Friday, October 29, 2004

Meta-Organizational Network Analysis

A couple weeks ago I attended an excellent seminar entitled "Realizing the Value from Social Networks," organized by Patti Anklam and Nat Welch under the auspices of KM Cluster.

The speakers were quite informative and stimulating. I especially enjoyed hearing about leader-to-leader networks from Jamie Millar at Tapestry Networks, and trust-based leadership from Rob Galford. Bill Ives and Kathleen Gilroy also had excellent advice on blog-based networking, which I have already found fruitful through this very forum.

During the day, I had the good fortune to sit and compare notes with Marion Kane and Roberto Cremonini of the Barr Foundation. Their new corporate slogan is "Using knowledge, networks, and funding to build a better Boston for all." The idea is to help nonprofit organizations collaborate more effectively at the inter-organizational level.

Talking with Marion and Roberto reminded me of a paper "Network Measures of Social Capital" that I mentioned recently but didn't finish discussing. In the paper, authors Steve Borgatti, Candace Jones, and Martin Everett neatly categorize social capital into four categories based on two criteria: Is the actor being considered an individual or group? And is the focus of analysis internal actor dynamics or external collaboration/competition?

The two most-studied categories in this scheme are (1) external competition among individual actors and (2) internal dynamics within a group. These may sound like two ways of saying the same thing, but as the authors point out so well, the issues come up differently in each case.

Outside the realm of typical SNA research, but certainly well studied, is internal dynamics of an individual, otherwise known as psychology.

Of the four categories, that leaves one remaining: external collaboration/competition among groups. This quadrant has certainly received plenty of attention under different guises (e.g., international relations and corporate competition come to mind). But in hard-nosed scientific terms, I would venture to say that this quadrant is the least understood of the four.

The authors describe several techniques for measuring social capital in this context ("external measures for collective actors"). Both techniques leave plenty of room for improvement.

One approach is to use standard measures of social capital. Since these techniques measure relations among many individuals, not among multiple groups, we have to represent each group as a single individual in order to use these techniques. This is very neat and tidy, but by collapsing one group to a single node we lose track of differences between individuals in that group. Do you care if the external relationships of your team all belong to one person, or are distributed among several individuals? If so, then you are losing something here.

Some researchers embrace this model fully. Some time ago I enjoyed meeting professors N. Venkatraman and Bala Iyer at Boston University's Systems Research Center. They are studying network-centric business strategy, uncovering intriguing trends based on complex networks of simple business-to-business relationships.

Another approach described by Borgatti et al is to use 2-mode networks. [Note: Explaining 2-mode networks is beyond this post, and there are no good links I can find.] Unlike the previous technique, these networks do represent all groups and all individuals. However, 2-mode networks only represent very limited kinds of relationships. Do you care about relationships between individuals other than "share membership in the same group"? For example, perhaps you also care who knows whom? Then you are losing a lot with a 2-mode network.

I'd say that the field of SNA has just scratched the surface in studying networks of organizations.

No comments: