Complexity and Social Networks Blog

« May 3, 2006 | Main | May 8, 2006 »

5 May 2006

Putting the network into politics (or at least political science)

Despite my best efforts, there is relatively little use of network analysis within political science. This was highlighted by my experiences at the Sunbelt and the Midwest Political Science Association meetings (the second largest meetings in political science, I believe) the last few weeks. I think the overlap between the two conferences was… me. There is a small upsurge in network-related research in the field—I am guessing there were several times as many papers at MPSA using network ideas than in previous years. But I doubt that I would need more than two hands to count the number of network-related papers at MPSA, which is a pretty big conference.

This despite, I would argue, the obvious applicability of network ideas to political science. How natural to look at politics as a product of the network of connections among agents with differential resources, where both the resources and connections make a difference to outcomes of interest. One simply has to look at the money that lobbyists make to infer that social capital may be converted into political capital. However, there is a rather meager amount of research in this vein within political science—instead, such analyses have appeared in sociology (e.g., Laumann and Knoke’s The Organizational State). There are some exceptions—e.g., the papers I wrote with Dan Carpenter and Kevin Esterling using the Laumann-Knoke data; Young’s book on the social organization of Congress as well as Patterson’s work on the personal networks of state legislatures (also see recent paper by Jim Fowler looking at co-sponsorship networks in Congress). There has also been some efforts bringing network ideas into the flow of things through the international system—including my own work, as well as Fabrizio Gilardi and others. Most prominent within the field has been a small but important vein of research by Robert Huckfeldt, John Sprague, and colleagues in the spirit of the Columbia studies on social networks and political opinion. Interestingly (this is just an impression), the social capital boom, instigated in significant part by my colleague Robert Putnam—a political scientist— has been far less in political science than in other social science fields, e.g., sociology, organizational behavior, and (surprisingly) now in economics. I think, probably, now, within political science, the field within which network ideas are most in vogue is public administration, where scholars such as Larry O’Toole, Brint Milward, Ken Meier, John Scholz and others, have been writing about the inter-linking of government and non-government actors for a while.

[If people can think of other examples, please post.]

Why this lack of network related research? I don’t have a definitive answer; and because of the feudal nature of political science, one has to do a fiefdom by fiefdom analysis. In part, it reflects (within American politics) the organization of the field in significant part around institutions—Congress, the Presidency, etc. Further, when one looks at actors, it is typically assumed that they are purposive/rational and independent. Public opinion research (with the exception of Huckfeldt, etc) relies on the essential independence of observations—an assumption that social influence, etc, peels away. International relations has focused in part on the (neo)realist vs (neo) liberal debate, which reifies states and again often assumes away any particular structure connecting them. There have been studies (e.g., on the “democratic peace�) which have done dyadic analyses of the predictors of conflict—however, these studies (the last I looked at them) assumed an independence of dyads—whether A and B are at conflict is essentially independent of whether B and C are at conflict. This is an assumption that is clearly at odds with casual observation and something amenable to analysis with modern network-based statistical methods (e.g., p*).

So, looking for a generalization across fields, there seem to be paradigmatic foci at the institutional level and actor levels (usually utilizing assumptions of rationality regarding the latter), but a relative paucity of theorizing at levels between—which is where network analysis exists.

Posted by David Lazer at 8:16 AM