June 13, 2005
Congressional Committees and Partisanship
Posted by Arcane Gazebo at June 13, 2005 2:20 PM
Those of you who read the comments regularly may have seen references to Mason's work on network analysis of congressional committees. The paper (subscription required) appeared in the May 17th PNAS, and some mainstream coverage has appeared in the form of an AP article:
Although the Congress study is incomplete, some early findings have emerged from the labyrinth of line graphs it has already produced. One major trend: Since Republicans took over control of the House in 1994, the connection between membership on various committees has become more defined.
Also, during the 107th Congress in 2002, the mathematicians found great carry-over from membership on the Rules Committee and a Homeland Security panel established after the terrorist attacks of Sept. 11, 2001.
No mathematical formula is necessary to explain that phenomenon, explains Rules Committee spokeswoman Jo Maney. In the early days of the Homeland Security Committee, lawmakers most familiar with the rules -- disproportionately from the Rules Committee -- were assigned to get it off the ground.
Science Now had a good writeup, but it is subscription-only. It mentioned an interesting result that didn't appear in the AP report:
Porter's team, which reports its findings online this week in Proceedings of the National Academy of Sciences, also related the network connections between committees to the political positions of their members. To avoid making any prior political assumptions, the group used a different mathematical method called single value decomposition to analyze the roll call votes of each member during a session of Congress. The method pinpoints groups that voted similarly on many votes and assigns each member two numbers that enabled the researchers to rank them from least to most partisan. Combining the partisanship measures with the network map, the team found that not only is the Select Committee on Homeland Security one of the committees most tightly linked to another committee (House Rules), but it is also among the most partisan.
Maybe this helps explain why our domestic security policy is so focused on scaring people rather than taking actual effective measures... or maybe that would be true with any set of current congressmen.
If you want the paper sans subscription (because when it's for academic purposes, I can give it out to whomever I want), go to http://www.its.caltech.edu/~mason/papers/congress.pdf
Right now, I think the AP report was only picked up in Daytonia, so the chance of making major newspapers seems basically nil at this point. (I'm becoming such a media slut. It's ridiculous.)
One comment on the last AP paragraph (which may or may not have made it into that article---I don't remember): The key with the spokesperson's comment is that they _admit_ that they blew off the rules (or at least official guidelines or whatever) in forming select committees, so that comment actually works in our favor, because it confirms that something the mathematics suggested actually did happen. (The point of the article, by the way, is that we can use this perspective to find stuff without knowing any politics; a _lot_ more work will be needed to find things that are more subtle, although even in this case, the House didn't publicize the fact that they were breaking established norms.)
Oh, and if you look at the article, you won't notice any of my biases. I even used the word pentagon instead of pentagram in the appropriate spot...
Dude, I'm jealous. I definitely need to print this baby out and read it! (Probaby on the bus to and from campus, whee.) One can mix Applied Math with politics, and get it published! It's inspiring!
Well, there was also a crapload of work involved (a necessary component to getting it published, or at least in my willingness to submit it for publication). And you're welcome to the data if you want it. It took 6 weeks to convert some of it to a usable format, so there's no point in anybody else going through that part of it.