More Than a Feeling: Personality, Polarization, and the Transformation of the US Congress

The University of Chicago Press, Chicago 60637

The University of Chicago Press, Ltd., London

© 2017 by The University of Chicago

All rights reserved. No part of this book may be used or reproduced in any manner whatsoever without written permission, except in the case of brief quotations in critical articles and reviews. For more information, contact the University of Chicago Press, 1427 E. 60th St., Chicago, IL 60637.

Published 2017.

Printed in the United States of America

26 25 24 23 22 21 20 19 18 17       1 2 3 4 5

ISBN-13: 978-0-226-45584-6 (cloth)

ISBN-13: 978-0-226-45598-3 (paper)

ISBN-13: 978-0-226-45603-4 (e-book)

DOI: 10.7208/chicago/9780226456034.001.0001

LCCN: 2016048672

This paper meets the requirements of ANSI/NISO Z39.48-1992 (Permanence of Paper).

More Than a Feeling

Personality, Polarization, and the Transformation of the US Congress

ADAM J. RAMEY, JONATHAN D. KLINGLER, GARY E. HOLLIBAUGH JR.

THE UNIVERSITY OF CHICAGO PRESS

CHICAGO AND LONDON

FOR THE MANY PERSONALITIES OF THE UNITED STATES CONGRESS

Contents

List of Figures

List of Tables

Acknowledgments

PART I. Foundations

CHAPTER 1. Introduction

1.1. A Tale of Two Senators: Chuck and Roy Disagree on the Shutdown

1.2. Traits and Elite Behavior in Institutions

1.2.1. Translating Individual Differences into the Language of Institutions

1.3. The Elite Behavior in Institutions Agenda and Plan of the Book

CHAPTER 2. Modeling Individual Differences: Translating Personality Traits into Mathematical Parameters

2.1. The Five-Factor Model

2.1.1. The Lexical and Questionnaire Schools of Thought

2.1.2. Causal Foundations and Stability in the Big Five

2.2. Challenges to the Five-Factor Model

2.3. Personality and Political Science

2.4. Modeling Personality

2.4.1. Defending Models of Personality

2.4.2. Parameterizing Core Cognitive Constraints

2.4.3. Measuring Personality-Based Cognitive Constraints

2.5. The Big Five Traits

2.5.1. Openness (to Experience)

2.5.2. Conscientiousness

2.5.3. Extraversion

2.5.4. Agreeableness

2.5.5. Neuroticism

2.6. A Framework for Political Choice

2.7. Considerations for Strategic Interactions

2.8. Modeling Individual Differences: Conclusion

CHAPTER 3. Read My Lips: Measuring Personality Through Legislative Speech

3.1. Limitations of Existing Approaches for Elected Officials

3.2. Using Text to Measure Personality Traits

3.3. Measuring Personality: From Speeches to Scores

3.4. Validity of the Estimates

3.4.1. Strategic Misrepresentation and Authorship Concerns

3.4.2. Face Validity

3.5. Read My Lips: Conclusion

3.6. Appendix

PART II. Revisiting the Textbook Congress

CHAPTER 4. Securing Reelection: Deterrence and Disbursements

4.1. Who Attracts Quality Challengers?

4.2. Who Spends?

4.3. Individual Differences and Seeking Reelection: Conclusion

CHAPTER 5. Committee Assignments

5.1. Congressional Committees and Core Cognitive Constraints

5.2. Plum Assignments

5.3. Becoming Chair

5.4. Committee Assignments: Conclusion

CHAPTER 6. Proposing and Passing Legislation

6.1. Personality, Proposals, and Passage

6.2. Putting Bills on the Agenda

6.3. Workhorses and Show Horses

6.4. Predicting Legislative Success

6.5. Proposing and Passing Legislation: Conclusion

CHAPTER 7. Cooperation, Obstruction, and Party Discipline: Shifting Norms in the US Congress

7.1. Rebellion, Obstruction, and Polarization

7.2. Party Brands, Loyalty, and the Big Five

7.3. Bucking the Party: Working Across Party Lines

7.4. Holding the Floor: Filibustering and Obstruction

7.5. Norms and the Shattering Thereof: Conclusion

CHAPTER 8. Media Presence and Home Style

8.1. Who Tweets?

8.2. Press Releases

8.3. Media Usage: Conclusion

CHAPTER 9. Moving On

9.1. Moving On or Moving Out?

9.2. Lame Ducks and the Shadow of Irrelevance

9.3. Moving On: Conclusion

9.4. Appendix: A Model of Legislative Voting

PART III. Bringing It All Together

CHAPTER 10. More than a Conclusion: Personality, Politics, and Polarization

10.1. Personality and the Congressional Life Cycle

10.2. Personality and Congress as an Institution

10.3. Personality and the Future Study of Elites and Institutions

Notes

Bibliography

Index

List of Figures

1.1. Estimates of Congressional Ideology

1.2. Senator Grassley (R-IA) Tweet Example

3.1. LIWC Comparison of Bon Jovi and Nirvana Songs

3.2. Comparing LIWC (2001) Usage between the Pennebaker Corpus and Floor Speeches

3.3. Senate Scores over Time (Selected Members)

3.4. Word Count and Precision

4.1. Predicted Probabilities of Quality Challenger Entry

4.2. Predicted Campaign Disbursements

5.1. Personality and Plum Committee Assignments

5.2. Personality and Committee Chair Assignments

6.1. Personality and Bill Proposals

6.2. Personality and Ceremonial Bill Proposals

6.3. Conscientiousness and Legislative Effectiveness

7.1. Personality and Bipartisan Cosponsorship

7.2. Predicted Rates of Supporting Cloture

8.1. Twitter Adoption by House Members (2007–2010)

8.2. Predicting the Number of Days to Adoption of Twitter

8.3. Press Release Issue Focus and Agreeableness

9.1. Personality and Running for Election

9.2. Personality and Other Electoral Decisions

9.3. Personality and Lame Duck Absenteeism

10.1. Personality and Polarization in the House and Senate over Time

10.2. Personality and Tenure in the House and Senate over Time

List of Tables

2.1. Defining Terms for the Big Five

2.2. Core Cognitive Constraints

3.1. OLS Models of Personality and House ADA Score (1996–2008)

3.2. LIWC (2001) Categories

3.3. MRCPD Categories

4.1. Strategic Probit Model of Candidate Competition in US House Races, 1996–2012

4.2. Transfers from Campaign Committees to Other Candidates and Party Organizations, 1996–2012

4.3. Linear Regression of Logged Campaign Disbursements

5.1. Logistic Regression Models of Personality and Plum Committee Assignments

5.2. Logistic Regression Models of Personality and Chair Assignments

6.1. Negative Binomial Models of Personality and Bill Proposals

6.2. Binomial Regression Models of Personality and Ceremonial Bill Proposals

6.3. Tobit Models of Legislative Effectiveness

7.1. Predicting Loyalty—Party Votes, 104th–109th Congresses

7.2. Binomial Regression Models of Bipartisan Cosponsorship

7.3. Predicting Minority Party Support for Cloture (104th–112th Congresses)

8.1. Who Tweets and How Often (2007–2010)?

8.2. Predicting Substantive vs. Credit-Claiming Behavior (2005–2007)

9.1. Multinomial Logit Regression Model of Personality and Electoral Decisions

9.2. Balance Statistics

9.3. Binomial Regression Models of Personality and Lame Duck Absenteeism

Acknowledgments

It is often the case in science that the best and most novel avenues of inquiry arise by accident. This project counts itself among the works (of all qualities) that have emerged in this venerable tradition. While completing another project, two of us (Ramey and Klingler) were sipping Turkish coffee in Toulouse, France, on a fine autumn day. Klingler was conveying with great excitement his recent forays into the voluminous and fascinating literature on personality and decision-making. Ramey was equally excited about his own recent explorations in the area of quantitative text analysis. In the midst of this exchange, an idea was born from a simple question: could we measure theoretically relevant personality traits of individuals using text? We gambled that the answer was yes. Ramey and Klingler then set up a Skype call with Hollibaugh to see if (a) he thought they were crazy, and (b) if he was willing to forgo another joint project and instead pursue this idea. While he was not (and still is not) fully willing to give Ramey and Klingler a pass on the insanity question, Hollibaugh nevertheless rolled the dice on the project.

In the years since, we have worked tirelessly toward a monumental goal—bridging the diverse literatures on personality, political institutions, and quantitative text analysis. Though much of that time was spent penning this manuscript, a nontrivial time was spent evangelizing our new set of ideas. Both tasks have been challenging, and without the support of those closest to us and this enterprise the final product would not have been possible. It is thus fitting that we begin by acknowledging those who have played significant roles in the development, growth, and maturation of this project.

Much of this book was written in unlikely locations. When together, we rarely sat in an office, typing away. Rather, much of our time fleshing out and executing the ideas expressed in this project was done in cafés, restaurants, cigar rooms, and the like. While it is impossible to give a full account of all such places, we give special thanks to the staff and management at those establishments where we took up significant table space over the course of our endeavor. In Toulouse, we thank the staff of La Rose de Sables, Ras La Tasse, and Al Diwan for a nearly endless supply of caffeine, sandwiches, raisin et menthe, and impeccable service. We offer particular recognition to Ras La Tasse for providing great air-conditioning during the canicule of summer 2015, and Al Diwan for providing us with a comfortable and productive spot to work during the night hours of the same visit. The final push on this book (as well as two articles written almost entirely at those locations) would have been stymied without them. In Abu Dhabi, the location of the largest number of joint visits, we thank the staffs of Tarbouche, the Marina al Bateen Lebanese restaurant (where many of our original ideas were formed), and Almaz by Momo. The constant supply of Turkish coffee and delicious food cannot be overlooked. In Chicago, we thank the staff at Iwan Ries & Co. for providing us with fine cigars and a comfortable spot to work mere steps away from the Palmer House Hilton. Last, we thank the entire staff at Nat Sherman’s in New York. The last push of this project was facilitated by their excellent Wi-Fi, impeccable ambiance, and unrivaled cigars.

On the academic front, we thank those who have taken the time to read the manuscript (or patiently listen to us ramble on about it) and provide us with constructive feedback. These individuals include David Austen-Smith, Ken Benoit, Matt Blackwell, Richard Bonneau, Sean Bottomley, Sam Bowles, Dave Campbell, Charlotte Cavaille, Chris Chabris, Tyson Chatagnier, Heidi Colleran, Carsten de Dreu, Erik Dickson, Drew Dimmery, Conor Dowling, Ray Duch, Dominik Duell, Andy Eggers, Armin Falk, David Gelman, Michael Gibilisco, Michael Gill, Justin Grimmer, Mike Gurven, Matt Hall, Andy Harris, Pablo Hernandez-Lagos, Marc Hetherington, Astrid Hopfensitz, John Jost, Bethany Lacina, Geoff Layman, Daniel Magleby, Cesar Mantilla, Michael McDonald, Slava Mihkaylov, Jeff Mondak, Rebecca Morton, Jonathan Nagler, Tommaso Nannicini, David Nickerson, Bruce Oppenheimer, Elena Panova, John Patty, Michael Peress, Dave Primo, Ben Radcliff, Jason Reifler, Molly Roberts, Larry Rothenberg, Paul Seabright, Maya Sen, Jo Silvester, James Snyder, Arthur Spirling, Jonathan Stieglitz, Jean Tirole, Josh Tucker, Karine van der Straeten, Yannis Vassiliadis, Erik Voeten, Christina Wolbrecht, Elisabeth Wood, Jon Woon, Hye Young You, and many others. This list is long, mostly because of the many visiting scholars at the Institute for Advanced Study in Toulouse, whose early and continuing feedback allowed this project to take root and develop. Additionally, we received helpful critiques on different parts of the manuscript from seminar participants at Binghamton University, Columbia University, New York University, Northwestern University, the University of Notre Dame, the University of Rochester, and Vanderbilt University as well as from participants at the 5th Annual New Directions in Analyzing Text as Data Conference, the 2015 and 2016 Pyrenean Interdisciplinary Research Events, and various American, European, Midwest, and Southern Political Science Association meetings.

Additionally, our home institutions have each contributed in their own ways. Ramey thanks Muataz Al Barwani, Benoit Marchand, and the entire team of staff working behind the scenes of NYU Abu Dhabi’s high-performance computing environment, BuTinah. Klingler acknowledges support received through ANR-Labex at the Institute for Advanced Study in Toulouse. Hollibaugh is grateful for support received through the University of Notre Dame’s Institute for Scholarship in the Liberal Arts in the College of Arts and Letters.

We were extremely fortunate to work with the University of Chicago Press (with particular thanks to Rodney Powell and Holly Smith) and to have John Tryneski guiding us every step of the way. His guidance was invaluable, and we are particularly grateful that someone with his wisdom and expertise was willing to invest so much time, effort, and trust into three young—and relatively green—scholars. Not only has the book grown (both in scope and clarity) from the original prospectus and sample chapters that Ramey and Hollibaugh peddled to him several years ago, but the three of us have also grown as scholars. For this and other reasons too numerous to list here, we are very grateful.

We also acknowledge that portions of Chapter 3 are drawn from our forthcoming (and more technically oriented) article in Political Science Research and Methods, Measuring Elite Personality Using Speech.

Most of all, we would like to thank our families, who served as sources of support during the highs and lows of writing as well as sounding boards for complaints about the other coauthors. This work would not have been possible without the personal support of Adam and Lisa Guebert, Gary Hollibaugh, Sr., Katie Hollibaugh, Steven and Marijo Klingler, Dominic and Stacy Perry, Bruce and Valerie Quinnell, Youssef and Corinne Ramey, Tania Ramey, Lara Ramey, Mariah Ramey, and the other countless colorful personalities that populate each of our lives. This is for you.

PART I

Foundations

CHAPTER ONE

Introduction

On October 16, 2013, the United States Senate voted 81–18 in favor of a bipartisan agreement ending a 16-day partial government shutdown. This resulted from a partisan impasse over the continued implementation of the Affordable Care Act, alternatively referred to as Obamacare. A few months later, Speaker John Boehner (R-OH) was interviewed on The Tonight Show with Jay Leno. When asked about the aftermath of the shutdown, Speaker Boehner had this to say:

Listen, I told my colleagues in July, I didn’t think shutting down the government over Obamacare [would] work because the President said I’m not going to negotiate. And so I told them in August. Probably not a good idea. Told them in early September.

So I said, do you want to fight this fight? I’ll go fight the fight with you. But it was a very predictable disaster. And so the sooner we got it over with the better. But remember the issue. The issue was, we wanted to delay Obamacare for a year because it wasn’t ready. Then we asked them to delay the individual mandate for a year. So we were fighting for the right things; I just thought tactically it was not the right way to do it.¹

Republicans agree on policy. We know this. Figure 1.1 presents a graphical depiction of the ideologies of all members of Congress during the 104th through the 112th Congresses, as measured by Poole and Rosenthal’s (1997) DW-NOMINATE scores, which are estimates of the latent ideology of legislators generated via examination of their recorded (or roll call) votes. As the graphic shows, Republicans are clustered on the right side of the ideological scale, and Democrats are clustered on the left side, indicating significant degrees of preference similarity within parties but little to none across parties.

FIGURE 1.1. Estimates of Congressional Ideology

As the Speaker said on Leno, the caucus disagreed on tactics, even though the underlying policy preferences—to stop the implementation of the Affordable Care Act—were the same.

1.1   A Tale of Two Senators: Chuck and Roy Disagree on the Shutdown

We managed to divide ourselves on something we were unified on, over a goal that wasn’t achievable.—Senator Roy Blunt (R-MO)²

There’s been a lot of talk about the negative impact of not raising the debt limit, but there’s too little focus on the negative consequences of ignoring the $17 trillion debt. Government spending has exploded since 2008, increasing the national debt by $6 trillion. Obamacare is a drag on the economy and hurting workers’ ability to find full-time jobs. Yet the President refuses to lead for fiscal responsibility, both short and long term, even with a government shutdown. This agreement raises the debt limit with no action on the debt. It’s a missed opportunity for forcing action to limit government and increase economic opportunities.—Senator Chuck Grassley (R-IA)³

On the day of the vote to end the shutdown, Republican senators Chuck Grassley (R-IA) and Roy Blunt (R-MO) entered the chamber with identical policy goals—to defund or delay implementation of the Affordable Care Act. Yet Senator Blunt voted for the agreement to end the shutdown and Senator Grassley voted against it. Additionally, several weeks prior, Senator Blunt had voted to end debate on H.J. Res. 59, a House-passed continuing resolution to fund the government—a vote many conservative activists decried as being essentially a vote against delaying implementation of the Affordable Care Act. Senator Grassley had voted to sustain the filibuster. In a press release issued the day of the vote to end debate on H.J. Res. 59, Senator Ted Cruz (R-TX) claimed that far too many Republicans joined Harry Reid in giving the Democrats the ability to fund Obamacare.⁴

These votes are particularly interesting because the spatial model of voting assumes legislators vote purely on the basis of the proximity of their ideal points to the positions of the alternatives under consideration. This approach carries the assumption that the tactics and procedures on which legislators vote achieve a particular policy position with certainty, that legislators are indifferent between tactics, and that the eventual policy outcome is known a priori to all legislators. This simplification—while useful for modeling purposes and tractability—is inherently flawed, as individuals who share common policy objectives frequently debate among themselves over which tactics and procedures are preferred, and which are most likely actually to achieve mutually agreed-upon ends.

FIGURE 1.2. Senator Grassley (R-IA) Tweet Example

Given the context of the vote to end the shutdown as well as the votes that preceded it, the spatial model seems lacking. Indeed, while most of the support for continuing the shutdown came from those on the more conservative end of the spectrum, Senator Blunt is actually more conservative than Senator Grassley, as measured by their Common Space scores (Poole and Rosenthal 1997).⁵ However, if we look beyond their ideologies and instead at their preferences for how they conduct business, the votes might make more sense. For starters, Senator Grassley is reasonably well-known for his prolific use of Twitter (having adopted it relatively early, in November 2007, only a few months after President Barack Obama) and is notable for handling his account himself as opposed to appointing a designated staffer.⁶ He maintains this practice despite the occasional unclear or arguably strange tweet, an example of which is presented as Figure 1.2.⁷

This willingness not only to be an early adopter of a new technology (at least relative to his colleagues) but to manage his account himself—despite the occasional misstep—speaks volumes about Senator Grassley’s willingness to take risks. Senator Blunt, on the other hand, joined Twitter in 2009—over a year after Senator Grassley joined—and a new media director manages his more official (and less personal) Twitter account.⁸ In contrast to Senator Grassley, Senator Blunt’s Twitter adoption timing and his more conventional social media management style paints the picture of someone less willing to partake in risky behavior. Considered in this light, their decisions to vote differently on the agreement to end the shutdown—as well as their different cloture votes prior to the shutdown—make more sense. Senator Grassley, being less risk averse, was willing to take the riskier option of letting the shutdown occur and the debt ceiling be reached in order to force a deal on the Affordable Care Act, whereas the more risk-averse Senator Blunt—despite having the same policy goals and being more conservative than Senator Grassley—was not. If this is even remotely true, then scholars of legislative behavior—and elite behavior more generally—need to look beyond the spatial model and account for varying preferences and beliefs over tactics.

As such, we argue that legislators do not act solely upon policy positions per se but also on intermediary tactical actions that bridge individual policy preferences and legislative outcomes.⁹ Individuals have heterogeneous beliefs and uncertainty regarding which tactics are most likely to achieve their desired ends, and may also derive utility from how policy is produced. Accordingly, legislators use decision-theoretic methods to evaluate competing tactics. In the Blunt/Grassley example, both senators wanted to prevent implementation of the Affordable Care Act. However, they disagreed over whether the tactical mechanism—the shutdown—was the best and most appropriate means by which they could achieve their shared goal. Each senator drew upon his own preferences/beliefs pertaining to the translation from tactics to policy and came to a different decision, with one voting to continue the shutdown and the other voting to end it. Thus, the key to understanding legislative behavior is to characterize and estimate the beliefs and preferences of legislators over the tactics available to them. But how?

1.2   Traits and Elite Behavior in Institutions

This question of how to understand legislators’ behavior as a function of beliefs and preferences over tactics is ripe for exploration, as scholars of institutions have, implicitly and explicitly, acknowledged the roles played by nonpolicy traits within institutional structures. For example, congressional scholars already acknowledge that nonpolicy individual differences are often important to legislative behavior and elections. Indeed, we frequently incorporate office motivation into our models of legislative behavior, and often incorporate terms for the valence characteristics of candidates into models of elections. These terms are often discussed in terms of personal character but also in terms of leadership ability.

Within the institution of Congress itself, some theoretical models have incorporated nonpolicy qualities of individuals, including character, ability, and alternative motivations. However, there has been no systematic attempt to incorporate more broadly influential individual differences into these models. This lack of theoretical development is strange, since the classic and foundational works of American politics focused on describing individual differences in elite behavior. These works generally predate the new institutionalism of the 1980s. Indeed, a focus on personal style permeated early studies of elite behavior in Congress. Fenno’s classics Home Style: House Members in Their Districts (1978), and Congressmen in Committees (1973) are excellent examples of works that examine the individual differences that lead politicians to approach their institutional roles differently.

Additionally, the study of the presidency was, until only fairly recently, dominated by studies of leadership traits and typologies of personal leadership styles. While Barber (1972) is a prime example of this style of work, more recent works (Rubenzer and Faschingbauer 2004) across political science, psychiatry, and psychology have since continued in this research tradition. The same is true for studies of the judiciary, as the structure of judicial decision making—either one person making decisions or a small group making decisions via deliberation—has lent itself to personality-based analysis (Gibson 1981). However, in contrast to the history of model-based analysis of legislative behavior, and the more recent history for executive and bureaucratic behavior, studies of the judiciary have only very recently begun transitioning from a qualitative focus on individual differences and legalism to an institutional approach centered on modeling and policy preferences, possibly in part due to the reluctance of some to treat the courts in ideological terms.

These classics have inspired several important research agendas, but to a large extent they cannot meaningfully communicate with contemporary students of political institutions. While the classics were often qualitative studies that created interesting typologies of leadership, the dominant paradigm is quantitative and theorizes on the basis of rational choice-based models, both formal and informal. Nonetheless, a tremendous amount of accumulated qualitative knowledge exists on the influence of individual differences on elite behavior within American political institutions. Models have allowed scholars of American political institutions to clarify theories of institutional behavior, but they would gain from developing a language allowing individual differences to be comprehensively modeled.

1.2.1   Translating Individual Differences into the Language of Institutions

In this book, we seek to identify a Rosetta stone that will allow students of political institutions to begin a dialogue with the rich literature on individual differences. Ideally, this approach would comprehensively and tractably include the most important persistent individual differences into models of institutional politics. The field most dedicated to studying these individual differences is personality psychology (Borghans, Duckworth, Heckman, and Weel 2008). There are many definitions of personality, but scholars generally argue that personality consists of several interrelated components, including differences in motivations, reputations, expectations, values, interests, and attitudes, rather than a single monolithic difference (Almagor, Tellegen, and Waller 1995; Benet-Martínez and Waller 1997; McAdams 2006; McAdams and