Social Network Analysis: Methodology
How does one measure “social capital?” It is difficult to do so directly. Social trust is not lying about waiting to be observed. As a proxy, social network analysts have used the concept of a social “tie” (cf. Serageldin and Grootaert, 1999, p. 13) A social tie is simply a connection between two entities in which information is shared. For instance, Person A may be “tied” to person B by the fact that “A” routinely provides some kind of information to “B” (or vice versa). Now suppose a person “C” exists who is tied to “B” but not to “A.” In this case, we can say that “C” is one degree removed from “A,” but possesses at least a weak “tie” to “A” in the sense that “C” might ask “B” for an introduction to “A,” and so quickly and easily develop a more vigorous tie. The assumption here, of course, is that the simple sharing of information is a proxy for social trust. Not all scholars agree on this point. In a review of Putnam’s work, for instance, Levi (1996) argues that Putnam “never offers a precise definition of trust,” and so has an undifferentiated conception of social capital (p. 46). In general, however, researchers in this field agree that social interaction—in effect, social ties—lower the costs of participation in public life and so are likely to enhance a community’s reserves of social capital (cf. Coleman, 1988; Dasgupta and Serageldin, 2000; Grix, 2001). Social ties, in other words, are a measure of social trust, and social trust is a proxy for social capital.
Thus, our study tracks ties between civic organizations at Lake Tahoe. To this end, our method was fairly simple. First, we generated a list of civic organizations around the lake. We compiled our list through several means: the phone book, websites developed to civic life at the lake, and a snowball method in which we asked every organization to name other organizations doing work around the lake. As a criterion for inclusion, every group had to have a current phone number and at least one staff member. Some groups we found did not fit even this loose criterion. For example, the local Alzheimer’s support group, while doing good work for the community, did not qualify because they are handled by someone who works out of her home and does not have a staff that answers phones and sets official policy for the group. On the other hand, Sierra Watch has a staff that sets policy and reports to an office each day. Those characteristics qualify Sierra Watch for inclusion in the survey. Overall, we developed a database of 148 organizations.
As a second step, we developed a short questionnaire to survey the organizations in our database (see Appendix A). To keep our list of questions short, we focused on whether the organizations shared information with or received information from other organizations about civic life around the Lake. Our idea was that sharing information is a proxy for social interaction, interaction for trust, and trust for social capital. We also included a set of demographic questions related to the organization’s size, tenure in Tahoe, and the like, so that we could correlate kinds of groups with kinds of ties. Finally, we asked a few reputational questions concerning “most influential” civic activists” around the Lake.
As a third step, we called the organizations. This turned out to be more difficult than it sounds. On the rare occasion, a person actually picked up the phone and answered our questions on the first call. More commonly, we had to do many call-backs. This may be a drawback to doing this kind of work as journalism. The simple act of getting the information is very time consuming and pain staking. News organizations, many of which are already short-staffed, may be unwilling to devote precious resources to this exercise. After doing it ourselves, we have some ideas for how it might be done more efficiently. More on that in our discussion section. As a demonstration of how difficult this process is, though, consider that of the 148 organizations in our database, we succeeded in gathering information on 49—roughly one-third of the total. We simply never got hold of some groups. Other groups refused to answer our questions, and still others couldn’t answer our questions. This may seem like a small number then, but after repeated call-backs over several weeks, to us it seems like a grand success.
We entered all of our data into an Excel spreadsheet (available at TBA). As is common practice in SNA, we listed every organization as an “ego” in rows and as “alter-egos” in columns. We placed a “1” in every cell in which an organization either received or gave information to another organization. We then established separate pages for “information to” and “information from” data. After completing each grid, we totaled the numbers of each row and column and eliminated the groups that had zero connections.
Once this process was complete, we took one more step. One of the dangers in this kind research is that people will exaggerate the truth. In other words, they will say that they are more connected with others than they actually are. To limit the danger of this bias in our data, we decided to construct a final database, which included only groups that were mentioned by other organizations as information providers. Our reasoning was that organizations were less likely to exaggerate the number of groups they received information from than they were the number of groups to which they gave information. By adding how many times a group was mentioned as an information giver, we determined with better accuracy which organizations were most connected in the web of information around Tahoe.
We used the SNA program UCINET to analyze our data. UCINET is a networking software application that takes the data we plugged in and uses it to build connections for a visual display. UCINET displayed the network and we saw that there were connections and we had successfully built a social network map of the data we had collected.