I spent last week at an ICPSR workshop on network analysis methods in Bloomington, Indiana, and while I wouldn’t say I’m exactly an expert now on network analysis, I did learn to do a few interesting things with my human security data.
For example, below is a visual representation of the “human security issue agenda” as represented by 88 websites of organizations in the “human security network” circa May 2008. I identified the organizations in that network through a hyperlink analysis using the tool Issuecrawler, and then we downloaded their mission-statements and “issue lists” and tagged each document with as many different “issues” as we could find in the text data. When two issues occur in the same document, I consider that a “tie” between two “issues.” So on the map, links between issues represent co-occurences in the text.
As you can see, lots of organizations do lots of issues, so when you first look at the graph the ties are so many it’s just a big mess. But if you drop out half the ties, you end up only with the most densely connected issues, and then it starts to look pretty interesting:
What’s interesting is that the issues cluster based on how often they co-occur a lot: some issues cluster together more than others. You notice that the human security network is really made of a few separate issue clusters – a cluster around peace or conflict-prevention and resolution; one around conflict mitigation or humanitarian affairs; one around repressive or violent practices (what human security specialists like to call the “freedom from fear” agenda) and another dealing more with economic and social rights, which is adjacent to a cluster dealing with development and poverty reduction, and another dealing with environmental security and health, both of which connect to weapons issues which connect to the security sector / arms controls organizations, which connect again to the “peace and security”/conflict prevention folks.
Lots of people have written abstractly about what “human security” means, but as far as I know this mine is the first analysis that actually maps the term empirically. If you add in a few more of the links (using a “cut-point” of .40 instead of .50 for example) you can see these clusters emerge in somewhat sharper relief, but also see them beginning to connect together.
You’ll notice that these clusters are linked to each other to a greater or lesser extent. For example, there are a lot of connections between “human rights” and “humanitarian affairs” and relatively fewer connections between the “arms control” cluster and the “economic and social rights” cluster. There’s also a distinction between issues associated with “arms control” and those associated with “disarmament” – because everyone knows the arms control folks are manly realists and the disarmament folks are peace-loving hippie feminists.
Finally, there are some interesting cases where an issue doesn’t show up where you might imagine it would in the network. “Depleted uranium weapons,” for example, is mentioned in connection with health and the environment (and is most closely associated with “disease” issues like cancer) but is not on the agenda of organizations concerned with limiting the use of weapons in warfare (like landmines and chemical weapons). “Sovereignty,” interestingly, is most often mentioned not as a counterpoint to military intervention in humanitarian crises (as the literature would suggest) but in regards to natural resource disputes (or, shown here, “minerals”).
What does it all mean? For my project, I’m interested in what gets on the agenda and what doesn’t, and so how agenda space is carved up within a network is a really interesting piece of that question.