Having imbibed a great deal of Pierre Bourdieu during my graduate study in Sociology, I have a reflexivity reflex that some might regard as excessive. During my two forays into the AALS meat market, I couldn't help regarding the experience as ethnographic fieldwork for a study of legal academia (a mental whirlpool that was not at all helpful during the interview process). My interest in homo academicus juridicus as an object of study has not abated since I joined the tribe.
This summer, in preparation for a SEALS conference panel on academic legal blogs, I began to pursue this line of research in earnest, examining network ties in the academic legal blogosphere. The architecture of the web, based on links among sites, lends itself readily to network analysis. Using a crawler, it is a fairly simple matter to collect all the links among a defined set of sites, which can then be fed into a program for data analysis.
For a first glance, I started with Dan Solove's most recent "Law Professor Blogger Census" (August 2007). I updated that list by eliminating any inactive blogs (i.e. those that either have vanished from the web or had no posts during 2009), and adding a few that I happened to be aware of that began since the August 2007 census. I then collected all the links among these sites, using SocSciBot (a free and relatively simple to use program developed by Michael Thelwall, a professor of information science at University of Wolverhampton in England). I then used Pajek (a free network analysis program developed by Professors Vladimir Batagelj and Andrej Mrvar at the University of Ljubljana in Slovenia) to create network graphs.
Of the 150 blogs in the updated Census set, there were 35 with 10 or more inbound links from others in the set. These are plotted on a graph (click to enlarge), drawn (in Pajek) using an algorithm that arranges the nodes such that sites appear closer to one another the more links they share, and those appearing closer to the center of the graph are more "central" to the network (in the sense of having more links to and from other nodes in the graph).
Inbound links are of interest because, like citations in legal writing, they indicate some type of "authoritative" status on the part of the receiving site. Unlike site visits -- which reflect a site's influence or status among the general blog-reading public -- in-links from other law professor blogs reflect influence or status more specifically within the legal academy (or at least the portion thereof that writes and reads law professor blogs). Thus, several of the 35 law professor blogs with the greatest overall readership (based on data provided by Paul Caron) are absent from this graph, and some of those that are included appear relatively peripheral to this network, while some of the sites near the center of this network appear to have smaller overall readership (though some of these are omitted from Caron's figures because they lack public site meters).
As a further way of exploring relationships among these site, I created a multi-dimensional scale graph (using SPSS to analyze data exported from Pajek):
In this two-dimensional model, it is possible to distinguish among four clusters (one of which, marked in green here, might further be broken into sub-clusters), identified in the following table (with the numbers in parentheses corresponding to the "variable" numbers in the MDS graph):
The two dimensions in the graph are derived statistically from the network data. They represent some underlying factors by which the various sites resemble or differ from one another. Typically, in an MDS analysis, the researcher will seek to define the dimensions inductively. I have not done so here, as there do not appear to be any obvious characteristics (such as subject-matter, institutional affiliation, or political ideology) that might distinguish among the clusters. I would certainly welcome any thoughts readers might have on that score.
In addition to linking patterns across sites, I plan to collect and analyze data on guest bloggers to see whether there are any interesting patterns. I have also thought about doing a similar analysis of law student and legal practitioner blogs. The aim, in each case, is to shed further light on the social structure of these segments of the legal profession.
By way of illustration: A researcher interested in consumer perceptions of automobile brands might collect data using a survey that presents pairs of brands and asks respondents to indicate, for each pair, whether the brands are similar or dissimilar. After creating an MDS graph from this data, the research would then attempt inductively, based on how the brands cluster in the graph, to identify the characteristics according to which consumers differentiate among brands -- perhaps style and price, for example.
 I hope that Faculty Lounge readers will charitably view my stint here as an outlier and not draw any adverse inferences about the authority and status of this most valuable site!