for-profits must be selective in how they focus their energies. Thus, energies get focused on serving their clients, and not on connecting with actors outside their organizational settings. The exception to this rule, however, would be those organizations that provide resources such as funding.
The findings in Table 2 do not show a correlation between popularity and program development. One reason may have been that the measure for program development was too vague. From my background interviews in March, 2000, I had gained the impression that funding and program development were significant ways in which agencies collaborated with one another. However, I noticed as I was administering the survey that the two items, "funding" and "program development," confused a number of the participants. Many asked if there was a difference between the two. In response, I tried to not impose definitions upon respondents, but rather asked them what they thought those terms meant and then suggested they answer the question based on their own understanding.
Thus, respondents may have become confused by both of these items, although funding seems to have been a more salient term for respondents than program development. Further, when I later analyzed qualitative data gathered at the same time during the study, I became aware that respondents discussed other inter-agency activities that might also translate into forms of resource exchange or reciprocity: these included one-time events, such as fairs or festivals, the mutual referral of clients, and writing grants together. As I had not asked respondents about these activities, information may have been lost as to the unique forms of reciprocity found within this not-for-profit community.
Finally, the restriction I placed on respondents in nominating with whom they had the most frequent contact may have skewed these results. The fact that respondents could only nominate three actors may have eliminated additional communication ties. Thus, although I was able to gather information on the respondents' strongest communication links, weaker links were most likely lost. These weaker links may have been ones through which respondents engaged in program development. Thus, the lack of social interaction data may have affected the results.
Further interpretation will be given to these findings shortly. First, however, the results for the betweeness centrality scores need discussion. The next table, Table 3, shows the results for the betweeness centrality scores.
| Social interaction | |
| Trustworthy information | .73** |
| Trustworthy collaborator | .50** |
| Program development | .31 |
| Funding -outgoing nominations | .29 |
| Funding – incoming nominations | .15 |
| (n = 33) | * p < .05, ** p < .01 |
Table 3. Correlations Between social interaction, trust, and resource exchange for betweeness centrality.
Table 3 shows the betweeness scores for social interaction having significant positive correlations with both trust measures. Thus, actors who occupy positions between others, based on social interaction, are also the same actors situated between flows of trust. Unfortunately, the betweeness centrality score does not reflect the direction of the nominations; thus, it is difficult to determine whether the trust nominations flow towards the between actor or away.
Betweeness Centrality
This correlation between trust and betweeness centrality is surprising, given the theoretical framework outlined earlier. Recall Burt's (1997, 2000) conceptualization of social capital: actors holding between positions are seen as powerful in that they gain a competitive advantage over their peers in accessing and manipulating resources. Yet critics of Burt have noted that in-between actors could equally be seen as being constrained versus powerful. According to this argument, in-between actors must answer to a variety of alters, and this pressure to respond to several different alters constrains an actor's mobility. Thus, in-between actors are less powerful according to this line of argument (Krackhardt, 1999).
This challenge to Burt's (1997, 2000) view highlights the interpretive flexibility of betweeness centrality: the environment in which a network is placed and the motivations of actors constituting that network have much to do with the meaning of a network's structure. Thus, the organizational settings for Burt's (1997, 2000) work limit the generalizability of his findings. Burt's between actors are managers in large firms seeking promotions and raises. In contrast, the community of not-for-profits in Troy is smaller in scale than those typically studied by Burt, and the actors involved in this study are not being asked questions about promotion and incomes. Thus, a different scale of network is being studied along with a different set of motives.
The motives of not-for-profit actors relevant to this study pertain to the general lack of staff and resources not-for-profits typically must face. As discussed earlier, not-for-profits tend to forgo connecting with other agencies in order to concentrate time and staff on intraorganizational needs. Thus, in such a climate, a not-for-profit actor might choose to connect with a more between and trustworthy person because doing so would permit that actor to accomplish many things at once, namely a) stay connected with the rest of the network, and b) receive news about the rest of the network (including what all have in common, i.e. youth).
This new interpretation shifts the focus away from the between actor to those connected to the between actor. Whereas Burt (1997, 2000) focuses on the benefits of being a between actor, this new interpretations shifts the focus toward the benefits of being connected to a between actor. In doing so, one can argue that these not-for-profit actors make choices on whom to interact with based on very practical reasons.
These results from Tables 2 and 3 also indicate that betweeness centrality and in-degree centrality are highly correlated with one another. In fact, collapsing the measures into subscales and calculating Spearman Correlation Coefficients between these subscales reveals that, overall, betweeness centrality and in-degree centrality are highly intercorrelated across relations of trust, resource exchange, and social interaction, as shown in Table 4:
| In-degree subscales | Betweeness subscales | ||
| 1 | 2 | 3 | |
| 1. Trustworthy | .498** | .540** | .033 |
| 2. Social interaction | .477*8 | .777** | .214 |
| 3. Resource exchange | .463** | .588** | .686** |
| ( n = 33) | * p < .05, ** p < .01 | ||
Table 4. Intercorrelations between subscales for in-degree and betweeness centrality.
Thus, these two centrality measures, although theorized as being two competing indicators for social capital, are in essence interchangeable. In addition, these results reinforce earlier conclusions drawn in this paper that a few actors are dominating the network. Finally, these results also show that in certain scenarios, betweeness and degree centrality are virtually synonymous with one another (Freeman, 1979).
Conclusions
This paper shows that popular people are also people holding the strategic position of being between actors. Theoretically, these findings coincide with Coleman's (1990) notion of the debit/credit system where those surrounding a central actor tend to be reliant on that actor and not on one another. However, in terms of Bourdieu's (1987) and Putnam (1993, 2001), these findings are not supportive of these two scholars' claims: both argue that individual actors benefit from holding a great many ties because these ties provide access to resources. As these findings show, popular actors, i.e. those with whom actors like to interact, are the ones giving funds, not receiving. Further, Bourdieu (1987) and Putnam (1993, 2001) claim that it is more important that the overall network is interconnected, i.e. that ties are being distributed more or less evenly, as opposed to the individual's being well-connected. As the descriptive statistics show in Table 1, a wide spread exists among actors' quantity of ties. The network, in other words, is not very cohesive.These findings also show, more specifically, that government agency employees are the most popular and powerful actors in the network. Ties are concentrated around these actors, indicating that the network is structured vertically rather than horizontally. I have already attempted to account for the meaning of these findings: not-for-profits have historically been understaffed and not in positions to network throughout the community. Relying upon a few powerful figures makes sense, given these agencies' circumstances.
One significant limitation of this paper is its failure to capture the variety of ways not-for-profits may be reciprocating with one another. Additional interviews revealed that not-for-profits are in the process of changing how they pursue funding as a result of a change in funding rules established by federal and state governments. In particular, these not-for-profits are now required to establish collaboratives among themselves to apply for grants jointly. These collaboratives thus provide a potentially powerful source of reciprocity. In addition to collaboratives, interviews revealed that many not-for-profits engage in other forms of reciprocity, e.g. client referrals. Thus, future research needs to account for these alternative forms of reciprocating.
Finally, the fact that respondents to my survey were not given the opportunity to list all with whom they communicate has limited my ability to measure the levels of interconnectivity in this network fully. Thus, although government agency employees may still be the most popular and powerful in terms of funding, other actors might figure as more central given other forms of reciprocity. Future research will have to explore these possibilities.
In terms of the differences between the two measures of centrality used for this study - in-degree and betweeness centrality - the in-degree measure has proven to be the more suitable of the two. Betweeness centrality, when used for asymmetric data, fails to make distinctions between the directions of nominations. In contrast, in-degree centrality sums the total number of nominations an actor receives. Knowing the direction of nominations is important to understanding how trust and resource exchange flow throughout the network. Thus, in-degree centrality offers the researcher a better sense of which actors are getting nominated for which aspects of social capital.
This study has been as much an exploration of the kinds of reciprocity that exist among the not-for-profits as well as understanding how two different forms of centrality relate to trust, reciprocity, and each other. As my measures for reciprocity and interaction improve, I hope to capture the full richness of interactions within this community. In doing so, I hope to ascertain the role a networking technology such as the Connected Kids database might play in the ongoing dynamics of this community.
Footnotes
1. Burt (1997, 2000) discusses this same view in terms of "structural holes" versus "betweeness centrality." Yet as Brass (1992) notes, betweeness centrality, in essence, reflects the same theoretical stance as structural holes. Thus, as this study is interested in contrasting different measures of centrality, and as no real theoretical difference exists between structural holes and betweeness centrality, the betweeness centrality measure is being used.2. Symmetrizing the data reduces the data to only reciprocal ties in the network. For example, if Jim nominates Sal, and Sal nominates Jim, then this link is kept within the dataset. On the other hand, if only Jim nominates Sal, and not vice versa, then Jim's nomination gets thrown out of the dataset, as it is an unreciprocated tie.
References
Blanchard, A. & Horan, T. (1998). Virtual communities and social capital. Social Science Computer Review, 16, 293-307.
Borgatti,
S.P., Jones, C., &
Boxman, A.W., De Graaf, P.M., & Flap, H.D. (1991). The impact of social and human capital on the income attainment of Dutch managers. Social networks, 13, 51-73.
Bourdieu, P. (1986). The forms of capital. In J.G. Richardson
(Ed.), Handbook of theory and research for the sociology of education.
Brass, D.J.
(1992). Power in organizations: a social network perspective. In G.
Moore and J.A. Whitt (Eds.) Research in politics and society,( pp. 295-323).
Burt, R.S. (1997). A note on social capital and network content. Social networks, 19, 355-373.
Burt, R.S. (May, 2000). The network structure of social capital. Pre-print
for a chapter in R.I. Sutton & B.M.Staw (Eds.) Research in
organizational behavior, 22.
Coleman, J.S., Katz, E., & Menzel, H. (1966). Medical
innovation: A diffusion study.
Coleman, J.S. (1990). Foundations of social theory.
Coleman, J.S. (1993). The design of organizations and the right to act. Sociological forum, 8 , 527-546.
Foley, M.W., & Edwards, B. (1999) Is it time to disinvest in social capital? Journal of public policy, 19, 141- 173.
Freeman L C (1979). Centrality in social networks: conceptual
clarification. Social networks 1,
215-239.
Friedland, L. (1996). Electronic democracy and the new citizenship. Media, culture, and society, 18, 185-212.
Garton, L., Haythornthwaite, C,
& Wellman, B. (1999). Studying on-line social betworks. In
S. Jones (Ed.), Doing internet research (pp. 75-106).
Harrison, T.,
Zappen, J., & Prell, C. (forthcoming). Transforming
new communication technologies into community media. In N. W. Jankowski
and O. Prehm, Community media in the information age: Perspectives and
prospects.
Ibarra, H. (1993). Network centrality, power, and innovation involvement: determinants of technical and administrative roles. Academy of management journal, 36(3), 471–501.
Knack, S., & Keefer, P. (1997). Does social capital have an economic payoff? A cross-country investigation. The quarterly journal of economics, 3, 1251-1288.
Krackhardt, D. (1999). The ties that torture: Simmelian tie analysis in organizations. Research in the sociology of organizations, 16, 183-210.
Putnam, R.D. (1993). Making democracy work: Civic traditions in
modern Italy.
Putnam, R.D. (2001). Bowling Alone: the
collapse and revival of American community.
Roberts, R.A. (1996). The
diffusion of innovation: Dualities of one electronic free community network. Dissertation for
Schuler, D. (1996a). New
community networks: Wired for change.
Schuler, D. (1996b). How to kill community networks. Hint: We may have already started. The Community network movement, (consulted on February 2003): http://www.scn.org/ip/commnet/kill-commnets.html.
Scott, J. (1991). Social
network analysis: A handbook.
Tsai, W. (2000). Social capital, strategic relatedness, and the formation of intraorganizational linkages. Strategic management journal, 21, 925–939
Wasserman, S., Faust, F. (1994). Social network analysis:
methods and applications.
Wellman,
Wellman, B. & Gulia, M. (1999). Virtual communities
as communities: Net surfers don't ride alone. In M.A.
Smith & P. Kollock (Eds.), Communities in cyberspace (pp. 167-194).
About the Author
Christina L. Prell received her Ph.D. from Rensselaer Polytechnic Institute in Communication and Rhetoric. She is an Assistant Professor in Communication at McDaniel College. Her research interests include computer-mediated-communication, social capital and IT, social network analysis, and the social construction of technology.
Address: Christina Prell, Ph.D. Candidate, Language, Literature, and Communication, Rensselaer Polytechnic Institute, Troy, NY, 12180.
©Copyright 2003 Journal of Computer-Mediated Communication