In a fascinating bit of research from MIT (covered here by the BBC), social network researchers have used mobile phones to collect high-resolution data on communication (via call-logs), location (via GSM base-stations) and proximity to other people (via continuous Bluetooth scanning) in a study of 94 people. The Beeb's report majors on the disparity between how people describe their own interactions with other people (self-reporting) and the data from the phones reveal, but the main point is really the implications of being able to accurately map these relationships within communities.
Segment of a Social Network
Analysis (Creative Commons,
courtesy of Wikipedia user
One of the analyses the researchers performed on the data was to find out whether they could predict the nature of relationships between people (defined in the charming terms: reciprocal friends, non-reciprocal friends and reciprocal non-friends) from mobile phone data. Using the volume of phone communication and the proximity data from the bluetooth scans, they were able to predict friendship with 95% success.
You can have a look at the full paper here. It contains phrases like "Using a multiple regression quadratic assignment procedure, common to the analysis of the adjacency matrices representing social networks, we can assess the significance of the predictive value of variables".
The implications of this could be pretty staggering to social network analysis. As the paper concludes:
"...these methods allow for an inspection of the dynamics of macro networks that were heretofore unobservable. There is no technical reason why data cannot be collected from hundreds of millions of people throughout the course of their lives. Further, while the collection of such data raises serious privacy issues that need to be considered, the potential for achieving important societal goals is considerable."
This particular study was based on under 100 people at MIT, but imagine what an insight it could give to researchers and policy makers interested in social capital, or (dare I say it) behaviour change initiatives. Some behaviour-change research that I heard about, for example concentrates on identifying people in communities who are the influencers (or mavens to use the Gladwell-ism).
For more on social capital and social networks (online and offline) - do have a look at the RSA's Connected Communities project blog.