As I noted earlier this week, I attended a 5-day ICPSR course on Social Network Analysis. Yesterday, I decided to take some of the network software out for a spin by analyzing a dataset I have on branch offices of National Law Journal 250 firms (the 250 largest firms by number of lawyers) for 2006.
The chart below [click to enlarge] is a "2-mode" network sociogram that depicts degrees of interconnectedness between cities (blue squares) and NLJ 250 firms (red circles).
Here are some quick interpretative guides:
- Cities. You'll note that lots of firms tend to have offices in New York, Washington, Chicago, L.A., and San Francisco. Hence, the network algorithm places these locations in the center. City nodes are scaled by the number of NLJ 250 HQ's in the city. New York has the most (35), followed by Chicago (19) and Washington (17). Why is London in the center? Because firms with offices in London overwhelming have offices in NYC and Washington. For some contrast, I added labels for some outer-periphery cities.
- Firms. Larger firms, depicted by larger circles, are clustered in the space between or around the important cities. In other words, large firms are integrally linked to large cities--very large firms all have a presence in most or all of the key cities. The outer periphery represents firms with more distinctive networks of less important cities. The farther apart the circles, the more dissimilar the firms' networks.
There is nothing truly new or novel in this graph (those will come later), though it does suggest that large firm practice in a larger city probably entails substantively different (probably more specialized) work than large firm practice in smaller markets. A lawyer contemplating where to begin his or her career might contemplate the glass ceiling imposed by taking a job at an outer-periphery location or firm. Among other modes of analysis, I will eventually generate 30 charts like the one above (from 1978 to 2007) to create an animated "flip book" showing how this constellation has changed over time.
Kudos to Stan Wasserman, Bernice Pescosolido, and Ann McCranie for putting on a terrific ICPSR workshop. Thanks to their class, I have five years worth of research ideas.