Partition Decouplingfor roll call dataScott PaulsDepartment of MathematicsDartmouth Collegescott.pauls@dartmouth.edu5th An...
Partition Decoupling for roll call dataThis is joint work with Greg Leibon, Dan Rockmore,and Robert Savell,...
Inference from roll call data Aye! Nay! Aye ...
A legislator is merely a bundle of votes. 5th Conference on Political Networks June ...
Motivational model 5th Conference on Political Networks June 13-16, 2012
Partition Decoupling Method (PDM) NETWORK ...
Comparisons Minority Random Poole- Poole- % of PDM: one PDM: % of ...
Example: 108th Senate“Conservative Republicans” Sessions, Kyl, Cornyn, Santorum, ...
Example: 88th Senate Party Civil RightsOuter shape: red=midwest, blue=northea...
Layer twoRegional identification dominates highestcorrelations (particularly in recent years).Clustering on the residual d...
Example: 108th SenateThree clusters of mixed party.Four sets of issues distinguish the clusters effectively:1. Infrastru...
Interaction of the two layers 5th Conference on Political Networks June 13-16, 2012
Interaction of the two layers 5th Conference on Political Networks June 13-16, 2012
Conclusions• PDM decomposition reveals multiple layers of structure associated to roll call voting.• Taken together, thes...
Distinguishing clusters: 108th Senate• Coarse picture: one dimensional ideology (“liberal/conservative”). ...
Distinguishing clusters 108th Senate• Coarse picture: one dimensional ideology (“liberal /conservative”). ...
Distinguishing clusters 108th Senate• Coarse picture: one dimensional ideology (“liberal conservative”). ...
Distinguishing clusters 108th Senate• Coarse picture: one dimensional ideology (“liberal/conservative”). ...
Distinguishing clusters 108th Senate• Coarse picture: one dimensional ideology (“liberal/conservative”). ...
Distinguishing clusters 108th Senate• Coarse picture: one dimensional ideology (“liberal/conservative”). ...
Distinguishing clusters 108th Senate• Coarse picture: one dimensional ideology (“liberal/conservative”). ...
I suspect that some of the issues raised by thereferees reflect disciplinary differences. For a paperlike this to be attra...
of 22

Political networks

My talk for the 2012 Political Networks conference on empirical mathematical models of ideology.
Published on: Mar 4, 2016
Source: www.slideshare.net


Transcripts - Political networks

  • 1. Partition Decouplingfor roll call dataScott PaulsDepartment of MathematicsDartmouth Collegescott.pauls@dartmouth.edu5th Annual Conference on Political NetworksJune 13-16, 2012University of Colorado at Boulderand the University of Denver.
  • 2. Partition Decoupling for roll call dataThis is joint work with Greg Leibon, Dan Rockmore,and Robert Savell, all from Dartmouth.http://arxiv.org/abs/1108.2805 5th Conference on Political Networks June 13-16, 2012
  • 3. Inference from roll call data Aye! Nay! Aye Aye 5th Conference on Political Networks June 13-16, 2012
  • 4. A legislator is merely a bundle of votes. 5th Conference on Political Networks June 13-16, 2012
  • 5. Motivational model 5th Conference on Political Networks June 13-16, 2012
  • 6. Partition Decoupling Method (PDM) NETWORK COMMUNITY LEARNING LOW DIMENSIONAL REPRESENTATION ITERATION 5th Conference on Political Networks June 13-16, 2012
  • 7. Comparisons Minority Random Poole- Poole- % of PDM: one PDM: % of model model Rosenthal: Rosenthal: residual layer two residual 1 dim. 2 dim. captured layer capturedHouse APRE 0 0.4561 0.534 0.593 13 0.839 0.856 11Percent 67.3 [72,88] 84.5 86.5 13 94.7 95.3 11correct(House)Senate 0 0.4834 0.476 0.563 17 0.809 0.822 7APREPercent 66.6 [70,90] 82.3 85.2 16 93.6 94.1 8correct(Senate) 5th Conference on Political Networks June 13-16, 2012
  • 8. Example: 108th Senate“Conservative Republicans” Sessions, Kyl, Cornyn, Santorum, “Moderate Republicans”: etc. e.g. Snowe, Chaffee, Frist, Lott, Brownback, Collins, Specter, etc. Hagel, etc. Fitzgerald, Gregg, McCain, Sununu, Warner “tax cuts”Zell Miller (D-GA) “Liberal Democrats”: e.g. Kennedy, Feingold, Boxer, Leahy, Reed, etc. “Conservative Democrats”: e.g. Pryor, Lincoln, Bayh, 5th Conference on Political Networks June Breaux, Landrieu, etc. 13-16, 2012
  • 9. Example: 88th Senate Party Civil RightsOuter shape: red=midwest, blue=northeast,green=south, 5th Conference on Political Networks June yellow=west black=southwest, 13-16, 2012
  • 10. Layer twoRegional identification dominates highestcorrelations (particularly in recent years).Clustering on the residual data provides a newpartition of network which is (often) completelydifferent than the first layer.In particular, clusters are not dominated byparty identification. 5th Conference on Political Networks June 13-16, 2012
  • 11. Example: 108th SenateThree clusters of mixed party.Four sets of issues distinguish the clusters effectively:1. Infrastructure: Three amendments (86, 214 and 230) to H.J. Res. 2, the Appropriations Bill, relating to infrastructure projects.2. Energy: Seven amendments (515, 843, 844, 851, 853, 856, 884 and 1386) to Senate Bill 14, a bill concerning the energy security of the United States. One amendment (272) to S. Con. Res. 23, relating to drilling in the Arctic National Wildlife Refuge.3. Homeland Security: Two amendments (515 and 3631) pertaining to Homeland Security.4. Trade: The passage of the US-Chile Free Trade AgreementThe first and second clusters are well separated by theEnergy votes, the first and third by Energy andInfrastructure votes and the second and third by oneenergy vote, Homeland Security and Trade votes. 5th Conference on Political Networks June 13-16, 2012
  • 12. Interaction of the two layers 5th Conference on Political Networks June 13-16, 2012
  • 13. Interaction of the two layers 5th Conference on Political Networks June 13-16, 2012
  • 14. Conclusions• PDM decomposition reveals multiple layers of structure associated to roll call voting.• Taken together, these form a mathematical description of ideology.• The coarse version of the first layer is close to the results of spatial models but even the first layer significantly outperforms spatial models with respect to standard metrics.• The use of multiple layers allows us to capture a more nuanced picture of ideology while still retaining the parsimony of the NOMINATE-type models.• Our dimensionality results confirm those of Poole-Rosenthal while simultaneously incorporating contradicting evidence (e.g. Heckman-Snyder) – the dimensions appear at different scales. 5th Conference on Political Networks June 13-16, 2012
  • 15. Distinguishing clusters: 108th Senate• Coarse picture: one dimensional ideology (“liberal/conservative”). Y N N N N N N N N Y Y N N N Y Y Y Y Y Y N Y/N Y/N Y/N Y/N N/Y N/Y N/Y Y Y Y Y Y N N Y Y Y N N Y N Democrats Republicans 5th Conference on Political Networks June 13-16, 2012
  • 16. Distinguishing clusters 108th Senate• Coarse picture: one dimensional ideology (“liberal /conservative”). Y N N N N N N An amendment to N N Y Y N N N an appropriations Y Y Y Y Y Y N bill which would eliminate tax cuts. Y/N Y/N Y/N Y/N N/Y N/Y N/Y Y Y Y Y Y N N Y Y Y N N Y N Democrats Republicans 5th Conference on Political Networks June 13-16, 2012
  • 17. Distinguishing clusters 108th Senate• Coarse picture: one dimensional ideology (“liberal conservative”). Y N N N N N N An amendment to N N Y Y N N N repeal authorities Y Y Y Y Y Y N and requirements for a base closure Y/N Y/N Y/N Y/N N/Y N/Y N/Y Y Y Y Y Y N N Y Y Y N N Y N Democrats Republicans 5th Conference on Political Networks June 13-16, 2012
  • 18. Distinguishing clusters 108th Senate• Coarse picture: one dimensional ideology (“liberal/conservative”). Three votes: 1. Sense of the Y N N N N N N Congress re: global AIDS funding N N Y Y N N N 2. Cloture: Safe, Accountable, Y Y Y Y Y Y N Flexible and Efficient Transportation Act of Y/N Y/N Y/N Y/N N/Y N/Y N/Y 2004 Y Y Y Y Y N N 3. Amendment to provide a Y Y Y N N Y N brownfields demonstration for qualified green/sustainable Democrats Republicans design projects 5th Conference on Political Networks June 13-16, 2012
  • 19. Distinguishing clusters 108th Senate• Coarse picture: one dimensional ideology (“liberal/conservative”). Two votes: 1. Extend Y N N N N N N Unemployment N N Y Y N N N Benefits Y Y Y Y Y Y N 2. Sense of the Senate re: Y/N Y/N Y/N Y/N N/Y N/Y N/Y imposition of an Y Y Y Y Y N N excise tax on tobacco Y Y Y N N Y N lawyer’s fees that exceed Democrats Republicans $20,000/hr 5th Conference on Political Networks June 13-16, 2012
  • 20. Distinguishing clusters 108th Senate• Coarse picture: one dimensional ideology (“liberal/conservative”). Amendment to Y N N N N N N protect US workers N N Y Y N N N from foreign Y Y Y Y Y Y N competition for performance of Y/N Y/N Y/N Y/N N/Y N/Y N/Y Federal and State Y Y Y Y Y N N contracts. Y Y Y N N Y N Democrats Republicans 5th Conference on Political Networks June 13-16, 2012
  • 21. Distinguishing clusters 108th Senate• Coarse picture: one dimensional ideology (“liberal/conservative”). Y N N N N N N N N Y Y N N N Amendment to vest sole jurisdiction Y Y Y Y Y Y N over Federal budget Y/N Y/N Y/N Y/N N/Y N/Y N/Y process in the Y Y Y Y Y N N Committee on the Budget Y Y Y N N Y N Democrats Republicans 5th Conference on Political Networks June 13-16, 2012
  • 22. I suspect that some of the issues raised by thereferees reflect disciplinary differences. For a paperlike this to be attractive to a political scienceaudience, it is necessary to motivate the methodtheoretically. This can be of the type "this model isbased on these behavioral axioms" (like theNominate and Clinton-Jackman-Rivers model) or"this method allows us to answer an important andpreviously unanswered question." Either of theseframes would make the paper of greater interest topolitical scientists. If the method doesnt providestand on its own substantively, then its notsomething that will have much traction in ourdiscipline, nor, in my mind, should it. 5th Conference on Political Networks June 13-16, 2012

Related Documents