|Home | Vita | Data & Code | Papers | Contact|
TEXT PREPROCESSING FOR UNSUPERVISED LEARNING: WHY IT MATTERS, WHEN IT MISLEADS, AND WHAT TO DO ABOUT IT (With Matthew Denny) Replication materials, including software here.
Political Analysis, forthcoming.
CLASSIFICATION ACCURACY AS A SUBSTANTIVE QUANTITY OF INTEREST: MEASURING POLARIZATION IN WESTMINSTER SYSTEMS (With Andrew Peterson) Online Appendix here. Replication materials here.
Political Analysis, vol 26(1): 120--128 (2018) .
Proceedings of the 2017 Conference on Empirical Methods in Natural Language Processing, 1558--1572
British Journal of Political Science, vol 48(2): 343--368.
Journal of Politics, 79(3): 903--920
Journal of Politics, 78(1): 120--136.
Journal of the Royal Statistical Society, Series A, 179(1): 56--57.
British Journal of Political Science, 46 (3): 567--589 (2016).
Political Analysis, 23 (2): 299-305 (2015)
I am the organizing contributor/editor of a special issue of Legislative Studies Quarterly on British Political Development. This special edition consists of six papers, on various aspects of UK legislative behavior in the nineteenth century. The table of contents is here. My introduction to the special issue can be found here:
My coauthored paper in the special issue can be found here:
Legislative Studies Quarterly, Vol 9, No 4.
Quarterly Journal of Political Science, 9, no. 3 (2014): 337--370
American Journal of Political Science, 58, 873--887
Journal of Conflict Resolution, Vol 52 (2), 207--238 (2015)
Brookings Papers on Economic Activity, Fall 2012
American Journal of Political Science, Vol 56, No 2, 413--432
American Journal of Political Science, Vol 56, No 1, 84--97. Replication materials are here
Electoral Studies, Vol 31, No 1, 212--221
American Political Science Review, Vol 105, No 2, 337--358.
Journal of the American Statistical Association Vol. 105, No. 489: 71–83.
Journal of the American Statistical AssociationVol. 105, No. 490: 447–45
The American Statistician, 61:4, 2007
Political Analysis, 15(4), 2007
PS: Political Science and Politics, 41(2), 2007
Government and Opposition, 41(4), 561--568.
Political Analysis 15 (1) Winter, 2007.
The Political Methodologist 14 (2)
NONE OF THE ABOVE: THE UK HOUSE OF COMMONS VOTES ON REFORMING THE HOUSE OF
LORDS, FEBRUARY 2003 (with Iain McLean and Meg
`Power laws' suggest that events of a large magnitude will be rare, whilst small events will be much more common, and that a simple mathematical law relates `severity' with frequency. We find that a wide variety of phenomena in political science are power law distributed. These empirical regularities are both unexpected and unexplained. More work on a general explanatory theory for these patterns is desirable.
Note: this (version of the) paper was cited in Common ecology quantifies human insurgency appearing in Nature, 17 December 2009.
REBELS WITH A CAUSE? LEGISLATIVE ACTIVITY AND THE PERSONAL VOTE IN BRITAIN, 1997--2005 (available on request)
Does a Member of the British Parliament's voting record have any effect on their constituency electoral performance? Scholars have assumed not, else they have tested the proposition with an extremely limited number of roll calls. Congruent with public opinion findings we contend that, paradoxically, voters conditionally reward both 'party unity' and 'independent mindedness' in their elected representatives. Using novel non-parametric 'random forest' classification procedures, and a new data set recording behavior on over 2000 roll calls from 1997--2001 and 2001--2005, along with commensurate constituency controls, we thus show that MPs' popularity is indeed effected by their legislative activity in small but significant ways. In particular, government-party voters demand unity on votes that are key parts of the government's programmatic agenda, while welcoming more 'maverick' behavior on less important issues.
The measurement of power, even in structured settings like legislatures, has proved elusive. We discuss the problems with traditional, a priori voting indices approaches and suggest a data-driven, actor-based, (logistic regression) method that is straightforward to implement. This treatment is consistent with systematic theoretical models and discussions of power, and formally allows the separation of 'power' from its causes. To illustrate the strengths of this new technique, we apply the model to the 108th United States Senate. We find that institutional, ideological, personal and geographic variables all influence senators' power.