‘r i ;M 3 Liam? 517" ” out“: .38 3 in?“ .r J: .LQEW "N“... This is to certify that the dissertation entitled PROBABILITY MODELS OF CONGRESSIONAL CAREER DECISIONS presented by PAUL BRACE has been accepted towards fulfillment of the requirements for PH.D. degreein POLITICAL SCIENCE flauflf (Vat/(mm Major professor / t (‘ Datei’ [(1] ' /Clg // n MS U i: an Affirmative Action/Equal Opportunity Institution 0-12771 MSU LIBRARIES .___ RETURNING MATERIALS: Place in book drop to remove this checkout from your record. FINES will be charged if book is returned after the date stamped below. PROBABILITY MODELS OF CONGRESSIONAL CAREER DECISIONS by Paul Brace A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Political Science 1982 ABSTRACT PROBABILITY MODELS OF CONGRESSIONAL CAREER DECISIONS By Paul Brace Underlying the analysis presented here is the assumption that the motivations of elected officials, whether assumed or discovered, play an important role in shaping interpretations of the activities of representative bodies, and understanding of representation. This thesis seeks to enhance the under- standing of representative bodies by evaluating the career decisions made by members of the United States House of Representatives. These decisions are plaCed in two categories. Decisions by members of the House to purSue higher office manifest progressive ambition. Alternatively, those deciding to leave the House voluntarily rather than seek reelection or election to higher office constitute a second category or class of decisions. ' Both types of decisions were assumed to be the result of a rational choice calculation. Hence, these decisions were held to be a function of the utilities, risks, and costs that were operative at the time of the decision. Paul Brace Beyond these factors, personal attributes such as age and risk taking propensities were posited to influence these decisions. Hypotheses concerning each category of decisions were derived and tested using decisions of members of the House between 1950 and 1976. These hypotheses were first evaluated using crosstabular analysis and were then further scrutinized using multivariate probit models. One of the most substantial influences upon decisions to seek higher office was the presence or absence of an incumbent seeking reelection. Also influencing these decisions were the degree of electorate the House member shared with the higher office, and the type of higher office. These factors point to the importance of understanding the nature of the contest and the resulting prize in understanding decisions to seek higher office. Among factors found to influence retirement decisions, advanced age was found to be one of the most substantial contributors to an increased likelihood of retirement. The influence of age, however, was found to be conditioned by the age at which a member entered the House for the first time and thus pointed to the role variable seniority has upon retirement decisions. ,For Klaudia ii .For Klaudia ii ACKNOWLEDGEMENTS This project was my responsibility, but many others have been instrumental in its completion. Joseph A. Schlesinger, my chairman, instilled in me an interest in political careers and shared with me his wealth of insight into American politics. The other members of my committee, Paul R. Abramson, John H. Aldrich, and David W. Rohde all contributed guidance and channeled my thinking during the course of this endeavor. Four other persons were of immense help in this project. For technical aid and editorial assistance I thank Dolores Siefert. Harriet Dhanak provided assistance in manipulating and maintaining my data set. For assistance in solving an immense computer task and for providing support, refuge, and comic relief I thank my fellow graduate student and friend, Rob Hymes. Special thanks also to Phil Baumann whose constant presence in our office provided me with the incentive to finish. The data utilized in this dissertation were made available by the Inter-University Consortium for Political and Social Research. The Roster of United States Congressional Officeholders and Biographical Characteristics of Members of the United States Congress, 1789-1980 that iii were merged into one file by the Consortium was further merged by me with Candidate and Constituency Statistics of Elections in the United States, 1788-1978, another Consortium data set. The Consortium bears no responsibility for the analysis or interpretations presented here. To my typist, companion, and wife, Klaudia, I am forever grateful. iv TABLE OF CONTENTS LIST OF TABLES CHAPTER I Introduction . Progressive Ambition Retirement The Importance of Career Decisions Conclusion . CHAPTER II Introduction . Decisions to Seek Higher Office The Value of the Higher Office . . The Probability of Winning the Higher Office The Value of the House Seat . Risk Acceptance and Progressive Ambition . Age . . Decisions to Retire The Utility of House Service The Probability of Reelection Plan of Analysis . . . . CHAPTER III Introduction . . The Value of the Higher Office . . The Probability of Winning the Higher Office The Value of the House Seat. . . . . Age. . . . . . . . . . . Conclusion . CHAPTER IV Introduction . . The Value of the Higher Office The Probability of Winning The Value of the House Seat. Personal Attributes. Conclusion Appendix I . Appendix II vii 16 25 32 41 42 47 51 52 57 62 63 69 77 79 96 108 114 119 127 130 135 140 145 148 150 CHAPTER V Introduction . . The Value of the House Seat. The Probability of Returning to the House. Personal Attributes. Conclusion . CHAPTER VI Introduction . . The Value of the House Seat. The Probability of Returning to the House: Personal Attributes. Conclusion . Appendix III CHAPTER VII Introduction . . The Nature of Representation The House. Longitudinal Comparisons Cross-National Comparisons Conclusion . BIBLIOGRAPHY. vi 157 160 170 175 181 185 188 195 198 203 207 209 209 211 214 215 217 10 11 12 13 14 15 16 17 18 LIST OF TABLES House Members Seeking Higher Office by Year, Office Sought, and Election Outcome . Hypotheses Concerning Decisions to Seek Higher Office . Percentage of Members Running for Higher Office, Controlling for Risk Taking and Type of Office Office Seeking When Confronted with Simultaneous Opportunities . Percentage of Members Running for Higher Office, Controlling for Risk Taking and Incumbency Status of Opposition . Percentage of Controlling Percentage of Controlling Percentage of Controlling Incumbency Percentage of Controlling Incumbency Percentage of Controlling Risk Taking Percentage of Controlling Party . . Percentage of Controlling Percentage of Controlling Percentage of Controlling Percentage of Controlling Percentage of Controlling Percentage of Controlling Percentage of Controlling Members Running for Higher .Office,. for Incumbent' s Tenure and Office Members Running for for Competitiveness Members Running for for Competitiveness Members Running for for Competitiveness Members Running for for Competitiveness Members Running for for Competitiveness Members Running for for Office and Size Members Running for for Seniority . . Members Running for Higher Office, of State and Office Higher Office, of State and HigherOOffice,o of State and Higher.0ffice,. of State and HigherOOffice,. of State and Higher.0ffice,. of State . . . Higher Office, Higher.0ffice,. for Seniority and Party . Members Running for for Vulnerability . Members Running for Higher.0ffice,. Higher Office,. for Marginality and Risk Taking . Members Running for for Redistricting Members Running for Higher Office,. Higher Office,. for Redistricting and Office vii 70 75 78 80 81 83 86 87 90 92 93 95 98 99 102 103 106 107 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 4o 41 42 43 44 45 Percentage of Members Running for Higher Office, Controlling for Age and Risk Taking . Percentage of Members Running for Higher Office, Controlling for Age and Party . Percentage of Members Running for Higher Office, Controlling for Age at which they First Entered Congress and Party . . . . . . . . . . Results of Probit Estimation . . . The Effect of the Value of the Higher Office Upon the Likelihood of a Member of the House Seeking Higher Office . . . . The Effect of Probability of Winning Upon the. Likelihood of a Member of the House Seeking Higher Office . . The Effect of. Probability of Winning Upon the Likelihood of Members of the House Seeking Higher Office . . . The Effect of Risk Taking Propensities Upon the Likelihood of Members of the House Seeking Higher Office The Effect of Age Upon the Likelihood of Members of. the House Seeking Higher Office First Occupation of Member by Means of Leaving House Hypotheses Concerning Decisions to Retire . Percentage of Members Retiring, Controlling for Seniority Percentage of Members Retiring by Age at Entering Congress for the First Time Percentage of Members Retiring, Controlling .for Party Decisions to Retire by Year . . Retirements per Year and Congressional Salary in 1967 Dollars . Percentage of Members Retiring, Controlling for. Margin of Victory . . . . . . . . . . . . . Percentage of Members Retiring, Controlling for Redistricting . . . . . . . . . . . . . . Percentage of Members Retiring, Controlling for Risk Taking . . . . . . . . . . . . . . Percentage of Members Entering Congress at Age of Entry by Risk Taking . . . . Percentage of Members Retiring by Age Results of Probit Estimation . . The Effect of Utility of House Seat Upon the Likelihood of Members of the House Retiring The Effect of Probability of Returning to the House Upon the Likelihood of Members of the House Retiring . . The Effect of Risk Taking Propensities Upon the Likelihood of Members of the House Retiring . The Effect of Age Upon the Likelihood of Members of the House Retiring Estimated Impact of Age, Conditional Upon Seniority on Retirement Decisions of Members of the House viii 109 111 113 126 128 132 136 196 200 201 204 Chapter I Congress is a unique organization. The Constitution authorized the body to organize itself as it saw fit. Over time it elected to follow principles of organization such as rules, hierarchy, and division of labor that make it corres- pond quite closely to the Weberian image of bureaucracy in many respects. A distinctive feature of Congress as an organization is the body's inability to control who is and who is not a member. The ultimateauthority for determining membership lies outside the organization. As Anthony Downs has stated, "[e]very man's self interest leads him to be most responsive to those who decide whether he shall retain his primary position..." and "persons elected by constituents outside of the organization employing them have a primary 'loyalty' to those constituents rather than to the organiza- tion itself."1 The incentive system that the member confronts resulting from this arrangement and the ways members respond to this incentive system are at the heart of the question of the nature of representation and the characteristics of representative institutions. Differing views on the motivations of elected repre- sentatives have had an impact on Congressional research. For example, David Mayhew has placed primary emphasis on the goal of reelection as a motivation of congressmen. By assuming this to be the preeminent goal of congressmen, he argues that the members of Congress have organized the institution in a manner that allows them to fulfill their reelection needs remarkably well. Congressmen, in pursuit of reelection, use various means to satisfy their constit- uents; and the institution, to Mayhew, becomes less an instrument for public policy decision making than it is a vehicle for their reelection, ultimately rendering it in- Capable of coming to terms with important national issues.2 Morris Fiorina has also used the reelection goal as a founda- tion for a broad argument concerning Congress.3 The motiva- tion to be reelected has led Congress to establish and finance massive federal programs in an effort to appeal to the demands of their constituents. Once established, these massive endeavors have not always functioned well; and, according to Fiorina, congressmen have seized this as an opportunity by shifting their role to that of ombudsman, helping their constituents cope with the massive bureaucracy. Congress, then, in its narrow interest with reelection, created a monster and now does battle with this monster, again helping to ensure reelection and, furthermore, doing so without having to engage in the potentially disadvanta- gous (to them) practice of making decisions on controversial and important national issues. While few, if any, students of Congress would argue that the reelection goals of congressmen are unimportant, some give them less prominence in their interpretations of legislative behavior than do the above authors. Richard Fenno, for example, acknowledges the reelection goal but also evaluates the roles desire for influence in the House, good public policy, and a career beyond the House plays.4 He found that members' giving prominence to the goal of reelection, influence in the House, or good public policy leads to distinct patterns that serve as a basis for pre- dicting gross similarities and differences in committee behavior. These differences, combined with differing environ- mental constraints, lead each committee to adopt sets of’ strategic premises designed to achieve their members' goals. The output of committees, and of Congress (these being the decisions they reach), are explainable in terms of differing member goals and their interaction with environmental constraints. Lawrence C. Dodd is another Congressional scholar willing to ascribe goals to congressmen other than reelection. As Dodd states, were members solely preoccupied with reelection, we would expect them to spend little time in Washington and devote their personal efforts to constituent speeches and district casework. If this were the case, "one would expect Congress to be run by a centralized, efficient staff who, in league with policy-oriented interest groups, would draft legislation, investigate the issues, frame palatable solutions, and present the members with the least controversial bills possible."5 To Dodd, reelection is only the first orientation in a four-tiered career path congressmen climb in their quest for power. Although the first tier is never fully completed, many members are capable of shoring up their electoral base through casework and move on to a second stage in which they broaden their horizons and seek service on key committees. Some move on to a third stage and seek service on power com- mittees; and a very few move on to a fourth stage in which they seek party leadership. Dodd believes that the members of Congress have actively sought to design a congressional structure and process that would maximize their ability to exercise personal power within Congress, this structure being the committee system. Committee government, while attempting to satisfy members' individual desires for personal power, also serves in Dodd's view to disperse internal Congressional authority so widely "that the resulting institutional impotence cripples the ability of Congress to perform its constitutional roles..." and leaves Congress "with the inability ... to make national policy or ensure policy implementation.”6 As these works suggest, how you view the motivations of Congressmen ultimately has a tremendous impact upon how you interpret the activities and outputs of the body as a whole. It would appear that the degree to which one emphasizes re- election as a motivation for congressmen ultimately shapes one's interpretation of the function and disfunctions of Congress and, ultimately, politics. Elections are, of course, important to both interpretations of Congressional behavior; for, even in views that do not emphasize reelection as a sole or even primary motivation, it remains a necessary condi- tion for the fulfillment of other goals. Hence, an implicit assumption underlying each perspective is that elections matter, although disagreement may remain as to their ultimate impact upon congressmen's behavior. That elections matter in any meaningful way presupposes that a representative's constituents will vote for or against him on the basis of their Opinion of his performance in Congress. Roughly three decades of scholarly examination of voting behavior have cast serious doubt on the assumption that the electorate votes in an informed manner on the performance of elected representatives. Miller and Stokes, for example, stated that the electorate sees very little al- together of what goes on in the national legislature.7 Their research suggested how little is known about the two parties“ legislative records among constituents, and even less is known about the part played by their own congressman. The general lack of information held by the electorate, as well as the substantial degree of apathy demonstrated by the general public concerning most public issues, presents a substantial challenge to normative democratic theories that suggest that elected representatives will act in the public interest because of the existence of elections that can be invoked as sanctions against elected officials whose perfor- mance in office does not correspond to the public's wishes. As has been noted, "the individual voter was not all the 8 theory of democracy requires of him." The question of the relationship between leaders and those led is at the heart of democratic theory as well as a key feature of how one interprets the activities of representative bodies and of politics itself. That representatives are responsive at all to the public's wishes in the absence of widespread political participation is one of the more enigmatic features of American democracy. The key to this enigma has been provided by Joseph A. Schumpeter in his Capitalism, Socialism and Democracy. To Schumpeter, ”the Democratic method is that institutional arrangement for arriving at political decisions in which individuals acquire the power to decide by means of a competitive struggle for the people's vote."9 The crucial element is competition for office. Schumpeter suggests that leaders will naturally attempt to mobilize public Opinion to ensure election. Once elected, leaders are immune from popular pressure unless they choose otherwise. As Schumpeter states, "Democracy means only that the people have the opportunity of accepting or "10 From this perspec- refusing the men who are to rule them. tive, then, "rule by the people” merely means deciding on rulers, not necessarily government by or for the people. Within this conception of democracy, it is the threat of a potential challenge and the electorate's ability to remove them that induce the elected official to be cognizant of the public's interest. For this inducement to function, it is necessary for elected officials to wish to retain their position. If they are not motivated to retain their position, there is no necessary reason for them to respond to their constituents' wishes. As Joseph Schlesinger'has stated, "Representative government, above all, depends upon a supply of men..." driven by ambition for "the desire for election and, more important, for reelection becomes the electorate's "11 A fundamental con- restraint upon its public officials. dition for the functioning of democracy is the presence of office holders who are motivated by the desire to retain their position. Motivations, or goals, are difficult to measure realis- tically. We typically make assumption about the goals of individuals and evaluate their behavior in light of the assumptions we make about their goals. More specifically, we evaluate the decisions individuals make in light of goals we assume them to hold. As noted at the outset, the assumptions one makes about the goals of members of the United States House of Representatives greatly influence the explanations one forwards for the operating characteristics of that body. A necessary condition for fulfilling any of the postulated goals of the members was reelection, while the differences between the perspectives seem attributable to the degree of sufficiency they attach to the reelection goal. In each case behavior is evaluated in terms of the decisions indivi- duals make in light of goals they are asserted to hold. The decisions of elected representatives in light of goals we postulate them to hold are important to the under- standing of the institution they labor in and the nature of representation. An important decision to all of the above is that individuals should decide to try to retain office. The explanations of Congress would all fail if the predominant decisions of elected representatives were not to seek reelec- ion. Of course, the decision to seek reelection is in contem— porary times the most common career decision manifested by members of the House. As Morris Fiorina points out, "[slince World War II nearly 90 percent of all incumbents have sought reelection in any given election, and approximately 90 percent "12 These figures of all those who ran were successful. certainly testify to the veracity of reelection as a prominent and perhaps a sufficient goal of members of the House. None- theless, in a given year between roughly six and thirteen percent of the members decide to leave the House voluntarily, and it would appear that fuller understanding of these decisions could greatly enhance our picture of the House and of representation in general. The decisions to leave the House are not, of course, all for the same apparent reasons. Some decide to leave to pursue higher office; some leave for state, local, or federal office; and, finally, some leave to return to private life. For analytical purposes it would seem reasonable to delineate these decisions into two basic categories. The first, seeking higher office, concerns those decisions that are oriented to the pursuit of what is clearly a higher office. Specifically, decisions to seek a Senate seat or governorship are placed in this category while other offices, such as judicial, federal, or state and local, are deleted due to their conjectural status as being higher of lower than the United States House. The other category concerns decisions to leave the House either to return to private life or to pursue some lower political office. The former category could be said to contain members of the House who exhibit progressive ambition while the latter might be classified as those who chose "opting out," a term applied by Stephen Frantzich to members who left the House for another career or relaxation, but here extended to include members who 13 At this point it would sought lower offices as well. prove useful to examine some of the findings of past research on progressive ambition and retirement. Progressive Ambition The logical starting point for the consideration of progressive ambition is the work of Joseph Schlesinger. Schlesinger created a threefold categorization of ambition: those seeking higher office were classified as having progressive ambition; those seeking to maintain their current position were classified as having static ambition; and those whose ambition was limited to serving a current term before retiring were classified as having 14 discrete ambition. "The central assumption of ambition theory," writes Schlesinger "is that a politician's behavior is a response to his office goals."15 Schlesinger thus points to the importance of motiva- tion, but this tells only part of the story. Ambition interacts and is conditioned by the situation the 10 individual finds himself in and the opportunities he con- fronts. Although a politician may desire a higher office, his situational status may lead him to attenuate his ambition to retaining his current office or leaving public life altogether. The opportunity structure serves to condition ambition. Schlesinger finds a multitude of outlets for political ambition in the great number of offices contested. This opportunity structure is an open system, with lines of advancement unregulated legally and with few prerequisites for most offices. However, offices are held by fixed terms which ensures a constant possibility of turnover which works both to increase opportunities and at the same time serves to create risks for the politician; especially in the case where one office must be sacrificed to seek a higher one.16 Schlesinger's theory Of ambition points to the importance of evaluating a politician's behavior, not so much in terms of where he has been or where he is currently, but in terms of where he wants to be. The actions a politician takes today are assumed to be oriented toward the electorate he seeks to appeal to in the next election. Schlesinger's work serves to underscore the importance Of the politician's ambition, and how the ambition interacts with the politician's situation and the political Opportunity structure in shaping his behavior. As Kenneth Prewitt argues, ambition theory adds another dimension to research on leadership and 11 representation by holding that expectations of the future shape current behavior.l7 What are the specific factors that shape ambition and how do they Operate? Schlesinger provided initial investi- gations and suggestions that served to guide subsequent research. One dimension of ambition Schlesinger acknowl- edged but did not investigate was the influence that social background and personality traits might have upon 18 John Soule explored these dimensions in his 19 ambition. sutdy Of 97 of 110 Michigan state legislators. In terms of social background, Soule found that legislators with progressive ambitions were more likely to have had an early socialization to politics and were more likely to report family or school as the agents Of their initial political socialization. In terms of personality traits, the progressively ambitious were found to be more likely to value leadership or power and less likely to value doing things for other peOple, being treated with under- standing, Or doing what is socially correct. Soule also found that ambition was related to the legislator's representational role orientation. Ambitious members were found to be more likely to use their own judgments in decision situations than to take instructions from their constituents. He also found that the ambitious were more likely to base their decisions on what they believed best for the entire state than for their constituency. This finding led Soule to conclude that, 12 "Edmund Burke's classical conception of a worthy legis- lator is most closely reflected by legislators who hold progressive ambition."20 David Rohde has provided insight into another 21 psychological aspect of progressive ambitions. Rohde assumes, in his study of the members of the United States House of Representatives, that if upon their first day in that chamber they were offered a Senate seate or governorship without cost or risk, they would take it. To Rohde, it is the costs and risks associated with Opportunities that shape progressive ambitions. Cost and risks do not tell the whole story, however, Rohde contends that two members could be confronted with identical Opportunities, in terms of costs and risks, and one member might take the opportunity while the other does not, because it is still possible that they differ in their intensity or preferences for risky alternatives. Rohde thus invokes the concept (or psychological trait) of risk-taking to explain differences in the numbers of Opportunities taken for higher Office by members of the House. He categorizes as high risk-takers those members who, when they first ran for the House, challenged an incumbent or ran in a district substantially favorable to the other party. He finds that high risk-takers are indeed proportionately more likely to seek higher Office than are the lower risk-takers. Risk-taking and its relation to progressive ambition could, it would seem, 13 have a notable impact upon the nature of a politician's representational posture. Riker and Ordeshook contend that risk-taking is a component of leadership.22 They feel that leaders may have a higher prOpensity to seek risky prospects than do followers, who prefer the costless alternative of doing nothing. Perhaps Burke's worthy legislator was possessed of the characteristic of being willing to bear the risks associated with ignoring the immediate and narrow interests of his constituency in pursuit of broader solutions to national concerns. Burke himself must have been willing to bear substantial risks, since in the first election after giving his now famous_ speech to the electors of BriStol, he was voted out of office.23 Another important influence upon ambition suggested by Schlesinger is age. He stated that the age cycle restricts a person's political chances; reasonable expectations in one period of life are unreasonable at another. The individual politician, according to Schlesinger ". . .constantly faces the problem of reconciling two everchanging schedules, the timetable of Office and the timetable of age."24 Age operates in two ways. On the one hand, Schlesinger as well as Harvey Lehman25 and David Walker26 all argue that age in recruitment is a factor that affects the professional outlook of a politician. In Schlesinger's opinion, "the younger a man is when he enters politics, the 14 greater the range of his ambition and the likelihood of H27 On his developing a career committment to politics. the other hand, advancing age either in recruitment or in office can work to diminish expectations concerning higher Office. Jeff Fishel, for example, presents evidence that age is related to the career commitments of congressional challengers.28 Paul Hain, in an examination of state legislators, found that "the older a politician is the less likely he is to express ambitions to advance, to 29 seek advancement, or to advance." Hain argues that the attenuation in ambition resulted from what he labeled a "middle'age crisis," with the resulting decline in ambition being abrupt as Opposed to gradual or continuous. A similar pattern between age and ambition is provided by Kenneth Prewitt in his examination of city councilmen.3O At the same time, however, Prewitt's subjects from this lower level of government were prOportionately less likely to express progressive ambition across all age categories, suggesting the potential importance of locational factors in further conditioning ambition. Locational or structural determinants of ambition were examined by Gordon Black.31 Schlesinger had earlier asserted that some offices were substantially more likely to promote political advancement than other offices, and he characterized these as the "opportunity structure" of the political system. Black, in his study of 435 city 15 councilmen in the San Francisco Bay area, elaborates a model of political ambition based upon the principles of utility theory. Within this framework, Black suggests that politicians tend to make decisions on the basis of the costs, benefits, and probabilities that operate at the time of their decisions. He is led to conclude that the structural characteristics of systems of political Offices can shape both the risks that politicians face and the investments that are required to reach political office. To Black, structural variations within systems of offices create barriers that aspiring politicians must overcome. The difference in ambition found by Hain and Prewitt would be explained by Black as the result of differential investments in political activity between state legislators and city councilmen. Black asserts that "as a politician invests in one Office, even if he has little desire at the time to seek higher Office, he is altering his evaluation of other offices in the political career sequence."32 To Black, as the politi- cian's investment increases, his evaluation of political alternatives is likely to become more positive while his evaluation of nonpolitical alternatives remains about the same. As Schlesinger's original work suggested, and as subsequent research has.supported, ambition is an important explanatory factor in understanding the behavior of politicians. Ambition has been shown to vary as a 16 function of age, structural arrangements, and psycho- logical traits. Ambition and factors influencing it, can work to shape how a politician views his representa- tional role. To understand variations in the motivation of politicians and the nature of representation, we would do well to attempt to understand the independent influences upon ambition and relate their impact to the political process. Before considering the impact of ambition on the political process, however, we need to consider another type of decision or behavior manifested by some politicians. Retirement Retirement, or more specifically, voluntarily leaving a legislative body as Opposed to being defeated, is at the heart of a number of important issues. The question of representation itself could be influenced by retirement. For an election to serve as a potential sanction that can be invoked by the electorate, it must exist as a potential threat to take from the Officeholder something he wants. This something is the office he holds; and for the representative who has decided to retire, the election poses a hollow threat. If a representative has decided to retire, there is no necessary reason to expect him to be responsive to his constituents or to act in what he perceives to be the public's best interest. l7 Retirement also speaks to the nature of service. High turnover in a legislative body could indicate one of two things. If such turnover were a function of defeat, we might conclude that a high degree of competition were present. Such competition would suggest that a high value is placed upon service. On the other hand, a high degree of turnover due to retirement might also suggest that service was worth attaining but not worth retaining, a career structure resembling "a fast game of musical chairs."33 While retirement might not speak to the exact motivations of retirees, it does suggest very strongly that continued service was not their primary motivation. In the contemporary House efforts at reelection are the norm and success is very high. This, however, has not always been the case. Douglas Price, in examining two nineteenth century periods, (1811-1820 and 1887-1896) finds that both were marked by very high turnover. In the earlier period, however, only 49 of 465 departures could be attributed to electoral defeat, while in the latter period defeat accounted for 309 of 750. He asserts that "what the early House lacked was not safe seats, but a desire and incentive to retain one's seat."34 Price points to the realignment of the 18905 and iother factors as reducing two-party competition. The high degree of competition prior to this realignment suggests that a high value was in fact placed upon House 18 service by the end of the nineteenth century. He concludes that "as the possibility of winning repeated elections went up . . .the desire to pursue a career in the House also went up."35 Nelson Polsby considers the decline in turnover to have resulted from the institutionalization of the House of Representatives. By institutionalization, Polsby means that over time it has become noticeably more bounded, more complex, and more universalistic and automatic in its internal decision making. In terms of voluntary retirement, Polsby believes institutionali- zation led to increased membership stability; and this increased stability worked to produce a more stable and attractive institution for its members, presenting more Of an incentive to stay.36 - Fiorina, Rohde, and Wissel, in their extensive analysis of congressional turnover from 1789 to 1973, conclude that the level of party competition characteris- tic of each party system helps to explain the secular 37 In terms of decline in congressional turnover. retirement, the authors contend that a member's decision on whether or not to voluntarily retire depends on the value of the House seat to him and the likelihood of defeat if he chooses to run again. Although finding support for the latter proposition, the authors could not, either due to the weakness of the measure or because of counteracting effects, find substantial influence 19 upon retirement attributable to the value of the House seat. Samuel Kernell discounts the role played by declining competition. He found that the percent of incumbents running for reelection increased in districts where there had been recent party turnover as well as in those where there had not.38 Kernell feels that forces in addition to declining competition were converting the House into a body of professionals. He believes ambition to be the primary source of declining turnover. The primary force for membership stability, Kernell believes, can be found in the men who ran for Office. He asserts that the "musical chairs Of the early nineteenth century were replaced by a tacit consensus among 39 politicians about appropriate career development." A hierarchy emerged, and no longer was every politician eligible and equally likely to seek every other Office. With the emergence of hierarchy and the limitations it imposed upon Options and probability, congressional careers became "static" in Schlesinger's terminology. While rates of turnover, and the characteristics of that turnover (i.e., defeat or retirement) are an important component of these and other explanations of the development of careerism and the institutionalization of the House around the turn of the century, they have also been a concern for students of the contemporary Fkn1se. Charles Bullock, for example, examines patterns 20 of attrition and longevity in the House in the forties, fifties, and sixties.40 He found that defeat caused few of the senior members to leave the House during the 19405 (15.2 percent) and 19505 (12.3 percent). During these decades, voluntary departure was the most frequent reason for career termination. This pattern was greatly altered during the 19605, when primary and general election defeats came to account for more departures than either death or retirement among what he defines as careerists.41 Bullock attributes this change to redistricting, newly mobilized groups pushing new issues, and the increasingly national character of two party competition. Contemporary Congresses have witnessed an increase in voluntary retirements. A recent piece by Joseph Cooper and William West provide figures showing that in every Congress except one since 1966, voluntary . . 42 ret1rement has 1ncreased. The authors explore some traditional explanations for retirement; specifically, age, political vulnerability, and political ambition, and find none of these satisfactorily explain the increase. In terms of age, they find that for the period they examine (1957-1979) the average age of voluntary retirees declined along with the decline in the general age of House members. They also find that there was a somewhat greater dispersion of voluntary retirements across age cohorts. 21 In terms of political vulnerability, the authors note that most voluntary retirees come from safe districts. They also point out that the proportion of safe retirees has increased during the period they examine. They also dismiss redistricting, because the pace of redistricting does not correspond to the increase in retirement. Finally, the authors dismiss political ambition as an explanation for the increase in voluntary retirements. (Note: The authors include the progressively ambitious among voluntary retirees.) They find only a modest increase in members' leaving the House to pursue or be appointed to higher Office during the 19705 as compared with earlier periods in their study. They also note that the success rate of those pursuing another office had declined in the 19705, leading the authors to conclude that something other than high expectations for another Office was responsible for leading members to leave. ‘The authors posit disaffection with House service as the predominant factor influencing the recent increase in voluntary retirement. They cite numerous retirees' quotations appearing in newspapers, magazines, and the like that point to the increasing demands placed on House members' time, the strains put on their family lives, the diminishing utility derived from chairmanships, the proliferation of single issue groups, and other comments that point to increasing costs and diminishing 22 rewards of House service. They also note that retirement has grown among members under sixty-five who have achieved leadership positions. This increase among members who should be most resistant to retirement serves, the authors believe, to substantiate the notion that disaffection is a primary factor in increasing retirement. To Cooper and West, voluntary exits are a functiOn of disaffection, and this disaffection stems from growing costs of service stemming from increased work- loads as well as decreasing benefits in the form of less powerful chairmanships, salaries that have not kept pace with inflation, and other factors. The authors believe that this alteration of benefit-cost ratio of House service has had an impact upon the norms and structure of the House. They assert that as disaffection for continued service grows, so too will reluctance to accept formerly dominant norms such as seniority, reciprOCity, and civility, characteristic of the House when members wanted long and stable careers. Stephen Frantzich examines data concerning members who left the House between 1964 and 1974; and, although he doesn't address the temporal issues cited by Cooper and West, his choice of years to examine was fortuitous in that it provides further insight into retirements during this period.43 Like COOper and West he also dismisses the political vulnerability hypothesis, finding that the electoral record of retirees is generally quite 23 strong. Also, although voluntary retirement is found to increase with age, Frantzich notes that over half those retiring were under the traditional retirement age of 65. He also finds that a major influence upon those who decided to leave the House was job frustration. In a subsequent article, Frantzich goes into more depth in exploring the basic notion that diminishing job desirability leads to most voluntary retirements.44 He finds that disability associated with age is an important component of some retirement decisions, but fails to tell the whole story of retirement. Vulnerability, not in objective terms, but in terms of the member's perceptions, is another component of retirement for some members. Redistricting, causing some members to face other incumbents in a combined district or making them face a drastically altered and largely new electorate, according to Frantzich, explains why retirees come disprOportionately from recently redistricted constit- uencies. Finally, declining desirability of House service, stemming from dissatisfaction with the Oppor- tunities House service provides, and outside opportunities constitute a third factor in some retirement decisions. Frantzich believes the decision to retire results from an amalgam of influences but that disability (age), desirability, and vulnerability are important. The contentions of Cooper and West, and of Frantzich are supported by the results of interviews conducted by 24 John Hibbing with a majority of House retirees in 1978.45 Hibbing delineates the causes of retirement into two categories: direct costs and opportunity costs. Among direct costs of House service he finds to influence retirement decisions are familial sacrifices, constant public scrutiny, disrespect for public officials, the lack of security inherent in having to seek reelection every two years, and the distastefulness with with retirees viewed the fund raising process. Among Opportunity costs, he finds that the congressmen's salary compared with what they perceive they could be earning outside of politics, and legislative frustrations stemming from the inability to accomplish tasks in the post-reform House are both antecedents of many retirement decisions. Another reason for at least some of the more frequent retirements in the 19705 has been presented by Albert Cover.46 He suggests that the growing careerism of the fifties and sixties had to give at some point. Growing careerism led to a large number of individuals in the group most prone to retire - the elderly. The result, declares Cover, is a "greening" of the Congress as relatively junior members flow into the House. While most authors testify that electoral vulner- ability is not a primary cause of retirement, some authors assert Otherwise. Hibbing, for example, contends that when the effects of age are controlled for, "we 25 find that those who are not as safe as we would expect on the basis of their age are indeed more likely to retire 'voluntarily."47 James L. Payne examines the margin of victory of members who retired between 1968 and 1976. He finds that retirees serving more than three terms, on average, experience a net decline in electoral margin between their next to last and last elections. This result suggests that long term electoral performance may very well be related to retirement decisions.48 The Importance of Career Decisions As the discussion to this point suggests, the motivations of elected representatives, as manifested in their career decisions, play an important role in our understandings of the Operating characteristics of representative institutions. At this point it might do well to address the implications progressive ambitions and retirements have upon the legislative process. We can begin by examining past research upon the impact of progressive ambition. Perhaps the most important implication of progressive ambition stems from a major tenet of Schlesinger's ambition theory. He states that "the constituency to which a legislator is responding is not always the one from which he has been elected, and that it is more important to know what he wants to be than how he got to be where he is not . . .a politician's behavior is a 26 49 This fundamental tenet response to his office goals." has led numerous authors to evaluate the effect differing ambitions have upon the roles politicians assume. Prewitt and Nowlin, in a manner consistent with the major tenet of Schlesinger's theory mentioned previously, hypothesize in their study of San Francisco Bay area city councilmen that ambitious councilmen are more likely to have broader policy perspectives that reflect their orientation toward the level of government to which they aspire.50 They find that an ambitious Office holder will take on characteristics ”he perceives to be those of incumbents in more elevated posts and will take stands on issues Which reflect their views."51 They also hypothesize that incumbents who aspire to and expect to gain a higher political office will be more favorable to polities which expand the powers Of the higher Office. They found that those who had expressed ambitions toward federal, state, and regional offices were more favorable toward expanding the prerogatives ot those offices. Just as the above authors found ambition to shape governmental orientation, John Soule found a relationship between the ambitions of officeholders and their representational orientations.52 As mentioned previously, he found that progressively ambitious state legislators were more oriented to the state as a whole, or were more likely to try to balance the interests of their district and the state than were the state legislators who were 27 not progressively ambitious. The less ambitious were found to attach greater importance to the interests of their districts. Soule argues that the less ambitious are more likely to assume the role of delegate, while the progressively ambitious are more likely to assume politico or trustee orientations. Prewitt and Eulau evaluated the interplay between the public's power to select those who govern and the existence of responsible behavior toward the public on 53 Their findings suggest the part of elected officials. that the electoral sanction can serve to keep legislators responsive to the public. They find that responsiveness as a function of the electoral sanction is based upon ambition. Members who attach little importance to a political career tended to use their own preferences as a guideline for policy making. Also characteristic of such officeholders was turnover by resignation rather than electoral defeat. In situations where electoral defeat was more common, members tended to exhibit a heightened concern for public Opinion. To the authors, the ability of council members to follow their own views without electoral retribution is a function of the level of community support. When community support for councilmen is high, they can follow their own judgment without fear of electoral retribution. In another piece, Prewitt examines the presence of volunteerism, or more specifically, the situation 28 where the movement into and out of Office is regulated by self-selection and self-elimination patterns rather 54 He feels that this than electoral challenges. volunteerism in the city councils he examined is a function of four characteristics. First is appointment rather than election. Second is low voter turnout. Third is the infrequency of electoral defeat; and fourth are high rates of voluntary retirement. Prewitt feels that volunteerism undermines accountability. Furthermore, he believes that volunteerism in political recruitment, trusteeship in political representation, and ritualism in political elections form a compatible package. He concedes that volunteerism only impacts upon electoral accountability and not on whether representatives consult the preferences of their constituents. He believes that the degree to which such consultation takes place when there is volunteerism, ”is for reasons other than accountability forced by electoral sanctions.SS As the above research serves to emphasize, ambition bears an important influence upon modes of representation and the content of politics. While the impact of progressive ambition is relatively clear, the role of discrete ambition or retirement is less clear. Prewitt's work suggests that retirement, as Opposed to electoral defeat, is a component of volunteerism; and he views this to be compatible with a trustee orientation. This is indirectly compatible with Davidson's finding 29 that the proportion of congressmen who perceive themselves to be trustees is higher among senior members than among junior and freshmen members.56 It is also compatible with Bullock's notion that as the number of senior members in the House swells, "the likelihood is greater that congressional decisions will be based on legislator preference rather than on perceived constituent desires."S7 These are compatible, since long service necessarily suggests the absence of electoral defeat and frequently entails low voter turnout in the respective careerists' elections; and careerists often engage in self-elimination, although Bullock did notice an increasing trend toward defeat in the 19605. The research concerning retirement basically examines it in two ways: one as the result of other processes, the other as a causal factor in its own right, influencing other processes. Our knowledge concerning the former is much fuller that that concerning the latter. Schlesinger provides us with perhaps the most general notion of what widespread retirement could indicate for a political system. He notes that "a political system unable to kindle ambitions for office is in as much danger of breaking down as one unable to 58 While it would be unrealistic restrain ambitions." to suggest that our political system, and the United States House in particular, was unable to kindle ambitions, the 19705 did witness increased retirements from that 30 body, suggesting that for an increasing number that institution did not provide sufficient conditions for the kindling of their office ambitions. There is little empirical evidence concerning the impact of retirement upon the House, although some authors have supplied conjecture as to its impact on that body. Stephen Frantzich provides some findings concerning the influence of retirement. He contends that "retirement removes from Congress some definine types of representatives, and allows their replacement with others."59 He finds some support for the notion that retirees reduce their effort prior to leaving the House. He also finds that retirement in the period he studied (1966-1974) clearly removed some of the most conservative members from the House, opening the door to potential change through turnover.6O His findings suggest that when ideological change does occur (although a minority of the time), it is in the liberal direction. He contrasts this pattern with the tendency for successors for non-retirees to be more conservative in the cases where ideological change occurs, although the tendency is minimal. It should be stressed that what Frantzich observes, minimal as it is, could well be timebound phenomena and the result of long term trends leading up to the periods he studies and not necessarily a function of retirement itself. 31 Other authors, without empirical support, are willing to speculate as to the likely impact of retirement. COOper and West feel that if the trend toward increased retirement continues, it will result in new types of members being recruited into the House, members "more attuned to the benefit-cost ratio that now prevails."61 They feel this new type of member is more likely to be self-oriented and their presence will serve to intensify the current tendencies toward fragmentation, fractiousness, and exploitation of position. Albert Cover believes increased retirements will have an impact external to the House as well as an impact within the House.62 Externally, he believes increased retirements will reduce the number of well-entrenched incumbents who owe nothing to the President by virtue of the latter's coattails and could lead to a more faborable atmosphere in which the President can operate. Also enhancing the President's position will be the concomitant replacement with less secure junior members who might be more susceptible to Presidential persuasion. In terms of representation, retirement necessarily leads to more contests in which no incumbent is running, which to Cover could lead to a Congress more responsive to shifts in public Opinion. Internally, Cover contends that the trend toward increased retirements, with its corollary depletion Of senior ranks will serve to reinforce the undermining Of the seniority system. With this, he contends, will come 32 heightened internal conflict as automatic rules for committee leadership give way to Open contests. He also believes the induction of more junior members will further reduce the socialization of new members, expediting the collapse of established norms. This Opinion is, of course, shared with Cooper and West. Conclusion This chapter was begun with the assertion that the motivations of elected representatives, whether assumed or discovered, play an important role in shaping our interpretations of the activities of representative bodies, and our understanding of representation. Joseph Schlesinger's ambition theory provided a foundation upon which to interpret the behavior of elected representatives by pointing to the benefit of viewing this behavior in terms of where the elected representative wants to be rather than where he is at present. Subsequent research has shown that differential office goals and different factors influencing office goals have an important impact upon political processes. Also examined in this chapter is research concerning discrete ambition. Retirement, it was seen, suggests something about the degree of competitiveness as well as the utility derived from holding Office. Furthermore, the increasing frequency of retirement from the House is held by some authors to be a potentially important influence upon the operating characteristics of that body 33 as well as a potential factor influencing the body's responsiveness to the public and its susceptibility to Presidential influence. Clearly, understanding the career decisions of elected representatives can aid our understanding of representation and of representative bodies. These decisions, it must be assumed, are manifestations of underlying attitudes and processes that impact upon the representatives' behavior and ultimately upon the operating characteristics of th institutions in which they serve. In the chapters that follow, a theory concerning career decisions will be developed. The propositions will be evaluated in light of empirical evidence. Conclusions will be forwarded as to the factors influencing these decisions and by inference influencing the motivations of members, the nature of representation and some of the Operating characteristics of the House of Representatives. NOTES 1Anthony Downs. Inside Bureaucracy. (Boston: Little, Brown, and Co., 1967), p. 28. 2David R. Mayhew. Congress: The Electoral Connection. (New Haven: Yale, 1974). 3Morris P. Fiorina. Congress: Kgystone of the Washing- ton Establishment. (New Haven: Yale, 1977). 4Richard F. Fenno, Jr. Congressmen in Committees. (Boston: Little, Brown, and Co., 1973). 5Lawrence C. Dodd. "Congress and the Quest for Power” in Congress Reconsidered, vol. 1, eds. Lawrence C. Dodd and Bruce I. Oppenheimer. (New York: Praeger, 1977), pp. 269- 307. 6Ibid., p. 281. 7Warren E. Miller and Donald E. Stokes. "Constituency Influence in Congress," American Political Science Review 57 (March 1963). See also, Stokes and Miller, ”Paity Government and the Saliency of Congress," Public Opinion Quarterly (Winter 1962). 8Berelson, Lazersfeld, and McPhee. Voting. (Chicago: University Press, 1954). 9Joseph A. Schumpeter. Capitalism, Socialism, and Democragy, 3rd ed. (New York: Harper and Row, 1950), p. 269. 101bid., p. 285. 11Joseph A. Schlesinger. Ambition and Politics: Politi- cal Careers in the United States. (Chicago: Rand’McNally and Company, 1966), p. 2. 12 Fiorina, Congress: Keystone of the Washington Establishment, p. 5. 13Stephen Frantzick. "Opting Out: Retirement from the U.S. House of Representatives, 1966-1974," American Politics Quarterly 6 (July 1978). 14Schlesinger. Ambition and Politics: Political Careers in the United States, pp. 9-10. 15 Ibid., pp. 5-6. 16Ibid., pp. 16-20. 35 17Kenneth Prewitt. The Recruitment of Citizen Politicians. (Indianapolis: Bobbs-Merrill, 1970). 18Schlesinger, Ambition and Politics, pp. 172-174. 19John W. Soule. "Future Political Ambitions and the Behavior of Incumbent State Legislators," Midwest Journal of Political Science 13 (August 1969). 20 Ibid., p. 452. 21David W. Rohde. "Risk-Bearing and Progressive Ambition: The Case of Members of the U.S. House of Representatives," American Journal of Political Science 23 (February 1979). 22William H. Riker and Peter C. Ordeshook. An Intro- duction to Positive Political Theory. (Englewood Cliffs: Prentice-Hall, l973),pp. 75-77. 23Alfred Cobban. Edmund Burke and the Revolt Against the Eighteenth Century. (London: George AlIen and’Unwin, Ltd., 1962), p. 61. 24Schlesinger, Ambition and Politics, pp. 16-20. . 25Harvey Lehman. Age and Achievement. (Princeton: Princeton University Press, 1953). 26 David B. Walker, "The Age Factor in the 1958 Con- gressional Elections," Midwest Journal of Political Science (February 1960), pp. 1-26. 27Schlesinger, Ambition and Politics, p. 176. 28Jeff Fishel. Party_and Opposition: Congressional Challengers in American Politics. (New York: David McKay Company, Inc., 1973). 29 Paul Hain. "Age, Ambition and Political Careers: The Middle-Age Crisis," Western Political Quarterly 27 (June 1974), p. 274. 30 Kenneth Prewitt, The Recruitment of Citizen Politicians. 31Gordon S. Black. "A Theory of Political Ambition: Career Choices and the Role of Structural Incentives," American Political Science Review 66 (March 1972). 32Ibid., p. 159. 33Samuel Kernell. "Congressional Careerism and the Emergence of a Political Career Structure," paper presented at the meeting of the Social Science History Association, Cambridge, Massachusetts, 1979, p. 12. 36 34H. Douglas Price. "Careers and Committees in the American Congress: The Problem of Structural Change," in The History of Parliamentary Behavior, ed. William O. Aydelotte. (Princeton: Princeton University Press, 1977), p. 36. 351bid., p. 39. 36Nelson Polsby. "The Institutionalization Of the U.S. House of Representatives," American Political Science Review 62 (1968). 37Morris P. Fiorina, David W. Rohde, and Peter Wissel. "Historical Changes in House Turnover," in Congress in Change, ed. Norman J. Ornstein. (New York: Praeger, 1975). 38Samuel Kernell. "Toward Understanding Nineteenth Century Congressional Careers: Ambition, Competition and Rotation," American Journal of Political Science 21 (Novem- ber 1977). 391616., pp. 692-693. 40Charles Bullock. "House Careerists: Changing Patterns of Longevity and Attrition," American Political Science Review 66 (1972). 41Ibid., p. 1297. Bullock defines careerists as members victorious in ten or more elections. 42Joseph Cooper and William West. "Voluntary Retirement, Incumbency and the Modern House," Political Science Quarterly 96 (Summer 1981). 43Stephen E. Frantzich. "De-Recruitment: The Other Side of the Congressional Career Equation," paper presented at the annual meeting of the American Political Science Association, September 1977. 44Frantzich, "Opting Out". 45John R. Hibbing. Voluntary Retirement from the U.S. House: The Costs of Congressional Service," Legislative Studies Quarterly 7 (February 1982). 46Albert D. Cover. "The Greening of Congress: Patterns of Congressional Retirement," paper presented at the annual meeting of the Midwest Political Science Association, Cincinnati, Ohio, April 1981. 47 Ibid., p. 73. 48James L. Payne. "Career Intentions and Electoral Performance of Members of the U.S. House," ngislative Studies Quarterly 7 (February 1982), p. 97. 37 4gSchlesinger, Ambition and Politics, p. 6. SOKenneth Prewitt and William Nowlin. "Political Ambitions and the Behavior of Incumbent Politicians," Western Political Quarterly 22 (June 1969). 511616., p. 301. 52Soule, "Future Political Ambitions and the Behavior of Incumbent State Legislators," p. 452. 53Kenneth Prewitt and Heinz Eulau. "Political Matrix and Political Representation: Prolegomenon to a New Departure from an Old Problem,” American Political Science Review 63 (June 1969). 54Kenneth Prewitt. ”Political Ambitions, Volunteerism and Electoral Accountability,” American Political Science Review 64 (March 1970). 55161d., p. 17. 56Roger Davidson. The Role of the Congressman. (New York: Pegasus, 1969)p. 136. S7Bullock, "House Careerists." p. 1300. 58 Schlesinger, Ambition and Politics, p. 2. 59Frantzich, "Opting Out," p. 267. 601bid., p. 269. 61Cooper and West, "Voluntary Retirement, Incumbency and the Modern House," p. 299. 62Cover, "The Greening of Congress". Chapter II In this chapter, a theory of progressive ambition devel- oped by David Rohde will be illustrated and evaluated. Some of the hypotheses he derives from his theory will be examined, and other hypotheses that stem from his theory will be derived. After examination of this theory of progressive ambition, attention will be turned to retirement decisions. A decision calculus will be posited concerning retirement that is in many ways similar to the one Rohde posits for decisions concerning higher office. From this, hypotheses concerning retirement decisions will be developed. David Rohde has provided a deductive theory concerning progressive ambition: a priori reasoning about the relation- ship between variables influencing progressive career decisions of members of the United States House of Represen- tatives. His theory is grounded in the rational choice tradition. As such, it carries with it the traditional assumption that the subjects of his inquiry are rational, in that they are assumed capable Of ranking alternatives and of selecting the alternative that yields the highest expected utility. Beyond this, Rohde explicitly assumes that progres- sive ambition is held by all members of the House except those who begin service in the House with the intent of simply filling out the present term (i.e., the discretely ambitious). 39 Specifically, Rohde assumes that "if a member of the House, on his first day of service, were offered a Senate seat or 1 governorship without cost 93 risk, he would take it." This assumption carries with it a number of important implications. First, it suggests that upon entering the House all members are equal in terms of ambition. Second, by a55uming this to be true only upon entering the House, it implies that additional service in the House can alter this equality, indirectly paraphrasing Sam Rayburn's admonition to Lyndon Johnson that ”nobody who ever amounted to anything in the 2 Third, it shifts the House ever left here for the Senate.” 'orientation away from the individual to the opportunities he confronts. Specifically, it implies that progressive and static ambition is behavior manifested because of the risks inherent in the Opportunity structure in which the individual finds himself and his willingness to bear those risks. Rohde assumes a particular decision calculus for his subjects. It is a variant of the decision calculus first 3 developed by Riker and Ordeshook. The calculus has the following form: E(al) = P1(01)U(01) + P1(02)U(02) + P1(03)U(03) - C(31) E(a2) = P2(01)U(Ol) + P2(OZ)U(OZ) + P2(03)U(03) - C(az) where E(a1) is the expected utility of choosing alternative 1 Pi(0.) is the probability that outcome j will occur if 3 alternative 1 is chosen U(O.) is the utility the actor receives if outcome j occurs 40 C(a.) is the direct cost incurred by choosing alterna- tive i and where specifically: 01 = no office is occupied after the election 02 = the presently held Office is occupied after the election 03 = the higher office being considered is occupied after the election a1 = the actor runs for the presently held Office a2 = the actor runs for the higher office The outcomes are mutually exclusive (i.e., P1(03) = P2(OZ) = 0) which is to say that an actor cannot occupy the present office if he runs for higher office, nor occupy the higher one if he runs for reelection. Rohde further assumes that occupying no Office after the election is the least preferred outcome and arbitrarily sets the value of that out- come to zero (i.e., U1(Ol) = 0). This simplifies the original equation concerning decisions to seek higher office to: and that concerning reelection decisions to: E(a1) = Pl(OZ)U(OZ) - C(al) Assuming, as he does, that representatives are rational maximizers of expected utility, a representative will run for higher Office only if E(a2)>'E(a1). Rohde limits considera- tion to members who had an Opportunity to run for higher office; an Opportunity in this context is when no incumbent is running for higher office or, if an incumbent is running, 41 he is not of the representative's party. He also excludes from consideration opportunities for governorships in states with odd year gubernatorial elections, since an explicit component of the decision calculus is the sacrifice of the currently held House seat, which would not be necessary in these states. Rohde considers four basic categories of factors influen- cing decisions to seek higher Office. At this point, it will prove useful to evaluate each of these categories. Further- more, within these categories, Rohde's original hypotheses will be restated. To these will be added hypotheses that are believed to be logical extensions of the theory. I. Decisions to Seek Higher Office The Value of the Higher Office Utilizing the results of extensive surveys he and others had conducted with members Of the House, Rohde notes the greater attractiveness perceived in a Senate seat as compared to a governorship. One dimension of this was the Senate's smaller size and greater potential for influence. The other dimension was the relative security a six-year term promised. He also notes Schlesinger's original contention about the similarity of function between offices being an important manifest tie.4 This leads him to his first hypothesis. H1: Among House members, the proportion of Opportuni- t1es to run for the Senate that 15 taken w1ll be greater than the proportion of opportunities to run for governor. 42 Rohde's next area of concern is the difference among governorships. He expects a difference in attractiveness to exist on the basis of differential power but acknowledges the lack of data concerning the governor's power during the period he studies. In the absence of such data, he is left with a more obvious difference between governorships, this being length of term. He posits the following hypothesis: H2: Among House members, the proportion of Opportuni- t1e5 to run for governorsh1ps w1th a four-year term that is taken will be greater than the proportion of opportunities taken to run for gover- norships with a two-year term. The Probability of Winning the Higher Office In this category, attention turns to factors posited to influence the P2(03) term in Rohde's calculus. The first hypothesis in this category is reflective of the widely accepted advantage incumbents have in securing reelection.5 His third hypothesis is thus: H3: Among House members, for both Senate and guberna- torial races, the proportion of opportunities to run for higher offices that is taken in situations where no incumbent is seeking reelection will be greater than the prOportion of opportunities taken in situations where an incumbent is seeking reelection. This hypothesis, it is believed, can be extended to include differences in incumbency. Schlesinger suggests that there very well might be electoral differences between senators and governors. In terms of senators, he suggests that they may accumulate electoral support with increasing years of service.6 This implies that senators become less vulnerable over time. This, in fact, was asserted by Donald Matthews, who stated that "if the senator survives the first challenge 43 to his position, then he becomes more secure than before."7 These assertions suggest the following hypothesis: HSA: Among House members, for Senate races only, the prOportion of opportunities taken when an incum- bent is seeking reelection will be greater when the incumbent is running for reelection for the first time than the proportion of Opportunities taken when the incumbent is seeking a third or greater term. In terms of governorships, there is a different expecta- tion. While senators accumulate electoral support with in- creasing years of service, Schlesinger suggests that governors accumulate grievances which produce the rejection of incum- bents.8 This suggests that with increasing tenure a governor may become more vulnerable, unlike senators who become less so. This notion is forwarded by Stephen Turrett who contends that "[d]e1ay for a governor may be tantamount to defeat, while time is a senator's ally."9 Thus, the following hypothesis is: H33: Among House members, for gubernatorial races only, the proportion of opportunities taken when an incumbent is seeking reelection will be greater when the incumbent is running for reelection for the second or greater time than the proportion of opportunities taken when the incumbent is seeking reelection for the first time. The next hypothesis Rohde develops concerns the partisan bias of the electoral situation. The likelihood of a candidate's winning in a state that is "safe" for the other party is expected to be less than in competitive situations or when the state is "safe" for the representative's party. Because of this, Rohde posits the following: H4: Among House members, for both Senate and guberna- torial races, the proportion of Opportunities to 44 run for higher office that is taken in states which are "safe” for the Opposition party will be less than the proportion of opportunities taken in states which are competitive or "safe" for their own party. The basic logic of this hypothesis is obviously sound. However, an extension or amendation is in order. The competi- tiveness of a state not only tells us something about the likelihood of winning the eventual election for higher office but also suggests something about the nature of the competi- tion for a party's nomination for the higher office. V.O. Key, for instance, has shown that the incidence of primary contes- ting is related to the strength of parties within districts at the congressional level. For example, in districts in which the Republican party was strong and victory in the general election seemed certain, Key found that there were often two or more candidates bidding for the parties' nomination (unless an incumbent was running). On the other hand, in districts in which the Democratic party's prospects in the general election were slight, there was ordinarily much less interest in capturing the Democratic nomination, and hence there are fewer contests for the party's nomination. Finally, he found that in highly competitive districts there were likely to be contests in the primaries of both parties.10 Thus, while degree of safeness suggests something about the ultimate likelihood of winning the higher office, it also suggests something about the degree of difficulty in securing a party's nomination to run for that higher office. In the case of the safe seat, we may expect a higher proportion of opportunities to be taken because of the near certainty of 45 obtaining the office if the nomination battle is won. On the other hand, in states that are relatively safe for the other party we may expect a fairly high proportion of Oppor- tunities to be taken because, although the eventual likeli- hood of winning may be rather low, the likelihood of obtaining the party's nomination is high. It is expected here that the lowest proportion of opportunities will be taken in the competitive situation because of the relatively high degree of uncertainty surrounding both the nomination for higher Office and the eventual electoral outcome. Hence: H4A: Among House members, for both Senate and guber- nator1al races, the proport1on of Opportun1t1es taken to run for higher office in states whose partisan bias is either favorable or unfavorable to the representative's party will be greater than the proportion of opportunities that are taken in compet1t1ve states. ' The next factor Rohde considers as influencing the probability of Obtaining higher office stems from one of Schlesinger's contentions about manifest ties that link Offices together. In this case, it is the degree of shared electorate that is of concern.11 Stephen Frantzich had previously argued that congressmen from small states, where congressional dis- tricts include either all or a substantial portion of the state-wide electorate, would be more likely to exhibit progressive ambitions than would congressmen from large states. 12 Rohde acknowledges the His data confirm this notion. possibility that for governorships there may be an inverse relationship between the power and assumed attractiveness of a governorship and the likelihood of winning that governorship. An examination of Schlesinger's index of the powers of 46 governors13 in 1969 suggests that in 1969, at least, large states had governorships with relatively more power while smaller states had relatively weaker ones. Because Of this, Rohde restricts his hypothesis concerning the degree of shared electorate to the Senate. Thus, he hypothesizes: H5: Among House members, for Senate races, the proba- b1l1ty that a House member w1ll run w1ll be directly related to the proportion of the state's population that the population of his House constituency comprises. While it is agreed here that the attractiveness of a governorship and the probability of winning, in all likelihood, does interact, leading to an attenuation of the relationship between the degree of shared electorate and decisions to seek governorships, it is also felt that power is only one dimension of the value of a higher office. A four-year governorship would still relieve the more frequent electoral effort required by the two year term characteristic of the House. Furthermore, a governorship might make an eventual transition to the Senate more likely than one from the House to the Senate. As Schlesinger has noted, "the speculation about a sitting governor is almost universal: will or will he not 14 make a try for the Senate." Schlesinger, as well as Donald Matthews, provides data that suggests that governors are 15 The reduction indeed a major source Of Senate candidates. in electoral frequency promised by a four-year governorship, as well as the potential utility a governorship may hold for an ultimate transition to the Senate, suggests that, although not as strong, the relationship between shared electorate and decisions to seek governorships should be in the same direction 47 as that with decisions to seek the Senate. Hence: HSA: Among House members, for gubernatorial races, the probability that a member will run will be direct- ly related to the proportion that the pOpulation of his House constituency comprises of the state's pOpulation. The Value of the House Seat In this category, factors are considered by Rohde that influence the U(OZ) term. Specifically, these are factors that would influence the amount that would be sacrificed if a higher office were pursued. The discussion will begin at this point but will consider factors considered to influence P1(Oz), the probability of retaining the House, and the way that changes here could alter the magnitude of the expected utility of seeking reelection. Rohde notes that power in the House rests largely in committees and that the way to power in committees is through the seniority system.16 Although not a direct measure of power, seniority provides a useful surrogate if it is assumed that the more senior a member is, the more powerful he becomes. If it is further assumed that members derive utility from power, seniority and the value of the House seat should be positively correlated. This leads Rohde to the following hypothesis: H6: For both Senate and gubernatorial races, the probab1l1ty that a House member w111 run w111 be inversely related to his seniority. The inverse relationship posited to exist between senior- ity and prOportion of opportunities taken to seek higher office suggests that the first-term member is more likely to seek higher office than is the second-term member, and the second 48 more likely than the third, and so on. This proposition seems disagreeable for a number of reasons. The first-term representative has had little time to acclimate himself with either Washington or his district. Granted that after one term the value of his House seat is no doubt less than that of seats held by his senior counterparts, the first-term repre- sentative also has not had enough time to engage in activities that would gain for him the name recognition and prominence useful in seeking higher office. Thus, while in agreement with the notion that the value Of the House seat increases with increased seniority, it is also believed that seniority interacts with the probability of obtaining higher office. Thus: H6A: For both Senate and gubernatorial races the . probab111ty that a House member w111 run w111 increase with increased seniority over the early portion of his House career but will then diminish with increasing seniority. If we may assume that seniority leads to increased power in the House, we need not look far for another factor that. enhances power and thus the attractiveness of a House seat. Party is a factor Rohde acknowledges but does not test. The minority party in the House has less power than the majority party. Since 1950 the Republicans have been in the minority in the House in every Congress with the exception of the Eighty-Third (1953-55). This leads to the following hypothe- sis: H6B: For both Senate and gubernatorial races, the . proportion of opportun1t1es taken to run for h1gher office will be greater for Republicans (the minority party) than the proportion of opportuni- ties taken by Democrats (the majority party). 49 Another factor posited to influence the expected utility of returning to the House is the probability a member attaches to that outcome. Something that could be expected to influence this subjective probability is the member's margin of victory in his last House election. Although a crude indicator, since it ignores intervening factors since the last election, it still seems plausible that a member with a narrow margin of victory would estimate a lower expected utility on returning to the House than would a member with a wide margin of victory, other things being equal. Of course, a diminishing P1(02), the estimate of the probability of returning to the House, could very well lead to the representative's lowering his estimate of P2(03), the probability of obtaining the higher office. Nonetheless, the following hypothesis would seem worthy of evaluation: H For both Senate and gubernatorial races, the proportion of Opportunities taken to run for higher Office will be greater for members who had narrow margins of victory in their last election than the proportion of Opportunities taken by members who had wide margins of victory in their last election. 6C: As noted, margin of victory suffers from being retro- spective. What would be more useful is some prospective indicator of the probability of returning to the House. This would be quite difficult to obtain for all members. We do have something that could serve as an indicator for some members, however, and this is how they fared in redistricting. Albert Cover quotes one member as saying the only thing that would have hurt them more than redistricting would have been to have a certain part of their vital anatomy laid on a 50 17 This candid and colorful stump and hit with a mallet. account suggests that members who are harmed by redistricting may indeed attach a lower probability to returning to the House. Another aspect of redistricting is a party's effort to help a member obtain a nomination for higher office when they are forced to alter or remove that member's district. An insightful case is that of former Michigan Representative James J. Blanchard. The Michigan legislature, in the control of the Democratic party, was forced, due to the declining relative population of the state, to eliminate a Congressional district. A Congressman from the Detroit area was a logical choice because of the area's severe decline in pOpulation during the 19705. Unfortunately for the Democrats, all of the Congressmen in the Detroit area were Democrats, forcing them to commit "cannibalism." James Blanchard was encouraged to run for governor with pledges of support and resources. He eventually decided to seek the governorship, making the job of state representative Mike Griffin, chairman of the legis- lative redistricting committee "easier."18 The moral is that redistricting and the underlying political processes it generally entails may have the effect of lowering the proba- bility a member attaches to returning to the House, raising the probability he attaches to obtaining higher office, or both. This leads to the following hypothesis: For both Senate and gubernatorial races, the proportion of opportunities taken to run for higher office will be greater among members who have been hurt substantially by redistricting than among members who were hurt only slightly, not affected, or helped by redistricting. H6D: 51 Risk Acceptance and Progressive Ambition The final concern of Rohde, and perhaps his most innova- tive, is risk-taking. Before considering his utilization of 19 the concept, a brief review of the topic is in order. The notion of risk in economic literature is actually quite Old.20 Contemporary treatment of risk has its origins in the work of Milton Friedman and L.J. Savage. Their concern was with the application Of expected utility theory to risk-taking. They sought to explain why low-income consumer units would buy insurance (thus taking a certain small loss to avoid a very small probability of a large loss) and also purchase lottery tickets (gambling on a small chance of a large gain against a 21 'Their concern was whether great chance of a small 1055). there was any consistency among choices in different risk situations. Friedman and Savage found that these differences in choice could be explained if those selecting insurance were posited to have utility functions that were convex from above, while those selecting the lottery had utility functions that were concave from above. Riker and Ordeshook extend the concept of risk-bearing to their consideration of political participation. They assert that "some peOple are more likely to select risky alternatives than others."22 They contend that differential risk-taking propensities would allow a risk taker to have a rational motive to participate and a risk averter to have a rational motive to reject participation. Rohde applies the concept of risk bearing to decisions to seek higher office by noting that costs and risks inherent in such a decision do not tell the 52 whole story. It is possible, with two individuals with identi- cal preference orderings concerning higher office, for one of them to seek higher office and the other not because, as Rohde states, "[pJeOple with the same preference orderings will "23 Rohde further differ in the intensity of those preferences. contends that it is "differences in intensity of preference, and thus willingness to take electoral risks in seeking office, that distinguishes the ambitious politician from the nonambi- tious."24 This belief leads Rohde to posit the following: H7: If two House members are presented with similar opportun1t1es to seek h1gher off1ce, and one 15 a "risk taker" and the other is not, then the "risk taker" will have a greater probability of running for higher office than the other. Ag; A central concern of Schlesinger, but one not addressed by Rohde, is the role age plays in shaping ambition. Schlesinger contends that the age cycle restricts a person's political chances; reasonable expectations in one period of life are unreasonable at another. The individual politician, according to Schlesinger ”. . . constantly faces the problem of reconciling two everchanging schedules, the timetable of "25 Office and the timetable of age. The research of Paul Hain serves to substantiate this notion. Hain states that ". . . of the many variables affecting a politician's 26 ambitions, age is central." In his study of state legisla- tors, Hain found support for the hypothesis ". . . that among occupants of a given office the older a person is the less 27 likely he is to advance." He went on to conclude that 53 ”. . . age greatly affects the political ambitions and careers of American politicians."28 Rohde suggests that it is the coSts and risks associated with seeking higher office, and the risk-taking propensities of the individual, that distinguishes the ambitious politician from the non-ambitious. Schlesinger and Hain provide compel- ling evidence that suggests age plays an important role in distinguishing the ambitious from the non-ambitious politi- cian. How can these differing perspectives be reconciled? On the one hand, one could reasonably assume that relative youth or relatively advanced age could substantially reduce the probability a member attaches to obtaining higher office, thereby reducing the expected utility of that alternative. On the other hand, age may be related to risk-taking propensi- ties. For example, J.S. Slotkin describes early middle age as ". . . the testing stage . . .[a]. . .period in which the individual examines his career to determine the extent to which he has obtained the gratification he has hOped to gain from his life course.29 Such introspection may lead members of early middle age to be less satisfied with their current career status quo and lead them to have more intense prefer- ences for the risky alternative of higher office. Whether we posit members' ages to influence the probabil- ities they attach to obtaining higher office, or to influence their risk-taking propensities over the course of their lives, or some combination of both, the expected relationship between age and decisions to seek higher office would be the same in each case: the relatively young and the relatively old will 54 be less likely to seek higher office than members of middle years, other things being equal. This reasoning leads to the following hypothesis: H8: Among House members, for both Senate and guberna- torial races, the proportion of opportunities taken to run for higher office will increase with age up to some point in middle age after which the propor- tion of opportunities taken will decrease with decreasing age. II. Decisions to Retire Up to this point we have relied heavily upon Rohde's theory of progressive ambition. We now move into an area for which no formal theory has yet been formulated; this concerns decisions to leave the House for reasons other than seeking higher office. Although no explicit formal theory of retirement has yet been developed, Joseph Cooper and William West, in their con- sideration of the trend towards increasing retirement in the 19705, suggest that the benefit-cost ratio of service in the House has declined.30 John R. Hibbing also examines retire- ments from a cost-benefit perspective.31 Previously, Fiorina, Rohde, and Wissel had analyzed voluntary retirement in benefit- cost terms.32 Their notion of a benefit-cost ratio of service Iimplies that some type of rational choice calculus underlies these decisions, but they do not go so far as to formalize the characteristics of the calculus. It should prove useful, at this point, to attempt to formalize a decision calculus concerning retirement decisions to see what this decision calculus leads us to expect about retirement decisions. 55 We may begin by assuming again that our actors are rational maximizers of expected utility. Next, we may posit a decision calculus that is assumed to underlie retirement decisions. This calculus is similar to that employed by Rohde concerning decisions to seek higher office but has some notable differen- ces. The terms in the calculus are as follows: 00 = no office occupied after the election: actor retires 01 = no office occupied after the election: actor defeated 02 = the currently held office is occupied after the election a0 = the actor retires a1 = the actor runs for the currently held office (i.e., reelection) P = the probability of defeat 1 - P1 = the probability of reelection Let us assume, as Rohde does, that occupying no office after the election is the least preferred outcome, and arbi- trarily set U(Oo) = U(Ol) = 0. Let us further assume that retirement is an act chosen with certainty and involves no costs pg; s3. Hence: E(a = U(OO) = 0 0) Let us begin by considering two extremes concerning reelection and see how they might relate to retirement. Our rationality assumption implies that a rational actor will retire only if E(al) < E(a = U(OO) = 0 0) The first extreme we may evaluate is when the probability 56 of defeat approaches being a certainty (i.e., P = l, 1 - P l 1 = 0). Under this condition 1) = 'C(al) Assuming there is some direct utility cost incurred by E(al).= 1(01)U(01) + 0(O2)U(OZ) - C(a seeking reelection, the actor under this extreme condition would clearly retire since E(a = -C(a1) < E(a = U(OO) = 0 1) 0) Research concerning the incumbency advantage enjoyed by members of the House would suggest that this extreme is not realistic for many, if not most, retirements from the House. This extreme does, however, point to the theoretical impor- tance to the understanding of retirement decisions of the probability of defeat in relation to the utility costs incurred in seeking reelection. Let us now eValuate another extreme and see what it suggests about the conditions under which we might expect retirement. Consider the case where the probability of re- election is a certainty (i.e., P1 = 0, l - P1 = 1). In this case E(a = 0(01)U(Ol) + 1(OZ)U(OZ) - C(a = U(OZ) - C(a 1) 1) 1) At this extreme we would expect a rational actor to retire only if C(al) > U(02) Only when the cost incurred in seeking reelection is greater than the utility derived from reelection, at the extreme when the probability of obtaining reelection is a cer- tainty would our decision rule, E(al) < 0, hold. This latter extreme suggests that for some, even though they may return 57 to the House with certainty, either the Costs they incur in retaining their House seat or the utility they derive from holding that seat is not sufficient to lead them to seek re- election. The extremes outlined above taken independently embrace only a narrow range of retirement decisions. In combination, however, it is believed they can provide a more general under- standing of retirement decisions. In other words, retirement can be viewed as a function of some combination of relatively low probability of reelection, relatively low utility derived from House service, and/or relatively high costs incurred in seeking reelection. For the sake of simplicity, it is assumed here that there is an inverse relationship between the proba- bility of reelection and the costs incurred in seeking re- election. By so doing we can narrow our discussion to factors hypothesized to influence the utility of House service, and those hypothesized to influence the probability (and con- versely costs) associated with returning to the House. Each will be considered separately below. The Utility of House Service In this category, we may examine many of the same factors that were considered to impact upon the value of the House seat and hence the degree of sacrifice in seeking higher office. Many of the expectations will be the same. As noted previously, power in the House rests largely in committees, and the way to power in committees is through seniority. If we again assume that members derive utility 58 through power and that power is a function of seniority, we are led to the following hypothesis: H9: Among members of the House, the probability of retirement will be inversely related to seniority. As is apparent, there is a problem with this hypothesis. This problem stems from the fact that members with very high seniority no doubt come from safe districts. This is a problem in that, for a member from a very safe district, defeat is unlikely; and retirement becomes the only way short of death that he can leave the House. This being the case, it is quite possible that members with relatively high seniority will be as likely to retire as those with relatively low seniority. There is a way around this problem. The age at which a member enters Congress ultimately determines the upper limit of seniority which that member could accumulate. The member who enters the chamber at 65 certainly has different expectations from the member who enters at 35 concerning seniority. Thus, the age at which a member enters Congress shapes both the potential and the ultimate value of a House seat, assuming this value to be a function of seniority. This leads to the following hypothesis: H10: For members of the House, the probability of ret1rement w111 be pos1t1vely related to the age at which the member entered Congress for the first time contiguous to the session in which retirement is considered. We may also expect age to have another influence upon retirement. Stephen Frantzich notes that a number of the retiring members he examined revealed an acute awareness of age and disability. Their comments are informative: 59 It seemed like a good time to quit. This feeling was motivated in part by observing some of my elderly colleagues . . . forcing themselves to continue their congressional duties despite physical incapacities. I determined not to let myself get into that category. 33 There are just too many people who stay around too long. I didn't want peOple looking at me from the gallery and whispering, "He used to be Congressman ." 34 As with any career, life expectancy places an upper limit upon its ultimate length. For members of Congress, increasing age does not necessitate retirement but the statements above suggest that for at least some members it is an important factor. Thus, we have the following hypothesis: H11: For members of the House, the probability of retirement will be positively related to their age. As in the case with decisions to seek higher office, we would expect a member's political party to have an influence upon the utility he derived from House service.' The minority party has less power in the House than the majority party. We would expect members from the minority party to derive less utility from House service than do the majority members, other things being equal. Hence: H12: For members of the House the probability of retirement will be greater for Republicans (the minority party) than for Democrats (the majority party). It is believed here that risk-taking propensities should bear a relationship to retirement decisions. When evaluating the alternatives, seeking reelection or retiring, the member confronts a risky alternative and a certain alternative. As mentioned previously, Riker and Ordeshook note, "some people are more likely to select risky alternatives than are others."35 60 Each attempt at reelection involves risk; and as in the case of decisions to seek higher office, where two members may have identical preference orderings but may differ in the intensity of their preferences, it is believed that in retirement decisions the inherent risks in seeking reelection tell only part of the story. Some members will be more likely than others to bear the risks of seeking reelection. This suggests the following hypothesis: H13: Among members of the House, the probability of ret1rement w111 be less for the h1gh r1sk taker than for the less risk acceptant. Perhaps one of the most explicit benefits derived from service in the House is salary. It would be unrealistic to assume that members of the House serve in that body primarily for the remuneration represented by their salary; an examination of member biographies suggests that many, if not most, could earn more elsewhere. Nonetheless, it would seem that salary, for members without an outside source of income, would represent their primary source of income; and its decline in real dollar terms might make it relatively difficult for them to maintain lifestyles their qualifica- tions warrant. Furthermore, while congressional salaries may decline in real dollar terms, this has not been the case for many alternative occupations for which congressmen are frequently qualified. The following comments by retirees serve to illustrate this point. Private indugtry salaries are so much higher it's ridiculous. 6 61 The relationship of the congressional salary to private industry salaries has done a reversal. When I entered Congress, the salary was $10,000. At that time there weren't but two or three lawyers in all of my hometown making $10,000. Now the salary is $60,000, but there must be two hundred lawyers in this town making that, maybe more. Seeking my tenth term in Congress next year would too severely limit my opportunities in the private sector for my remaining active years.38 In an indirect manner, it appears that the congressional salary represents an opportunity cost. As it climbs slowly, or even declines, in real dollar terms, outside opportuni- ties have increased rather rapidly; that which is foregone increases in magnitude and the cost of House service in- creases. It would be exceedingly difficult to estimate each member's potential earning power outside the House and develop an indicator of opportunity cost to predict retirement. In the absence of such an indicator, congres- sional salary in real dollars would serve as a crude but useful surrogate for assessing the relative monetary rewards of service in the House. This leads to the following hypothesis: H14: Among members of the House, there is an inverse relat1onsh1p between the proport1on of members retiring and congressional salary measured in constant (1967) dollars. The final area of concern with regard to the utility of House service is related to the proposition that the nature of House service has changed in the 19705. As mentioned previously, Cooper and West have asserted that a. number of mutually reinforcing trends have led to increased dissatisfaction during the 19705. These and other authors 62 point to changes in or around 1970 that have diminished the utility of House service or made retirement more attractive. Among these factors are improved pensions for retired members, House provisions limiting outside incomes and requiring financial disclosure, declining reliance upon seniority for the distribution of rewards for House service, increasing difficulties in getting legislation passed and the increasingly demanding nature of the job of congressmen, as well as other influences. We would expect, and in fact find, that more members are retiring after 1970. The following hypothesis is forwarded, not to restate the obvious, but as a prOposition to test while other factors are controlled in the multivariate analysis to be discussed later. Hence: H Among members of the House, the probability of retirement will be greater after 1969 than it was prior to 1970. 15‘ The Probability of Reelection In this section we will consider factors posited to influence the P(Oz) term in our retirement calculus. The most obvious influence would be some indicator of pro- spective vulnerability. This, of course, would be virtually impossible to deveIOp. In the absence of a prospective indicator of vulnerability, we are left with a retrospec- tive indicator, margin of victory in the last election. This indicator is problematic for a number of reasons. First, it ignores intervening influences that could drastically alter a member's estimate of the probability of being re- elected. Second, much as in the case of members with high 63 seniority, for members from safe districts retirement becomes the only way short of dying that they can leave the House. With this being the case, we might find members from safe districts over time to be more likely to retire than members from more marginal districts. However, since the turnover of members from safe districts is low, we might also expect that proportionately more retirements come from marginal districts. Thus, although our expecta- tions are confounded by the above factors, the following hypothesis nonetheless seems worthy of evaluation: H16: Among members of the House, the probability of ret1rement W111 be an inverse funct1on of the margin of victory in their last election. As noted, margin of victory in the last election ig- nores the prospective aspect of the probability of reelec- tion. For some members, however, we do have a means by which to evaluate the impact of electoral prospects upon retirement decisions. As noted with decisions to seek higher office, we would expect members who were harmed by redistricting to attach a diminished probability of returning to the House. This leads to the following and final hypothesis: H Among members of the House, the probability of retirement will be greater for those hurt sub- stantially by redistricting than for those who were hurt only slightly, not affected, or helped by redistricting. 17‘ Plan of Analysis In the preceding pages I have restated the hypotheses David Rohde derived from his theory of progressive ambition 64 and provided extensions or amendations that flow from his theory. I have also developed a formal calculus underlying the cost-benefit ratio of service in the House and have derived hypotheses concerning retirement decisions. Chapter Three will begin by considering the frequency of attempts for higher office by members of the House. From there, the hypotheses concerning progressive ambition developed in this Chapter will be evaluated in light of tabular results. Then, having examined the tabular results, these hypotheses will be evaluated in light of the separate variable influences when incorporated into multivariate probit models in Chapter Four. This will allow for the evaluation of the relative independent influence of these factors as well as to discount the possibility of spurious- ness. In Chapter Five the pattern of analysis will be much the same as that in Chapter Three except that attention will be turned to decisions concerning retirement. In Chapter Six retirement decisions will be further scrutin- ized using probit. Within each of these chapters, the relevant concepts will be operationalized and the method of analysis will be more fully explored. The data by which these hypotheses are to be evaluated consist of all members of the House elected between 1950 and 1974. The origins of the data concerning these members will also be detailed in these chapters. In Chapter Seven, conclusions will be drawn in light of the results obtained in Chapters Three through Six. 65 It is hoped that through the use of a multivariate technique, the relative influences of each of the theoretically derived variables can be disentangled, providing a better picture of the factors influencing the career decisions of members of the House. From this picture a better understanding of the motivations of representatives can be obtained that ultimately may contribute to a better understanding of the operating characteristics of representative bodies and of the nature of representation itself. 66 NOTES 1Rohde, "Risk Bearing and Progressive Ambition," p. 3. 2Houston Chronicle, May 31, 1981. 3William H. Riker and Peter C. Ordeshook. An Intro- duction to Positive Political Theoyy. (EnglewoodICliffs: Prentice HalI] 1973). 4Schlesinger, Ambition and Politics, pp. 99-100. 5See, for example, Barbara Hinckley, "Incumbency and the Presidential Vote in Senate Elections: Defining Parameters of Subpresidential Voting," American Political Science Review 64 (1970) and Warren Lee Kostroski, “Party and Incum- Bency in Postwar Senate Elections: Trends, Patterns and Models," American Political Science Review 67 (December 1973). 6 Schlesinger, Ambition and Politics, pp. 68-69. 7Donald R. Matthews. U.S. Senators and Their World. (New York: W.W. Morton and Company, Inc., 1973), pl 242. 8Schlesinger, Ambition and Politics, p. 69. 9Stephen J. Turrett. "The Vulnerability of American Governors: 1900-1969," Midwest Journal of Political Science 15 (February 1971), p. 109. 10V.O. Key, Jr. American State Politics: An Intro- duction. (New York: Alfred A. KnOpf, Inc., 1956). See also W1111am H. Slanding and James A. Robinson, "Inter-Party Competition and Primary Contesting: The Case of Indiana," American Political Science Review 52 (December 1958). 11 Schlesinger, Ambition and Politics, p. 99. 12Frantzich, "De-Recruitment: the Other Side of the Congressional Career Equation". 13Joseph Schlesinger. "A Comparison of the Relative Positions of Governors," in The American Governor in a Behavigral Perspegtiye, eds. Thad Beyle and J. Oliver Williams. (New York: Harper and Row, 1972). 14 Schlesinger, Ambition and Politics, p. 103. 15Schlesinger, Ambition and Politics, found that 22 percent of all individhals elected to the Senate between 1914 and 1958 were formerly governors. Donald Matthews, U.S. Senators and Their World, found that 22 percent of senators between 1947 and 1957 came to that office via a governorship. 67 16Rohde, "Risk Bearing and Progressive Ambition," p. 11. 17Cover, "The Greening of Congress, " p. 4. 18Detroit Free Press, September 8, 1981. 19Decision making under risk, as Opposed to uncertainty, occurs under the assumption that an individual has an a riori probability distribution over the states of nature, while under certainty the states of nature are assumed finite and have no a priori probability assignments over the states. See, R. Duncan Luce and Howard Raiffa, Games and Decisions: Intro- ductions and Critical Survey. (New York: John Wiley and Sons, Inc., 1957), p. 325. 20See, for example, F.H. Knight, Risk, Uncertainty and Profit. (New York, 1921). 21Milton Friedman and L.J. Savage. ”The Utility Analysis of Choices Involving Risk," The Journal of Political Economy 56 (August 1948), p. 294. ' 2Riker and Ordeshook, An Introduction to Positive Political Theory, p. 75. 23 Rohde, "Risk Bearing and Progressive Ambition," p. 11. 24Ibid., p. 12. 25Schlesinger, Ambition and Politics, p. 174. Z6Hain, "Age, Ambition and Political Careers," p. 269. 27Ibid., p. 271. 28Ibid., p. 274 29J. A. Slotkin, "Life Course in Middle Age," in Problems of the Middle Aged. (Springfield: Thomas Publishers, 1965), p. 44. SOCOOper and West, "Voluntary Retirement, Incumbency and the Modern House," p. 288. 31Hibbing, "Voluntary Retirement from the U.S. House," p. 65. 32Fiorina, Rohde, and Wissel, "Historical Changes in House Turnover," p. 39. 33Quoted in Frantzich, "Opting Out," p. 259. 34Ibid., p. 259. 68 35Riker and Ordeshook, An Introduction to Positive Political Theory, p. 75. 36Hibbing,"Voluntary Retirement from the U.S. House," p. 65. 3 7Ibid., p. 65. 381bid., p. 65. Chapter III In this chapter, hypotheses concerning decisions to seek higher office, deveIOped in the first part of Chapter Two will be examined. In the process, questions concerning the operationalization of concepts embodied in these hypotheses will be addressed. Before turning attention to these hypotheses, however, it will first prove useful to examine the frequency of these decisions and their outcome in the period under study (members elected between 1950 and 1974). Table 1 displays the number of congressmen deciding to seek higher office by the office they sought, outcome, and year. Perhaps the most striking initial impression is the relatively high failure rate among high office seekers. Overall, 95 of 157 (60.5 percent) lost in their bid for higher office. If we ignore primary losers and concentrate only upon those who made it to the general election, the success rate improves appreciably, with 62 of 125 (49.6 percent) winning. In general elections in the period studied, higher office seekers had almost an exactly even chance of success. Overall, it appears that for many members these decisions were made with much less than certainty about the 70 ANN NN N6 N6 NN NN eN NN om Ne Neeee NN eN N e N o N m N N ee-NNeN eN e e e N N N N N N eN-NNeN NN N N N e e N N N N Ne-NNeN NN e NN N e N N N N e ee-meeN NN e N N N N N N N e Ne-eeeN N N e N N N N e N N ee-NeeN e N m N e N N N e N ee-NemN eN e e e N e N N e N Ne-NemN e N N N o N e N N N ee-mNeN m o N e e N o e N e NN-NNNN N e N e e N e o N e eN-NmeN NN o N e e o N e N N em-NNeN NN e e e o N N e N N Nm-NmmN lmfimpo>o .ENNm .cou .ENNA .coo .ENNQ .:oo ”Nmo> umOA :oz umoa :oz umoq so: Noo me >m mummmo mmzom: ozaxmmm mammzmz mmDO: H mqm Second Term Total .76 1.20 .90 (395) (245) (640) 4.52 3.40 4.04 (642) (445) (1087) 3.1 2.60 2.89 (1937) (690) (1727) 84 the period studied, there was a tendency for House members to be more likely to challenge governors seeking beyond their second term than to challenge those seeking reelection for the first time. The differences noted suggest that there are distinctions between governorships and Senate seats that impact not only upon the value House members attach to these offices, but also upon the probability members attach to attaining these offices over time when incumbents are seeking reelection. Our next area of concern centers on the partisan bias Of a state in which a member is contemplating pursuit of higher office. As an indicator, average statewide vote margin for Senate and gubernatorial races over the four years preceeding the member's opportunity for higher office were computed.5 It should be acknowledged that this is a rather crude indicator in that it does not take into account trends or national events, such as presidential elections, that could very well impact upon a member's estimate of the electoral climate of his state. Although crude, this indicator should still allow us to evaluate the more blunt manifestations of partisan bias upon decisions to seek higher office. Hypothesis 4 provides our first expectations: members should be less inclined to run for higher Office in states substantially favorable to the 85 the Opposition party. Table 7 provides data relating to this hypothesis. For both governorships and Senate seats, members are less inclined to run when the partisan bias favors the opposition party. However, this Table does not take into account another major factor influencing the probability of attaining higher office, the incumbency status of the opposition. Table 8 controls for both partisan bias and incumbency status. As can be seen, a different pattern emerges. When no incumbent is running in a state favoring the other party, members are most likely to seek higher office. The notion embodied in Hypothesis 4 holds only when an incumbent is seeking reelection. When no incumbent is running, members are more likely to seek higher office in situations where the partisan bias is unfavorable. How can this be? As mentioned in Chapter Two, the partisan bias of a state tells only part of the story concerning the probability of attaining higher office. A necessary, but not sufficient, condition for attaining higher Office is securing one's party nomination to run for the higher Office. Recalling Table 1, members were by and large successful in making it to the general election: 82 percent of those seeking Senate seats and 71 percent of those seeking governorships were successful in attaining their party's nomination. Members seeking higher office are quite successful 86 TABLE 7 PERCENTAGE OF MEMBERS RUNNING FOR HIGHER OFFICE, CONTROLLING COMPETITIVENESS OF STATE AND OFFICE Average Statewide Vote: >10% 110% Office: Unfavorable Unfavorable Total Governorship .97 1.50 1.45 (206) (1793) (1999) Senate 4.80 5.40 5.33 (269) (1740) (2009) Total 3.10 3.40 3.39 (475) (3533) (4008) 87 TABLE 8 PERCENTAGECW'MEMBERS RUNNING FOR HIGHER OFFICE, CONTROLLING FOR COMPETITIVENESS OF STATE AND INCUMBENCY Average Statewide Vote: >10% 510% Opposition is: Unfavorable Unfavorable Total Not Incumbent 4.15 3.70 4.21 (303) (1978) (2281) Incumbent 1.16 3.10 2.89 (172) (1555) (1727 Total 3.10 3.40 3.39 (475) (3533) (4008) 88 in securing nomination. This success may be related to the nature of primary contesting, first articulated by Key.6 They may, in fact, be so successful in’ securing nomination, at least in part, because in cases where no incumbent is running, they are pur- suing the nomination in a situation with the least competitive primaries: states that favor the other party. As noted in Table 7, the second highest proportion of members seeking higher office is where no incumbent is running and the partisan bias is less unfavorable. In this circumstance, we might suspect that the primary contest is more heated, causing proportionately fewer members to run, but still a relatively high proportion due to the high value of the nomination. The fewest members run where there is an incumbent seeking reelection in a state whose partisan bias is unfavorable to the member. Here the nomination might be relatively easy to obtain, but the incumbent's advantage perceived by the member leaves the nomination to be of little value. Overall, it appears that there is a tradeoff between the attractiveness of a nomination and the ability to secure that nomination. Hypothesis 4A suggests something about the nature of this tradeoff. The expectation is that a member will be most likely to seek higher office in situations where the combined expectations about securing the nomination and 89 winning the general election are high. Table 9 displays data that relate to this hypothesis. When no incumbent is running, or when an incumbent is running, members are least likely to run in the competitive situation. In such situa— tions, primary contesting could be very high and the general election outcome uncertain. Members are somewhat more likely to run when the partisan bias is favorable, a situa- tion where the nomination is highly contested but the eventual electoral outcome would yield favorable expectations. Members are most likely to run when the partisan bias is unfavorable; although when an incumbent is running, there is only a very slight difference between this category and that when the partisan bias is favorable. This similarity in magnitude is suggestive because in both we might antici- pate the combination of expectations concerning primary and general election outcomes to be similar: high primary expectations and low general election expectations in the one case, low primary expectations and high general election expectations in the other. When no incumbent is running, substantially more Opportunities are taken for higher office in the unfavorable situation than in either the favorable or competitive situations. Under this condition we might suspect that members derive the highest combined expectation of success in attaining higher office; the nomination may be easier to secure and the opposition would have only the favorable partisan bias of the state, not incumbency, as an advantage. 90 TABLE 9 PERCENTAGE OF MEMBERS RUNNING FOR HIGHER OFFICE, CONTROLLING FOR COMPETITIVENESS OF STATE AND INCUMBENCY >5% 5% Unfav. to >5% Opposition is: Unfav. 56 Fav. Fav. Total Not Incumbent 4.76 3.07 3.68 4.21 (608) (750) (923) (2281) Incumbent 3.18 2.61 3.11 2.89 (377) (805) (545) (1727) Total 4.16 2.83 3.47 3.39 (985) (1555) (1468) (4008) 91 The notions forwarded above are tentative. We would do well to evaluate these decisions in light of partisan bias controlling for other factors. For example, partisan bias would certainly suggest something about the degree of risk inherent in a potential bid for higher office. In Table 10, risk taking is controlled for. Interestingly, the more risk acceptant are most likely to run in the favorable situation while the less risk acceptant are most likely to run in the unfavorable situation. If we continue to assume that competitiveness impacts upon primary contesting, it appears that the high risk takers are more willing to risk their House seats for an uncertain nomination, while the less risk acceptant will make this sacrifice more often for a more certain nomination but with a less certain eventual outcome. Party is controlled for in Table 11, and it is evident that differing patterns exist between the two parties. For Democrats there appears to be an inverse relationship between the favorableness of partisan bias and the prOportion of members seeking higher office. Republi- cans, on the other hand, are slightly more likely to run in the favorable situation than in the unfavorable situation; but in each case they are substantially more likely to run than in the competitive situation. These differences tentatively suggest that there may be differences in nomina— ting patterns within the two parties, but without further evidence this remains as no more than speculation. Our discussion of competitive bias suggests that its effect upon decisions to seek higher office is much more 92 TABLE 10 PERCENTAGE OF MEMBERS RUNNING FOR HIGHER OFFICE, CONTROLLING FOR COMPETITIVENESS OF STATE AND RISK TAKING Average Statewide Vote: >5% 5% Unfav. >5% Member is: Unfav. to 5% Fav. Fav. Total Risk Taker 5.26 3.54 6.47 4.90 (323) (537) (402) (1262) Other 3.62 2.45 2.34 2.69 (662) (1018) (1066) (2746) Total 4.16 2.83 3.47 3.39 (985) (1555) (1468) (4008) 93 TABLE 11 PERCENTAGE OF MEMBERS RUNNING FOR HIGHER OFFICE, CONTROLLING FOR COMPETITIVENESS OF STATE AND PARTY Member is: Democrat Republican Total Average Statewide Vote: >5% 5% Unfav. >5% Unfav. to 5% Fav. Fav. Total 3.21 2.95 2.60 2.86 (498) (745) (998) (2241) 5.10 2.72 5.32 4.07 (487) (810) (470) (1767) 4.16 2.83 3.47 3.39 (985) (1555) (1468) (4008) 94 complex than the relationship suggested in hypothesis 4. In every instance except one, members were shown to be least likely to run in the competitive situation, lending support to hypothesis 4A. This suggests that decisions to seek higher office may be not only a function of partisan bias but also a function of the relative certainty of nomination which has been assumed here to interact with partisan bias. In the absence of more detailed information about the charac- teristics of primary contesting, the nature Of this relation- ship and its impact upon decisions to seek higher office will reamin only speculative. It is hoped, however, that the multivariate analysis to be performed in the next chapter might shed further light upon the relationship between partisan bias and decisions to seek higher office by system- atically controlling for the influence of other variables, allowing us to disentangle the influence of this particular factor. Attention is now turned to hypothesis 5. Recalling the discussion in Chapter Two, we initially limit our considera- tion of degree of shared electorate to Senate seekers only because of the assumed tradeoff between size of state and value (i.e., power) of governorships. Examination of Table 12 shows that there is indeed a very strong linear decline in attempts for Senate seats as the size of state (or number of Congressional districts) increases. This, of course, has been demonstrated before. But what of governor- ships? Again, examining Table 12, we can see that hypothe- sis 5A is at least partially incorrect. Members in the 95 TABLE 12 PERCENTAGE OF MEMBERS RUNNING FOR HIGHER OFFICE, CONTROLLING OFFICE AND SIZE OF STATE Office: Number of Governorship Senate Seat Total Congressional Districts 1 or 2 3.10 31.00 15.68 (129) (107) (236) 3 to 6 4.50 10.80 6.87 (290) (176) (466) 7 to 10 1.10 4.60. 2.82 (357) (351) (708) 11 to 19 .72 4.10 2.40 (416) (418) (832) 20 or more .62 2.30 1.53 (807) (957) (1764) Total 1.40 5.30 3.39 (1999) (2009) (4008) 96 'states with the smallest number of congressional districts are less likely to seek a governorship than are those from states with three to six districts. Furthermore, the relatively high proportion of members seeking governorships in the smaller states could stem from the inclusion in this category of members with at-large districts in larger states.7 At the extreme of the smallest state, there does appear to be a tradeoff between the value of a governorship and the probability of attaining that office as measured by degree of shared electorate. However, ignoring this extreme, we can also see that there is an inverse relationship between size of state and proportion of members seeking governorships. Although not as pronounced as in the case of Senate races, within limits the shared electorate hypothesis would also appear valid for attempts at governorships in spite of the assumed differences in power among governorships in states of different sizes. The Value of the House Seat We now turn our attention to factors posited to influence the U(Oz) term in the decision calculus. These factors are important in that they reflect the degree of sacrifice a member must make to pursue higher office. As mentioned in Chapter Two, it was assumed that the utility derived from a House seat was a function of power; and this power was shaped by a member's seniority. Hypothesis 6, provided by Rohde, suggests that there is an inverse relation- ship between a member's seniority and the likelihood of his 97 seeking higher office. Examining Table 13, we can see that when we combine categories of seniority, there is a linear decline in the likelihood of seeking higher office as member's tenure in the House increases. However, as noted in Chapter Two, the notion embodied in hypothesis 6 seems disagreeable, for it leads us to expect the first term member to be most likely to pursue higher office. This seemed disagreeable because the first term member may more often than not be relatively unknown outside his district. More time in the House might allow him to overcome this. This would, admitted- ly, increase the value of the House seat but would also serve to increase the probability of attaining a higher office. Looking at Table 14, we can see in the column totals that when members are disaggregated into individual terms, first term members are less likely to run than second term members, lending credence to hypothesis 6A. Further examination shows, however, that there remains a problem in that taken as a whole, fifth term members were the most likely to seek higher office. This may be more easily explained if we consider hypothesis 6B which incorporates another dimension of power, and hence value, of a House seat. Members of the minority party have less power in the House and derive less utility, in terms of power, from increased seniority. We would expect more Republicans than Democrats to seek higher office, and the row totals of Table 14 show that they were almost one and one-half times more likely to run than Democrats, lending support to hypothesis 6B. Reexamining the proportion of Opportunities taken within categories of seniority, it can 98 TABLE 13 PERCENTAGE OF MEMBERS RUNNING FOR HIGHER OFFICE, CONTROLLING FOR SENIORITY Number of Terms Member has Served in the House 1-3 . 4-6 7-9 :10 Total Total 4.40 3.60 2.50 .74 3.39 (1648) (1119) (665) (540) (4008) 99 NNOONV Noqm NNNNN NNNNC NONNV NNNNV NNNNN NNNNN NONNN NNNNN NoNNV NN.N NN. NN.N NN.N NN.N NN.N NN.N NN.N NN.N NN.N oN.N Nance NNNNNN NNNNV NNNN NNNN NNNNV NNNNV NONNN NNNNN NNNNV NNNNV NNoNV No.4 NN.N NN.N oN.N NN.o NN.N NN.N NN.N NN.N NN.N NN.N cauNanaom NNNNNV NNNNV NNoNv NNNNV NNNNN NNNNN NNNNV NNNNN NNNNV NNNNN NNNNV NN.N NN.o om.N NN.N NN.N NN.N NN.N NN.N No.4 oN.N NN.N NmNooeoa quNmm , m.NO£Eoz Nance oN.M m N N o N N N N N Nomzo: ecu :N wo>hom mm; Nonaoz mENOH mo Nonesz >N¢HHmOHzmm mow OZHANOmHZOU .mu~mm0.mm:c~: mom oszzsm mammzmz «H mqm mo :NmNmz >HHAHm mom UZHAAOmHZOU .mUHmmO mmIUH: mom OZHZZDm mmmmZmz no mo20% Total 4.34 5.74 ' 4.90 (738) (524) (1262) 3.47 2.36 2.69 (836) (1910) (2746) 3.87 3.08 3.39 (1574) (2434) (4008) 104 to be a further manifestation of their heightened intensity of preference for higher Office. The less risk acceptant are more unwilling to sacrifice the safe seat, choosing instead to seek higher office when the probability of their returning to the House is less certain. As noted, the retrospective nature of marginality leaves much to be desired as an indicator of the prospective probability a member attaches to securing reelection. We would ideally like to have a more sophisticated and compre- hensive indicator of all members' prospects for reelection. In the absence of such a refined measure, I have turned to how members fared in redistricting to assess how at least some members responded to lowered or heightened prospects of reelection. To compile this, Congressional Quarterly's Pre-Election Reports from 1958 to 1976 were examined. Members were categorized as being harmed if the narrative account of the member's district alteration was basically unequivocal. In a similar manner, members were categorized as being helped if the description of redistricting left no doubt as to its beneficial (to the member) effects. Cases not mentioned and those covered in speculative terms were categorized as being affected minimally or not at all. Although subjective, this method is quite conservative in that some members who may very well have been harmed or helped by redistricting were nonetheless relegated to the middle category in the absence of an absolute commitment one way or another by the reporter commenting on the impact Of redistricting. 105 In spite of the conservative nature of the coding scheme, Table 17 provides compelling evidence for the notion embodied in hypothesis 6D. In the period studied, members who were harmed by redistricting were almost three and one- half times more likely to pursue higher office than members who were affected minimally or not at all. Furthermore, members who were harmed were almost five times more likely to seek higher office than those who benefited from redis- tricting. These resultsvare important for two reasons. First, while not hitting at the center of the debate on the effects of redistricting, these results nonetheless provide evidence that redistricting does influence some members' careers. Second, these results suggest that the Pl(OZ) term in our decision calculus exerts an important influence upon the career decisions of at least some members of the House. This evidence is encouraging in that it suggests that further examination of prospective vulnerability, as. opposed to retrospective, might ultimately help us understand the role assessments of reelection success have upon the decisions of all members to seek higher office. When office is controlled for, an interesting difference emerges in the relationship between redistricting and decisions to seek higher office; the impact of redistricting on those decisions is isolated to Senate candidacies as can be seen in Table 18. Certainly the small number of oppor- tunities available to those affected one way or another renders only cautious speculation about this difference. A tentative explanation for this difference, however, might 106 TABLE 17 PERCENTAGE OF MEMBERS RUNNING FOR HIGHER OFFICE, CONTROLLING FOR REDISTRICTING Member's District was Altered: Minimally or Adversely not at all Favorably Total 11.54 3.43 2.38 3.39 (52) (3914) (42) (4008) 107 TABLE 18 PERCENTAGE OF MEMBERS RUNNING FOR HIGHER OFFICE, CONTROLLING FOR REDISTRICTING AND OFFICE Office: Governorship Senate seat Total Member's District was Altered: Minimally or Adversely not at all Favorably Total 0.00 1.47 0.00 1.45 (24) (1965) (10) (1999) 21.40 5.10 3.10 5.32 (28) (1949) (32) (2009) 11.54 3.43 2.38 3.39 (52) (3914) (42) (4008) 108 be that incumbent governors, who generally have some input on redistricting, might be inclined to see that potential challengers from Congress are not provided with an additional incentive to mount a campaign for the governorship. Without more information, this difference remains a mystery; but the fact that one in five members hurt by redistricting who have opportunities will seek a Senate seat, while none of the members so affected will pursue a governorship, at least suggests that something systematic may be involved. Age We now turn to one of Schlesinger's central concerns: the impact of age upon ambition. His work and others' leads us to expect a curvilinear relationship between age and decisions to seek higher office. This notion is embodied in hypothesis 8. It is assumed that the relatively young and the relatively old have attenuated expectations concerning higher office compared to those in middle years. Turning to the column totals presented in Table 19, it can be seen that in generalthe likelihood of seeking higher office does peak in the category of 36 to 45 years, although these categories are basically arbitrary. Members in this category are almost one and one-half times more likely to seek higher office than are those less than 35 and those 46 to 55. Furthermore, members in this age group are almost three and one-half times more likely to pursue higher office than are those over 55. Table 19 also shows that when risk-taking propensities are controlled for, a difference emerges. The high risk takers appear more likely to run for higher office with 109 TABLE 19 PERCENTAGE OF MEMBERS RUNNING FOR HIGHER OFFICE, CONTROLLING FOR AGE AND RISK TAKING Member's Age <35 36-45 46-55 :56 Total Member is: ‘ Risk Taker 5.26 6.23 6.32 1.87 4.90 (76) (369) (443) (374) (1262) Other 2.25 4.74 2.72 1.45 2.69 (89) (633) (991) (1033) (2746) Total 3.64 5.29 3.83 1.56 3.39 (165) (1002) (1434) (1407) (4008) 110 increasing age up to age 55, although this increase is minimal. The less risk acceptant, on the other hand, correspond to the curvilinear relationship posited in hypothesis 8. The high risk-takers appear to be less con- strained by the assumed disadvantages Of relative youth or relative old age. We might once again attribute the differ- ential behavior of the high risk taker to the assumed higher intensity of preference for higher office. This higher in- tensity of preference appears to lead them to be willing to pursue higher office over a wide span of the life cycle, while the less risk acceptant, with less intense preferences, appear to have their expectations shaped more by their age. This difference is interesting. A question concerning ambition is whether it develops within Office or, as Rohde suggests, it is constant with the decisions to seek higher office shaped by the inherent costs and risks in the Oppor- tunities confronting the member. The results in Table 19 suggest that for the high risk taker, ambition is almost constant through roughly age 55, although we must again acknowledge that these categories are arbitrary. On the other hand, for the less risk acceptant, it appears that ambition, as manifested in decisions to seek higher office, does heighten and then diminish with age, providing support for the notion that ambition develops and then diminishes with increasing age. Another distinction emerges in Table 20. Although the relationship between age and decisions to seek higher office remains curvilinear, Republicans appear to be much less 111 TABLE 20 PERCENTAGE OF MEMBERS RUNNING FOR HIGHER OFFICE, CONTROLLING FOR AGE AND PARTY Member's Age: .335 36-45 46-55 ‘356 Total Member's Party: Democrat 2.25 4.98 3.47 .77 2.86 (89) (562) . (807) (783) (2241) Republican 5.26 5.68 4.31 2.56 4.07 (76) (440) (627) (624) (1767) Total 3.64 5.29 3.83 1.56 3.39 (165) (1002) (1434) (1407) (4008) 112 constrained in their office seeking by age. Perhaps because of their minority status within the House, Republicans are more willing to sacrifice their House seat over a wider range of age categories. Another reason for this might be differen- ces between parties in the age at which their members first attain office. Numerous authors have noted that Republicans tend to attain office at an older age than Democrats. this holds for members of Congress studied here, it might help to explain why Republicans are less constrained by age in their decisions to seek higher office. For the young Republicans there is little current power in the House and, due to their minority status, little hope for power in the future. For the Republican entering at an older age, pros- pects for power in the House are just as dim, and decisions to seek higher office will, of necessity, come at a later age. Examination of Table 21 shows that almost 58 percent of Republicans entered Congress after age 40 while only 52 per- cent of Democrats did so. In terms of decisions to seek higher office, for both parties there is an inverse relation- ship between age at entry and proportion of members pursuing higher office, although the relationship is quite weak. For Republicans, the decisions to seek higher office begin with members who enter between ages 30 and 35 and extends to members who enter between 61 and 65. Decisions by Democrats begin with members who entered under age 30 and extend only to the 56 to 60 category. For both parties, the probability of seeking higher office is highest among members entering at earlier ages, but the decline with higher entry age is less omeCOONOQ zom«* :oxmb moNuchupoaao mo :ONNNOQOng o.ooN N. o.N o.¢ o.oN N.NN N.NN N.NN N.NN N.N NNOONN NoNN NNNV NNNNN NNGNN NooNV NNNNV NomoNN NNNNN NNNV Nanak N.N o.c N.N N.N N.N 4.N N.N N.N N.N N.4 o.ch N. N.N N.N N.NN N.cN N.NN N.NN N.NN N.N NNNNNV NoNS NNNN NmNV NNNNN NNNN. NNNNN NNNqV NcNNV NNNN mcauNanaom NN.N o.o N.N N.N N.N N.N N.N N.N N.N c.o o.ooN N. N. N.N N.N N.NN N.NN N.NN c.NN «NN.N ”u NNNNNN NoNN New NNNN NNCNN NNNNV NNNNN NNNNV NNNNV NNNN NNmNOOEo: 1 N.N o.o o.o N.N o.N o.N N.N N.N N.N «N.N qupmm m .hmflfimz NmNON NNA NN-NN oo-om NN-NN om-oe NN-NN oe-oN NN-NN ONV prucm um ow< m.NonEoz >Hmmzh :UHIZ H< mu< mom oquqomHzou .muHmmO mmon: mom UZHZZDm mammZmz no mo<92mummm Hm m4m 0 1.768 .038 Age2 -.0012 82< 0 -2.18 .014 Constituency Overlap 2.15 B3> 0 10.59 .0001 Risk Taking .21 B4> O 2.08 .019 Party .16 B5> 0 1.56 .059** Competitiveness of State -.007 B6> 0 -2.84 .002 Redistricting .78 B7) 0 2.83 .002 Office .33 B8> 0 2.90 .002 Twoyr -.92 B9< 0 -3.73 .0001 Margin .002 B10< 0 .798 .212** Senate-lst Reelection -.56 B11< B12< 0 -4.76 .0001 Senate -> lst Reelection -.86 B :>B (<0 -6.05 .0001 12 11 Governor-lst Reelection -1.20 813<0 -5.34 .0001 Governor->'lst Reelection -1.07 B14< o -3.28 .0005 Change .211 B15>.0 1.98 .024 **Not statistically significant at the commonly accepted level of signif'cance of'¢= .05 Estimated R = .4847; N = 3231 127 2 yielded by the constellation of.values of the other variables. One means to evaluate the estimated effect of a given variable would be to introduce an infinite number of values for the other variables and then evaluate the change in probability elicited by the variable of interest, given these other values. The number of combinations, of course,‘ is technically infinite; and this method would not go far in aiding our interpretation of the model. Rather than do this, a technique has been employed similar to one used by Rosen- stone and Wofinger.7 Let us assume that the constellation of all variables leads each member to have a certain predis- position, or probability, to seek higher office. By this assumption, we can evaluate the change in probability in seeking higher office associated with a given variable, at given levels of predisposition assumed to be a function of all other variables, which are held constant. The discussion that follows will employ this method to our four categories of theoretical interest: the value of the higher office, the probability of winning, the value of the House seat, and personal attributes (risk-taking and age). The Value of the Higher Office In the extreme left hand column of Table 23, the assumed probability of seeking higher office generated by all other variables is displayed. The probability of seeking a four‘year governorship is assumed to be the same as the yalues in the extreme left-hand column because this is our referent.8 128 TABLE 23 THE EFFECT OF THE VALUE OF THE HIGHER OFFICE UPON THE LIKELIHOOD OF A MEMBER OF THE HOUSE SEEKING HIGHER OFFICE Likelihood Increase in likelihood Decrease in likelihood of a member of seeking higher office of seeking higher seeking attributable to an oppor- office attributable higher tunity for a Senate seat to an opportunity for office (%) a 2-year.governorship 10 6.85 -8.60 20 10.50 -16.10 30 12.50 -22.50 40 13.20 —27.90 50 12.93 -32.12 60 ‘ 11.90 -34.80 70 10.23 -35.50 80 7.90 -33.20 90 4.74 -22.57 129 This provides us with a means of comparison. In terms of Senate seats, opportunities for this Office generate a sub- stantially higher probability of seeking higher office than opportunities for the other two offices. Looked at another way, a member with other attributes leading to a 40 percent probability of seeking higher office would not run when con- fronted with an Opportunity for a four-year governorship, while the same member confronted with a Senate seat Opportun- ity would run (P = .532). This difference in offices is clearly estimated to be enough to tip the balance in favor of running for members who would otherwise not run. Just the opposite is true for two~year governorships in more ex- treme terms. All other conditions might yield members with an 80 percent likelihood for running; but, if their only opportunity were for a two-year governorship, they are estim- ated not to run (P = .468). Only in the very extreme case where all other factors led a member to have a 90 percent prior likelihood of running would the member confronted with an Opportunity for a two-year governorship pursue it. With the effects of all other variables controlled for, the value of the higher Office is seen to be a substantial influence upon the office-seeking behavior of members of the House. These differential values can be seen to be determining factors for a wide range of prior likelihoods; those otherwise unlikely to run might do so if presented with an opportunity for the Senate, while under almost no condition would a member seek a two-year governorship. Of course, these assertions could have been made without such 130 a sophisticated technique. However, we have isolated here the specific effect of differential office value and on the basis of these results can dismiss the possible influ- ence of other variables in the model shaping the differences in decisions among offices. As examination of other varia- bles will demonstrate, the magnitude of change in probability associated with differential office values is among the more substantial of all variables considered; and this serves to underscore its importance in understanding decisions to seek higher offices. The Probability of Winning We now turn to factors posited to influence a member's estimate of securing higher Office. An initial concern was with the effects an incumbent's pursuing reelection would have upon a member's decision. In Table 24 are displayed the influence of different types of incumbents upon a mem- ber's running for higher office. Taken as a whole, the presence of an incumbent can be seen to substantially reduce the probability that a member seeks higher office. Although less so for the Senate, in all cases other conditions would have to contribute substantially to the probability of a member's seeking higher office before he would challenge an incumbent. In the case of a first term senator seeking reelection, members who would run with a 70 percent proba- bility with no incumbent would not run against a first-term incumbent (P = .484). As expected, the difference is even greater with the more senior incumbent, leaving the member 131 with only a 36.7 percent likelihood of running. Clearly, incumbency makes a substantial difference; and in the case of the Senate, the tenure of the incumbent works to further diminish the likelihood of a member of the House challenging the incumbent. The effect of an incumbent governor is even more pronounced. A member otherwise likely to run, with no incumbent, with a probability of .80 would not run against the first term governor (P = .3594). With increasing tenure, there does appear to be a slight increase in the probability that a member of the House would challenge the incumbent, which serves to substantiate the expected differ- ence between Senate amigubernatorial incumbents. In terms of magnitude, incumbency has one of the largest impacts upon decisions by members of the House to seek higher office. Although more so in the case of governors, in all cases the model suggests that the constellation of values of all other variables would have to yield a substantial probability to outweigh the influence of the presence of an incumbent. Stated differently, we would expect a member to challenge an incumbent only under extreme conditions. As noted above, the coefficient associated with the competitiveness of state was problematic at least in the extremes. Displayed in Table 24 are the estimated effects that partisan bias has upon the probability of a member's seeking higher office. In general, its effects are minimal over the range 10 percent favorable to 10 percent unfavorable. The results suggest that members are more likely to run in states where the partisan bias favors the Opposition party. 132 TABLE 24 THE EFFECT OF PROBABILITY OF WINNING UPON THE LIKELIHOOD OF A MEMBER OF THE HOUSE SEEKING HIGHER OFFICE Decrease in likelihood Decrease in likelihood Likelihood of seeking higher office of seeking higher office of a Member attributable to an attributable to an Seekin incumbent senator incumbent governor HighergOffice seeking his: seeking his: (%) lst lst lst lst Reelection Reelection Reelection Reelection 10 -6.80 -8.42 ~9.36 ~9.10 20 -ll.92 ~15.54 -l7.90 -17.20 30 -15.99 -21.60 ~25.70 -24.40 40 -19.10 -26.65 -32.60 -30.66 50 -21.20 -30.50 -38.50 ~35.77 60 -22.17 -32.90 -42.90 -39.40 70 -21.60 -33.30 -45.10 -40.88 80 -18.97 ~30.80 -44.06 -39.09 90 -13.00 -23.36 -36.40 -31.30 133 TABLE 24 (cont'd.) Influence of partisan bias on the proba- . . bility of :ikgliggggr seeking higher Increase in likelihood of seeking Seekihg office higher office attributable to the Higher Avg. Statewide degree of shared electorate Office Margin was: Number of Congressional Districts (%) 10% 10% in State: Unfavor. Favor. 1 S 10 25 10 1.12 -1.31 70.50 9.50 4.12 .56 20 2.06 -l.86 70.49 14.00 6.60 1.18 30 2.60 -2.20 64.80 16.40 8.00 1.56 40 2.80 -2.50 57.30 17.14 8.60 1.68 50 2.80 -2.80 48.40 16.64 8.50 1.60 60 2.50 -2.80 39.18 15.17 7.90 1.40 70 2.20 -2.60 29.60 13.00 6.90 1.20 80 1.86 -2.06 19.86 9.80 5.43 1.06 90 1.31 -1.10 9.90 5.73 3.38 .82 As mentioned in Chapter Two, there is reason to believe that levels of primary contesting could lead to this type of re- lationship, but due to problems in measurement and specifi- cation these results remain inconclusive. The next area of concern is with the effect shared electorate has upon pursuing higher office. Unlike the variables considered up to this point, it makes no sense to consider the absence of this variable, for all members share at least some of their state's electorate. Hence, this var- iable will always have a positive influence, and it is the difference in magnitude that is of interest. Consider members who, based on the influence of other factors, would have a 40 percent likelihood of seeking higher office, ignoring the size of their state. A member from a state with 25 congres- sional districts would have a 41.68 percent likelihood of running and hence would not run. Similarly, the member from a state with 10 districts would have a probability of .486 and also would not run. However, the member from a state with five districts would run with a likelihood of 57.14 percent, and the member from an at-large or single district state would run with a likelihood of 97.13 percent. Clearly the member from a small state or an at-large district exists in one of the extreme conditions necessary to overcome the disadvantage to candidacy posed by incumbency. The advantage of shared electorate for members from the single district state or at-large district is estimated here to consistently outweigh the disadvantage posed by incumbent senators or governors of any tenure. A completely shared electorate 135 appears sufficient to outweigh all other negative factors which is to say members in such situations may be impervious to other influences. Members in larger states, depending upon their size, are more subject to other influences; and in the largest states their decisions are based almost en- tirely upon the influence of other factors. The Value of the House Seat In examining Table 25, the estimated effects of factors posited to influence the U(OZ) and P(OZ) terms can be seen. (The reader may note the absence of seniority; see Appendix II.) The influence of being in the minority party can be seen to contribute to the probability of seeking higher office, although its effect is relatively small. Other factors would have to contribute rather substantially to the prior proba- bility of seeking higher office before minority party status could play a determining role. Although not of tremendous magnitude, it is at least suggestive of the role lessened power in the House has upon a member's willingness to leave that body. Thus, while Congress does not possess sufficient means to retain its membership, the means by which it allo- cates authority within that body does appear to have an influence upon who leave of their own volition. Another dimension of this argument can be seen by exam- ining the change that occurred after 1969. As mentioned previously, a number of authors have pointed to reforms and other changes that increased disaffection with House service among members. Increasing workloads,_reduced value of 136 om.m cN.w ce.N v. m©.~ o ca em.m en.VN NN.N m. o~.e o cw on.c om.o~ oc.~ m. 5N.m c on on.n ow.e~ om.~ o. om.m o co mm.m o~.m~ wN.m n. om.o o om 3 .mmz :Nwhmz m.NopEoz .m mm: #0350: mwv ecu :N wcwon >HNmNucmumnzm :ONNOOHO OONwwo Nosz: on manmusn poENm: mcwon ummN mN: :N NNONON> OONwwo mewxoom -NNuum oonwo ow oanuanNNuum mo :NwNmE m.NonEoE Noan: chxoom posse: m mo Nocwfln chxoom oonmo NocmN: ow oNnmuanNpum oonwo «0 moonwaexNH woocNHexNN mo coozwfloxNN wcwxeom mo woozNN NonmN: chxoom mo may :0 xuhmm :N emmONOcN -oxNN :« ommONOcN coonNNoxNN :N ommONO2N we ou:o:~w:~ mUHmmo mmIUH: oszmmm mmDOI mzb mo mmmmzmz no Goozmquwq mzb 20¢: quzzHZ mo >HHAHmpom mm; Nonsoz,mENOH we Nonszz >HHmOHzmm mom oquqombzou .UszHHmz mammzmz mo mum mqm 10% Total 94.3 94.8 95.6 95.4 (478) (603) (4156) (5237) 5.7 5.2 4.4 4.6 (29) (33) (191) (253) 9.2 11.6 79.2 100.0 (507) (636) (4947) (5490) 173 TABLE 36 PERCENTAGE OF MEMBERS RETIRING, CONTROLLING FOR REDISTRICTING Member's District was Altered: Minimally or not at Adversely all + Favorably Total Member decided to: Seek 82.9 95.6 95.4 Reelection (58) (5179) (5237) Retire 17.1 4.4 4.6 (12) (241) (253) Total 1.3 98.7 100.0 (70) (5420) (5490) 174 redistricting caused some retirement decisions requires at a minimum that redistricting temporally preceded these decisions. While in some cases this is no doubt true, an alternative scenario may be evident in some cases. A member's intentions of retiring may be known within a state before a redistricting plan is drawn. This knowledge could make the state legisla- ture's job much easier. Rather than harm the electoral chances of either party's incumbents, by drastically altering their districts, they could instead choose to make drastic changes in a district of a member who has previously announced his retire- ment, allowing them to make more minimal changes in other incumbents' districts. In this sense, a retirement decision may act as a magnet that attracts adverse alteration of a member's district. In all, vulnerability does appear to influence the re- tirement decisions of some members. As the threat of elec- toral sanctions becomes clear, it appears that some members are unwilling to bear the burden of the additional costs of a close race and instead choose to voluntarily relinquish their positions. It appears that, although relatively few in number, some members are incapable of securing or maintaining a reliable reelection coalition; rather than become more responsive, a condition normally assumed of marginal districts,2 some choose to retire. In the absence of a desire to retain their position, there is no necessary condition for assuming they will be responsive to their constituency's wishes in their remaining days in office. This is even more apparent when a member's reelection coalition is disrupted by 175 redistricting. Unable to identify or appeal to a reelection coalition within the confines of an altered district, a fairly high number of members Opt out and in the process have rendered the responsiveness inducing effects of the electoral sanction inoperable. Vulnerability may produce a burden that is too great for some members to carry and leaves them attaching their primary goals to something other than the interests of their constituents. Personal Attributes In this final category we examine the effects of differing personal attributes upon the likelihood of retirement. We turn our attention first to risk-taking propensities. As suggested in hypothesis 13, it is expected that high risk takers will be less likely to retire than less risk acceptant members. Seeking reelection involves costs and risks. Willingness to bear risks should influence whether a member decides to pursue the risky alternative of seeking reelection or of doing nothing and retiring from the House. One might suspect that the high risk taker, emerging in a district previously favored by the other party or gaining office by defeating an incumbent, would be more susceptible to defeat and because of this fewer would be retiring. However, although they may emerge originally in a competitive situa- tion, results concerning vulnerability suggest that this might lead them to be somewhat more likely to retire, other things being equal. Furthermore, seeking reelection is a conscious decision; and even if defeat were the ultimate 176 reason for the high risk takers to leave the House, they still could have chosen retirement. Retirement is an alternative open to all members. The expectation is that high risk takers will be less inclined to pursue this alternative. Table 37 provides evidence in support of hypothesis 13. Less risk acceptant members are over one and one-third times more likely to retire than are high risk takers. Although lending support to hypothesis 13, these results should be interpreted cautiously due to the possible presence of other explanatory factOrs. One plausible alternative explanation for the differential retirement rates between groups is that high risk takers enter Congress at a proportionately younger age. As operationalized, the high risk takers emerged from districts where more cautious, and perhaps older, candidates may have been less likely to tread. To challenge an incum- bent or run in a district favored strongly by the other party may be characteristics that identify not only risk taking propensities but also relative youth. Table 38 shows that indeed members categorized as risk takers were proportionately more likely to enter Congress at younger ages than were the less risk acceptant. Where 43.04 percent of the less risk acceptant entered Congress by age 40, 49.47 percent of high risk takers entered by this age. The tendency to enter younger may leave the high risk takers with higher expectations concerning power in the House and lessen their likelihood of retirement. This age difference is moderate, however, and more sophisticated multivariate analysis will be required to disentangle these separate influences. 177 TABLE 37 PERCENTAGE OF MEMBERS RETIRING, CONTROLLING FOR RISK TAKING Member dedided Risk Taker Other Total to: Seek Reelection 96.4 95.0 95.4 (1469) (2768) (5237) Retire 3.6 5.0 4.6 (55) (198) (253) Total 27.8 72.2 100.0 (1524) ' (3966) (5490) PERCENTAGE OF 178 TABLE 38 MEMBERS ENTERING CONGRESS AT AGE OF ENTRY BY RISK TAKING Age at Entering Congress for First Time: :40 41-55 :56 Total Member was: Risk Taker Other Total 49.47 43.04 94.83 (754) (1707) (2461) 43.04 50.76 48.61 (656) (2013) (2669) 7.49 6.20 6.56 (114) (246) (260) 100.00 100.00 100.00 (1524) (3966) (5490) 179 The final variable of concern is the member's age. Ad- vanced years and retirement go hand in hand in virtually any occupation. In many cases this is the result of formal manda- tory requirements. Members of the House face no such require- ment. They do, however, face the disabilities of advanced years. Furthermore, the members' estimate of their probability of returning to the House or the utility they derive from House service may both decline with advanced years. In either case, it is expected that the likelihood of retirement should increase with advancing years. Table 39 shows, not surpris- ingly, that this is in fact the case. From ages 30 to 55, retirement is very rare, accounting for no more than 2.2 percent of the career decisions in any given category. Beginning with the 56-to-60 category, retirements pick up, and this group is better than twice as likely to retire than any of the younger categories. In the 61-to-65 category, retirements increase even more, this group being at least four times more likely to retire than any group under age 55. Finally, we have the 65-and-over category which, not too surprisingly, is the most likely to exit the House via retire- ment. Over one in ten members in the 65-and-over category choose retirement rather than a reelection attempt. Another way of looking at the impact of age is to evaluate how differing age categories account for the total of retire- ments in the period studied. Members 55 and under constitute 60.24 percent of the Opportunities to seek reelection or retire, yet this group accounts for only 27.7 percent of all retirements during the period. Conversely, members over 55 «£111.15? r. ,. .4.“ a. .4 , 1 a, ,- h~41 « . ‘- a.“ ..N N t I ..r - .‘fi.a§lm'..ul O. #315 ..r.4.,_......... . . N. .. o. lo. . 47.1.0.3,0. .. rum. . .rv . 7%..“ .. . 9| .. ...”: NNNNNN NNNNN NNNNN NNNNN NNNNV NNNNN NNNNN NNNNN NNNNN NNN N.NNN N.NN N.NN N.NN N.NN N.NN N.NN N.N N.N N. NNNON NNNNN NNNN NNNN NNNN NNNN NNNN NNNN NNV Nee New N.N N.NN N.N N.N N.N N.N N.N N.N N.N N.N oNNNNN NNNNNN NNNNN NNNNN NNNNN NNNNN NNNNN NNNNN NNNNN NNNNN NNNN :oNNUNNNNN MW N.NN NNNN N.NN N.NN N.NN N.NN N.NN N.NN N.NN N.NNN Noam .1. ”OH wowwumw hoe—52 NNNON NNA NN-NN NN-NN NN-NN NN-NN NN-NN NN-NN NN-NN NNv Nom< N.Nonaoz mm m4mm oszHHmm mmmmzmz mo mo 11.592 .0001 Risk Taking -.l49 B < -1.949 .0256 Party .152 B3> 2.227 .0129 Redistricting .809 B4> 4.202 .0001 ' Mar‘gin ‘.0014 BS< -1.252 ,10* Real Salary -.015 B6< -2.611 .0045 Change .384 B7> 5.078 .0001 Entry Age .0289 B8> 7.305 .0001 Estimated R2 = .151 N = 5490 *Not significant at a = .05 187 separate influences of these two variables. Seniority was dropped from the equation because it was estimated to have a positive relationship to retirement decisions. This is dis- agreeable for a number of reasons. First, there is no reason to expect advanced seniority, per se, to lead to increased likelihood of retirement. This positive relationship is no doubt a function of advanced age that corresponds to high seniority. Second, by incorporating entry age it is felt that the effects seniority embodies are at least partially captured; the greater the entry age, the lower the expecta- tions concerning seniority and the greater the likelihood of retirement. Due to the implausibility of its estimated co- efficient, its concomitant collinearity with age, and the ability to capture the underlying theoretical concept it embodies with a surrOgate variable, seniority was deleted from the equation. In evaluating the overall model, we can see that the 2 is quite low at .151. This relatively low value, estimated R it is believed, is the result of the lack of specificity of the dependent variable. As mentioned previously, retirement as an alternative embodies many diverse possibilities. With- out knowledge of these alternatives, it is more difficult to predict a member's decision. In the case of decisions to seek higher office, a great deal of information about the alternatives confronting the member was incorporated into the model and, not surprisingly, that model performed much better. Nonetheless, considering the relative rarity of the dependent variable and the lack of specificity it embodies, the model 188 performs reasonably well. Taking each coefficient independently, it can be seen that they all correspond to theoretical expectation. Further- more, with only one exception, all the coefficients are statistically significant with the level of Type I error less than .05. The exception is the coefficient associated with the member's margin of victory in his last election. However, even this coefficient is significant at the somewhat less acceptable and less stringent, but not unreasonable, level of a = .10. All in all, the model corresponds well to theoreti- cal expectations. To fully evaluate the influence of the factors in the model requires the implementation of the procedure utilized in Chapter IV. Because of the assumed curvilinear function of the underlying dependent behavior, the estimated influence of the independent variables depends upon the levels taken by other variables in the euqation. As in Chapter IV, the levels of the other variables are set at prespecified levels, and the influence of the independent variable of interest will be evaluated at each of these different levels. The discussion that follows will be divided into factors influen- cing the value a member attaches to House service, the proba- bility of returning to the House, and personal attributes of the member. The Value of the House Seat We begin our consideration of value of the House seat and its impact upon retirement decisions by evaluating the 189 influence of the age at which a member enters the House. The estimated effects of this variable are presented in Table 41. For the sake of comparison, the effects of entry age are pre- sented with the average entry age of House members as a refer- ent. Thus, the figures presented represent the differential probability of retirement for members entering below and above the average entry age for the period studied. Beginning with the relatively young entrants of age 30, we can see that they display a substantially lower likelihood of retirement than members entering at the average age of roughly 43. While the average member in a situation that would lead him to retirement with a probability of .60, the member who entered at age 30 in an otherwise equal situation would have an estimated probability of retirement of .4541, and it is thus estimated that he would not retire. The member who enters at 40, being close to the mean entering age, does not differ substantially from the average; but as we move to the member who enters at 50, the effect of increased entry age on retirement can be seen. While the member who entered at the average entry age might be in a situation where he was teetering on retirement with a probability of .50, the member in the same situation who entered at age 50 is estimated here to retire with a probability of .5842. Finally, we turn to the member who enters at age 60. Entering the House at this later age is estimated to substantially increase the likeli- hood of retirement. Whereas the member entering at the average age in a situation with a probability of retirement NN.N- NN.N NN.N NN.N 1490 o em.c mm.m «N.N- cN.N- cm om.m- oo.N . ma.m em.m c No.NN mm.m o~.~- NN.NN- ow cm.o- om.N ch.NN ~m.v o mc.vN wo.N om.~- mw.mN- on oN.N- c~.~ oN.mN Nc.m c ov.nN NN.N cN.m- mm.vN- so VN.N- oe.~ mm.vN cv.c c N~.mN Nv.m mc.m- o~.ep- em eo.o- om.N mm.mN mo.c c va.m~ NN.N mN.~- ma.m~- cc om.m- cm.~ mm.vN eo.m c om.mN mm.n mv.~- c~.~N- cm Nc.v- NN.N NQ.NH hm.e c ON.oN mm.c mo.~- oc.m- cm ~m.~- mm. e~.m on.~ c NN.NN No.v ov.N- NN.N- oN .fiwm am...- . . eN NaueeNeNNNa NeNNN eneeN sea Ne eN 3 Ne NN AmthNoc NomNV 0:» :« omzoz ueosouwueu N0u< um weNeuzm Nessa: may thamm ON ecu :« mafia; on NO coogmwoxmy vo.~e mo chhmuoz emnmusnmuuum age so Nahum peoEONNNON mo oocosz:_ mo coozNNoxNN :« emmehocN oNnmuanNuum acoEONNNON mo coozmmoxNN ecu :m moucopowwN: ewe zhuco eueho>m one o>one Nonsoz a me No xenon mmoeucou mcmuoNeo coocNNexNN ON o~amu=nmuuum «coachmueu No pooszexNN one :« moocmhemNN: oz—z~hmm mmzo: m:b mo mammZmz mo noozmamxmq mzb 2oz: mu_>:mm mmzo= no >H~A~b2 no Hummum 2:? .av mgm mo :Nmpmz N.NenEoz wow :mzp mmoN No Noumoum NNONON> mo mewwpma ow oNnmpznwhuum NCOEONNNON mo poozwaoxNN ecu :N moocopommwn chNNuom Noose: m we woosNNoxNN oszHHmm mmzoz mzh mo mmmmzmz no GOOIHmeHA m2? 20m: mmDOI mzh OH quszHmm no >HHAHm