Network Positions
This week’s readings explored the factors that affect centrality and prestige, and the ways that status or position in a social network can impact society. Wasserman and Faust in “Centrality and Prestige” discuss “most important actors” and the measures that can describe importance and prominence. A prominent actor is one whose ties are visible to other members of the network. Choices of ties and the effects of indirect ties both impact the level of prominence. Centrality and prominence are related according to the authors. A central actor is one who is involved in many ties. They focus their discussion on the example of Padgette’s Florentine Families; a table analyzes the levels of prominence of families in Florence. I could not easily make connections between the tables and indexes and the examples used? The way that they were able to manipulate mathematical data and individual characteristics in the example where they were tracking the movement of medical information in the network was interesting. Though the results were not generalizable to the overall study of network science, it seemed like a creative way to come to conclusions about what affects centrality and prestige.
Expanding on the first article, Linton C. Freeman’s “Centrality in Social Networks” discusses the different types of centrality. Point centrality was most interesting to me. It is based on distance from other points, the degree number, strategic location, and the ability to avoid the control potential of other points in the star. The last two criteria were particularly interesting. Strategic location seems to be related to resources possibly. Ability to avoid control potential is a different idea to me, but I agree that the freedom to pass one’s own messages and be able to act independently within the network with a bird’s-eye view is an important aspect of centrality that may be overlooked. Discussion Question: What do you think influences strategic position of a central actor?
“Uncovering Terrorist Networks” is an example of the ways in which network science can be useful in other areas. Krebs’ goal is to map the organizational networks of terrorists. The author analyzes the known ties for closeness, acknowledging that there will be problems with incompleteness in her data because of the group that she is studying. Her statement that there are no ties outside of the hijacker network could be a result of incompleteness. It seems like it would be impossible to conclude that there are no ties outside of the network when it would be difficult to access people who know terrorists; they might not be forthcoming with information. She points out that even though the ties are dense, they provide the redundancy, trust, and resilience that a covert terrorist network depends on.
The Valente et al. article sets out to figure out the connection between smoking and popularity among adolescents. They discover that there is a correlation and they suggest that popular kids are more likely to smoke. They are using a limited sample, so they do not try to generalize, but they do stand by their data because there is a strong connection. Possible explanations include the fact that they contribute more to supporting social norms which in this case means smoking. Also, the smokes name fewer friends, which could suggest that they are making more friends outside of school and those people could be the smoking influence. Discussion Question: What do you think about the apparent paradox between isolates and popular kids?
Mattapa et al. logistical regression analysis seeks to find the predictors of bullying and victimization in social networks. Again, this group is using a small sample, but still finding important and dramatic results. They uncover longitudinal evidence that friendships prevent victimization, which supports the earlier reading about the basics of social networks and the ability to provide resources and protection. Another important finding that supports prior research is the data that bullies and aggressive victims choose aggressive friends, which could be homophily. In addition to the small sample size, the researchers are also concerned about biased self-reports and the 160-item survey they base their conclusions on.