.. ~r. .r a, . WW0? . 345! 7:1... .1, .11.. 4» . .y. NM... . .9... a . erfrdnixmvnfl . 42.82%? an I 1...... . Cr 14. T “munummmwfi .NH .— ,/ ' I?! T ‘ '3 a Wwwmra ‘ "9 Po COMPET'. ALYSIS OF STATE PARTY . “v.17! .I .. .. r . w..w.....w.¢...n....,.¥ . 13...". . r1 1.. . . 1 .4 gunfirérurfiuniyvmu... .. g .. s . . . .uwnfihfiuuufi. mm. 9.». u (I CY AND . . , . . a? a. air . EN , , . . . , . Lfiufiafimwfifiwwfi . . , . r , . . . . v . dirt r..f . v 4m. u». . A . . _ .rl. «mm... Hg. . .gww , .n. , , . . .. . . . o «Janna... 1113.3“. .1!qu v. Iva! ; than 931.)! I . 3.": r Sis} .1 trri BEN I mouM ‘AN‘AN '7 "The . . Iii/5P.» rt... 7 157:7... :1. .. I: r: a.mr..lr.....lll.r...hvu ...l. . . 1%.. mgfifiévg .. .fi This is to certify that the thesis entitled INCUMBENCY AND COMPETITION: AN ANALYSIS OF STATE PARTY POLITICS presented by Mark Sage Hyde has been accepted towards fulfillment of the requirements for _Ii.i)_.___degree in LQLiLiQaLSCience fl #4:... Major professor Date 1% LIBRARY Michigan State University a; 9.... .n. -v-— “-1 ABSTRACT INCUMBENCY AND COMPETITION: AN ANALYSIS OF STATE PARTY POLITICS BY Mark Sage Hyde The two hypotheses under examination in this dis- sertation are: (A) The level of intereparty competition in state politics is an inverse function of (1) the number of candidacies for reelection by incumbent state legislators (2) the candidacy of an incumbent governor for reelection. (B) The decisions of incumbent state legislators and an incumbent governor to seek reelection are a direct function of the level of inter-party com- petition. The relevant theoretical and empirical literature with respect to the two hypotheses is reviewed, and opera- tional definitions of the variables of incumbency and com- petition are explained and described. The hypotheses are tested using parametric statis- tical techniques with data collected for approximately a 50 year period on all gubernatorial and state legislative races in four states: Connecticut, Iowa, Kansas, and Ohio. A model of the interaction of the variables of in- cumbency and competition over time, based on the two Mark Sage Hyde propositions above, is developed. The model is examined in light of the empirical findings with respect to the two hypotheses, and these empirical findings are related to theoretical formulations and empirical studies in the area of political recruitment. INCUMBENCY AND COMPETITION; AN ANALYSIS or STATE PARTY POLITICS BY Mark Sage Hyde A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Political Science 1972 g... T0 MARCIA... H- ACKNOWLEDGEMENTS A great number of people contributed to this work and I am grateful to all of them. First, I must thank my dissertation committee. Professor Paul Conn, my chairman, always gave me enough freedom to develop my own ideas, to make and learn from my own mistakes, and in general, to feel that the work I was doing was my own. 0n the other hand, his guidance in every facet of the work is clear to me and greatly appre- ciated. Professor David.Meltz mixed exactly the right quantities of criticism and praise to get the most work from a harried graduate student. For his voluminous work on my writing style, David Meltz probably deserves to be listed as a ghost writer. Professor Joseph Schlesinger provided a great deal of substantive criticism of this work; a short paragraph of his criticism usually contained enough substance to keep me busy for an extended period. Other faculty members also made substantial con- tributions to this study. Liz Powell wrote the program that organized and made sense out of mounds of raw data; she also provided help with mathematical problems. Susan Lawther provided substantive comment on the work, but more important, she listened. iii My fellow students Al Arkley, Paul Hain, and David Klingman all contributed in various ways. Paul and I had many discussions concerning ambition theory and its possi- bilities for generating testable hypotheses. David did a considerable amount of work of my computer program, while Al produced some much needed money at a crucial point. Undergraduates Tim.ReynoldS and Harry Barman helped in collecting and coding the data; Mrs. John Stinchfield not only produced a well typed, but a well edited manuscript. Finally, I express my greatest appreciation to Marcia and Jennifer.‘ They watched without complaint as my role of graduate Student and doctoral candidate came to more and more dominate my role and responsibilities as husband and father. They adapted to the circumstances better than I--they have contributed much to this work. iv TABLE OF CONTENTS CHAPTER 1. INCUMBENCY AND COMPETITION . . . . . . . . . 1 2. THE EFFECT OF INCUMBENCY ON ELECTORAL COMPETITION: THE RESEARCH DESIGN AND EMPIRICAL FINDINGS . . . . . . . . . . . . 31 3. THE EFFECT OF ELECTORAL COMPETITION: THE INCUMBENT'S DECISION . . . . . . . . . . . 50 A. Governor o o o o o o o o o o o o o o o 55 B. Legi313ture O O O O O O O O O O O O O 56 C. MajorityéMinority Differences . . . . 58 4. INCUMBENCY: SOME CONCLUSIONS . . . . . . . 78 I. Status of the Model . . . . . . . . 78 II. Suggestions for Future Research . . 83 III. Implications of the Findings . . . . 86 IV. COnCIUSion o o o o o o o o a o a o o 89 BIBLIOGRAPHY O O O O O O O O O O O O O O O O O O O O 93 GENERAL! REFERENCES 0 O O O O O O O O O O O O O O O O 97 APPENDIX A O O O O O O O O O O O O O O O O O O O 0 O 102 Table 2.1 2.2 3.1 3.2 3.3 3.4 LIST OF TABLES Percentage of Time That Parties Retain Offices Running Incumbent and Non- Incumbent Candidates . . . . . . . . . . Effect of Incumbent Legislators Seeking Reelection Upon Party Competition for Legislative Seats and Office of Governor Effect of Incumbent Legislators Seeking Reelection Upon Inter-Party Competition for Legislative Seats and Office of Governor-Incumbent Governor Seeking Reelection . . . . . . . . . . . . . . . Effect of Incumbent Legislators Seeking Reelection.Upon Inter-Party Competition for Legislative Seats and Office of Governor--Non—Presidential Years . . . . Effect of Incumbent Legislators Seeking Reelection Upon Inter-Party Competition for Legislative Seats and Office of Governor--Incumbent Governor Seeking Reelection in Non-Presidential Years . . Mean Level of Competition and Mean Percentage of Incumbents Running-«All Years 0 O O O O O O O O O I O O O O O O Inter-Party Competition and Incumbency Decision to Seek Reelection--State LegiSIatorS00000000000000 Inter-Party Competition and Incumbency Decision to Seek.Reelection--Office of Governor O O O O O O O O O O O O O O O O Inter-Party Competition and Incumbency Decision to Seek Reelection, for Non- Presidential Years--State Legislators . vi 37 38 4O 41 43 61 63 64 66 Table 3.5 Inter-Party Competition and Incumbent Decision to Seek Reelection, for Non- Presidential Election Years--Office of Governor................. 68 3.6 Decision of Incumbent Governor to Seek Reelection as a Function of Changes in the Level Of Competition 0 o o o o o o o o 73 3.7 Competition for Legislative Seats and Incumbency Decision to Seetheelection, Controlled for Party-«State Legislators . 75 vii Figure 3.1 LIST OF FIGURES COMPETITION AND INCUMBENCY . . viii 52 CHAPTER 1 INCUMBENCY AND COMPETITION A substantial body of literature has accumulated within the discipline of political science that may be roughly classified under the heading of ”political re- cruitment." These studies have attempted to isolate the psychological, social, and political variables that in- duce or inhibit men to run for public office. The study of recruitment is of substantial inter- est to the political scientist who seeks to understand (explain) why an individual chooses to run for office rather than merely vote, contribute money to a campaign, or actively work to see a candidate other than himself elected. We should note that the answer to this and re- lated questions has important normative implications. Some of the very early empirical work in political science was directed, for normative purposes, to the study of the decision to run for and remain in elective office. Writing in 1938 with data gathered from eight states, Charles Hyneman attempted to explain the high turnover rate in state legislative bodies.1 He was deeply con- cerned with the common pattern of transient membership in legislatures that prevented the emergence of a stable, 1 experienced, and therefore more capable, group of leaders. One of the most noted political scientists who studied American political parties, V. 0. Key, Jr., stated that one of the functions a party Should perform is the recruitment of candidates for public office. Key points out: A prime function in the achievement of party purpose consists of the recruitment, development, and support of candidates for public office. In the governance of large and complex societies, parties provide a means for sifting from the available candidates behind whom support can be amassed in campaigns for office. Ob- ‘viously, this operation of veto, of cajoling, of trial, of choice lies close to the heart of the govern— ing process.2 According to Herbert Jacob, modern investigations of political recruitment can be grouped in three cate- gories: 1. elite theories 2. studies of social backgrounds 3. psychological theories.3 The cumulative result of these approaches provides the following information. A pool of potential candidates for elective office, characterized by common social back— grounds and political socialization patterns, can be iden— tified. Within this pool, it is possible to predict which individuals will want to run for office on the basis of psychological variables. But between the pool of poten- tial candidates and the holding of elective office lies politics, the process determining which members of the group will be selected to run and which will be finally elected to office. e. . _ u _ o o . O a p I - nJ . I . I . r . a. t n 0 — l ' l n .. . y n I. a 9 u r... . _ . .. . — . r I. I n . I . . . An alternative approach to the above three is to examine those political variables which influence the selection of leaders in the American political system. This approach, unlike the previous ones, places emphasis upon the politics of recruitment, i.e., the identifica- tion, description, and measurement of political variables that, in the context of political theory, are believed to affect decisions to seek either initial or later career offices. Representative of this approach is Joseph Schlesinger. Schlesinger's major work, Ambition and Politics, is a detailed study of the careers of a large sample of 4 Schlesinger recognized politicians in the United States. the emphasis on social and psychological research in the field of political recruitment and commentedr A wealth of data exists about the backgrounds of po- litical leaders...yet, deSpite the industry which has gone into the accumulation of these data, they have been notably fruitless in producing predictive propo- sitions in relating data to behavior. And this failure has been due primarily to the lack of a useful political theory. Theory there isé but it is generally social or psychological. The sample under consideration in elite, social, or psychological studies is, in general, the entire popu- lation of the United States (or some other unified politi— cal system), and those who choose to seek elective office are identified by some set of characteristics that differ- entiate them from the general population. Most of the re- cruitment literature has been concerned with the initial decision to run for office, concentrating largely on new- comers to the political scene. But the sample changes when the group under examination becomes those who have actually attained elective office. The researcher is.no longer trying to pick and choose from among a general population of those who will run for office, but rather is now working with a sample of individuals who are already a subset of the general population, officeholders. These people are apparently differentiated in at least one reSpect from the larger sample: they respond more readily to political stimuli. Office holding, and therefore politics, plays a large, perhaps a dominant role in a politician's public life. Unless the politician has what Schlesinger calls "discrete ambitions" (the desire to hold an office for one term and then retire from politics), he will respond to political stimuli as the businessman responds to economic stimuli. There has been very little research on the conditions under which those politicians holding a particular office will choose to run for that same office again. This would seem to be a fruitful area of research for political scientists, and emphasis on political variables seems especially appropriate for this specialized sample. The study of the electoral behavior of incumbent officeholders is a valuable undertaking whether, like Hyneman, one prefers, in a normative sense, the development of an experienced leadership at several levels of government, or on the other hand, if one's I . 1. I . L . interest is purely theoretical, i.e., measuring the fac- tors that lead to rapid or slow turnover in public office. Now let us examine more closely the first three approaches mentioned above, detailing the general criti- cisms that have been made. Elite theories range over analysis as diverse as Marx and Lasswell, who both emphasized the importance of elites in political affairs. Marx, of course, perceived an elite ruling class based upon economic power, suggesting that those who control the means of production enjoy po- litical as well as the economic power in the state. For Lasswell, the elite are those who get the most of what there is to get; they are the "influential" who have more of such representative values as deference, income, and safety.6 Gaetano Mosca concentrated upon showing the dominance of social positions in a political system, while Vilfredo Pareto contended the ruling class had psychologi- cal traits which enabled them to dominate the community.7 Many important empirical studies of political leadership begin with elitist assumptions. C. Wright Mills identifies certain elite groupings which he claims have dominant political, economic, and social power in American society.8 Floyd Hunter, one of the first to in- vestigate power structures at the local level, finds elite groups holding the dominant positions in this stratum of society.9 Political scientists such as Robert Dahl and Robert Presthus have attempted to overcome some of the . . . D . n u. l . . . . . o n . I u . . . . - . .. . . . C i r o . — I . c . . I | . n u . . r . . l a. . . . l .. n D . U . Ix . . , simplifications of the earlier studies and argue that sev- eral elite groups share power at the community level. Studies of the social characteristics of decision- makers have been the most frequently used approach in the analysis of political recruitment. In two books, Donald Matthews demonstrated that United States Senators and other high level politicians in the country “...normally come from upper-status occupations, have a better than average occupation, are predominantly Protestant, white, and native born."11 Earlier writings in political Science described the social origins of the machine politicians of Chicago in the 1920's and early 1930's.12 Other authors employ the concept of political socialization in exploring the background of political decision-makers. They isolate variables such as father's orientation toward politics that are part of the overall socialization process of an individual and may be useful predictors of future politi- cal activities.1 AS Schlesinger indicates in his criticism of the lack of political theory, the difficulty common to these two approaches to the study of political recruitment is that neither of them say much about recruitment or poli- tics. The elitist and sociological approaches do little more than estimate some soci0¢economic status parameters of American political officeholders. These studies, with varying degrees of precision, are limited to defining the pool of potential candidates for political office. Their focus is generally at the macro-level and does nothing more than identify sub-groups within the social, economic, and political sub-systems that have provided the majority of leaders in the country. The lack of any theoretical framework in such a book as‘Matthews',‘Thg Social gagg- ground 2; Political DecisionrMakers, for example, leaves the reader with a great deal of data to which he can assign little or no significance.14 The psychological approach goes one step further. These researchers, rather than trying to define the social and economic backgrounds from which particular individuals come, have attempted to determine what psychological drives motivate them to run for elective office. The emphasis here is on isolating personality traits; given a pool of potential candidates (those who satisfy the appropriate background and socialization criteria discussed above), this approach attempts to determine why some individuals within the pool seek elective office and why others do not. Working with the individual rather than the group as the unit of analysis, they attempt to determine the rela- tionship between personality traits and the decision to run for office. In a series of articles, Rufus Browning has con- cluded that two variables are characteristic of those who seek elective office 1. They are marked by high power and achievement moti- 'vation, and 1 2. By relatively low affiliative concerns. Those individuals who entered politics by accepting the re— cruitment requests of party leaders (rather than actively seeking candidacy)”...were distinguished only by high levels of affiliative motivation,” i.e., empathy for other people. The reason for the difference between those who openly seek candidacy and those who simply accept an offer will become clear in the discussion of Lester Seligman's work below. We should note, however, that the emphasis in this psychological approach is on volunteerism and psychologi- cal drives and not on recruitment or politics 22; gs. The question to which research of this type is directed is: Which people are motivated to seek political office, and, as such, usually overlook the crucial political variables that may be systematically related to the process by which a party seeks a candidate or prevents a motivated person from attaining a nomination. Joseph Schlesinger, who is representative of the approach to political recruitment emphasizing political variables, provides a sound theoretical base in his "ambi- tion theory."l7 Rather than asking how one advanced in politics, Schlesinger's theoretical perspective leads him to seek an answer to the question: 'Who will want to ad- vance and under what political circumstances? The primary assumption of his ambition theory is that a politician's behavior is a response to his office goals, and that these goals or ambitions are related to the specific situation confronting the politician. His analysis begins where the psychological approach ends by assuming that politicians are more or less ambitious, and that these ambitions are nurtured or constrained by the political opportunities available to him. Articulating this theory with data on political careers, Schlesinger maps out what he terms the "political opportunity structure” in the United States and then proceeds to analyze the relationship between this structure and such variables as party organization and inter-party competition.18 Lester Seligman has also concentrated closely on a systematic investigation of political variables influencing recruitment. In one of his studies, Seligman divided re- cruiting into two stages: 1. Certification, the social screening and political channeling that results in eligibility for candi- dacy, and . . 19 2. Selection, the actual ch01ce of candidates. He selected four Oregon state legislative dis- tricts on the basis of their respective levels of inter- party competition and examined candidate entry into the primary race. Two of the districts were competitive; one was Republican dominated, and the last Democratic domi- nated. While admitting that he is working with an ex- tremely small sample that may be biased by local factors, Seligman arrives at some interesting generalizations. In one party dominated areas, officials of the majority party were least active in instigating or supporting candidates. 10 The political marketplace was relatively accessible to "self-starters“ and interest groups. In competitive dis- tricts, the candidate marketplace was "wide open"; in these areas a direct relationship was found between party competition and recruitment diffusion. In districts Safe for one party, the minority party had to actually con- script candidates. Seligman concludes that when a party is in this “hopeless minority" position, "...only in this setting was fully centralized party recruitment to be found.”20 Seligman's generalizations shed some light on the findings of Browning, discussed above. Browning found that those who actually sought candidacy were marked by high power and achievement motivation and low affiliative concerns, and that those individuals who entered politics at the request of party leaders scored high on affiliative concerns. Seligmans has hypothesized that the only time party leaders request people to run for office, i.e., the only time they have centralized control over recruitment, is when the party is a hopeless minority. It is under- standable that the person who agrees to run for office when he haS'virtually no chance of winning rates high on affiliative concern. He accepts the overtures of party leaders as a favor for a friend or because he wants to help an organization which faces possible extinction. Conversely, in Situations of majority party dominance or high competition, when the marketplace for candidates is 11 more open, the type of person attracted to run for office is either the self-starter or the representative of special interests, both of whom seem likely to rate high on power and/or achievement motivation, but not Show great empathy for their potential constituents. Within the approach concentrating on political variables, Schlesinger and Heinz Eulau have at least ap- proached the topic of incumbents' electoral behavior. Schlesinger talks about incumbency in general terms, while Eulau has specifically formulated a series of propositions with respect to the office of state legislator. Schlesin- ger points out that the rate of turnover of personnel (rather than party) in an elective office is greatly af- fected by the structure of political opportunities. The more often an office is vacated by an incumbent (for what- ever reason), the greater is the opportunity for someone else to run for that office, i.e., the larger is the op— portunity structure. If there is a large number of ambi- tious newcomers to the lower levels of elected office- holders, there will be a greater demand for rapid turnover of personnel in higher offices to provide opportunities for these newcomers. Schlesinger comments: ...there is evidence that, where parties face increas- ing competition, they value incumbents more highly for their proven vote-getting abilities....In a competi- tive situation, what might be considered the instinc- tive demand of other politicians for rapid turnover in high office is counteracted by the need for partisan politicians to subdue their personal office ambitions in order to reap the benefits accruing to their party from continued control of the state's principal of- fice.... 12 Schlesinger is arguing that incumbents may prove to be better candidates for the office they already hold than either newcomers to the political scene or a poli- tician from another office, because the incumbent has proven a winner with that particular constituency. ‘When a party faces stiff competition, some politicians may be forced to subdue their ambitions for higher office by per- mitting incumbents to hold the fortress when the battle gets rough. Similarly, those at the higher levels of office holding may have to postpone their movement up the career ladder in order to run as an incumbent at the lower level against a severe challenge. While in this instance (high level of competition) a particular politician's of- fice goals may be sacrificed, party control of an office or set of offices may be retained. A politician might sacrifice his personal career goals for the sake of overall party success under several conditions. He might see his long run personal goals closely tied to the success of the party, and in this in- stance, what is beneficial to the party is beneficial for the individual politician. On the other hand, when party and personal goals are clearly at odds, the party might offer the incentive of support for future, attractive nominations for higher office if the politician would presently put off his conflicting career goals. Obviously, the party cannot continue this delaying tactic for any ex- tended period of time. 13 Heinz Eulau, in The Legislative System, discusses the relation between inter-party competition at the state level and the decision of incumbent state legislators to . 22 . . return to their seats. Eulau and hlS co-authors had In- terviewed state legislators in four states: California, Ohio, Tennessee, and New Jersey. He summarized the data on competition and incumbency this way: In order to determine just how committed our respond- ents were to the legislative office they were occupy— ing, we asked them whether they expected to run again for the state legislature...£ang/ the data Show a con- sistent inter-state pattern. It appears that the more competitive the party system of a state, the greater is the proportion of legislators who expect to run again for their legislative seat....In general,...the data suggest that politicians in more competitive situations seem to-be more inclined to see their legislative career as a continuing enterprise than do legislators in less competitive situations.23 Eulau concludes with the following proposition: The more competitive the structure of the political party system, the more likely it is that state 1egis~ lators will be committed to their legislative career by planning to run for their present seat again.24 Beyond stating that politicians in competitive situations seem more committed to a legislative career as a continuing enterprise, no explanation is offered along with the predictive proposition. There is a short discus- sion of “...the consequences of the difference between competitive and non-competitive systems on career commit- ment..." indicating that it is likely to affect legisla- tive deliberation and action, but as for an explanation of why legislators in competitiveasituations say they will choose to run again, Eulau has little to say. Pointing .. _ .. .. . .. . L: _ .. _ .- . , .(‘a‘ -. ' . v: 14 out that the project is concerned with the analysis of state legislatures as role systems and that the discussion of legislators' political careers is only peripheral to the study, he leaves further exploration and fuller ex- planation to future researchers. The primary purpose of this dissertation is to provide a systematic test and satisfactory explanation of the relation between the independent variable of level of inter-party competition and the dependent variable of the proportion of state legislators seeking reelection. The major hypothesis, following Eulau, is that a high level of competition will lead to a greater likelihood of incumbent state legislators and an incumbent governor seeking re— election. The theoretical defense of the hypothesis is fo— cused more upon Schlesinger's contention that incumbents may provide continued party control of a particular of- fice, or set of offices, and less upon Eulau‘s argument that legislators in a competitive situation are more com- mitted to their seats. That is, based on.Schlesinger's ambition theory and nuclear unit theory of party organiza- tion, an incumbent will be viewed as responding to pressure from the party to seek reelection in a competitive situa- tion. However, an important point must be made here. In his statement concerning the reasons that incumbents may be induced to seek reelection, Schlesinger assumes that incumbents will win elections more often than 15 non-incumbents. This is a critical assumption for the major hypothesis of this dissertation, and therefore, it will be put in proposition form and subjected to empirical verification. This is our first task: the political ef- fect of incumbency decision to seek reelection must be determined before we can predict when incumbents will choose, or be induced, to seek reelection. The predicted effects are set forth in the following prOposition: The level of inter-party competition in state politics is an inverse function of (1) the number of candi- dacies for reelection by incumbent state legislators, and (2) the candidacy of an incumbent governor for reelection. Before we discuss incumbency advantage at the state level, let us turn to an examination of legislators at the national level, where a substantial amount of work has been done in determining the success of incumbent Con- gressmen and Senators. Although there are obvious differ- ences between state and national legislators (e.g., size and makeup of constituency, prestige, income, substantive issues, etc.), they are enough alike that parallels in behavior patterns may be found. In a legal, formalistic sense the U. S. Congress is a state dual legislature writ large. The most comprehensive work on the fate and effect of Congressmen seeking reelection has been done by Milton Cummings.26 Cummings analyzed election returns for the period 1924-1964 for both the House of Representatives and President in an attempt to pinpoint the voting patterns of 16 American citizens. In a general appraisal of the elec— toral advantage held by incumbent Congressmen, Cummings classified legislative districts with reSpect to how the Presidential candidate of the incumbent's party ran in- those districts. It would be expected that the advantage enjoyed by an incumbent Congressman would appear not only in districts where the Presidential candidate of his party ran well, but especially in a situation where the Presi- dential candidate of his party ran poorly, offering him no electoral help. Cummings found that the incumbent had a marked advantage in both situations. From 1924 to 1964, only 18 incumbents were defeated in districts where incum- bent candidates of the same party as the winning President were running, and in which the winning President carried the district.27 Cummings concluded that, "In districts carried by the winning President, even those where he won by a narrow margin, the House candidates of his party who were already in Congress were nearly invincible."28 For Congressional candidates who faced an opposi- tion victory for the Presidency both at the nationwide and district level, incumbency was a crucial variable. Cum- mings says, "...at practically every level of opposition presidential strength, incumbent House nominees of the party that lost the presidency did better than their non— incumbent fellow partisans."29 The outlook for non- incumbent House Candidates was indeed bleak: in districts where the winning President's margin climbed over 52% p r . . . up. .. .. . .s — . u. r .r. . . I n. .- . .— I .r . . . — I. 17 “...the chances that a non—incumbent House candidate could be elected dropped steadily until they were virtually nil.”3o The advantage of Senate incumbents at the national level has also been documented. Lewis Froman, Jr., pre- Sents an 85% success figure for Senate incumbents.31 Barbara Hinckley has used Senate incumbency as one inde- pendent variable in explaining deviations from expected party-line voter in Senate elections.32 Her conclusions are that incumbency and the Presidential vote can explain 64% of Senate voting fluctuation. At least at the national level, then, there is abundant evidence that incumbent legislators of both houses have a distinct electoral advantage. However, systematic explanations of this advantage are not offered by these authors. Cummings focuses on some of the rela- tionships between Congressional and presidential voting, without offering any explanation for incumbency advantage. Froman presents his figures on incumbency success and con- cludes the Senate and House have fairly stable membership, but again no explanation is forthcoming. Hinckley sug- gests two alternative possibilities for incumbency advan— tage--"recognition" explanation and the "experience" ex- planation posited by Joseph Schlesinger. These two alternatives both have merit for ex- plaining the advantage of a particular legislative incum- bent when he enters the electoral arena, at either the 18 state or national level. The “recognition" explanation is based simply upon the assumption that an incumbent is more widely known than his opponent. With generally low levels of voter interest and information about issues, voters will choose the more familiar (or as Hinckley points out, the less unfamiliar) name. Schlesinger's explanation is that legislators, during their years of office holding, accumulate experience and seniority which can be used to build electoral support.33 In comparison, an executive officeholder accumulates grievances, rather than electoral support, as a consequence of his seniority. The executive office is more visible, and the voter finds it less diffi- cult to assign governmental responsibility to one execu- tive rather than numerous legislators, and orders his voting preferences accordingly. Also, the legislator's constituency (especially at the state level) is likely to be more homogeneous and less pluralistic than that of statewide officeholders. An explanation of incumbency advantage as it af— fects the level of statewide inter-party competition (as stated in the above hypothesis) is slightly different than an explanation of a particular incumbent's advantage. Of equal importance to presenting an explanation for the in— cumbency advantage enjoyed by both governors and state legislators is the necessity to explain in what way the decision of incumbents to run again for their offices af- fects other candidates in other races. That is, in 19 addition to explaining why incumbents have greater success than non—incumbents in a particular race, it is necessary to also explain that a generally lower level of competi- tion in other offices is produced. As Schlesinger sug- gests, party organization and success are determined in large part by the interaction between party candidates and factions. An explanation of the finding that an incum- bent's decision to run depresses the level of intereparty competition in office races (other than his alone) would necessarily have to take account of the relationship be- tween these several offices. What follows is a theoretical defense of the first hypothesis, combining both the recog- nition and seniority explanations, and taking into account the relationship between what Schlesinger calls ”nuclear units." Schlesinger contends that a party consists of politicians seeking election to specific public offices and that the form or structure of a party depends upon the relationship between the efforts of these individuals and their supporters. In terms of party organization, Schlesinger introduces the concept of ”party nucleus." A nucleus is the "...basic unit of party organization.[3n§7 is the collective effort devoted to the capture of a single public office."34 In other words, a nuclear or- ganization is built around a single candidate for elective office. The major problemcf political party organization becomes the relation of these nuclei to each other, and 20 the fundamental factor relating party nuclei, according to Schlesinger, is the electorate. There are three possible relationships between electorates (i.e., constituencies) and their party nuclei: electorates may be congruent, disjoint, or enclaved. If the set of voters for two offices is the same, for example those of United States Senator and Governor of a particu- lar state, the constituencies are congruent; if there are no overlapping voters, the constituencies are disjoint. Finally, one constituency may be a subset of another, and in this case, the smaller constituency is said to be en- claved within the larger. An example of an enclaved con- stituency would be the electorate of a state legislator to that of the governor, the relationship under examination here.35 It is in this sense that political parties, and an officeholder's relation to the party, will be considered throughout this study. The level of cooperation among these officeholders is largely determined by the sharing of constituencies, the level of inter-party competition within these constituencies, and the politician's response to the opportunities (or lack thereof) with which he is confronted. First, an explanation of why incumbent governors and state legislators win more often than non—incumbents will be presented; then, the effect of legislative races upon gubernatorial elections and vice versa, will be 21 discussed. An incumbent legislator starts with the advan- tage of a recognizable name within the constituency. Assuming low voter interest for races at the state legis- lative level, the voter may cast his lot with a name that is recognizable, and an incumbent's name would be recog- nizable in a political context. Furthermore, the incum- bent has had an opportunity to use patronage payoffs in his district to build direct electoral support and main- tain at least a small, informal organization (or nuclear unit). Even if the assumption of low voter interest and information is not strictly held, it is difficult to imagine a challenger unseating an incumbent by an attack on the ”issues." The responsibility of a legislative body is so diffused among its many members that pinning failure of a pOpular measure or passage of an unpopular one to a single legislator is a difficult task for an opposition candidate. The legislator, unlike the governor, is not held responsible for the implementation of entire legisla- tive programs, and consequently, he is not held personally (i.e., electorally) responsible for unpopular developments in state government. As Schlesinger points out, he can hide in the "group atmosphere" of legislative politics. An incumbent governor has several advantages over a non~incumbent challenger, but accumulated liabilities may outweigh the advantages after several terms in office. Anthony Downs addresses himself to a situation of 22 accumulated liabilities in this theoretical work, An Economic Theory 9; Democracy. In a discussion of how a party might unseat an incumbent, Downs describes several strategies open to the chaflenger, including that of build- 36 ‘When no majority of ing a ”coalition of minorities." 'voters reaches a consensus on all issues, the challenging party can attempt to build a winning coalition across several issues, i.e., forging a majority from several separate, distinct minority groupings. Individually, each group does not have the strength to upset the incumbent, but their allied strength may provide at least a minimum winning coalition. However, at least the one term incum- bent governor is probably in an advantaged position. As the chief executive of the state he will gain wide pub- licity in the news media, and this gives him a headstart in terms of voter recognition on most nominees that might oppose him.37 In addition, the incumbent governor may not need a successful record of solving state problems to maintain voter support. Murray Edelman claims a leader must Show a willingness to cope with problems, but he is not held responsible for successfully solving those prob- 38 lems. He quotes George Gallup in a discussion of opinion polls: People tend to judge a man by his goals, by what he's trying to do, and not necessarily by what he accom- plished or by how well he succeeds. People used to tell us over and over again about all the things that Roosevelt did wrong and then they would say, "I'm all for him, though, because his heart is in the right place: he is trying."...If people are convinced you 23 are trying to meet problems and that you are aware of their problems and are trying to do something about them, they don't hold you responsible for 100 percent success. Nor do you have to have any great ideas on how to accomplish the ends.3 Another advantage for the incumbent is the gover- nor's role (perhaps in concert with a U. S. Senator) as titular head of the party in his state. As such, he wields a large amount of patronage power, and an effective use of patronage during his tenure in office can provide a strong electoral base from which to begin his campaign. Because the incumbent has these electoral advantages, and has proven his vote getting ability, finding an attractive opposition candidate may be difficult. The potential electoral strength of an incumbent may inhibit an ambi- tious politician of the opposite party from mounting a challenge. Few ambitious politicians will enter a race if they feel the chance of electoral success is marginal; if possible, under the limitations of time and age, they will choose to wait for a better opportunity. An incumbent governor without major political liabilities may find him— self confronted with only a weak challenge from a ”sacri- ficial lamb.“40 As pointed out in the discussion of minority coalitions above, these advantages held by an in- cumbent governor may decrease over time: the publicity and responsibility of the governor's office illuminate mis— takes and liabilities, as well as positive characteris- tics. David Leuthold has argued that incumbency 24 advantage is due largely to the ability to raise and wisely employ the resources needed by any candidate.41 In a study of Congressional campaigns, he found incumbents easily acquired necessary resources, while for non- incumbents ”...the acquisition of resources was one long 42 In addition, non- series of requests and pleas." incumbents who could secure resources reduced the number of requests because a losing effort in the race would mean several years of repaying financial and social debts in- curred in their campaigns. Leuthold also found that in- cumbents had more experience in using the available re- sources, citing several instances in which newcomers used expensive campaign tactics which the incumbent knew from previous elections were ineffective. The decision of state legislative incumbents to seek reelection has ramifications for the gubernatorial race, and the candidacy of an incumbent governor has a direct effect on legislative races. The candidacies of state legislators will provide organizational strength for the gubernatorial candidate at the district level, where direct voter contact occurs. A gubernatorial candidate (incumbent or non-incumbent) will need an effective state- wide organization to meet any serious chalenge, and the nuclear units of his party's incumbent state legislators provide bases around the state on which to build his cam~ paign. Instead of attempting to build a statewide organi- zation of his own for the race, the gubernatorial 25 candidate can rely on district organizations of legisla— tive incumbents to help campaign in certain areas. The nuclear organization of an incumbent legislator is already in existence and functioning at the start of a campaign: a non-incumbent hopeful usually must build his unit from scratch and cannot expend much effort on behalf of those higher on the ticket. Furthermore, an incumbent legisla- tor has a decided electoral advantage, as well as an or- ganized nuclear unit, giving him and his organization even more time to contribute to those at the top of the ticket. The candidacy of an incumbent governor also pro- vides electoral benefits for the legislative candidates of that party. With a strong candidate at the head of the ticket, many legislative aspirants may ride into office on the ”coattails" of the governor. If those who vote a straight party ticket on the basis of the governor's race are sufficiently impressed with the gubernatorial candi- date, the legislative candidates of that party will be simultaneous benefactors in the voting returns. This coattail effect has been documented at the national level (Presidential--lower offices), and there is no reason to believe that it does not operate at the state level as well.43 Thus, the sharing of constituencies places the candidates of the enclaved and larger electorate in a mu— tually dependent situation. The gubernatorial candidate's need for statewide organization strength is partially met u T 1': 26 by incumbent legislative candidates, and the coattail ef- fect of an attractive candidate heading the ticket polls votes for legislative candidates. The following chapter is devoted to developing the research design used to test this first hypothesis and also analyzing the relevant data with respect to that hypothesis. The chapter concludes with a short section on the implications of these findings for the electoral strategies of state parties. In Chapter 3, the major hypothesis is developed, which predicts the decisions of incumbents to seek reelection as a function of inter-party competition. Also, a model of the inter- action between incumbency and competition is presented. The final chapter is devoted to an examination and evalua- tion of this model and the empirical findings of the study. FOOTNOTES Charles S. Hyneman, ”Tenure and Turnover of Legislative Personnel," The Annals, Vol. 195, 1938, p. 21. 2V. 0. Key, Jr., American.State Politics: An Introduction (New York: Alfred Knopf, 1956), p. 11. 3See Herbert Jacob, ”Initial Recruitment of Elected Officials in the United States-—A Model,” Journal 2; Politics, Vol. 24, No. 4, 1962, p. 704. 4Joseph Schlesinger, Ambition and Politics (Chicago: Rand McNally, 1966 . 51bido , pp. 12.13. 6See for example, Harold O. Lasswell, 32 al., The Comparative Stud 9; Elites (Palo Alto: Hoover Institute Studies, 196 and Harold Lasswell, Politics: Egg Gets What, When, Hgy (New York: World Publishing Company, 1958 . 7Gaetano Mosca, The Ruling Class (New York: McGraw Hill, 1939): Vilfredo Pareto, The Mind and Society (New York: Dover Publications, 1935). 8C. Wright Mills, The Power Elite (New York: Oxford University Press, 1956). 9Floyd Hunter, Community Power Structure (Chapel Hill: University of North Carolina Press, 1953). 10Robert A. Dahl, Who Governs? (New Haven: Yale University Press, 1961), and Robert Presthus, Men at the Top (New York: Oxford University Press, 1964). 11The quote is from Jacob, 22.‘gi2., p. 705. See Donald.Matthews, The Social Background 2; Political DecisionéMakers (New York: Random House, 1954), especial- ly pp. 20-32. Also, by the same author, g. g. Senators and Their World (Chapel Hill: University of North Car0* lina Press, 1960), eSpecially pp. 11-67. 27 ...Htuw.“ r ..: t . -.~ ~ ' ..._. — . - ..pwfl— 28 12See Harold Gosnell, Machine Politics (Chicago: University of Chicago Press, 1937). 3 . . . For example, Kenneth PrewItt, pp p;., "Political Socialization and Political Roles," Public Opinion Quarterly,‘Vol. 30, 1966, p. 91. 14Matthews, pp. cit. 1SRufus P. Browning, Businessmen in Politics: Motivation and Circumstance ip ppp'Rise pp —Power, unpub- lished Ph. D. dissertation, Yale University, 1960, ”The Interaction of Personality and Political System in Deci- sions to Run for Office: Some Data and a Simulation Tech- nique," Journal pp Social Issues,‘Vol. 24, 1968, p. 93: ”Hypotheses About Political Party Recruitment: A Partially Data Based Computer Simulation,” in William D. Coplin, Simulation in the St tud of P litics (Chicago: Markham Publishing Co., 1968 p. -303, with Herbert Jacob, "Power Motivation and Political Personality," Public Opinion Quarterly,‘Vol. 28, 1964, p. 75. 16Browning in Coplin, Simulation, pp. cit., p. 308. 17See Schlesinger, Ambition, pp. cit., eSpecially Chapter 1. 18Ibid. 19 Seligman' 5 main works on recruitment are: “Polit— ical Recruitment and Party Structure," American Pplitical Science Review, Vol. 55, 1961, p. 77: “Party Roles and Political Recruitment,” Western Political Quarterly, ‘Vol. 11, 1958, p. 361: "Leader Selection in the Oregon Legisla- ture, ” Western Political Quarterly,‘Vol. 12,1959, p. 67, "Political Parties and the Recruitment of Political Leaders,” in Lewis Edinger (ed. ) Pplitical Leadershi ip Industrialized Societies (New York: John‘WIley and Sons, Inc., 1967). 20Seligman, ”Political Recruitment and Party Structure," pp. cit., p. 77. 21Schlesinger, Ambition, pp. cit., p. 68. 22John‘Wahlke, Heinz Eulau, William Buchanan, Leroy Ferguson, The Legislative System (New York: HJohn Wiley and Sons, Inc., 1962). See especially Chapter 6, drafted by Eulau. ZBIbido 9 pp. 121‘1220 29 0.. hi H 26Milton Cummings, Congressmen Lnd the Electorate, Elections for the U. S. House of Representatives Lnd w Pres1dent, 1920-1964 TFree Press: New York, 1966). 27 Ibido, p. 720 31Lewis A. Froman, Jr., The Long Process (Little, Brown and Company: Boston, 1967), p. 170. 2Barbara Hinckley, ”Incumbency and the Presiden- tial Vote in Senate Elections: Defining Parameters of Sub-Presidential'Voting," American Pplitical Science Review, Vol. LXIV, No. 3, September, 1970, p. 836. 33Schlesinger, Ambition, pp. pip. 34Joseph Schlesinger, "Political Party Organiza- tion” in.James G. March (ed.), Handbook pg Organizations (Chicago: Rand McNally, 1965), p. 784. The concept of the party nucleus is intuitively very satisfying, or in technical terms, has a lot of face validity. A Democratic Party professional is quoted as saying that the Republicans attempt to run Congressional campaigns from the top of the party (i.e., trying to force cooperation between units), but ”...we [Ehe Democrat§7 try to get a candidate who matches his district. In effect we run 435 separate campaigns." As quoted by A. P. writer John Beckler in.Michigan State News, March 9, 1970, p. 2. 35For a discussion of enclaved disjoint, and con— gruent electorates, see ibid., pp. 787-793. 36Anthony Downs, pp Economic Theopy pp Democracy (New York: Harper and Row, 1957), see pp. 55-60. 37The data from this study (latest returns, 1950) are not affected by what might presently be an important variable for building voter recognition--television. Large amounts of money Spent on television by a relatively unknown may overcome the advantage of candidates that rate higher on voter recognition. For example, in the Demo- cratic primary for U. S. Senator in Ohio in 1970, a rela- tively obscure businessman, Howard Metzenbaum, defeated a 30 national hero, former astronaut John Glenn; both men at- tributed.Metzenbaum's success to an effective and expen— sive use of television. 38Murray Edelman, The Symbolic Uses of Politics (Chicago: University of Illinois Press, 1962), es- pecially Chapter 4. 39Ibid., pp. 78-79. From Opinion Polls: ILter- ‘views py Donald McDonald with” ElmoiRoper Lnd Gporge Gallup (Santa Barbara, California, 196 65), pp. 34-35. 40Of course, electoral success can never be as- sumed. In 1968, while the Republicans were winning the Presidency, incumbent Republican Governor John.Chafee of Rhode Island lost to just such a sacrificial 1am-—circuit court Judge Frank Licht, a candidate conscripted by the Democrats who thought they were the "hopeless minority." 41Da'vid Leuthold, Electioneering in a Democracy: Campaigns for Congress (New York: John‘wiley and Sons, Inc., 1968). 2 4 Ibid., p. 122. 43For a discussion of the "coattail" effect see Milton Cummings, pp. pip., especially Chapters 1— ~33 v. 0. Key, Jr., Parties, Pol1tics, Lnd Pressure Groups, 5th ed. (New York: Thomas Crowell Co., “1964); Barbara Hinckley, "Interpreting House Mid-Term Elections: Toward a Measure— ment of the In-Party's 'Expected' Loss of Seats," American Political Sgience Review (September, 1967), p. 694; Charles Press, "Voting Statistics and Presidential Coat- tails,“ American Political Science Review (December, 1958), p. 1041, and "Presidential Coattails and Party Cohesion," Midwest Journal of Political Science (November 1964), p. 320, Warren E. Miller, "Presidential Coattails: A Study in Political Myth and Methodology," Public Opinion Quar- terly (Winter, 1955-1956); Angus Campbell and Warren E. Miller, "The Motivational Basis of Straight and Split Ticket'Voting,” American Pglitical Science Review (June, 1957), p. 293. CHAPTER 2 THE EFFECT OF INCUMBENCY ON ELECTORAL COMPETITION: THE RESEARCH DESIGN AND EMPIRICAL FINDINGS The hypothesis to be tested in this chapter is as follows: The level of inter-party competition in state politics is an inverse function of (1) the number of candi- dacies for reelection by incumbent state legislators, and (2) the candidacy of an incumbent governor for reelection. The independent variable of incumbency decision to seek reelection is divided into two parts: decisions by legis- lators and the decision by the governor. The operational definition and measurement of this variable is straight- forward. For the legislators it is simply the percentage of the elected legislature at time t who choose to run in the next regular election, time t + 1. The percentage was calculated by dividing the number of incumbent legislators running at time t + 1 by the total number of legislators elected at time t, making the rate of incumbency decision to run for the legislature a continuous variable. In the governor's case, incumbency decision to seek reelection is a dichotomous variable; that is, either the governor does or does not seek reelection. The governor's decision to seek reelection was assigned a value of 1; a negative 31 32 decision by the governor was given a value of O. The operational definition and measurement of the independent variable, inter-party competition, is a bit more complex. For the governor's office, the level of competition is defined as the difference between the per— centage vote of the two top parties, and subtracting that value from 1, the result is a measure of competition that ranges from .001 (lowest level of competition) to .999 (perfect competition).1 The measure of competition for the lower house was devised by aggregating separate figures for each legisla— tive race. This measure, with three slight modifications, was adopted directly from the doctoral dissertation written by David B. Meltz.2 Meltz needed a long term measure of competition for a time study analysis of cohesion in state legislatures. He had found the initial idea for the measure in a dissertation written by Mark Stern, who had developed a competition index to study two party competi- tion in townships over several years. The Meltz measure is a continuous variable (meaning parametric statistical analysis can be employed), measures perceived competition, and in an elegant touch by Meltz, ranges between zero and one. The final equation for the measure (JL.) is as follows 33 Mi‘os _ _ Lo xwl = Jae _ (1 ——9§———)r, where x 6 i , and political unit under study number of observations mean division of the party vote over r years i r p. 1 As Meltz points out,.fl.- can be used to measure two party competition in any voting unit--electoral, legislative, or committee. For the derivation and a more complete ex- planation of the measure see the Appendix at the end of this volume.3 In constructing his measure, Meltz experimented with assigning differing weights to each legislative session's partisan division in computing,ui and 6;, arguing quite correctly that assignment of equal weights to each of their sessions was as arbitrary as any other. Positing that current parameters affect present behavior more than events longer in the past, he assigned weights first in arithmetic, and then in geometric, progression from the first to the last session under analysis. For the purpose of this study, a measure of long term competition is not needed, but the Meltz measure can be readily adapted for our purposes. The need is for a measure that will aggregate all state legislative races in one election year into a statewide measure of electoral competition for legislative seats in that year. First, the mean division of the two party vote was computed across all districts as a whole for each election year, 34 rather than over a time period of n years. In this way, .IL- becomes a Spatial rather than a temporal measure. Meltz‘ procedure of weighting relatively more im- portant events is also employed in the adaptation of his measure. Many of the state legislative districts under analysis each have two, three, or more representatives. In fact, one Ohio state legislative district has 18 repre- sentatives elected at large.4 Therefore, in computing the mean and standard deviation of the division of the two party vote, weights were assigned to the districts equal to the number of representatives elected from that dis- trict. Single member districts received a weight of 1, dual member districts, 2, and so on. By this procedure, the relatively greater importance of districts electing more than one representative is taken into account. Need— less to say, the weight for any particular district was changed as its number of representatives was reapportioned. There was one other difficulty encountered in adapting the.IL- index to the present study. In collect- ing the data for each district race, the vote of each party was calculated as a percentage of the 5223; vote cast, and the percentage figures of the two top parties did not total 100% if candidates of more than two parties received votes in the general election. The limit of .5 in computing the value of x becomes meaningless for these districts. A new value called "winning latitude” (WL) was defined such that: 35 WL = 2 96 of Democratic vote + Z i of Republican votes Number of Districts Using this new value WL in place of .5 renders the measure applicable to a three (or more) party situation, when only two of the parties are considered. If WL is calculated for a two party situation, in which the percentage votes of the two necessarily total 100%, the value of WL is of course equal to that in Meltz' original equation, .5.5 We now have Operational definitions and descrip- tion of measurement procedures for both the independent and dependent variables. In order to test the associa- tion between the variables, the following statistical analysis was performed. Pearson product moment correla- tions were run between the independent variable of per- centage of legislative incumbents seeking reelection and the dependent variable of measures of competition for governor and legislative races, for each year under analysis. Then two variables were employed as controls; first, the correlation was calculated for only those years in which an incumbent governor was running, and second, the correlation was calculated for only non-presidential years. Finally, correlations were calculated employing both control variables. The hypothesis was tested using data from four states: Connecticut, Iowa, Kansas, and Ohio. These states were chosen for three reasons. First, the returns 36 were readily available, having been included in an analysis of state politics done by V. 0. Key, Jr.6 Second, a systematic tabulation of incumbents seeking re- election required that district changes over time be kept to a minimum. In three of the states; Ohio, Kansas, and Iowa, there were no additions to, deletions from, or boun- dary changes in the number or makeup of state legislative districts in the time period under study. All three elected state representatives from counties rather than from Specially drawn districts. Connecticut, on the other hand, elects their general assembly from towns rather than counties. In the 50 year period under analysis in Connec- ticut, two towns disappeared from and two other simul- taneously appeared on the roll of municipalities who had representation in the general assembly. These four towns are completely left out of the data.7 Finally, the four states display diversity along at least two dimensions. Kansas is one of only six states that elect all state legislators from single member dis- tricts, while Connecticut has all single or dual member districts. Iowa and Ohio both have a significant portion of multi-member districts. In addition, the four states provide a fairly diverse geographic distribution. The time period under analysis for each of the states is Connecticut, 1900-1948; Iowa, 1906-1946; Kansas, 1912-1950; and Ohio, 1908—1948. These data include a total of 13,815 separate legislative races and the 37 election of 86 governors.8 We must first compare the success rate of incum- bent candidates by state officeholders to the level of success of those at the national level, discussed in Chap- ter 1. It will be remembered that Cummings, Froman, Hinckley, and Leuthold all found that incumbents had a marked electoral advantage over non-incumbent opponents, and the same is true at the state level.9 Table 2.1 clearly demonstrates that incumbents win more elections than would be expected by chance. Table 2.1 Percentage of Time That Parties Retain Offices Running Incumbent and Non-Incumbent Candidates §2§2§ WITH INCUMBENTS WITHOUT INCUMBENTS Leg. Gov. Leg. Gov. Conn. 82.2 72.7 62.9 57.1 Iowa 80.9 79.3 75.3 66.7 Kansas 80.5 63.6 70.7 57.1 Ohio 75.5 61.5 63.0 28.6 It should be noted that Schlesinger's contention that legislators benefit from incumbency more than execu— tive officeholders holds true. ‘While incumbent guberna- torial candidates show an advantage over non-incumbent opponents, they do not win as often as incumbent state legislators. The main point, however, is that both state legislative and gubernatorial incumbents win more often than non-incumbents.10 38 We must now turn from an examination of the simple dichotomous variable of winning-losing to the effect of incumbent legislator candidacies upon the level of inter- party competition for both legislative seats and the office of governor. Table 2.2 shows this effect. Table 2.2 Effect of Incumbent Legislators Seekinijeelec- tion upon Party Competition for Legislative Seats and Office of Governor* fi OF LEGISLATURE SEEKING REELECTION STATE LEGISLATURE GOVERNOR Connecticut -.038 .296 (N=24) Level of Inter- Iowa -.088 -.184 (N=19) Party Competition Kansas -.329 .107 (N=18) Ohio -.280 -.109 (N=18) * ProductéMoment Correlations The data in Table 2.2 show mixed results. In each of the four states the relationship between incumbent can- didacies and competition for legislative seats is in the predicted direction—-competition is an inverse function of the number of legislators seeking reelection. However, the relationship is very weak, and in no case does the correlation ever reach a level of .4. In the case of competition for the governor's chair, the data from Iowa and Ohio show a very weak relationship in the predicted direction, but the other two states show a weak relation- ship opposite to that predicted. 39 The deviating states are Connecticut and Kansas, and they have two common structural elements that may ex— plain the unexpected findings. Both states, in relation to Iowa and Ohio, have a relatively large number of state legislative districts, and Kansas elects all representa- tives from single member districts while Connecticut com- bines only single and dual member districts. In addition, as mentioned above, Connecticut elects its 188 assemblymen from towns rather than counties, as in the other three states. With a large number of districts and only one or two representatives elected from each district, the prob- lem of communication and cooperation at election time is likely greater in these states. The predicted relation- ship between the number of legislative incumbents seeking reelection and competition for the office of governor was based upon a certain level of cooperation between these nuclear units. If lack of cooperation caused by a large number of separate legislative races undermines this joint effort, it is not surprising that the predicted relation— ship does not hold. The findings presented in Table 2.2 measure the effect of legislative incumbents seeking reelection upon competition for legislative seats and the governor's race, but the hypothesis being tested also predicts competition as an inverse function of an incumbent governor seeking reelection. Thus, we must next discover the additive ef- fect of an incumbent governor seeking reelection upon 4O competition for the governor's office and legislative seats (coattail effect). That is, Table 2.2 does not get at the interaction effect between the governor's race and races for the legislature. Table 2.3 presents the rela- tionship between the percentage of legislative incumbents seeking reelection and the level of competition, con- trolled for years when an incumbent governor is running. Table 2.3 Effect of Incumbent Legislators Seeking Reelec- tion Upon Inter-Party Competition for Legisla- tive Seats and Office of Governor-Incumbent Governor Seeking Reelection* PERCENTAGE OF INCUMBENT LEGISLATURE SEEKING REELECTION STATE LEGISLATURE GOVERNOR Connecticut -.542 .223 (N=11) Level of Iowa —.361 -.291 (N=13) Competition Kansas -.515 -.167 (N=11) Ohio -.533 -.655 (N=10) * ProductéMoment Correlations It can be seen immediately that the relation is greatly strengthened when the control variable of incum- bent governor seeking reelection is employed. This is especially true for inter-party competition for legisla- tive seats; while in Table 2.2 there was only one r value above the .3 level, now there is only one value below .5. The predicted coattail effect is clearly shown in this table; the decision of the incumbent governor to seek 41 reelection greatly aided the candidacies of state legisla- tive candidates. The relationship for the governor's races is also strengthened, with Kansas now displaying results in the predicted direction. Only Connecticut shows re- sults opposite to those predicted, but less so than in Table 2.2. Another control variable employed was that of iso- lating non-presidential election years. As suggested by V. 0. Key, Jr., the tides of national politics greatly af- fect political events in the several states, and thus the data from the four states were analyzed for only non- presidential election years.11 The results are shown in Table 2.4 Table 2.4 Effect of Incumbent Legislators Seeking Reelec- tion Upon Inter-Party Competition for Legisla- tive Seats and Office of Governor--Non- Presidential Years* PERCENTAGE OF INCUMBENT LEGISLATURE SEEKING REELECTION STATE LEGISLATURE GOVERNOR Connecticut -.455 .140 (N=12) Iowa -.522 -.555 (N=10) Level of Competition Kansas -.054 .331 (N=10) Ohio -.629 -.643 (N=10) * ProductéMoment Correlations Once again the relationship (with the exception of Kansas) is strengthened over treating the data as a whole 42 in Table 2.2. The absence of a presidential race makes it more likely that state electoral politics will follow the pattern predicted in the hypothesis. The various state nuclear units can interact in a "normal" manner, without worrying about either relying upon or defensively reacting to the national ticket of either party. Again, Connecticut and Kansas show weaker relationships than Iowa and Ohio. In fact, Ohio and Iowa have a stronger relationship be— tween the variables when the control variable of non- presidential years is employed than when the data are con- trolled for the candidacy of an incumbent governor. This is another indication that fewer total districts with more multi-member districts provides a political structure more conducive to cooperation between legislative and the gubernatorial nuclear units. In the absence of the unifying force of a nationwide election, these states have parties that can put forth a cooperative electoral effort. Finally, both variables of candidacy of incumbent governor and non-presidential election years were con- trolled, and the results are presented in Table 2.5. 43 Table 2.5 Effect of Incumbent Legislators Seeking Reelec- tion Upon Inter-Party Competition for Legisla- tive Seats and Office of Governor--Incumbent Governor Seeking Reelection in Non-Presidential Years* PERCENTAGE OF INCUMBENT LEGISLATURE SEEKING REELECTION STATE LEGISLATURE GOVERNOR Connecticut —.609 .472 (N=4) Level of Iowa -.464 —.510 (N=7) Inter-Party Competition Kansas -.742 .147 (N=4) Ohio -.560 -.754 (N=7) * ProductéMoment Correlations The results in this table are mixed, probably due to the small number of cases under analysis. The pattern of results in the predicted direction, with the exceptions of governors' races in Connecticut and Kansas, holds again, but, for example, the high correlations in Kansas legislative races (-.742) and Ohio gubernatorial races (-.754) are clearly suSpect. In general, the results indicate that incumbent candidacies indeed have a negative effect on competition; i.e., inter-party competition is an inverse function of incumbent legislative candidacies and the candidacy of an incumbent governor, and the hypothesis is supported. How- ever, the relationship is more clearly defined for states that elect representatives from a relatively smaller num- ber of districts, at least some of which are multi-member. 44 The crucial distinction is probably fewer number of dis- tricts, making organization of a collective electoral effort less difficult. The existence of multi-member dis- tricts is most likely a result of having fewer total dis- tricts, some of which are populous enough to have more than one representative. And multi-member districts may be a further advantage to a cooperative effort. Because all multi‘member districts included in the data elected representatives at large, cooperation within districts was forced upon candidates. On the whole, the data showed that multi-member districts usually elected an entire slate of party candidates, rather than splitting the vote between candidates of different parties. In other words, state legislative candidates of one party from a multi- member district usually had their electoral fate decided as a whole. Promising young candidates, such as Robert Taft of Ohio, were the notable exceptions to this rule. The implications of these findings in relation to the value of incumbent candidates for state parties is clear; incumbents win more often and provide electoral support for other party candidates. The critical assump- tion, mentioned in Chapter 1, upon which the theoretical argument for the major hypothesis is based has proven valid. We can now move to a preliminary discussion of the electoral strategy of state parties with regard to incum- bent candidacies. Rather than assuming the benefits of incumbents seeking reelection, (this having been 45 demonstrated in the above hypotheses), we must now only assume that the parties are aware of them. The advantages of incumbents running for office clearly exists; the as- sumption is that the party perceives these benefits. While it is doubtful that any state party has undertaken a systematic study of incumbency electoral advantage, it seems unlikely that party officials and candidates, based upon direct experience in electoral politics, would be un— aware of these advantages. The recognition of incumbency advantage leads to the following electoral strategy, as suggested by Schlesinger. In periods of intense competition, the party (officeholders and their nuclear units) will rely on in- cumbents to assure party control of the governor's chair and the state legislature as a whole. The direction of reliance of one nuclear unit upon another, in the case of an enclaved-larger constituency set as that of state legislator-governor, is dependent upon the level of com- petition in the separate constituencies. Schlesinger pro- vides the following table to explain the dependency of one nuclear unit on the other, or the interdependency of both. 46 LARGER CONSTITUENCY Competitive Non-Competitive ,‘ 1. 2. 3 Competitive Interdependent Smaller Depen- > dent on larger at) 351 m "u 3. 4. 3:2 Not Larger Depen- No Interde- —uo ompetitive dent on pendence 3‘) smaller 1% In the first situation, candidates in both the larger and enclaved electorates are facing a competitive race. The two nuclear units will be interdependent, each relying on the other to help build at least a minimum winning coalition. In the second situation, the state legislator is faced with a serious challenge, while the gubernatorial candidate is fairly sure of winning his race. The enclaved unit of the state legislator is then in the dependent situation; he needs the help of the larger nuclear unit to face serious opposition, while the gubernatorial candidate can probably win regardless of the performance of the enclaved nuclear unit. The third situation is precisely the reverse; the gubernatorial can- didate is in electoral trouble while the state legislative aspirant needs little or no help. Finally, in the fourth situation, the electoral outlook for both nuclear units is either so promising or so bleak that neither unit offers 47 help to the other. We would expect, then, that parties facing intense competition at either level (or both) would attempt to induce, or coerce, incumbents to run. The following chap- ter will develop the major hypothesis of the dissertation, based upon the empirical findings presented above. The research design used to test the hypothesis, and the findings with reSpect to that hypothesis, are also in- cluded in Chapter 3. FOOTNOTES 1The percentage figures were calculated in such a way that they totaled .9999..., instead of 1.0. Thus the measure ranges from .001 to .999 rather than from 0 to 1.0. 2 David B. Meltz, Com etition and Cohesion: A Model g£_Majority Party Legislative Bargaining (Unpus- lished Ph.D. Dissertation, University of Rochester, 1970). 3Ibid., pp. 55-56. Meltz developed his index as a measure of perceived competition, to be used in measuring the independent variable. However, the index is open to other interpretations. In this instance,.n_ is used to measure the dependent variable of inter-party competition and is interpreted simply as a closer approximation of reality than, for example, a simple percentage difference. The measure, by use of the logarithmic transformation, accounts for the decreasing marginal utility of X as X in— creases. In other words, the politician cares more about increasing his vote total by 1%.when he is confident of gaining 50% of the vote than when he is confident of gaining 90% of the vote. However, the case can be made that there is an objective as well as perceived difference in such situations. When the index is used in the next chapter to measure the independent variable, it will be interpreted as measuring perceived competition. 4The district is Cuyahoga County. 5I wish to thank.Miss A. Elizabeth Powell for help in overcoming the problem of dealing with party votes cal- culated as a percentage of the total vote. 6See‘V. 0. Key, Jr., American State Politics: Ag Introduction (New York: Alfred Knopf, 1956). 7The Connecticut towns left out of the data are Chatham, East Hampton, Huntington, and Shelton. 8Several years of the Ohio election returns were missing from the Key data and were supplied by the State Library of Ohio at Columbus. 9See Chapter 1. PD. 48 49 10The rate of success for holding onto offices with or without incumbent candidates is somewhat differ- ent for the majority and minority parties. Although the returns were not controlled for party in the original analysis of the data, a spot check of the raw figures in- dicates a party clearly in the majority can be highly successful with either type of candidate. However, the conclusion to be drawn from the data in Table 2.1 remains the same-~incumbents win more often than non-incumbents. Furthermore, the basic thrust of the chapter is to measure the effect of incumbent candidacies upon the level of inter—party competition, not upon the simple won-loss dichotomy. That is, we are more interested in the margin of victory and the relationship between offices than in which party won or lost a particular election. 11Key, 22. cit. CHAPTER 3 THE EFFECT OF ELECTORAL COMPETITION: THE INCUMBENT'S DECISION In Chapter 2, we examined the effect of incumbent candidacies upon the level of inter-party competition and found that 1) incumbent state legislators and governors are more likely than non-incumbents to win elections, and 2) incumbent candidacies generally produce lower levels of inter-party competition both for the incumbent's own race and for those of his party involved in an enclaved or congruent relationship with him. Based on this informa- tion, the attempt will be made in this chapter to predict when incumbents will choose to seek reelection. The following hypothesis will be tested. The decisions of incumbent state legislators and an incumbent governor to seek reelection are a direct function of the level of inter-party competition. The two variables of the first hypothesis have been reversed; competition is now the independent vari- able, while an incumbent's decision to seek reelection is the dependent variable. For this second hypothesis, the relationship is conceived of as involving a time lag ef- fect. The proposition will be tested with data for contiguous electoral years within the four states under 50 51 analysis; incumbency decision to seek reelection (depen- dent variable) at time t + 1 is predicted to vary as a result of the level of inter-party competition (indepen— dent variable) at time t. For example, competition in 1940 is predicted to have certain effects upon the number of incumbents running in 1942, in turn, the level of com- petition in 1942 would influence the number of incumbents running in 1944, etc. The two hypotheses of the study considered to- gether present a theoretical statement of an ongoing relationship over time between competition and the deci- sion of incumbent state legislators to seek reelection. The relationship is predicted to be of the following orf der: a large number of incumbent candidacies produce a low level of competition in that election year; the low level of competition will lead to fewer incumbent candi- dacies in the following year which, as shown in Chapter 2, produces a high level of inter-party competition; the high level of inter-party competition leads to a large number of incumbent candidacies in the next election, and the cycle begins to repeat itself. In graph form (with the curve smoothed out between data points), the relationship would appear as intersecting sine curves. 52 Figure 3.1 COMPETITION AND INCUMBENCY High Competition Low t t+1 t+2 t+3 t+4 t+5 Time Competition: ---------- Incumbency: This predicted relationship of Incumbency and Competition over time is obviously dependent upon the second hypothesis being substantiated, and further discussion of this model must await a discussion of that proposition. There are both methodological and substantive problems that cult; some of outset of the of the data. switching the first to the second hypothesis. make testing the second hypothesis diffi- these difficulties were foreseen from the study while others developed in the analysis The first, and most obvious, problem is independent and dependent variables from the The two propositions, despite the introduction of a one election time lag in testing the second, may be contradictory: if more 53 incumbents running lowers the level of competition in that election year, then demonstrating that the greater the competition, the more incumbents will run in the next election year may prove impossible. However, the rela- tionship between competition and incumbency may occur over longer periods of time than one election. Second, the primary statistical tool of data analysis is once again the Pearson Product-Moment Corre- lation. It was recognized at the very initial stages of the study that low level correlations would produce an ambiguous situation in interpreting the data. Low corre- lations may indicate either a lack of relationship between the level of competition and the decisions of incumbents to seek reelection, or opposite behavioral tendencies among a fairly equal number of incumbent legislators which would balance each other and disguise any relationship. The distinction is crucial as the interpretation of no relationship might force us into a Type II error, accepting the null hypothesis when in fact it should be rejected.1 Finally, the office of governor and state legis- lator may not be strictly comparable in the context of this study. Legislator's behavior is considered in the aggregate (the percentage of a legislature seeking reelec- tion) while the behavior of a governor is examined in- dividually; individual idiosyncrasies will show up much more readily in the behavior of the governors. Second, sample size for state legislators is much more than 1,000 54 times as great as that for governors. Both these factors will make explanation and prediction of gubernatorial be- havior more difficult than that of state legislators. Although there are expected differences in elec- toral behavior on the part of majority and minority party candidates, both parties view incumbent candidacies in a favorable light. The advantages to either party, majority or minority, of having incumbents run is clear cut. A political party, it will be remembered, is defined as a group of aspiring candidates and elected officeholders and their respective nuclear organizations involved in a more or less cooperative effort to gain election to Specific public offices. The basis for cooperation is the desire for political success (the fulfillment of political ambi- tion) on the part of individual candidates. Schlesinger argues that in a competitive situation involving enclaved electorates, parties will look to improve the quality of all candidates on the ticket because the careers of candi— dates in this type of relationship are closely tied to- gether. ...a party which is competitive for statewide office will attempt to find and improve upon its candidates for the state legislature and Congress, even when there is no expectation that they can win, because their activities will assist the state candidacies. Similarly, legislative candidates in competitive dis- tricts will want an attractive gubernatorial candi- date even if he cannot win, because of the marginal benefits to them.2 We can conclude, in general, that parties (especially at a time of intense competition) will attempt to nominate an 55 entire slate of attractive candidates, and based on the success rates and effects on competition of incumbent can- didates shown in Chapter 2, we will argue that incumbents are the most attractive and most available candidates for a party to have on the ticket. We will now spell out more specifically the ad— vantages of incumbent candidacies and then discuss dis- tinctions between majority and minority electoral posi- tions which might lead to behavioral differences among the party incumbents. A. Governor Along with a United States Senator, the guber- natorial candidate leads his party's ticket in the state. It was shown in Chapter 2 that an incumbent governor is more likely to win an election than a non-incumbent, and more important, provides powerful coattails for the state legislators of his party, eSpecially in non-presidential years. Thus, when an incumbent governor seeks reelection the party has a greater opportunity to hold onto the state's highest elective office and secure a maximum num- ber of seats in the legislature. The rewards for elec- toral success at the gubernatorial level produce opportu- nities and resources for additional later successes. The ability of a party to enact a systematic program of party bills (including the pet bills of in- dividual legislators) during a legislative session is 56 greatly enhanced by the incumbent governor seeking re- election-ethey are more likely to have a friendly face in the governor's mansion and a majority in the legislature. Those legislators who expect to be in positions of legis- lative leadership need the gubernatorial coattails to assure a majority for their party in the legislature in order to gain or retain the leadership positions. In ad- dition to providing coattails on election day, the deci- sion by the governor to seek reelection may convince some legislators to seek reelection when under other conditions they would not, and encourage attractive non-incumbent candidates to fill posts in legislative districts where the party has no incumbent or in which the incumbent does not seek reelection. Finally, the reelection of an in- cumbent governor insures the continued flow of patronage from the governor's office to those lower on the political ladder. All these direct and potential benefits occurring to the party (i.e., other officeholders and candidates) from an incumbent candidacy at the top of the ticket will lead to pressure on the incumbent governor to seek reelec- tion at a time of increasing or intense inter-party com- petition. B. Legislature An incumbent governor will normally attempt to convince state legislators of his party to seek reelec— tion, especially if the governor is running for more than 111:». "mu .1: J i {J:.’!! 57 his second term. There is some evidence that an executive officeholder, after a certain point in time, becomes more 3 As pointed vulnerable the longer that he holds office. out in the previous chapter, the executive cannot veil his actions in the group atmosphere of legislative politics, and it is easier for the individual voter to assign governmental responsibility, and orient his voting behav— ior accordingly, with respect to a Single executive than to a large group of legislators. In order to aid his cam- paign, the incumbent governor will work toward organiza- tional strength at the legislative district level, where more direct voter contact occurs. At that local level, incumbent state legislators would have at least the rudi- ments of a nuclear organization. A non-incumbent gubernatorial candidate will need both the organizational strength at the local level and the publicity and endorsements provided by elected offi- cial lower on the ticket. Either an incumbent or non- incumbent gubernatorial candidate will urge state legisla- tive incumbents of his party to such reelection in order to assure a majority in the legislative body, thereby paving the way for the governor to enact a legislative program that will enhance his image for future career goals. As the state legislature becomes more competitive, the value of incumbents as successful candidates increases markedly. Minority legislators will retain the hope of 58 becoming part of the majority and the majority legislators will fear losing their dominant position. Both sets of attitudes will lead to pressure on incumbents to seek reelection. C. Majority - Minority Differences Although there will be pressure on both majority and minority party incumbents to seek reelection at times of intense intereparty competition, the majority party will probably be more successful in its efforts. It must be noted that increasing competition is perceived quite differently by the two parties--for the minority it means the opening up of the political opportunity structure and possible control of key offices, while for the majority it means the blocking of political opportunity and possible loss of control of the office of governor and of the state legislature. From these differing perSpectives of increasing competition might develOp different attitudes concerning the decision to seek reelection. A major assumption of Schlesinger's ambition theory is that a politician's ambi- tions are formed by the opportunities confronting him.4 As statewide electoral competition intensifies, i.e., as available electoral offices are more closely contested by the minority party, the structure of opportunity for the minority party politician opens up and fosters political ambitions. On the other hand, the political ambitions of "( e A 59 the majority party politicians are somewhat tempered as his opportunity for advancement is blocked. The minority party is confronted with the task of convincing incumbents to seek reelection just when politi- cal ambitions are being fostered. Party leaders will be hard pressed to control the officeholders and parcel out nominations in a way that the party can present its strongest challenge. As Lester Seligman has pointed out, as inter—party competition increases, the political mar- ketplace opens up and party leaders have less control over nominations.5 Inter-party competition over the limited number of nominations that mean career advancement might lead to a breakdown in established patterns of intra-party decision-making, including the question of nominations. The majority party, however, does not have to con- tend with ambitious politicians jumping on the bandwagon. As inter-party competition increases, the opportunity structure for the party and development of political ambi— tions among its officeholders are both blunted, and the holding of offices by running incumbents for reelection would be perceived as an appropriate strategy by both the party as a whole and individual officeholders. It is when competition is decreasing, i.e., when the opportunity structure of the majority party is opening up, that majority politicians would seek career advancement nomina- tions. In sum, we would expect the hypothesis under examination to be a more accurate description of majority 60 party behavior, and while the hypothesis will be tested with data encompassing all incumbents, the majority - minority distinction must be kept in mind as a possible control variable. The principal test of the hypothesis is to examine the relationship between changes in the level of competi- tion and the decisions of incumbents to seek reelection from one election to the next within individual states. However, an overview of all the states, comparing them on the two variables, will give us the broad outline of the relationship. The data were analyzed across states; the mean level of competition for legislative seats as measured by the adapted Meltz-IL. index and average percentage of legislative incumbents seeking reelection over the entire time period under study were calculated and paired for each state. These values, shown in Table 3.1, indicate a positive relationship between level of competition and the number of incumbents seeking reelection. The data were not collected with an inter-state comparison in mind, and the sample of only four states is extremely small. How- ever, the time period under examination (c. 1900-1950) and the number of elections within that period are substan- tial. The four states scale perfectly on competition and percentage of incumbents seeking reelection, and the r value between the variables is .755. In the rather gross figures of Table 3.1, there is some evidence that higher levels of competition leads to a large percentage of 61 Table 3.1 Mean Level of Competition and Mean Percentage of Incumbents Running--All Years Mean Level of Mean Percentage of Competition Incumbents Running Conn. .592 36.9 Iowa .652 59.6 Kansas .648 59.2 Ohio .769 62.7 legislative incumbents seeking reelection. Now we must focus our analysis on the individual states and examine the variable on an election to election basis. The individual state analyses use the same mea- surement procedures employed in Chapter 2, with one addi- tional measure of competition. The now dependent vari- able of incumbency decision to seek reelection remains the same; the decisions of state legislators to seek reelec- tion is measured by the percentage of legislative incum- bents running, while the decision of the incumbent governor is assigned a value of 1 for seeking reelection and 0 for not doing so. The independent variable of intereparty competition is measured for legislative seats by the Meltz..fL~index, as adapted, and for governor by the same simple percentage difference.6 Moreover, an additional measure of competition is employed in testing this propo- sition. Several points made in the theoretical discus- sion of the hypothesis rest upon competition as measured 62 by the number of seats in the legislature won by the two major parties. Competition within the legislative body itself, as distinct from the electoral competition leading to the makeup of the body, is defined as the percentage difference between the number of seats held by the two major parties. As in the index of competition for the office of governor, the percentage difference is sub— tracted from 1, resulting in an index of competition that ranges from .001 (lowest level of competition) to .999 (perfect competition). In order to test the degree of relationship be- tween the three measures of competition and the percentage of incumbent legislators seeking reelection, Pearson product-moment correlations were used once again. As noted above, pairs of values for contiguous electoral years were compared; competition at time t was paired with incumbency decision to seek reelection at time t + 1. The result is a set of paired values appropriate for correla- tion analysis. Testing/the*relationship between the three measures of ébmpetition and the decision of the incumbent governor i olved the comparison of interval scale data with a sim le dichotomous variable. Due to the dichoto- mous variabl , the Pearson r was useless in this instance. A statistical technique which does provide a measure of association between a continuous variable and a dichoto- mous variable is the point biserial correlation. This 63 statistic "...is a product-moment correlation...and can always be interpreted as a measure of the degree to which the continuous variable differentiates, or discriminates, between the two categories of the dichotomous variable”.7 The paired sets of values were set up in the same way as in dealing with the legislature; competition at time t was paired with incumbency decision by the governor at time t + 1. Table 3.2 presents the data with respect to the level of competition and the decision of incumbent legis- lators to seek reelection. Table 3.2 Inter-Party Competition and Incumbency Decision to Seek Reelection--State Legislators* Level of Competition 8‘ 'H for for within g Governor Legislative Seats Legislature ; Conn. .124 -.086 .059 (N=23) 5 Iowa -.088 .126 -.045 (N=19) 3:, Kansas .060 .030 -.003 (N=19) 5 Ohio -.032 -.315 -.181 (N=20) a: * Pearson Product-Moment Correlations The relationships are weak; only one t value reaches the level of 2, and that is opposite in direction to that predicted in the hypothesis. The other correla- tions are quite low; eight of the twelve values presented in the table are at the 0 level. Because the values are 64 so low, we need not consider the direction of the rela- tionships. The relationship between competition and the de- cision of incumbent governors to seek reelection is not much stronger. (See Table 3.3.) Table 3.3 Inter-Party Competition and Incumbency Deci- sion to Seek Reelection--Office of Governor* Level of Competition -§ ‘6 for for within .21} Governor Legislative Seats Legislature gig Conn. .441 .381 .424 (N=23) 5:33 Iowa -.141 -.110 -.029 (N=19) gg Kansas .217 .144 .121 (N=19) 53 Ohio .001 .194 .283 (N220) * Point-Biserial Correlations Except for Connecticut, the correlations show little if any relationship between the variables. Connecticut, at least in comparison to the other states, shows a rela- tively strong relationship in the predicted direction. As was pointed out earlier in this chapter, low level correlations are difficult to interpret. The first impulse is to accept the low correlations at face value and reject the hypothesis; certainly, under most condi- tions, the correlation values in Table 3.2 and 3.3 would be justification for abandoning the analysis. However, two factors may be operating that are disguising or 65 masking the relationship—-1. as in the first hypothesis, national elections may be influencing elections in the several states, or 2. as mentioned above, there may be opposite behavioral tendencies among certain types of legislators that dnguise any relationship. In order to control for the effect of national elections on elections in the states, the correlations were computed for non-presidential election years only, just as we did in the analysis of the data with reSpect to the first hypothesis. By isolating state from national elections, the relationship may come more clearly into focus. Second, one of the cleavages on which opposite tendencies of legislators to seek reelection may divide is the majority-minority Split, and the data were analyzed with the dependent variable controlled for minority ver- sus majority party. Table 3.4 presents the data with respect to the level of inter-party competition and decisions of incum‘ bents to seek reelection, considered for non-presidential years. 66 Table 3.4 Inter-Party Competition and Incumbency Deci- sion to Seek Reelection, for Non-Presidential Years-«State Legislators* Level of Competition o c '2 for for within 3 Governor Legislative Seats Legislature m m Conn. -.124 -.438 -.235 (N=12) 4.) F; Iowa .067 .249 .051 (N=10) g Kansas ‘0015 ‘0546 “0632 (N210) s H Ohio “’0143 “‘0553 ‘0424 (N=10) I“ B: * Pearson Product-Moment Correlation Because the paired sets of values in the correlations are for contiguous electoral years, the level of competition is for non-presidential years, while the decision to seek reelection is for presidential years. The independent, rather than the dependent, variable was isolated for non— presidential years for two reasons. It was assumed first that competition within a state could be better read by politicians in nonepresidential years, and second, that a politician would act on factual electoral information (i.e., the previous election) before he would base his behavior on a hypothetical future (i.e., who the presiden- tial candidate might be and how he might run within the politician's own constituency). The relationship for non-presidential years is much stronger and has implications for supporting or dis- proving the hypothesis. Competition for governor is once 67 again a poor predictor of whether incumbent legislators will seek reelection. Evidently, a gubernatorial candi- date facing a competitive challenge can do little about convincing state legislators to join him on the ticket. However, competition for legislative seats and within the elected legislature Show a much stronger relationship in this Table. Excepting Iowa, there is a definite pattern of incumbency decision pp; to seek reelection when compe— tition stiffens. According to these figures, neither party can induce or coerce incumbent state legislators to run again if they are faced with a serious challenge. Legislators, perceiving this challenge, may decide to forego competitive politics. State legislators especially have little incentive to expend much time, effort, or money on a relatively unattractive political position, and would not readily respond to pleas for another candi— dacy in order to help others on the ticket. The data for the office of governor (Table 3.5) are not as clear, but in part this may relate to the na- ture of measurement, i.e., individual vs. aggregate be- havior and the size of the sample. 68 Table 3.5 Inter-Party Competition and Incumbency Deci- sion to Seek Reelection, for Non-Presidential Election‘Years--Office of Governor* Lgvel of Competition 5 x: for for within 3.3. Governor ngislative Seats Legislature E130 Conn. .462 .066 .109 (N=12) 31$ Iowa -.347 -.244 —.059 (N=10) §§ Kansas -.038 -.494 -.482 (N=10) g: Ohio .259 .156 .037 (N=10) o e * Point-Biserial Correlation Connecticut and Ohio both show an extremely weak relationship in the predicted direction, while Iowa and Kansas Show a somewhat stronger relationship Opposite to that predicted. As mentioned earlier in this chapter, the difficulty in finding any strong relationship or definite pattern with respect to the behavior of governors lies partially in the size of the sample. The number of legislative races included in the data is more than 1,000 times as great as the number of gubernatorial elections. There is a greater likelihood of personal idiosyncrasies and extraneous variables affecting the data on guberna- torial behavior than that concerned with legislative be- havior. In short, in a particular electoral year we are examining legislative behavior in the aggregate (percent- age of the elected body seeking reelection), while fo- cusing on individual behavior in relation to the office 69 of governor (the decision of one man to seek reelection). In order to assess the total impact of competition upon the incumbent's decision to seek reelection, rather than competition for governor, within the legislature, and for legislative seats considered separately, multiple and partial correlations were employed. Multiple correlations were calculated both for all the data and for non- presidential election years only. One of the problems in using multiple correlation is that of multicollinearity, i.e., working with independent variables that are highly correlated with each other. This problem was immediately evident in the relationship between competition for legis- lative seats and competition within the legislature it- self. Competition as measured by the party split in the legislature was employed to cover the unlikely situation of a great number of highly competitive legislative races all won by the same party. A hypothetical example of such a situation would be Republicans capturing all 125 seats in the Kansas legislature with each candidate winning by a 50.l% to 49.9% margin. 0n the other hand, the situation might have been exactly reversed; the two parties might have equally Split the total number of legislative seats, but each winning candidate faced no opposition in the general election. Nothing even remotely approaching this Situation occurred and the two measures of competition were highly correlated, ranging from a low of .880 in Ohio to a high of .936 in Connecticut. It was assumed that the 70 two measures were in essence measuring the same thing; the legislative Split index was dropped, and the multiple cor- relations were run with only two independent variables, 1) competition for governor, and 2) competition for legisla- tive seats. Next, the possibility of high correlations between the two remaining measures of competition had to be ex- plored. Before the multiple correlations between the two measures of competition and incumbency decision to seek reelection were calculated, partial correlations were cal- culated for each independent variable, controlling for the other measure of competition. These partials were cal- culated for all years and for non-presidential election years only. Except for the state of Connecticut, where considered only for non-presidential election years, the relationship between each independent variable and the dependent variable was not substantially altered by con- trolling for the second independent variable. The result for non-presidential election years in Connecticut was expected, as the correlation between the two independent variables in that state for non-presidential years was .717. For the other three states, the correlation between competition for governor and competition for legislative seats ranged from a low of .119 (Ohio for non-presidential years) to a high of .661 (Iowa across all years). Except for the particular case of Connecticut in non-presidential years (and possibly Iowa), the possibility of 71 multicollinearity producing unreliable multiple correla— tions is minimal. Multiple correlations can tell the researcher nothing about the direction of relationships. The cor- relation always takes on a positive value and is inter- preted as indicating how much of the variance in the de- pendent variable can be explained by the multiple impact of more than one independent variable. The percentage of the variance explained is equal to the square of the mul- tiple correlation (R2). The dual impact of the two in- dependent variables still explained less than 10% of the variance for the state legislature; in only one state, Connecticut, could more than 10% of the variance in the dependent variable be explained by the level of competi- tion. Multiple R's were also calculated for non~ presidential election years only. These values indicated a somewhat stronger relationship, explaining up to 34% of the variance in state legislators' decisions to seek re- election. The multiple R's for the office of governor, calculated only for non-presidential years, could still explain only about 10% of the variance. At the beginning of this chapter, we mentioned the possibility of contradictory hypotheses. The very similar results in the analysis of the data with respect to the hypotheses is a strong indication that the two are in fact measuring the same thing, and negative findings 72 for the second hypothesis are almost inevitable. How- ever, there are two more controls which must be put on the data. As mentioned above, the dependent variable will be divided as to majority and minority party, but first we will examine the data for increases in competi- tion rather than static levels of a low competition. The test of the second hypothesis employed a one election (two year) time lag for competition to have an effect on the behavior of incumbents, which was an at- tempt to overcome the tendency toward contradictory propositions. Because there is a tendency in American politics for competition to shift slowly over relatively long periods of time, a one election time lag may not be sufficient to measure the effect of competition upon in- cumbency behavior. Therefore, the data were examined for years in which there was an increase in competition over the previous election; and compared to years in which a decrease in competition was recorded. Increased competi- tion from time t to time t + 1 was predicted to have an effect on the number of incumbents at time t + 2.8 For example, if competition in 1944 showed an increase from the previous election (1942), a relatively large per- centage of incumbents should run in 1946. Increasing competition, rather than static levels of high or low competition, seems to be a definite factor in the decision of incumbent governors to seek reelection. Table 3.6 shows how often incumbent governors choose to 73 run when competition for the office of governor itself, legislative seats, or both, was either increasing or de- creasing. Table 3.6 Decision of Incumbent Governor to Seek Reelec— tion as a Function of Changes in the Level of Competition Level of Competition Increasing for for for Governor Legislative Seats Both 3 % N* % N % N c: 3; Conn. 53.9 (13) 54.5 (11) 57.1 (7) :> 3 Iowa 57.1 (7) 57.1 (7) 80.0 (5) fig Kansas 83.3 (6) 83.3 (6) 100.0 (4) .04) ggohio 87.5 (8) 87.8 (9) 85.7 (7) OH GO H0 . . . +,“ Level of Competition DecreaSing om fi§0onm 44.4 (9) 45.4 (11) 40.0 (5) O 5'” Iowa 50.0 (10) 50.0 (10) 60.0 (5) 4.) m Kansas 36.4 (11) 36.4 (11) 33.3 (9) O 32 Ohio 50.0 (10) 22.9 (9) 42.9 (7) * N = total number of cases falling in that category This is without doubt a rather crude attempt to measure in- creasing competition. However, even this limited attempt demonstrates that politicians respond immediately to mounting opposition at the gubernatorial level. In all four states, incumbent governors more often choose to run when competition is increasing, and in all states but 74 Ohio, run most often when competition for both legislae tive seats and office of governor is increasing. How- ever, the data do not support the hypothesis that incum- bent legislators, of either the majority or minority party, respond as quickly. In all four states under analysis, there was never more than an 8% difference in either party between the number of incumbents running when competition was increasing or decreasing as measured for legislative seats, governor, or both. But this finding is not surprising. As was pointed out in Chapter 2, the office of governor is the highest elective office in the state and crucial for the coattail effect; poli- ticians would respond more readily to threats upon this office than to the larger, relatively less important group of state legislators. However, based on our discussion of the differ- ences in the electoral positions of minority and majOrity party politicians at a time of intense competition, we would expect to find some relationship between competi- tion and incumbency decision to seek reelection for the majority party legislators. The relationship between static levels of competition at time t and the percentage of incumbents seeking reelection at time t + l, is pre- sented in Table 3.7. 75 Table 3.7 Competition for Legislative Seats and Incum- bency Decision to Seek.Reelection, Controlled for Party--State Legislators* Competition for Legislative Seats m $5 Conn. Iowa Kansas Ohio 35 Maj. Party -.111 .319 .044 .289 H (8% Min. Party ‘0162 0112 “0115 0107 34.3 * Pearson Product-Moment Correlations The Table shows consistent positive results in only two states--Iowa and Ohio. In these two states, at least for the majority party, there is weak evidence that a high level of competition leads to a greater number of incum— bents running. The evidence becomes more convincing when all four states are contrasted as to the average level of competition for the entire time period under study (see Table 3.1); Iowa and Ohio have consistently been more competitive in state legislative races than either Connec- ticut or Kansas. The data analysis with respect to the second hypo- thesis leaves us without firm conclusions. The negative findings when the data were examined only for non- presidential years are questionable because of methodo— logical difficulties. The weak positive results when the data were examined for increasing competition and minority-majority party differences, coupled with the inter-state data from Table 3.1, are some indication that 76 the hypothesis is at least partially supported. If we accept the premise that the negative findings are due to methodology and not substantive relationships, we can conclude that the evidence leans very slightly toward re- jection of the null hypothesis. In the next Chapter, we will examine the relation- ship over time of the two variables of incumbency and competition to see if our model of intersecting sine curves has any validity whatever. FOOTNOTES 1For discussion of Type I and Type II errors, see George Ferguson, Statistical Analysis ip Psychology and Education (New York: McGraw~Hill, 1971). 2Joseph Schlesinger, "Political Party Organiza- tion” in James G. March (ed.), Handbook.g§ Organization (Chicago: Rand McNally, 1965), p. 789. 3This point may seem contradictory after demon— strating how valuable an incumbent governor is for his coattail effect. However, there is some evidence that an incumbent governor's marginal benefits to those lower on the ticket decreases each time he runs as an incumbent. Unfortunately, there were not enough instances of ex- tended tenure in the office of governor to systematically test this hypothesis. 4See Joseph Schlesinger, Ambition gpg Politics (Chicago: Rand McNally, 1966). 5Lester Seligman, "Political Recruitment and Party Structure," American Political Science Review, Vol. 55, 1961, p. 77. 6The measure for governor is the percentage vote for the losing candidate subtracted from the percentage for the winning candidate, with that value subtracted from 1. The result is an index of competition that ranges from .001 to .999. 7Ferguson, pp. cit., pp. 2404241. 8The author would have liked to examine increases in competition over a longer period of time, but no such pattern of increasing competition appeared in the data. Competition for the legislature increased two elections in succession only 11 times in all the states, and only 9 times for the office of governor. In general, there was no pattern of increasing competition to follow. 77 CHAPTER 4 INCUMBENCY: SOME CONCLUSIONS In this final chapter we examine the status of the model suggested in Chapter 3, examine possibilities for further research concerning incumbency behavior, and dis- cuss both the normative and empirical implications of the findings with respect to the two hypotheses. I. Status of the Model The empirical evidence only weakly supports the second hypothesis of the model. The lack of empirical support may be as much a result of weak methodology as lack of a true substantive relationship; the problems mentioned in the previous chapter--the possibility of contradictory hypotheses, the alternative interpretations of a zero level correlation, differences in sample size-- may be masking a much stronger relationship than that which the data suggest. We will, however, limit our dis- cussion primarily to the empirical findings; we do not attempt to attribute more to the data than the methodo- logy, however limited, permits. It can be concluded from the data that incumbent 78 79 candidacies lower the level of competition, but the con- tention that high levels of competition necessarily lead incumbents to seek reelection is not supported. A sub- stantive problem related to this latter contention may be the researcher's choice of offices. The office of state legislator is a relatively unattractive position, and is generally considered less prestigious than the state senate, for example, which could have been the focus of this study. We examined the lower house rather than the senate precisely because it is an unattractive position. The major proposition of the study is that the level of competition is a key variable in the decision of incumbents to seek reelection; if the hypothesis could be validated for the lowly position of state legislator, it would probably hold for other, more attractive offices. On the other hand, testing the hypothesis for more pres- tigious offices severely limits our ability to generalize to more mundane positions. Furthermore, the choice of the lower house has a more pragmatic basis--the sample of officeholders is considerably larger than it is for other offices. There is some indication that the model (and specifically the second hypothesis) might be a more powerful explanatory device for the behavior of state senate incumbents. The distinction between lower and upper houses is largely in the size of the two bodies and the resulting size of individual constituencies. There is 80 no reason to believe that the electoral advantage enjoyed by incumbents, and the resulting lower level of inter- party competition which a large percentage of incumbents seeking reelection brings about, is at all different for the senate than for the lower house. Because the senate is a smaller body and each officeholder represents a larger constituency, however, the office would be per- ceived as relatively more important by the party hier- archy and the gubernatorial candidate. It follows that the pressure on incumbents to seek reelection at times of intense inter-party competition might be greater for sena- tors than for house members. The large size of the general assembly in Connecticut (279 members), for exam- ple, diminishes the importance of an individual seat within the body. If the party and the politician both view a seat in the senate as preferable to one in the house, the second hypothesis of the model might better predict political behavior in this contest. There is some available literature based upon cost-benefit analyses which provides a possible explana- tion for the negative findings with respect to the second hypothesis. Gordon Black, following James Q. Wilson, attempts to delineate the professional from the amateur politician.1 . According to Black, ”Professionalization...refers to the assimilation of the standards and values prevalent in a 2 given situation." In the context of Schlesinger's 81 ambition theory, a professional politician is one whose behavior is a function of attaining a particular office, now or in the future, i.e., a politician's behavior is a response to his office goals. Black agrees with Schlesinger, commenting, ...we suggest that the presence or absence of ambition itself is the critical intervening factor between an individual's political past and his political future and between his initial political values and those of the professional politician....3 Black suggests that the professional politician has a com- mitment at two levels: 1) commitment to the position he presently holds, and 2) commitment to seek other posi- 4 One measure of commitment to the office, ac- tions. cording to Black, is the set of investment costs asso— ciated with that office, and one of these costs is the level of intereparty competition. An incumbent state legislator, usually a part-time politician and perhaps not 'completely "professionalized," may refuse to pay the in- vestment costs of retaining his position when faced with a serious challenge. Frolich, Oppenheim, and Young also approach the decision to seek office (election or reelection) from a 5 cost-benefit approach. In a tightly argued theoretical work, they provide the following definition of a politician. Any individual who acts to supply a collective good without providing all of the resources himself we will call a political leader or political entrepreneur. Such an individual will only find this role valuable when the total resources he can collect as a leader exceeg his costs, thereby producing a leader's sur- plus. 82 The authors point out that one of the costs borne by a political leader is that of providing a "collective or- ganization," in Schlesinger's terms forming and maintain- ing a nuclear unit that will assure election of a can- didate. When a state legislator faces mounting political opposition, the costs of maintaining that collective or- ganization may be greater than the resources he can collect as a leader, thereby producing a Shortage rather than a surplus. Under such conditions, the political entrepreneur will prefer to leave the marketplace. Alfred O. Hirschman Specifically talks about the ”exit option" available to members of varied organiza- tions.7 While Hirschman's analysis of political organiza- tion is mainly devoted to the voter option of choosing among parties, elements of his analysis are of value in explaining incumbency behavior. Hirschman argues that the response to decline in organizations may take several forms, including ”voice" (verbal protest of some sort) or "exit” (leaving the organization). The exit option be- comes less likely the stronger the commitment to the or- ganization. As already pointed out, the position of state legislator is usually low-paying, part-time and relatively non-prestigious, hardly qualities that lead to loyalty to or the ability to be heard in the party. Hirschman's analysis would lead us to conclude that the state legis- lator would choose the "exit option" when faced with serious electoral opposition. 83 II. Suggestions for Future Research In addition to examining the variables of incum- bency and competition in differing contexts, research on the impact of incumbency upon electoral behavior and party organization should proceed in three important areas. Two of these are theoretical and one is methodological. In Chapter 2, we describe and combine two explana- tions for incumbency advantage--the recognition or visibility theory posited by Barbara Hinckley and the ex- perience or seniority explanation offered by Joseph Schlesinger.8 The recognition theory assumes low voter interest and information, while the Schlesinger explana- tion assumes a more active, informed voter. There is need for more research to measure the relative value of these two explanatory propositions. Schlesinger argues that seniority builds politi- cal experience and expertise that are invaluable in elec- tion campaigns. To be successful, however, an incumbent should not be SO'ViSible as to be quickly identified and held responsible for political mistakes; this particular disadvantage plagues highly visible executive offices such as mayor and governor. On the other hand, the visi- bility explanation credits incumbency success to the visibility of one or more terms in office. We may test the visibility explanation by ranking several offices with respect to the level of visibility and by then measuring the rate of incumbency success. .I—sqh—r— 84 The ranking could be done by surveying voters, tapping their information level concerning particular offices and/or officeholders. If the recognition theory is valid, the two rankings of visibility and incumbency success should show a positive correlation. In this study, we have taken a preliminary step in this direction. The rate of incumbency success for the offices of governor and state legislator were compared with the rate for candidates for the U. S. House of Representatives for the same years in the same states.9 The offices were arbitrarily ranked by the author from more to less visible in the following order: governor, congressman, state legislator. No clear pattern emerged, but initial results show that incumbency success may in fact be inversely related to visibility. The greater success rates of both congressman and state legislators in relation to governors in winning reelection is taken as preliminary evidence supporting the Schlesinger ex- planation. Such a study leads to another, rather complex, empirical problem that plagues all research on incumbency. How much of incumbent's performance at the polls can be attributed to party identification rather than the incum- bency of the candidate? In short, how can one separate party from person? A first step in this direction is made in Chapter 2 when the success rate of incumbents is com— pared to the success rate of non-incumbent candidates for 85 the same office over a specified period of time. It can be argued that the difference between the rates is directly attributable to the variable of incumbency. One way to further isolate the importance of in— cumbent candidacies is to more carefully control for the effects of party and incumbency. Further, we must assume that the effect of other variables is random. The inves- tigator would first chart the rate of party turnover for an office when no incumbent was running, preferably over an extended time period. This calculation provides a measure of the base strength of the two (or more) major parties in the political system. Then we would calculate the rate at which the parties held onto the office when they ran incumbents with varying lengths of tenure. .A comparison of the figures would measure the relative ad- vantage of incumbents in general, and indicate the strength of more experienced officeholders for a particu- lar office. It would also partially indicate whether in— cumbents are more valuable for the majority or minority party. Furthermore, by ranking incumbents as to length of tenure, a test of Schlesinger's seniority thesis would result. Research on the explanation of incumbents' behav- ior (as well as other areas) would be greatly stimulated by more work on the measurement of competition. This is a methodological concern. The construct of competition is important in maintaining democracy because democracy 86 requires a healthy opposition party. It is also a power- ful independent variable in empirical theory. There is an urgent need for a measure of change in levels of com- petition in order to test hypotheses which incorporate the variables of increasing and decreasing competition. The crucial problem is the assigning of weights to the same percentage change occurring at varying distances from the critical 50% + 1 minimum winning coalition. The problem becomes even more complex when one moves beyond a simple two party dichotomy..J\_. was adapted for this study to apply to a three party situation whereby two parties are considered at the same time, but even this static measure of competition has yet to be fully adapted to the three or more party system. III. Implications of the Findings Various findings of this study are substantively, theoretically, and normatively significant. We examine each area separately and indicate how this study has con- tributed to each. The most important, substantive finding of the study concerns the relationship between the office of governor and that of state legislator. We have shown that the coattail effect, demonstrated by such people as Press and Cummings at the national level, operates about as strongly at the state level.10 An incumbent governor, running at the head of a ticket, is a great bonus to the 87 state legislative candidates of his party. The analagous relationship of governor and legislator at the state level to president and congressman at the national level was most immediately evident in the non-presidential election years, i.e., when the state electoral systems are not directly influenced by national political candidates and issues. In these years, each state operates on a fairly independent basis, and the long coattails of an incumbent governor are evident in every state we analyzed. The hypotheses in this study were derived directly from.Schlesinger's ambition theory and the derivative theory of political party organization. In short, the theoretical basis of the workzests primarily on Schlesinger's formulations. The evidence which supports the first hypothesis further indicates the value of Schlesinger's theories for generating hypotheses and pro’ viding explanation for important areas of political behav- ior. Given that the decision to enter a political race has been made, it is clear that we have substantiated the basic tenet of his party organization theory, i.e., that cooperation between nuclear units is dependent upon the level of competition within the districts of those units, for the offices of state legislator and governor. Co- operation between units prior to the decision to seek election (or in this case reelection) was not evident; cooperative behavior in the form of incumbents seeking reelection at times of intense inter-party competition was 88 not observed. To repeat, the fault may lie not in the theory, but in the method. Furthermore, Schlesinger's explanation of incumbency advantage--experience and a low profile--seems to have greater validity than the more par- simonious visibility hypothesis. To summarize, this study demonstrates the value of using Schlesinger's formulations in further research on party and electoral politics. Another aSpect of the work that has some sig- nificance is the use of competition as a dependent vari- able. As mentioned previously, competition has been used in the study of American politics as a powerful independent variable in empirical theory, and has been identified as a key element of democracy (the existence of a viable oppo- sition) in normative theory. While being recognized by both normative and empirical theorists as an important variable, there has been little empirical work done on defining the conditions under which competition will flourish. This study has indicated that one set of condi- tions, incumbents seeking reelection, sharply reduces the level of party competition. This finding has important implications for contemporary electoral politics in America. The most strongly supported conclusion in this study is that, ceteris paulus, an incumbent candidate for governor or state legislator will defeat a non-incumbent candidate most of the time. If two opposing candidates conduct campaigns for the state legislature with roughly 89 equal resources at their disposal, and no unique handicaps evident for either candidate, the incumbent has between a .8 and .9 probability of reelection. If he wants to win, a challenger to an incumbent must find a way to overcome the advantage of incumbency enjoyed by his opponent. A traditional answer has been to outspend the incumbent, with the result that money, especially for mass media advertising, has become an important variable in deter- mining the outcome of American elections. Legislation has been introduced in Congress that would limit the amount of money spent on political cam- paigns.11 With the cost of electioneering so high, it is argued that the costs of running for office excludes all but the rich or makes candidates too dependent upon large contributors to the campaign. However, this study indi- cates that any attempt to equalize the resources of the candidates will work directly for the advantage of the in- cumbent. Incumbents already win a convincing number of elections and to place a limit on the amount of money a challenger may Spend to unseat an incumbent only increases the probability of an incumbent victory. We make no value judgment on the desirability of incumbency advantage but feel that decision-makers should be cognizant of this factor when considering the issue of campaign spending. IV. Conclusion The main hypothesis of this study was not 9O validated, but as described above, we believe this work has made contributions to the discipline along several dimensions and has opened possibilities for further re— search. We believe the study of incumbency behavior has normative importance for the politics of America. Writing 35 years ago, Charles Hyneman used empirical methods attempting to solve the normative problem of high turnover rates in state legislatures, preventing the emergence of a stable legislative leadership. This study has indicated that incumbents have such electoral strength that the normative problem may now be reversed--lack of legisla- tive turnover may lead to a stagnant leadership. We feel the normative and empirical questions raised in this work are important enough that the effort expended in producing this study was well directed. FOOTNOTES 1Gordon Black, "A Theory of Professionalization in Politics," American Political S ience Review V. LXIV, No. 3, p. 865; Also see James Q. Wilson, The Amateur Democrat: Club Politics ip Three Cities Tahicago: The University of Chicago Press, 1962). 2Ibid., p. 865. 31bid., p. 867. 4Black's discussion of different levels of com- mitment closely parallels Schlesinger's ideas on varying levels of ambition. 5Norman Frolich, Joe A. Oppenheimer, Oran R. Young, Political Leadershi and Collective Goods (Prince- ton, N.J.: Princeton University Press, 1971). 61bid., p. 6. 7Albert 0. Hirschman, Exit, Voice, and Loyalty: Responses pg Decline ip Firms, 0 anizations, and States (Cambridge: Harvard University Press, 1976. 8Barbara Hinckley, "Incumbency and the Presiden- tial Vote in Senate Elections: Defining Parameters of Sub-Presidential Voting," American Political Science Review, V. LXIV, No. 3, p. 836, and Joseph Schlesinger, Ambition.gpg Politics (Chicago: Rand McNally, 1966). 9The data on Congressional races was provided by Professor Neil Romans, Department of Political Science, Providence College. loCharles Press, "Voting Statistics and Presi- dential Coattails," American Political Science Review (December, 1958), p. 1041, and "Presidential Coattails and Party Cohesion," Midwest Journal pf Political Science (November, 1964), p. 320; Milton Cummings, Congressmen.gpg Egg Electorate, Elections pg; pygmy. g. House pf Represen- tatives and President, 1920-1964 (New York: Free Press, 1966). 91 92 11 See Time M - Not Ceiling." ___.. . aY 17, 1971, "Campaign Cash: Floor, BIBLIOGRAPHY BIBLIOGRAPHY Black, Gordon. "A Theory of Professionalization in Politics,” American Political Science Review, V. LXIV, No. 3, p. 865. Browning, Rufus. Businessmen in Politics: Motivation and Circumstances in the Rise“ to Power. Yale University: unpublished Ph. D. dissertation, 1960. . ”The Interaction of Personality and Political System in Decisions to Run for Office: Some Data and a Simulation Technique,“ Jgurnal pf Social Issues, V. 24, 1968, p. 93. . "Hypotheses About Political Party Recruit- ment: A Partially Data Based Computer Simulation," in.William O. Coplin, Simulation in the Study of Politics. Chicago: Markham Publishing Co., 1968. Browning, Rufus, and Herbert Jacob. “Power Motivation and Political Personality,“ Public Opinion Quarterly, V 28, 1964, p. 75. Campbell, Angus and Warren Miller. ”The Motivational Basis of Straight and Split Ticket Voting," American Political Science Review, V. LI, No. 2, p. 293. Cummings, Milton. Congressmen and the Electorate, Elections for the U. 8. House of Representatives and President, 1920- 1964. New York: The Free Press, 1966. Dahl, Robert. ‘Who Governs? New Haven: Yale University Press, 1961. Downs, Anthony. Ag Economic Theory g; Democracy. New York: Harper and Row, 1957. Edelman, Murray. The Symbolic Uses g: Pglitics. Chicago: University of Illinois Press, 1964. Ferguson, George. Statistical Analysis ip Psychology and Education. New York: McGraw-Hill, Inc., 1971. 93 94 Frolich, Norman, and Joe Oppenheim, Oran Young. Political Leadershi and Collective Goods. Princeton, N.J.: Princeton University Press, 1971. Froman, Lewis A., Jr. The Long Process. Boston: Little, Brown, and Co., 1967. Gosnell, Harold. Machine Politics. Chicago: University of Chicago Press, 1937. Hinckley, Barbara. “Interpreting House Mid-Term Elec- tions: Toward a Measurement of the In-Party's 'Expected' Loss of Seats," American Political Science Review, V. LXI, No. 3, p. 694. Hirschman, Albert. Exit, Voice, and Lo alt , Responses 29 Decline ip Firms, Organizations, and States. Cambridge: Harvard University Press, 1970. Hunter, Floyd. Community Power Structure. Chapel Hill: University of North Carolina Press, 1953. Hyneman, Charles. ”Tenure and Turnover of Legislative Personnel," The Annals, V. 195, 1938, p. 21. Jacob, Herbert. "Initial Recruitment of Elected Officials in the United States--A Model," Journal g; Politics, V. 24, No. 4, 1962, p. 704. Key, V. 0., Jr. Parties, Politics, and Pressure Groups. New York: Thomas Crowell Co., 1964. . American State Politics: 5p Introduction. New York: Alfred Knopf, Inc., 1966. Lasswell, Harold, gp'gl. The Comparative Study pf Elites. Palo Alto, California: Hoover Institute Studies, 1952. . Politics: Who Gets What, When, How. New York: World Publishing Co., 1958. Leuthold, David. Electioneering ;p g Democracy: Campaigns for Congress. New York: John Wiley and Sons, Inc., 1968. Matthews, Donald. The Social Background g; Political pecisionéMakers. New York: Random House, 1954. . Q. S. Senators and Their World. Chapel Hill: University of North Carolina Press, 1960. 95 Meltz, David. Competition Lnd Cohesion: A Model of Majority Party Legislativem Bargaining. University of Rochester: unpublished Ph. D. dissertation, 1970. Miller, Warren. "Presidential Coattails: A Study in Political Myth and Methodology," Public Opinion Quar- terl , Vol. 29, p. 126. Mills, C. Wright. The Power Elite. New York: Oxford University Press, 1956. Mosca, Gaetano. The Ruling Class. New York: McGraw Hill, 1939. Pareto,‘Vilfredo. The Mind and Society. New York: Dover Publications, 1935. Press, Charles. "Voting Statistics and Presidential Coat- Tails," American Political Science Review,‘V. LII, No. 4, p. 1041. . "Presidential Coattails and Party Cohesion," Midwest Journal of Political Science, November, 1964, Presthus, Robert. Men at thJ Top. New York: Oxford University Press, 1964. Prewitt, Kenneth, pp pi. ”Political Socialization and Political Roles," Public Opinion Quarterly, V. 30, 1966, p. 91. Schlesinger, Joseph. "Political Party Organization" in James G. March (ed.), Handbook pg Organization. Chicago: Rand McNally, 1965. . Ambition and Politics. Chicago: Rand McNally, 1966. Seligman, Lester. "Political Recruitment and Party Struc- ture,“ American Political Science Review, V. LV, No. l, p. 770 . ”Party Roles and Political Recruitment," Western Political Quarterly, V. 11, No. 2, p. 361. "Leader Selection in the Oregon Legislature," Western Political Quarterly, V. 12, No. 1, p. 67. “Political Parties and the Recruitment of Political Leaders,” in Lewis Edinger (ed.), Political Leadership ip Industrialized Societies. New York: John Wiley and Sons, Inc., 1967. 96 Wahlke, John, and Heinz Eulau, William Buchanan, Leroy Ferguson. The Legislative System. New York: John Wiley and Sons, Inc., 1962. Wilson, James Q. The Amateur Democrat: Club Politics ip Three Cities. Chicago: UniverSity of Chicago Press, 1962. “u in“ Budd'- "II-0‘ GENERAL REFERENCES GENERAL REFERENCES Abraham, Henry J. The Judicial Process. New York: Oxford University Press, 1962. Almond, Gabriel A. and James S. Coleman (eds.). .The Politics 9; Develo in Areas. Princeton, N.J.: Princeton University Press, 1960. Almond, Gabriel A. and G. Bingham Powell. Comparative Politics: A DevelOpmental Approach. Barber, James D. The Lawmakers. New Haven: Yale University Press, 1965. Beckett, Paul and Celeste Sunderland. "Washington State's Lawmakers: Some Personnel Factors in the Washington Legislature," Western Political Quarterly, V. X, No. 3, p. 180. Biddle, William W. The Cultivation pg Community Leaders: prFrom the Grass Roots. New York: Harper andeow, 1953. Bowman, Lewis and G. R. Boynton. "Activities and Role Definitions of Grassroots Party Officials," Journal 2: Politics, V. 28, No. 1, p. 121. Bowman, Lewis. ”Recruitment Patterns Among Local Party Officials: A Model and Some Preliminary Findings in Selected Locales,” American Political Science Review, V0 IDCCI, N00 3, p0 1530 Corson, John H. and R. Shale Paul. 1:12p Near 2133 T__op: Filling Ke Posts ip Ape Federal Service. Baltimore: John Hopkins Press, Committee for Economic Develop- ment Supplementary Paper NO. 20, 1966. Eldersveld, Samuel. Political Parties: A Behavioral Analysis. Chicago: Rand McNally, 1964. Eulau, Heinz and John D. Sprague. Layyers l2 Politics: A Study i2 Professional Convergence. Indianapolis: The Bobbs-Merrill Co. Inc., 1964. 97 98 Fiellin, Alan. “Recruitment and Legislative Role Concep- tions: A Conceptual Scheme and a Case Study,” Western Political Quarterly, V. 20, No. 2, p. 271. Frank, John P. Marble Palace: The Supreme Court 1p American Life. New York: Alfred A. Knopf, 1958. Freeman, G. L. and E. K. Taylor. H w pg Pick.Leaders: A Scientific Approach.pp Executive Selection. New York: Funk and Wagnalls, 1960. Gates, Janet and Charles 0. Jones. Legislative Turnover. Vermillion, South Dakota: Governmental Research Bureau, State University of South Dakota, 1957. Goodman, Jay S., and Wayne Swanson and Elmer Cornwell. ”Political Recruitment in Four Selection Systems," ‘Western Political Quarterly, V. XXIII, No. l, p. 92. Harris, Louis. "Why The Odds Are Against a Governor Becoming President,” Public Opinion uarterl ,'v. 23, NO. 3, 3610 Holt, Robert T. “Age as a Factor in the Recruitment of Communist Leadership,” American Political Science Review, V. XLVIII, No. 2, p. 486. Hyneman, Charles S. “Legislative Experience of Illinois Lawmakers,” University 2; Chicago Law Review,‘V. 3, p. 104. . ”Tenure and Turnover of the Indiana General Assembly,” American Political Sgience Review, Part I, V. XXXIII, NO. 1, p. 513 Part II, V. XXXII, N00 2, p. 311. Hyneman, Charles S. and Edmond F. Ricketts. "Tenure and Turnover of the Iowa Legislature," Iowa Law Review, V. XXIV, NO. 4, p. 6730 Hyneman, Charles S. and George W. Carey. "The Iowa Legis- lature: A General Description,” Monograph, Indiana University, 1963. Jacob, Herbert and Kenneth N. Vines (eds.). Politics 22 the American States: A Comparative Analysis. Boston: Little, Brown and Co., 1965. Jennings, M. Kent. Community Influentials. New York: The Free Press, 1964. Jewell, Malcolm E. and Samuel C. Patterson. The Legisla- tive Process in the United States. New York: Random House, 1966. 99 Keefe, William J. and Morris 5. Ogul. ThJ American Legis- lative Process: Congress Lnd thJS TSates. Englewood Cliffs, N.J.: Prentice-Hall, 1964. Keller, Suzanne. Beyond thJ Ruling Class: Strategic Elites in.Modern Soc1ety. New York: Random House, 1963. Kim, Yung C. "Political Recruitment: The Case of Japanese Refectural Assemblymen," American Political Science Review,‘V. LXI, No. 4, p. 1036. Kornberg, Allan and Norman Thomas. "The Political Sociali- zation of National Legislative Elites in the United States and Canada,” Jgurnal 2; Politics, V. 27, No. 4, p. 761. Krislov, Samuel. ThJ Supreme Court in the Political Process. New York: The MaCMillanfi Co., 1965. Lederle, John W. ”Legislative Personnel Given Careful Study,” Michigan State Bar Journal, V. XXVIII, July, 1949, p. 23. Lewis, Gordon F. "A Comparison of Some Aspects of the Backgrounds and Careers of Small Businessmen and American Business Leaders,” American.Jgurnal pg Socio- lo , V. LXV, January 1960, p. 348. Mann, Dean E. ”The Selection of Federal Political Execu- tives," American Political Science Review, V. LXVIII, No. 1, p. 81. Mann, Dean E. with James W. Doig. ThJ Assistant Secre- taries: Problems Lnd Processes of Appointment. Washington, D.C.: The Brookings Institution, 1965. Marvick, Dwaine (ed.). Political Decision-Makers. New York: The Free Press of Glencoe, 1961. Matthews, Donald R. "United States Senators and the Class Structure," Public Opinion Quarterly, V. 18, Spring, 1964’ p. 50 Menzey, Michael C. "Ambition Theory and the Office of Congressmen," Journal g; Politics, V. 24, No. 4, p. 704. Micaud, Charles A. "Organization and Leadership of the French Communist Party," World P litics, V. 4, April 1952, p. 318. 100 Miller, Warren E. and Donald Stokes. "Party Government and the Saliency of Congress," Public Opinion Quarterly, V. XXVI, Winter 1962, p. 531. McHenry, Dean E. "Legislative Personnel in California,” The Annals pg the American Academy 9; Political and Social Sciences, V. CXCV, January 1938, p. 45. Olson, Mancur. Jr. The Lo ic pg Collective Action. Cam- bridge: Harvard University Press, 1965. Patterson, Samuel C. "Characteristics of Party Leaders," Western Political Quarterly, V. 16, 1963, p. 332. Peabody, Robert L. Organizational Authorit : Superior- Subordinate Relations Ln ThrJe Public SerVice Organi- zations. New York: Atherton Press, 1964. Prewitt, Kenneth. ”Political Socialization and Leadership Selection,” ThJ Annals pg thJ American Acadegy Lf Political aLd Soc1al Sciences, V. CCCLXI, September 1965, p. 96. Ranney, Austin. Pathways Lg Parliament. New York: Alfred A. Knopf, 1961. Reid, Ira and Emily Ehle. ”Leadership Selection in Urban Locality Areas,“ Public Opinion Quarterly, V. 14, Summer 1950, p. 262. Riddick, Floyd M. and Neal A. Hollander. "Undergraduate Origin as a Factor in Elite Recruitment and Mobility: The Foreign Service--A Case Study," Western Political Quarterly, V. 19, June 1966, p. 354. Ripley, Randall B. Part y Leaders Ln thJ House Lf Represen- tatives. Washington, D.C.: The* Brookings Institu- W tion, 1967. Rosenweig, Robert N. "The Politician and the Career in Politics,” Midwest Journal 9; P litics, V. 1, 1957, p. 163. . ”Motivation and Careers in Politics," Midwest Jgurnal pgyP litics, V. III, 1959, p. 16. Schlesinger, Joseph. How The Became Governor: A Study Lf Comparative State Politics. East Lansing: Governmental Research Bureau, Michigan State Univer- sity Press, 1957. Schmidhauser, John R. "The Justices of the Supreme Court: A Collective Portrait,” Midwest Journal pg Politics, V. III, February, 1959, p. 1. 101 Seligman, Lester G. ”A Study of Leadership Selection in Oregon,”‘Western Political Quarterlx, V. 12, March 1959, p. 153. . "Elite Recruitment and Political Development," Journal 2g Politics, V. 26, August 1964, p. 617. . Leadershig‘ig‘g New Nation: Political Develogment in Israel. New‘York: Atherton Press, 1964. Snowiss, Leo M. ”Congressional Recruitment and Represen- tation,“ American Political Science Review, V. Lx, NO. 3, p0 6270 Warner, W. Lloyd and Paul‘Van Riper, Norman Martin, Orins F. Collins. The American Federal Executive: A Study of the Social and Personal Characteristics of Egg Civilian and Militarz Leaders of the United“ States Federal Government. New Haven: Yale Univers ty Press, 1963. Wildavsky, Aaron. Leadershi in a_Sma11 Town. Tatowa, N.J.: The Bedm1n1ster Press, 1964. APPENDIX APPENDIX A Description and Explanation of-fL. Measure quoted from: David B. Meltz, Competition and Cohesion: A Model‘gg Ma'orit Party Legislative Bargaining Unpublished Ph.D. dissertation, University of Rochester, 1970), pp. 55-56. 102 THE MELTZ .0. MEASURE He A's-tern? computed for every town the mean [3“] division of the two party vote for a period of ten years. He then defined the majority party as the party having the higher mean vote over time. He computed for the majority party the standard deviation.[7f47 of its vote and developed the following taxonomy: if for town i, 1. Ai - 20’i > 50% then i is a "safe" town 2. ”i - di < 50% theni is a competitive town 3. pi - 261 < 50% (”i - 6i then i is a marginal town While this index is quite ingenious and represents a sig- nificant advance over previous measures, it appears inade- quate for the purpose of this study. Sternfs index lacks sophistication because his use of one and two standard deviations as representing, in some way, meaningful sub— stantive cutoff points is really indefensible. Further- more, a continuous variable is not only substantively more interesting but, unlike Stern's taxonomy, is amenable to parametric statistical analysis. In order to transform Stern's index into a continuous variable, the following was done: 103 104 define a value where J1- -XG. = 05 1 1 solving for X d, r r = number of legislative sessions In order to satisfy the requirement the index be a measure of perceived safeness a mathematical technique which would take into account the decreasing marginal utility for X as X increases...is the logarithmic trans- formation. Define a new value, say 5 (eta) as log X....In order to satisfy our other requirements that the proposed index lies between zero and one, we perform another trans- formation. Define.Jl. = £3; or .n— = £293§il. Mathe— matically, this transformation insures that for all values of x between .1 and 10.43. lies between zero and one. Finally, in order to create an index of competition rather than one of safeness, we define the value 41- (omega) which is equal to 1-5. JL.=(1-(E29_2<:_1.)) x=31__-_-__._§ 2 r 63 r = number of years considered. Thus.JL. is continuous, ranges between zero and one, and explicitly measures perceived competitiveness....It is 105 worth noting that.11. ...can be used as a measure of two party competition for any voting body, be it electoral, legislative, committee, and so forth. 4 ”71111111711111"!£1158