Friday, October 8, 2010

Facebook, Social Networking and Social Network Analysis

When I tell folks that one of the subjects that I teach is “Social Network Analysis,” most (not all) mistakenly confuse this with social networking, Facebook and the like. While it is true that the network of individuals, groups and organizations that are a part of Facebook constitute a social network, it is only one type of social network that folks like myself study.
What is social network analysis? Well, it would be misleading to suggest that there is a unified social network theory. Instead, it is better to view social network analysis (SNA) as a loose collection of theories and methodologies that analysts use to better understand and predict social behavior.  A key assumption of SNA is that our behavior is profoundly affected by our ties to others and the networks in which we are embedded.  And it isn’t just our direct ties that affect our beliefs and behavior; our indirect ties affect them as well. Indeed, social network theories assume that our structural location (i.e., where in a social network we find ourselves – center, periphery, member of core group, social isolate, and so on) affects what we say and do.
For instance, we are much more likely to join a church, mosque, secular movement, etc. to which we have a tie (i.e., we know someone) than we are to one that we don’t.  I believe it was the Presbyterian Church USA that found that 85% of the individuals they surveyed (who attended Presbyterian churches) joined the particular church to which they belong because they already knew someone. Similarly, Marc Sageman found in his study of terrorist networks that 83% of the folks who joined what he calls the global salafi jihad (what most folks refer to as Al-Qaeda and Al-Qaeda related groups) joined through personal ties (friendship, kinship, mentorship). And, these percentages keep showing up, whether they are studies of other religious movements such as the Mormons or secular ones such as the Civil Rights movement.  
Researchers have also found that we are more likely to be happy if we have friends who are happy (or friends of friends who are happy), we are more likely to be obese if we have friends who are obese (or friends of friends who are obese), and we are more likely to get divorced if we have friends who have gotten divorced (or friends of friends who have gotten divorced).
SNA differs from other more traditional approaches to data analysis because while the latter tends to focus on attributes (e.g., race, ethnicity, gender, age, education level), SNA focuses more on the pattern of relationships. The reason for this is simple: while peoples’ behavior often changes from one social context to the next, their attributes do not. Thus, we cannot explain such changes in behavior in terms of their attributes. What does change, however, is the social context (i.e., the pattern of relationships) in which their behavior occurs, and SNA is one method of empirically mapping social context.  This is not to say that social network analysts regard attributes as unimportant.  Nevertheless, most do believe that the patterns of ties in which actors find themselves is a very important factor in understanding and predicting behavior and beliefs.
One of the most entertaining introductions to the power of social networks can be found in the book by Nicholas Christakis, M.D. and James Fowler, Ph.D., Connected: The Surprising Power of Our Social Networks and How They Shape Our Lives (London: Little, Brown and Company).  Unlike a lot of the more technical introductions, this one targets a popular audience, making it highly readable. If you’re unsure whether you want to plunk $$s down on a book just now, you may want to check out the 18-minute talk that Christakis gave last February at the prestigious TED talks:





Two other books worth considering are Albert-Laszlo Barabasi’s
Linked: The New Science of Networks (Cambridge, MA: Perseus Publishing) and Duncan Watts’s Six Degrees: The Science of a Connected Age (New York: W. W. Norton & Company).

2 comments: