Books

  • Representational Style in Congress: What Legislators Say and Why It Matters In Press. Cambridge University Press. 
  • The Impression of Influence: How Legislator Communication and Government Spending Cultivate a Personal Vote. Under Review. With Sean Westwood and Solomon Messing. 

Publications

Congressmen in Exile: The Politics and Consequences of Involuntary Committee Removal (with Eleanor Neff Powell) Forthcoming, The Journal of Politics

  • Abstract: We show how preferred committee assignments act as an electoral subsidy for members of Congress---empowering representatives' legislative careers. When holding the assignments, legislators are able to focus more on legislative activity in Washington. But when the subsidy is removed, legislators are forced to direct attention to the district. To test our theory of legislative subsidy, we exploit committee \emph{exile}---the involuntary removal of committee members after a party loses a sizable number of seats, and the losses are unevenly distributed across committees. Legislators are selected for exile using members' rank on the committee (committee seniority), causing exiled and remaining legislators to appear strikingly similar---motivating our estimation strategy built around exile. We show that exile has only limited electoral consequences, but this is partly due to compensatory efforts---exiled legislators shift attention away from Washington and towards the district. Exiled legislators raise and spend more money for reelection, author less legislation, are absent for more days of voting, and vote with their party less often.  
  • Supplementary Information

How Words and Money Cultivate a Personal Vote: The Effect of Legislator Credit Claiming on Constituent Credit Allocation (with Solomon Messing and Sean Westwood) American Political Science Review, 2012. 106 (4), 703-719.

  • Abstract: Particularistic spending, a large literature argues, builds support for incumbents. This literature equates money spent in the district with the credit constituents allocate. Yet, constituents lack the necessary information and motivation to allocate credit this way. We use extensive observational and experimental evidence to show how legislators' credit claiming messages---and not just money spent in the district---affect how constituents allocate credit. Legislators use credit claiming messages to influence the expenditures they receive credit for and to affect how closely they are associated with spending in the district. Constituents are responsive to credit claiming messages--they build more support than other non-partisan messages. But contrary to expectations from other studies, constituents are more responsive to the total number of messages sent rather than the amount claimed. Our results have broad implications for political representation and the study of U.S. Congressional elections. 
  • Supplementary Information
  • Presentation Slides

Appropriators Not Position Takers: The Distorting Effects of Electoral Incentives on Congressional Representation Forthcoming, American Journal of Political Science

  • Abstract:  Congressional districts create two levels of representation. Studies of representation focus on a disaggregated level: the electoral connection between representatives and constituents. But there is a collective level of representation---the result of aggregating across representatives. This paper uses new measures of home styles to demonstrate that responsiveness to constituents can have negative consequences for collective representation. The electoral connection causes marginal representatives---legislators with districts composed of the other party's partisans--to emphasize appropriations in their home styles. But it causes aligned representatives--those with districts filled with co-partisans--to build their home styles around position taking. Aggregated across representatives, this results in an artificial polarization in stated party positions: aligned representatives, who tend to be ideologically extreme, dominate policy debates. The logic and evidence in this paper provides an explanation for the apparent rise in vitriolic debate and the new measures facilitate a literature on home styles. 
  • Supplementary Information
  • Presentation Slides
  • (Previously Appropriators Not Statesmen: The Distorting Effects of Electoral Incentives on Congressional Representation)

Text as Data: The Promise and Pitfalls of Automatic Content Analysis Methods for Political Texts (with Brandon Stewart) Forthcoming, Political Analysis

  • Abstract: Politics and political conflict often occur in the written and spoken word. Scholars have long recognized this, but the massive costs of analyzing even moderately sized collections of texts have hindered their use in political science research. Here lies the promise of automated text analysis: it substantially reduces the costs of analyzing large collections of text. We provide a guide to this exciting new area of research and show how, in many instances, the methods have already obtained part of their promise. But there are pitfalls to using automated methods--they are no substitute for careful thought and close reading and require extensive and problem specific validation. We survey a wide range of new methods, provide guidance on how to validate the output of the models, and clarify misconceptions and errors in the literature. To conclude, we argue that for automated text methods to become a standard tool for political scientists, methodologists must contribute new methods and new methods of validation.

Elevated Threat-Levels and Decreased Expectations: How Democracy Handles Terrorist Threats (with Tabitha Bonilla) Forthcoming, Poetics (Special Issue on Topic Models in Social Science) 

  • Abstract : A persistent concern in democracies is that terror threats make the public willing to restrict freedoms for increased safety. But a large literature has struggled to determine how terrorist threats affect the public's policy preferences. To more credibly estimate the effects of terror threats, we exploit elevations of the U.S. government's color coded alert system. Using this design, a new statistical model for texts, and a new collection of news stories, we show that media outlets allocate substantially more attention to terrorism after an alert. This sudden shift in media attention, though, has only limited effects on the public. The terror alerts raise the public's perceived likelihood of a terror attack, but opinion about President Bush's job performance, preferences for foreign intervention, or willingness to restrict civil liberties changes little in response to the alerts. Rather, the only consistent result is decreased economic expectations---consistent with the strong economic downturn after the 9/11 attacks and the types of stories published after the terror alerts are elevated. Terror alerts, then, did not exercise direct influence on the public's policy preferences. Instead, they changed the topic of conversation. 

Evaluating Model Performance in Fictitious Prediction Problems.  Discussion of ``Multinomial Inverse Regression for Text Analysis" by Matthew Taddy.  To Appear in the Journal of the American Statistical Association.  

General Purpose Computer-Assisted Clustering and Conceptualization (With Gary King)  Proceedings of the National Academy of Sciences, 2011.  108(7), 2643-2650.

  • Abstract: We develop a computer-assisted method for the discovery of insightful conceptualizations, in the form of clusterings of input objects. Each of the numerous fully automated methods of cluster analysis proposed in statistics, computer science, and biology optimize a different objective function. Almost all are well defined, but how to determine ex ante which one, if any, will partition a given set of objects in an “insightful” or “useful” way for a given user is unknown and difficult, if not logically impossible. We develop a metric space of partitions from all existing cluster analysis methods applied to a given data set (along with millions of other solutions we add based on combinations of existing clusterings), and enable a user to explore and interact with it, and quickly reveal or prompt useful or insightful conceptualizations. In addition, although uncommon in unsupervised learning problems, we offer and implement evaluation designs that make our computer-assisted approach vulnerable to being proven suboptimal in specific data types. We demonstrate that our approach facilitates more efficient and insightful discovery of useful information than either expert human coders or many existing fully automated methods. We (will) make available an easy-to-use software package that implements all our suggestions.
  • Slides from Polmeth
  • Selected media coverage: NPR "On the Media" http://ow.ly/4FhmV ; NPR "Here and Now" http://j.mp/eGBUQy ; NewsMax http://ow.ly/4BeAG ; Wisconsin Public Radio "Joy Cardin" http://j.mp/eUAZ2C ; Forbes Magazine http://t.co/qiaU0nR ; New York Times http://www.nytimes.com/2011/04/10/opinion/10kristof.html?_r=2&hp ; Washington Post http://j.mp/etHP8D ; PBS (Top Ten Data Mining Ideas of 2011) http://www.pbs.org/idealab/2012/01/the-top-10-data-mining-links-of-2011006.html

An Introduction to Bayesian Inference via Variational Approximations Political Analysis, 2011. 19 (1) 32-47

  • Abstract: Markov Chain Monte Carlo (MCMC) methods have facilitated an explosion of interest in Bayesian methods. MCMC is an incredibly useful and important tool, but can face difficulties when used to estimate complex posteriors or models applied to large data sets. In this paper I show how a recently developed tool in computer science for fitting Bayesian models, variational approximations, can be used to facilitate the application of Bayesian models to political science data. Variational approximations are often much faster than MCMC for fully Bayesian inference and in some instances facilitates the estimation of models that would be otherwise impossible to estimate. Variational approximations have guaranteed, fast, easily assessed convergence and provide accurate estimates of the expected value of the posterior (given sufficient sample size), but have some limitations. Therefore, variational approximations are best suited to problems when fully Bayesian inference would otherwise be impossible. Through a series of examples, I demonstrate how variational approximations are useful for a variety of political science research. This includes models to describe legislative voting blocs and statistical models for political texts. The code that implements the models in this paper is available in the supplementary material.  
  • Included in Political Analysis virtual issue on Big Data in Political Science
  • Supplemental Notes

Approval Regulation and Endogenous Provision of Confidence: Theory and Analogies to Licensing, Safety, and Financial Regulation (with Dan Carpenter and Eric Lomazoff)  Regulation and Governance, 2010. 4(4) 383-407

  • Abstract: Recent years have witnessed renewed political and scholarly interest in consumer protection regulation; in the financial sector, most notably, legal scholars have proposed that safety regulation should govern financial markets (Warren 2007, 2008). In this paper, we ask whether the effects of safety regulation go beyond safety and might affect consumers’ beliefs about the distribution of products they can use. We model approval regulation, where a government regulator must approve the market entry of a product based upon observable, unbiased and non-anticipable experiments. We show that even if regulator and firm disagree only about quality standards, the disagreement induces the firm to provide more information about its product than it would in the absence of regulation. Put differently, purely first-order disagreements in regulation will generate second-order consequences (more certainty about product quality).  These second-order consequences of regulation are sufficient to generate first-order effects among end-users (more consumption of superior products), even when users are risk-neutral. In other words, even if approval regulation produces little or no improvement in safety or quality, it still aggregates information useful to ‘downstream’ product users; these users will exhibit higher consumption and will more readily switch to superior products. In contrast with libertarian analyses of entry regulation and licensure, the model predicts that entry restrictions may be associated with greater product or service utilization (consumption) (Law 2003; Law and Marks 2009), as well as greater price sensitivity among consumers. Because contemporary cost-benefit analyses ignore these second-order effects, they are unlikely to capture the possible confidence effects of approval regulation.

A Bayesian Hierarchical Topic Model for Political Texts: Measuring Expressed Agendas in Senate Press Releases  Political Analysis, 2010. 18(1) 1-35.

  • Abstract: Political scientists lack methods to efficiently measure the priorities political actors emphasize in statements. To address this limitation, I introduce a statistical model that attends to the structure of political rhetoric when measuring expressed priorities: statements are naturally organized by author. The expressed agenda model exploits this structure to simultaneously estimate the topics in the texts, as well as the attention political actors allocate to the estimated topics. I apply the method to a collection of over 24,000 press releases from senators from 2007, which I demonstrate is an ideal medium to measure how senators explain their work in Washington to constituents. A set of examples validates the estimated priorities and demonstrates their usefulness for testing theories of how members of Congress communicate with constituents. The statistical model and its extensions will be made available in a forthcoming free software package for the R computing language.
  • Awarded 2010 Robert H Durr Award for the best paper presented at 2009 Midwest Political Science Association meeting applying quantitative methods to a substantive problem.
  • Awarded 2011 Warren Miller Prize for the best paper published in Political Analysis in 2010.
  • Included in Political Analysis virtual issue on Bayesian methods in Political Science
  • Supplemental Notes
  • Slides from QSSI New Faces II

Working Papers

Are Close Elections Random? (With Brian Feinstein, Eitan Hersh, and Dan Carpenter) Under Review

  • Abstract: Elections with small margins of victory represent an important form of democratic competition and, increasingly, an opportunity for causal inference. When scholars use close elections for examining competition or for causal inference, they impose assumptions about the politics of close contests: campaigns are unable to systematically determine the outcome. This paper calls into question this model and introduces a new model that accounts for strategic campaign behavior. We draw upon the intuition that elections that are expected to be close attract greater campaign efforts before the election and invite legal challenges and fraud after the election. Our theoretical models predict systematic differences between winners and losers in extremely close elections. We test our predictions using all House elections from 1880-2008, finding that structurally advantaged candidates are more likely to win close elections. But the structural advantages that predict winners shift over time: from 1880 to the 1960's, candidates from strong parties are systematically more likely to win close contests, but the advantage dissipates in more recent contests. After the 1940's, incumbent candidates are much more likely to win close elections. Our findings suggest a new research agenda on the systematic determination of close contests.
  • Supplementary Information

A Class of Bayesian Semiparametric Cluster-Topic Models for Political Texts (with Rachel Shorey, Hanna Wallach, and Frances Zlotnik)

  • Abstract: We introduce a new Bayesian cluster-topic model for political texts and a novel methodology for model selection in statistical models for text. We first develop a fully parametric model that simultaneously estimates the topics articulated in texts as well as partitions of documents based on their attention to topics. In the context of this model we then outline a new and general methodology for model selection that combines the strengths of both statistical and substance based approaches. First, we extend our fully parametric model to a group of semiparametric models using three nonparametric priors: the Dirichlet process prior, the Pitman Yor process prior, and the uniform process prior. These models \emph{estimate} the number of clusters used in the model, but we show the estimated number of clusters and the number of documents per cluster are heavily model dependent. Therefore, while the statistical guidance in selecting models is essential, human judgment is necessary to make a final model selection. To use human input to make this final selection, we introduce a battery of experimental methods that provide carefully-elicited subject expert evaluations of the models and explain how to extend these methods to new models and data sets. We implement our models and experiments on a new collection of over 19,000 House press releases from 2010. Using our results, we show that ideologically extreme representatives dominate policy debates---a finding with widespread consequences for policy deliberation and lawmaking.

The Downside of Deadlines (with Dan Carpenter). Under Revision 

  • Abstract: Deadlines are ubiquitous institutions in government decision making, constraining both agencies and courts. Yet these institutions are almost entirely ignored in formal models in instituional political science. We analyze deadlines as exogenously imposed instiutions upon a government decision maker, as a means of exercising control over the bureaucratic agent. Our formal model demonstrates how deadlines are successful at lowering the time to approval during regulation. But, our analysis also illuminates the downside of deadlines. The effect of deadlines on regulatory behavior is highly non-linear, making imposition of deadlines a difficult taks for even highly rational agents. Further, our formal model predicts that deadlines will increase the variance in the approval time distribution under a large set of conditions and that deadlines will increase the error rate of regulatory behavior, often in exponential fashion. Our formal analysis explains an expanding empirical literature about the effects of deadlines and suggests some of the limits of deadlines as an effective tool of control over policymaking and bureaucratic agents.
  • Popular Press Reaction: Boston Globe , Washington Post
  • Robert Wood Johnson Foundation Working Paper # 41