1 POLARIZATION RECONSIDERED: THE INFLUENCE OF INSTITUTIONAL STRUCTURE ON LEGISLATIVE NETWORKS By PAULINA S. RIPPERE A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 201 3
2 201 3 Paulina S. Rippere
3 To Mom, Abuela, and Grampy
4 ACKNOWLE DGMENTS T his dissertation would never have come to be without the guidance and c are of my advisor, Larry Dodd. Larry puts his heart and soul into everything he does, especially his role mento ring graduate students. His research and teaching open the minds of students to concepts and theories we never before considered, and his confidence in us inspires us to achieve what we never thought possible. him, an I am also grateful to my colleagues at UF. This includes my dissertation committee members: Beth Rosenson, Dan Smith, Michael Heaney, Ken Wald, and Sean Adams. It brother Josh Huder and Jordan Ragusa Always willing to talk, to brainstorm, and to give useful critiques, these are some of the most talented and hardworking people I have ever met. I cannot thank them enough for their guidance through the y ears. Special thanks go to Chris Tecklenburg, Tristan Vellinga and Matt Harrigan for sharing in the daily struggles and helping me keep my sense of humor. My experience in the graduate program at Florida has been enriching and rewarding, and it is because of these people. Most importantly, I am thankful for m y wonderful little family. My mother is beautiful and strong and inspires me ever y day to fight the good fight. Abuela and Grampy pushed me from the beginning to get a good education and to accomplish all that I can. They are in my heart always; this world is not the same without them I am grateful to Steve and Susan for their role as my devoted cheering section. Finally, I am forever thankful to Travis for continually providing me with overflowing love and support
5 TABLE OF CONTENTS page ACKNOWLEDGMENTS ................................ ................................ ................................ .. 4 LI ST OF TABLES ................................ ................................ ................................ ............ 7 LIST OF FIGURES ................................ ................................ ................................ .......... 8 ABSTRACT ................................ ................................ ................................ ................... 11 CHAPTER 1 INTRODUCT ION ................................ ................................ ................................ .... 12 Research Questions ................................ ................................ ............................... 22 Theory ................................ ................................ ................................ ..................... 26 Plan of the Dissertation ................................ ................................ ........................... 37 2 THEORY AND LITERATURE REVIEW ................................ ................................ .. 41 New Institutionalism: Decision Making across Contexts ................................ ......... 43 The Institutional Structure of Congress: House and Senate Differences ................ 49 Chamber Size ................................ ................................ ................................ ... 49 Length of Terms ................................ ................................ ............................... 50 Rules and Procedures ................................ ................................ ...................... 51 Party Loyalty ................................ ................................ ................................ ..... 53 Traditional E xplanations of Party Polarization in Congress ................................ ..... 55 An Alternative Measure of Party Polarization ................................ .......................... 57 Sensemaking in Organizations ................................ ................................ ............... 59 Sensemaking in the House and Senate ................................ ........................... 61 Sensemaking and Bill Cosponsorship ................................ .............................. 65 3 MEASURING COSPONSORSHIP NETWORKS AND BEHAVIOR ........................ 71 Data and Method ................................ ................................ ................................ .... 73 Incoming vs. Outgoing Cosponsorship ................................ ............................. 75 Network Partisanship vs. Cosponsorship Partisanship ................................ ..... 76 A Composite Measure of Bill Cosponsorship Ties ................................ ............ 81 Measuring the Depth of Cosponsorship Relationships ................................ ........... 86 Comparing Roll Call Voting to Bill Cosponsorship ................................ .................. 88 4 SENSEMAKING ................................ ................................ ................................ ... 104 Party Leaders vs. Non Leaders ................................ ................................ ............ 105 Ideological Extremity and Bill Cosponsorship ................................ ................. 107
6 Strength of Cosponsorship Ties ................................ ................................ ..... 114 Years of Service ................................ ................................ ................................ ... 117 Regression Analyses: Exploring Differences Within and Between Legislators ..... 120 Cohort Effects ................................ ................................ ................................ ....... 128 5 NETWORKING IN THE SENATE ................................ ................................ ......... 153 Previous House Service ................................ ................................ ........................ 155 The Gingrich Senators ................................ ................................ .......................... 162 6 CONCLUSION ................................ ................................ ................................ ...... 184 Roll Call Voting and Bill Cosponsorship: Why the Discrepancy? .......................... 187 Use of the Conflict Frame ................................ ................................ ................................ ................ 193 LIST OF REFERENCES ................................ ................................ ............................. 198 BIOGRAPHICAL SKETCH ................................ ................................ .......................... 207
7 LIST OF TABLES Table page 3 1 Characteristics of bill cosponsorship, 1979 2010 ................................ ............... 92 3 2 The Four Faces of Bill Cosponsorship ................................ ................................ 93 4 1 legislator effects ................................ ................................ ................................ 136 4 2 partisanship, between legislator effects ................................ ................................ ................. 137 4 3 Descriptive statistics of cohorts ................................ ................................ ........ 137 4 4 Cohort Effects on Partisan Network an d Cosponsorship Behavior, by Chamber ................................ ................................ ................................ ........... 138 5 1 between legislator effects ................................ ................................ ................. 171 5 2 The Gingrich Senators ................................ ................................ ...................... 172 5 3 Comparison of total network and cosponsorship partisanship between Gingrich Senators and their Senate colleagues, between legislator effects ..... 173
8 LIST OF FIGURES Figure page 2 1 Distributions of Nokken Poole DW NOMINATE scores in the House, by Congress ................................ ................................ ................................ ............ 69 2 2 Distributions of Nokken Poole DW NOMINATE scores in the Senate, by Congress ................................ ................................ ................................ ............ 70 3 1 Example of a sponsor by cosponsor matrix, 110 th Senate ................................ .. 93 3 2 Distributions of incoming network and cosponsorship partisanship variables, by chamber ................................ ................................ ................................ ......... 93 3 3 Distributions of outgoing network and cospon sorship partisanship variables, by chamber ................................ ................................ ................................ ......... 94 3 4 Total network partisanship, by Congress, House of Representatives ................. 95 3 5 Total network partisanship, by Congress, Senate ................................ .............. 96 3 6 Total cosponsorship partisanship, by Congress, House ................................ ..... 97 3 7 Total cosponsorshi p partisanship, by Congress, Senate ................................ .... 98 3 8 Average depth of partisan cosponsorship ties, by Congress .............................. 99 3 9 Average depth of bip artisan cosponsorship ties, by Congress ......................... 100 3 10 Ideological extremity vs. total network partisanship in the House, by Congress ................................ ................................ ................................ .......... 101 3 11 Ideological extremity vs. total network partisanship in the Senate, by Congress ................................ ................................ ................................ .......... 102 3 12 Ideological extremity vs. total cosponsorship partisanship, by chamber and Congress ................................ ................................ ................................ .......... 103 4 1 Ideological extremity among party leaders and non leaders, 96 th 110 th Congresses ................................ ................................ ................................ ...... 139 4 2 Network partisanship among party leaders and non leaders, 96 th 110 th Congresses ................................ ................................ ................................ ...... 140 4 3 Cosponsorship partisanship among party leaders and non leaders, 96 th 110 th Congresses ................................ ................................ ................................ ...... 141
9 4 4 Ideological extremity vs. total network partisanship for party leaders and non leaders, by chamber ................................ ................................ ......................... 142 4 5 Ideological extremity vs. total cosponsorship partisanship for party le aders and non leaders, by chamber ................................ ................................ ........... 143 4 6 Depth of partisan ties for party leaders and non leaders, by chamber ............. 144 4 7 Depth of bipartisan ties for party leaders and non leaders, by chamber ........... 145 4 8 Total network partisanship over number of years served, by chamber ............. 146 4 9 Convergence of total network partisanship over number of years served, by chamber ................................ ................................ ................................ ........... 147 4 10 Total cosponsorship partisanship over number of years served, by chamber .. 148 4 11 Differences in networking patterns across House cohorts, by Congress .......... 14 9 4 12 Differences in networking patterns across Senate coh orts, by Congress ......... 150 4 13 Differences in cosponsorship patterns across House cohorts, by Congress .... 151 4 14 Differences in cospo nsorship patterns across Senate cohorts, by Congress ... 152 5 1 partisanship, by Congress ................................ ................................ ................ 174 5 2 partisanship, by Congress ................................ ................................ ................ 175 5 3 Change in the total network partisanship of senator s with and without previous House service experience, by Congress ................................ ............ 176 5 4 Change in the total cosponsorship partisanship of senators with and without previous House service experience, by Cong ress ................................ ............ 177 5 5 service for those who served in both chambers, by Congress .......................... 178 5 6 Senate service for those who served in both chambers, by Congress ............. 179 5 7 an total network partisanship in the House and in the Senate ................................ ................................ ................................ ........ 180 5 8 the Senate ................................ ................................ ................................ ........ 181
10 5 9 Total network partisanship of the Gingrich Senators and their Senate colleagues ................................ ................................ ................................ ........ 182 5 10 Total cosponsorship partisanship of the Gingrich Senators and their Senate colleagues ................................ ................................ ................................ ........ 183
11 Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy POLARIZAT ION RECONSIDERED: THE INFLUENCE OF INSTITUTIONAL STRUCTURE ON LEGISLATIVE NETWORKS By Paulina S. Rippere May 201 3 Chair: Lawrence C. Dodd Major: Political Science Although the literature on congressional change has established that political parties in Congress have become increasingly polarized over time, this conclusion is limited to the study of voting behavior. The analysis of another legislative practice, bill cosponsorship, reveals that while the parties in the House have divided, those in the Sena te have not. This suggests that (1) the extent of party polarization may vary across stages of the legislative process and (2) there are important differences in the personal interactions of legislators in each chamber. Overall, this dissertation argues th at using an alternative measure of party polarization produces results different from what the traditional literature suggests.
12 CHAPTER 1 INTRODUCTION On August 25, 2009 Edward Moore Kennedy passed away after a year long battle with brain cancer. While serving the state of Massachusetts in the U.S. Senate for almost 47 years (1962 2009), Kennedy became known as unabashedly liberal and areas of immigration, civil right s, education, and health care, Kennedy was no ideological moderate. He consistently received interest group ratings, or ideology scores based on roll call voting, between 90 and 100 percent liberal from groups like Americans for Democratic Action and the National Journal left issue positions and pattern of voting so consistently with his party, Ted Kennedy is remembered as a legislator who was willing to build relationships that crossed party lines and to compromise, qualities which helped him work with his partisan opponents to pass hundreds of bills of significance over the course of his career. alities: his ability to turn even ideological opposites into allies. A committed liberal known for his thundering oratory, Kennedy was also a quiet dealmaker who worked cooperatively with even the include his work with Republican President George W. Bush on the major overhaul of national educational standards, now known as No Child Left Behind, and his partnership with John McCain (R AZ) on
13 immigration legislation. Kennedy also became known for wor king with Phil Gramm (R President Bill Clinton, one of the most highly polarized and contentious proceedings in recent history (Bayh 2010). Most notable is his long time personal friendship and work with Orrin Hatch (R UT) on issues ranging 09) remarked, more often than not, I was trying to derail whatever big government scheme isagreements over policy, however, were never constructive and shrewd lawmaker. He never lost sight of the big picture and was willing to compromise on certain provisions in order to move forward on issues he believed important. And, perhaps most importantly, Congress, on the airwaves, in cyb erspace, and in the public square will learn that being faithful to a political party or a philosophical view does not preclude civility, or even friendships, with those on the other side. These remarks suggest that Hatch and Kennedy, two senators who regu larly took opposite positions on significant and contentious issues, were able to work together and compromise even if they typically did not vote together. However, this is not the picture that is usually painted of bipartisan relationships in the House and Senate. news broadcasts and in newspapers across the country which tend to focus on the sad politicians, increasing party polarization, and (particularly at the end of 2012) the continual failure of the years go by, nothing is being done in Congress. Med ia reports on the decline of
14 bipartisanship are supported by the work of political scientists who study roll call voting patterns. These quantitative analyses (as well as the interest group scores mentioned above) reveal that, as time passes, members of e ach party in both the House and the Senate tend to vote more regularly with their fellow partisans and against their partisan opponents (this topic is discussed in detail in Chapter 3). The days of successful bipartisan coalitions on major legislative pro posals appear to be long gone. Indeed, party polarization is a very pervasive force in Congress today, and this is particularly true of relationships in the House. From the first day that a newly elected representative arrives in Washington, D.C. he is ta ught that party identification dictates almost everything from where one sits on the floor of the chamber and in committee, to to vote. In recent years, a legislato freshman orientation session he will attend, influencing his approach to policy making and the relationships he establishes even before the legislative session begins. Since 1972, the Kennedy School at Harvard has held an orientation for newly elected members of the House. During the session, legislators attend lectures on topics including the economy, global politics, health care, and the environment. They also learn about legislative rules and procedures and re ceive advice on matters like how to organize their office and hire staff. In 1992, the Heritage Foundation, a conservative think tank, began holding an orientation for new Republican legislators intended to serve as an alternative to the Kennedy School se ssions which some legislators had complained were too liberal (Ross 1992). Although the orientation at Harvard was officially sanctioned by the House and paid for with government funds, it has become
15 attended almost exclusively by Democrats. In 1994, giv en the large class of Republican representatives elect, the Kennedy School even canceled their orientation due to lack of interest; in contrast, approximately 60 newly elected Republican House members attended the orientation at the Heritage Foundation (Se elye 1994). In 2012, this division between members of the two major political parties is even greater. All freshmen House members sit in on morning orientation sessions together, but they are separated by party from the afternoon into the night. Party le aders Speaker John Boehner and Minority Leader Nancy Pelosi hold segregated dinners for newly elected members of the Republican and Democratic parties, respectively. This separation makes it difficult for legislators to meet and build relationships with t heir colleagues across party lines. In the few cases where they do make the acquaintance of their partisan opponents, freshmen legislators can be afraid to admit these party fraternization is frowned upo with like (Carney and Renacci 2011). As newcomers, they avoid making waves and strive to fit in an d learn the prevailing norms of the institution. Furthermore, the congressional chambers have been no stranger to name calling, swift retorts, and other types of uncivil behavior exchanged between partisan opponents. Perhaps the most recent widely publici zed example is when, during Wilson (R Representatives (Abdullah 2012). In July 2003, Representative Bill Thomas (R CA)
16 gave a tearful apology to his colleagues on the floor of the House for his actions in response to a conflict that bubbled up days earlier during the meeting of the Ways and Means Committee, of which Thomas was Chairman. When Thomas and Republicans on the committee made overnight changes to a pension bill and tried to push it through for a vote, Democrats on the committee felt they had not been given enough time to read the new version of the bill and adjourned to a library adjacent to the committee ro om. In response, Chairman Thomas called Capitol police and asked them to remove his Democratic colleagues from the library (Eilperin 2003). During the same committee meeting, Thomas made a separate phone call to the House sergeant at arms requesting that CA) had heard a challenge from Rep. Scott McInnis (R CO), the 71 year old Stark said, ( Associated Press 2003). Some steps have been taken to try to repair relationships in the House. In March 1997, 200 House members and their families attended the first Bipartisan Congressional Retreat in Hershey, Pennsylvania. Coordinated by the Aspen Institute, a Washington, D.C. think tank, and funded by a grant from the Pew Charitable Trusts, the event was a rare opportunity for legislators to get to know one another on a personal level in an attempt to improve relations while back at the Capitol (House Committee on Rules 1999). The theory behind this approach, which has been echoed by several
17 continued instances of incivility over t he last fifteen years, the success of the se endeavor s is debatable. In 2011, Representatives Nan Hayworth (R NY) and David Cicilline (D RI) Congress to build relationships wi (Hayworth 2011). During their first meeting, Hayworth and Cicilline were joined by only nine of their colleagues (Hayworth 2011), and because Hayworth lost her bid for reelection in 2012, it is unclear whether or not the Caucus will continue meeting during the 113 th Congress. An example of a more informal effort to create bipartisan friendships in the House James Re nacci (R OH) and John Carney (D anything, accomplished without the rancor, posturing and divisiveness that has two to three weeks in a room in the Cannon House Office Building, members of the group eat and discuss legislation they would like to work on together; some have said with new friends from acr What spawned the friendship between Renacci and Carney which eventually led to the monthly breakfast meetings? The co for months in the H ouse Financial Services Committee, we decided to get lunch. We
18 quickly realized that, while we disagreed on many issues, we had a similar outlook on a host of others. After getting together a few more times, we started inviting colleagues to talk over brea (Carney and Renacci 2011). While these attempts at building bridges across party lines in the House do exist, they involve a very small minority of House members and are the exception to th e rule. However, the story is quite different in the Senate. In the smaller chamber where legislators serve six year terms and often operate in complete agreement on the basis of unanimous consent, newly elected senators are taught that this chamber is d ifferent from the House. Traditionally, freshmen senators would learn the ways of the Senate from the senior senator from their home state or from the Secretary of the Senate. In 1976 and into the 1980s, when much larger freshmen classes of 17 20 people were given by the Senate historian followed by a dinner for all of the senators elec t. Unlike the dinners for newly elected House members which are segregated by party membership, the Senate dinner is for all freshmen senators and is sponsored by the President Pro Tempore and the majority and minority leaders (Senate Historical Office n. d.). As a result, freshmen senators are given time to meet and get to know members of both parties from the very start of their time in the Capitol. During the 1996 orientation session, Senator Robert Byrd (D WV), who became known as the longest serving s enator, addressed the class of 1996 and called their
19 attention to some of the most important ways in which the Senate differs from the House (Baker 2006): The Senate and, therefore, Senators were intended to take the long view and to be able to resist, if need be, the passions of the often intemperate House. Few, if any, upper chambers in the history of the western world have possessed the Senate's absolute right to unlimited debate and to amend and block legislation [Its] deference to minority views sharply distinguishes the Senate from the majoritarian House The Senate is often soundly castigated for its inefficiency, but in fact, it was never intended to be efficient. Its purpose was and is to examine, consider, protect, and be a totally independent source of wisdom and judgment on the actions of the lower house and on the executive. As such, the Senate is the central pillar of our Constitutional system. I hope that you, as new members, will study the Senate in i ts institutional context, because that is the best way to understand your personal role as a United States Senator. In this passage, Byrd conveys to his newly elected colleagues the qualities toward which each of them should strive over the course of their career. He points out that the Senate, unlike the House, is supposed to uphold the rights of the minority, often resulting in a lengthy, but necessary, deliberative process. Moreover, Byrd emphasizes the institution above all else; this stands in stark contrast to an alternative view of their responsibilities as a party loyalist or power seeker. All of this plays a role in imparting on the freshmen the image of the Senate as a body which necessitates bipartisan communication and compromise. Later in his speech, Byrd e laborated on how the institutional structure of the Senate, its rules and traditions, incentivize bipartisanship (Byrd 1996): Now I would like to turn for a moment to the human side of the Senate, the relationship among Senators, and the way that even that face of service The requirement for super majority votes in approving treaties, involving cloture, removing impe ached federal officers, and overriding vetoes, plus the need for unanimous consent before the Senate can even proceed in
20 many instances, makes bipartisanship and comity necessary if members wish to accomplish much of anything. Realize this. The campaign is over. You are here to be a Senator. Not much happens in this body without cooperation between the two parties. Indeed, from the very moment they begin their first term, new senators are instructed in the importance of working with their colleagues on the other side of the aisle. Likewise, they are reminded that the Senate is an institution composed of people who must maintain constructive working relationships with one another if they are to be successful. This introduction is drastically different from the stories gleaned from the discouraged. In 2004, two junior senators, Lamar Alexander (R TN) and Thomas Carper (D DE), began a further attempt to foster healthy working re lationships between members of the two political parties. Designing a freshman orientation schedule which focused where twenty sitting senators served as mentors to the freshmen (Broder 2004). Each discussion was led by a bipartisan pair of senators, and the orientation included sessions which specifically addressed pulled into their party caucuses and start exch an berexclusive social spot 11). Meeting
21 group of senators (10 Democrats, 9 Republicans), made an effort to improve bipartisan relations by getting to know one another personally (Raju 2011). Furthermo re, Alexander and Senator Chuck Schumer (D unused, empty, never While the increased ease of air travel and the subsequent shortening of the legislative week to about three days (Tuesday through Thursday) did cause a decline in the amount of socializing that went on between senators, these efforts to establish bipartisa n working and personal relationships have occurred frequently in the Senate in recent years. Moreover, these reports illustrate that (1) the Senate cannot function without bipartisan compromise and (2) sitting senators consistently make efforts to sociali ze newly elected legislators to become hard working and fair negotiators rather than hard nosed partisans. Given these anecdotes contrasting experiences in the House and Senate, we are left to wonder: Why are relationships in these two chambers so differen t? Why does the House emphasize to its newly elected members that partisan opponents should be relationships across party lines? Furthermore, how do these different experien ces stack up against the empirical evidence that party polarization in roll call voting has been increasing significantly over the last 30 to 40 years? While the Senate is not void of cross party animosity, news reports, anecdotes, and legislator testimon ies give the impression that there is something distinctively different about that chamber which sets
22 it apart from the more raucous House. Perhaps there is something unique about the institution and its rules that make it more likely to foster civil beha vior and the formation of bipartisan friendships? This is the question this dissertation addresses. Research Questions For many years, congressional scholars have sought to understand the factors which influence and motivate the behavior of legislators. These puzzles have introduced numerous questions regarding many aspects of legislative decision making, including bill sponsorship and cosponsorship, roll call voting, committee membership, service in the party leadership, the influence of partisanship and ideology, and constituency service. Scholars have explored these questions from a variety of perspectives using multiple theoretical lenses and methodological tools. Theories which have been particularly useful include rational choice theories which foc us on the individual goals of members of Congress, structural approaches which emphasize the historical and institutional context, and theories of social learning and adaptation that consider changes in member behavior over time. While these theories, and others like them, have helped us learn much about the legislative process, they have been developed, in large part, based on evidence from the House of Representatives. This is because the House is a much larger body than the Senate, providing scholars w ith a much larger number of cases for quantitative analysis, and because the more fluid and personalistic nature of the Senate makes developing generalizable theories more difficult. Part of the problem with this trend is that studying the House alone tell s us only half of the story. Members of the two chambers must work together to pass legislation, and the features unique to each chamber play a role in this process. Another concern is
23 that when scholars do analyze the legislative process in the Senate, they often do so by viewing the institution through theoretical lenses designed to view and understand the House. To utilize this approach, we must assume that the goals and behavior of senators and the context in which they operate are similar to those o f members of the House. However, because the two bodies are very different, we have reason to believe that the assumptions made for the House do not necessarily hold for the Senate. For example, differences in the structures of the institutions produce different working environments. The world of House members is one that is characterized by predictability. There is a clear hierarchy of power structured by party leadership posts. There are well defined rules that govern every day proceedings in the ch amber, such as who may speak on the floor and for how long. More broadly, there are regular two year election cycles to which all members are subjected. This means that at any given point in the congressional session, every House member (with the excepti on of those who plan to retire) has exactly the same amount of time before he or she must face an election, giving all legislators in the chamber a common goal on which they are focused simultaneously. This stability and predictability provides legislator s with a consistent framework for interpreting social relationships. With everything laid out for them, there is very little guesswork regarding how to behave or with whom to associate. In contrast, senators are charged with a variety of tasks and operate in a fluid environment where they have more room to develop their own unique goals and innovative approaches to problem solving. By virtue of Senate rules governing the filibuster and unanimous consent agreements, among others, each senator has power as an individual without having to hold an official post in the party leadership. At any
24 given moment a single senator can bring legislative decision making to a halt. A former Kennedy staffer notes, s legislation. In the HELP [Health, Education, Labor, and Pensions] Committee, the vast bulk of the health legislation we worked on did not qualify for precious and limited floor time which meant that it had to be passed by unanimous consent. This required bipartisanship even when we were in the majority. And when without Republican support. This reality makes compromise and the formation of a dense network of personal relationships abs olutely necessary (Nexon 2009, emphasis added). Due to these factors and the resulting lack of predictability in the chamber, navigating life in the Senate is much more difficult than is decision making in the House. To overcome these challenges, senators in comparison to members of the House, rely less on their partisan identities and more on the social relationships they cultivate during their time in office (Dodd 2002). Although writers have commented for decades on the differences between the structur es of the two chambers, many scholars have continued to use theories designed for the House to understand the behavior of senators. When decades of scholarship suggest that the two environments are different, how can these assumptions be valid? How can w e ignore an entire area of literature which discusses the fundamental differences between the two legislative chambers and argue that the theories which explain the behavior of representatives also explain the behavior of senators? For some scholars the s olution has been to study the House alone; while this may keep us from misinterpreting what we observe in the Senate, we learn nothing about the policymaking process in the chamber. What is needed to remedy this situation is a theoretical approach that wil l help us to explain the behavior of legislators in both the House and the Senate. Such an
25 approach would allow us to account for the unique contexts of each chamber and the nt. While the House setting is structured, regularized, and predictable leading to the adoption of generic roles and the development of common goals, the Senate is fluid and unpredictable, requiring members to take on a variety of tasks and generate innov ative solutions to diverse challenges (Dodd 2002). Although a rational choice perspective may be useful in understanding the routinized behavior and widely shared goals of members of the House, it is not necessarily appropriate for interpreting the priori ties of senators which fluctuate along with the changes in their working environment. Moreover, while the institutional and historical context influence the actions of legislators in both chambers, rational choice and structural theories focus on either t he individual or the background, respectively. Rather than limiting our study by utilizing a theory which forces us to choose between these two perspectives, why not identify an approach which considers elements of the individual, the institutional struct ure, and the temporal dimension altogether? The goal of this dissertation is to explore the influence of institutional structure on the behavior of members of Congress and to develop a theory of member behavior which takes into account the unique environme nts of each chamber. Specifically, the research questions are: How do the different institutional structures and working environments of the House and Senate affect the interactions and behavior of members of each chamber, and how have changes in institut ional structure influenced the goals and decision making processes of members over time?
26 Theory To answer these questions, I develop a theory of sensemaking, or how members of the House and Senate interpret and adapt to the unique environment of each chamb er. I use this theory to explain differences in the bill cosponsorship behavior and networks of legislators. The motivation for exploring bill cosponsorship rather than other legislative behavior comes from another weakness of the literature. Specifical ly, scholars who study the development of party polarization in Congress rely almost exclusively on data on roll call voting. Based on these analyses, scholars have concluded that members of the House of Representatives and the Senate have become sharply divided along party lines over the course of the post War period. While this finding is accepted widely both in the academic literature and within popular culture and the media, a question which has not yet been considered is: What does party polarization look like within other types of legislative behavior? To answer this question, this dissertation explores the connections made by members of Congress through bill cosponsorship. While roll call votes serve as he decision to cosponsor legislation captures a much more personal aspect of the legislative process. Often performed when one legislator asks another for support, the act of cosponsorship signifies that a relationship exists between legislators. What is important in this context is not the type of bills which a legislator cosponsors, but whether or not the individuals with whom s/he cosponsors cosponsorship across numerous b ills paints a picture of the types of interactions, either partisan or bipartisan, in which s/he tends to engage over time.
27 Another reason for utilizing bill cosponsorship data is that reliable measures of more personal connections are difficult, if not im possible, to obtain. While it would be ideal to collect data on more illuminating indicators of House and Senate friendships, for example, which legislators play basketball together or which go to the bar together, this information is not available public ly. Likewise, other informal stages in the legislative process, such as private conversations among legislators and staff and vote trading, may also give us valuable insight into the nature of bipartisanship in each chamber, but they, too, are not accessi ble to most researchers. While legislators may cosponsor bills simply because they have a personal interest in the policy or because they are trying to claim credit for a popular action which they can tout to their constituents in a stump speech, there is important evidence that common bill cosponsorship is a significant indicator of a meaningful relationship between legislators. Specifically, legislators invest time recruiting colleagues to cosponsor their bills cosponsorship requires social interaction and collaboration among members (Bratton and Rouse n.d.; Cho and Fowler 2010; Lipinski 2009). It is an indicator that legislators trust one another (Fowler 2006b) and a sign that m embers of Congress work together by choice Moreover, when members of both parties are willing to cosponsor a bill, this signals to other members that the legislation is ideologically balanced and is not a tool of a particular party. Bipartisan cosponsor ship is so important that it was part of typical strategy was to team up with one or more Republican partners. By making comm on cause with at least a limited number of Republicans, while holding the Democrats together, the senator could put together the sixty
28 votes generally needed to pass legislation. His unique stature as the leading liberal champion helped him lead progressiv e Democrats to support compromises that they would not necessarily swallow from other, less trusted Democrats. At the same time, Republican support made it much easier to hold moderate or conservative Democrats who would be nervous about supporting the kin d of ambitious legislation Senator Kennedy worked for, without Republican support to provide cover. For progressive Democrats, it was hard to be to the left of Senator Kennedy. For moderate and conservative Democrats, it was hard to be to the right of Repu blican senators like Orrin Hatch of Utah or John McCain of Arizona (Nexon 2009). As a loyal progressive, Kennedy easily drew the support of his liberal Democratic colleagues. By including Republicans as cosponsors, Kennedy was also able to attract the sup port of moderate Democrats and Republicans who were key to passing the legislation. The success of these signals relied on the reputation Kennedy built for himself as a trustworthy legislator with a vast bipartisan personal network. One of the most promin ent bipartisan bills of recent years is the Bipartisan Campaign Reform Act of 2002 (BCRA), informally referred to as McCain Feingold. A major legislative effort intended to increase restrictions on campaign financing, the lead cosponsors on the bill were Senators John McCain (R AZ) and Russ Feingold (D WI) and, in the House, Representatives Christopher Shays (R CT) and Martin Meehan (D MA). McCain, who belongs to the class of hyper partisan warriors known as the erest of his party and its leadership in the Senate and was able to pass reforms the pair had attempted for the previous seven years (Waldman 2001; for more on the Gingrich senators, see Chapter 5). During this process, the senators involved attended nume rous bipartisan meetings and engaged in weeks of debate and negotiations that paired conservatives like McCain with liberals such as Dianne Feinstein (D CA) and then Senate Minority Leader Tom Daschle (D SD) (Lancaster and Dewar 2001). Over the years, the cosponsors had to accept
29 profile A more dramatic case is the story of the long fought battle over the issue of men tal health parity. Two unexpected allies, Senators Pete Domenici (R NM) and Paul Wellstone (D MN), occupied drastically different ideological positions but worked together for ten years to pass legislation requiring insurance companies to cover mental hea lth issues equally with physical diagnoses. Domenici, an unlikely advocate for the control and same sex marriage and supported school vouchers, tax cuts and mandatory t hree strikes sentencing. He was no bleeding heart, no cause However, the senator was attracted to the issue of mental health when his daughter, t was finally diagnosed as atypical schizophrenia, it is improbable that Pete Domenici, Mr. On the Democratic side, Wellstone became interested in the topic because his older brother suffered from mental illness: alliances with former Senator Alan Simpson, whose niece commit ted suicide, and Senator Harry Reid, whose father killed himself, and Tipper Gore, who has suffered depression, and Representative Marge Roukema [R NJ], whose husband is a psychiatrist, and Representative Patrick Kennedy [D RI], who has also battled depres sion, and Senator Edward wish it didn't have to work that way, that all of us would care deeply anyway about peopl e who were vulnerable and not getting the care they need. But this kind of thing happens a lot in politics for fully human reasons (Sontag 2002).
30 These cross party ties helped Domenici and Wellstone grow their support from just over 20 Senate cosponsors wh en they initially introduced legislation in 1992 to the 66 cosponsors they had ten years later (Sontag 2002). In 1996, the senators were able to pass a limited version of their bill but continued to work until 2002 when Wellstone died tragically in a plan e crash. In 2004, Domenici continued his work to expand the legislation, this time with Senators Kennedy and Michael Enzi (R unusual consensus finally pass the legislatio n in 2008 as the Mental Health Parity and Addiction Equity Act (Pear 2008). The final examples illustrating the significance and influence of bipartisan cosponsorships involve the work of Senator Ted Kennedy. While the list of successful bipartisan bills In the first case, Kennedy joined forces with Senator Nancy Kassebaum (R KS) to pass the Kassebaum Kennedy Health Insurance Portability and Accountability Act of 1996, more commonly known as HIPAA. This law allows workers to keep their insurance when they change jobs and requires insurance companies to cover preexisting Kennedy knew that an y reforms related to this issue would have to be less aggressive and would require bipartisan support. As a result, he chose the widespread problem of preexisting conditions and insurance coverage as an issue that members of both parties could get behind. Working with Kassebaum and her Republican colleagues required
31 compromise, a major This bill was seen by Kennedy as one of his most important efforts (Grove 1996) and was also identified by David Mayhew as a landmark law (Mayhew n.d.). Also in the area of health care, Ted Kennedy and Orrin Hatch were the primary supporters of the Ryan White Comprehensive AIDS Resources Emergency (CARE) Act passed in 1990. This law provided funding for the care and treatment of low income, uninsured, and under insured peopl topic would pass the Senate by the large margin of 95 to 4 was the result of successful Resources and Services Adm inistration n.d.). Another unusual partner for Kennedy was Senator Lauch Faircloth (R NC), with whom Kennedy cosponsored the Faircloth Kennedy Church Arson Prevention Act of 1996 which strengthened penalties for arson at houses of worship. As with Orrin H atch, Kennedy stood at the opposite end of the ideological spectrum from Faircloth, but that Kennedy's co sponsoring a bill with Faircloth -not lost on anyone -is all th e more piquant because the arch conservative from North Carolina is a staple demon of intervi
32 represents his causes well, and is very hard working and very thorough in what he -conservative Rep ublicans and a fervent foe of just about everything Kennedy stands Kennedy bill passed the Senate 98 0, cleared the House, and was signed into law by President Clinton in 1996. What these examples illustrate is the effecti veness of building personal relationships in the Senate. In so many of his interactions with his Republican colleagues, Ted Kennedy showed that he was willing to give in order to receive; moreover, his repeated success shows that this strategy works. Ken nedy consistently approached potential Republican cosponsors with an attitude of openness and a willingness to ensure that they, too, got something out of the law they were working to pass. Why is this approach so important in the Senate? The reasons rel ate to the and less than that for most of American history and a historically weak party system, personal conn ections mean 100 members the Senate still encourages the forging of cross party per sonal relationships inside the chamber (Loomis 2000, 2). The lack of a strong hierarchical party leadership structure and the presence of rules giving power on an individual basis to each senator give members of this chamber the freedom to associate with whomever they choose. It is these rules that allow any individual senator to obstruct and put a stop to the process altogether. This
33 configuration is what requires bipartisan compromise for the successful passage of legislation, and it is the bipartisan From an historical institutionalist perspective, it is against the background of this structure that senators are highly motivated to build friendships with their colleagues. When they are first elected to the chamber and are learning the ways of the institution, peculiar S enate friendships are very much the product of individual calculations by individual senators who recognize that their political objectives cannot be achieved senator party ties. Considering the Senate as a whole, the existence and maintenance of strong bipartisan relationships can also help keep the chamber functioning, or at least keep it f rom imploding, during times of heightened party polarization and conflict. These the components of institutional kinship serve to provide an important, perhaps indispensable, force fo period when the institution risked being torn apart was the impeachment trial of President Clinton. Embedded in memory are the notorious fights that took place in the House between members of op posing parties, one of which involved Representatives Patrick Kennedy (D RI) and Bob Barr (R GA). When Barr gave a speech on the House floor against President Clinton and cited a quote from the late President John F.
34 became furious. As Barr exited the House giving a speech at a convention put on by the Council of Conservative Citizens, a group that promotes the preservation of the white race (Merida 1999). The compromises on impeachment were worked out by bipartisan teams such as Slade Gorton (R WA) and Joseph Lieberman (D CT), Dianne Feinstein (D C A) and Robert Bennett (R UT), Tom Harkin (D IA) and Susan Collins (R ME), and even the unlikely pair of Edward Kennedy (D MA) and Phil Gramm (R to differentiate themselves from the House, senators refuse to let their party loyalty in the Senate conducted its business in a much more civil manner than the infor mal groups and worked out compromises that prevented the sort of outbursts that Senators and jour long standing tradition of courtesy, to the bipartisan friendships that permit Senators to reach agreement even in the face of policy disagreements, to the six year term that gives Senators more fr eedom to repel outsiders who would push them to extremes, and to simple fear of replicating the contentiousness of the House debate (Uslaner 2000, 33) It is the personal relationships built across party lines that help keep the institution together during the darkest of days. In his analysis of the observance of norms of civility and comity in each chamber of Congress, Eric Uslaner argues that the differences between the House and Senate
35 are really a matter of degree. It is not that the Senate is civil whi le the House is not; it is more that the Senate is less uncivil block legislation so easily, so they make a lot of nois implication of members because there is something about the institution which incentivizes good behavior. Instead, senators are less uncivil than represent atives simply because they have other, more palatable alternatives for getting their way. While there is no doubt that the Senate has witnessed its own share of skirmishes and indecencies, I argue in this dissertation that there is something more going on under the surface. While partisan battles are broadcast and reported widely in the media and are severe enough to cause legislators to lament the state of membership, longer terms, and unique rules, which promotes and sustains an undercurrent of bipartisanship flowing just beneath the surface of what is witnessed by most observers. Moreover, this phenomenon is decidedly absent from the House. To explain this, it is nec essary to understand that civility and comity are different from collaboration. While of course civil behavior is more likely to attract allies than is name calling or personal attacks, it is possible for legislators who may have spoken against one anothe r in the past to develop and maintain a strong collaborative relationship. To this point, Uslaner attributes incivility observed in the Senate, as well as the decline in comity in both chambers, to the decline in the number of ideologically moderate legis lators. As the number of conservative southern Democrats and the
36 number of liberal eastern Republicans declined, both cha mbers of Congress have lost the sources of relative moderation within their parties. And it is moderates who reach out to people in t he other party to compromise on legislation and who live and die by good personal relations. The famed Senate friendship of Republican Orrin Hatch and Edward Kennedy is rather unusual. Ideological opposites may get along well enough, but they do not neces sarily work well together on issues (Uslaner 1999, 4 5). As we have observed in the stories recounted above, it is indeed very possible for legislators who occupy opposite ends of the ideological spectrum to work together on legislation, often for years at a time. While the close friendship of Kennedy and Hatch may be unique, they are joined by numerous bipartisan pairs of senators who managed to get along well enough, despite their polarized ideologies, to cosponsor legislation and see it through to enact ment. Yet another example illustrates that civility in no way requires an absence of partisanship. Senators Phil Gramm [R TX] and Daniel Patrick Moynihan [D NY] were passionate ideologues from opposite sides of the spectrum. But both typically made thei r arguments politely and with comity. Perhaps that is why they were frequently part of the negotiating team when final compromises were struck on important legislation (Sununu 2011). In this case, two senators who certainly were not moderate were known fo r their ability to collaborate and steer important bills to their successful passage. Likewise, Senators John McCain and John Kerry (D MA) started out on bad terms due to their opposing views on the Vietnam War (McCain saw Kerry as a traitor for participa ting in anti War protests and throwing away his medals), and McCain even campaigned against Kerry in Massachusetts in 1984, an action that was deemed taboo, or even uncivil (Baker 1999, 7 8). However, serving on the Senate Select Committee on P.O.W./M.I.A Affairs
37 and personally at odds, found themselves on the same side of an issue that both cared together opponent, McCain refused. Plan of the Dissertation As we will see in the ch apters that follow, party polarization in voting behavior does indeed plague the House and the Senate. However, when we examine the stages of the legislative process that precede the vote, we find that occupants of the two chambers behave very differently Specifically, while House members learn to associate almost exclusively with their fellow partisans at the expense of building relationships with their partisan opponents, senators tend to associate almost equally with members of both political parties. Indeed, ideological opposites who frequently cast votes on opposite sides of legislation also work together through bill cosponsorship. One does not need to be a moderate or to operate in a chamber with perfect civility to collaborate with members of th e other party. Furthermore, the reason for these inter chamber differences rests with the institutional design of each chamber. Because the House has 435 members, it must operate as a strict rule based hierarchy; without clear roles for each legislator an d rules that keep the chamber to a strict schedule, chaos would break out. This clearly defined structure leaves very little room for individual legislators to break from the patterns laid out for them. New House members are taught from the beginning to associate with members of their party and to stay away from those who are unlike them. It is not often
38 allows for the creation of institutional rules that give individua l senators much more freedom and power relative to their House counterparts. Seeing that their ability to easily obstruct creates gridlock and keeps them from accomplishing their policy goals, senators learn that they must build relationships with their c olleagues of both parties if they are to achieve success in passing legislation. As a result, the formation of bipartisan working relationships is incentivized and enshrined within the structure of the Senate. Guided by the theory of sensemaking, we can m ake several predictions about cosponsorship networks and behavior in each chamber. First, I expect that the rigid structure of the House and the routinized behavior it encourages forces members to rely on partisanship as a behavioral cue. As a result, th e cosponsorship networks of House members will include more of their fellow partisans relative to the networks of senators. The increase in party polarization during recent years has no doubt had an effect on the networking behavior of House members. Whi le I expect that higher levels of polarization will also influence senators, the institutional structure and rules of the Senate force senators to interact and forge relationships with their colleagues on both sides of the aisle more regularly than represe ntatives must. As a result, the networks of senators will be composed of a smaller ratio of their fellow partisans to their partisan opponents and will be more diverse than those of representatives. This expectation stands in sharp contrast to the finding s presented in the existing sensemaking processes vary according to the institutional setting in which they serve suggests that it is not appropriate to make an ob servation about legislative behavior in
39 the House and apply it to the work life of the Senate. Perhaps more importantly it implies that policymaking strategies that bring success for members of the House may not work for legislators when they move to the Senate. In Chapter 2 I explore the sensemaking literature and explain why it is appropriate to use this theory to shed light on House and Senate behavior. I elaborate adapt ing to the different institutional settings influence their strategies for networking and cosponsorship. Furthermore, I delve deeper into the literature on House and rules influence decision making processes and why do they have this effect? Finally, I explore research on bill cosponsorship to show the significance of this activity in s. In Chapter 3, I describe in detail the data and methods of analysis used throughout the dissertation. I provide preliminary evidence which indicates that the cosponsorship networks of House members have indeed polarized over time, while those of senato rs have not. I take the analysis further in Chapter 4 where I divide members of each chamber according to their institutional roles and individual characteristics and show that even though some legislators may occupy similar ideological positions based on their roll call voting patterns, their cosponsorship behavior can be very different. The results of the analysis illustrate differences in the behavior and networks chang e over time the longer they serve in each chamber.
40 Finally, having demonstrated that the pattern of cosponsoring legislation and Senate, the analysis in Chapter 5 focu ses specifically on this chamber. Here, I show that patterns of behavior learned when a legislator serves in the House do not necessarily carry over when they are elected to the Senate. Instead, the unique institutional structure of the Senate forces leg islators to adapt their strategies and become more bipartisan in their cosponsorship choices. Likewise, I find that even the group of hyper cosponsorship behavior once they leave the rout inized and predictable environment of the House for the more fluid and individualistic Senate chamber. Ultimately, the goal of this dissertation is twofold: to (1) emphasize the importance of theorizing about each chamber of Congress independently, and (2) illustrate that, while roll call voting is a very important and visible stage of the legislative process, the choices made at that time are not necessarily representative of the decision making processes used during other stages of policy making.
41 CHAPTER 2 THEORY AND LITERATUR E REVIEW This dissertation is grounded broadly in the literature on historical institutionalism and more specifically in the study of sensemaking in organizations. Relying on the body shape its actors goals and decision and behavior vary across the House and the Senate. In the process, I address two main problems exhibited in research on Congress. First, scholars of the institution have often stud ied the House while neglecting to consider corresponding relationships in the Senate. have been conducted, the most prominent o f which may be Donald Matthe (1960) the transformation of Senate n orms and procedures (1989, 2009) few scholars have compared and contras ted the two chambers together. Th e tendency to focus on the lower chamber likely oc curs because the House, with a much larger membership compared to the Senate, better facilitates large N quantitative analyses. It is also possible that scholars prefer to study the House because the working environment it establishes is much more predict able, making the development of coherent theories much simpler. At its core, the chamber is based on a strict hierarchy, and its members are largely driven by the same goal: their desire for se members to stray outside of behavior in this chamber more straightforward than it might be for a more individualistic and less structured institution, such as the Se nate.
42 The second concern in theorizing about Congress is that, in an attempt to overcome the first problem, researchers often take theories developed to explain legislative behavior in the House and apply them to the Senate. One important example of this has been the attempt to frame the development of party polarization in both chambers in terms of the same theoretical mechanisms. Delving deeper into the literature on party polarization, I explain yet another problem with this literature: the conclusions of scholars in this genre are based strictly on the analysis of one stage of the legislative process: roll call voting. Although anecdotal evidence, usually in the form n provided to bolster the argument that the two major political parties have become more polarized in both chambers of Congress over the last 30 years, it is troubling to note that the only empirical evidence provided in support of this trend focuses speci fically on one isolated stage in the policy process: the vote. W hat this dissertation offers is a test of the argument that polarization has progressed similarly in the House and in the Senate over the post Reform period. To explain these observations, I present a theory which accounts for the unique structure and working environments that exist in each chamber. Specifically, variation in the rules and norms established in the House and Senate cause legislators in each chamber to make sense of their env ironments differently. While House members are able to rely on party loyalty to help them navigate the decisions they must make throughout their two year term, senators are forced to develop more informal decision making processes. In the Senate, strict adherence to party lines is not enough; by virtue of rules such as the filibuster and unanimous consent agreements, and the fluid and unpredictable
43 environment they create, senators must build working relationships with their colleagues on both sides of th e aisle if they are to be successful legislators. Ultimately, I find that the extent of party polarization varies according to the stage in the legislative process under examination (i.e. roll call voting or bill cosponsorship) and, equally important acr oss legislative chambers. By examining bill cosponsorship, a stage of policymaking which is less visible and less widely publicized compared to roll call voting, this research uncovers evidence of bipartisan relationships in the Senate which have not been considered previously in the literature. Moreover, t h is conclusion supports the need for theorizing about each chamber independently. New Institutionalism: Decision M aking across Contexts The idea behind new institutionalism is that political actors do n ot operate in a vacuum. Choices and strategies are not based solely on the wishes of an individual; rather, they are shaped by the environment in which that individual operate s In fact, the ons within new institutionalism historical institutionalists and rational choice institutionalists is that institutions provide the strategic context in which political actors make policy choices. gic choices and thereby shape these authors mention? bill. As will be discussed below there is already a significant literature which illustrates the growing influence that party loyalty has on this stage of the legislative process. Another strategic choice is the decision to cosponsor legislation. This practice, too, has received atten tion from scholars. However, what has been neglected is the study of
44 how the institutional context influences the process of building relationships which subsequently leads to common cosponsorships. This is an area where theories utilizing the new instit choice institutionalists and historical institutionalists presume that organizationally embodied routines play a crucial role in allocating resources and structuring the incentives, options, and constraints procedures, such as committee and staff meetings, hearings, and floor speeches, that kee p each chamber running. All of these activities provide legislators with the which direct these interactions have a direct and significant effect on the people (and parties) each leg islator gets to know. Taking a broader view of institutions, some scholars include in their analysis sets of relationships that persist, although in an inherently conflictual and tension filled way. Institutions may be formal organizations or informal net works. They have shared meetings and relatively stable bundles of resources attached to them. I [Theda Skocpol] take an organizational realist approach to institutions, viewing them as actual patterns of communication and activity, rather than seeing the m primarily as values, norms, ideas, or official rules (Skocpol 1995, 105). Here, Skocpol explains that the essence of an instituti on is captured not only in its rules and norms, but also in the relationships built and interactions engaged in by the member s of the organization. Furthermore, the process by which these interactions are translated into norms involves the socialization of new members. For example, l metaphor, we assume that political actors associate certain actions with certain
45 situations by rules of appropriateness. What is appropriate for a particular per son in a particular situation is defined by the political and social system and transmitted through passed down and shaped over time. Unlike individuals who serve f or relatively short periods of time, institutions typically survive for decades, if not centuries. Therefore, existing rules and norms are the product of the preferences of individuals serving hat has built up over time. Given that the House and Senate institutions and their rules are very different, the patterns of behavior that are passed through the socialization process also vary by chamber. Taking the example of party p olarization in roll call voting as described below, the literature states broadly that the growth in polarization has occurred consistently in both chambers. Given this observation, we have no reason to suspect that the patterns of behavior or socialization are any different in the Senate than they are in the House. However, as will be shown in the chapters that follow, significant differences are This begs the question: if senators, as in dividual political actors, possess roughly the same individual goals of reelection, making good public policy, and obtaining power that House members do, why do we observe differences in their cosponsorship behavior? I argue that the answer lies in importa nt differences in the structure of the two chambers, as well as differences in the strategies of decision making and relationship building that are formed as a result of these structures. Moreover, a theory based on the new institutional perspective is ap propriate to explain this phenomenon because the
46 events, or arrangements rather than on accounting for human behavior without regard to context or modeling very gener al processes presumed to apply at all times and 7). Contrary to the traditional literature on party polarization which relies strictly on the analysis of roll call voting behavior, I find that while bill cosponsorshi p networks in the House mirror the polarization they exhibit in almost evenly split between their same party colleagues and their partisan opponents. One possible indi vidual level explanation is that there is something unique about legislators who run for and are elected to seats in the House compared to those who are elected to the Senate. Perhaps House members are more narrow minded or hard headed or have less politi cal experience and, therefore, less willingness to compromise personality and his or her choices that result from this disposition. However, as I show in Chapter 5, ther e is nothing unique about senators as individuals when it comes to cosponsoring legislation. Comparing the bipartisan cosponsorship behavior of senators who previously served in the House to senators who never served in the lower chamber, I find that ther e are no significant differences between the two groups; senators with previous House service are equally likely to interact with their partisan opponents as are senators without prior House experience. Lacking evidence to support the idea that there is s omething fundamentally different about legislators in the House and Senate as individuals, how can we explain the higher levels of bipartisan cosponsorship in the Senate relative to the House?
47 The answer lies in the constraints created by each institution in which rational actors must operate. It is these differences in the structu re of each chamber which help explain inconsistencies in what would otherwise be rational behavior. In other words, theories developed to understand the behavior of legislators in the House suggest that members consistently make decisions, voting and otherwise, based on their party loyalty. Viewing the bipartisan cosponsorship decisions of senators with this theoretical lens would lead us to believe that senators who cosponsor b ills with members of the opposing party are not acting rationally. However, a new institutionalist approach tells us that individuals who do not seem to be acting rationally may, in fact, be maneuvering around institutional rules or norms while still seek ing to reach their goals. What seems (voting and associating with colleagues based almost strictly on party loyalty), is s imply not rational for senators given the unique institutional con text in which they are placed. Unlike House members, senators operate in an environment where the key to supporters, including colleagues from both sides of the aisle. The motivations for the development of these bipartisan networks are Senate rules which incentivize bipartisanship. Senate rules, particularly the filibuster and unanimous consent agreements, require senators to build bipartisan coalitions in order to pass legislation. Referring to the norm o f reciprocity broader structure of the institution make the Senate unique: It is natural to think of senators huddling toget her in the well of the chamber and making deals with one another. The Senate is, above all, a chamber of
48 individuals. The Senate has few rules, and routine legislative business is usually accomplished through unanimous consent agreements. A single senat or can tie the Senate up in knots and a minority (forty one) can sustain a filibuster even in the face of a determined majority. Without a commitment to reciprocity, the Senate could well come to a complete halt The House, on the other hand, is more majo ritarian. It has an elaborate set of rules that in recent years has been increasingly used to restrict minority party rights. The House membership is four and a half times as large as he two year election cycle means that members have less time to chat with their colleagues since they must spend more time courting the folks back home (2000, 35 6). whic h encourages the formation of bipartisan friendships based on the norm of reciprocity. It is reciprocity and the trust it produces which give senators the confidence and security to build relationships with their colleagues across the aisle. Rather than prioritizing party loyalty over all else, as is the practice in the House, the Senate environment dictates that rational actors pursuing their individual reelection, policy, and power goals will find it necessary to build strong ties with their partisan op ponents, as well as their same party colleagues. In the remainder of this chapter, I review the literature on the differences in House and Senate structures. I also explain what scholars of party polarization have found when they conducted analyses using data on roll call voting patterns. Given the theory set out above, I situate my examination of bill cosponsorship behavior in the context of the relevant literature and explain why bill cosponsorship and the personal relationships and networks built durin g its practice are important to understanding differences in legislative behavior in the House and Senate. Specifically, I look to a theory of organization al behavior known as sensemaking to explain how legislators in each chamber learn strategies for suc cess through the socialization process.
49 The Institutional Structure of Congress: House and Senate Differences Given the argument that environmental context influences individual behavior, it is important to review differences between the institutional stru ctures of the House and Senate. Within scholarship on Congress, there is a large body of literature emphasizing important institutional differences between the House and Senate. Much of this work explores how differences in the structure, rules, and norm s of each chamber produce diverse outcomes in the behavior of their members. Chamber Size First, the most basic difference between the House and Senate, their size, may influence the relationships built among legislators within each chamber. With 435 memb ers of the House, it is nearly impossible for legislators to know or even recognize all of their colleagues; unless members serve on the same committee, are from the same state, or have similar qualities in common, two individuals may never meet. In contr ast, with only 100 members of the Senate, it is easier for senators to know each other on a more personal level. their ability to interact with other members through face to f ace contact, with even th e senior and freshmen members communicating frequently. In addition, the size of each body affects the way its committees operate. Because the House is so large, there are usually more than twenty people serving on a single House committee, while in the Senate that number is typically smaller. Another contrast of the committee systems is that most work in the House is done in subcommittees, while most work in the Senate is done in full committees or on the floor (Baker 1995). With 435 members, the bulk of legislative work cannot be done on the floor of the House; because of this, the institution is forced to subdivide and delegate its
50 tasks to the appropriate committees. In doing so the House has developed a specialized committee system and strong par ty organizations, as well as a formal system of rules and procedures (Dodd 2002). Indeed, the institution is organized (Carmines and Dodd 1985). Although the roles of H ouse and Senate committees are identical, membership on prestigious and powerful committees in the House gives a representative more institution. Ross Baker (1995) assert s that membership on a prestige committee in the Senate does not define who you are the way that it does in the House. Because a coveted as they are by their House cou nterparts (Baker 1995, 62). As a consequence, seniority matters much more in the House than it does in the Senate giving individual rank and file senators more power and influence in the chamber relative to their House counterparts Finally, the differen ce in chamber size also affects the physical organization of each institution. In the Senate, each member has his or her own desk. However, in the House, the use of individual desks was given up years ago when the size of the membership grew too large to accommodate everyone. Similarly, the higher status of the Senate is reflected in the greater number and quality of amenities and size of office spaces given to senators in comparison to representatives (Baker 1995). Length of Terms Because a single term in the House lasts only two years, House members spend most of their time focused on winning reelection rather than investing their time in
51 building lasting relationships with one another. The connections that are made between House colleagues tend to exi st between members of the same party; this occurs, in orientation sessions to committee meeting seating arrangements, encourages interaction among fellow partisans and provides few opportunities for members of opposing parties to establish working relationships (Lipinski 2009, 344). In contrast, the six year terms of senators give them time to establish more meaningful connections with one another. Additional anecdotal evidence exi sts which supports these claims. During interviews with former Senate Majority Leader Tom Daschle and Florida State Senator Nan Rich, both legislators commented on the influence that chamber size and length of term have on the ability of legislators to bu ild relationships with one another. 1 In particular, Sen ator Rich commented that the Florida Senate is characterized by a much more bipartisan atmosphere than the House due to the smaller size of the Senate. Likewise, Sen ator Daschle suggested that the lo nger terms of service of U.S. senators gives them more opportunities to get to know one another, personally. In the House, he said, legislators are forced to focus on their campaigns rather than on networking with their colleagues. Moreover, compared to past congresses, legislators in the House today have little to no interaction with members of the other party. Rules and Procedures Due to its large size, the House has been designed as a highly structured, hierarchical institution that relies on establis hed rules and procedures to maintain 1 The interviews with Sen. Daschle and Sen. Rich were conducted on January 20, 2010 and November 30, 2009, respectively.
52 efficiency and order. Over time, as the House has grown in size, its membership has become more selective (turnover has declined and incumbency success has risen), it has become more complex internally (committees have gained autonomy and the number of leadership positions available has increased), and it has developed universalistic norms and practices (clear codes of conduct apply to everyone) (Polsby 1968). The Senate, on the other hand, is a much more informal, flu id, and individualistic chamber (Baker 1995; Dodd 2002). In the House, the Speaker may choose to re cognize a member to speak, or he may ignore the individual. In the Senate, it is easier to gain the opportunity to speak on the floor because members who ar e ignored can cause big problems later by stalling the legislative process. Furthermore, House rules limit the time a member can speak ). Perhaps the greatest procedural difference is that senators enjoy the use of the filibuster as a legislative tool. Any member, no matter how inexperienced or how veteran, can use this technique for his or her own purposes. As a he possibili ty of filibusters encourages the Senate to seek consensus whenever possible and to conduct business under the terms of unanimous consent agreements that limit the time available for debate and amending Another advantage held by senators is that they are not burdened by germaneness rules. In the House, a bill must be presented to the Rules Committee before it can be advanced to the floor. Rules Committee members have the opportunity to decide the formal procedure for handling the bill o n the floor and write rules that limit the types of amendments that can be offered. In the Senate, however, there is no
53 Rules Committee, so members are able to have a greater influence on legislation once it reaches the floor. By introducing an amendment Senate vote on them, even if they have not been studied and evaluated by the As an alternative, party leaders can decide what the floor procedure will be, and members must agree through unanimous consent to these stipulations (Baker 1995). Again, this process gives more power to individual senators than is given to representativ es. Most importantly, the rules and procedural features unique to the Senate force senators of both parties to work together on a regular basis in advance of bringing legislation to a vote. Unlike members of the House, senators face obstacles, such as the filibuster, unanimous consent agreements, and non germane amendments, which the process of legislating in the Senate requires more than twisting arms and enforcing part y loyalty during roll call voting; it necessitates the development and maintenance of functional working relationships among a variety of senators (Sinclair 2009). Party Loyalty Finally, differences in the power structure in each chamber suggest that House members may have more incentive than senators to toe the party line. According to the theory of Conditional Party Government, assertive leaders of the House majority party actively influence the behavior of rank and file members (Aldrich and Rohde 2001; Rohde 1991). This is due, in part, to the relatively rigid hierarchical structure of the House (Smith and Gamm 2009, 144), as well as to changes made to House rules during the 1970s, 1980s, and 1990s. During this period, individual legislators ceded
54 powe r to their party leaders to make the legislative process more efficient. Over time, norms of seniority were replaced by an emphasis on party loyalty; to obtain power in the House, a legislator had to show that s/he was committed to the party organization (Aldrich and Rohde 2001; Cox and McCubbins 2005; Rohde 1991; Theriault 2008, 53). In comparison, the power held by party leaders to coerce and alter the behavior of the rank and file is much weaker in the Senate (Lazarus and Steigerwalt 2009, 350 352). Th 144). new member orientation in which he highlight ed the differences between party leadership in the House and Senate. In his speech, Byrd notes that the positions held by Senate par ty leaders are relatively new and grant little real power compared to the leadership in the House: In this now 208 year old institution, the positions of majority and minority leaders have existed for less than 80 years. Although the positions have evolved significantly within the past half century, still, the only really substantive prerogative the leaders possess is the right of first recognition before any other member of their respective parties who might wish to speak on the Senate Floor. Those of you who have served in the House will now have to forget about such things as the Committee of the Whole, closed rules, and germaneness, except when cloture has been invoked, and become well acquainted with the workings of unanimous consent agreements. Those o f (Byrd 1996). Moreover, Byrd instructs former House members that they should forget what they learned in that chamber folkways.
55 Traditional Explanations of Party Polarization in Congress This review of the inter chamber differences in the power of party leaders speaks to a broader discussion about inter party diff erences. A development which is often lamented publicly by politicians and members of the media is the increase in party polarization, or the growing ideological gap between the parties. The traditional view of polarization in Congress states that legisl ators in both the House and Senate have become more sharply divided along party lines over time (Fleisher and Bond 2004; McCarty, Poole, and Rosenthal 2006; Poole and Rosenthal 1984; Theriault 2006). Some scholars have explained this trend by pointing to p artisan realignment and their ideology (Jacobson 2000; Butler 2009; Theriault 2008). Others suggest that, in recent years, voters have elected fewer moderate candidat es to Congress, choosing replace to their party (Fleisher and Bond 2004; Poole and Rosenthal 1984; Theriault 2006). A third perspective states that societal changes such as the growth of income inequality, advanced party polarization in both chambers by placing the interests of low and high income citizens, represented by the Democratic and Republican parties, respectively, in direct competition with one another (Garand 2010, 11 10; McCarty, Poole, and Rosenthal 2006). This encouraged representatives from each party to leave the ideological middle in favor of the extremes of the spectrum. Although other explanations for the growth of congressional party polarization do exist (see for example, Carson et al. 2007; Fiorina, Abrams, and Pope 2004; Ladewig 2010; Masket 2007; Theriault and Rohde 2010), what these theories have in common is
56 used most commonly was created by Keith Poole and Howard Rosenthal (and later contributed to by Timothy Nokken) and is presented in Figures 2 1 and 2 2 in the form of first dimension Nokken Poole DW NOMINATE scores. Essentially, using a method of ideal point estim based on their pattern of roll call voting. 2 Like the more commonly used Poole and Rosenthal scores, the Nokken Poole values range from 1 (most liberal) to 1 (most conservative). Figures 2 1 and 2 2 tell the traditional story of party polarization: the parties in both the House and the Senate have diverged over time. This is indicated by the gradual separation of the two modal peaks (representing liberals on the left and conservatives on the right). Although during the most recent Congresses the two parties in the Senate were not divided as markedly as they were in the House, it is clear that liberals and conservatives in both chambers drifted apart. Virtually all arguments about the de velopment of party polarization in Congress are made based on these data. In sum, the literature has introduced a litany of hypotheses to explain a phenomenon (party polarization) which they have observed within only one stage of the legislative process. This begs the question: If polarization is so prevalent within Congress, do we also observe a growing division between the parties across other activities performed by its members? 2 Because the analysis in th is dissertation requires that ideology scores be compared (and exhibit variation) across time, the Poole and Rosenthal W NOMINATE and DW NOMINATE scores cannot be used. However, the Nokken and Poole DW NOMINATE scores allow for such an analysis. For an e xplanation, see Ladewig 2010, p. 503. To obtain the data, visit: .
57 An Alternative Measure of Party Polarization As an alternative to studying congressional voting behavior, I examine the cosponsorship. The importance of bill cosponsorship has been well established (see, for example, Campbell 1982; Fowler 2006a, 2 006b; Kessler a nd Krehbiel 1996; Koger 2003). The ability to assemble and maintain a network of colleagues on whom a legislator can draw support often determines capacity in the policy making process (Lipinski 2009). Legislators invest much time re cruiting colleagues to 1982), a practice which makes bill cosponsorship an important indicator of established relationships between legislators. Cosponsorship requires so cial interaction and collaboration among members (Bratton and Rouse 2011 ; Cho and Fowler 2010; Lipinski 2009) and can indicate that legislators trust one another (Fowler 2006b). Perhaps most importantly, cosponsorship is a sign that members of Congress ha ve chosen to work together during the policy process, rather than having this decision made for them, for example by appointment to a common committee. Cosponsorship can also be used as a signaling device to influence other legislators (Kessler and Krehbi el 1996), and can be affected by a variety of individual, electoral, and institutio nal factors (Bratton and Rouse 2011 ; Cooper and Young 1989; Koger 2003; Garand and Burke 2006; Platt and Sinclair Chapman 2008; Rocca and Sanchez 2008). By measuring the int eractions of legislators, scholars have sought to explain the more personal and previously unquantifiable factors which influence legislative productivity. Based on measures of these networks, researchers have concluded that Congress possesses the qualiti
58 are densely connected (Cho and Fowler 2010; Watts and Strogatz 1998). Further work cosponsorship are more successful i n passing their amendments to bills (Fowler 2006a, 2006b; Cho and Fowler 2010). Moreover, research suggests that the cosponsorship behavior of legislators varies with qualities such as partisanship, seniority, and service as a committee chair or in the pa rty leadership (Koger 2003). Comparing the two chambers, most House members are typically connected through cosponsorship to less than 25% of their colleagues in the chamber while most senators are connected to more than 75% of their fellow senators. This the Senate cosponsorship network is much more densely interconnected than the Senate relative to the House, networks scholars have failed to examine inter c hamber differences beyond this observation Similarly, studies of cosponsorship networks illustrate that while polarization in the Senate has diminished, in the House it remains high relative to past leve ls (Zhang et al. 2008). However, this conclusion i s limited and requires an explanation. Using bill cosponsorship as an indicator of party polarization is different from analyzing roll call votes because cosponsorship measures the personal connections legislators make with one another, rather than their p olicy positions. Moreover, bill cosponsorship patterns paint a picture of the network of individuals with whom a legislator chooses to interact throughout each Congress. To measure party polarization within the process of bill cosponsorship, this dissert ation explores the extent to which legislators cosponsor with their fellow partisans compared to their partisan opponents
59 If the pattern of party polarization observed in roll call voting holds for patterns of bill cosponsorship across both chambers of C ongress, we would expect to see legislators cosponsoring bills much more frequently with their fellow partisans at the expense of collaborating with members of the opposing party. Studying legislative behavior from this perspective reveals conclusions diff erent from those drawn by the traditional polarization literature: Although members of the House have become more polarized in their cosponsorship behavior over the last 30 years, choosing to associate through bill cosponsorship with a larger number of the ir same party colleagues relative to their colleagues on the other side of the aisle as well as cosponsoring more bills with their partisan colleagues compared to their partisan opponents, members of the Senate have not. Instead, senators tend to maintain cosponsorship networks in which approximately half of their connections are made with 50 and the sensemaking process in which they engage when they enter the institution. Having outlined the literature which guides this analysis, I return to a more detailed discussion of the process of sensemaking as it is applied to members of the House and Senate. Sensemaking in Organizations The theory presented in this dissertation relies heavily on the work of working environments of the House and Senate. The literatu re on sensemaking conceives of the phenomenon in different ways, but generally, scholars argue that individuals rely on sensemaking most when they are placed in an unfamiliar situation;
60 ed point ortantly, argues Karl Weick, sensemaking should be understood as a process rather than as an outcome. Namely, it is retrospective, social, ongoing, grounded in identity construction, focused on and by extracted cues, and driven by plausibility rather tha n accuracy (Weick 1995, 17). Alternatively, sensemaking can also be understood as a social activity. Relating this concept to the experiences of newcomers joining and being socialized into an events that may be discrepant from predictions. Discrepant events, or surprises, trigger a need for explanation, or post diction, and, correspondingly, for a process through which s that when individuals enter a new environment that is not well defined for them, they will rely on sensemaking to understand each new (and perhaps surprising) event they encounter. depending on other words, the more random or the less predictable a scenario, the more an individual rong influence on the manner by which individuals within organizations begin processes of transacting with
61 members to understand and to share understandings about such featur es of the organization as what it is about, what it does well and poorly, what the problems it faces are, and how it should resolve them (Feldman 1989, 19). This indicates that sensemaking is an important aspect of how members of an organization build re lationships with one another and come to understand the goals and strengths of the group to which they belong. This discussion is an introduction to why congressional scholars should develop theories which account for the unique structures of the House and Senate. In the remainder of this chapter, I present my interpretation of the process of sensemaking specific to the House and Senate which is the approach used to guide the empirical analyse s in the chapters that follow. Sensemaking in the House and Sena te It is appropriate to use this concept to help us understand the behavior of legislators in the House and Senate precisely because election to Congress places individuals in an unfamiliar environment within an organization where they must build relations hips with their colleagues and gain a clear understanding of the goals and strengths of the institution. Specifically, work on sensemaking suggests that the consistent framework which characterizes the House (i.e. its hierarchical structure, emphasis on t he connection between party loyalty and individual success, and its strict rules) allows representatives to interpret social relationships more easily. In contrast, because the Senate is a much more fluid and less hierarchical institution, it is much more difficult for senators to develop routines that can help them interpret their surroundings.
62 When considering the process of sensemaking in each chamber, it is important to understand how the daily work lives and concerns of House members differ from those of senators, and how these differences translate into different patterns of relationship building and networking through bill cosponsorship. First, b ecause all legislators in the House are required to face reelection every two years, they have a common e lectoral focus; when the campaign season arrives, all House members running a campaign divert their attention to the goal of reelection. This introduces consistency and predictability to the House chamber. Similarly, because the House is charged with cra debates, leading to the formation of a routine This makes the House world regularized and predictable, structured by the rules of the institution. This predictability leads to t he development of generic roles for members and a relatively straightforward sensemaking process. In contrast, senators have multiple roles and are assigned a variety of tasks. This gives them the freedom to take their own innovative and unique approach t o lawmaking. Although the Senate shares with the House the duties of writing legislation and performing oversight, senators also must debate the ratification of treaties and confirm presidential nominations and appointments (Dodd 2002). All of these duti es representatives have an easier time developing expertise in specific policy areas (Baker 1995). While the House typically provides the detailed expertise in the legislative process, the Senate focuses more on the broad concepts of national policies (Carmines and Dodd 1985). House members devote much attention to the local interests of their
63 districts and link the people to the government, while senators typically have a bro ader, national perspective of politics in which they have a consensus on the national interest. All of these factors contribute to the regularized and predictable routine of the typical House member and the more fluid and individualistic nature of the lif e of a senator. Moreover, the staggered, six year terms of senators means that the priorities of the members vary according to how distant their reelection campaign is; those who must mount a campaign in the upcoming election will be driven more by their d esire to please their constituents, whereas those who have most recently been reelected may be more concerned with other goals, such as policymaking or obtaining a place of power within the chamber. Rather than creating a predictable and consistent enviro nment where members can rely on the security of their decision making rules (as is done in the House), these factors force senators to move with the tide of the particular political scenario and place their trust in their relationships with their colleague s, rather than the security of their work environment and pre defined decision making processes. One consequence of th ese differences is that House members typically rely on belief driven sensemaking, while senators use action driven sensemaking (Weick 199 5; Dodd 2002). With belief driven sensemaking, members of an organization have a common belief system and do not need to constantly reevaluate how relationships work (Dodd 2002). In the House, this common belief system is encompassed largely by the sense that they can achieve their personal, policy, and reelection goals by placing loyalty to party and constituency service above all other concerns. Compounding this interpretation of their environment is the fact that all members of the chamber must face r eelection after two years of service. The predictability of this scenario makes House
64 members feel more secure in their environment and, as a result, makes them more likely to rely on the processes of belief driven sensemaking they developed during their earliest years in office. In comparison, senators must rely on action driven sensemaking because their particular working environment is much less stable and predictable. Senators utilize ongoing adaptive learning because it is difficult for them to gain reinforce this learning process are the existence of staggered six year terms, the discrepancies in population sizes of senate constituencies, the s poradic involvement of senators in presidential campaigns, and the variety of opportunities to obtain power and Overall, t he Senate is characterized by complexity, confusion, and uncertainty (Dodd 2002). Because of the more tumultu ous and unpredictable environment created in the Senate, members of the chamber are not able to develop clear decision rules about the choices they should make in daily Senate life. Navigating the legislative process in the Senate is not as simple as rely ing on constituency service and party loyalty to get ahead. Instead, the job of a senator requires extensive social networking and intra group communication to improvise new social relations hips and governing strategies. In his book on Senate friendships Ross Baker explored the phenomenon Unlike institutional norms which are received standards, enforced by the frowns of disapproval of senior members, institutional kinships derive from the experience of s enators and are established by living senators for the advantage and convenience of living senators. They depend not on tradition but on current usage. Based upon a track record of trust derived from shared experiences, these peculiar friendships enable senators to
65 save time in gathering information and cues and reduce transaction costs in their dealings with colleagues. Life in the Senate is simply easier in the company of a few trusted colleagues who can supply cues and information on issues with which you are unfamiliar and, of even greater importance, to give timely warning of impending threats (Baker 1999, 7). From this perspective, forming friendships in the Senate is a tool used to save time and make the duties of daily life easier. As we have see n throughout this literature review, t associate and build relationships with members of both parties because it i s in their interest to do so. Just as we can consider a strict adherence to party loyalty to be the chosen sensemaking device for members of the House, we can understand the formation of bipartisan relationships and reciprocity to be the sensemaking device most appropriate for senators. This reliance on rel ationship building, or networking, is what this dissertation explores in detail. Sensemaking and Bill Cosponsorship Delving further into the process of sensemaking in each chamber, I elaborate on the various stages in this process and explain how inter cha mber differences in the p rocess of sensemaking translate into inter chamber differences in cosponsorship networks and behavior. First, as discussed above, the Senate is a more individualistic institution; senators have more individual power, which gives t hem more freedom (compared to House members) to vote and behave according to their individual ideological or policy preferences, rather than according to what the party leadership (or the institutional power holders) want. This freedom comes from the uniq ue institutional structure of the Senate including the institutional rules, such as unanimous consent agreements and the filibuster, on which I have already elaborated.
66 Second, because senators tend to have more individual freedom, it is possible for them to maintain positions as ideological moderates. Rather than being heavily incentivized to toe the party line (as are House members), senators have enough individual power and security to break ranks with their party, often during roll call voting, but esp ecially throughout the much longer process of writing, cosponsoring, and cooperating on policy making that precedes the actual vote. Examples of senators who take advantage of this freedom include Olympia Snowe (R ME) and Richard Lugar (R IN), Max Baucus (D MT) and Ben Nelson (D NE) Other interesting cases include the well known bipartisan pairs of senators who cooperate, sometimes for years, on writing legislation despite the fact that they are partisan opponents; examples here include the bipartisan pa irs mentioned in Chapter 1, particularly Ted Kennedy and Orrin Hatch, and Pete Domenici and Paul Wellstone. Yet another factor in their ability to form these enduring bipartisan relationships is the length of their terms, discussed above. While it is tru e that there are fewer ideological moderates serving in the Senate during recent Congresses (as exhibited by the more polarized distribution of roll call votes in recent years), those senators who do tend to be more moderate have the independence and indiv idual clout to cosponsor, interact, and vote how they like, largely due to the unique structure of the Senate. The point here is not that the Senate is not polarized; the evidence in terms of roll call voting is very clear. Instead, my argument is that th structure allow senators who do want to be bipartisan to actually do it; in the House, representatives have virtually n o freedom to behave in this way The evidence of this is the observation, discussed throughout the chapters that follow, that the cosponsorship
67 networks of House members tend to be composed primarily of relationships between same party colleagues, while cosponsorship ties in the Senate tend to be made almost to the same extent between partisan colleagues and pa rtisan opponents. Why, then, do we observe the discrepancy between the increasingly polarized pattern of Senate voting and the maintenance (for the most part) of relatively moderate Senate cosponsorship networks? I explore this question in more detail in the concluding chapter of this dissertation, but a brief explanation may relate to the visibility of the vote versus the relative invisibility of bill cosponsorship (and other stages of the legislative process). It is possible that the increasingly polari zing political atmosphere, both within Congress but particularly in the media and the public, pressures all legislators to go along with the tide. Because roll call votes are the most visible ices are routinely attacked in the media and during political campaigns, all legislators, even senators, may feel obligated to toe the party line when it comes to their roll call voting decisions in order to avoid condemnation and retaliation from con stitu ents and interest groups. However, it is within the less visible or publicized stages of the legislative process, such as committee markup sessions, personal conversations with colleagues on the floor or in the dining room, and through bill cosponsorship w institutional norms of individualism exert their strongest influence. Here shielded from making processes are most shaped by institutional traditions of individualism, recipro city, and cooperation. In comparison, members of the House know that institutional norms of party loyalty govern all of their behavior, whether publicly visible or not, so their pattern of working only with
68 fellow partisans is expressed through roll call voting as well as their more private work activities. In sum, by identifying the similarities and differences between the House and Senate, we are able to see that the two bodies are inherently different. The importance of this revelation is that we must not attempt to view one chamber with lenses created to understand the other. Comparing the types of relationships built by members of the House to those created by senators can help us understand how legislators in each chamber make sense of their environ ment. By taking a new approach to the study of Congress, specifically one which accounts for the structural, procedural, and behavioral differences of the two branches of the institution, we can begin to develop a more accurate understanding of each chamb er of the national legislature. In the following chapter, I turn to an explanation of the variables used in the analyses throughout this dissertation.
69 Figure 2 1 Distributions of Nokken Poole DW NOMINATE scores in the Hou se, by Congress
70 Figure 2 2 Distributio ns of Nokken Poole DW NOMINATE s cores in the Senate, by Congress
71 CHAPTER 3 MEASURING COSPONSORS HIP NETWORKS AND BEH AVIOR The traditional story of polarization in Congress during the Post War period is a familiar one: members of the House of Representatives and the Senate have become sharply divided along party lines. However, this conclusion relies almost exclusively on the analysis of legislative voting behavior. Chapter 2 present ed the traditional argument on party polarization: that the two major parties have been polarizing in terms of their voting behavior in both the House and the Senate over the last 30 40 years. In Chapter 3, I will turn to answering a question which has no t yet be en considered in the literature : What does party polarization look like within other types of legislative behavior? The purpose of this chapter is to begin to explore the connections made by members of Congress through bill cosponsorship. I will among legislators through their cosponsorship of legislation. I will explain how each measure is created and present descriptive st atistics on these variables. Finally, I will condense these variables into two composite measures of cosponsorship relationships. Overall, this chapter will present a detailed explanation of the construction of the dependent variables that will be used i n the analyses presented in later chapters. decision to cosponsor legislation captures a much more personal aspect of the legislative process. Often performed when one legislat or asks another for support, the act of cosponsorship signifies that a relationship exists between legislators. What is important in this context is not the type of bills which a legislator cosponsors, but whether or not the individuals with whom s/he cos ponsors are fellow partisans or
72 bills paints a picture of the types of interactions, either partisan or bipartisan, in which s/he tends to engage over time. To measure t he extent of party polarization exhibited through bill cosponsorship in the House and Senate, I focus on whether legislators form cosponsorship ties with their fellow partisans or with their partisan opponents This measure estimates the types of relation ships (either same party or opposite party ) maintained by each legislator. I assume that legislators in a highly polarized chamber will tend to associate with and create more cosponsorship ties with their fellow partisans, while legislators in a less pola rized chamber will tend to have a more bipartisan network and make similar numbers of cosponsorship ties with members of both parties. Overall, the analysis examines the factors which may contribute to varying levels of same party ties in bill cosponsorsh ip across legislators and over time, during the 96 th through 110 th Congresses (1979 2008). Studying legislative behavior from this perspective reveals conclusions different from those drawn by the traditional polarization literature: Although members of th e House have associated more frequently with their partisan allies through their cosponsorship behavior over the last 30 years, members of the Senate have not. Instead, senators tend to maintain cosponsorship networks in which approximately half of their connections are made with fellow partisans and half are made with partisan context in which senators operate. In presenting this argument, I explore several measures of
73 legislative chamber. Data and Method The data used to create the measures of bill cosponsorship come from the work of James Fowler (Fowler 2006a, 2006b). 1 Using the Library of Congress Thomas legislative database, Fowler and his colleagues collected data on all public and private bills, resolutions, and amendments introduced in the House and Senate durin g the 93 rd through 110 th Congresses, or from 1973 through 2008. This series includes all of the years accessible on the Thomas database at the time Fowler and his colleagues collected the data. For each case, the researchers coded a number of pieces of i nformation, including the name of each bill, the Congress number and date during sponsor, and any cosponsors who signed on to the bill. While private bills and amendments re ceive fewer cosponsorships relative to public bills, Fowler and his and a cosponsor contains information about the degree to which legislators are socially eral, the observation that a piece of legislation of any type has a Moreover, b ecause bill cosponsorship was not allowed in the House until the 91 st Congress (1969 70) and was limited to 25 cosponsors per bill until the 96 th Congress (1979 1980), I have limited my analysis to the 96 th 110 th Congresses. Doing so ensures that the rules of bill cosponsorship (and thus, the potential for building relationships 1
74 through cosp onsorship) are the same across all Congresses under consideration. More importantly, considering changes in the cosponsorship networks and behavior of legislators during this period allows us to observe the transition in these patterns as members of the i nstitution moved from the Subcommittee Era into the era of party government. Statistics describing basic characteristics of bill cosponsorship are presented in Table 3 1. In Table 3 1, we see first that senators tend to sponsor more bills, on average, com pared to House members. 2 This may occur because, compared to House members, senators represent much larger districts, have more staff members, and sit on and chair more committees, thus making them more active in the policymaking process (Fowler 2006a, 46 0). However, the average number of bills cosponsored by a legislator does not vary much by chamber: House members cosponsored between 168 and 376 bills, while senators cosponsored between 137 and 376 bills during each Congress. Finally, each bill introdu ced in the House tends to attract more cosponsors than bills introduced in the Senate. This is likely due to the fact that the House is a larger chamber with more people available to cosponsor bills. Considering the relative size of each chamber, bills o n average tend to carry cosponsors from about 3 5 percent of the members in each setting. Perhaps most importantly, given the very large number of bills introduced in each chamber during each Congress, legislators are presented with literally thousands of opportunities to express their support for their colleagues through cosponsorship. 2 The values in Table 3 1 vary from those in Table 1 of Fo article, Fowler and his colleagues have updated the data to make corrections and to include
75 However, seeing as they choose to cosponsor only a few hundred bills (at the most) during each Congress, this suggests that their time is limited and they make their cospon sorship choices carefully. This observation lends support to the idea that bill cosponsorship matters and is a representation of a meaningful relationship between colleagues. Rather than cosponsoring bills at any opportunity, legislators carefully select a small percentage to which they lend their support. Having discussed the data, I Incoming vs. Outgoing Cosponsorship In constructing a measure of the relatio nship established between legislators through bill cosponsorship, it is important to note the direction of the tie. For any given legislator, a cosponsorship can be given or it can be received This point represents the first two faces of the measure: in coming and outgoing cosponsorships. For this analysis, incoming cosponsorships are the cosponsorships received by a legislator from his or her colleagues on bills that the legislator sponsored. In contrast, outgoing cosponsorships are the cosponsorships given out by a legislator in the act of supporting a bill sponsored by his or her colleagues. Each cosponsorship decision represents an individual choice made by a legislator, and these decisions may be made for a variety of reasons. For example, an outgo cy position taken in the bill he cosponsors, or it may be given with the intention of helping a colleague whom the legislator respects and trusts. Likewise, a legislator may cosponsor a bill in order to signal to his colleagues that he may cosponsor a bill that is favored by the leadership in the hopes of receiving some
76 type of perk, perhaps support from a party leader or other influential colleague on one o f support; in other words, from whom does this legislator receive support, or who trusts and respects this legislator? The analysis of the cosponsorshi ps received by each legislator can tell us something about how influential each legislator is. For example, legislator is in a position of power within the party, possibly as a party leader. Alternatively, receiving more support from partisan opponents may indicate that the legislator is someone who frequently cooperates across party lines, thus building trust with colleagues across the aisle and making them more willing t work. Although individual cosponsorships may be given for a number of different reasons, the patterns of these decision making processes created over time paint a blished relationships. Network Partisanship vs. Cosponsorship Partisanship The third and fourth faces of bill cosponsorship address the methods for relationships. To calculate the values for the dependent variable, I constructed a sponsor by cosponsor matrix for each chamber of each Congress in the dataset. In each matrix, cell values indicate how many bills sponsored by legislator i (listed in the rows) were cosponsored by le gislator j (listed in the columns). An example is provided in Figure 3 1. The image presents the first five senators (sorted alphabetically) who served during the 110 th Senate (2007 variable indicating whether t
77 3 Looking in cell D3, we see that Lamar Alexander (a Republican) cosponsored two bills that were sponsored by Daniel Akaka (D). Likewise, cell C5 indicates that Daniel Akaka (D) cosponsore d four bills that were sponsored by Wayne Allard (R). Cell values along the diagonal of the matrix equal 0 because a legislator cannot cosponsor a bill that s/he already sponsored. Using these matrices, we can determine the number of outgoing and incomin g cosponsorships given and received by each legislator. Furthermore, we can also calculate how many of these cosponsorships opponents. This measure captures what I will r cosponsorship partisanship Taking this analysis a step further, we can dichotomize the cell values in these network Rather than focusing on the bil ls being cosponsored, we can center our analysis on the relationships established among the individuals serving in each Congress. In this case, rather than observing a range of values indicating how many times legislator j cosponsored bills written by leg islator i a relationship exists between each pair of colleagues. This measure expresses the types of cosponsorship connections, either partisan or bipartisan, which each legislator tends to make. Through network partisanship Having clarified the construction of the sponsor by cosponsor matrices, it is also necessary to explain the calculation of the dependent variables used in the analyses 3 Independents are alwa ys coded as the party with which they caucus.
78 network partisanship (incoming and outgoing) and cosponsorship partisanship (incoming and outgoing). These four variables are depicted in Table 3 2. First, I define e incoming network as all of the legislators who cosponsored at least one bill sponsored by that legislator during a given Congress. In l of the colleagues from whom he received at lea outgoing network is defined as all of the people to whom the legislator gave at least one cosponsorship. legislator reflec ts in part the degree to which the average member is integrated into the network when legislators have more cosponsors, it may indicate they are operating in 4 The network partisa nship variables calculate the percentage of these cosponsors (either For example, during the 107 th Congress (2001 2002), Representative Nancy Pelosi sponsored 13 bills and received at lea st one cosponsorship from 170 of her colleagues; 154 of these cosponsors were fellow Democrats. This makes her incoming network partisanship value for the 107 th Congress approximately 90 percent. During the same Congress, Pelosi cosponsored 447 bills tha t were written by 180 of her colleagues; 128 of these individuals whose bills Pelosi cosponsored were her fellow Democrats, making her outgoing network partisanship value approximately 71 percent. 4 cosponsors, or cosponsorships received by each legislator (2006a, 462).
79 To find the incoming and outgoing cosponsorship partisanshi p values, we calculate the percentage of cosponsorships (rather than cosponsors) which are received from or given to fellow partisans, respectively. Returning to the example of Nancy Pelosi and the 107 th Congress, we find that, of the 251 cosponsorships r eceived by Pelosi, 234 came from fellow partisans. This makes her incoming cosponsorship partisanship value approximately 93 percent. Likewise, of the 447 bills Pelosi cosponsored, 351 of those cosponsorships went to her fellow partisans, making her outg oing cosponsorship partisanship value approximately 79 percent. To reiterate, the important distinction between these two faces of cosponsorship is that network partisanship measures the type of people (either partisan or bipartisan) with whom each legisl ator tends to associate, while cosponsorship partisanship measures the partisan nature of the bills each legislator cosponsors. 5 To describe the distributions of these variables, I present box plots for each Congress by chamber. The graphs for the two inc oming partisanship variables are shown in Figure 3 2. In the first graph we see that most senators have network partisanship values ranging from about 45 to 65 percent. This tells us that 45 to 65 percent of the people from whom a senator receives cospon sorships tend to be from his or her own party. While there are some outliers on either side of this range, the pattern is fairly consistent across Congresses. In comparison, the cosponsorship partisanship values are slightly higher, ranging from about 45 to 75 percent. This suggests that although senators tend to build incoming networks close to the 50 50 range (receiving 5 The variables measuring incom ing network and cosponsorship partisanship correlate at 0.97 for the House and 0.77 for the Senate. The variables measuring outgoing network and cosponsorship partisanship correlate at 0.97 for the House and 0.86 for the Senate.
80 cosponsorships from approximately the same number of fellow partisans as partisan opponents), when it comes to the percentage of cospo nsorships they receive, the pattern is slanted slightly more in favor of receiving cosponsorships from members of their own party. With that said, these ranges are much lower than those we observe in the House. In contrast, the percentage of fellow partis ans with whom members of the House tend to associate and from whom they receive cosponsorships is much higher than that of the Senate. In the House, network and cosponsorship partisanship values are similar to one another and range from about 60 to 90 per cent of their incoming cosponsors and cosponsorships. 6 This indicates that, for House members, much more than half of the people with whom they associate (and more than half of the cosponsorships they receive) come from their fellow partisans at the expen se of building relationships with their colleagues from the opposite side of the aisle. Finally, for every Congress in the series, the top quartile of House members includes legislators who receive cosponsorships exclusively from their fellow partisans (i .e. the top quartile includes people with 100% partisan cosponsors and cosponsorships). For the Senate, this is only true of the cosponsorship partisanship variable during the 110 th Congress (2007 2008). Turning to the distributions for outgoing cosponsor s and cosponsorships, we see similar box plots presented in Figure 3 3. For these values in the Senate, network partisanship ranges from about 45 to 70 percent, while cosponsorship partisanship is 6 In the House graph, the o utliers with zero percent of their cosponsors and cosponsorships from fellow partisans tend to be legislators who received between one and three cosponsorships on bills they sponsored, all of which came from members of the opposing party.
81 again slightly higher, between 45 and 80 percent. In the House, the values for both variables are again similar, falling between 45 and 90 percent. Again, the decisions by legislators in the House in terms of whose bills they prefer to cosponsor skew much more in favor of their fellow partisans compared to the behavior of senators. To begin to unfold an explanation of these inter chamber differences, I now turn to explain the construction of the composite cosponsorship variables and an elaboration on the differences between roll call voting patterns and the ext ent of party polarization in cosponsorship decision making. A Composite Measure of Bill Cosponsorship Ties While recognizing that the direction (incoming or outgoing) of each bill cosponsorship decision is important to consider, this particular question is not the primary focus of this dissertation. Rather, what I am most concerned with illustrating and explaining is the partisan or bipartisan nature of the complete set of ties entered into by each legislator. The bulk of the dissertation will set aside t he question of the direction of each tie in favor of painting a more complete picture of these relationships. incoming and outgoing ties separately, I will focus most of the analysis on all of a outgoing and incoming cosponsorships separately (see Chap ter 6). In conducting this composite analysis, I maintain the distinction between the third and fourth faces of bill cosponsorship (network partisanship and cosponsorship partisanship) and combine the first two faces of bill cosponsorship (the incoming and outgoing ties). This gives me two composite measures: total network partisanship and
82 total cosponsorship partisanship To create the measure of total network partisanship I calculate the percentage of partisan cosponsors by adding the number of incomin g and outgoing partisan cosponsors and dividing this value by the total number of incoming and outgoing cosponsors. Likewise, to create the measure of total cosponsorship partisanship I simply add the number of incoming and outgoing partisan cosponsorshi ps and divide this by the total number of incoming and outgoing cosponsorships. 7 Having explained the measurement of the key dependent variables, it is now necessary to return to the question introduced in the preceding chapter: that is, given the prevalen ce in the literature on party polarization of the argument that the two major parties in Congress have become extremely polarized over the last 30 years, does this trend extend beyond roll call voting to other types of legislative behavior, particularly bi ll cosponsorship? Observing the distributions for the composite partisanship variables gives us the first piece of our answer. First, when we graph the distributions of the total network partisanship variable for each chamber, we find that members of the parties in the House have polarized, while those in the Senate have not. In the figures below, the x axes range from 0 ). Taking each chamber in turn, we see in Figure 3 4 that the peak of each distribution (or the mode of the data) gradually moves further and further to the right over time. During the 96 th through 102 nd Congresses, the 7 As an alternati ve composite measure, I tried taking the average of incoming and outgoing network partisanship as well as the average of incoming and outgoing cosponsorship partisanship. These values correlate with total network partisanship and total cosponsorship parti sanship at 0.94 and 0.935, respectively. I choose to use the variables for total partisanship because they are, theoretically, a more appropriate depiction of the sum of all ties in which a legislator is engaged.
83 distributions appear to be slightl y bimodal. This suggests that a portion of House members maintained more bipartisan cosponsorship networks; of all the legislators to whom members of this group were connected through cosponsorship, about 50 percent were fellow partisans and 50 percent we re partisan opponents. At the same time, there also appears to have been a (slightly larger) group of legislators serving in these Houses which maintained more partisan cosponsorship networks. Of the cosponsors connected to these individuals, about two t hirds to three fourths were members of their own party. However, beginning in the 103 rd Congress, this bimodal pattern disappears. In fact, what seems to happen is the size of the more bipartisan group of House members declines, while the size of the more partisan group increases. By the 110 th Congress, the majority of House members fall easily around the 75 percent partisan network mark. These observations fit with the traditional argument about party polarization during the post War period: the parties in the House have grown further apart, with legislators in the chamber choosing to associate almost exclusively with their fellow partisan colleagues. In the House, bill cosponsorship associations seem to have gone the same way as roll call voting patter ns. The results for the Senate, however, are very different. In this chamber (shown in Figure 3 5), a bimodal pattern exists through about the 103 rd Congress, but the overall percentages are much lower compared to those of the House. In the Senate, the l ess partisan group tends to have cosponsorship networks of which less than half of the members are fellow partisans; members of the more partisan group have networks of which 50 60 percent are fellow partisan. As with the House, the bimodal nature of
84 the distribution disappears during the 104 th 110 th Congresses; however, unlike the House, the new mode for the Senate sits only slightly above 50 percent These observations suggest an important inter chamber difference. Specifically, not moved in tandem with the networks of House members over the last 30 years. Rather than the two modes of the Senate distributions converging at the highly partisan 75 percent range as they did in the House, legislators in this chamber have continued to build and maintain cosponsorship relationships with almost equal numbers of their fellow partisans and their partisan opponents. This point is exceptionally important because it stands in contrast to what we would expect to find given the enormous existi ng literature on congressional party polarization. By observing patterns of behavior other than roll call voting, we have already begun to dig up evidence that the traditional argument may not apply to all types of legislative decision making. Turning now to look at the same distributions for the total cosponsorship partisanship variable for each chamber, we see in Figure 3 6 that the trend for network partisanship in the House is even more exaggerated for cosponsorship partisanship in the House. Over tim e, the percentage of cosponsorship connections with fellow partisans increases significantly. By the 110 th Congress, a majority of House members exhibit behavior in which 75 90 percent of the cosponsorships they give and receive link them to their fellow partisans. Again, it is important to note that the network partisanship ranges for the House (60 80 percent) are slightly lower than the cosponsorship partisanship ranges (70 90 percent). This means that a House s/he associates through cosponsorship, is
85 slightly more bipartisan than the distribution of actual bill cosponsorships in which s/he takes part. From this, we can surmise that, in addition to associating with more of their fellow partisans as cosponsors, House members also tend to cosponsor bills more frequently with these fellow partisans. In other words, relationships between fellow partisans in the House are deeper than relationships between partisan opponents. The concept of the depth of cosponsorshi p relationships, or the strength of cosponsorship ties, will be explored later in this chapter. Moving again to the Senate (Figure 3 7), we find that legislators in this chamber have moved slightly in the more partisan direction over time, but the bulk of the distributions remain in the 50 75 percent range, much lower than for the House. The partisan degree of cosponsorship patterns began inching their way upward beginning during the 100 th Congress when the mode of the distribution sat between 60 and 70 pe rcent. However, even during the 104 th and 105 th Congresses, legislative sessions known for their extreme party polarization, the Senate distributions remained quite even, rather than becoming distributed heavily toward the higher partisan range. In other words, although there were senators for whom 75 percent of their cosponsorships connected them to fellow partisans (making their behavior look much like that of House members), there were similar numbers of senators who maintained 50 50 networks during th is extremely polarizing time. 8 focused on their same party colleagues during the 107 th through 109 th Congresses, it began to moderate again during the 110 th Congress as evidenced by the inc reasing dispersion of 8 This is especially true of the 104 th thr ough 106 th Congresses.
86 the distribution in the less partisan (or more bipartisan) direction. Moreover, even though senators may have begun to engage in bill cosponsorship more frequently with their party allies during these periods, it is important to reme mber that the networks and relationships maintained by these individuals remained moderate. The significance of this is that, despite the increasingly polarized political atmosphere in which senators operated during this period, something caused them to c ontinue to build connections with their partisan opponents and bridge the otherwise widening gap between members of the two major parties. It is this phenomenon with which this dissertation is most concerned. Measuring the Depth of Cosponsorship Relations hips The final facet to the puzzle of congressional cosponsorship relationships deals with a point mentioned above: the depth or strength of the ties between legislators. Because this dissertation is concerned with the relationships established between me mbers of Congress through bill cosponsorship, we can assume that legislators with more cosponsorship ties have stronger relationships than those with fewer cosponsorship ties. To calculate the strength of the ties between fellow partisans, I take the tota l number of partisan ties (incoming and outgoing) and divide by the total number of partisan cosponsors (incoming and outgoing). This gives me, for each legislator, the average number of partisan cosponsorships engaged in per partisan cosponsor. For exam ple, if a legislator was involved in 100 cosponsorships with 50 different partisan cosponsors, the average depth of his partisan ties would be equal to 2 (or 100 divided by 50). This indicates that, on average, the given legislator tended to cosponsor two bills for every partisan cosponsor with whom he had a cosponsorship relationship.
87 Looking at Figure 3 8, we find that the depth of partisan cosponsorship ties in the House is very consistent across Congresses. Throughout the period, the vast majority of House members engage in approximately 1 2 cosponsorships per partisan cosponsor. The black circles in the plot tell us that legislators who cosponsor an average of 3 5 bills with each of their partisan cosponsors are outliers and, thus, rare. Keeping in mind that these are averaged values, and a given legislator could feasibly cosponsor 11 bills with one colleague and only 1 bill with another, (making his average depth of ties equal to 6), this suggests that partisan cosponsorship ties in the House are we ak. In the Senate, we again find a very different result. The left panel in Figure 3 8 illustrates the much stronger ties found between partisan cosponsors in this chamber. Most senators average between 3 and 6 cosponsorships per partisan cosponsor, and the number of senators with much higher averages has been increasing over time. Beginning around the 106 th Congress, we see more senators engaging in more cosponsorships with their same party colleagues (on average); an increasing number of legislators in this chamber have an average tie depth of approximately 5 15 cosponsorships per same party cosponsor. This suggests that ties between fellow partisans in the Senate have been growing stronger in recent years. However, perhaps the most interesting inter c hamber differences are found when we observe patterns of cosponsorship among part y opponents. Using the same technique that was described above to create the measure of average number of cosponsorships per same party cosponsor, I create a measure for the average number of cosponsorships per bipartisan cosponsor. This allows us to observe the depth of
88 relationships between members of opposing parties. Specifically, Figure 3 9 indicates that bipartisan relationships in the House are even weaker than same p arty relationships in the House. In this chamber, most legislators cosponsor about 1 bill, on average, with each of their bipartisan cosponsors. Even the outliers in this case barely reach the level of 5 cosponsorships per bipartisan cosponsor. Again, t his suggests that bipartisan relationships in the House are very weak. In the Senate, bipartisan ties do appear slightly less frequent compared to partisan ties; however, most senators do tend associate with their bipartisan colleagues on multiple bills. Throughout the period, most senators tend to have an average of 3 4 cosponsorship ties with members of the opposing party, and several outliers even stretch into the 5 15 cosponsorships range. This suggests that the picture of interaction between bipartis an pairs of senators is much like that between partisan pairs of senators. More importantly, it indicates that, despite the highly polarized atmosphere of the most recent Congresses, senators continue to talk to, cooperate with, and build connections with their partisan opponents. This finding is significant because it runs contrary to what has been reported in analyses of roll call voting patterns and daily discussions in the media. Rather than shutting down communication with their partisan opponents, senators associate with their colleagues on the other side of the aisle on a repeated basis by supporting and receiving support on legislation in the form of bill cosponsorship. It is this consistent and repeated interaction which, I argue, is the product of the very different institutional environments within the House and Senate. Comparing Roll Call Voting to Bill Cosponsorship As one final test of the difference between roll call voting and bill cosponsorship, I graph scatterplots of the two variables f or each chamber and Congress. As a measure
89 Nokken Poole DW NOMINATE scores used in Chapter 2. Rather than using the raw scores which run along a scale from 1 to 1, I calculate the absolute values of the scores to determine how ideologically extreme (rather than how liberal or how conservative) each legislator is. The relationship between ideology and total network cosponsorship appears in Figure 3 10 for the House an d 3 11 for the Senate. As Figure 3 10 indicates, there is a generally positive relationship between ideological extremity and total network partisanship for the House: members of the House who tend to vote more frequently in the same ideological direction (consistently taking either the liberal or the conservative position on votes) also tend to have cosponsorship networks with higher percentages of fellow partisans. There is little fluctuation in this trend over time, with the exception of the Republican Revolution Era Congresses of the 1990s, which exhibit a slightly more positive relationship between the two variables. Figure 3 11 depicts the same relationship for the Senate. However, in this case, the relationship is only slightly positive. Here, ther e is a positive correlation, but senators who are the most ideologically extreme have cosponsorship networks that are only slightly more partisan than those of the most moderate senators. In particular, there are several Congresses (e.g. the 96 102 and 10 6 108) during which the most ideologically extreme senators (those which fall around the 0.75 and higher mark on the x axis) have cosponsorship networks with a partisan percentage approximately equal to the most moderate senators (around 50 60 percent). T his is important because it suggests that senators who are the most extreme in their roll call voting patterns still
90 manage to maintain cosponsorship networks which are split almost evenly between their fellow partisans and their partisan opponents. This evidence lands strongly in favor of the argument offered by this dissertation. Approximately the same pattern appears when we use the total cosponsorship partisanship variable; however, the inter chamber differences are smaller. I have graphed the relatio nship between ideological extremity and cosponsorship partisanship for the House and for the Senate in the same figure, presented below as Figure 3 12. Here, we see that the relationship between the two variables is positive for both chambers, but for mos t Congresses, the correlation is a bit steeper for the House. Although using the percentages of actual bills cosponsored with fellow partisans (rather than the network percentages) suggests that ideology is a stronger predictor of cosponsorship behavior t han we would have expected given the results in Figure 3 11, there is still very little difference between the cosponsorship behavior of the most ideologically extreme senators compared to that of the most moderate senators. During most Congresses, even t he most ideologically extreme senators continue to associate through bill cosponsorship with their partisan opponents. Having elaborated on the measurement of the key dependent variables, I now turn to a more detailed analysis of the factors contributing t o these inter chamber differences. Specifically, I will explore the differences across legislators in each chamber, comparing party leaders to non leaders and looking at variation across legislators based on the number of years they have served in the cha mber. Later, I will limit the analysis to the Senate and discuss any differences which appear between senators who previously served in the House and those who were elected directly to the
91 Senate. In Chapter 5, I will look even more closely at the influe nce of history and the institutional structure and norms in place during each recent congressional era to determine how these factors influenced the cosponsorship and networking behavior of different cohorts of legislators. These results of these analyses will help us to see and understand how the structures of the House and Senate shape the behavior of legislators in each setting and over time.
92 Table 3 1 Characteristics of bill c osponsorship, 1979 2010 Congress Years To tal bills Mean bills sponsored by each legislator Mean bills cosponsored by each legislator Mean cosponsors per bill House 96 1979 1980 10,396 23 188 8 97 1981 1982 10,180 23 225 10 98 1983 1984 9238 21 300 14 99 1985 1986 8841 20 335 17 100 198 7 1988 8228 18 344 18 101 1989 1990 8522 19 373 19 102 1991 1992 8671 19 342 17 103 1993 1994 7591 17 262 15 104 1995 1996 6722 15 168 11 105 1997 1998 6916 15 222 14 106 1999 2000 8010 18 280 15 107 2001 2002 7663 17 274 16 108 2003 2004 7765 17 2 79 16 109 2005 2006 9374 21 302 14 110 2007 2008 10,613 23 376 16 Senate 96 1979 1980 4194 42 137 3 97 1981 1982 9694 96 219 2 98 1983 1984 11,228 111 293 3 99 1985 1986 7597 76 326 4 100 1987 1988 7782 77 360 5 101 1989 1990 7405 74 376 5 1 02 1991 1992 7686 76 337 4 103 1993 1994 5830 58 232 4 104 1995 1996 8101 80 176 2 105 1997 1998 7001 70 212 3 106 1999 2000 8265 81 289 4 107 2001 2002 8745 87 261 3 108 2003 2004 7804 78 284 4 109 2005 2006 10,160 101 316 3 110 2007 2008 10,327 1 01 348 3 Note: Bills include any bill, resolution, or amendment offered in the House or Senate.
93 Table 3 2 The Four Faces of Bill Cosponsorship Partisanship Measure Direction of Tie Network (Incoming) Cosponsorship (Incoming) Network (Outgoing) Cosponsorship (Outgoing) Figure 3 1 Example of a sponsor by cosponsor m atrix, 110 th Senate Figure 3 2 Distributions of incoming network and cosponsorship p artisanship variables, by c hamber
94 Figure 3 3. Distributions of o utgoing n etwork and c os ponsorship p artisanship v ariables, by c hamber
95 Figure 3 4 Total n etwork p artisanship, by Congress, House of Representatives
96 Figure 3 5 Total network p artisanship, by Congress, Senate
97 Figure 3 6 Total c osponsorship p artisanship, by Congress, Hous e
98 Figure 3 7 Total c osponsorship p artisanship, by Congress, Senate
99 Figure 3 8 Average d epth of p artisan c osponsorship t ies, by Congress
100 Figure 3 9. Average d epth of b ipartisan c osponsorship t ies, by Congress
101 Figure 3 10 Ideological e xtremi ty vs. t otal n etwork p artisanship in the House, by Congress
102 Figure 3 11 Ideological e xtremity vs. t otal n etwork p artisanship in the Senate, by Congress
103 Figure 3 12 Ideological e xtremity vs. t otal c osponsorship p artisanship, by c hamber and Congress
104 CHAPTER 4 SENSEMAKING The data presented in Chapter 3 offer an initial illustration of the argument which is the focus of this dissertation: while patterns of roll call voting among legislators may be very similar in the House and Senate, cosponsorsh ip behavior is not. In this chapter, I will elaborate on why this is the case by dividing the members of each chamber by their institutional roles and individual characteristics. I will begin by comparing party leaders to members of the rank and file in order to explore how changes in the roles of leaders during the Party Government Era of the last 30 years may have evolved differently in each chamber. Second, I will provide evidence for the sensemaking behavior changes the longer they serve in each chamber. This particular analysis will also serve as a stepping stone to the work discussed below which finds that each cohort of legislators adopts the patterns of behavior structured by the rules and norms of the institution in place at the time of their initial election. In other words, knowing when a legislator was elected can help us predict how partisan their networking behavior will be throughout their career. Throughout these discussions, I also consi der the role of factors unique to each ition and the number of bills he or she sponsored and cosponsored during each Congress. How liberal or conservative a legislator is undoubtedly affects his or her l ikelihood of interacting with fellow partisans; we would expect rabid ideologues to be less willing tha n ideological moderates to compromise on policy by cosponsoring bills written by their partisan opponents. o be either an active or a passive policymaker (measured in terms of the number of bills
105 willingness to work with others (number of b ills cosponsored) may also help us predict he aisle when cosponsoring bills. Finally, to control for the distribution of party seats in each chamber, I measure the strength of Congress. Clearly, if there are more mem chamber, there will be more opportunities for the legislator to cosponsor bills with his party allies same party network or cosponsorship percentage. Th ese variables and the results of these analyses are described in detail below. Party Leaders vs. Non Leaders One of the most important changes to take place within the halls of Congress over the last 30 years has been the transformation in the roles of mem bers of the party leadership. As the literature on Conditional Party Government suggests, in the wake of the Southern Realignment (which sorted southern conservatives into the Republican Party and northern liberals into the Democratic Party, unifying each group internally and creating an ideological gap over policy between them), rank and file legislators became willing to cede much of their own individual power to their party leaders ( Rohde 1991 ). One purpose for this was to increase the power of Congres s as an institution which had found itself threatened by the Executive Branch throughout the Subcommittee, or Reform, Era ( Dodd and Schott 1979 ). Another purpose was to increase institutional efficiency by creating a more organized hierarchy of authority, or chain of command, through which the policymaking process could be executed. These efforts resulted in the gradual transformation of the role of party leaders from the figureheads and facilitators who ruled rather passively during the Textbook Congress Era to the
1 06 dominating masters of the chamber who demand party loyalty and have maintained control since the 1980s. W hile these descriptions of institutional transformations are accurate, they largely describe developments within the House during this peri od. This is, in part, due to the fact that studies of party leadership focus largely on the Speaker rather than on the Senate leadership, but it may also be because party leaders in the Senate play a role that is unique to that chamber. As discussed in C hapter s 1 and 2 given the structure of the House and the existence of majority rule practices, a strict hierarchical organization based on loyalty to party goes a long way to increasing efficiency and legislative output. However, the Senate is simply not structured in the same way. Because of the existence of institutional rules and traditions, such as the filibuster and unanimous consent agreements, majority rule is not feasible. Even though the two major parties are unified internally along ideologica l lines, and many Senate partisans may find themselves to be very distant from their partisan opponents in terms of their policy preferences, this does not change the fact that the entire legislative process in the Senate can be stopped by one individual. This is precisely where the traditional literature on party polarization falls short. Unlike in the House where a rank and file legislator possesses little individual power to alter the outcome of legislation, senators have retained the ability to influen ce legislative outcomes single handedly. As a result of this individual power, if Senate party leaders hope to accomplish their policy goals, they must maintain connections to senators on both sides of the aisle in order to forestall complications through out the policy process. This arrangement means that although Senate party leaders have
107 throw a wrench in the system. Despite their elite status, party leaders in this chamber do not have the luxury of associating primarily with their fellow partisans. Perhaps even more so than the Senate rank and file, leaders must make extra efforts to establish bipartisan ties. What this suggests for patterns of bill cosponsorship an d networking is that Senate party leaders do not exhibit the hyper partisanship we might expect from their position as leaders of their party. Instead, they maintain more moderate networks, associating almost equally with fellow partisans and partisan opp onents. In contrast, party leaders in the House feel less pressure to maintain cross party relationships, thanks in large part to the majority rule structure of the chamber. While the House rank and file may need to work with their partisan opponents to build relationships and win favors in order to get ahead in the chamber, party leaders have already achieved this goal. Sitting at the top of the totem pole, their job is to enforce party loyalty, not create bipartisanship. Because of this, we can expect House leaders to exhibit even more polarized networking behavior compared to their rank and file colleagues. With this in mind, I will now explore the data to compare party leaders and non leaders in each chamber. Ideological Extremity and Bill Cosponsor ship Based on the traditional literature, what we expect to find when we look at changes in the roll call behavior of legislators is that members of both chambers of Congress have become more ideologically extreme (or more polarized) over time. Ideologica Poole DW NOMINATE score (explained in detail in Chapter 2). Because network and
108 cosponsorship partisanship are measured as a percentage (rather than on a directional scale), it is of using the strict measure of ideology on the 1 to 1 scale. For the purposes of these analyses of party leader ld the following positions are con sidered to be party leaders: Speaker of the House, President Pro Tempore of the Senate, Majority Leader (House and Senate), Minority Leader (House and Senate), Majority Whip (House and Senate), and Minority Whip (House and Senate). Indeed, when we graph t he distributions of the ideological extremity of legislators, shown in Figure 4 1, we obtain the result we expected. Here, we see first that the overall values for ideological extremity are similar in both chambers. The values for the House are slightly h igher compared to the Senate during the last several Congresses in the series, but this fits with the traditional polarization argument (presented in Chapter 2). Moreover, across time we see the distributions for each Congress move steadily upward, sugges ting that, on the whole, the members of both chambers have become more ideologically extreme over time. Essentially, this is an alternative representation of what is shown in the histograms in Figure s 2 1 and 2 2 where we see the distributions of the raw DW NOMINATE scores (rather than their absolute values) moving farther apart over time Finally, for the Senate, although the bars representing party leaders (in black) tend to fall at approximately the same height as those representing non leaders (in gra y), when they do differ from one another, the values for party leaders tend to be higher. This suggests that the ideological positions of Senate party leaders are about the same or more extreme than those of rank and file senators. A similar pattern is o bserved for party
109 leaders in the House, who often fall even farther above their rank and file colleagues. Again, this fits with the traditional argument about the rise of Conditional Party Government and party polarization. Turning to the data on bill cos ponsorship, Figure 4 2 presents the distributions of total network partisanship for members of the Senate and House across all Congresses under study. On the left side of the figure, we see that Senate party leaders maintain levels of network partisanship almost identical to those of non leaders. While there is some fluctuation in these levels over time, particularly a slight increase in partisanship during the 103 rd 105 th Congresses (early 1990s), the distributions remain between the 50 and 60 percent le vels. What this suggests is that serving in the role of party leader in the Senate does not with either fellow partisans or partisan opponents. In contrast, beginning in the 104 th Congress, party leaders in the House have maintained much more polarized cosponsorship networks compared to their rank and file colleagues. Through the first half of the series, there appears to be little difference between the nature of cosponsorship relationship s established by party leaders compared to non leaders. However, with the Republican Revolution and the rise of Newt Gingrich to the Speakership in 1994, this change d drastically. Throughout the second half of the series, party leaders could not be more different from the rank and file. For several of these Congresses (particularly the 104 th 106 th ), the middle quartiles for party leaders do not even fall within the middle quartile range for non leaders. This means that House party leaders during this pe riod were more polarized than a majority of their rank and file colleagues.
110 Looking at Figure 4 3, we observe a similar trend for the measure of total cosponsorship partisanship Here, the overall partisanship percentages are slightly higher compared to t he network partisanship percentages (as discussed in Chapter 3), but the pattern remains: Senate party leaders are no different from their rank and file colleagues, while the party leaders of the 104 th 110 th Houses are distinctly more polarized than their colleagues. negotiations. As 2012 came to a close and 2013 began, members of Congr ess and President Barack Obama found themselves mired in debate over how to reduce the pieces of legislation, including an amendment to a bill that increased income tax ra tes. This particular amendment passed with 89 votes, including those of 40 Republicans. The next day Senate Majority Leader Harry Reid took to the chamber floor and pushed through a number of confirmations of presidential nominations. Although many sena tors had already left town after a major vote on the fiscal cliff, the significance of this motion is that Reid had the full support of Minority Leader Mitch McConnell: Harry Reid, who every day does much more than most people in the news media realize, an d definitely accomplishes much more than the news media ever reports, pushed through pages and pages of nominations for President Obama yesterday when everyone was focused on what the House would do on the fiscal cliff vote. And Harry Reid did that with th e active but invisible help of Mitch McConnell who did his part to make sure that no Republicans would vote against any of those nominations. And what did the United States Senate do today, that dysfunctional United States Senate? According to the news med ia, absolutely nothing (Muller 2013) This observation highlights an issue to which I will turn in the conclusion of this dissertation: the role of the media in framing the legislative process in terms of conflict
111 rather than compromise. However, it also illustrates the bipartisan negotiations that go on between party leaders in the Senate that are often invisible to everyone besides a few behind the scenes players. Overall, these patterns are likely due to the inter chamber structural differences discusse d above. The incentive for Senate party leaders to work with their colleagues across the aisle is simply not present for leaders in the House. Focused more on enforcing party loyalty among their rank and file, House leaders have little need to cosponsor bills written by their partisan opponents. Rather than building bridges across parties, the hyper partisan nature of the House is exacerbated when we look more closely at House leaders who choose to associate almost exclusively with their same party colle agues Another approach to this is to compare the ideological positions of leaders and non leaders in each chamber to their total network and cosponsorship partisanship. In Figure 4 r total network partisanship percentage, including legislators from all Congresses in the series (96 110). Doing so allows us to compare a legislator voting patterns against his bill cosponsorship patterns. In Chapter 3, I presented graphs of the same relationship for each Congress without distinguishing between party leaders and non leaders (see Figures 3 10 and 3 11). Here, I combine all the Congresses in the graphs in order to obtain a more reliable prediction (or regression fit) for the party leade rs. Because there are only five party leaders serving in each chamber during each Congress, there is a lot
112 Congresses gives us a large enough number of cases from which we can draw conclusions about the general behavior of party leaders versus non leaders. the partisan degree of his cosponsorship network. This tells us that senators who are more id eologically consistent in their policy preferences (by taking the consistently liberal or consistently conservative positions in their roll call votes), are likely to have a higher network partisanship percentage. In other words, senators who are more ide ologically extreme in voting on policy are also less willing to work with their partisan opponents by cosponsoring policy. Moreover, the results support what was found in Figures 4 2 and 4 3. For the Senate, party leaders (in black) fall well within the i deological and partisan network range of non leaders (in gray). There is virtually no difference in the relationship between ideological extremity and network partisanship for Senate party leaders and non leaders, as illustrated by the lines depicting the values fitted to a regression rank and file member of the Senate has no influence on the relationship between his ideology and network partisanship. However, in the H ouse we find that, while the positive correlation between ideological extremity and network partisanship remains, there are again important distinctions between House leaders and non leaders. In particular, the overall network partisanship of House party leaders tends to be higher than that of the rank and file, as illustrated by the placement of the fitted line for party leaders (in black) above that for non leaders (dashed, in gray). This tells us that, given two House members, one who is
113 a party leader and one who is not, but who share the same value of ideological extremity, the party leader will have a more partisan cosponsorship network than will the non serving as a par ty leader will tend to make a legislator more partisan than will being in the rank and file. Looking at the relationships between ideological extremity and total cosponsorship partisanship we find similar results. Figure 4 5 provides further support that the cosponsorship patterns of Senate party leaders are almost identical to those of non leaders, while the behavior of House leaders is more partisan than th at of non leaders, controlling for ideological extremity. W hat is most important to note about this analysis is that (1) service as a party leader is an important predictor of partisan cosponsorship behavior above and beyond a patterns of cosponsorship by party leader s in each chamber. If institutional structure had no influence on the roles of party leaders in each chamber, we would expect to find that the effect of serving as a party leader to be the same in both the House and the Senate. Indeed, this is what we co nclude from Figure 4 1 when we observe the influence of leadership on roll call voting behavior: party leaders in both chambers tend to be more loyal partisans in their voting patterns compared to their rank and file colleagues. However, this is not the c ase for bill cosponsorship ties. Rather than mimicking the hyper partisan patterns of relationship formation exhibited by party leaders in the House, members of the Senate leadership often tend to be even less partisan in their networking strategies than are their rank and file counterparts. This is
114 likely due to the necessity of bipartisan cooperation created by the unique rules and norms of the Senate. Strength of Cosponsorship Ties As a final descriptive analysis of the differences between party leader s and non leaders in each chamber, I return to an exploration of the depth, or strength, of cosponsorship ties between legislators. Figures 4 6 and 4 7 illustrate the depth of logical extremity. Party leaders appear in black (labeled with the number of the Congress during which they served); non leaders appear in gray. Looking at both figures, we see a stark contrast in the depth of cosponsorship ties in the House and Senate. Because they serve in a much smaller chamber, senators are able to associate with their colleagues through cosponsorship much more frequently than are House members. Having fewer colleagues to choose from lends to associating with the same colleagues rep eatedly. However, this does not explain the differences in the behavior of party leaders in each chamber or the differences in the strengths of the partisan and bipartisan ties. If the traditional argument about party polarization were applicable to bill cosponsorship behavior, we would expect to find that legislators in both the House and the Senate have weaker ties to their partisan opponents than to their fellow partisans. In other words, if the parties have been polarizing, we would expect to find mem bers of both chambers associating most frequently with their same party allies while neglecting to build relationships with their other party opponents. What we actually find is that this is true for the House, but not so much for the Senate. While most House members have between one and five cosponsorship ties to their fellow partisans, they have
115 between only one and three ties to their partisan opponents. Moreover, the distribution of the points for partisan ties in the House resembles a slight bell sh ape, suggesting value of their ideology score) have the strongest relationships with their partisan colleagues. In contrast, the most ideologically moderate House membe rs (with a value of 0) and the most ideologically extreme House members (with a value approaching 1) tend to have the weakest relationships with their same party colleagues. However, when we turn to Figure 4 7 we find that the relationship between ideologi cal extremity and depth of bipartisan ties in the House is not a bell curve, but has a negative slope. This indicates that the most ideologically moderate House members (with a score around 0) have the strongest ties to their bipartisan colleagues, while the most ideologically extreme House members (with a score approaching 1) have the weakest ties to their bipartisan colleagues. Again, this is what we would expect according to the traditional argument about party polarization. A final point to note befo re turning to the Senate graphs is that party leaders in the House tend to fall toward the bottom portion of the plot, suggesting that they tend to have weaker relationships with their partisan and bipartisan colleagues compared to non leaders. Of course, this could be due to a quirk of the chamber in which party leaders choose to spend their time engaging in activities other than bill cosponsorship. Nevertheless, the point is worth mentioning, especially when we observe the behavior of Senate party leade rs. When we compare the plots of partisan and bipartisan ties for the Senate, we find that they look very similar. While there is a slight positive slope to the plot of partisan ties and a slight negative slope to the plot of bipartisan ties (indicating t hat the
116 most ideologically extreme senators have stronger ties to their fellow partisans and weaker ties to their partisan opponents), there is a number of legislators who have very strong ties (between 5 and 15 cosponsorships per tie) to their colleagues; this is true for both the partisan and bipartisan plots. Moreover, more than half of these senators are members of the party leadership. This observation is even more surprising when you consider that party leaders make up only 5 percent of the observat ions in each graph (5 party leaders out of 100 senators during each Congress). What this tells us is that, while party leaders in the House build relatively weak relationships with their colleagues through bill cosponsorship, a disproportionately high amo unt of Senate leaders cooperate through bill cosponsorship frequently with both their partisan and bipartisan colleagues. Furthermore, the senators with the strongest bipartisan relationships are party leaders from the 109th and 110th Congresses, which oc curred during a period of extremely high party polarization. To elaborate on this further, if the traditional polarization argument held, we would expect the plot of bipartisan ties in the Senate to match the plot of bipartisan ties in the House; in an atm osphere of high party polarization, members of both chambers would associate on an extremely limited basis with their partisan opponents. Just as legislators vote only rarely with their colleagues across the aisle, so would they cosponsor legislation. Ho wever, this is not what we find. Instead, we see that senators, particularly Senate party leaders, cooperate with their partisan opponents through bill cosponsorship between 5 and 15 times during a given Congress. This observation does not paint a pictur e of a chamber in which Democrats and Republicans and their respective party leaders refuse to speak to and interact with each other during the
117 policymaking process. Rather, it suggests that Senate party leaders (even during the most recent and highly pol arized Congresses) reach out to their colleagues across the aisle by either cosponsoring bills their opponents have sponsored or soliciting cosponsorships from them. In either case, this is a sign that bipartisan connections are made, and they are not a f luke, but an established pattern of behavior. Years of Service role can shape his or her cosponsorship behavior, it may also be the case that service in the particular in stitution overall may also have an effect over time. As discussed in Chapter 1, the theory on which we rely to understand this dynamic is sensemaking. When a legislator is elected to either the House or the Senate, he is introduced to an environment with which he is unfamiliar. Relying on cues from his colleagues and institutional leaders, the freshman legislator gradually adopts the norms of the institution. This is a process of trial and error; over time, the legislator learns which strategies help hi m get ahead in the chamber and achieve his policy and power goals, and which strategies are not as helpful. Based on the assumption that legislators choose to adopt strategies that make them more successful in achieving their goals, we can infer that the longer a legislator has served in the chamber, the more his behavior will look like the type of behavior that makes one successful within that particular a legislator to pass his preferred policies and obtain positions of power in a chamber, then we would expect to see legislators becoming more loyal to their party as the number of years they have served increases. In contrast, if success is determined by how well conn ected a legislator is within the larger chamber network, then we would
118 expect to see his personal network partisanship decline the longer that he has served in the chamber. An exploration of the data reveals some very important inter chamber differences. First, Figure 4 the number of years each legislator served in the particular chamber. If a legislator was elected in 1994 to the 104 th Congress, his value for years of service during th at Congress would be two, for having served two years by the end of that term. Likewise, if a legislator served in the House and later moved on to the Senate, his value for years of service would return to two after the first Congress in which he served i n the Senate. Finally, the plots in the figures below include all legislators who served during all the Congresses in the series (96 110); this means that legislators who served during multiple Congresses will appear as one point for each Congress during which they served. While these graphs do not allow us to observe change within an individual legislator over time (a trend which we will explore in the regression analyses below), they do allow us to draw comparisons across legislators controlling specifi cally for the amount of time each spent in the chamber. One thing that the graphs for both chambers have in common is the large amount of variance in total network partisanship for legislators who have entered the chamber most recently. Freshmen senators possess a total network partisanship percentage anywhere between 30 and 85 percent. In the House, newly elected members vary between 25 and 100 percent. This suggests that, upon entering either chamber, legislators lack a clear understanding of the type of networking behavior
119 necessary to succeed, and each legislator pursues a different strategy. However, over time, the variance in these plots decreases, and the distributions of the points narrows. W e see that legislators who have served over 20 years in the Senate tend to exhibit a total network partisanship percentage between 35 and 60 percent. In the House, there are more observations and, consequently, more variance, but the general trend is for the points to converge between the 60 and 85 percent ma rks by about 40 years of service. Much of the explanation for this decrease in variance in both chambers is due simply to the fact that more legislators drop out of the plots as the number of years of service increases (because more individuals serve in t he House for example, for 2 10 years compared to 40 60 years). However, it is important to note that the observations which do remain in the dataset for these higher values of years of service are not scattered all over the map; instead, they tend to be clustered within the same general area. Again, because some of the points are repeated observations of to be somewhat steady across Congresses, but this does not exp lain the overall narrowing of the plots as the value of x increases for all legislators in the dataset. Figure 4 9 includes only the legislators who served more than 40 years in the chamber, the number of observations is small, we can see that the points represent a number of legislators who all tend to exhibit similar degrees of partisanship in their cosponsorship networks. When we look at the values for the total cosponsorship partisanship v ariable, we find similar results. For both chambers, the graphs illustrate high variance at the low values of x, and the points gradually converge as x increases. In the Senate, the points
120 converge around the 50 75 percent range; in the House, they conve rge at much higher values, around 75 100 percent. I argue that this is the result of the process of socialization within each chamber. As members serve longer in the House, they learn, through the process of sensemaking, to build more connections with me mbers of their own party (at the expense of building bipartisan relationships) and to support (and solicit support from) their partisan colleagues through bill cosponsorship. A final point to make about the relationship between years of service and cospons orship behavior in each chamber is related to the differences in the behavior of party leaders in each chamber. For both the network and cosponsorship partisanship variables, we see important differences in the placement of the points representing party l eaders (the black circles) in each chamber. In the Senate, members of the party leadership are scattered across the range of the entire Senate membership. This observation fits with our previous conclusions (presented above) about how Senate party leader s do not differ much in their cosponsorships or network partisanship compared to the Senate rank and file. However, in the House party leaders are again clustered at the top extreme of the scatterplot, with most leaders located between 75 and 100 percent for their total cosponsorship partisanship throughout their tenure. This, too, fits with our previous observations, providing further evidence for the argument that the nature of the roles of party leaders is different in each chamber of Congress. Regress ion Analyses: Exploring Differences Within and Between Legislators As a further test of these relationships, I regress the total network and cosponsorship partisanship variables on the explanatory and control variables discussed above. I run separate mode ls which control for fixed effects (within cluster variation) and random effects (between cluster variation). This means that we can
121 observe the effects of the independent variables on the dependent variables both within clusters (considering how each leg between clusters (comparing legislators to one another). Based on the theory which drives this behavior, and that there may be important changes in this behavior the longer each legislator serves. This justifies the use of the within legislator analysis. However, we or his ideologic al position, might distinguish each member of Congress from his colleagues. This makes the use of the between legislator analysis necessary, as well. The dependent variables, total network partisanship and total cosponsorship partisanship, have already be en described in detail, as have the following independent variables: service as a party leader, years of service, and ideological extremity. I also include three additional control variables in the model: party size, number of bills sponsored, and number of bills cosponsored. Party size is a measure of the percentage variable is included because we would expect that the more fellow partisans a legislator has with whic h to cosponsor, the higher his network and cosponsorship partisanship values will be. More fellow partisans in a chamber means more opportunities to cosponsor with them. The bills sponsored and cosponsored variables are included as controls for a legislat time in the realm of writing policy will also invest time in building support for these bills in order to help them pass. This process of building support involves tal king to colleagues
122 and asking for their votes; in other words, it involves networking. Therefore, I expect that legislators who are more active in the process of policymaking, (or who sponsor and cosponsor more bills), will have more extensive (and divers e) networks as a result of the work they invest in getting these bills to pass. Clearly, for the Senate, this means that these more active senators will have less partisan (or more bipartisan) networks and cosponsorships; as I have already established, bi partisan networking is necessary to be a successful policymaker in that chamber. However, in the House, although the general atmosphere is one of hyper partisanship, it is possible that the most active policymakers, or those who work the hardest to see th eir bills pass, will pull out all the stops to be successful. This may mean that they are more willing to cosponsor bills written by their partisan opponents in order to have the favor returned to them for the passage of their own legislation. This would mean that increased policy activity (in the form of bill sponsorship and cosponsorship) makes even a House member less partisan compared to his less active colleagues. The results of the within and between legislator analyses are presented in Tables 4 1 and 4 2, respectively. As shown in Table 4 1, the trends in the data uphold the hypotheses presented in this chapter. Overall, we find similar results for both dependent variables, suggesting that the measures are robust. The number of legislators inclu ded in the House models is 1,335. Because many of these legislators served in multiple Congresses and, as a result, appear in the dataset multiple time s the number of observations is much higher, at 6,537. The average number of Congresses in which a leg islator served is 4.9. For the Senate models, the number of legislators is 260, the number of observations is 1,509, and the average number of Congresses served by each senator is 5.8.
123 Turning to the relationships between the explanatory variables and leg network and cosponsorship partisanship, we see first that, in terms of service as a party leader, the results confirm what we have already observed: the nature of party leadership varies in the House and in the Senate. Specifically, for a given legislator, as service as a party leader varies across time (from belonging to the rank and file to percent and total cosponsorship partisanship increases by 5 percent (p< 0.01 in both models). Substantively, this means that when a member of the House becomes a party leader, he also becomes more partisan in terms of the people with whom he cosponsors legislation. In the Senate however, becoming a party leader makes no signi ficant with same party or other party colleagues Considering this in terms of change over time, Senate party leaders are no more (or less) loyal to their party when they assume a leadership role than when they are simply members of the rank and file. Furthermore, the results also indicate that longer service in the House correlates with more partisan cosponsorship network and behavior, while longer service in the Senate has the opposite effect. For any given legislator, each additional year served in the House corresponds to a 0.07 percent increase in network partisanship and a 0.06 percent increase in cosponsorship partisanship (p<0.01 in both models). At first glance, this seems miniscule; how ever, it is reasonable that only one additional year of service length o
124 partisanship increases by 1.2 percent. Again, this change is small, but it is statistically si gnificant. In the Senate, each additional year of service corresponds to a decrease of 0.04 percent in network partisanship and 0.07 percent in cosponsorship partisanship (p<0.05 in both models). Again, though these changes are small, they are statistica lly significant. The remaining variables are control variables, and most of the relationships end up as expected. Party size and ideological extremity are both very good predictors of r. In both chambers, both variables have a strong positive and statistically significant (p<0.001) to associate more with his same party colleagues and to do so mo re frequently, relative to his associations with his partisan opponents. Each one percentage point increase in percentage point increase in his partisan network and cosponsors hip associations. Likewise, as legislators move further to the ideological extremes over time, based on the absolute value of their Nokken Poole DW NOMINATE scores, they tend to increase their network and cosponsorship partisanship. If a legislator were to move from the absolute moderate position (with a value of 0 as his ideology score) to the absolute ideological extreme (with a value of either 1 or 1 as his DW NOMINATE score), his network and cosponsorship percentages would increase by between 6 and 1 9 percentage points. These percentages are so large because they measure movement
125 from one ideological extreme to the other, a change which is unlikely to occur during heless. Finally, the results for the variables measuring legislative activity are less consistent. In terms of the number of bills sponsored, as any given House member sponsors more bills over time, his network and cosponsorship partisanship tend to decre ase by 0.06 and 0.02 percentage points, respectively (p<0.001, p<0.06). For a given senator, bill sponsorship and network and cosponsorship partisanship are not significantly related. In terms of cosponsorship in either chamber, as any given House member or senator cosponsors more bills, his network partisanship decreases significantly. However, this type of change has no statistically significant effect on cosponsorship partisanship. Overall, the results for these variables are inconsistent, but it is important to include them in the model in order to control for any ways in which colleagues. Going further, Table 4 2 presents the results for the between legislator models. To reiterate, the between legislator effects variables allow us to compare each legislator to his colleagues during a given Congress, rather than looking at the change within each legislator over time. First, when we look at the relationship betw een service as a something puzzling. In both chambers of Congress, during any given Congress, legislators who hold party leadership positions have network and cosponsorship pa rtisanship values significantly higher than those of their rank and file colleagues. In the case of the House, it is not surprising to find that the total partisanship values for
126 party leaders are 7 and 10 percentage points higher than those of their coll eagues (p<0.01). However, we find similar results for the Senate: here, party leaders have partisanship values 2.8 and 4.8 percentage points higher than their rank and file colleagues (although here, p<0.10). This result is particularly curious given the distributions we reviewed earlier in this chapter showing that party leaders tend to be safely in line with t heir rank and file colleagues. Second, the results for the years of service variable illustrate precisely why it is important to run both the with in and between legislator effects models. While we found that the influence of lengthier service in a chamber on an individual legislator tended to make House members more partisan and senators less partisan, we find in Table 4 2 that, during a given Con gress, in both the House and the Senate, legislators who have served longer are less polarized than those who have been elected more recently. The likely cause of this is related to what I call cohort effects Essentially, I suspect that the era in whic h a legislator is first elected to the during his formative years, or during his f irst several years of service. Because Congress, as an institution, has developed over the last 30 years into a body that prioritizes party unity over other factors, such as seniority and apprenticeship, it makes sense that the cohorts elected, for exampl e, during the Textbook Congress Era, have developed less polarized patterns of networking and cosponsorship compared to the cohorts elected during the Party Government Era of the 1990s and 2000s. Therefore, during any given Congress, we can expect the mos t senior members of each chamber
127 to exhibit less partisan network and cosponsorship patterns compared to the most junior members. These patterns are explored further below. Again, the party size and ideological extremity variables have the predicted effec t. In the between effects analysis, the party size variable essentially serves as a measure of whether or not the legislator in question serves in the majority party. If a n Congress, then the legislator has more fellow partisans available as cosponsors and s/he tends to utilize them as such. In terms of ideological extremity, legislators in both chambers who fall further to the extremes also tend to have much more partisan networks and cosponsorship patterns. Finally, the results for the variables for the number of bills sponsored and cosponsored are again inconsistent. Generally, during a given Congress, legislators who sponsor more bills tend to have less partisan networ k and cosponsorship patterns, although the only relationships that are significant are network partisanship for the House (p<0.01) and cosponsorship partisanship for the Senate (p<0.06). In contrast, in both the House and the Senate, legislators who cospo nsor more bills have less partisan networks and more partisan cosponsorships. What this suggests is that legislators who cosponsor more bills during a given Congress associate through cosponsorship with a more bipartisan range of colleagues, but tend to g ive and receive actual cosponsorships more frequently with their partisan colleagues than their partisan opponents. In other words, as they cosponsor more bills, their networks become more diverse, but they exchange more frequent support with members of t heir own party. However, while the results are statistically significant at at least the 0.05 level, the coefficients are small,
128 suggesting that we probably should only think of these variables in terms of control variables, rather than as having any majo behavior or network. Cohort Effects The final factor I will discuss which distinguishes legislators from one another is the time period during which they were first elected to the chamber. Different historical e ras possess different characteristics and are characterized by different norms and patterns of behavior. For example, the Textbook Congress Era of the 1930s 1960s is the chamber above all else. Other important practices were deference and apprenticeship: junior legislators were not to speak unless spoken to, and they were to bide their time, working diligently until they were finally called upon to serve in a positio n of power. Over time, however, these traditions changed. Junior legislators in the 1970s decided that the system did not serve their interests sufficiently. As a result, they changed the rules of the institution, and the norms of behavior followed suit Eventually, due in part to societal and policy developments and changes to the broader party system (in the form of the Southern Realignment), Congress enter ed the Party Government Era in which it operates today. These historical developments are import ant for our purposes because of the impact that contemporary rules and norms have on the socialization of legislators. If congressional norms were constant across history, we would expect there to be no differences in the behavior of legislators across co horts. In essence, history would not matter. However, historical change has occurred; thus, it is important to consider the possibility that historical traditions and events have varying effects on legislators as they
129 are introduced into the institution across time. To explore this concept, I categorize legislators in the dataset according to the year in which they were elected to the chamber. I have created four cohorts, each matching a distinct period in congressional history. The various cohorts are outlined in Table 4 3. The first cohort includes all legislators who served during the 96 th through 110 th Congresses but were elected during the Textbook Era of Congress, or between 1940 and 1972. There are no legislators in the dataset who were elected prior to 1940. I have chosen 1973 as the first cut point because this year marks the period leading up Watergate Babies were a large class of Democratic legislators, mostly from the North, scandal. It was the election of this cohort (including 49 Democrats in the House and 5 in the Senate) which greatly increased the power of liberal northern D emocrats vis vis the southern conservative Democratic committee chairmen and solidified the successful passage of congressional reforms, such as the Subcommittee Bill of Rights and other rules changes, during the mid to late 1970s. These reforms transf ormed the institutional norms and procedures of Congress and ushered in a period of dominance by subcommittee chairmen and the weakening of the long held seniority norm. The second cut point occurs in 1980, in line with the election of Ronald Reagan. The 1980s represent a period of Republican dominance and the initial movement toward a party dominated system in Congress. The 1980 election marked the first time since 1954 that the Republican Party had controlled a chamber of Congress (in 1980, they won a m ajority in the Senate), and the grip the Democratic Party had held for
130 decades began to weaken. Moreover, the early 1980s are the period during which we saw the rise of hyper partisan, and often uncivil, behavior, to which many observers point as sign of increasing party polarization. For example, during the 98 th Congress (1983 1984), the members of the Conservative Opportunity Society, a group of young, conservative House Republicans organized by Newt Gingrich in 1983, began delivering speeches on the re cently formed C SPAN network in an attempt to embarrass the Democratic majority. In response to this, then SPAN camera operator pan the chamber to reveal that the representatives were speaking to an empty audience (Ragusa 2012 ). This event triggered one of the many partisan battles for which this period became known. The final cut point is situated in 1994, the election year during which the Republican Party reclaimed majorities in both the House and the Senate for the first t ime in forty years. This Republican Revolution, as it has become known, marked the rise of Representative Newt Gingrich (R GA) to the House Speakership and the beginning of the hyper partisan era we have today. Since 1994, party unity scores in roll call voting have increased, and norms of party loyalty have solidified. Based on these important changes in congressional development, we should expect to find that the cosponsorship behavior of each cohort of legislators (1) is different from that of the othe r cohorts and (2) reflect the era during which they were and cosponsorship percentages will increase the later they were elected; legislators elected during the Text book Congress Era will have the lowest partisan network and cosponsorship percentages, those elected during the Modern Party Era will have the
131 highest partisan percentages, and members of the Reform and Early Party cohorts will fall in the middle. To expl and cosponsorship partisanship by cohort for each Congress and each c hamber. First, Figure 4 1 1 illustrates the total network partisanship values for each cohort in the House during each Co ngress in the series. During the first several Congresses, the network partisanship for members of the Textbook Era cohort (in black) is higher than expected. The longest serving members of this cohort tend to have network partisanship values around 75 p ercent. Because members of the Textbook Congress were socialized during a period of relatively weak parties, I expected them to be the most willing to reach across the aisle and work with their partisan opponents. While there are members of this cohort w ho were elected after 1960 who tend to be less partisan, with partisan network values around 50 percent, it is somewhat surprising to see the Textbook Era House members networking in such a polarized manner. Moreover, the last remaining members of this co hort retained their higher partisan values through the 110 th Congress Members of the next cohort, the group elected during the Reform Era of the 1970s, exhibit a range of behavior during their earliest years of service, anywhere between 25 and 75 percent partisan. Over time, however, they tend to cluster between the 50 and 75 percent range; this is true even during the highly partisan 104 106 th Congresses. Even during the Republican Revolution of 1994 and the Clinton impeachment hearings of 1999, the Ref orm Era cohort in the House continued to associate somewhat regularly with their partisan opponents.
132 Moving into the Early and Modern Party Era cohorts, we begin to notice some changes. Specifically, House members elected during the Early Party Era of the 1980s (the red circles) demonstrate an increase in their partisan network behavior over time. Beginning with the 99 th Congress and moving through the end of the series, we see that each additional cluster of legislators elected to the House during this d ecade has partisan network percentages that are slightly higher than the clusters of their colleagues who preceded them. The range for this group extends from 50 to 100 percent partisan, and inches upward slightly over time. From this graph, it appears t hat the most partisan members of this cohort are the ones who were elected during the late 1980s and early 1990s. The final cohort is represented by the gray circles and includes all legislators elected to the House between 1994 and 2008. This cohort incl udes the most partisan legislators elected throughout the entire period. Although the range includes members with partisan network percentages as low as 50 percent, the entire cohort includes people who consistently reach the 100 percent range. This is n ot true of any of the other cohorts. Thus far, this analysis suggests that legislators elected to the House tend to retain the partisan cosponsorship network practices they developed when they were first elected to and socialized into the chamber. For ad ditional perspective, we turn now to the Senate. For the Senate, the results are somewhat different. Rather than observing a networking pattern unique to each cohort, what we see is that although the later cohorts may exhibit higher partisan network perce ntages when they are first elected, most senators return to an institutional mean around the 50 60 percent mark. For example,
133 the Textbook and Reform Era cohorts (in black and in blue, respectively) tend to stay around the 50 percent mark and never move a bove the 75 percent mark throughout the series. Members of the first half of the Early Party Era cohort (in red), or those elected during the 97 th and 98 th Congresses of the early 1980s, maintain this trend. However, some members of the second half of th e Early Party Era cohort, or those elected to the 102 nd and 103 rd Congresses (1991 1993), have partisan network percentages closer to and above the 75 percent mark. This seems to suggest the same trend that we observed for the House: legislators elected t o the chamber during later eras tend to have (and maintain) higher partisan network percentages. However, the most important point to notice is that, while these senators elected during the Early Party Era may have higher network percentages initially, th ey do not maintain this pattern over time. When we observe the distribution for these Early Party cohort members (the red circles) for the 104 th through 110 th Congresses, we see a stabilization of their partisan network percentages, and a gradual convergen ce within the 50 60 percent range. 1 This is particularly remarkable (and unexpected given the traditional literature on party polarization) because these later Congresses took place during the most highly polarized period in recent history. Furthermore, we see the same trend occurring with the Modern Party Era cohort (in gray). When these senators are first elected (during the 104 th 106 th Congresses), there are many within the cohort who have partisan network percentages around the 75 percent mark. Howe ver, during the 107 th 110 th Congresses, members of the first half of this cohort fall in line with the rest of their Senate colleagues 1 This pattern is also true of the Textbook and Reform Era cohorts (the black and blue dots), although it is less surprising as these senators already exhibited this more bipartisan network behavior prior to this period.
134 (around the 50 percent mark), while the more newly elected cohort members (elected during the 109 th and 110 th Congresses) are the senators who exhibit the higher partisan network behavior. 2 Overall, what these plots seem to suggest is that members of the Senate experience a process of socialization, or learning, during the first few Congresses in which they serve (or during their first term in the Senate). Elected during a more polarized era (the 1980s 2000s), senators in the Early and Modern Party Era cohorts start out practicing a more polarized pattern of networking, choosing to favor building relationships with their sam e party colleagues over their partisan opponents. However, as time passes, senators learn that this tendency is not beneficial to achieving their policy and power goals in this chamber and adjust their networking practices accordingly, instead choosing to maintain cosponsorship networks much closer to the 50 50 range. To explore this further, I present the results of a regression analysis including dummy variables for each era in Table 4 4 below. Because the dummy variables for all four of the eras cannot be included in the model, one era must be dropped. I expect that the Modern Party Era is the most partisan of the four, therefore I have chosen to use it as the reference category. In reading the regression results presented in Table 4 4, the coefficien ts for the era dummies should be read as being in comparison to the level of partisanship exhibited during the Modern Party Era. Moreover, because each ership in the cohort to which he was elected does not vary over time 2 Plots for the total cosponsorship partisanship variable show similar results and are included as Figures 4 13 and 4 14
135 (i.e. the year on e was first elected does not change over his career), we can only observe the between legislator effects. Overall, what we find is that the influence of being elected to a chamber during a terns is much stronger in the House than it is in the Senate. For example, in the House we see that legislators elected during the Textbook, Reform, and Early Party Eras all have network and cosponsorship partisanship values significantly lower than legis lators elected during the Modern Party Era (the reference category). In the Senate, however, only members elected during the Textbook Era have values significantly less partisan than the Modern Party group on both dependent variables. This suggests that senators elected during the Reform and Early Party Eras practice cosponsorship behavior that is not significantly different from those elected during the Modern Party Era despite the increased party polarization observed in roll call voting patterns during this latter period. Furthermore, when controlling for the period during which a legislator was elected, we find that party leaders in the House are significantly more partisan than rank and file House members, while the cosponsorship networks and behavio r of party leaders in the Senate are not significantly more or less partisan than their rank and file colleagues. The results for all of the other variables are consistent with the findings presented in earlier tables. The purpose of this chapter has been to illustrate not only that there are important differences between the networking and cosponsorship patterns of legislators between each chamber of Congress, but also that the institutional roles and individual characteristics of each legisl ator matter w ithin each chamber. I begin Chapter 5 by
136 focusing specifically on the Senate in an effort to determine if legislators who previously served in the House, and were socialized in its highly polarized atmosphere, later take these patterns of behavior with th em when they serve in the Senate. What I find is that behavior learned previously; seeking to achieve their policy and power goals, newly elected senators abandon thei r previous notions of networking which they learned in the unique working environment. Table 4 1. hin legislator e ffects Network partisanship Cosponsorship partisanship House Senate House Senate Coeff. (S.E.) p value Coeff. (S.E.) p value Coeff. (S.E.) p value Coeff. (S.E.) p value Party leader 4.250 (1.274) 0.001 0.141 (0.686) 0.838 5 .147 (1.398) 0.000 0.661 (1.052) 0.530 Years of service 0.076 (0.021) 0.000 0.045 (0.020) 0.027 0.061 (0.023) 0.006 0.074 (0.031) 0.019 Party size 1.028 (0.021) 0.000 0.965 (0.025) 0.000 1.017 (0.023) 0.000 1.113 (0.038) 0.000 Ideological extremity 6 .330 (1.414) 0.000 9.294 (1.865) 0.000 7.650 (1.552) 0.000 19.072 (2.860) 0.000 Bills sponsored 0.066 (0.010) 0.000 0.001 (0.001) 0.285 0.020 (0.011) 0.059 0.000 (0.002) 0.809 Bills cosponsored 0.008 (0.001) 0.000 0.013 (0.001) 0.000 0.000 (0.001) 0 .686 0.002 (0.002) 0.202 Constant 13.232 (1.144) 0.000 8.062 (1.534) 0.000 13.747 (1.255) 0.000 0.001 (2.352) 1.000 Observations 6,537 1,509 6,537 1,509 Number of legislators 1,335 260 1,335 260 T (avg.) 4.9 5.8 4.9 5.8 R 2 (within) 0.34 0.60 0.30 0.42
137 Table 4 2. between legislator e ffects Network partisanship Cosponsorship partisanship House Senate House Senate Coeff. (S.E.) p value Coeff. (S.E.) p value Coeff. (S.E.) p valu e Coeff. (S.E.) p value Party leader 7.510 (2.595) 0.004 2.863 (1.530) 0.062 10.407 (2.864) 0.000 4.806 (2.442) 0.050 Years of service 0.246 (0.033) 0.000 0.154 (0.033) 0.000 0.311 (0.036) 0.000 0.272 (0.052) 0.000 Party size 0.949 (0.027) 0.000 0.853 (0.082) 0.000 0.998 (0.030) 0.000 0.893 (0.130) 0.000 Ideological extremity 29.997 (1.078) 0.000 16.941 (1.465) 0.000 36.493 (1.190) 0.000 35.557 (2.339) 0.000 Bills sponsored 0.051 (0.017) 0.002 0.008 (0.005) 0.120 0.020 (0.018) 0.273 0 .015 (0.008) 0.059 Bills cosponsored 0.003 (0.002) 0.030 0.006 (0.002) 0.011 0.005 (0.002) 0.001 0.015 (0.004) 0.000 Constant 10.266 (1.494) 0.000 10.829 (4.380) 0.014 6.544 (1.649) 0.000 5.280 (6.994) 0.451 Observations 6,537 1,509 6,537 1,509 Numbe r of legislators 1,335 260 1,335 260 T (avg.) 4.9 5.8 4.9 5.8 R 2 (between) 0.60 0.58 0.62 0.61 Table 4 3. Descriptive s tatistics of c ohorts House Senate Network partisanship years n mean min max n mean min max Textbook 1940 1972 1,036 63.7 23.9 96 .9 298 53.6 36.4 70.5 Reform 1973 1979 1,311 63.6 14.1 100 405 54.3 33.7 73.6 Early Party 1980 1993 2,826 64.8 24.6 95.7 520 56.9 33.6 84.8 Modern Party 1994 2008 1,463 69.7 27 97 288 57.7 36.5 85.3 Cosponsorship partisanship Textbook 1940 19 72 1,036 66.5 21.1 97.6 298 58.4 29.3 80.3 Reform 1973 1979 1,311 67.4 11 100 405 60.2 31.1 83.5 Early Party 1980 1993 2,826 68.6 21.7 96.5 520 64.7 17.3 88.5 Modern Party 1994 2008 1,463 74 20.3 97.6 288 65.8 28 90.5
138 Table 4 4. Cohort Effects on Par tisan Network and Cosponsorship Behavior, by Chamber Network partisanship Cosponsorship Partisanship House Senate House Senate Coeff. (S.E.) p value Coeff. (S.E.) p value Coeff. (S.E.) p value Coeff. (S.E.) p value Textbook 5.671 (0.645) 0 .000 3.624 (0.708) 0.000 6.703 (0.720) 0.000 4.998 (1.157) 0.000 Reform 6.161 (0.601) 0.000 2.056 (0.646) 0.002 5.980 (0.671) 0.000 1.659 (1.056) 0.117 Early party 3.689 (0.468) 0.000 0.910 (0.579) 0.117 3.697 (0.522) 0.000 0.528 (0.947) 0.578 Party leader 6.450 (2.506) 0.010 1.927 (1.466) 0.190 8.939 (2.797) 0.001 2.806 (2.395) 0.242 Party size 0.960 (0.026) 0.000 0.877 (0.081) 0.000 1.005 (0.029) 0.000 0.937 (0.132) 0.000 Ideological extremity 25.988 (1.127) 0.000 16.119 (1.487) 0.000 32.4 54 (1.258) 0.000 35.277 (2.430) 0.000 Bills sponsored 0.056 (0.016) 0.001 0.007 (0.005) 0.126 0.028 (0.018) 0.127 0.015 (0.008) 0.056 Bills cosponsored 0.004 (0.002) 0.018 0.007 (0.002) 0.002 0.005 (0.002) 0.003 0.014 (0.004) 0.000 Constant 12.530 (1.492) 0.000 10.125 (4.326) 0.020 8.719 (1.666) 0.000 2.332 (7.069) 0.742 Observations 6,537 1,509 6,537 1,509 Number of legislators 1,335 260 1,335 260 T (avg.) 4.9 5.8 4.9 5.8 R 2 (between) 0.63 0.59 0.63 0.61
139 Figure 4 1. Ideological e xtremity among p arty l eaders and n on l eaders, 96 th 110 th Congresses
140 Figure 4 2. Network p artisanship among p arty l eaders and n on l eaders, 96 th 110 th Congresses
141 Figure 4 3. Cosponsorship p artisanship among p arty l eaders and n on l eaders, 96 th 110 th Congresses
142 Figure 4 4. Ideological e xtremity vs. t otal n etwork p artisanship for p arty l eaders and n on l eaders, by c hamber
143 Figure 4 5. Ideological extremity vs. t otal c osponsorship p artisanship for p arty l eaders and n on l eaders, by c hamber
144 Figure 4 6. Depth of partisan ties for p arty l eaders and n on l eaders, by c hamber
145 Figure 4 7. Depth of b ipartisan t ies for p arty l eaders and n on l eaders, by c hamber
146 Figure 4 8. Total n etwork p artisanship over n umber of y ears s erved, by c hamber
147 Figure 4 9. Convergenc e of t otal n etwork p artisanship over n umber of y ears s erved, by c hamber
148 Figure 4 10. Total c osponsorship p artisanship over n umber of y ears s erved, by c hamber
149 Figure 4 11. Differences in n etworking p atterns across H ouse c ohorts, by Congress
150 Figure 4 12. Differences in n etworking p atterns across Senate c ohorts, by Congress
151 Figure 4 13. Differences in c osponsorship p atterns across House c ohorts, by Congress
152 Figure 4 14. Differences in c osponsorship p atterns across Senate c ohorts, by Congress
153 CHAPTER 5 NETWORKING IN THE SE NATE When Senator Robert Byrd addressed the class of incoming Senate freshmen during the 1996 orientation session, he spoke about many topics, including Senate traditions and norms, the need to maintain bipartisan relationship s that foster compromises on legislation, and the tools the newly elected senators should use in order to be successful over the course of their careers. However, in this speech he also instructed the freshman class on the unique nature of the Senate and the chamber: A Senator must really be much more than hardworking, much more than conscientious, much more than dutiful. A Senator must reach for noble qualities honor, total dedication, self discipline, extreme selflessness, exemplary p atriotism, sober judgment, and intellectual honesty. The Senate is more important than any one or all of us more important than I am; more important than the majority and minority leaders; more important than all 100 of us; more important than all of the 1 ,843 men and women who have served in this body since 1789. Each of us has a solemn responsibility to remember that, and to remember it often (Byrd 1996) These words speak to the story of the Senate as an institution. We often forget that, behind the fra y of day to day politics and negotiations, there is a code of conduct, a standard to which all senators are held. This tradition has been handed down from one generation of senator to the next, an informal transition invisible to the public eye and embodi ed in the oft The analyses conducted in Chapter 4 suggest that patterns of networking and cosponsorship behavior in the Senate are distinct from those in the House. Whether we look at the roles of party leaders or the overall partisan degree of the ties made between legislators in each chamber, cosponsorship relationships in the Senate are consistently
154 more bipartisan than those in the House. To explore this further, I look more closely in this chapter at the co sponsorship networks and behavior of senators. First, I explore the influence that serving and being socialized in the House has on the networks and cosponsorship patterns of legislators when they are elected to the Senate. Do they retain the hyper partis an patterns of relationship building they learned in the House, or do they make sense of their new surroundings in the Senate by adapting to the different institutional structure and the bipartisan relations it promotes? To test this, I divide senators in each Congress by whether or not they previously served in the House. I find that previous House service is not a significant predictor of increased partisan cosponsorship patterns, suggesting that legislators who move from the House to the Senate adapt t o their new surroundings and learn to build more bipartisan cosponsorship relationships. Second, I discuss an argument presented in the literature which suggests that the behavior of legislators can be shaped by the people with whom they serve. Specifical be aggressive partisan warriors during their service in the House and stand out, in terms of their roll call voting behavior, from their colleagues even once they moved on to the Senate. After explaining this argument, I will test the hypothesis against the data on bill pa rti san voting patterns extend to their cosponsorship behavior. Here I find that although the Gingrich Senators do tend to occupy more ideologically extreme positions (which are based on their roll call voting patterns), compared to their Senate colleagues the networks they
155 form through bill cosponsorship are not significantly different from senators who do not belong to their unique cohort. Rather than adopting a hyper partisan style of networking through bill cosponsorship while serving in the House and carrying this pattern of behavior with them to the Senate, the Gingrich Senators are neither more nor less partisan than their colleagues during either their House or Senate service. It appears that, in terms of bill cosponsorship, they are simply averag e. Previous House Service In the first section of this chapter, I limit the analysis to the Senate and explore differences between senators who previously served in the House and those who did not. I have already shown in previous chapters that cosponsors hip networking and behavior in the House tends to be based more frequently on partisan (rather than bipartisan) ties. It is reasonable to expect that, once legislators learn a pattern of behavior and, through the process of sensemaking, establish habits o f cosponsorship in the House, they may take it with them if and when they are elected to the Senate. Because we know from the evidence provided in the previous chapters that members of the Senate, as a whole, tend to exhibit more bipartisan networking and cosponsorship behavior, we would expect to find that senators who began their legislative service in the Senate (without previously serving in the House), would reflect the more bipartisan cosponsorship patterns of the Senate, especially because they were never exposed to the more partisan processes of the House. Therefore, we could expect that the senators who previously served in the House (and learned networking patterns that fit the legislative process in that chamber) would continue to practice these more partisan patterns of behavior when they enter the Senate, thus making them distinctly different from those senators who never served in the House.
156 One caveat of this argument is that we assume that former House members would retain their more parti san cosponsorship patterns only if those strategies actually work in the Senate. However, if the patterns of cosponsorship behavior and networking that members learn in the House do not work in the Senate, then we would expect to find that senators who pr eviously served in the House will adapt their behavior and networking strategies to fit the patterns that make them the most successful in the Senate; that is, a more bipartisan means of forming relationships. In other words, in order to illustrate the in decision making processes regarding their cosponsorship networking and behavior, I would have to show that senators who previously served in the House, particularly the Gingrich senators, are not different in their Senate networking behavior from those wh o had not served in the House. House and the chamber relies so much on social and political networking across parties in order to fu nction, even the most polarized senators who were socialized in a hyper partisan House and learned to cosponsor and network primarily within their polarized group in the House shift their cosponsorship networking and behavior in the Senate in order to oper ate effectively there. Because of the individual power retained by each the legislative process, regardless of his or her tenure or institutional position of power ), the practice of networking with colleagues on both sides of the aisle and maintaining relationships built on trust and reciprocity is critical to the success of senators.
157 Turning to the data, when we compare senators according to whether or not they pre viously served in the House, we find that there is little difference between the two groups. In Figure 5 1 illustrates the range of values for the total network partisanship of senators with no previous House service (in gray) and those who did serve in t he House prior to their election to the Senate (in black). Across all of the Congresses, the median values for total network partisanship fall within about the 55 to 60 percent range. There is little variation in these values over time, and, most importa ntly, the values for those with and without previous House service are almost identical. Moreover, it is not just the medians that are the same; the entire range of values for each Congress (as represented by each box and the bars on the upper and lower e nds of it) are almost identical, suggesting that all senators (not only those who occupy the median positions) have very similar networking patterns regardless of their status as previous House members. Turning to the data on total cosponsorship partisansh ip, we find similar results. Overall, the medians and ranges for both groups of senators are very similar. There is some divergence beginning during the 104 th Congress, where some senators who previously served in the House (the black bars) appear to prac tice more partisan networking patterns (as reflected by the higher peaks of the boxes in the plots). However, the median values for these Congresses (with the exception of the 108 th and 109 th Congresses) are almost identical to those for the senators with no previous House service. partisanshi p against the number of years he served to determine whether the influence
158 of time spent in the Senate has a different effect on each of the two groups of senators. What we find in Figure 5 3 is that this is not the case. Instead, all senators, regardless of whether or not they previously served in the House, fall within approximately the same range and remain within that range throughout their careers. Across all Congresses in the series, all senators sit slightly above the 50 percent partisan mark. The two groups of senators are so similar that the regression fit lines for each plot are almost indistinguishable. This supports what is illustrated in Figure 5 1, that when legislators enter the Senate, the institution has an overpowering effect on shaping their cosponsorship behavior above and beyond any influence that previous service in the House may have had. We find similar results wh en we create the same plot using the total cosponsorship partisanship variable, as shown in Figure 5 4. Again, comparing senators based on whether or not they previously served in the House produces no major differences. If service in the House had a las ting effect on legislators and their hyper partisan patterns of cosponsorship persisted once they began serving in the Senate, we would see in the scatterplots that the group of senators who previously served in the House (in black) have much higher total network and cosponsorship partisanship values compared to those who did not serve in the House (in gray). Instead, we see that the partisanship values for both groups are very similar. One possible critique of this argument is that perhaps the network and cosponsorship values for all senators are lower because of self selection into the chamber by people who are simply more willing to practice bipartisan cosponsorship. If this is the case and more cooperative individuals are attracted to Senate service, t hen
159 previous House service may be irrelevant. To answer this question, we can compare the total network and cosponsorship partisanship values of this group of senators who also previously served in the House for their House service against the values for their Senate service. If these legislators were equally likely to cooperate in the House as they are in the Senate, then this will tell us that the se individuals are indeed more bipartisan, and the Senate as an institution does not necessarily foster this more bipartisan cosponsorship behavior. Restricting the data to include only the senators who previously served in the House, Figure 5 5 plots the total network partisanship values against the years of service for all senators with previous House service. The gray circles represent the the legislator moved onto the Senate); the black circles represent the values for each erved in the Senate. In other words, it allows us to compare the House and Senate network values across Congresses for only the senators who served in both chambers at some point in their career. We find that the total network partisanship values for the se legislators during their House service tend to be higher than the values for their Senate service. This evidence goes against the hypothesis that more cooperative and bipartisan legislators self select into the Senate and have a history of bipartisan c osponsorship in both chambers. Instead, it indicates that, on the whole, these senators tend to network in a more partisan fashion while serving in the House, and later moderate their networking practices when they enter the Senate. Figure 5 6 illustrate s that the results are similar for the total cosponsorship partisanship variable.
160 As a final illustration of the difference in the cosponsorship networking and behavior in each chamber, I plot the mean of the total network partisanship values for each le cross all Congresses in which he served in the House against these values for his Senate service. This allows for a direct comparison of each s axis. If House network partisanship were a good predictor of network partisanship once a legislator moved on to the Senate, the relationship would be strongly positive, with the plot extending from the bottom left to the top right of the graph. However, this is not what the data show. Rather, the regression fit line is flat, suggesti ng that House networking is not a good predictor of Senate networking. In other words, regardless of 0 percent or 100 percent when elected to the Senate fall slightly above the 50/50 network mark. This indicates that even the House members who cosponsor most exclusively with their fellow partisans tend to build more bipartisan ties, relative to partisan ties, once they enter the Sen ate. When we create a sim ilar plot using the mean total cosponsorship p artisanship values, we find a slightly more positive relationship, but one that is not nearly as strong as we would expect if House cosponsorship behavior was a strong predictor of Sena te cosponsorship behavior. Overall, this evidence supports the argument that House members form more partisan patterns of cosponsorship networking and behavior and later adjust these patterns when they serve in the Senate.
161 As a final test of this hypothes is, I regress the total network and cosponsorship partisanship variables on the independent variables used in the regression analyses in Chapter 4 as well as a dummy variable for previous House service. Because this dummy variable only varies across legis lators, rather than within over time, (i.e. If a senator served in the House prior to entering the Senate, this will be true for all the years that s/he served in the Senate.), it is not appropriate to run a fixed effects regression. Therefore, I run only a random effects regression which determines the difference in the dependent variables across legislators during a given Congress. Table 5 1 presents results that support what we have already observed in the figures above. For both dependent variables, there are no significant differences in the network or cosponsorship partisanship of senators who previously served in the House and those who did not. Again, during a given Congress, Senate party leaders tend to have network and cos ponsorship partisanship percentages slightly higher than non leaders (p < 0.10 for total network partisanship; p < 0.05 for total cosponsorship partisanship). Likewise, we again observe the cohort effect described in Chapter 4 where, during a given Congre ss, senators who have more years of experience in the chamber tend to be less partisan than their more recently elected colleagues (p < 0.001 in both models). Next, the party size and ideological extremity variables are also as we expected: the more seats for both variables in both models). Finally, the control variables of number of bills sponsored and numb er of bills cosponsored are again inconsistent. The number of bills sponsored is negatively
162 latter case (p < 0.10). The number of bills a senator cosponsored is re lated negatively to his or her network partisanship, positively to his cosponsorship partisanship, and significant in both cases (p < 0.05, p < 0.001). Ultimately, what is most important to take from these models is that the network and cosponsorship patt erns of senators with and without previous House service are not significantly different. In general, senators tend to network and cosponsor bills with approximately equal percentages of their same party colleagues and partisan opponents regardless of whe ther or not they were previously socialized in the hyper partisan environment of the House. The Gingrich Senators In 2009, Senator Lindsey Graham (R SC) met several times with President Barack Obama to discuss a variety of issues, the most prominent of whi ch was climate change. Ideologically, Graham is a staunch Republican, receiving a 90% rating from the American Conservative Union. Elected to the House in 1994, Graham was part of the Republican Revolution and one of 32 candidates who signed the Contract with America pledge led by then Speaker elect Newt Gingrich. He was a manager during 2002. Despite his ultra conservative record, Senator Graham has become known in Washin gton for his willingness to negotiate bipartisan compromises. This reputation the name of bipartisanship continues to weaken the Republican brand and tarnish the ideals of Small 2009). However, I've never felt threa tened by people who say s a crime or sin to work with the other side because most Republicans and
163 Democrat s understand that for the good of the co Small 2009). Senate Majorit TN) efforts to end the filibuster (Newton Small 2009). In 2006 they led a bipartisan effort with Ted Kennedy to enact immigration reforms (Swarns 2006). In 2009, Graham cosponsored an alternative health care bill with Democrat Ro n Wyden (OR), became the lone Republican supporter in the cosponsored a climate and energy bill with Senators John Kerry (D MA) and Joe Lieberman (I CT), all in addition to his work with President Obama on climate change (Newton a trustworthy leader helps him to bring additional Republican senators along with him on these compromise measures; one examp le of these followers is Richard Burr (R NC). What do Senators Graham, McCain, and Burr have in common aside from their willingness to work with legislators on both sides of the aisle? They are all Gingrich Senators. In 2001, during negotiations over the legislation that would become known as No Child Left Behind, the White House tapped Senator Judd Gregg (R NH) to work with Senators Ted Kennedy, then Democrat Joe Lieberman, and Evan Bayh (D IN) and Representatives John Boehner (R OH) and George Miller (D CA) to reach a compromise. Doing so led to the eventual passage of the bill (Broder 2001). In 2012,
164 amidst the squabbling over the weak economy, the debt crisis, and the approaching election, another display of bipartisanship occurred: a senior Republi can, Sen. Saxby Chambliss (Ga.), even stopped by a fundraising event for a Democrat, Sen. Mark R. Warner (Va.), to show his support, an extremely rare display of bipartisanship during which the two promoted their work together on a substantial debt reducti on package. Warner and Chambliss convened another bipartisan huddle Tuesday with Erskine Bowles, the Democratic co chairman of the Simpson Bowles debt commission, in preparation for the post election session, in which Congress is expected to tackle the fis cal issues (Kane 2012) Despite the atmosphere of highly polarized parties that pervaded Washington at the time, senators whose constituents would be affected by legislation like the farm bill, highway bill, Postal Service, and Food and Drug Administration were able to come together, regardless of their party loyalties, to reach an agreement. In 2008, the extent of bipartisan support for the farm bill was enough to override a veto by the Bush White House; in that case, 82 senators voted in favor of the fin al draft of the bill (Kane 2012). However, the same was not true for the House where legislators could not agree on bipartisan relationships like the one between Chamb liss and Warner, who spent most of two years working together to find common ground were a sign that the Senate would be able to make it through the looming fiscal cliff negotiations without much difficulty (Kane 2012). What do Judd Gregg and Saxby Chambl iss, noted here for their willingness to reach across the political aisle, have in common? They, too, are Gingrich Senators. Setting aside these anecdotes, a n argument made in the literature recently is that the party polarization observed in the Senate ( DW NOMINATE scores) has been caused largely by a particular group of legislators. It
165 is said that members of this group, known as the Gingrich Senators, brought the hyper partisan behavior they learned in the Hous e with them when they were elected to the Senate (Theriault and Rohde 2011). Examples of this behavior tend to involve the Gingrich senators opposing any and all actions supported by the Democrats (particularly President Obama) simply for the sake of oppo sing them: Senator Jim DeMint (R proposing the creation of a budget commission for reducing the deficit o nce President Obama announced his support for the legislation; and in 2011, Jon Kyl (R AZ) changed his position of favoring all tax cuts to opposing the extension of the payroll tax reduction once Obama made the proposal to do so (Theriault 2010, 2011). I n a more humorous account, the Gingrich Senators were also the least likely to participate in a Secret Santa exchange, an effort proposed by Senators Al Franken (D MN) and Mike Johanns (R NE) to improve comity in the chamber (Theriault 2013). To be classif ied as a Gingrich Senator, a legislator must have served in the House an d subsequently in the Senate; she or he must be Republican; and she or he must have been elected to the House after 1978, the year during which former Representative and House Speaker Newt Gingrich was elected. According to this 2011, 1011). The list of the 33 Gingr ich Senators is included in Table 5 2 This table has been adapted from Table 1 in Theriault and Rohde 2011, page 1017.
166 Specifically, comparing Gingrich Senators to three other types of Republican senators (those without previous House experience who serv ed in the Senate before Gingrich was elected to the House, those without previous House experience who served in the Senate after Gingrich was elected to the House, and those with previous House experience who served before Gingrich was elected to the Hous e), Theriault and indicates that it is the combination of the three qualities whi ch make up a Gingrich senator (previous House service, Republican P arty identification, and election after 1978), rather than the effect of any singular component, which contributes to the hyper partisan roll call voting patterns. The authors conclude tha polarization in the Senate since the 1970s lies overwhelmingly at the feet of previous Based on these findings, the authors argue that it was not ju st any former House member s who made the Senate more polarized ; it was the Gingrich Senators, specifically. In light of this, although I showed above that previous House service does not set the cosponsorship behavior of legislators apart once they move t o the Senate, it is possible that there is something unique about the cosponsorship behavior of the Gingrich Senators. If they truly are different from their Republican House and Senate colleagues in terms of the partisan and polarized degree of their vot ing patterns, then we would also expect them to be hyper partisan in terms of their cosponsorship behavior. To test this hypothesis, I compare the cosponsorship patterns of Gingrich Senators against that of their Senate colleagues.
167 Figure 5 9 illustrates the distribution of total network partisanship scores compared to ideological extremity scores for each Congress for the Gingrich Senators (in black) Democratic senators (in blue), and Republicans who do not meet the other criteria to be Gingrich Senators (in red). The results indicate that, unlike with their roll call voting (where Gingrich Senators are much more partisan compared to their Senate colleagues), the cosponsorship networks of the Gingrich Senators fall well within the range of the networks of their Senate colleagues. If the hyper partisan roll call voting behavior of the Gingrich Senators extended into the realm of networking through bill cosponsorship, we would expect to see the black circles floating above the rest of the distribution in ea ch of the sub graphs. The closest we get to this ideal is during the hyper partisan 104 th 106 th Congresses. However, even during these years, when Newt Gingrich himself was at his peak and we might expect the Gingrich Senators to be most partisan, we fin d that the cosponsorship networks of the Gingrich Senators are at the high end of the range, but not abnormally high, compared to those of their Senate colleagues. In fact, although the Gingrich Senators tend to be more ideologically extreme than their Re publican colleagues (falling further to the right in each of these sub graphs), their total network partisanship is not much higher than that of their colleagues, particularly that of their Republican colleagues. More importantly, this is true even for th e Gingrich Senators at the most extreme ideological ranges (e.g. at or above 0.75). Overall, while there does appear to be a slight positive relationship between ideological extremity and network partisanship during the hyper partisan Congresses of the 19 90s, it appears that the Gingrich Senators exhibit cosponsorship networking patterns strikingly similar to those of their Senate colleagues.
168 Even when we graph the relationship between total cosponsorship partisanship and ideological extremity, the Gingric h Senators do not appear to be much more polarized than their Senate colleagues. Although the results presented throughout this dissertation for the total cosponsorship partisanship variable have tended to reflect higher partisan values (compared to the r esults for the total network partisanship variable), we see in Figure 5 10 that the partisan cosponsorship values for the Gingrich Senators are very similar to those of their Republican colleagues in the Senate. While they may cause or contribute to the i ncreasing party polarization in roll call voting in the Senate, the Gingrich Senators seem to maintain cosponsorship networks and practice cosponsorship behavior that is not significantly different from their Senate colleagues. This suggests that any hype r partisanship that may have been learned by members of this group while they served in the House dissipated once they began their terms in the Senate. To explore this further, I include a dummy variable for belonging to the group of Gingrich Senators in a regression on the dependent variables. As with the regression in Table 5 1 which considers only the between Gingrich Senator does not change over the course of his or her service in the Senate; a legislator was either previously elected to the House as a Republican after 1978 or not. Because of this, it is appropriate to run only a random effects regression. The results are presented in Table 5 3 The results for the control variables are consistent with th ose presented in Table 5 1 and the regression tables discussed in previous chapters; including the dummy variable for Gingrich Senators does not change our observations of these other
169 relationships. What is new is the finding that Gingrich Senators are no t significantly different from their Senate colleagues in terms of either their total network partisanship or their total cosponsorship partisanship. This confirms what we observed in the scatterplots above. Although the Gingrich Senators may have learne d to practice hyper partisan roll call voting behavior while serving in the House and later brought these practices with them to the Senate, the same is not true of their cosponsorship behavior. Despite the hyper partisan roll call voting exhibited by thi s group of senators, they still manage to practice building relationships through bill cosponsorship with their fellow partisans and their partisan opponents. Moreover, these results suggest that the hypotheses introduced in this dissertation have passed a n important test. If anyone in the Senate were likely to exhibit highly polarized patterns of networking and cosponsorship, we might expect it to be the Gingrich Senators. Because these individuals are known for being not only ideologically extreme, but also combative, we would expect them to be the least likely of any senators to be willing to cosponsor legislation and work with their Senate colleagues across the aisle. However, as the results indicate, when these legislators enter the Senate, they beha ve, at least in terms of bill cosponsorship, very much like the rest of the members of the Senate. In pa rticular, as shown in Figures 5 9 and 5 10, the Gingrich Senators fit right in with their Republican Senate colleagues. Finally, it is also important to recall that this is true even of the most ideologically extreme senators. It is perhaps this point that most strongly differentiates roll call voting behavior from networking through bill cosponsorship.
170 These observations conform to the theory of sense making. When the Gingrich Senators served in the House, it made sense for them to behave in a hyper partisan hese legislators were socialized into an atmosphere where one had to exhibit fierce party loyalty in order to get ahead. The more a legislator obstructed the goals of the opposing party and campaigned openly against its members, the better. No one exhibi ted this practice better than Speaker Gingrich. However, as has been repeated several times throughout this dissertation, this strategy simply does not work in the Senate. Even the accounts presented by Sean Theriault as evidence of the hyper partisan be havior practiced by the Gingrich Senators reflect this norm: whenever Jim DeMint and other Gingrich Senators behave uncivilly, they are chastised by their fellow Republican senators ( Theriault 2010, 2011 ). While the hierarchical structure of the House and approach fostered by Newt Gingrich and his followers may have allowed Republicans to maintain power and enact their legislative agenda successfully in the House, this type of behavior does not bring legislators success in the Sen ate. Perhaps as the Gingrich House Members eventually became Gingrich Senators they initially thought their old strategies would work. Perhaps the combative side of their personalities and their old hyper partisan ways even come out in public from time t o time. It may even be the case that these staunch conservatives personally dislike or, at the very least, want to see their Democratic opponents lose on policy and in elections. However, the evidence presented here suggests that their patterns of bill c osponsorship tell a different story.
171 Regardless of their statements in the media, their opposition to participating in Secret Santas, and even their roll call behavior, Gingrich Senators continue to adapt to the unique Senate environment. They build long term working relationships, like that of Judd Gregg and Mark Warner, with their colleagues on the other side of the aisle, and they cosponsor legislation with their partisan opponents. They do this because this is what makes for a successful career in th e Senate. Table 5 1. Influence of p revious House s ervice on s b ill c osponsorship, b etween l egislator e ffects Total network partisanship Total c osponsorship partisanship Coeff. (S.E.) p value Coeff. (S.E.) p value Previous House service 0.252 (0.465) 0.588 0.530 (0.742) 0.476 Party leader 2.874 (1.532) 0.062 4.830 (2.445) 0.049 Years of service 0.154 (0.033) 0.000 0.273 (0.052) 0.000 Party size 0.854 (0.082) 0.000 0.894 (0.130) 0.000 Ideological extremity 17.089 (1.492) 0.000 35.867 (2. 381) 0.000 Bills sponsored 0.008 (0.005) 0.124 0.015 (0.008) 0.062 Bills cosponsored 0.006 (0.002) 0.013 0.016 (0.004) 0.000 Constant 10.831 (4.386) 0.014 5.284 (7.001) 0.451 Observations 1,509 1,509 Number of legislators 260 260 T (avg.) 5.8 5.8 R 2 (between) 0.58 0.62
172 Table 5 2 The Gingrich Senators Senate House of Representatives Name State Tenure Tenure Allard Colorado 105 110 102 104 Allen Virginia 107 109 102 Brown Colorado 102 104 97 101 Brownback Kansas 105 111 104 Bunning Kentucky 106 111 100 105 Burr North Carolina 109 present 104 108 Chambliss Georgia 108 present 104 107 Coats Indiana 101 105, 112 present 97 100 Coburn Oklahoma 109 present 104 106 Craig Idaho 102 110 97 101 Crapo Idaho 106 present 103 105 DeMint S outh Carolina 109 112 106 108 DeWine Ohio 104 109 98 101 Ensign Nevada 107 112 104 105 Graham South Carolina 108 present 104 107 Gramm2 Texas 99 107 98 Grams Minnesota 104 106 103 Gregg New Hampshire 103 111 97 100 Hutchinson Arkansas 105 107 103 10 4 Inhofe Oklahoma 104 present 100 103 Isakson Georgia 109 present 106 108 Kyl Arizona 104 112 100 103 Mack Florida 101 106 98 100 McCain Arizona 100 present 98 99 Roberts Kansas 105 present 97 104 Santorum Pennsylvania 104 109 102 103 Smith New Ham pshire 102 107 99 101 Sununu New Hampshire 108 110 105 107 Talent Missouri 108 109 103 106 Thomas Wyoming 104 110 101 103 Thune South Dakota 109 present 105 107 Vitter Louisiana 109 present 106 108 Wicker Mississippi 110 present 104 110 Note: Gramm was first elected as a Democrat to the 96 th Congress. In January 1983, he resigned his seat, switched parties, and won reelection as a Republican. The data include only his service as a Republican.
173 Table 5 3 Comparison of t otal n etwork and c ospons orship p artisanship b etween Gingrich Senators and their Senate c olleagues, b etween l egislator e ffects Total network partisanship Total cosponsorship partisanship Coeff. (S.E.) p value Coeff. (S.E.) p value Gingrich Senator 0.110 (0.747) 0.883 1.416 (1 .190) 0.235 Party leader 2.869 (1.533) 0.062 4.726 (2.441) 0.054 Years of service 0.153 (0.033) 0.000 0.284 (0.053) 0.000 Party size 0.853 (0.082) 0.000 0.895 (0.130) 0.000 Ideological extremity 16.864 (1.560) 0.000 36.560 (2.484) 0.000 Bills sponso red 0.008 (0.005) 0.119 0.014 (0.008) 0.071 Bills cosponsored 0.006 (0.002) 0.014 0.015 (0.004) 0.000 Constant 10.825 (4.389) 0.014 5.332 (6.988) 0.446 Observations 1,509 1,509 Number of legislators 260 260 T (avg.) 5.8 5.8 R 2 (between) 0.58 0.62
174 Figure 5 1 The i nfluence of p revious House s ervice on s t otal n etwork p artisanship, by Congress
175 Figure 5 2 The i nfluence of p revious House s ervice on s t otal c osponsorship p artisanship, by Congress
176 Figure 5 3 Change in the t otal n etwork p artisanship of s enators with and w ithout p revious House s ervice e xperience, by Congress
177 Figure 5 4 Change in the t otal c osponsorship p artisanship of s enators w ith and w ithout p revious House s ervice e xperience, by Congress
178 Figure 5 5 T otal n etwork p artisanship v alues d uring l s ervice for t hose w ho s erved in b oth c hambers, by Congress
179 Figure 5 6 Total c osponsorship p artisanship v alues d uring l Senate s ervice for t hose w ho s erved in b ot h c hambers, by Congress
180 Figure 5 7 Comparing l m ean t otal n etwork p artisanship in the House and in the Senate
181 Figure 5 8 Comparing l t otal c osponsorship p artisanship in the House and in the Senate
182 Figure 5 9 Total n etwork p artisanship of the Gingrich Senators and their Senate c olleagues
183 Figure 5 10 Total c osponsorship p artisanship of the Gingrich Senators and their Senate c olleagues
184 CHAPTER 6 CONCLUSION The story of Senator Ted Kennedy and the success he achieve d in policymaking throughout his years in the Senate illustrates how the structure and rules of the chamber influence the behavior of its members. As has been shown in the preceding chapters, newly elected legislators to both the House and the Senate are participants in a process of socialization. In the House, freshmen members are segregated along party lines from the very first day they enter the Capitol and are taught to associate strictly with members of their own party. The representatives who do ma ke friends with their partisan opponents feel as though they have to keep these relationships a secret or risk being chastised by more senior members of their own party (Felde 2012). As time passes, they learn that they must vote, cosponsor legislation, a nd fundraise with their same party colleagues in order to win the favor of their party leaders. This loyalty is eventually repaid in the form of favorable committee appointments, leadership posts, support for legislative proposals, and even campaign funds These practices, combined with its large membership and majoritarian rules, make the House a regimented In the Senate, however, new members are provided with the time and opportunity to meet and converse with all members of the chamber, not just those who share their ideological views. While many roll call votes are structured along party lines, leading the DW NOMINATE sc ores of senators to reflect the growing party polarization witnessed using the same measure in the House, legislation typically requires the support (or at least the lack of obstruction) of all senators to even survive to the voting
185 stage. This reality re quires senators to build bipartisan coalitions long before a bill goes to the floor or even to committee. Rather than operating based on party loyalty, the Senate relies on the trust established among legislators of all ideological perspectives and the re sulting reciprocity practiced by them over the course of their service in the chamber. The result of this socialization has often been the formation of cross party friendships, like that of Ted Kennedy and Orrin Hatch, and a willingness to compromise with policies you want: The importance of personal relationships in the Senate cannot be overemphasized, and here, too, Senator Kennedy was a master. He was by nature a charming person, and he worked at establishing and maintaining friendly relationships with his colleagues. It was a rare senator, of whatever Kennedy charmed people into supporting things that they oppose d He and Senator Hatch, for example, were very good friends, but any time Senator Kennedy partnered on a bill with Senator Hatch, it ended up as a genuine compromise representing the views of both (Nexon 2009) Likewise, personal bonds be tween senators can form between two people whom the public might perceive as the most unlikely of friends. I n a statement to reporters and in his memoir, Vice President Joe Biden, who served in the Senate from 1973 to 2009, remarked on the relationship be tween two political rivals, former Vice President and De mocratic senator from Minnesota Hubert Humphrey and Barry Goldwater, Republican presidential nominee and Arizona senator : One of the most emotional moments I saw was when Hubert Humphrey was dying of cancer. He made his last appearance on the floor and he could hardly walk. He walked into the well [of the Senate] and Barry Goldwater got out of his seat and they both embraced each other for a good three minutes, crying. These were arch arch arch id eological enemies. Stone and Green 2012; see also Biden 2008, 133 ).
186 Although Humphrey and Goldwater possessed drastically different legislative priorities and went head to head on the campaign trail, potentially giving some the impression that they were bitter enemies, their common service in the Senate led them to respect and appreciate each other as colleagues. Of course, i t is important to note that there can be exceptions to the rule. In the House legislators are not always bitter partisan rivals who place loyalty to party above all else. One example is the famous friendship between former Democratic Speaker men would fight bitterly on the House floor during the week, then play golf together on Stolberg 2003 ). On the other hand, senators are not always shining examples of honorable behavior. I n 2003 there was an exchange between then Senate Minori ty Whip Harry Reid (D NV) and Senate Majority Leader Bill Frist (R TN) in which ve never seen such amate 2003). And perhaps the most recent instance of cross party, cross chamber animosity Speake r John Boehner (R OH) told now Bresnahan et al. 2013 ). Nevertheless, the core of the argument remains: the rules of the Senate have the uniq ue effect of requiring that its members work together to pass legislation. The power given to each individual senator by virtue of their ability to halt every day proceedings
187 requires that bipartisan coalitions be built long before legislation can be brou ght to the floor for a vote. Moreover, the conditions under which senators serve particularly their six year term and their small number of colleagues provide them with the opportunity to establish these relationships. So why is it, then, that we observe the growth in party polarization in roll call voting in both the House and the Senate? Moreover, is this trend absence of bipartisanship? I believe the answer to the first q uestion lies in the legislative process, itself. The answer to the second question results from the propensity for the media to sensationalize conflict in order to sell their product to the public. I now turn to a discussion of each of these explanations Roll Call Voting and Bill Cosponsorship: Why the Discrepancy? It has been shown clearly throughout the literature on congressional party polarization that party unity on roll call votes has been increasing in both the House and the Senate over time. Att empting to refute this argument would be a fruitless endeavor; however, the growth in party polarization in voting behavior does not automatically imply that legislators in both chambers have stopped cosponsoring bills together, as well. Having shown that despite their increasing party unity in voting, senators continue to build relationships through bill cosponsorship with colleagues on both sides of the aisle, the challenge now is to explain why cosponsorship has not gone the way of voting. First, the e xplanation that seems most obvious is that roll call votes are an activity that is much more visible to the public. This means that, in a world where voters and interest group leaders criticize legislators for voting against their party, an incumbent sena tor who faces a strong challenger during a party primary risks being ousted if he does not vote the party line:
188 to vote in primary elections and who show up in mid [Senate historian Donald Ritchie] about being re elected which is all of them has to be concerned about the more ex treme factions in their parties (Strickland 2010). Such was the fate of Senator Richard Lugar (R IN), the longest serving senator nomination in 2012 to Tea Party candidate Richard Mou rdock after facing charges of being too moderate and not keeping in contact with his constituents, as well as failing to 1 In a similar twist of fate, although with happier endings for these incumbents Senator Joe Lieberman (I CT) lost the Democratic primary in 2006 but recovered his seat by running as an Independent, and Lisa Murkowski (R AK) was reelected in 2010 as a write in candidate after losing her primary to Joe Miller ( Sullivan 2012 ). In each of these cases, the sitting senators were accused of not being conservative (or for Lieberman, liberal) enough to keep their seat. Their ideological and party loyalty was measured largely by their voting record 2 and, particularly in the case of Lugar, m assive amounts of money from outside groups was pumped into the campaign to see the senator defeated. Senator Byron Dorgan (D [W] ithin the past decade, many factors outside the Capitol have changed the way senators vote, legislate, and campaign played to either please or avoid causing problems with the bleacher political radio programs, internet blo gs, and outside groups many of which align with a pol itical party (Strickland 2010). 1 Mourdock ended up losing the general election to Democrat and former U.S. House member Joe Donnelly (Associated Press 2012). 2 For the 2011 congressional session Murkowski and Lieberman were identified by the National Journal as among the senators closest to the ideological middle ( National Journal 2012).
189 These examples illustrate the tremendous influence that factors external to the Senate However, as has been discussed, strict part y loyalty throughout all stages of the legislative process is not a winning strategy in the Senate. While senators may vote the party line in order to maintain their image as a party loyalist, this type of behavior and an unwillingness to build bridges ac ross party lines more generally cause senators to lose the respect and support of their colleagues. At the same time, bill cosponsorship behavior is much less visible to the general public compared to roll call voting. The fact that this practice is some what hidden from view, or at least not as widely reported in the media, gives senators the freedom to be active policymakers, interacting with members of both parties, without risking the potential censure by interest groups and party activists who may pre While it is true that senators often write press releases touting their participation through cosponsorship in bipartisan policymaking efforts, these reports are typically not quantified and relea sed as data to be used as campaign fodder. Most likely due to the ease with which roll call votes can be turned into quantifiable measures of legislative behavior, it is votes, not bill cosponsorship, that have been used by political scientists, interest much easier for activists to explain to the average voter compared to the decision to cos ponsor legislation and its effects on party success. In other words, it is much easier for interest groups to charge legislators with disloyalty by pointing to their voting record and percentage of party line votes than for these activists to convince vot ers that a
190 senator should be replaced because he cosponsored legislation with members of the other party too frequently. Because of this relative invisibility, senators are free to satisfy their electoral goal by playing the role of party loyalist in thei r roll call voting, while also achieving their goal of making good public policy by maintaining bipartisan relationships with their colleagues through bill cosponsorship. In the House, due to the majoritarian rules and hierarchical structure, legislators can rely on a strategy of across the board party loyalty to satisfy their electoral and policy goals simultaneously. Another possible explanation comes from the work of Laurel Harbridge (2011). Harbridge studies party polarization in roll call votes and b ill cosponsorship in the House and shows that while polarization in voting has increased over the last 30 40 years, this is largely due to the fact that House party leaders have been selecting fewer bills that have the support of bipartisan cosponsorship c oalitions to be placed on the congressional agenda for a vote. Perpetually seeking to increase intra party unity and create inter party divisions, the majority party leadership in the House selects bills that will have these intended effects. When the ca lendar is occupied by measures inspiring party line votes, we observe an increase in party polarization. While Harbridge does not study the corresponding relationships in the Senate, we can use her theory to help us understand better the argument of this d issertation. include more polarizing legislation because of the majoritarian system which governs decision making in that chamber. Members of the majority party leadership have the power to decide which bills are voted on and which are not because the Speaker of the House makes these decisions in conjunction with the members of the majority party
19 1 dominated Rules Committee. If the Speaker is not in favor of a bipartisan bill he faces little pressure from his rank and file colleagues to make sure it goes to the floor for a vote. However, the Senate is very different. There is no Rules Committee, and senators can propose amendments that are not germane to the bill being debat ed. This gives individual senators the power to alter the policy agenda on the floor. Although the majority leader does have the ability to propose what bills and resolutions should be considered, it is very easy for a rank and file senator to tie the ch amber up in knots by objecting to a unanimous consent agreement or threatening a filibuster. This indicates that both the majority and minority leaders in the Senate must work together, and with their party rank and file, to reach agreement on the process of bringing bills to a vote. As Majority Leader, Harry Reid does not have the luxury of scheduling bills that will unify his own party along a party line vote and ignoring bills that will potentially divide his party. He has much less control over the a genda compared to Speaker John Boehner. Moreover, knowing that this is a delicate process, senators themselves build bipartisan coalitions through bill cosponsorship in the hope that this will preempt party battles that may occur later on and help their l egislation pass successfully. If it is the case, then, that bipartisanship dominates the legislative process leading up to a floor vote, why is it that party polarization has increased in the Senate in terms of roll call voting? In other words, if rank an d file senators are motivated to establish bipartisan coalitions through bill cosponsorship and the Senate Majority and Minority Leaders must work with each other to see legislation adopted, why do we still observe increases in the polarization of votes in this chamber? One explanation is
192 found in the discussion provided above regarding the role of interest groups and activists serving as outside pressures on the voting decisions of senators. However, another explanation may be that senators divide along p arty lines on roll call votes because they, like most people active in politics, have ideological preferences that divide them along these lines. If the average partisan within the electorate votes loyally for candidates of their own party, why should we expect senators to not engage in similar behavior? While Senator Kennedy may have been willing to forge compromises on legislation with Senator Hatch, despite their divergent views, should Kennedy also have been willing to consistently vote against his ow n agreed came to a vote? Absolutely not. In fact, Harbridge investigates the question of how frequently House bills with the support of bipartisan coalitions actually rec eive a bipartisan vote, and she finds that between 25% a nd 50% of bills with bipartisan cosponsors that reach roll call votes result in a bipartisan vote 1973 and 2000 was 33 percent. This me ans that even when members of the House are able to cross party lines and work together on cosponsoring legislation, on average, only one third of the bills that have bipartisan cosponsors and receive a floor vote actually receive bipartisan support during the vote. This suggests that, at least for the House, bipartisan support during the pre vote stages of the legislative process does not guarantee that there will be bipartisan support when a measure reaches a floor vote. Granted, we might expect the lev el of bipartisan support during a vote to be higher in the Senate, given the amount of effort put forth leading up to the voting stage to solicit
193 support from members of both parties. However, it seems unreasonable to predict that work together and compromise will necessarily result in a corresponding willingness to vote together, as well. The point is that no matter how hard a bipartisan pair of senators works to reach a middle ground on a particular bill, there may always be peo ple for whom the compromise is unacceptable and not worth their vote. Media Bias and the Over Use of the Conflict Frame To answer the second question presented at the start of this chapter, I turn to an analysis of the media and their practice of sensationalizing conflict. One reason why public concern with party polarization is so high is because members of the media prefer to report on fistfights and policy failures, rather than subtle successes. It is more interesting, from a Evan Bayh (Bayh 2010), than to focus on those who return, year after year, to quietly hammer o laced outburst toward Democratic Senator Patrick Leahy (Dickinson and Seabrook 2004) makes for a much more entertaining story than one detailing how two senators from opposing parties met ov cosponsored. More recently, media reports have covered the conflict and obstruction involved with the debate over the fiscal cliff, while ignoring the small victories that have been accompl ished during the same period. The work accomplished between Harry Reid and Mitch McConnell mentioned in Chapter 4 is a good example of this. Rather than reporting on both the successes as well as the failures present in the day to day
194 life of Congress, t he media prefer to dwell on the conflict, failures, and ineffectiveness for which Congress has become known. This view is echoed by one former senator: said Sen. George Voinovich. Of a ll the senators serving their final months in Congress, it's the Republican from Ohio who speaks with the most passion noticed productivity. No doubt, the Senate could novich said. blaming that on the bills that provide the most political drama (Strickland 2010) us to believe that the party polarization observed in roll call voting patterns also pervades other phases of the legislative process. As we have seen, this is not exactly true. Another reason why bipartisan negotiations in Congress are reported so infre quently is because the media and the public often give the perception that compromise is a sign of weakness. Legislators who are willing to give up a little of what they want in order to see their bills adopted are viewed as people who do no t stick to the ir principles: Ritchie says those outside factors influence a fundamental tenet of compromise with disloyalty, a sena tor can be bombarded with negative ads, stories, commentary, or protest (Strickland 2010) Some media reports of bipartisan friendships do exist (see Raju 2011) but in general s enators prefer not to publicize their efforts at bipartisanship and their cros s aisle friendships for fear that their constituents will see them as weak or disloyal During in depth interviews with nine senators who would retire from Senate service in 2010, Ken Strickland, a reporter for NBC News, spoke with the legislators
195 about th the point that many of the good things the Senate accomplishes are not discussed in the media: On Aug. 5,  the Senate passed the largest child nutrition bill in history affecting what students eat every day in public schools. On that same day, the chamber also authorized $600 million for border security and confirmed more than 100 ambassadors, judges, U.S. attorneys, and members of the military. Each was approved unanim ously, and it was all over in a matter of minutes. And you probably didn't hear much, if anything, about it. These measures, none of which was trivial or purely symbolic, such as the re naming of a post office or the creation of a commemorative stamp, are perfect examples of the influential work that goes on in the Senate on a regular basis. Also important to note is that these bills were passed through unanimous consent agreements, a level of support which could not be obtained if the Senate were as polar ized along party lines as the traditional literature would have you believe. Moreover, the examples listed above are not special cases; rather, Strickland reported that during an 18 month period i n 2007 and 2008, were passed through UC [or unanimous consent] passed in the Senate is done so with the support of 100 percent of the chamber. To obtain this supp ort, senators rely on their dense network of personal relationships: command er giving orders to his troop s. have a problem with it. You flush them out on the telephone. You go see them. You tweak it a little bit boom it gets done (Strickland 2010).
196 Senators are capable of engaging in this process because the small size of the c able to make those phone calls. Yet another example of how bipartisan relationships help the Senate to work, even on the most contentious issues, is the vote on the stimulus bill, or TARP. An emergency piece of legislation intended to address the 2008 fiscal crisis and costing the federal government $700 billion was passed by the Senate in a bipartisan vote, 75 24. This success occurred with a Democratic Congress, a Republic an President (George W. Bush), and during the months leading up to the 2008 presidential election between Senators Barack Obama and John McCain. Indeed, the vote was so contentious that it even cost some Republican senators their careers (Strickland 2010) Moreover, These anecdotes illustrate well the argument of this dissertation. Even in times of crisis, the institutional structure of the Senate operates so that members of both parties can come togeth er and compromises can be made. Ultimately, Ross Baker said it best: basic dynamics have changed little. Periodic alarms that warn us that the Senate is becoming like the House usua lly turn out to be false. Complaints from senators themselves, that partisanship threatens to overcome the enforced civility of the ins titution, are usually overdrawn (Baker 1999, 13 4). While particular personalities may temporarily shatter the general te nor of civility in the Senate chamber and the ideological polarization of the parties may give the impression
197 that senators no longer talk to their colleagues on the other side of the aisle, the Senate as an institution persists, continuing to link legisla tors from all ranges of the ideological spectrum in an effort to steer the course of the ship of state toward good public policy.
198 LIST OF REFERENCES CNN 15 June. http://www.cnn.com/2012/06/15/politics/obama interrupted/index.html (December 20, 2012). Aldrich, John H., and David W. Rohde. Government: Revisiting the Electoral Connection, in Congress Reconsidered 7 th ed. Lawrence C. Dodd and Bruce I. Oppenheimer, eds. Washington, DC: Congressional Quarterly Press. Associated Press in House as Republicans, Democrats Go Toe to Lubbock Avalanche Journal 19 July. http://lubbockonline.com/stories/071903/nat_071903039.shtml (December 20, 2012). Challen New York Post 8 May. http://www.nydailynews.com/news/politics/u s senator richard luagar loses primary tea party challenger richard mourdock article 1.1074696 (February 1, 2013). The Traditions of the United States Senate Washington, D.C.: Government Printing Office. http://www.senate.gov/reference/Sessions/Traditions/Orientation_programs.htm (December 10, 2012). Baker, Ross K. 1995. House and Senate NewYork: W.W. Norton. Washington Post 21 March. http://search.proquest.com.lp.hscl.ufl.edu/washingtonpost/docview/409240710/1 3B8CA4F19C683F8A5/2?accountid=10920 (January 3, 2012). New York Times 20 February. http://www.nytimes.com/2010/02/21/opinion/21bayh.html (December 20, 2012). Biden, Joseph. 2008. Promises to Keep: On Life and Politics NY: Random House. Bratton Kathleen A., and Stella M. Rouse 2 Legislative Studies Quarterly 36:423 60. Politico 2 Ja nuary. http://www.politico.com/story/2013/01/the fiscal cliff deal that almost wasnt 85663.html (January 2, 2013).
199 Broder, David S. Washington Post 17 December. http://search.proquest.com.lp.hscl.ufl.edu/washingtonpost/docview/409278258/1 3B1E83FBFF12A611DF/2?accountid=10920 (January 10, 2013). ____________. Washington Post 7 November. http://www.washingtonpost.com/w p dyn/articles/A30857 2004Nov6.html (December 10, 2012). Roll Call Legislative Studies Quarterly 34:297 318. United States Senate: Art and History http://www.senate.gov/artandhistory/history/common/generic/Feature_Homepage _ByrdOrientation.htm (December 10, 2012). Campbell, James E Legislative Studies Quarterly 7:415 22. Cantril Hadley. 1941 The Psychology of Social Movements NY: John Wiley and Sons. Carmines, Edward G., and Lawrence C. Dodd. Congress Reconsidered 3 rd ed., Lawrence C. Dodd and Bruce I. Oppenheimer, eds. Washington, D.C.: Congressional Quarterly Press. Washington Post 18 November. http://search.proquest.com.lp.hscl.ufl.edu/docview/904435353?accountid=10920 (December 10, 2012). Carson, Jamie, Michael H. Crespin, Charles J. Finocchiaro, and David W. Rohde. 2007. American Politics Research 35:878 904. Journal of Politics 72:124 35. Legislative Studies Quarterly 14:67 105. Cox, Gary W., and Mathew D. McCubbins. 2005. Setting the Agenda: Respo nsible Party Government in the U.S. House of Representatives New York: Cambridge University Press.
200 National Public Radio 27 June. http://www.npr.org/templates/st ory/story.php?storyId=1980130 (February 2, 2013). Dodd, Lawrence U.S. Senate Exceptionalism Columbus, OH: Ohio State University Press. Dodd, Lawrence C., and Richard L. Schott. 1979. Congress and the Administrative State NY: Wiley. Washington Post 24 July. http://search.proquest.com.lp.hscl.ufl.edu/docview/409595310?accountid=10920 (December 20, 2012). Represent! Politics, Government, and Public Life for Southern California 27 November. http://www.scpr.org/blogs/politics/2012/11/27/11241/segregation alive and well capitol hill democrats / (December 10, 2012). Feldman Martha S. 1989 Order Without Design: Information Production and Policy Making Stanfor d, CA: Stanford University Press. Fiorina, Morris P., Samuel J. Abrams, and Jeremy Pope. 2004. Culture War? The Myth of a Polarized America New York: Pearson Longman. British Journal of Political Science 34:429 51. Political Analysis 14:456 87. ____________ Social Networks 28:454 65. Call Voting in Journal of Politics 72:1109 28. Republican Takeove American Politics Research 34:159 188. Washington Post 9 July. http://search.proquest.com.lp.hscl.ufl.edu/washing tonpost/docview/307991354/1 3B1E54D33A7C960934/5?accountid=10920 (December 10, 2012).
201 Joint Government/Public Economic s Workshop at Cornell University, 26 August. http://faculty.wcas.northwestern.edu/~lmh735/documents/harbridgeagendacontro lmanuscript7.26.2011cornell.pdf (January 10, 2013). Politico 26 August. http://www.politico.com/news/stories/0809/26482.html (December 20, 2012). The Ryan White HIV/AIDS Program: A Living History http://hab.hrsa.gov/livinghistory/legislation/index.htm (December 10, 2012). Congressional Research Service 28 November. http://www.senate.gov/CRSReports/crs publi sh.cfm?pid=%26*2D4Q%5CK3%0A (February 2, 2013). Hearings of the Subcommittee on Rules and Organization of the House 29 April. http://democrats.rules.h ouse.gov/archives/rules_hear02.htm (December 20, 2012). Washington Post 31 Jan. http://search.proquest.com.lp.hscl.ufl.edu/washingtonpost/docview/408413345/1 3B1E333 0D87E778EB7/14?accountid=10920 (January 10, 2013). Polarized Politics: Congress and the President in a Partisan Era eds. Jon R. Bond and Richard Fleisher. Washington, D.C.: Congressional Quarterly Press. American Political Science Review 90:555 66. USA Today 26 August. htt p://usatoday30.usatoday.com/news/washington/2009 08 26 ted kennedy obit_N.htm (December 20, 2012). Legislative Studies Quarterly 28:225 46. Lancaster, John, and Helen Dewar. 200 Washington Post 31 March. http://search.proquest.com.lp.hscl.ufl.edu/washingtonpost/docview/409106171/1 3B8C96AF412397EBB2/10?accountid=10920 (January 3, 2013).
202 zation and the Vanishing of Marginals: Retrospective Roll Journal of Politics 72:499 512. Legisl ative Studies Quarterly 34:347 73. Congress Reconsidered 9 th ed. Lawrence C. Dodd and Bruce I. Oppenheimer, eds. Washington, D.C.: Congressional Quarterly Press. Louis Meryl R. 1980 Surprise and Sensemaking: What Newcomers Experience in Entering Unfamiliar Organizational Settings Administrative Science Quarterly 25:226 51. March James G., and Johan P. Olsen 1984 Institutionalism: Organizational American Political Science Review 78: 734 49 Partisanship in the California Assembly, 1849 American Journal of Political Science 51:482 97. Matthews, Donald R. 1960. U.S. Senators and their World Chapel Hill: University of North Carolina Press. Datasets for Divided We Govern (1991) http://pantheon.yale.edu/~dmayhew/dataset_DWG04_laws_1991_2002.doc (Accessed December 10, 2012). McCarty, Nolan, Keith T. Poole, and Howard Rosenthal. 2006. Polarized America: The Dance of Ideology and Unequal Riches Cambridge, MA: MIT Press. Washington Post 29 March. http://www.washingtonpost.com/wp srv/poli tics/daily/march99/lott29.htm (January 3, 2013). MSNBC.com http://tv.msnbc.com/2013/01/02/what harry reid and mitch mcconnell did while you werent looking/ (January 10, 2013). National Journal National Journal 23 February. http://www.nationaljournal.com/magazine/in the middle 20120223 (January 10, 2013).
203 Newton Time 23 December. http://www.time.com/time/politics/article/0,8599,1949766 2,00.html (January 10, 2013). fts Health Affairs 28:1040 8. Washington Post 11 December. http://articles.washingtonpost.com/2012 12 11/politics/35767466_1_first lady mckernan political spouses (January 10, 2013). ical Ailments is Required in Washington Post 6 October. http://search.proquest.com.lp.hscl.ufl.edu/washingtonpost/docview/433958699/1 3B8CC7FB424BF1CB34/7?accountid=10920 (January 3, 2013). Pierson, Paul and Theda Skocpol 2002. Historical Institutionalism in Contemporary Political Science, in Political Science: State of the Discipline Ira Katznelson and Helen V. Milne r, eds NY : W.W. Norton. Platt, Matthew B., and Valeria Sinclair Chapman. Solving: Exploring Bill Sponsorship in Post Meeting of the Southern Political Science Association New Orleans, Louisiana. Journal of Politics 46:1061 79. American Political Science Review 62:144 68. Rule 22 Blog 26 January. http://rule22.wordpress.com/2012/01/26/newt gingrich the adversarial news media and camscam/ (January 10, 2013). Politico 7 November. http://www.politico.com/news/stories/1111/67807.html (December 10, 2012). Ring Peter S., and Gordon P. Rands 1989 Sensemaking, U nderstanding, and C ommitting: Emergent I nterpersonal Transaction P rocesses in the E volution of 3M's M icrogravity R esearch P n Research on the Management of I nnovation: The Minnesota S tudies A ndrew H. Van de Ven, Harold.L. Angle, and Marshall S. Poole, eds NY : Ballinger. Ring Peter S., and Andrew H. Van de Ven 1989 n Research on the Management of I nnovation: The Minnesota S tudies A ndrew H. Van de Ven, Harold.L. Angle, and Marshall S. Poole, eds NY : Ballinger.
204 Effect of Race and Ethnicity on American Politics Research 36:130 152. Rohde, David W. 1991. Parties and Leaders in the Postreform House Chicago: University of Chicago Press. School Christian Science Monitor 16 December. http://www.csmonitor.com/1992/1216/16071.html (December 10, 2012). New York Times 11 December. http://www.nytimes.com/1994/12/11/us/gop freshmen enter ready for their exits.html (December 10, 2012). United States Senate: Art and History http://www.senate.gov/artandhistory/history/common/generic/Baker_OrientationP rogram.htm (December 10, 2012). Sinclair, Barbar a 1989. The Transformation of the U.S. Senate Baltimore, MD: Johns Hopkins University Press. _____________ Congress Reconsidered 9th ed Lawrence C. Dodd and Bruce I. Oppenheimer, eds. Washi ngton, D.C.: Congressional Quarterly Press. Skocpol Theda. 1995 Polity 28:103 6. Congress Reconsidered 9th ed. Lawrence C. Dodd and Bruce I. Oppenheimer, eds. Washington, D.C.: Congressional Quarterly Press. New York Times Magazine 15 September. http://search.proquest.com.lp.hscl.ufl.edu/nytimes/docview/21548644 8/13B8CAD F4FD6D84C9B4/4?accountid=10920 (January 3, 2013). Starbuck William H., and Frances J. Milliken 1988 The Executive Effect: Concepts and Methods for studying Top Managers Donald Hambrick, ed. Greenwich, CT: JAI Press. New York Times 17 November. http://www.nytimes.com/2011/11/18/us/politics/congressional breakfast club tackles bip artisanship.html (December 10, 2012).
205 Steinmo Sven, and Caroline Tolbert 1998 Do Institutions Really Matter? Taxation in Industrialized Democracies Comparative Political Studies 31:165 87. New York Times 30 November. http://www.nytimes.com/2003/11/30/weekinreview/the nation the high costs of rising incivility on capitol hill.html?pagewanted=all&src=pm (January 10, 2013). Stone, Daniel, and Miranda Green. Congressional Name Calling, Incivility on the Daily Beast 11 June. http://www.thedailybeast.com/articles/2012/06/11/congressional name calling incivility on the rise.html (December 20, 2012). NBC News 14 September. http://www.msnbc.msn.com/id/39060755/ns/politics the_exit_interviews/#.UOyV1uRi6Sp (January 10, 2013). Washington Post 28 November http://www.washingtonpost.com/blogs/the fix/wp/2012/11/28/ousting a senator in a primary not so easy/ (February 2, 2013). Boston Globe 17 January. http://www.boston.com/bostonglobe/editorial_opinion/oped/articles/2011/01/17/g ood_manners_are_smart_politics/ (December 10, 2012). New York Times 26 May. ht tp://search.proquest.com.lp.hscl.ufl.edu/nationalnewscore/docview/433342378/ 13B1EFA70573DFA7CD/1?accountid=10920 (January 10, 2013). Lost Angeles Times 28 April. http://sea rch.proquest.com.lp.hscl.ufl.edu/nationalnewscore/docview/193782072/ 13B1EF011DA2641C972/4?accountid=10920 (January 10, 2013). Party Politics 12:483 503. _______________. 2008. Party Polarization in Congress New York: Cambridge University Press. Huffington Post 1 September. http://www.huffingtonpost.com/sean theriault/post_782_b_702304.html (January 10, 2013).
206 CNN.com 21 December. http://www.cnn.com/2011/12/21/opinion/theriault gingrich congress gridlock/ind ex.html (January 12, 2013). http://www.utexas.edu/law/colloquium/papers public/2012 2013/01 31 13_theriault_ginrich_senators.pdf (F ebruary 10, 2013). American Political Science Association Washington, D.C. __________________________ ______ Journal of Politics 73:1011 24. Conference on Civility and Deliberation in the Senate, spo nsored by the Robert J. Dole Institute and the Pew Charitable Trusts, U.S. Capitol, Washington, D.C. Esteemed Colleagues: Civility and Deliberation in the U.S. Senate ed. Burdett A. Loomi s. Washington, D.C.: The Brookings Institution. Washington Post 4 March. http://search.proquest.com.lp.hscl.ufl.edu/washingtonpost/docview/409092673/1 3B8C877F74578A2D1D/3?accountid=10920 (January 3, 2013). Nature 393:440 2. Weick, Karl E. 1995. Sensemaking in Organizations Thousand Oaks, CA: Sage. W Paper presented at the Annual Meeting of the Midwest Political Science Association http://www.allacademic.com/meta/p363984_index.html. Zhang, Y. A.J. Friend, Amanda L. Traud, Mason A. Porter, James H. Fowler, and Peter J. Mucha. 2008. Community structure in Congressional Cosponsorship Physica A 387:1705 1712.
207 BIOGRAPHICAL SKETCH Paulina S. Rippere is originally from Key West, Florida. She graduated summa cum laude from th e University of Central Florida with a B.A. in political sci ence and was awarded Honors in the Major for her completion of an undergraduate thesis. At the University of Florida, Paulina earned an M.A. and Ph.D. in political science graduating in May 2013 Her primary areas of specialization are American political institutions, Congress and the legislative process, and political parties. In addition to her dissertation, Paulina is involved in an ongoing research project with Professor Stephen C. Craig which examine s the most effective ways that political candidates can respond when they are attacke d in a negative advertisement. During the 2012 2013 acad e mic year, P aulina served as Visiting Assistant Professor of Political Science at Flagler College in Saint Augustine, Florida In August 2013 s he will join the faculty at Jacksonville University in Jacksonville, Florida as Assistant Professor of Political Science.