.vx/Sz.‘ . "WK" ’-“' i ' ' 7'30'21'11813...‘ ' ." kw'? 3‘2 ,"..i\. v\)-'. )IV‘I -.' ..,,..«.,L_ 1“ .31)" r 3'“:'.l".'\ I ' I" . . 374.731. ‘5‘", ' . If; _ l G .'I' ’ -.~.~ “A , 'rfirlfl 41,133); . 691' .3 .3, .2- ., 7. I. .2 _. . _, .‘ _w{’ J A. .. _ (Hugh‘u‘: « ‘d “I"; Gk". I 1‘). 'L‘IU 2333'"! '1 r.’ .r 'l" .llt');.. "(I :1? ".1 .‘ (2‘: ' .-'a “$1.2. .llv I "If; «if-"m? d ”r.- ”aura 1"", I w , . 52:5} ' J" ‘ ' (If. -\f¢ fish-«33:3,;- A, .- . .z , :2»; «g; 41-} ’3" Wk} :..'.v #5531 :er*“'¥£} Cw) L’V‘H‘y "‘V'Y‘{ ‘ '51». ”911.31%.” 2.5.15 31'1": gigg'h’ 2" ’1‘." ‘ .3”). 35% ‘5 flu.) ‘ (“k-am Wu‘fl‘i‘fi “‘26:; 752 330$ Y "’ «1;» "(1'31 4 .2' 1A / {"1 3. ~ " ‘ 4C :‘u . I ,'..? . . ,. ’-,- >43. ‘ . I. . .Xf/‘l: 6.“, .9 xx. gl/t f. ‘, ¢ - T..(‘I .‘;\"“\‘ ‘ v” "1"}; uni. l.' (‘1'? 1‘ . ‘1'." .1 . (a; A; firm "Va-"‘6 511""; a. ‘ {2'}: . W35 #21 «153.3 ' \“fii’: " . ’Qij/Zv‘, (’1‘) X ”’11 ;€‘u we) .1 '1‘” AV 3 ' "1'???”- I'I/Au. .: .. LJ‘ ‘4‘ ’7 ""l ‘1:- $3.4" 431mg.” .150 ""3 ,; 3.95353“, If“ 81 «$3.6 "'3! K’f‘k ,3‘ I f’...‘ { I':'V'“"";1)<»,%:1‘f;tv‘ "3,1 ' ”A? ‘36 3“”. l“ ' ((1. I". v“. . AAA-1%”? (194"; Hm I .lg' "‘ an" a“. \. \‘ . 1. .4 l I , )‘ I." v2" f {12%;} 0‘4; M'j‘f . ’12:": ("11 “I?" A 31",“ .5371)”. :c fit‘fiz’é ‘1 “If ',1, ' (‘y’i'guifi 9"1? 1' ‘ ' V'} ">29 . "NM-4‘- a“? .1 5.4 I i' ‘ ’ (Hf '\. “$46; 2'“. H ‘1'"; 3.? ,3 h' (”n (vii; . 5:2} "y’jjg‘f'lfliJl'Vl "(.l’ “j. 7“: gall} 1.: '3'“) ' ‘flé \;¢:‘P" Ah. {9“}, lift/a?! 533,16 ‘ 2:?31. 'Aylfl‘zjwfléjq V.“ I'm-£3} '5: or MAHny'fiVfi’fi“11312324WNWV-WA(A:A»? a”: - - 1:325} 18'“ “1’: '5' 2/fo ”A gift [Al’flf‘f‘ ' :VAS???‘ .231". 37...}ng .f'fV'ffiA . ' . v." 3x'”*"‘ix‘-r:7'..m 3‘729'. 51‘ 11"“ #1le 1,? .3_. ~ “17-. $3 "I é/J,“"“KI\"IJ “"111" ’2 '. ' {<5 I, a“ 4" I" " 1"!) (1,; All' 2‘: .4 4‘1"]. ,"mflc/ “1“" (1‘1'1'2‘ . - . «fie-33:. .- ,’y,;. 53,. _ -A, 3 1 45's - n. al' N; . r , 3:11.”: [£111. \YW - ' ‘l. ._ . 23 37797} 11111111111111.1111 111 11 11111 11111111111 11 1.113211“: | University This is to certify that the dissertation entitled CONGRESSIONAL ELECTIONS AND THE STRATEGIC BEHAVIOR OF ELITES: THE ROLE OF PARTY ORGANIZATION presented by Calvin Jay Mouw has been accepted towards fulfillment of the requirements for Ph.D. degreein Political Science a 6 ~ .. '1 ‘1 if“? LU. Uixk V Major professor MS U i: an Affirmative Action/Equal Opportunity Institution 0-12771 PLACE IN RETURN BOX to remove this checkout from your record. TO AVOID FINES return on or before date due. DATE DUE DATE DUE DATE DUE ..... v. 14 ;: Mb"? 2 J 81"? ”In: 5' in ~1 - 1 n I. 1‘ F; a I MSU Is An Affirmative Action/Equal Opportunity Institution CONGRESSIONAL ELECTIONS AND THE STRATEGIC BEHAVIOR OF ELITES: THE ROLE OF PARTY ORGANIZATION BY Calvin Jay Mouw A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Political Science 1988 ABSTRACT CONGRESSIONAL ELECTIONS AND THE STRATEGIC BEHAVIOR OF ELITES: THE ROLE OF PARTY ORGANIZATION By Calvin Jay Mouw This study focuses on the role of American political party organizations in U.S. House of Representative elections. The study takes as its theoretical base a theory of party organization developed by Joseph Schlesinger, which portrays parties as rational organizations that develop calculi based upon individual and collective decision- making. Using this theoretical approach the thesis attempts to construct a model which differs from existing literature in its ability to explain party organizational activity as a response to varying national and district level determinants. Jacobson and Kernell's ”Strategic Choice Theory” of congressional elections is criticized on the role of the party as a strategic contributor in their election analysis. Two models are developed that reflect the interactive set of determinants which allow us the test various sets of hypotheses which have been only previously subject to speculation. Results suggest that party organizations have both different determinants and resource allocation strategies which differ considerably from that offered by Jacobson and Jacobson and Kernell. ACKNOWLEDGMENTS I am indebted to the four members of my dissertation committee: David Rohde, Charles Ostrom, Joseph Schlesinger, amd Jeffrey Hill. I would especially like to thank my dissertation chair David Rohde for valuable advice and critical judgment throughout the many stages of the project. I would also like to thank him for the opportunity to develop and use various data resources. I am also very grateful to Charles Ostrom who provided me with much appreciated insight throughout the project. Finally, I would like to thank Joseph Schlesinger and Jeffrey Hill for their time and contributions. p. p. p. TABLE OF CONTENTS Introduction ................................................... CHAPTER 1 Literature Review ................................... Introduction ............................................. Review of the Decline Thesis: Political Parties and Competition in Congressional Elections .................. UGU’?’ The Decline in the Role of Party Organization Thesis .......................... Review of the Resurgence Thesis: A Response to the Decline Thesis .......................................... UCUJI> Response to the Party Organization in Decline Thesis ............................... Theoretical Basis for Party Organizational Activity in Congressional Elections ................................. A. The Strategic Choice Theory of Congressional Elections ...................... B. Parties and the Strategic Choice Hypothesis .... C. Political Action Committees and the Strategic Choice Hypothesis .................. D. Individual Candidates and the Strategic Choice Hypothesis ................................... The Party Theoretical Perspective of Congressional Elections ............................................... Summary and Conclusion ................................... iv The ”Vanishing Marginals" Thesis ............... The Increase of Split-Ticket Voting Thesis ..... The Increase in Partisan Defections Thesis ..... Response to Mayhew's Conception of Competition . Response to the Split-Ticket Voting Thesis ..... Response to the Partisan Defection Thesis ...... . 6 . 6 . 9 10 12 14 15 16 18 19 20 20 21 26 CHAPTER 2 A Theory of Party Organization ...................... 28 Introduction ............................................. 28 The Study of Political Parties: Two Theoretical Approaches .............................................. 28 A. Parties as Membership-Based Organizations ...... 28 B. Parties as Electorally-Based Organizations ..... 30 Schlesinger's Office-Seeking Model of Political Parties ....................................... 31 A. Extension of the Downsian Model ................ 31 B. The Theory of Party Organization ............... 33 Office-Seeker Strategies: The Allocation of Resources .... 35 A. Individual Office-Seeker Strategies ............ 37 B. Collective Office-Seeker Strategies ............ 39 Differentiation with Jacobson and the Strategic Choice Model ................................................. . 40 A. The Symmetric Position: Jacobson on Collective Goals and Strategies ......................... 41 B. The Non-Symmetric Position: Collective Behavior as Individual Decisions ...................... 44 C. The Evaluation of National Conditions Procedure .................................... 50 D. The Determinants of Strategic Behavior ......... 51 Conclusion ............................................... 52 CHAPTER 3 Theoretical Hypotheses .............................. 53 Introduction ............................................. 53 Parties in the Electoral Market Structure ................ S3 The Role of the Structure of Political Opportunities ..... 55 Evaluation Procedures for National Conditions ............ 56 The Symmetric Hypotheses ................................. S8 The Non-Symmetric Hypotheses ............................. 62 Conclusion ............................................... 66 CHAPTER 4 The Models of Party Organizational Behavior ......... 67 Introduction ............................................. 67 From Theory to Model ..................................... 67 V Operationalization of Constructs ......................... 70 A. The Structure of Political Opportunities ....... 70 B. The Party System ............................... 71 C. Party Collective Strategies .................... 75 The Models of Party Organizational Behavior .............. 76 A. Model I: A Three-Equation Pooled Cross-Sectional Block-Recursive System ....... 77 B. Model II: A Two-Equation Cross-Sectional Model of Party Expenditures .................. 78 C. Data-Related Issues ............................ 79 Methods of Estimation .................................... 84 A The Probit Model ............................... 84 B The Tobit Model ................................ 85 C. Instrumental Variable Techniques ............... 87 D Simultaneous Equation Estimation Techniques .... 88 CHAPTER 5 Results and Discussion .............................. 93 Introduction ............................................. 93 Presentation of Results for Model I ...................... 93 A. Model I Results ................................ 94 B. Discussion of Results .......................... 96 Presentation of Results for Model II .................... 100 A. Model II Results .............................. 101 B. Discussion of Results ......................... 107 CHAPTER 6 Conclusion ......................................... 116 Introduction ............................................ 116 Theoretical Consequences ................................ 116 A. Advantages of Party Model over Strategic Choice Model ................................ 119 B. Congressional Election Literature ............. 121 Further Research ........................................ 124 Appendix A Measurement Rules and Data Sources ................ 126 Appendix B McDonald and Moffitt's ”Uses of Tobit Analysis" ... 130 List of References ............................................ 131 vi Table Table Table Table Table Table Table Table LIST OF TABLES Probit Coefficient Estimates -- Pooled Cross—Sectional Model ................. 94 Tobit Coefficient Estimates -- Pooled Cross-Sectional Model ................. 95 Tobit Estimates -- Party Expenditures -- 1976 Election ............................... 101 Tobit Estimates -- Party Expenditures -- 1978 Election ............................... 102 Tobit Estimates -- Party Expenditures -- 1980 Election ............................... 103 Tobit Estimates -- Party Expenditures -- 1982 Election ............................... 104 Tobit Estimates -- Party Expenditures -- 1984 Election ............................... 105 Party Contributions/Expenditures by Candidate Status, 1978-1986 ................. 107 vii INTRODUCTION One of the prevailing themes of congressional election analysis is that there has been a decline in the role of the political party. Proponents of the decline thesis have focused on numerous aspects ranging from changes in partisan identification and defection to the role of party organization. In more recent literature there is evidence for a ”party in resurgence” thesis. Some of the changes taking place are in the increased power of the national organizations for the Republican and Democratic parties. The resurgence thesis as posited in the organizational mode has emphasized two major phenomena: the increase in activities that resolve around getting candidates elected, and the centralization and nationalization of the party structure. While the increased role of the parties has been documented, a coherent theoretical discourse on their organizational activities as they relate to congressional elections has been noticeably lacking. Two attempts have focused on the organizational advantage of the Republican Party over the Democratic Party and have speculated about the effects this might have on specific election years. (Jacobson, (1985-1986), for the 1982 congressional elections; and Herrnson, (1987), for the 1984 congressional elections.) The Herrnson study does not attempt to model party organizational behavior and defines ”effectiveness” only in response to a single district-level competitiveness measure. The Jacobson article attempts to incorporate an explanation of party 2 activities by using a theory of congressional elections developed by himself and Samuel Kernell. (1983) The Jacobson and Kernell theory presents how macroelectoral behavior can be derived from individual decisions on whether to run for office, which in turn reflect partisan electoral conditions. The strategic calculus of individual candidates and the contribution strategies of individuals, interest groups and parties are posited to respond to the same sets of determinants. Their theory shows how the collective strategic choices of politicians and contributors are shaped by national events and conditions; these decisions in turn determine the kinds of candidates and campaigns the voters are offered. It is the primary linkage of political parties as contributors and the relationship to individual candidates that we are concerned with. The description of events Jacobson (1985-1986) provides is not validated by the Strategic Choice theoretical orientation. This thesis proposes to study these relationships from a party theoretic base, specifically the theory of party organization by Joseph Schlesinger (1984, 1985) This theory also allows us to clarify two important considerations: first, the development of a decision calculus which takes into account the interaction of the individual and collective units in their response to varying conditions over the nomination/election time period; and second, it allows for the development of a model that includes both national political conditions and district level competitiveness measures as determinants for party organizational behavior. These two considerations are not adequately dealt with in the Jacobson and Kernell thesis. 3 The thesis focuses on three main theoretical concerns. First, a redefining of the determinants for party organizational behavior at both the individual candidate choice level and the collective contribution decision level. Second, this thesis differs from previous research on the characterization of the evaluation of national conditions and performance. Jacobson and Kernell’s Strategic Choice position offers an evaluation of current conditions hypothesis, whereas this thesis offers a prospective evaluation or ”extrapolation” hypothesis. The third concern focuses on the strategies undertaken by the two levels of party organization. Jacobson (1985-1986) and the conventional wisdom is based upon a symmetric interpretation of the ”offense” and ”defense” metaphor. This thesis in turn hypothesizes that the metaphor should in fact be defined in non-symmetric or non-congruent terms. The thesis proceeds as follows: Chapter One is the literature review with the focus on the role of party in the study of congressional elections. Chapter Two presents the foundation of Schlesinger’s theory of party organization and how it is adapted for our purposes. Chapter Three provides the relevant theoretical hypotheses which are to be considered. Chapter Four provides the specification of the models, various data-related issues, and a discussion of the methods of estimation. Chapter Five both presents and discusses the results in terms of the hypothetical considerations. The conclusion consists of an analysis of the theoretical consequences of the study. Appendix A offers the Operationalization, measurement rules, and sources of the data used for the models. Appendix B provides an extended presentation of the methods for interpreting Tobit coefficients. CHAPTER 1 Literature Review Introduction One of the prevailing themes of congressional election analysis is that there has been a significant change in the electoral advantage of incumbents. Congressional elections have become, according to most scholars, less competitive, meaning that fewer incumbents have been winning their seats by narrow margins, and more by ”safe” margins. Competitiveness as measured by electoral margins (Mayhew 1974) suggests a decline in competition. Related research that has offered explanations for this decline has also suggested a corresponding decline in the role of parties and a rise in candidate-centered campaigns. Incumbents and challengers act basically as individuals, separate from both the recruitment of political parties, and the accountability of belonging to a party. A quote from Jacobson (1983) represents the common perception: The evident atrophy of party organization and weakening of partisan ties over the past several decades is thus a crucial factor in the detachment of congressional elections from national political forces and in the rise of candidate-centered campaigns. (Jacobson 1983, 17-18) The result of this phenomenon has been what Fiorina (1980) referred to as "the decline of collective responsibility." "The only way collective responsibility has ever existed, and can exist given our institutions, is through the agency of the political party" and ”there is general 4 5 agreement that the formal party organizations have undergone a secular decline since their peak at the end of the nineteenth century." (Fiorina 1980, 26,28) This theme of the decline in competition and the decreasing role of the political party has been evident in four major areas of research. First, the decline in competition has been attributed to the electoral advantages that incumbents enjoy. Second, the decline is also evident in the increase in split-ticket voting during presidential election years and the ”diminishing” effect of presidential ”coattails”. Third, research has focused at the individual level on partisan defections in congressional elections. Incumbency has replaced the party as the dominant influence. Fourth, studies have focused on the decline in the role of party organizations in congressional elections. The literature suggests that the organizations are relatively ineffective, having never provided much financial assistance, and have played an increasingly secondary role as the influence of Political Action Committees has been on the rise. These four major contentions have resulted in a view of congressional elections as being essentially candidate-centered campaigns, with a decreasing role for the political party. A quote from Crotty and Jacobson exemplifies this view: Within their constituencies, members typically cultivate personal images surprisingly empty of partisan or even programmatic content. They present themselves to people in the district as trustworthy, hardworking individuals who merit support on the basis of personal qualities rather than political beliefs or goals. (Crotty and Jacobson 1980, 180) This chapter will proceed as follows: first, we will review the relevant literature that represents the aforementioned explanation for the decrease in competitiveness and the role of the political party. 6 Second, we will focus on the responses to these considerations. And third, we will focus on the prevailing theoretical basis for the political party and its role in congressional elections. Review of the Decline Thesis: Political Parties and Competition in Congressional Elections A. The "Vanishing Marginals" Thesis The primary analysis emphasizing the changes taking place in congressional election competitiveness was Mayhew’s ”Congressional Elections: The Case of the Vanishing Marginals”. Mayhew examined frequency distributions of vote percentages for incumbents in the congressional races between 1956 and 1972. The analysis showed that more and more congressional districts were becoming ”electorally safe” for House incumbents. Defining marginal districts as the incumbent elected by less than 55 percent of the vote, Mayhew’s histograms showed that the distributions which were normal in the earlier elections, were now becoming increasingly bimodal. "The number of incumbents running in the marginal zone has roughly halved over the sixteen-year period. For some reason, or reasons, it seems to be a lot easier now than it used to be for a sitting congressman to win three-fifths of the November vote.” (Mayhew 1985, 23-24) Data in the article by Cover and Mayhew (1977) suggests that if we assume getting 60 percent as being safe, then at no point between the period of 1956 and 1974 have fewer than half the incumbents won by safe margins. More importantly, the data shows that before 1966 about three-fifths of the House seats were safe, but after 1966 approximately three-fourths of the seats were safe. Mayhew’s 7 reasoning led him to suggest that the decline in congressional turnover could be attributed to the vanishing of the marginal districts. Explanations for this change in competitiveness have basically fallen into three areas. The first explanation is that of Tufte (1973) who argued that the institutional setting of congressional elections has worked to alter the outcomes of these elections. The institutionalized setting in question is redistricting. Some major Supreme Court decisions, specifically Baker v. Carr (1962), took place in the early 19605 regarding voting rights. Tufte looked at this coincidence and found it a plausible explanation. Upon looking at the results of certain redistricting patterns in some states, he found significant variation in some safe districts. Ferejohn (1977), Cover (1977), and Cover and Mayhew (1977) confirmed Tufte’s finding of a relationship between safe districts and redistricted states, but they also found an increase in safety in unredistricted states. Therefore, while Ferejohn agrees that redistricting plays a role in incumbency advantage, he and the others did not accept it as an explanation for the change in the incumbency advantage. The second area results from what Mayhew (1974) proposed himself. Mayhew contends that institutional change has resulted in the modification of voting behavior. He claims that the major factor is the level of visibility of candidates. Incumbents beginning in the late 19605 had a distinct advantage in the way information was circulated then they did previously. Mayhew’s argument was basically that incumbents learned how to use polls and publicity to their electoral advantage. Tools such as franking privileges, congressional staffs, and 8 travel allowances are now available to congressmen, and the incumbents have used them to their fullest advantage. In this same vein, Fiorina (1977) has also emphasized how some of these resources have helped the incumbents. He argues that it has been the expansion of the federal role during the 19605 that has led to an increase in the amount of federal services and the bureaucracy needed to implement them. As a result, the public needed assistance in dealing with this complex bureaucracy and the congressmen turned into mediators on behalf of their constituents. Fiorina claims "the decline of the marginal districts and the expansion of the federal role and its attendant bureaucracy are more than statistically related -- the growth of bureaucracy -- has encouraged behavioral change among congressmen, which in turn has encouraged behavioral change among voters.” (Fiorina 1977, 177-178) Mayhew and Fiorina argue that the role of the congressman has changed. Incumbents now serve as "ombudsmen", serving as go-betweens for the people and the bureaucracy, claiming credit for federal projects and money brought into their districts, and basically providing popular and non-controversial services to their constituencies. The third position attributes the changing nature of congressional elections to a shift in the behavior of the electorate. This position was initially developed by Burnham (1974), but has been more elaborately developed by Ferejohn (1977). Ferejohn analyzes survey data to show an increase in incumbency voting at the level of the individual voter. He also shows that increased incumbency results ”only partly from the increased informational advantage of incumbents over non-incumbents and the propensity of voters to cast their ballots in favor of candidates 9 who are known to them.” (Ferejohn 1977, 167) To emphasize this point, he also presents evidence which shows that the inclination for people to vote for the candidate they know has been increasing over the period under study for all levels of party identification. Fiorina (1977), in reference to Ferejohn, suggests that Ferejohn is correct, there is no question that electoral behavior has changed; however, Fiorina claims the change or at least part of it is "endogenous to the system". B. The Increase of Split-Ticket Voting Thesis The second major argument supporting the decline in competition and party thesis was that at the individual level, split-ticket voting in presidential election years has increased significantly since the advent of surveys in the 19505. According to Burnham (1975), the proportion of congressional districts producing split outcomes has increased tenfold between 1920 and 1972. The incumbency advantage has increased to a point where it is argued that presidential and congressional elections behave independently. Quoting Burnham (1975): Incumbents have become quite effectively insulated from the electoral effects, for example, of adverse presidential landslides. As a result, a once notable phenomenon, the so- called coattails effect, has virtually been eliminated. (Burnham 1975, 412) With this increase in split-ticket voting we see the "diminishing effect of presidential ”coattails"". Kritzer and Eubank (1979) attempt to analyze the lack of coattails in recent elections and propose that we have seen once again a secular change in the electorate, with this change being that voters are moving away from strong identification with the two parties, with the voting cue now being the incumbency effect. The 10 Analysis showed incumbency to be a significant confounding factor in presidential coattails; it may be that the phenomenon which was historically identified as a coattails’ effect has been supplanted by an incumbency effect. This analysis has once again demonstrated the potency of congressional incumbency; it may be on the way to becoming the single most important factor in congressional elections. (Kritzer and Eubank 1979, 624) Calvert and Ferejohn (1983) suggest that while a correlation still exists between Presidential and House outcomes, the correlation has diminished. They suggest that the decline in the coattail voting effect is not a result of the decline in the number of competitive seats in the House. They present a model which estimates ”the degree to which individuals associate their votes for president with their votes for House candidates.” (Calvert and Ferejohn 1983, 408) The principal source of decline lies in this relationship. Voters are relying less and less on presidential candidate evaluations to make decisions about voting at the congressional level. The Burnham, Kritzer and Eubank, and Calvert and Ferejohn findings suggest that House elections are becoming more detached from presidential politics and even more distant from being held responsible for national political conditions. C. The Increase in Partisan Defections Thesis The third area of research suggests that partisan defections in congressional elections have increased along with split-ticket voting. Cover and Mayhew (1977) found that the proportion of defections favoring House incumbents rose substantially after 1970. In 1958, challengers received 43 percent of the defections; in 1978, they received only 11 percent of the defections. Mann and Wolfinger (1980) in the analysis of the 1978 election survey suggest that incumbents have much greater 11 visibility than do challengers, and this enables them to win support of voters who identify with the challenger’s party. This defection hypothesis has been used to explain the differences in reelection rates between the House and the Senate. The 1978 CPS study for the first time included a "true” measure of name recognition, and this question showed that "House incumbents are almost universally recognized by voters in their districts, while the names of challengers are familiar to just under two-thirds of the voters." Mann and Wolfinger’s findings suggest that while the "ombudsmen" role of the incumbents is extremely important, "equally important are the public visibility and reputation of the challenger.” (Mann and Wolfinger 1980, 622,631) Once again, their findings also suggest that the effect of national political conditions and presidential assessments are decidedly secondary to the choices presented at the district level. Bullock and Scicchitano (1982) examine the ”defection hypothesis” by comparing the defection rate gains of Senate incumbents and House incumbents. Their findings suggest that the defection hypothesis can only partly explain the differences between the House and Senate in defection rates. This change in the challenger’s partisan vote has been used as evidence to support the increase of the power of incumbency and once again the decrease of the power of the party as a voting cue in congressional elections. This general impression which is also mentioned by the studies of Cover (1977) and Jacobson (1981), has become the basis for much of the literature supporting the decline in competition and party thesis. Evidence of this impression has heavily influenced the works of Jacobson. (1978, 1980, 1981, and 1983) 12 D. Decline in the Role of Party Organization Thesis Given these first three propositions, it is easy to see why the inherent role of the political party in congressional elections has been pictured as being in a state of decline. The party as an organization itself is considered ”weaker" today than it was when it supposedly achieved its greatest vigor in the period from the 18305 to the 19605. The picture was one of strong party organization, particularly at the local level, and occasionally at the state level. This party was one that was active in and had a near monopoly in such operations as nominating candidates for office, raising money, running campaigns, and voter mobilization. Since the 19605 the party organization has been viewed as ”weaker” since it has less of a monopoly on the aforementioned functions. Quoting Sundquist (1982): The new-style members of Congress ... reflect the individualism of the organizations from which they come.... The new-style members won their seats, in all probability, mainly by their own efforts. No party leaders handed them the nomination in the first place; they were self-selected, and they put together their own organizations, raised their own money, and ran their own campaigns. (Sundquist 1982, 49) Related to this aspect of the party decline thesis is party reform. A great deal of party reform has occurred; from the early changes of the adoption of the national delegate convention in the 18305, through the progressive movement, and through the attempts to make party conventions more representative after World War II. Reform has been active more recently also; with changes taking place in the party organization in Congress, the national committees, the direct primary system, and party finance. For some, party decline is the 13 result of reform. Kirkpatrick (1978) has argued that ”the most important sources of party decomposition are the decisions taken by persons attempting to reform the parties.” (Kirkpatrick 1978, 2-3) In terms of campaign finance reform, Alexander (1984) suggests that they have only helped to further serve the decline of parties. During the first half of the 19705, federal laws regulating election campaigns were changed dramatically through enactment of the Federal Election Campaign Act of 1971, the Revenue Act of 1971, and the FECA Amendments of 1974, which significantly altered both the 1971 laws. According to Alexander (1984): "The reform laws, which are intended to increase citizen participation in election campaigns and decrease so- called special interest influence, actually have had the largely unforeseen consequence of reinforcing candidate-centered politics to the detriment of the parties." (Alexander 1984, 99) The confluence of these four theses has served to present a picture of congressional elections as one in which the parties are playing a diminishing role. The lessening of the influence of parties as an electoral resource has increased the "safeness" of incumbents. The lessening of party influence in the electorate has resulted in increased split-ticket voting. The change in partisan defections largely falls to the advantage of incumbents and contrasts sharply with the invisibility of challengers. Individual members of Congress have little difficulty in satisfying constituency expectations; and their actions are largely carried out at the individual level. Organizationally, the structural characteristics are portrayed as decentralized and weak, unable to play a significant role in the 14 nomination and provision of resources to candidates; with the resultant lack of influence of party at the government phase of organization. Review of the Resurgence Thesis: A Response to the Decline Thesis While the decline thesis has more or less dominated the literature for a number of years, various components of it have come under criticism. Studies have shown that Mayhew’s conception of competition is perhaps not the best measure of competition in congressional elections; and on a wider scale, studies have also shown that there exists a resurgence in political parties, specifically at the organizational level. Various responses and criticisms have developed to the components of the decline thesis presented in Section II. Section III will detail some of these responses and criticisms. A. Response to Mayhew’s Conception of Competition It is Mayhew’s (1974) conception of ”the vanishing marginals” on which the majority of congressional research attempting to explain the incumbency advantage has been based. Jacobson (1987) has looked more closely at this conception. He suggests that competition, properly defined, shows that incumbents are no safer now than they were in the 19505. Vote margins have increased, but he suggests this is not the, proper way to define competition, which he defines in terms of ”swing ratios". The reason why incumbents are no more likely to win the election is because of something Mann (1978) first discovered, the increase in heterogeneity of inter-election vote swings across districts. Jacobson correctly asserts that it "makes no sense to claim that incumbents have become safer because they now win by larger margins 15 if they are just as likely to lose any given election as they were before." (Jacobson 1987, 128-129) Marginality criteria are arbitrary; a 60 percent margin in the 19505 as compared to a 65 percent margin in the 19705 does not suggest that the incumbent is "safer" in the 1970 election than he was in the 1950 election. The phenomenon to be explained is not a decline in competition for incumbent-held House seats; it is, rather, the greater idiosyncrasy and volatility of district-level voting that can expand reelection margins without making incumbents any safer. (Jacobson 1987, 139) Given this more appropriate definition of competition, the competitive status of House seats held by incumbents has not declined. B. Response to the Split-Ticket Voting Thesis The common understanding of split-ticket voting suggests that split-ticket voting ”disunifies” parties; while straight-ticket voting unifies them. The problem, as Schlesinger (1985) points out is that this argument ”works from the evidence of how constituencies vote rather than from an understanding of the reasons for that vote.” Schlesinger suggests that we ”distinguish between rigidities and flexibilities in the voting climate.” (Schlesinger 1985, 1166) Rigidities, he argues, represent long-run factors in voting; whereas flexibilities represent short-run factors such as candidate quality. In both flexible and rigid environments we can expect split-ticket or straight-ticket voting. However, the more flexible the electorate, the more candidates cannot rely upon party identification to gain election. As a result, the more flexible the attitudes in the electorate, the greater the need for organization. Therefore, an increase in split-ticket voting, Schlesinger argues, results in greater competition at the district level in congressional elections. 16 C. Response to the Partisan Defection Thesis The prevailing thesis of partisan defection rates (Cover 1977; Mann and Wolfinger, 1980) has been that since 1972 members of the challenger’s party have defected to the incumbent at increasing rates. Eubank and Gow (1983), Cow and Eubank (1984) and Eubank (1985) demonstrate that there is a pronounced proincumbent bias in the House vote in the 1978 and 1980 data. Eubank (1985) points out that the pre-l970 surveys measured party responses by those unable to recall the name of the candidate for whom they voted. The CPS surveys of 1970-1976 coded the congressional voting response in two separate variables: a candidate voting response and a party response variable. Eubank suggests that ”the most accurate voting estimates are ... obtained only by first employing the candidate response and then, for those unable to give such a response, utilizing the party responses.” (Eubank 1985, 960) The ”methodological discontinuities in the initiating studies of individual-level incumbent effects (Cover 1977; Cover and Mayhew 1977)” combined with "the severe pro-incumbent bias within the 1978 and 1980 surveys (Mann and Wolfinger)" has served to reinforce the misconception of this dramatic impact of incumbency at the expense of party as a voting cue in congressional elections. (Eubank 1985, 959) D. Response to the Party Organization in Decline Thesis E. E. Schattschneider wrote that "decentralization of power is by all odds the most important single characteristic of the American political party." (Quote from Arterton 1982) James Q. Wilson (1973) suggested that a reason American parties have remained decentralized has to do with the benefits offered. The material benefits are usually only 17 available at the local level; the solidary benefits can be ”thwarted by controversy.” According to Wilson, ”This leaves purpose, principle, and ideology as a major source of incentives for party organization.” (Wilson 1973, 96) Following this, a strengthening of the national party organization must depend on a different type of reward than offered at the local level. Kayden (1980) suggests that ”(i)t is the complexity of the law and technological developments which account for this change for two reasons: they allow the organization to acquire sufficient resources to hire professional staffs; and politics -- at least at the national and state levels -- is becoming more and more complex, requiring specialization in the performance of tasks.” (Kayden 1980, 276-277) Kayden posits that party organization will become like national bureaucracies, with the parties reaching the stage of organizational development that differs greatly from the normal structural characteristics of weakness and decentralization. The centralization of the party structure has coincided with a ”party in resurgence” thesis. This resurgence is in large part a reflection of the incentive structure developed by Kayden (1980). Quoting Adamany (1984): Their is evidence of new party vitality in fund raising, especially by the Republicans. National party committees, again on the Republican side, are providing a vast array of services as well as healthy campaign contributions to local, state, and federal candidates. There is greater professionalism in parties - - partly to meet the complexity of campaign finance laws, but also to provide the expertise required by new campaign technologies. (Adamany 1984, 70) The changes taken place have been in the increased power of the Democratic and Republican National Committees, and perhaps most extensively in the role of the Republican and Democratic Senatorial and 18 Congressional Campaign Committees. Bibby (1979, 1980) has reported on the activities of the Republican National Committees. Conway (1983) relates this Republican Party nationalism to implications in the party system. Herrnson (1986) also focuses on the role of the national party in the election campaigns of House and Senate candidates. He presents the national party organizations as "brokers" suggesting that this model: . depicts national party organizations as playing an important role in modern campaign politics.... (N)ationa1 parties use their financial and technical resources to assist candidates in communicating with and mobilizing groups of voters. The national parties use these and other assets ... to convince, induce, or otherwise persuade state and local party organizations, political consultants, PACs, and candidates to follow their advice, instruction, or informational cues when conducting election- related activities. (Herrnson 1986, 3) The resurgence thesis as posited in the organizational mode has emphasized two major phenomena: the increase in activities that revolve around getting candidates elected, and the centralization and nationalization of the party structure. (Reichley 1985) The picture being presented is one of the national parties being far removed from the mass state and local parties of earlier times; national parties more adept at campaign technology and the provision of services for candidates running for offices under their name. Theoretical Basis for Party Organizational Activity in Congressional Elections While the previous studies emphasizing the resurgence of the political party focused on the nationalization of the party organizational structure, they basically were descriptive in nature and lacked a coherent theoretical basis for many of their party in 19 resurgence contentions. One attempt to provide a theoretical base for party organizational activity is the Strategic Choice Hypothesis of Jacobson and Kernell (1983). A. The Strategic Choice Theory of Congressional Elections Jacobson and Kernell’s theory of strategic choice has been well documented. (1983) This theory presents how macroelectoral behavior can be derived from features of the opportunity structure. Individual decisions on whether to run for office relate to partisan electoral conditions. These decisions are based upon the probability of winning office. The collective result is that the better the party’s candidates, the better its performance. Their theory shows how the collective strategic choices of politicians are shaped by national events and conditions; these decisions determine the kinds of candidates and campaigns the voters are offered. They therefore propose that national economic conditions only have an indirect effect on congressional elections. The political party in their theory is one of the strategic actors. The strategic actors’ -- politicians, contributors, and parties -- "choices reflect ... the conventional wisdom that national events and conditions affect individual voting behavior.... Clearly, the choices presented to voters between a pair of candidates in the district are not at all independent of national conditions; indeed, they are a function of them." (Jacobson and Kernell 1983, 34) Not only do the strategic choices of individual candidates serve as the link between national events and individual voter behavior; but the strategies of parties and contributors play this role also. It is within this context that the party is placed in Jacobson and Kernell’s theory. The determinants of party organizational strategic 20 behavior are posited as the same for individual strategic behavior. The determinants in question are national aggregate economic and political conditions and district level conditions. B. Parties and the Strategic Choice Hypothesis The extension of Jacobson and Kernell’s thesis to party organization is expressed in an article by Jacobson (1985-86). In this article it is posited that the Republicans have achieved greater centralization in their national committee structure than have the Democrats. Because of this, Jacobson suggests that they also have been much more efficient in the distribution of their resources than have the Democrats, who Jacobson argues have ”remain wedded to a thoroughly decentralized system.” The Jacobson thesis in the article is to "present a theory of how congressional campaign finance serves to reinforce the effects of candidate strategies on choices presented to voters at the electoral level.” (Jacobson, 1985-86) Jacobson posits that the Republicans benefited by their more efficient distribution of resources in the 1982 House elections, and as a result lost only 26 seats whereas more were predicted. C. Political Action Committees and the Strategic Choice Hypothesis Aside from the initial assessment of PAC contributions presented by Jacobson and Kernell (1983), the next most direct application of the Strategic Choice position as it involves PACs is the study by Eismeier and Pollack (1986). Whereas previous research has focused on the aggregate spending of PACs, the Eismeier and Pollack study focuses on the aggregate flow of funds to candidates. They suggest that their model provides for direct evidence of how the strategic calculations 21 serve as a link between national conditions and election outcomes. A comparison of 1980 and 1982 -- with 1980 favoring Republicans and 1982 favoring Democrats -- supports a modified version of the Jacobson and Kernell model. They divide PAC motivational strategies into three categories: accomodationist, partisans, and adversaries. The accomodationists focus on securing access and influence in Congress. The partisans are pictured as "financial auxiliaries of the major parties.” Adversaries primary aim is to ”rid Congress of members who fail to pass one or another litmus test and perhaps by so doing to set an example for others." (Eismeier and Pollack 1986, 199) By analyzing the aggregate shifts in the distribution of PAC contributions, they find that the Republican partisan PACs did move from a challenger-oriented posture in 1980 to an incumbent posture in 1982. The accomodationists (specifically trade associations) did remain ”relatively impervious to political expectations;" and the adversarial PACs did show constant devotion to nonincumbents. D. Individual Candidates and the Strategic Choice Hypothesis Analysis of the determinants of individual candidate behavior has focused primarily on incumbent behavior and their particular activities designed to discourage strong opponents. (Mayhew 1974; Fiorina 1977; Fenno 1978) If the incumbent is doing a credible job, potential challengers will hesitate to run, at least given the present national conditions. (Jacobson and Kernell 1983) A more systematic version of related hypotheses has been offered by Ragsdale and Cook (1987). Their study specifically analyzes the interaction between the behavior of the incumbent and the resultant behavior of the challenger. They analyze three primary relationships: 22 (1) they test "whether challengers’ strength ... is a systematic product of incumbents’ resources;" (2) they test whether there exists a simultaneous relationship ”between the incumbents’ noncampaign resources (e.g. trips home, district staff allocation, and bills cosponsored) and challengers’ campaign money"; and (3) they "investigate the extent to which incumbents’ campaign spending and PAC contributions are affected by the use of their perquisites of office.” (Ragsdale and Cook 1987, 48) They use the elections of 1978 and 1980 to test these hypotheses. Their findings suggest that (1) incumbent activities in the interelection year do not have an effect on challenger funding efforts. (2) The relationship between incumbents’ noncampaign resources and challengers’ campaign expenditures is not reciprocal; the claim is for simultaneity not to exist. (3) They find that incumbent’s resources in office have a weak impact on both the members’ resources and the challengers’ resources. They therefore suggest that the probability of strong challengers is primarily related to preceding election and the strength of that challenger in that election. The individual decision is therefore not based on the resources of the incumbent. The determinants for individual behavior developed by Ragsdale and Cook closely approximate those developed by Jacobson and Kernell. The Party Theoretical Perspective of Congressional Elections While the previous literature has focused on the study of parties and congressional elections from a congressional election theoretical perspective, it will be argued in this thesis that many of the aforementioned issues can better be analyzed from a different 23 theoretical perspective. One attempt to provide a theoretical base from the party perspective has been Schlesinger’s "Office Seeking Model” of political parties. (1965; 1984; 1985) What follows is a brief description of the theory and the studies that have derived from it. Schlesinger argues that democratic parties can best be studied as primarily office seeking parties. He develops a theory of party organization that is based in a rational organizational perspective; one that accepts Downs’s (1957) definition of parties as teams. Since being on a team implies some form of ”deliberate cooperation," parties exist in the organizational form. While this research focuses on the organizational aspect of parties, it also integrates another area of study: the strategic behavior or choice of individual candidates. The theoretical basis for this construct originates in Schlesinger’s book Apbitipp apg Poligigs (1966). In Schlesinger’s model (1985), the party is shaped by important aspects of the environment, with the main consideration being those which shape the expectations for the office-seekers. There exist two main factors: the structure of politicgl ppporgunities and phe party sysggm. The party organization is a result of the interaction of these two variables. By the structure of political opportunities Schlesinger means ”the offices and the rules for attaining them” and ”the general patterns of behavior surrounding their attainment." (Schlesinger 1985) The construct of the party system focuses on the interaction between the two parties -- defined in terms of competition. It is in this way that Schlesinger incorporates the structural or institutional aspect to his 24 definition of the party. The goal of winning office is constrained by the institutionally defined structure of political opportunities. The opportunity structure affects the calculations of individual politicians. Black (1972) provides an extension of the Schlesinger model by looking at the individual level influences on the strategic calculations. He develops a model of political ambition based on utility theory which shows that the structural characteristics of systems can have an effect on both the perceived risks involved and the investments required to reach office. Black’s findings suggest that the structure of the political system does have a strong influence on the types of candidates that emerge. Rohde (1979) extends the work of Schlesinger and Black by analyzing the decisions of House members and by attempting to predict those who will run for higher office. While Schlesinger’s analysis focused on those who behaved ambitiously, Rohde’s analysis assumes members possess progressive ambition, and attempts to explain under what conditions and structures a politician will seek to run for a higher office. The decision calculus involves the probability of winning the office times the positive utility of the office, minus the costs associated with running for that office. Jacobson and Kernell’s (1983) theory of strategic politicians is also based on the principle of the opportunity structure. For them, national political forces which politicians expect will have an impact on the voters will play a key role in the strategic decision calculus. The platform of offices, peculiar to America, invites a large number of potential competitors, provides numerous opportunities for mobility, and raises some uncertainty about the prospects for success. Under these circumstances successful politicians must be acutely strategic in making career choices. How the structure of 25 opportunities determines the targets and, more importantly, the timing of career moves has important implications for a theory that views the collective strategic behavior of politicians as an important determinant of election outcomes. (Jacobson and Kernell 1983, 20) As a result, Jacobson and Kernell develop a collective strategic result based on the individual calculus developed by Black and Rohde. These collective (party) strategies basically reflect the short-term partisan forces. Since the individual candidates are rational actors, the prospects for the party are a crucial consideration in their decision calculus. This in short, is the nexus of their argument. A test of the Jacobson and Kernell thesis by examining individual candidacy decisions is provided by Bianco (1984). He finds that challengers in fact base their decisions on district level political factors and changes in economic conditions. His findings support Jacobson and Kernell’s contentions that: (1) ”candidates assume voters hold the party in power responsible for changes in economic conditions.” (2) That ”quality challengers from the party in power are more likely to run against an incumbent as the change in economic conditions becomes more positive.” And (3) that the converse is true, ”quality challengers from the party out of power are more likely to run against an incumbent as the change in economic conditions gets smaller or becomes negative.” (Bianco 1984, 361) Bianco also finds that unlike the Jacobson and Kernell thesis, these relationships do not hold for elections with open seats. Canon (1985) extends this research of the structure of political opportunity by ”merging” together Rohde’s individual level studies with the various structural considerations provided by Schlesinger. By doing this he also attempts to incorporate the ”amateur” politician into the 26 equation, solving the previous problem of not analyzing those without previous electoral experience. Canon therefore looks at the initial decision and considers not only the individual’s decision context but also the structural factors. By doing this he posits a more complete theoretical basis for candidate decision-making in congressional elections. Canon’s model is therefore an extension of Jacobson and Kernell in that the three main factors affecting strategic politicians’ calculations are: district specific conditions, national aggregate conditions, and his additive, structural/institutional variables. His results show the district level variable to be significant, the structural/institutional variable to have ”mixed results”, and the economic variables to be more or less insignificant. The results for open seat elections provide similar results to Bianco. Summary and Conclusion As Schattschneider (1942) writes, ”political parties concentrate on winning elections. This is the point at which modern parties make an effort to get control of the government.... The party becomes, therefore, a process formed about the elections.” (Schattschneider 1942, 61) The study of political parties should focus on the party as an organization that makes a bid for power. The requirement of winning elections determines what sort of an organization the party will be. The study of the role that parties play in congressional elections should come from this conceptualization. 27 The party, properly defined, is not in a state of decline. An approach which emphasizes the goals of the office-seekers within the party organization will allow us to conceptually clarify the candidate- centered politics that the literature has emphasized. Any analysis of party organizational behavior must include the aspect of individual candidate decision-making both as exogenous and endogenous factors. Individual candidates are a resource in the party organization. The strategic behavior of individual candidates operating within the structure of opportunities not only has an effect on congressional elections as Jacobson and Kernell correctly point out, it also has an effect on the collective behavior of the party. As a result, if as Jacobson and Kernell and Canon suggest, that district specific conditions, national economic conditions, and Structural/Institutional conditions have an effect on individual choice; then consideration of these factors as they interact with other determinants must be considered in attempting to analyze the collective behavior of the party in terms of resource distribution in congressional elections. CHAPTER 2 A Theory of Party Organization Introduction The object of this chapter is to present a theoretical base which serves two purposes: first, it provides a theoretical clarification for many of the varied contentions presented in Chapter I; and second, it provides a foundation for the development of testable propositions. The final section discusses how this theory differs with the Strategic Choice position on party organization and the distribution of resources in congressional elections. The Study of Political Parties: Two Theoretical Approaches The scholarship on political parties can basically be divided into two groups: those who believe the party belongs to its members, and those who view the party as an organization aimed at winning office. A. Parties as Membership-Based Organizations The seminal work in the conceptual framework of parties as membership-based organizations is Michels's hierarchical model, written in 1951 and presented in Eplitigpl Papties; A Spgiolpgigal Sppdx pf the Qliggpcpicgl Tendencies of Modern Democracy (1966). The problem of organizational democracy, according to Michels, is the contrast between the "norm of internal democracy and organizational practice." Parties 28 29 stress the norm of internal democracy, but in real life do not practice it. As an organization increases in size, the need for those with managerial and organizational skills to assume positions of importance develops. These positions gradually turn into positions of leadership with a resultant chasm developing between the leaders and the mass base. In short, there is a change in the party’s structural relationship with the leaders becoming a "new political class”, having taken over the party. The rank-and-file membership are then subject to exploitation by the political elites. This "party democracy" conception was further developed by Duverger (1963) who categorized mass parties in terms of their organizational unit: the caucus, the branch, and the cell. A more widely accepted distinction was the mass membepspip party and the Eggpg party. The mass-membership party (more representable of European parties) emphasized the role of the members; members who were active, dues-payers, and well-organized. The cadre party (more representable of American parties) is characterized by a looser organization which is structurally decentralized. The key characteristic is that the party is made up of individuals who are considered ”members" of the party, not just ”supporters.” Mass-membership parties are characterized by a high degree of organization; they are internally democratic, policy-oriented, and ideologically oriented. The "prime beneficiaries” of the party organization are the members. As a result, both Michels and Duverger are concerned about the level of intraparty democracy; for the party to maintain itself, it must be responsive to its members. 30 B. Parties as Electorally-Based Organizations Whereas the mass-based approaches of the study of parties emphasize the policy functions and internally democratic manner of the party, another strain of study emphasizes the party's electoral functions, with the emphasis on winning elections. Analysts in this mode assume the electorate to be the main beneficiaries of the party organization. Models in this mode stress competition among leaders who are socialized to uphold the democratic electoral process. The seminal work for this theoretical conception is Schumpeter's Capitalism, Soclalism, ppd Democracy (1950). Schumpeter suggests that the essence of democracy is found in the "institutional arrangement for arriving at political decisions in which individuals acquire the power to decide by means of a competitive struggle for the people's vote." (Schumpeter 1950, 269) Schumpeter draws a distinction between an electorate deciding on matters of policy (which they are incapable of doing), and choosing between teams of leaders offering alternative party platforms. Downs in An Economic Theory of Democracy (1957) offers a similar model where parties are characterized as unified teams of individuals whose sole aim is to win office. They do not possess ideological commitments, and they are prepared to adopt whatever platforms seem most likely to win elections. The party can be dominated by selfish office- seeking goals because the party system itself, and the competition for offices, forces the office-seekers to be responsible and for the party to be public-minded. 31 Schlesinger (1965) operates from the assumption that party organizations in democracies are dominated by their office drives. He argues that this position is derived from the "logic of democracy." There is nothing in the democratic system to deter the development and preservation of a political party dedicated to policy goals.... Yet if political power is gained through party competition, then it must be won by those parties which compete most effectively; only those parties which make their decisions in response to the competitive rules can, in the long run, be governing parties. (Schlesinger 1965, 767) The more competitive the party, the more it tends to be dominated by the office-seekers. It is in this way that Schlesinger's argument approximates Michel’s explanation of the oligarchic tendencies in political parties. Schlesinger's theory does not suggest however that the office-seekers will form a policy-defining oligopoly or elite. The assumption that the dominant goal of party is office leads one . to examine the conditions of office-seeking to find there the reasons for the development of various forms of party organization. (Schlesinger 1965, 767-768) It is on this basis that Schlesinger -- in the Schumpeter and Downsian mold -- develops his theory of political parties. Schlesinger's Office-Seeking Model of Political Parties A. Extension of the Downsian Model Downs’s (1957) analysis assumes that political parties act as though they are single individuals out to maximize their utility, in this case winning elections. Downs's model of the politician hypothesizes that he or she will always attempt to maximize his or her probability of acquiring or staying in office. Parties in Downs's system are individuals who all share the same values and goals, and who 32 seek election to office to obtain the advantages of holding the office for themselves. They do it in order to satisfy their personal needs for power, money, or status. In other words, parties in Downsian terms do not seek office in order to implement particular policies. They advocate policies in order to obtain votes and obtain selective benefits. Key to this position is the self-interest axiom that is present in all rational choice theory: rational behavior is directed primarily for self-maximizing ends. A crucial element in both Downs and Schlesinger's account is the role of the individual political actor. For Downs, democratic governments act rationally in order to maximize political support. Rational action is defined as "action which is efficiently designed to achieve the consciously selected political or economic ends of the actor." Schlesinger looks at the party as it is used to define personal goals. The important concept for Schlesinger is the nuclear party organization. The collective effort of the party is "devoted to the capture of a single public office." This is the ”building block" of the party structure. What holds the party together is this single nucleus, and its dependence on other nuclei. This corresponds with Downs's model in that the rational actor always attempts to maximize his probability of acquiring or staying in office. Schlesinger assumes two different types of ”contributions" to the party nucleus: ”minimal” and "connective". (1965) Minimal contributions consist of candidates and votes for the nuclear office, the requirements for winning elections. In order to do this, several kinds of connective contributions are required, among which are leadership, recruitment and nomination, technical services, and money. The organization only 33 functions to the extent that it ”facilitates vote mobilization tasks and promotes electoral success.” The party as an organization, exists only for this entity; without it, it would cease to be a party. The prospects for electoral success is what is required in order for a "true nucleus” to exist within the party structure. B. The Theory of Party Organization Schlesinger's research on political party organization focuses on organizational structure and the problem of goals. (1965, 1984, 1985) While his research focuses on the organizational aspect of parties, it also places an emphasis on the strategic behavior or career choices of individual candidates. (Schlesinger, 1966) In Schlesinger's model, the party organization operates in an open system, and thus is shaped by important aspects of the environment. The main environmental considerations are those that shape the expectations for the office- seeker. Two main factors exist: the structure of political opportunities and the party system. The party organization is a result of the interaction of these two variables. The structure of political opportunities derives from his Ambition Theory. By this he means, "the offices and the rules for attaining them”, and more importantly for our purposes, ”the general patterns of behavior surrounding their attainment." (Schlesinger 1985) The construct of the party system focuses on the interaction between the two major parties -- defined in terms of competition. These then are the three interacting variables: the structure of political opportunities, the party system, and the party organization. The theoretical basis for the structure of political opportunities appears in Schlesinger’s book Apbitipp 5nd Eolltlgg (1966). Political 34 actors behave in a way that is consistent with achieving higher office. The "political opportunity structure” represents the system which structures these individual behavioral patterns. Schlesinger focuses structurally on the various rules and procedural changes which lead to changes in career patterns; the increase or decrease of "linkages" between elective offices; and the interaction between the opportunity structure and the party system. He argues that the present day structure of opportunities, under ”The New American Political Party", is one of a more national or homogeneous nature; one with increased linkages between elected offices. The second important variable which interacts with the structure of opportunities is the party system, defined as the competition between parties for elective office. Party nuclei emerge in districts or constituencies where the level of competition suggests the party has a chance of gaining office. Depending on the competitiveness, Schlesinger distinguishes between party strategies in terms of long and short-run factors in voting decisions. In situations where one party is dominant the party strategically emphasizes long-run factors; with a specific emphasis on the retainment of party identification. Individual voters with strong party identification use this identification as the dominant influence in their voting decisions. Conversely, a party in a minority position emphasizes short-run factors; with an emphasis on individual candidate characteristics and relevant economic and political conditions. These factors are employed in order to encourage defections from the dominant party. The interaction of the structure of opportunities and the party system defines party organizational effort. In competitive 35 constituencies we expect to see greater organizational effort than in non-competitive constituencies. Constituencies in which a party is dominant results in the party possessing greater resources, and they will most likely use these resources to maintain that dominance. In these cases, the dominated party will likely invest very little in terms of resources. In competitive constituencies both parties are active in the use of resources. Need for resources is also partly defined by the opportunity structure, for constituencies which foster progressive ambition also provides instances for greater linkages amongst the party nuclei. Office-Seeker Strategies: The Allocation of Resources The role of the individual office-seeker determines the activities of the party organization. The goal of the office-seeker is realized (1) by the collective decision making on the allocation of the organization's resources, and (2) by the utilization of resources themselves. It is important to realize that these processes constrain individual behavior. Within this framework, collective party organizational behavior is taken to be the collectively constrained utilization of resources by individuals. Within the organization there is competition for the use of these resources. The choices of the resources and the pattern of allocation are determined by individual and collective office-seeker decisions; and these alternatives are a consequence of the characteristics of actual resources. In considering party 36 organizational resources, the focus is on such things as money and the individual candidates themselves. The party organization, according to Schlesinger, operates in a cycle consisting of three phases: the nomination, the electoral, and the government phase. The party organization allocates its resources in different patterns and strategies according to the environmental conditions at each phase. It is important to realize that the party's action at one phase has an influence on the party's behavior at the other stages. Conditions which result from policy decisions at the government phase affect individual and collective strategies. Likewise, individual choices made at the nominating phase affect strategies at the electoral phase. While the organization and the collective interests of the office- seekers are concerned with a diversified set of means in order to achieve their ends, the individual office-seeker is also concerned with strategic behavior. Strategies vary in their application according to the needs at the various phases of organization. Central to this argument is that the possibility for a collective action problem exists between the action of the individual office-seeker and the strategic use of resources; and the collective or organizational interests of the party, and their strategic use of resources. This is the basis on which Jacobson (1985-1986) bases his contentions. Key to this potential collective action problem is the amount of discretion that the office- seekers have in the determination of the way resources are allocated within the party organization. Given the present emphasis on candidate- centered campaigns, it is safe to argue that in the party organizational setting, the individual office-seekers possess a great deal of 37 discretion. This discretionary behavior is evident at all three levels of organizational behavior. The object of the individual office-seeker is to remain autonomous from collective party considerations. A. Individual Office-Seeker Strategies Schlesinger's theory states that the goal of the individual office-seeker is to gain or retain office. The office-seeker prefers to do this with the greatest amount of autonomy which is possible; therefore, the strategy of electoral competition is maximization of the vote and plurality. By maximizing the victory, the office-seeker provides for him or herself the greatest amount of independence from the collective interests of the nuclear and cross-nuclear organizations. Schlesinger's (1975) discussion of Riker’s "size principle” illustrates this point. Riker's "size principle" suggests that coalitions (parties) attempt to achieve a ”minimum winning coalition", thereby clarifying the ”rationality crisis" of Downs's formulation. Schlesinger suggests however, that by using a minimum winning coalition strategy, what is maximized is the number of people feeling they had a claim on the victory. . by achieving a minimum winning coalition we had maximized the number of marginal voters. And marginal voters, precisely because their votes are the ones needed to win, have a greater claim on the Officeholder. Thus the minimal winning coalition, rather than reducing the number of payoffs which the Officeholder had to make, actually appeared to heighten the expectation of payoffs. (Schlesinger 1975, 842) These payoffs in one way or another are evident at the government phase of party organization. Benefit-seekers and those who form the nucleus of the party organization expect to reap benefits which can be 38 gained from the control of office. By maximizing the victory the individual office-seeker increases the autonomy. Given the goal of maintaining or gaining office, the assumption is made once again that the individual candidate seeks to maximize his or her chances of winning; and in order to do so, the candidates seek to maximize the amount of resources needed to attain this goal. Using the same logic as Schlesinger’s maximization argument, the office-seeker can best maintain autonomy by the acquisition of large amounts of contributions. By acquiring contributions from a large number of contributors, the payoffs are minimized. The central question once again remains the size of the victory. If the best strategy of electoral competition is the maximization of the vote and plurality, then the best strategy of electoral resources is also to maximize for the office-seeker. By maximizing both strategies, the office-seeker maintains the greatest autonomy. This question of autonomy is also relevant to the collective interests of the party. The party organization and its relevant committees distribute resources to particular candidates. While individual candidates welcome the assistance of additional contributions, they prefer to maintain a measure of distance between themselves and the rest of the organization. This tension between the individual candidate and the collective interests of the party is most evident at the governmental phase of organization, For the individual, discontinuity is preferred, and non-disciplined governmental parties which allow for individual discretion allow for the system of candidate- centered politics to maintain itself. 39 B. Collective Office-Seeker Strategies A common criticism of the rationalist approach is that it is not useful when applied to collectivities. Political party behavior in the collective sense cannot be interpreted in the same way individual behavior can be interpreted. This perspective is built on the erroneous notion that individualist assumptions are not adequate in order to understand political behavior in the social context. A theory of collective behavior not built on individualist assumptions must be built upon collective rationality. The problem is that collective rationality must be built upon a group preference and Arrow (1963) shows that collective preferences in general cannot be transitive. A theory of party organization must assume that collective action is deduced from individual preferences and choices. In effect, the party organizational perspective allows us not to concentrate solely on individual actions, but to study the patterns that emerge from the collectivity itself. Within the framework, organizational behavior is defined in terms of the collective utilization of resources by individual office-seekers. For an individual, difficulty arises in the manipulation of these resources, because others are attempting to use them also. There are limits to individual office-seeker behavior in the collective organization. Actions of some individuals have consequences for others, and there exists a need for collective decision-making. These strategies of decision-making at the collective level consist of the allocation of resources. Being the integral part of the party organization, all office-seekers make decisions that affect the allocations of resources. For Schlesinger, the party is a group 40 organized to gain control of government; the party organization entails the cooperative activities or the linkages of cross-nuclear behavior. To a large extent the resources available in the party organization are controlled by the individual office-seeker. However, constraints exist over the uses, and the allocation must resort to certain criteria. From the viewpoint of the individual office-seeker, the party organization is an instrument for realizing the goal of gaining or maintaining office. An individual agrees to an allocation if he or she perceives it as directly contributing to his or her individual goal. Collective action is therefore contingent on some form of agreement on resource allocation. The crucial consideration is that for the individual office-seeker cooperative behavior is a means towards an end, and this end is winning office. Differentiation with Jacobson and the Strategic Choice Model The theoretical precepts of the Party Organizational Model and the Strategic Choice Model are built upon similar assumptions and principles. Differences do exist however and they are analyzed in the following four sections. The first section offers what is referred to as the ”Symmetric Position." This describes the theoretical base of Jacobson's position and the strategies derived from it. The second section offers a description of the ”Non-Symmetric Position” which builds off the party organizational theoretical base and critically analyzes Jacobson’s contentions. The third section focuses on the differentiation in terms of the decision procedure for the evaluation of 41 national conditions. The final section discusses the difference in the sets of determinants between the two approaches. A. The Symmetric Position: Jacobson on Collective Goals and Strategies Jacobson (1985-1986) posits a party organizational goal. "In congressional elections, a party's collective interest lies in winning as many seats as possible ..... (M)aximizing seats is the goal. In the abstract, it should make little difference to them which candidates win, only that as many as possible do.” (Jacobson 1985-1986, 604) Central to his argument is the perception that party competition exists and that each individual seat has a great deal of utility. For Jacobson the fact that the marginal utility of an additional House seat for the Democrats might not be that high is irrelevant. Concerns over the importance of the Conservative Coalition and the unpredictability of congressional elections lead the Democrats to want to maximize their seat total. What follows from this symmetric interpretation of congressional elections is a similar theoretical basis for resource distribution. An efficient distribution of resources by the organization suggests that the resources are provided to the candidates who are in the most competitive elections; i.e. incumbents in danger of losing their seat, challengers who have a chance of winning, and competitive open seat elections. An “efficient" or "effective" distribution does not suggest that party campaign assistance be provided for non-competitive candidates when there exists a situation of scarce resources. An individual office-seeker whose electoral strategy is to maximize the vote and plurality is proposed to want to maximize the amount of campaign resources available. By all candidates acting in their own interest -- specifically those in places of influence -- there 42 exists a distinct possibility that the best interests of the group will not be served. The result is a collective action problem. The good of the whole is not best served by the incorporation of individual interests. Resources provided for a "safe" incumbent are better distributed to a candidate in a competitive district where the resources provide a greater return. Jacobson argues that incumbents have a greater amount of influence in the allotment of party money and that the organizational advantage of the Republicans allows them to use their resources more efficiently than the Democrats. Effect of Symmetry Position on Strategies Jacobson's party organizational goal of maximization of seats is presented as being a symmetric argument. The two parties compete against each other every two years with the goal being to win as many seats as possible. The number of seats one party gains, the other party loses. Questions over the marginal utility of specific seats is deemed irrelevant. This symmetric presentation and the goals that are posited have a definite effect on the strategies of the party organization in terms of resource distribution. The symmetry of goal structure translates into symmetry for party resource strategies. The question is over how the dominant determinant of national conditions affects the allocation of resources to two subgroups of candidates: incumbents and challengers. In periods of favorable conditions for a party, the party allocates a proportionately higher amount of money to their challengers. This is known as "playing offense". In periods of unfavorable conditions, a party allocates a proportionately higher amount of money to incumbents. 43 This is known as ”playing defense". This is a highly symmetrical proposition. The problem with this formulation is that it is not particularly discriminating. The decision calculus for the organization is only slightly altered from the maximization position given the additional determinant of national conditions. It resembles a market system in which decisions are made on the simple notion of utility maximization. The party spends capital in order to receive a return. The defensive position suggests that parties seek to maximize the number of seats by trying to protect incumbents. The offensive position suggests that parties maximize by spending proportionately less on incumbents and more on challengers. The question is whether this makes any sense from a seat maximization perspective? Favorable or unfavorable conditions should affect incumbents and challengers of one party in an equal fashion. Conditions can be translated into either a certain percentage gain or loss for a particular candidate. By a strictly symmetric interpretation, party money offsets the unfavorable conditions and emphasizes the favorable. But this does not explain the differences between incumbents and challengers. Civen Jacobson's stipulations that a "competitive" situation exists between the parties in the House, and that all seats possess equal utility, a pattern of differentiation between challengers and incumbents should not be taking place. By the symmetric definition, to do so would be irrational. The metaphor is ill-conceived, lacks a theoretical foundation, and the preexisting evidence is in fact a function of some other forms of determinants. This provides the basis for the following discussion. 44 B. The Non-Symmetric Position: Collective Behavior as Individual Decisions Symmetry in this context implies a certain corresponding goal structure between two sides. For Jacobson and Jacobson and Kernell, symmetry exists in their theoretical perspective and in their explication of strategies. On the theoretical level, an alternative to the symmetrical position is offered. The positing of an organizational goal, and specifying that goal as the maximization of seats, leads to theoretical discrepancies in the ability to explain strategic behavior. The ppm; gymmgpplg position as it relates to theory and practice is the subject of the next two sections. e No - e c osition The difficulties in positing an organizational goal and the ramifications of doing so lead to a slightly different theoretical approach. Collective strategies consist of nothing more than the aggregation of individual strategies. This is the limit to which individuals allow the collective organization to constrain their own individual behavior. The focus is on individual office-seekers, and the aggregation of individual office-seeker goals does not translate into a collective goal of seat maximization. Olson, as stated, assumes a common goal and develops the collective action problem. ”(P)erfect consensus, both about the desires for the collective good and the most efficient means of getting it, will (not) ... always bring the achievement of the group". (Olson 1965, 59) Given Schlesinger’s conceptualization of the party organization however, a statement of a collective good for the organization might 45 merely be a statement on the limits of resource allocation that the individual office-seekers may tolerate, or as White (1974) suggests, "the minimal condition to which they agree in collective behavior." (White 1974, 375) The party organization is a multi-nuclear organization which shares a common objective, common in that the organization is dominated by office-seeker interests. Organizational continuity depends upon the need for the organization by the office- seeker who attempts to divorce him or herself from organizational considerations whenever prudent. The tension that exists between the individual and the collective is one of individual autonomy versus collective cooperation. The amount of cooperation between office- seekers is a reflection of their response to various conditions, specifically party system competitiveness. As previously noted, Schlesinger states that the collective goal of the office-seekers is to gain control of government. This is a different principle than seat maximization, and it leads to a number of different considerations. t n ar n Unlike the symmetric position, this position is in part non- symmetric because it distinguishes between minority and majority parties in Congress. The Democratic Party in the House does not operate under the same goal structure as the Republican Party. Seat maximization is not the appropriate goal definition for either party. The Democratic Party organization, controlled by office-seekers, does not go into each set of House elections trying to maximize seat total. Of crucial importance is the size of the respective party contingents. A competitive situation, as is evident in the Senate, is more conducive to 46 the maximization argument. Given the size of the Democratic party majority, and given the decline of importance of the Conservative Coalition, it is much more appropriate to describe Democratic behavior in terms of a "maintenance” strategy. The party is more concerned with maintaining their majority, and with maintaining the control of certain seats. The standing of the Republicans dictates a more ”expansive" type behavior which is more appropriate for a minority position. c a l Ut t The second consideration follows from the first. The non- symmetric position is preferable because it adopts the principle of marginal utility. All seats do not possess the same utility. The standard assumption in economic theory is that marginal utility declines as the supply of the good increases. For our purposes, this means that a seat which is instrumental in establishing a majority brings a high level of utility, and that each succeeding seat brings less utility. In reality, a certain threshold probably exists for the congressional parties. If the total number of seats for a majority is in danger of being reduced to minority status, these seats are important, and possess a great deal of utility. Given the present status of the Democratic majority in the House, seat number 257 possesses less utility than seat number 218. Spppclpy gpd Nop-Scarclty of Rgsoupces The non-symmetric position also takes into account the scarcity or non-scarcity of resources that exists. The situation varies given the two main party resources which are considered: individual candidates 47 (challengers) and monetary resources. These considerations vary given the party in question. Individual candidates as a resource are a far greater problem for the Republicans than the Democrats. This is the case for two reasons. The first is the Republicans have many more seats for which to find possible challengers. With the seat distribution in the House showing the Democrats with a 79 seat advantage, that puts the Republicans in the position of trying to recruit or be the recipients of 79 more challengers than the Democrats. The second reason is that the Republicans have less of a talent pool from which to choose potential ”quality" challengers. In many states the lower level political offices are dominated by Democrats. This puts the Republican Party in a position of scarce resources in terms of individual candidates. In terms of contributions and expenditures, the Republicans have a financial advantage over the Democrats. Given the campaign finance rules that exist for political parties, the Republicans are in a position to finance to the full amount any candidate, challenger or incumbent, who has the remotest chance of winning. For the Democrats, operating at a financial disadvantage, and to whom money is a scarce resource, it is a completely different picture. Given that this is the case, it is very difficult to determine set strategies for the Republicans in terms of how they respond to certain considerations. The Democrats, operating under constraints, allow for a greater definition of strategic behavior. Scarcity and non-scarcity of resources is an important determinant in the definition of strategies, and it is not considered in Jacobson's symmetric formulation of party strategic behavior. 48 The fourth and fifth considerations relate to two main sets of determinants for party organizational behavior: individual candidate choice and national conditions. v a d d o ce as et t Candidates for a party are a resource themselves, specifically ”strong” or ”quality candidates". While evidence of party recruitment patterns are on the rise (Herrnson 1986), the strategic decision of whether to run for office is still basically an individual decision. It is important to remember however, that this individual office-seeker is a crucial resource in the party organization, and therefore represents a strategic decision by the party organization. The choice of individual candidates on whether to run for office is defined as a separate decision procedure. It is hypothesized that parties at the collective level respond to what is taking place at the individual level. The decision of a potential "quality" challenger affects the magnitude of the monetary resources of the party organization. In that these individual decisions are made separately, they possess a different set of determinants than the collective choice. This also leads to non-symmetry in terms of collective strategies. The aggregation of individual choices displays a collective representation, but it can in no way be interpreted as a centrally controlled party decision, acting in response to whatever specific conditions exist for that particular election. e nal Co d t s a eter i t The final consideration of importance in defining the non- symmetric position is the role of national conditions. With 49 maximization of seats the party goal, both short-term forces such as national conditions and long-term forces such as party identification play an important role with both the majority and minority parties. Given the previous considerations, however, it is important to distinguish between majority and minority parties in their response to national conditions. Majority parties place more importance on long- term forces, and it is in their interest to shield themselves from potentially volatile short-term conditions. Conversely, minority parties place a greater emphasis on short-term forces like national conditions. The goal of gaining control places them in a position to try and take advantage of what short-term conditions provide them. As a result, minority parties place a greater emphasis on national conditions than majority parties. These differences are evident in terms of how the parties respond to national conditions in terms of resource allocation. These five conditions specify the difference between the symmetric and the non-symmetric position. e o No - et osition on t ate e The non-symmetric position results in several implications in terms of strategies. First and foremost is that there is a strategic difference between the parties. They respond differently to similar sets of determinants. The majority party responds to advantageous national conditions in a different way than the minority party would respond to similar advantageous conditions. The second implication is that the "efficiency" of an organization depends in part on the ability to acquire resources and the extent to 50 which the office-seekers control the organization. In this way the parties differ. The Republicans at the collective level do not work from a position of scarce resources, whereas the Democrats do. Given that this is the case, the common notion of ”offense" and "defense" if it exists should be more evident with the Democratic party. At the individual candidate level, the roles are reversed. The Republicans operate from a position of less resources than the Democrats. Likewise, evidence for "offensive" or "defensive" strategies if they exist, should be evident with the Republican party. C. The Evaluation of National Conditions Procedure The theoretical literature on individual voting offers a conceptual distinction between ”retrospective" and "prospective" voting. (Fiorina 1981) The theory of retrospective voting is based on the fact that voting behavior reflects the past and that elections "constitute a collective evaluation of pgpmml trends from past to present." Prospective voting emphasizes the future and focuses on ”ppoposed trends from present to future." (Fiorina 1981, 6) Fiorina’s differentiation of voters' evaluation procedures is translated for our purposes into an evaluation procedure for party elites. Jacobson and Kernell's theoretical presentation offers a third alternative. The "strategic choice" procedure emphasizes ggpmgl conditions at the ppgggpp time only. Their position is that individual candidates and contributors make decisions based upon a perception of how national conditions are at the present time. These conditions will in turn dictate their decisions on whether to run or contribute. The evaluation procedures differ in their presumptions about the character of the actors. The strategic choice procedure presumes a more 51 modest view and demands much less of the party elite than the prospective position. The prospective position presents the elite as capable of making calculated decisions based upon higher levels of information than the individual voter does in a retrospective procedure. The decision calculus for the party elite more closely represents a prospective procedure for the evaluation of national conditions. The individual candidates and contributors are making decisions based on their perceptions of future conditions or of how trends are dictating what the future may be at election time. D. The Determinants of Strategic Behavior The difference in terms of determinants is not as sharply defined perhaps because the party organizational model is in many ways an extension of the Strategic Choice position. In none of his writings does Jacobson ever specifically delineate a set of determinants for party resource distribution. However, in order for the Strategic Choice position to be viable theoretically, the dominant determinant has to be national conditions. In terms of strategies, for the offense and ' defense metaphor to have any viability, national conditions must play a dominant role. The party organizational approach differs by hypothesizing three sets of determinants. The structure of political opportunities interacts with the party system to define organizational behavior. As a result, the individual "quality" candidate deciding whether to run for office is an important determinant for both the party of choice and the opposing party. The party system is divided into two levels: the district level and the national level. The important considerations are those that define the competitiveness between the parties. At the 52 district level such variables as party strength in the district and various party organizational efforts are hypothesized to be important; and at the national level, a measure of national conditions are hypothesized to be an important determinant. It is in this final way that this study approximates the theoretical contentions of Jacobson and Kernell. Conclusion While this thesis is based upon a party organizational approach, it is nevertheless very similar in its precepts to Jacobson and Kernell’s theory of congressional elections. Differences do exist however and the focus of the theoretical hypotheses is on these differences. Several points are important to remember. Parties operate in a market system and thus are products of the environment of the system. This allows us to more sharply specify a total set of determinants. Also, it is best to study the activities of parties from a non-symmetric perspective. This results in a clearer definition of strategies and allows us to avoid many of the theoretical and practical inconsistencies of Jacobson's symmetric approach. CHAPTER 3 Theoretical Hypotheses Introduction The theoretical discussion of the preceding chapter allows us to infer several hypotheses. The hypotheses are presented in four sections. The first section emphasizes the structure of political opportunities as an important determinant for party collective behavior. The second section provides a discussion of the evaluation procedures for national conditions. The third section presents hypotheses related to the symmetric position, including the conventional wisdom of the offense and defense metaphor. The fourth section presents hypotheses related to the non-symmetric position. Parties in the Electoral Market Structure Party organizational behavior comes as the result of the interaction of two variables: the structure of political opportunities and the party system. The party system is characterized by interparty competition, which exists at the national level and the district level. The greater the degree of competition, both at the national and district level, the greater the effort of the party organization, both at the individual and collective level. 53 54 Within the electoral market a competitive relationship exists between the two major party organizations. Downs's concept of the “majority principle" and the coalition of minorities argument is based, at the issue level, on parties adopting platforms that result in winning majorities. At the issue level the competition between parties is represented by a causal structure which is not hierarchical, meaning that the time sequences are indeterminant. The same theoretical construction is used at the party organizational resource level. Party campaign activity, which is the endogenous variable, interacts with other endogenous variables, specifically the campaign activity of the opposing party. By postulating an electoral market structure, a party organization's strategy influences the other party organization. In this framework, the strategic behavior of one party is based in large part on the strategic behavior of the other party. This reciprocal relationship is evident at both the individual and collective levels of organizational activity. The construct of the party system focuses on the interaction between the two major parties. These organizations operate in an open system, subject to environmental characteristics, and the activities of opposing parties are an important aspect of this environment. The action and reaction to the strategic behavior of opposing party organizations is of major importance for the maintenance of a political party. It is therefore also an important factor in the determination of electoral and resource strategies. This behavior should be evident at the collective level and party organizational efforts are specified as simultaneous. 55 The Role of The Structure of Political Opportunities The theory specifies that this reciprocal relationship should also exist at the individual level. At the individual level, two different considerations are involved. In terms of inter-party competition, individual choice should be a significant determinant. A party organization should respond positively at the collective level to the opposing party offering a ”quality" candidate. In offering financial assistance to an incumbent, the party collective unit contributes more if the incumbent is encountering a serious challenge. In terms of intra-party behavior the decision of a quality challenger should also be a significant determinant. A party organization upon receiving the benefits of or recruiting a "quality" challenger should in return contribute greater financial assistance to that challenger. This maximizes the chance of that party actually overtaking that seat held by the opposing party. The individual choice reflects the construct of the structure of political opportunities and is hypothesized to be a significant determinant for both Republican and Democratic party collective efforts. Parties respond to individual choices whether they are taking place in the opposing party or their own party. The following hypothesis takes into account these considerations. r tu e of Po 1 a1 or t H oth i : The contributions and expenditures of the collective party organizations are a function of the decision to run of individual "quality" candidates. This variable should be a significant determinant when it is the party's own candidate and when it is the opposing party's candidate. 56 Evaluation Procedures for National Conditions An important area of differentiation between the two theoretical positions relates to characteristics involving the evaluation of performances. The Strategic Choice Model stipulates that current conditions and performance are the basis for evaluations at the party elite level. Fiorina (1981) in his theory of retrospective voting suggests that voters make evaluations based upon past conditions and performance. We in turn argue that party elites in their decision processes make judgments on future performances. Marra and Ostrom (1987) characterize future evaluation as the "extrapolation" of past and current performance. Their argument is based upon the fact that the environment is subject to change, and that sufficient evidence exists to suggest that presidential popularity and support is not subject to inevitable decline. While the evidence for the inevitability of decline is subject to question, events do take place which can cause shifts in the level of support for a president. As a result, it can safely be said that enough variation exists to show that there are periods of rise and decline, even if this exists in an overall setting of decline. The key is that presidential approval is subject to change. Since existing research indicates that the president's approval reflects voter satisfaction with the overall state of the environment, ghppggg in the level of approval over time provide one possible indicator of whether voters perceive environmental events and conditions to be getting better, staying the same, or getting worse. It is important to note that declining public support is not inevitable. (Marra and Ostrom 1987, 8) The "retrospective”, "strategic choice", and "extrapolation" positions are based primarily at the voter level. In this thesis, we apply the "extrapolation” principle to the behavior of party elites in the same way that Jacobson and Kernell apply the Strategic Choice 57 position to contribution strategies. The model specifies the extrapolation position at the individual and collective level of the party organization. The extrapolation position more accurately reflects the evaluation of performance of national conditions. In effect, the individual candidates and party decision leaders are speculating when they make their decisions on what the future conditions will be at the time of election. These actors are rational enough to realize that environmental conditions change over time, and that these changes can help or hinder their electoral chances. Jacobson and Kernell's theory states that individual candidate and contributors' strategies are based on a decision calculus which suggests the actors estimate election day conditions by substituting the relevant conditions at the time the decisions are made. Jacobson and Kernell hypothesize an average of economic and presidential popularity measures as characteristic of the decision processes of all elites. Party elites focus more on the degree to which performance evaluations have changed. Measures are needed which reflect a perception of things getting better or worse. The time periods provided by Jacobson and Kernell's theory for strategic decisions have limitations. They argue that for all strategies -- individual and contributions -- springtime economic and political performance represent the conditions on which elites base their decisions. In terms of contribution strategies, Jacobson and Kernell mainly look upon them as reinforcing the strategic decisions of individuals. At both levels, however, the time period is far too limited and does not provide an accurate estimate for the time considerations involved. Previous research has already emphasized the 58 limitations of this time period. Born (1986) in fact suggests that an average of an eighteen month time period is a better measure of the conditions that elites respond to in their individual calculi. A three month time period is far too restrictive for individual candidate choice. It is also inappropriate as a measure reflecting the contributional strategies of the collective party organization. A direct test of these competing positions should only be presented at the individual candidate level. We propose an ”Extrapolation Hypothesis" -- a change in conditions over an extended period of time -- as a measure which more accurately reflects how party elites view public satisfaction in the future. Three hypotheses follow from this discussion: Strategic Choice Hypothesis: The strategic choice of individual candidates deciding to run for office is a function of the average springtime popularity for the president. Extgapolation Hypothesis -- Indivldual Level: The strategic choice of individual candidates is a function of the change in presidential popularity from the third quarter of the non-election year through the first quarter of the election year. Emprepolaglop Hypothesis -- Qelleeplve Level: The strategies involving contributions and expenditures of collective party organizations is a function of the change in presidential popularity from the third quarter of the non-election year through the third quarter (time of election) of the election year. The Symmetric Hypotheses The final theoretical consideration results from looking at party distributional strategies in their specific role as resource contributors. Jacobson (1985-1986) admits that the party has an independent effect on outcomes and looks specifically at party organizations and the distribution of resources for the congressional year 1982. Herrnson (1987) analyzes the distribution of party resources 59 for the congressional election year 1984. Both Jacobson and Herrnson rely on a single measure of competitiveness to determine how "efficient" the respective parties are in the use of resources. Both of these studies, Jacobson in particular, make intuitive assumptions about the relationship between national conditions and party strategic behavior. Jacobson's interpretation of the distribution of party resources and the effects it had on the election of 1982 is not explicated by the principles of the Strategic Choice Model. Jacobson suggests that the Republicans initially perceived national conditions to be favorable and thus expended a great deal of resources in the recruitment of quality challengers. ”Party officials recruited vigorously in 1981, when the administration was riding high and a rosy Republican future could be envisioned." Jacobson also adds that "[t]hey worked successfully to hold on to most of those who had signed up even as conditions turned bleak. As a result, Republicans fielded their most experienced group of challengers since 1972." (Jacobson 1985-1986, 613-614) Following a i change in conditions which proved favorable to the Democrats, the Republicans were put into a quandary: either respond to the conditions and cease supporting these quality challengers, or continue the support and attempt to offset the effect of the conditions. Jacobson states that the Republicans chose the latter strategy. He argues that by choosing this strategy the Republicans were able to offset a potentially large number of seat losses. While this interpretation might have some feasibility in terms of actual events, the Strategic Choice Model does not account for this type of activity. It is hypothesized that at the collective level as well as the individual level, the extrapolation 60 position extended over a relevant time period allows for the parties to adapt to changing conditions. The 1982 congressional election scenario involves the specific actions that party organizations take in response to favorable or unfavorable conditions. Theoretically, the symmetric position suggests parties should attempt to maximize positive national conditions; it is at this time, according to Jacobson and Kernell, that parties should be ”playing offense," -- contributing a greater proportion to challengers. Conversely, when unfavorable conditions exist, parties should be on the defensive, -- contributing a greater proportion to incumbents. The conventional wisdom suggests that the offense/defense metaphor is symmetric, and can be stated as follows: positions taken by party organizations in response to favorable and unfavorable national conditions are congruent: favorable conditions lead to "offensive" behavior, unfavorable conditions lead to ”defensive" behavior. The Offensive Strategy Hypothesis: Party collective organizations respond "offensively" -- greater resources to challengers -- when the change in presidential approval is favorable for their party. efens ve trate o hesis: Party collective organizations respond "defensively" -- greater resources to incumbents -- when the change in presidential approval is not favorable for their party. This rationale is a much more plausible strategy in terms of individual candidate choice than in terms of party collective choices, where the former is much more concerned with a scarcity of resources than the latter, specifically in terms of the Republican Party. The limitations of this analogy are apparent in Jacobson’s description of the events for the 1982 elections. Given his account, it seems to have been beneficial to the Republicans to have responded "offensively" to 61 unfavorable conditions -- they were funding challengers during a period of unfavorable conditions, a time when the conventional wisdom suggests you protect incumbents. Not only does this go against the conventional wisdom, the results seemed to have been beneficial to the Republicans, and it therefore gives the appearance of a rational strategy. The Republicans were able to maximize seat distribution by adopting an irrational strategy. Minimizing unfavorable conditions does not imply that parties cease to expend resources. The offense/defense metaphor should not be defined symmetrically. If the metaphor has any application, it only applies in situations of scarce resources -- situations such as the supply of individual "quality" candidates. The contextual structure of campaign financing and the finite distribution of monetary resources does not allow for a correct application of the metaphor. A minimal application only applies given the scarcity of resources. The offense and defense metaphor as applied to collective party contribution strategies should be analyzed in a non-symmetric fashion. The contextual structure of campaign financing and the importance of non- symmetric determinants permit the parties to respond ”offensively" to unfavorable conditions. As previously hypothesized, collective party behavior is in part a function of the individual choice processes of potential quality candidates. Collective party behavior is also a function of national political conditions. The former of these considerations is a constant, whereas the national conditions as they take place in the collective time period are subject to change. Independent collective behavior is 62 therefore greatly restrained by the aggregate activity which has taken place at the individual level. Independent collective behavior depends in large part on the extent of the changes taking place in terms of favorable or unfavorable conditions at the national level. Unless significant variation is taking place at the national level, the collective party decision units in large part are basically responding to district level considerations -- candidates in competitive districts and those who possess previous electoral experience. The semblance of offensive and defensive behavior at the collective level is in large part an indirect reflection of the effect of individual level determinants on the choices of individual candidates. This explication also accounts for party activity in the 1982 congressional elections and explains some of the theoretical difficulties that Jacobson encountered. The Non-Symmetric Hypotheses The non-symmetric position emphasizes the differences in strategies between majority and minority congressional parties. We deduce from the symmetric position a congressional party goal of seat maximization, while the non-symmetric position contains a different goal structure . The principle of marginal utility leads us to conclude that the goal of the majority party is not seat maximization. Once a certain undefined threshold is met, individual congressmembers of the majority party no longer concern themselves with gaining more seats. A more appropriate goal or strategy for the majority party, especially with a 63 substantial seat lead like the present Democratic party, is maintenance of the seat advantage. Concern for seat maximization only exists for the majority party when it is in danger of losing this majority. Jacobson counters this argument by suggesting that party factionalism exists to the extent that a "reasonably safe" seat majority is not sufficient for party policy legitimization. ...at existing levels of party competition, every additional seat is valuable. Even though House Democrats may not expect their majority status to be seriously threatened, the larger their majority, the stronger they are as a party in confronting the conservative coalition or a Republican president wielding the veto. (Jacobson 1985-1986, 604) Historically, the alliance between Southern Democrats and the Republicans has played an important role in congressional voting. But its importance has been on a steady decline since the 19605 and 19705 when it was at its most influential. In 1987 there were fewer appearances of the coalition than in any year since Congressiepal gmapterly began keeping track. (Davis 1988, 110) The coalition is not nearly as relevant as it has been in the past, and it is questionable how seriously an individual Democratic congressmember takes this into consideration when thinking about a congressional election. Also, Jacobson's seat maximization argument does not differentiate among seats. By a strict definition of the symmetric position, Democratic seat maximization also includes the Southern Democrats who are causing the problem. A more extensive study would have to include an analysis of money distribution by region or ideological orientation. If party leadership is concerned about the conservative coalition, Southern Democrats or potential coalition members will be treated differently than non-conservative Democrats. 64 The majority and minority parties differ in their response to national conditions. The minority party attempts to take advantage of whatever the conditions offer. They also try to minimize the impact of unfavorable conditions to a much greater extent than the majority party. This does not mean that they play defensively. Ceasing to expend resources on quality challengers is an irrational strategy. The minority party, however interested in obtaining an eventual majority, expends its resources on incumbents and challengers. This is especially true given the non-scarcity of monetary resources of the Republican party. Two important factors define the non-symmetric position on party strategies. The first is environmental, the second is institutional. The enviromental condition is scarcity of resources. The institutional condition is majority or minority status. The party's status on each of these conditions dictates certain types of behavior. Non-scarcity of monetary resources suggests a party is not in a position where a strategic use of resources is imperative. Scarcity of resources results in a greater need for strategic behavior. The Democratic party operates under the condition of scarce monetary resources while the Republican party does not. Therefore, under the scarcity condition the Democratic party will be more strategically oriented. The institutional condition results in a different characterization. We hypothesize that parties with majority status operate in a different manner than parties with minority status. Once again, majority parties concern themselves with maintaining their majority. Individual members of a majority party place greater emphasis 65 on remaining autonomous from the collective whole. As a result, less strategic behavior takes place in terms of trying to maximize the total number of seats. Individual members of the minority party respond somewhat differently. The collective goal of maximizing seats and gaining a majority weigh more heavily in their individual calculus. Minority party members perceive the subgoal of autonomy in a different way than majority members. Autonomy is less important and collective behavior is easier to achieve in a party with minority status. The minority party does not have the task of passing a legislative agenda, and by definition the size of the contingent is less than the majority. Therefore, the Republicans as the minority party will behave more strategically at the collective level. The institutional status of being in the majority does not make strategic behavior at the collective level as achievable or evident. The scarcity of resources condition dictates non-strategic behavior for the Republican party. The condition specifying majority party status suggests less collective strategic behavior for the Democratic party. These two positions serve to cancel each other out in terms being able to explain one party’s strategic advantage over the other's. The non-symmetric position leads us to argue that strategic behavior, specifically in terms of responding to national conditions, is not to be completely evident from either party. If this is the case, the major determinants for strategic action appear from either the structure of political opportunities, or some district level characterizations. The preceding extrapolation hypotheses test for the type of decision procedure which best measures party decision processes. In 66 this section we hypothesize that no coherent or identifiable strategic behavior will be evident based upon national conditions. Arguing from a non-symmetric perspective, we hypothesize that the two other sets of determinants, individual choice and district level conditions, play a far more important role in the determination of strategic behavior. Conclusion The picture on both the theoretical and strategic level is nonsymmetric. The nonsymmetric position takes into account several considerations that the symmetric position does not. The role of several determinants and how they interact along with differentiating between majority and minority status suggests that parties do not, and in some cases cannot, respond in a similar fashion.. Finally, the consideration of scarce resources leads to a different description of the strategic behavior of party organizations. The Operationalization of the constructs along with the specification of the models is the subject of the next chapter. CHAPTER 4 The Models of Party Organizational Behavior Introduction The preceding chapters have developed a theory of the strategic behavior of party organizations. In order to quantify the theoretical developments, constructs are developed in a way which relate to their empirical correlates. By developing descriptive statistics a body of observational knowledge is developed against which the theory can be checked and from which parameters can be estimated. It is in this way that inferences can be drawn about actual political behavior. This chapter includes a discussion of the relevant theoretical constructs, the Operationalization of these constructs, the presentation of the models, and a discussion of the statistical inference techniques from which the inferences are drawn. From Theory to Model The primary function of the model of party organizational behavior is to explain the behavior of both the individual office-seeker and the decision unit for the collective office-seekers. These models will concentrate on the distribution of resources in the party organization with the emphasis on the resources of individual quality candidates and monetary funding. The strategic decisions of individual office-seekers 67 68 and the strategic decision of the party organization on whether to provide their connective contributions -- monetary assistance -- are the decisions explained by the models, and are the endogenous variables. The determinants of organizational behavior are in part assumed given from outside the model, and are the exogenous variables. Since the theory suggests an interactive effect between the levels of strategic choice, our model suggests that certain endogenous variables may also serve as exogenous variables. Also, given that the analysis concerns more than one strategic decision within the party organizational model, the first and most theoretically complete model involves a system of equations, with each equation representing an organizational strategic choice. The second major condition which the model must incorporate is the interaction between the structure of political opportunity and the party system. Individual strategies focus on the structure of political opportunities. Two key factors affect this structure: first, a two- party system dictates that an office-seeker must take into account the competitive chance of the party for which he or she decides to run. Second, the structure of political opportunities is not fixed, meaning that national political and economic conditions play an important role in the ”shifting tides of party strength.” Individual strategic decisions and the accompanying individual strategic calculus must take into account political party fortunes. This calculus reflects two major environmental characteristics: first, the predicted party fortunes, which are measured in part by district-level party competitiveness measures, and national political and economic conditions; and second, the calculus must reflect the interaction within party organizational 69 behavior which will also be determined in part by the resource considerations of both parties. Collective party strategies reflect their position within the market-based structure. As with the individual strategies, collective party strategies -- measured in terms of party expenditures -- also respond to the previously mentioned environmental considerations. Collective party strategic decisions are based upon district-level competition, resources in terms of expenditures, and national political and economic conditions. Given that an interaction exists between the individual and collective responses a final determinant of the collective response is the initial individual strategy. Parties attempt to recruit and are rewarded by quality challengers, therefore party resource strategies depend on the quality of a challenger. The final determinant for collective strategies involves the interaction between competitive party organizations in the electoral market. The theory suggests that parties will respond to both levels of opposing party organizational efforts: the individual choice of whether they will oppose a quality or non-quality challenger, and the opposing parties’ monetary resource expenditures. Given that the competitive party organizations are acting in zero sum market game, a major determinant of behavior is the strategic behavior of the opposing organization; subsequently, within the collective party organizational block a simultaneous relationship between organizational efforts is proposed. Party organizational behavior is analyzed by two models, one which is a block-recursive equation system, the other is a two-equation cross- sectional model. Block-recursive systems are groups of equations which 70 can be broken up into groups or blocks of equations. What characterizes a block-recursive system is that equations within each block are simultaneous but the groups of equations across blocks are recursive; which means that estimation of the endogenous variable in the first block permits the estimation of the endogenous variables in the second block. The reverse is not true however, because they are not reciprocal. (Pindyck and Rubinfeld 1981, 323) Operationalization of Constructs The theory of party organizational strategic behavior is based on three major constructs: the structure of political opportunities -- which embodies individual strategic behavior, the party system -- characterized by district and national level conditions, and party collective strategies. This section provides a description of the operationalizations. Sources for the specified measures are provided in the appendix. A. The Structure of Political Opportunities The focus in this study is on the choice procedures of individual candidates, specifically the decision of potential ”quality” challengers. Quality challengers are also defined as a party resource and consequently are an integral part of the total party’s strategic behavior for the election in question. The structure of political opportunities construct is represented by a single dichotomous variable, does the specific district have a quality challenger running against an incumbent, or is there a non- quality challenger running against an incumbent. This 71 Operationalization allows the inclusion of only elections in which an incumbent of one of the two major parties is opposed by a challenger of the other major party, all open seat elections are excluded. While it may be possible to quantify certain structural characteristics of the structure of political opportunities (Canon 1985), this study focuses primarily on the individual strategic decision, and its implications as a resource in the organizational model. A final theoretical consideration resulting from this construct is that in Model I it will be specified as an endogenous variable, whereas in Model II it will be specified as strictly an exogenous variable. The theory as constructed hypothesizes that it should be measured endogenously, the results from the two models will be compared in this light. Y1 - Non-Quality Challenger/ Quality Challenger Measurement Rule: quality measured by previous electoral experience and positions of visibility (broadcast journalism) - - 0 if non-quality challenger, 1 if quality challenger (Note: Sources for the data are provided in Appendix A) B. The Party System The party system reflects the competitive interaction between parties. Organizational strategic behavior is determined in large part by various aspects of this competitiveness; in general, the greater the degree of competition the greater the organizational effort. The party system and this competitive interaction is measured at two levels, at the district level and the national level. The district level variables reflect competitive characteristics which are district and candidate specific whereas the national level reflects the economic and political conditions for which one of the two parties is held responsible, thereby 72 establishing its competitive position at the national level. The party system in the context of this model is only measured exogenously. ev d t ons The district level competitiveness measures focus on two characteristics: (1) the relative strength of the party in the congressional district, and (2) the relative ability of the incumbent to get reelected. These factors are important in terms of individual strategic behavior -- a quality challenger deciding whether or not it would be to his advantage to run for a district seat at that particular time, and for party expenditures -- the party decision unit deciding whether or not making a contribution is an efficient use of resources, and after deciding whether to contribute, deciding how much to contribute. The party’s competitive position in the district and the status of the incumbent are two important factors in the determination of individual and collective strategies. Variables X1, X2, X5, X6, X9, and X10 are various measures for the incumbency variables. Variables X3 and X4 are the exogenous variables for quality challenger, and variables X7 and X8 are the district level party strength measures. X11 and X12 represent the non-party expenditures for each specific district. X1 - Republican Incumbent Measurement Rule: 0 if Democratic incumbent, 1 if Republican incumbent X2 - Democratic Incumbent Measurement Rule: 0 if Republican incumbent, 1 if Democratic incumbent X3 - Quality Challenger -- Republican Measurement Rule: 0 if non-quality or Democratic quality challenger, 1 if Republican quality challenger X4 - Quality Challenger -- Democrat Measurement Rule: 0 if non-quality or Republican quality challenger, 1 if Democratic quality challenger 73 X5 - Number of terms held by incumbent Measurement Rule: number of terms X6 - Party of Incumbent Measurement Rule: 0 if Democrat, 1 if Republican X7 - Republican Party Strength Measurement Rule: Republican presidential vote of nearest preceding election for that particular district X8 - Democratic Party Strength Measurement Rule: Democratic presidential vote of nearest preceding election for that particular district X9 - Number of terms held by Republican incumbent Measurement Rule: number of terms, 0 otherwise X10 - Number of terms held by Democratic incumbent Measurement Rule: number of terms, 0 otherwise X11 - Total Non-Party Political Committee Contributions: Republican Candidate Campaign Measurement Rule: summation of corporation, labor, non-party political committees which have not reported a connected organization, trade/member/health, cooperative, and corporation without stock contributions -- in hundreds of dollars X12 - Total Non-Party Political Committee Contributions: Democratic Candidate Campaign Measurement Rule: summation of corporation, labor, non-party political committees which have not reported a connected organization, trade/member/health, cooperative, and corporation without stock contributions -- in hundreds of dollars Neplppel Level Conditlone The theoretical orientation in this thesis suggests that district specific conditions are not the only consideration in the strategic behavior of party elites. National forces and the party in power ' responsible for the political and economic conditions also plays a crucial role. Schlesinger’s theory, as previously stated, proposes party organizational behavior at three interdependent stages. Party activity at the governmental phase affects the party organizational activity at the election phase. In terms of individual strategic 74 behavior, potential quality challengers will take into consideration the relative position of the party in terms of being held accountable for political conditions. Once again, this thesis suggests that political conditions and the strength or weakness of the party in connection with these conditions, is also an important determination in the distribution of financial resources. The relative political and economic conditions in terms of party responsibility are measured by the popularity of the sitting or incumbent President of that party for that election. The majority of studies which use this measure also include measures for economic conditions which include such variables as unemployment and inflation rates. It has been suggested that the presidential approval variable in effect is measuring the same thing or at least including the same considerations that are present in the economic variables. Various studies have focused on the determinants of public support, and there is evidence that the economy is a very important determinant of presidential public support. (Hibbs 1982; Kernell 1978; MacKuen 1983; Stimson 1976) Another study (Ostrom and Simon 1985) introduces the notion of a reciprocal relationship between public support and some of its determinants and focuses on the interdependence of the determinants. Presidential approval ratings are therefore used to measure national conditions for both specification and statistical reasons. The national level measures are as follows: variables X13 and X15 represent the Strategic Choice position. Variables X14, X16, X17, and X18 represent the Extrapolation position. X13 - Average Presidential Approval Rating for Democrat in Power Measurement Rule: average rating -- October of non-election year through March of election year 75 X14 - Change in Presidential Approval Rating for Democrat in Power Measurement Rule: change in rating -- October of non-election year through March of election year; highest to most recent X15 - Average Presidential Approval Rating for Republican in Power Measurement Rule: average rating -- October of non-election year through March of election year X16 - Change in Presidential Approval Rating for Republican in Power Measurement Rule: change in rating -- October of non-election year through March of election year; highest to most recent X17 - Change in Presidential Approval Rating for Republican in Power Measurement Rule: change in rating -- third quarter to third quarter; highest to most recent X18 - Change in Presidential Approval Rating for Democrat in Power Measurement Rule: change in rating -- third quarter to third quarter; highest to most recent C. Party Collective Strategies The party collective strategies reflect the organizational activity of the decision unit within the party that acts for the collective whole. This party organizational behavior is measured by the party expenditures for a congressional campaign, both direct and coordinated. Direct contributions represent the finances given directly to a particular candidate, while coordinated expenditures represent financial assistance on behalf of the candidate. These contributions and expenditures are primarily coordinated nationally and represent the party’s strategies in terms of the viability of the candidacy and the probability of winning the seat. The decision process consists of whether or not to contribute or spend, and deciding if they will, how much the contributions and expenditures will be. Model One focuses on the attempt to specify the interrelations between the individual and collective strategies, while Model Two focuses explicitly on party collective strategies. Both models however use party contributions and 76 expenditures endogenously and exogenously, producing a system that reflects simultaneity. Y2 - Republican Party Expenditures Measurement Rule: summation of direct contributions and coordinated expenditures -- in hundreds of dollars Y3 - Democratic Party Expenditures Measurement Rule: summation of direct contributions and coordinated expenditures -- in hundreds of dollars The Models of Party Organizational Behavior Party organizational behavior is measured with two separate models. Model I integrates the individual strategic equation and provides for national level political conditions as exogenous variables. In order to do this, the data for five elections -- 1976, 1978, 1980, 1982, and 1984 -- are pooled. Model 11 allows for cross-sectional analysis of the five election years in question, with individual strategic choice presented as strictly an exogenous variable. The Y’s represent endogenous variables with Y1 specified as the individual decision by a potential quality candidate, a reflection of the structure of opportunities. Y2 and Y3 represent the collective party unit decisions on contributions and expenditures. Endogenous variables on the right hand side (RHS) represent instances where instruments are used for issues involving simultaneity. The X’s represent the exogenous variables which in general reflect party system competitiveness characteristics. Variables X1 through X12 represent various forms of district level conditions, whereas variables X13 through X18 represent various forms of national level political conditions. 77 A. Model I: A Three Equation Pooled Cross-Sectional Block-Recursive System Individual Strategy Equation Y1" blO t b15X5 + bl6x6 + b17x7 + b1(13)x13 + b1(14)X14 + b1(15)x15 + bl(l6)xl6 (4'1) Collective Strategy Equations Yz' b20 t a21Y1 + a23Y3 + bzexa + b27x7 + b29x9 + b2(10)X10 + b2(12)x12 + b2(17)x17 + b2(18)x18 (“-2) Y3' b3o + a31Y1 + a32Y2 t bssxe + basxa t b39x9 + b3(10)x10 + b3(11)X11 + b3(17)x17 + b3(13)X18 (“13) Model I is a theoretically more complete model that specifies individual choice as an endogenous variable, allows for the interactive effects, and allows national level conditions as part of the explanatory variables. Since the national conditions as measured by the change in presidential popularity do not vary over a single election year, the model is pooled in order to provide for this important factor. Model II represents a simple two-equation simultaneous model that is estimated for the individual election years. The explanatory variables reflect the district level characteristics of the party system and the interactive effects of the electoral market system. The distinction between the two models involves another important consideration. By arguing that national conditions should be included as explanatory variables in order for the model to be theoretically complete, in effect the argument is that Model 11 is misspecified. By omitting explanatory variables, ”the influence attributed to the included variables is actually a combined influence of the included and excluded variables.” The result is biased estimates, and the ”biases in 78 the estimation become more severe as the excluded variable[s] becomes more important in explaining Yt.” (Hanushek and Jackson 1977, 81,82) Given the signs of the correlations between the excluded and included variables, the possible result is biased and inconsistent estimators; or at best, the estimator of the variance of the included variable will contain an upward bias, which affects the tests of significance and confidence intervals. (Kmenta 1971, 394) These problems potentially exist for Model II. As a result, the specification of Model I is argued to be a more complete estimation and is the focus for the hypothetical considerations to follow. Model II allows for the quality challenger variable to be measured exogenously and also disaggregates the five election years in order to account for possible yearly differences. B. Model II: A Two-Equation Cross-Sectional Model of Party Expenditures Yz' blo t a13Y3 + b11X1 + b13X3 + bl4x4 t b17x7 + b19x9 + b1(10)x10 + b1(12)X12 (4.4) b2(10)x1o + b2(11)X11 (4-5) where: Y1 quality challenger/ non-quality challenger Y2 Republican Party Expenditures Y3 Democratic Party Expenditures X1 Republican incumbent X2 Democrat incumbent X3 quality challenger/ Republican Kg quality challenger/ Democrat X5 Number of terms held by incumbent X6 party (0 if Democrat, 1 if Republican) X7 Republican party strength in district X8 Democratic party strength in district X9 Republican incumbent -- number of terms held in House X10 Democrat incumbent -- number of terms held in House X11 total non-party contributions to Republican candidate X12 total non-party contributions to Democratic candidate 79 X13 average presidential approval rating for Democrat in power, candidate decision time period X14 change in presidential approval rating for Democrat in power, candidate decision time period X15 average presidential approval rating for Republican in power, candidate decision time period X16 change in presidential approval rating for Republican in power, candidate decision time period X17 change in presidential approval rating for Republican in power, party decision time period X18 change in presidential approval rating for Democrat in power, party decision time period C. Data-Related Issues The Emdogemous Variables Three endogenous variables are measured in these models. Two variables Y2 and Y3 represent the collective party strategies which are operationalized by party expenditures. This Operationalization combines two different measures, the direct contributions that parties are allowed to give to individual campaigns, and party expenditures on behalf of the campaign. The summation of these two serve as the measure for collective organizational strategy, which is the same measure used by Jacobson (1986) in his analysis of party resource activity. The third endogenous variable Y1 is a dichotomous variable representing the ”quality” / ”non-quality” challenger. While most previous studies have not considered a quality candidate in terms of a party resource, there is general agreement on the use of previous electoral experience as the measure of ”quality”. It is usually operationalized to take into account any previous electoral experience, with the only debate being over the level to which these considerations take place. Any candidate who has or is holding any statewide, county, 80 or city level position is considered to be a ”better quality” challenger than one who has or is not. The way the construct is operationalized in this study is defined a bit more in general terms, reflecting the concept of ”visibility” as well as quality. Political background and thus a ”quality” challenger is one who not only has or is holding statewide, county, or city office, but also includes such ”visibility” positions as broadcast journalism and organizational advantages such as congressional or party staff. This differentiation from the more commonly used measure serves more as a theoretical clarification than as an empirical distinction. The differences in the distribution of the variables is minimal. The quality challenger variable is measured dichotomously with 0 representing a non-quality challenger (one with no previous electoral, visibility, or organizational experience) and 1 representing a candidate who fits any of the previous criteria. Questions over the use of quality challenger have been of a statistical nature in terms of correlation problems with other variables (specifically expenditure variables) and the question of simultaneity. The specification of the models in this study does not allow that to be a problem. The Ezegepeps Variables The principal data-related issues involved in the exogenous variables revolve around the presidential popularity variables X13 through X18. First however a brief review of the exogenous variables X1 through X12. Variables X1 and X2 are dummy variables used to determine the party of the incumbent in Model I. They are hypothetically related to 81 the discussion of under what conditions parties contribute to challengers and incumbents. Variables X3 and X4 are the variables which determine whether the quality challengers are Republicans or Democrats. Variables X3 and X4 are used in Model II and in effect represent the exogenous versions of ”quality challenger” which is Y1 in Model 1. Variables X3 and X4 are dummy variables with 0 representing a non-quality challenger and 1 representing a Republican or Democratic quality challenger. Variables X5, X9, and X10 represent the number of terms served by the incumbents. Variable X5 entails all incumbents of both parties and therefore is used with the dummy variables representing party of incumbent X6. X5 is used for the individual strategy equation in Model I. Variables X9 and X10 differentiate the number of terms served by the incumbent by party and are used for the expenditure equations in both Model I and II. This set of variables, along with X3 and X4, are hypothetically related to a discussion of the efficiency of the two parties in terms of resource distribution. Variables X7 and X8 represent the percentage of Republican and Democratic party presidential vote within the specific districts. This is used as a measure of district level party strength. Republican and Democratic percentages are differentiated because of the Anderson candidacy in 1980. The most recent measure of presidential vote is used for the following congressional election campaign. Again, these variables are also efficiency oriented and are used in combination with the other variables to determine party distribution of resources in terms of party strength in the district. 82 Variables X11 and X12 represent the total contribution to a Republican or Democratic campaign from non-party political committees. This measure is essentially a summation of the following types of contributions: corporations, corporations without stock, cooperative, trade/member/health, non-connected, and labor organizations. Non-party along with party contributions and expenditures are measured in hundreds of dollars. Non-party contributions represent behavior taking place within the political influence market, party efforts in general represent their behavior in the electoral market. Exogenous variables X13 through X18 all represent variants of presidential approval. As previously stated, national economic and political conditions are measured solely by the presidential approval measures. Since national economic conditions (inflation and unemployment) play a major role in the determination of presidential approval it is suggested the use of economic variables along with presidential approval leads to problems of autocorrelation. There is a difference in the measures of approval from those taking an average to those which focus on the change in approval over a specific time. The change in approval is an independent measure from an average approval rating. The change in approval represents a summary of the direction approval has been moving. (Marra and Ostrom 1987) The key consideration is that it provides an indication of how conditions will be at the election time period. This is essentially what the strategic actors are weighing in their decision calculus. This thesis argues that this measure is more appropriate for individual and collective decision calculi. A measurement made up of an average over a previous period does not accurately reflect the decision calculus of party elites. The 83 average approval measure does reflect the Strategic Choice Hypothesis of Jacobson and Kernell and is of integral importance to their thesis of elite behavior. As a result, X13 and X15 -- the average presidential approval for candidate time period -- is hypothesized as reflecting the Jacobson and Kernell thesis. The competing hypothesis which reflects this thesis suggests that variables X14 and X16 -- the change in presidential approval for the candidate time period -- more accurately reflects the elite calculus. These competing hypotheses are tested in the first equation of Model I. The final issue of importance involves the difference in time periods for the measures of presidential approval between individual (X14 and X16) and collective (X17 and X18) decision-making. Jacobson and Kernell hypothesize springtime presidential approval (average approval rating for March, April, and May) as the determinant for all strategic behavior, both individual and contributors. This thesis argues that the individual strategic choice of whether to challenge reflects a time period representing the change taking place from October of the non-election year through March of the election year. In order to accurately reflect the change taking place the highest approval rating from the first quarter through the third quarter of the non- election year is used and the difference is calculated (either negative or positive) for the time period at the end of March of the election year. To more accurately measure the Jacobson and Kernell hypothesis their measure is extended to take into account the average October through March approval ratings, assuming at this time that the original March through May measure is a totally inaccurate representation. 84 The time period for party collective strategies is presented as a third quarter to third quarter (election time) measure. The measure is developed by taking the highest approval rating and determining the difference by using the most recent approval rating before the election. This provides a more accurate reflection of how the party decision unit feels national conditions will serve them in the election. Variables X17 and X18 have been developed using these time periods and serve as determinants for collective party strategic behavior. Methods of Estimation The estimation of the preceding models involves the use of several techniques. This section first offers a brief description of the techniques and their properties, starting with the Probit model and continuing with a discussion of Tobit, instrumental variables, and simultaneous equation techniques. The section ends with a description of the estimation procedures for each model. An excellent discussion of limited-dependent variable techniques is provided by Maddala (1983). Much of the following including the notation is derived from this source . A. The Probit Model The individual strategy equation (4.3) is estimated by a Probit analysis model. It assumes that there is an underlying response variable Y1* which is defined by a regression relationship '1: Y1 -in + Ui (4.6) 85 Yi* is observed as a dummy variable y with y-l 1r Yi” > o y-0 otherwise (4.7) In this formulation, in is E(Y1*|Xi). From the relations (4.6) and (4.7) we get Prob(Yi-l) - Prob(Ui < -in) - 1 - F(-bxi) (4.8) where F is the cumulative distribution function for U. (Maddala 1983, 22) In our model: y-l if the challenger is a quality challenger y-0 if the challenger is not a quality challenger The properties of the maximum likelihood estimates using the Probit procedure exhibit the asymptotic properties of unbiasedness, efficiency and normality. The use of estimated standard errors as the measure of likely variation and the t-statistic with its relevant interpretation is analogous to ordinary least squares. Goodness of fit measures can be judged either in terms of the fit between the calculated probabilities and the observed frequencies or in terms of a likelihood ratio principle which is analogous to the F-test and is also approximated by the chi-square distribution. B. The Tobit Model The probit model assumes a dependent variable which is measured in discrete values. The collective strategies in both Model I and Model 11 are represented by dependent variables which are only observed in some of the ranges. In our models we have what approximates a ”censored regression model” or a ”tobit model.” It is defined as follows: 86 Yi-O ot erwise (4.9) b is a K x 1 vector of unknown parameters; X1 is a K x 1 vector of known constants; U1 are residuals that are independently and normally distributed, with mean zero and a common variance. (Maddala 1983, 151) This model was first developed by Tobin (1958). The model is similar to Probit in that it essentially consists of making predictions about the observed Y1, with the values being divided between those equal to zero, and those greater than zero. The Tobit procedure requires that the variable to be explained Y1 must consist of a significant number of observations which are equal to zero. This is the lower limit in the single limit equation function. It is applicable for our models because the expenditure measures for the parties has a limit of zero and the non-zero observations consist of a continuous range. An upper limit does theoretically exist, but the number of cases in which it is reached does not permit a change in the estimation procedure. The decision process consists of the parties either contributing or not contributing (0), and upon deciding to contribute, how much. As Maddala suggests, the literature on limited-dependent variable models has made some gains, but not much has been made in studying the effects of the violation of certain assumptions. In terms of heteroskedasticity, Maddala and Nelson (1975) showed that if it is ignored, the estimates will not be efficient or consistent. Special assumptions must be made about the residual variances, and even when this is done, the Tobit estimates are inconsistent. (Maddala 1983, 179) 87 In terms of autocorrelation, similar properties hold for Tobit as they do in OLS, the estimates are consistent but not efficient. The interpretation of coefficients in the Tobit model are not the same as in a regression model. McDonald and Moffitt (1980) point out that the coefficients obtained from using Tobit can be used ”to determine both changes in the probability of being above the limit and changes in the value of the dependent variable if it is already above the limit.” (McDonald and Moffitt 1980, 318) As a result, the total change in Y1 can be disaggregated into two parts: (1) the change in Yi of those above the limit, weighted by the probability of being above the limit; and (2) the change in the probability of being above the limit, weighted by the expected value of Y if above. Estimated tobit coefficients can be interpreted as regression coefficients after the weighting procedure has been conducted. A more complete explication of the procedure is offered in the Appendix B. C. Instrumental Variable Techniques The block recursive system being proposed for Model I is broken up into two blocks. The equations in the second block are simultaneous, whereas the endogenous variable in the first block permits the determination of the endogenous variables in the second block. In systems such as this, it is possible for some of the desirable properties to be lost when one or more of the right-hand variables in the equation is measured with error. Some of these lost properties can be regained if a new variable, called an instrument, replaces the variable measured with error. Correlation between an exogenous variable (dummy for quality challenger) and the error term leads, in general, to inconsistent maximum likelihood estimates. The technique of 88 instrumental variables estimation solves this measurement error problem. Instrumental variable estimation guarantees consistent estimation, but it does not guarantee unbiased estimation. In the first model the equations represent the way we believe the observed data on party organizational behavior were generated. The structural representation corresponds to the theoretical models of individual and collective behavior of party organizational units. Contrary to the second model, in the first model the direct effect is not the total effect of a change in the right-hand-side variable. The hypothesis is that changes in Y1, or any of the related exogenous variables, result in changes in the other endogenous variables, Y2 and Y3. Subsequently, a change in Xi is hypothesized to affect the Y’s directly, and indirectly by any influence each endogenous variable has on the other endogenous variables. These indirect effects transmit the direct effects of a change in an exogenous variable throughout the model. Exogenous variables that are present in both blocks of equations are therefore hypothesized to have both a direct and indirect effect on collective party organizational behavior. The endogenous variable Y1 when it appears as a RHS variable in equations two and three is created as an instrument. The use of instruments and how they are created is further explained in the next section on simultaneous equation estimation techniques. D. Simultaneous Equation Estimation Techniques The distinguishing feature of simultaneous models is that the causal structure is no longer hierarchical, meaning that at least one equation contains a higher ordered endogenous variable as an explanatory variable. (Hanushek and Jackson 1977, 246) An endogenous variable (Y2) 89 is interacting with another endogenous variable (Y3). Conceptually, the time ordering is that individual choice is made prior to the collective resource decision. The data in these latter decisions are collected at one point in time, subsequently the observations on the variables subject to reciprocal influence will appear simultaneous. Therefore equations (4.2) and (4.3) in Model I and equations (4.4) and (4.5) in Model II are argued to be simultaneous. As in the case of an OLS estimator, a MLE depends on whether or not the endogenous variables used as RHS variables are distributed independently of the error term in that equation. When this error term changes, the endogenous variable it determines directly changes, which in turn changes the other endogenous variables since they are determined simultaneously. As a result, the endogenous variable used as a RHS variable is contemporaneously correlated with the disturbance term in this equation; given this, a MLE is biased. For the models, by estimating Y2 using just the single equation, the estimates will be biased because Y3 is correlated with Y2, and as a result Y3 is correlated with U2. Also, it is possible that U2 and U3 are also correlated because the implication is that Y3 and U2 are correlated, which biases the estimates. Qphgltlops for ldenplfication in Simultaneope Eguatlons The identification of the individual equations is an important step which must precede any estimation of equations in a simultaneous system. ”If an equation is unidentified, there is an infinite number of possible coefficients ... which give the same relationships among observed variables, regardless of the presence or absence of error terms.” (Hanushek and Jackson 1977, 254-255) An equation can either be 90 underidentified, exactly identified, or overidentified. If the equation is underidentified it cannot be estimated. If it is exactly identified the same number of estimations exist for the simultaneous equations. If the equations are overidentified there is more than one estimate for some structural coefficients; procedures exist to combine these estimates into a better single estimate. There exist two conditions to test for single equation identification: the rank condition and the order condition. The order condition is a necessary condition whereas the rank condition is a sufficient condition. For the order condition, the number of exogenous (predetermined) variables excluded from the equation must at least equal the number of endogenous variables included on the RHS of the equation. Kmenta (1971) provides this relatively straightforward way of determining the identification of single equations using the order condition. Given Gx' # of endogenous variables K**- # of predetermined variables If K:* > Gx-l, we have overidentification If K - G -l, we have exact identification If K * < C -l, the structural equation is underidentified. (Kmenta, 1971, pp. 544-545) “NH Given that Model 11 is a block recursive system the question of identification and the use of instrumental variables comes into question. Hausman and Taylor (1983) and Judge et a1. (1985) suggest that for a single equation like (4.3) ”covariance restrictions aid identification if and only if they imply that a set of endogenous variables is predetermined in the equation of interest.” Hausman and Taylor show that ”if the system of equations is identifiable as a whole, covariance restrictions cause residuals to behave as instruments in the 91 sense that the full system estimator can be treated as an instrumental variable estimator where the instruments include residuals uncorrelated with the equation error due to the covariance restrictions.” (Judge et a1. 1985, 580) The order condition for both Models I and II shows that the equations are identified. If Y1 is used as a predetermined variable in the second and third equations, both equations are overidentified with the number of predetermined variables equal to one and the number of endogenous minus one equal to zero. Rank condition requirements are also met but given the complexity of the procedure are not explicated here. fleeimapiph Technigues for Simultaneous Eguatlen Models Techniques for the estimation of simultaneous equation systems where some of the endogenous variables are discrete or limited have been considered by Amemiya (1978; 1979), and specifically for simultaneous Probit and Tobit models Heckman (1978), Nelson and Olson (1978), and Sickles and Schmidt (1978). The model provided by Nelson and Olson (1978) is a general simultaneous-equations Tobit model. They estimate each reduced-form equation separately by various appropriate methods and obtain the estimates of the structural parameters from the estimates of the reduced-form parameters. The derivation of the asymptotic covariance matrix of the two-stage estimates for the Nelson-Olson model was developed by Amemiya (1979). In this article Amemiya showed that the Generalized Least Squares method which he used in his 1978 and 1979 articles based on the same reduced-form estimates was asymptotically more efficient. The techniques developed by Amemiya are quite complex 92 and cumbersome and the use of them is restricted by the availability of statistical packages. The model developed by Nelson and Olson (1978) is quite simple and straightforward and can easily be adapted to the procedures needed to run Models I and II. As applied to Model I the Nelson and Olson procedure would involve the following: (1) Estimate the parameters of the equation for Y1 by the Probit MLE and the reduced-form equations for Y2 and Y3 by the Tobit MLE. (2) Replace Y1 and Y3 in the RHS of equation 3.2 and Y1 and Y2 in the RHS of equation 4.3*by the Probit predictors Y1 and by the respective Tobit predictors Y2 and Y3 . (3) Estimate equation 4.2 by Tobit MLE with the instruments Y1* and Y3 (4; Estimate equation 4.3 by Tobit MLE with the instruments Y1* and Y2 . As Nelson and Olson suggest, since reduced form estimates are all consistent, the extension of proofs for equations with limited left hand side variables would establish both the consistency of the estimates and the fact that they are asymptotically normal. The five individual cross-sectional analyses of the congressional elections of 1976-1984 were estimated by the Nelson-Olson two-stage MLE procedure. The quality of challenger dummy variables are specified solely as exogenous variables in Model II. The pooled cross-sectional analysis for Model I was estimated by the procedure specified above with the quality of challenger variable specified as an endogenous variable. The hypothetical considerations involving these specifications were presented in Chapter 3, the results are discussed in the next chapter. Chapter 5 Results and Discussion Introduction This chapter is divided into two sections. It begins by presenting the results from the estimation of Model I. The Structure of Political Opportunity Hypothesis and the Evaluation Procedure Hypotheses are considered within this model. The second section consists of presenting the results from Model 11. Evidence for and against the Symmetric and Non-Symmetric Hypotheses are considered within this model. Presentation of Results for Model I The results from the estimation of Model I is presented in the following section. Section B consists of the results for Model I with Table 1 presenting the Probit Coefficient Estimates for Equation 4.1, and Table 2 presenting the Tobit Coefficient estimates for Equations 4.2 and 4.3. The summary statistics provided are the number of cases (N), the percent correctly predicted (X Pred.) -- the percentage of the total number of correctly predicted estimates for the value zero and non-zero, the standard error of the estimate, and the correlation between the actual and predicted estimates of the dependent variables (r). 93 94 A. Model I Results Table l. Probit Coefficient Estimates -- Pooled Cross-Sectional Model Dependent Variable: Non-Quality Challenger/ Quality Challenger Independent Variables: MLE T-stat Constant -1.673 -1.229* X5 Terms Inc. -0.055 -5.337* X6 Party 0.091 2.568* X7 Rep Strength 0.014 4.071 X13 Ave. App. Dem. 0.010 0.319 X14 Ch. App. Dem. Can. -0.001 -0.111 X15 Ave. App. Rep. 0.006 0.243* X16 Ch. App. Rep. Can. 0.015 3.882 N - 1640 -2(llr0 - llrl) - 537.64 2 Fred. - .76 (Null - .76) 95 Table 2. Tobit Coefficient Estimates -- Pooled Cross-Sectional Model Dependent Variable: Republican Party Expenditures Independent Variables: ZSMLE T-stat * Constant 147.000 2.727: Y1 Qual Ch Inst 70.772 2.430* X6 Party 40.475 2.096* X7* Rep Strength -l.876 -l.969* Y3 Dem Party Exp Inst 4.965 9.217* X9 Terms Rep. -7.896 -3.348 X10 Terms Dem. 1.621 1.277* X12 Non-Party Exp Dem 0.001 1.854* X17 Ch. App. Rep. Par. 4.348 5.705* X18 Ch. App. Dem. Par. -l.8ll -2.952 N - 1374 1 Fred. - .89 (Null - .90) Stan. Error - 1560 r - .41 Dependent Variable: Democratic Party Expenditures Independent Variables: ZSMLE T-stat * Constant 67.495 3 . 166: Y1 Qual Ch Inst 85.704 6.226* X6 Party -25.912 -2.899 X8* Dem Strength 0.520 1.472* Y2 Rep Party Exp Inst 0.156 3.825 X9 Terms Rep. 1.094 0.961 X10 Terms Dem. 0.048 0.076* X11 Non-Party Exp Rep 0.001 2.256* X17 Ch. App. Rep. Par. -l.800 -7.381* X18 Ch. App. Dem. Par. -0.921 -3.062 N - 1374 1 Fred. - .79 (Null - .81) Stan. Error - 370 r - .29 96 B. Discussion of Results In combining the previous five elections into a pooling scheme it is reasonable to expect a decrease in the goodness of fit and predictive measures, and this is the case. The correlation for the RPE equation is .41 whereas for the Democratic equation it is .29. The percent correctly predicted for the Republican estimation is .89 and not significant; for the Democratic equation it is .786 and again not significantly different from the null. For the Probit equation estimation in Model I the -2*loglikelihood ratio is 537.64 which is significant. The percentage correctly predicted is .757 and is not significant. To reiterate however, part of the value of the Probit estimation is its use as an instrument for the expenditure equations, and this is discussed when the individual hypotheses are considered. The Serpcpure pf Political Opportunities hypotheeis Interparty competition should also be visible at the individual level where candidates, acting as a party resource, make their strategic choices. Model I integrates the individual level equation and estimates it endogenously. Support for the contention that individual strategic choice should be measured endogenously is found by the significance of the instruments in Equations 4.2 and 4.3. In the Republican Party Expenditure Equation the instrument for the quality of challenger (Y1*) is statistically significant with a t-statistic of 2.43. In the Democratic Party Expenditure equation the instrument is significant at 6.23. The fact that minimal contributions should be measured endogenously suggests the possibility of some biasedness in the 97 estimations for Model II and any other studies that use the quality challenger measure as strictly exogenous in their specifications. The magnitude of the coefficients for the Quality Challenger Instrumental variable is also significant. Depending upon the party in question the results indicate that quality challengers receive considerable more financial backing than non-quality challengers. The results also show that a quality challenger of the opposing party also generates a significant contribution from the party in support of the incumbent. The size of the coefficients and the resultant t-statistics for the instrumental variable in the equations suggests the relative importance of this characteristic in the decision-making processes for the individual party units. In fact, for both the Democratic and Republican equations, the coefficients for the instrument are much larger than any of the others. As compared to national conditions and district party strength, the quality of the challenger is the dominant consideration when it comes to deciding money allocation. The §tpateglc Choice Hypothesis The Strategic Choice evaluation procedure is operationalized by using the average presidential approval rating. This hypothesis is directly tested in Equation 4.1 in Model 1. Variables X13 and X15 represent the strategic choice position. Neither variable is statistically significant at the .05 level. The average approval rating variables used in the model were an average over a longer time period than the overly short springtime time frame offered by Jacobson and Kernell. Models were run using the average springtime ratings for each party in power for both the individual and collective level equations. 98 The results of these estimations proved the variables to be not statistically significant and therefore not supportive of the Strategic Choice hypothesis. The Strategic Choice evaluation procedure is tested only at the individual level because it more closely represents the theoretical stipulations of Jacobson and Kernell. The central component of their theory is that the individual candidate makes the strategic choice. The models which they specify directly refer to this specific linkage. We can safely assume however, that the Strategic Choice procedure of using the average of some specified time period accurately reflects their position on evaluation procedures for all contributors. There is no evidence for the Strategic Choice position at the individual level. e o t n H othesis: Individual Level The Extrapolation Hypothesis states that the change in national conditions is a better measure of the party elite decision-making procedure than the Strategic Choice position. The change in presidential approval over an extended period of time provides the measure of how elites perceive public satisfaction to be in the future. This is a prospective evaluation. The individual choice equation 4.1 specifies both the average approval ratings and the change in approval ratings for the candidate time period, X14 and X16. Of the two variables, only the change in approval for a Republican in power (X16) is statistically significant. Variable X14, the change for Democratic President in power is not significant. ‘ An interpretation of the change in Republican presidential approval leads to very inconclusive evidence. Given the size of the coefficient, a dramatic change in the approval of a Republican president 99 would have to occur in order for this variable to have an important effect on an individual candidate deciding whether to run for office. Interpreting the approval coefficient along with the party coefficient tells us that it also is not very important which party the individual candidate is from. Overall, the evidence from this model is that national conditions as measured by presidential approval are not an important determinant of individual candidate choice. This leads to two important implications: first, this seriously undercuts a major premise of Jacobson and Kernell’s Strategic Choice position. Second, any proposed strategies based upon national conditions are open to serious question at the individual level. Offensive and defensive behavior is based upon a reaction to national conditions. This evidence shows that national conditions do not play an important role in individual candidate choice. e at on H othesiS' Collective ev The Extrapolation position at the collective organizational level (Eqs. 4.2 and 4.3) provides stronger evidence. Three of the four presidential approval variables are significant. The sign for the Democratic Party Approval variable is not what was hypothesized. As this relates to the individual level equation, two related questions are answered: one, empirical support exists for the extrapolation position in terms of both individual and collective calculi; and two, empirical support also exists for differentiating in the time periods between individual and collective decision-making. Support for the second consideration lends credibility to the proposition of differentiating the choice processes in terms of determinants of behavior. 100 Once again, as with the individual level coefficients, interpretation suggests that relatively large changes in approval are needed in order for it to have a significant impact on contribution strategies. National conditions as measured by change in approval do not play the most important role in determining money allocation. Only in instances where there is a significant change in approval, 12 to 15 points, do we see the parties responding in a significant manner. The specific hypothesis of offering the extrapolation evaluation procedure as a better measure than the Strategic Choice procedure is supported by the findings. Presentation of Results for Model 11 The results for Model II are presented with a separate table for the estimation of five congressional elections: 1976, 1978, 1980, 1982, and 1984. Each table presents the results from estimations of the Republican and Democratic party expenditure/contribution dependent variables. (Eqs. 4.4 and 4.5) The tables present the two-stage maximum-likelihood estimates (ZSMLE) and their respective t-statistics. The headings and summary statistics are the same for the tables in the Appendix as they are for Tables 1 and 2. The percent correctly predicted is interpreted the same for Probit as it is for the Tobit estimates. The asterisk next to the reporting of the t-statistic indicates that the coefficient is statistically significant at the .05 level. The number of cases for all of the estimations are considerably less than 435 because of the deletion of open seat races, races in which only one of the major parties was represented, and missing data. 101 A. Model II Results Table 3. Tobit Estimates -- Party Expenditures -- 1976 Election Dependent Variable: Republican Party Expenditures Independent Variables: 2SMLE T-stat Constant -11.974 -0.533* X1 Rep. Inc. 85.681 5.580* X3 Qual Ch Rep. 32.817 2.235* X4 Qual Ch Dem. -4l.l3l -2.235* X7 Rep Strength -0.552 -l.672* Y3 Dem Party Exp 2.396 9.699 X9 Terms Rep. 1.911 0.822* X10 Terms Dem. 9.288 5.273 X12 Non-Party Exp Dem -0.018 -l.429 N - 379 Z Pred. - .863 (Null - .824) Stan. Error - 72.8 r - .707 Dependent Variable: Democratic Party Expenditures Independent Variables: ZSMLE T-stat Constant 2.211 0.145* X2 Dem. Inc. 42.340 4.223* X3 Qual Ch Rep. -29.547 -1.789* X4 Qual Ch Dem. 21.547 2.081 X8 Dem Strength -0.568 -1.584* Y2 Rep Party Exp 0.675 3.156 X9 Terms Rep. -0.650 -0.435* X10 Terms Dem. '3.405 ‘3.386 X11 Non-Party Exp Rep -0.022 -1.009 N - 379 I Pred. - .784 (Null - .790) Stan. Error - 47.4 r - .587 102 Table 4. Tobit Estimates -- Party Expenditures -- 1978 Election Dependent Variable: Republican Party Expenditures Independent Variables: ZSMLE T-stat Constant 31.049 0.863* X1 Rep. Inc. 161.000 7.855* X3 Qual Ch Rep. 91.987 5.032 X4 Qual Ch Dem. -13.492 -0.739* X7 Rep Strength -1.796 -2.996* Y3 Dem Party Exp 3.066 11.291* X9 Terms Rep. 4.826 1.705* X10 Terms Dem. 6.916 3.859 X12 Non-Party Exp Dem -0.006 -0.805 N - 377 Z Pred. - .850 (Null - .821) Stan. Error - 91.6 r - .722 Dependent Variable: Democratic Party Expenditures Independent Variables: ZSMLE T-stat Constant 16.611 0.895* X2 Dem Inc. 78.133 7.194* X3 Qual Ch Rep. 17.561 2.199* X4 Qual Ch Dem. 43.258 4.877* X8 Dem Strength -l.638 -4.036* Y2 Rep Party Exp 0.377 4.725 X9 Terms Rep. -0.125 -0.087* X10 Terms Dem. -2.084 -2.921 X11 Non-Party Exp Rep -0.006 -0.823 N - 377 X Pred. - .790 (Null - .791) Stan. Error - 40.4 r - .546 103 Table 5. Tobit Estimates —- Party Expenditures -- 1980 Election Dependent Variable: Republican Party Expenditures Independent Variables: ZSMLE T-stat Constant -227.000 -5.06l* x1 Rep. Inc 131.000 4.931: X3 Qual Ch Rep. 55.452 2.156* X4 Qual Ch Dem. -150.000 -4.664* X7 Rep Strength 3.226 3.768* Y3 Dem Party Exp 3.548 9.126* X9 Terms Rep. 24.731 5.171* X10 Terms Dem. 5.852 2.470 X12 Non-Party Exp Dem 0.009 1.205 N - 391 Z Pred. - .824 (Null - .795) Stan. Error - 131.2 r - .709 Dependent Variable: Democratic Party Expenditures Independent Variables: ZSMLE T-stat Constant -33.637 -2.014: X2 Dem. Inc. 31.929 2.972 X3 Qual Ch Rep. 8.452 0.560* X4 Qual Ch Dem. 41.543 3.821 X8 Dem Strength 0.313 0.872* Y2 Rep Party Exp 0.128 1.658* X9 Terms Rep. -6.888 -3.728 X10 Terms Dem. -0.958 -l.082* X11 Non-Party Exp Rep 0.008 1.696 N - 391 I Pred. - .847 (Null - .851) Stan. Error - 48.8 r - .502 104 Table 6. Tobit Estimates -- Party Expenditures -- 1982 Election Dependent Variable: Republican Party Expenditures Independent Variables: ZSMLE T-stat Constant -24.l74 -0.3l9* X1 Rep. Inc. 110.000 2.656* X3 Qual Ch Rep. 122.000 3.019* X4 Qual Ch Dem. -7l.745 -l.794* X7 Rep Strength 2.320 1.739* Y3 Dem Party Exp 2.885 9.713* X9 Terms Rep. -10.572 -2.064 X10 Terms Dem. -l.566 -0.415* X12 Non-Party Exp Dem 0.013 1.849 N - 372 Z Pred. - .914 (Null - .841) Stan. Error - 192.1 r - .706 Dependent Variable: Democratic Party Expenditures Independent Variables: ZSMLE T-stat Constant -67.230 -2.554* X2 Dem. Inc. 40.225 2.088 X3 Qual Ch Rep. -6.837 -0.243* Xh Qual Ch Dem. 43.626 2.462 X8 Dem Strength -0.297 -0.457 Y2 Rep Party Exp 0.156 1.575 X9 Terms Rep. 2.235 0.915 X10 Terms Dem. -0.244 -0.l39 X11 Non-Party Exp Rep 0.016 1.607 N — 372 2 Pred. - .704 (Null - .703) Stan. Error - 82.9 r - .542 105 Table 7. Tobit Estimates -- Party Expenditures -- 1984 Election Dependent Variable: Republican Party Expenditures Independent Variables: 2SMLE T-stat Constant 266.000 3.548* x1 Rep. Inc. 219.000 4.053* X3 Qual Ch Rep. 45.605 0.954 X4 Qual Ch Dem. -5.967 -0.l2l* X7 Rep Strength -5.105 -3.547* Y3 Dem Party Exp 1.415 6.731 X9 Terms Rep. -2.676 -0.436 X10 Terms Dem. '2.866 'O.680* X12 Non-Party Exp Dem 0.028 3.346 N - 410 Z Pred. - .785 (Null - .762) Stan. Error - 225.8 r - .683 Dependent Variable: Democratic Party Expenditures Independent Variables: 2SMLE T-stat Constant -80.839 -3.574: X2 Dem. Inc. 72.780 2.726 X3 Qual Ch Rep. 19.345 0.747 X4 Qual Ch Dem. 34.607 1.428 X8 Dem Strength -0.317 -0.340* Y2 Rep Party Exp 0.271 3.151 X9 Terms Rep. -0.185 -0.058 X10 Terms Dem. -l.474 -0.678* X11 Non-Party Exp Rep 0.023 2.960 N - 410 1 Pred. - .641 (Null - .629) Stan. Error - 111.8 r - .674 106 For Model 11, estimates were obtained for the congressional elections of 1976 through 1984. For the equations with Republican Party Expenditures as the endogenous variable the correlations range from a high of .722 to a low of .683 for the two-stage estimates. For the equations with Democratic Party Expenditures as the endogenous variable the range is a high of .674 to a low of .502 for the two-stage estimates. The Republican party equations consistently provide a better fit than the Democratic party equations. The Republican party organizational advantage is a possible explanation for this difference. If this is the explanation of merit, it is interesting to note the convergence of the two correlations for the election year 1984. This is the year in which many observers viewed the Democrats as finally allocating funds in a collectively rational manner. These summary statistics are impressive not only for their relative numerical strength, but also for their substantive interpretation value. For Model II, the percent correctly predicted can be interpreted the same for the Tobit models as it can for Probit models. For the Republican Party Expenditures estimations, the two-stage estimates provide a range of .914 for the 1982 election to .785 for the 1984 election. For the Democratic Party Expenditure estimations, the two- stage estimates provide a predictive range of .847 for 1980 to .641 for 1984. The percent predicted is the total number of correctly predicted estimates for the value zero and non-zero. For the Tobit equation, the percentage states the percentage of times the model correctly predicted when the parties decided to contribute and when they did not. All the Republican equations in Model II offered predictive measures which are significantly different from the null; the Democratic equations provided 107 some measures which are not significant as applied to the Chi-square test. B. Discussion of Results e et c P s t on The symmetric hypothesis argues that offensive and defensive behavior are congruent. At the collective level ”defense” is associated with contributing a greater proportion to incumbents, "offense" is associated with contributing a greater proportion to challengers. Interestingly, the only relevant determinants for this type of behavior are national conditions -- favorable conditions lead to "offensive" behavior, whereas unfavorable conditions lead to ”defensive" behavior. This is the conventional wisdom of party responses to national conditions. The initial evidence from looking at Table 3 seems to support these contentions. Table 8: Party Contributions/Expenditures by Candidate Status, 1978-1986 Incumbent Challenger Open Seat Year Dem.2 Rep.Z Dem.2 Rep.1 Dem.1 Rep.2 1978 50.9 29.2 22.3 43.1 26.7 27.6 1980 57.0 27.6 26.4 53.4 16.6 18.9 1982 37.7 42.7 46.7 35.2 15.6 22.0 1984 52.7 29.7 30.4 56.8 17.0 13.5 1986 33.5 37.9 44.3 32.7 22.3 29.3 108 The percentages of total party expenditures/contributions going to incumbents or challengers shows a pattern. It can be assumed, at least in part, that these patterns are dictated by national conditions as shown by the statistical significance of the variables in the Model I Tobit estimates. The evidence shows the Republicans responding defensively in the midterm election years of 1982 and 1986, midterm years with a Republican president. Likewise, the evidence shows the Democrats responding more "offensively" in the years 1982 and 1986, where they expended 46.71 and 44.32 respectively. In 1980 and 1984 the Democrats only expended 26.4% and 30.4% of their total monetary resources on challengers. Questions of whether the parties are acting from a mindset or in response to national conditions can only be answered in part with a respecification of the models. Conditions for 1980 were perceived to be an advantage for the Republicans and the evidence shows them responding in an offensive manner. The conditions in 1982 were perceived to favor the Democrats, and thus they responded offensively. This is the context of Jacobson's interpretation. An analysis of this type conceals some of the greater intricacies that exist in these processes. The problems with explicating the events surrounding the 1982 congressional election is a prime example. The percentages given in Table 3 for 1982 do not explain the scenario involving the Republicans financing their quality challengers who were recruited at an earlier period under different conditions. The evidence from the table does not reflect this type of activity. 109 At the theoretical level, the offense and defense metaphor as conceived in the symmetric mode is inconsistent. The metaphor in general does not explain any type of behavior. 0n the strategic level, the evidence does not support the metaphor as it is generally understood. Model 11 results indicate that except for 1982, the Republicans consistently provide at least a two to one proportional monetary advantage for their incumbents over their quality challengers. This is highlighted by the fact that in 1982 the conditions dictated defensive behavior. The findings are similar for the Democrats. In years 1976, 1980, and 1982, the Democrats spent a greater proportion on quality challengers than they did on incumbents. Only in 1982 were the conditions defined enough to dictate offensive behavior on the part of the Democrats. The Non-Symmetric Position A non-symmetric interpretation allows for the possibility of party collective behavior to reflect what is normally considered "offensive" behavior in response to unfavorable conditions. There is a reason for this. The main determinants, as the results from Model I show, are the incumbent/challenger distinction and more specifically the quality/non- quality challenger distinction. These are the factors in which the party seems to respond. While the national conditions variables are statistically significant, the coefficients suggest that major changes either positively or negatively are needed in order for the parties to respond in a meaningful fashion. A threshold exists where either conditions or certain events take place that dictate a shift in monetary resource strategies. The findings do suggest that the collective decision unit acts on an independent basis, responding to candidate 110 characteristics and national level conditions; both of which reflect the competitive conditions between parties at the individual and collective levels. An interpretation of Table 3 does not allow for this type of manifestation. Evidence for the non—symmetric position can be found in the results from disaggregating the years in Model II. The results from estimating the model for 1982 show that the coefficient for quality challenger of their own party (X3 for Republican equation and X4 for Democratic equation) is larger than the incumbent of their own party variable (X1 for Republican equation and X2 for Democratic equation) for both parties. By purely a symmetric definition, this is not suppose to happen; and it is only in 1982 where results such as this occur. In seven out of the remaining eight cases the incumbency variable maintains a larger coefficient than the quality challenger. The results from the model for 1982 are extremely illuminating. While all five years suggest similar party responses, it is only in 1982 that we see both parties responding with greater resources to challengers than incumbents. The parties were not responding in the way Jacobson proposed. Late attempts to funnel money into certain races were based on district level factors in 1982. To the extent the Republicans responded to a change in national conditions, the resultant behavior could not be characterized by the offense and defense metaphor. The evidence shows that more than likely they were just responding to those races, incumbent or challenger, in which the money had any value whatsoever. The financial situation of the Republican party did not and does not restrain their strategic behavior. 111 The Democrats, operating from a position of scarce resources in terms of money, do not operate from a position of scarcity in terms of candidates. The autonomy of the Democrats also suggests that individual candidates are largely basing their decisions on their own perception of their individual chances. At the collective level, however, the quality challenger variable is still the dominant consideration. The key is that fact that quality challenges receive greater support than incumbents. For the Democrats this is the case in three out of the five elections studied. t e 5 us on Jacobson (1978) finds that the electoral consequences of spending are not the same for incumbents as they are for challengers. He finds that expenditures by challengers have a greater effect on the outcome of the election. If this is the case, the parties may perceive the situation as being one where in defensive times, the amount of expenditures that challengers need in order to offset the negative conditions might not be worthwhile. But since expenditures for challengers are supposedly worth more, it would seem rational for the parties to take this into account and make use of this challenger advantage. There is no apparent evidence for this. Once again, this suggests the importance of integrating the individual level equation into the system. Given the importance of the quality challenger instrument in Equations 4.2 and 4.3, and given the importance of the quality challenger variable in Model II, it leads us to suggest that the important decisions for the party as a whole are made at the individual level, and that party collective decision units 112 in large part respond to what takes place at the individual level. While elites at both the individual and collective level make rational calculations about how conditions will be at election time, at the collective level you see the possibility of greater flexibility, and therefore a greater ability to make efficient choices. The importance of national conditions is overemphasized at both levels. Jacobson is correct in asserting that the national collective unit plays an independent role, but the determinants suggest that characteristics involving challenger status plays a far more important role. This is the way that the collective party appears to be reinforcing individual decisions. This is what helps to explain Republican party behavior in the 1982 congressional elections. In effect, the key party resource is the individual candidate, and the collective unit responds in large part to individual candidate decisions. As a result, while parties at the collective level can have an independent effect, they seldom behave independently from the individual level choice. It is only in periods of great change -- significant changes in presidential popularity -- that we see noticeable idiosyncratic behavior at the collective level. The Barty and Non-Barty Distinction A distinction exists between party and non-party organizational behavior. This is evident in the action and reaction to strategic behavior between party and non-party expenditures. First we can focus on behavior within the electoral market and the strategies between party organizations. For Equation 4.2 with Republican Party expenditures as the dependent variable the two-stage maximum-likelihood estimate (2SMLE) 113 for Democratic Party expenditure is positive and statistically significant with a t-statistic of 9.217. Equation 4.3 with Democratic Party expenditures as the dependent variable provides a ZSMLE for Republican Party expenditures which is positive and significant with a t-statistic of 3.825. In Model II the results are similarly conclusive, the results from the ZSMLEs for variables Y2 and Y3 show that in nine out of ten cases the coefficient is positive and is statistically significant. Evidence exists that when controlling for simultaneity the party expenditure/contribution variables, Y2 and Y3, are major determinants of opposing party organizational action and reaction. The evidence also shows that while Democratic party behavior is a major determinant for Republican party behavior the reverse relationship, while statistically significant, is not nearly as strong. For Model II, in a comparison of t-statistics for the ZSMLEs, in all five cases the t-statistics for Democratic behavior (Y3) used as a right-hand-side (RHS) variable is much more significant than Republican behavior (Y2) as a RHS variable. This also supports the general conception of the Republicans having a significant organizational advantage -- and as a result a better developed decision calculus -- than the Democrats. Given that two competing parties are operating in the electoral market, opposing party strategies are based on determinants which differ from those of non-party organizational strategies. In Model I, for the Republican equation the t-statistic for Democratic Party and non-party contributions are 9.217 and 1.854 respectively. For the Democratic equation the t-statistic for Republican Party and non-party expenditures are 3.825 and 2.256 respectively. Using McDonald and Moffitt's (1980) 114 weighting technique the Model I coefficients suggest that the Republicans spend approximately $446 for every $100 spent by the Democratic Party. The Democrats, on the other hand, only spend approximately $3.88 for every $100 spent by the Republicans in terms of contributions and expenditures on behalf of candidates, holding everything else equal. These proportions are very similar to those provided in Model II. In 1982, for example, the Republican Party spent approximately $260 for every $100 of the Democratic expenditure, whereas they only spent approximately $1.00 for every $100 of non-party expenditures. For the Democrats, they spent approximately $12.00 for every $100 by the Republican Party, but only $1.30 for every $100 by non-party contributions. These findings once again support the contention of the importance of the interaction between parties as a determinant of party strategic behavior. It also, in part, lends credence to the differentiation between party and non-party organizational strategies in terms of their determinants for strategic choice. The same findings hold true for the results from Model II. A comparison of the t-statistics for the ZSMLEs for the party and non- party variables offers some interesting results. Six of the coefficients for the non-party variables X11 and X12 are not significant. In eight of the ten cases stronger significance exists for opposing party expenditures than for non-party expenditures; in many cases, the difference is quite large. Jacobson and Kernell do not differentiate between the strategic behavior of party and non-party organizations. PACs may differ in their motivation than the parties, but the final strategies result in similar 115 strategic results. The above findings suggest that major differences exist between parties and non-party organizations in terms of strategies and determinants. Jacobson and Kernell's stipulation that all forms of contributions be considered the same is a vast overgeneralization. Chapter 6 Conclusion Introduction This thesis finds support for several theoretical contentions. Evidence exists for the proposition that resource allocation strategies are better understood originating from a party theoretical perspective rather than a congressional election theoretical perspective. The study shows that the previous literature has not correctly specified party organizational activity within the electoral sphere. By using a theory of party organization that is electorally-based, we are able to conceptually clarify certain causal relationships, both external and internal to the party organization. Theoretical Consequences The difficulty in defining organizational goals and the strategies to achieve these goals is inherently a problem in discussions such as this. This difficulty is especially prevalent in the study of political parties. By emphasizing the dominant nature of the office-seeking goal in political parties, Schlesinger avoids many of the inconsistencies involving the collective benefits derived from holding office. This centrality of organizational purpose allows us to clarify some inconsistencies in the Jacobson studies. 116 117 Jacobson's (1985-1986) main contention is that a collective action problem exists among party officeholders when it comes to allocating monetary resources. Given that the Republicans are supposed to have an organizational advantage over the Democrats, they have less of a collective action problem. As a result, the Republican allocation has a greater effect on the goal of seat maximization. Conversely, the Democrats have a serious collective action problem, and this works to their disadvantage in terms of trying to maximize seats. A reconceptualization of the goal of party organization results in a different interpretation. Resource allocation in the Democratic Party is more intelligible given that its goal structure is not seat maximization but maintaining majority status. Individual behavior, it must be remembered, is constrained by collective organizational activity. That is why it is best for individuals to remain autonomous. Office—seekers in a competitive situation, and in need of the financial resources, are in a position to trade some of this autonomy for organizational assistance. It is in this way that competitiveness determines the level of organizational activity. Those in a position of relative security still attempt to maximize resources, but they are not in a situation where their autonomy is threatened. In fact, members that are safe, and possess institutional leadership ambitions, allocate portions of their overall contributions to members who are in need. This is done based on the congressional norm of reciprocity. The theoretical consequences present themselves over the distinction between Jacobson and Kernell's Strategic Choice Theory of congressional elections and Schlesinger's theory of party organization. 118 The common consideration involves the strategic use of resources by organizational entities. Schlesinger's theory of party organization defines organizational activity as the interaction of two constructs. The structure of political opportunities is individually career based and is defined as a separate subset of organizational behavior. Collective organizational behavior is conceived as essentially an aggregation of individual interests. While it is subject to a different set of determinants, the dominant determinant comes as the result of the individual acting within the structure of opportunities setting. Jacobson and Kernell do not differentiate between individual and collective determinants. The empirical difficulties that result are evident in Jacobson's (1985-1986) attempt at explaining party organizational behavior for the congressional election of 1982. By differentiating party organizational behavior into an individual and collective component it allows for the further clarification of the interaction of the theoretical constructs and provides the opportunity to analyze the role of the individual officeseekers in total party organizational behavior. The appropriateness of the Schlesinger model is evident in specifying the individual choice mechanism as an integral part of the theory. The empirical results show the extent to which the office- seekers' dominate party politics and party organizational behavior. The contention that is made in chapter two that the organization exists to serve the interests of the officeseekers is supported by the results. The evidence shows that the party organization is not only officeseeker dominated, but more importantly is dominated by those with "progressive" ambition, as exemplified by the role of quality 119 challengers. Schlesinger (1966) argues that ambitions vary in their direction. Ambitions can be "discrete", "static", or "progressive." A discrete ambition is one in which "the politician wants the particular office for its specified term and then chooses to withdraw from public office.” A static ambition suggests that ”the politician seeks to make a long-run career out of a particular office." A progressive ambition is when "the politician aspires to attain office more important than the one he now seeks or is holding.“ Schlesinger goes on to assume that "progressive ambitions dominate and are suppressed only when they appear unreasonable in terms of the chances." (Schlesinger 1966, 10) Central to his theory is that the level of organizational effort or "multinuclear transfers" depend in part on the lines of the structure of opportunities. When the structure dictates progressive ambitions, the result is a maximizing of transfers. When the structure dictates discrete ambitions, the result is a minimizing of transfers. The role and importance of the quality challenger in the study serves to emphasize the importance of progressive ambition in the distribution of organizational effort. A. Advantages of Party Model over Strategic Choice Model One of the main advantages of the party model lies in theoretical completeness. Jacobson and Jacobson and Kernell's presentations build on many of the same theoretical principles, but they fail to draw linkages between many of the constructs. While they base the Strategic Politicians thesis in the structure of opportunities argument, they do not relate it to any notion of party system competitiveness. The result is an incomplete representation of parties. This translates into various inconsistencies when trying to explain party resource activity. 120 The linkages provided by Schlesinger allow us to fully account for the dominance of the office-seeker in party organizational activities. The structure of political opportunities defines the individual candidate choice process. The interaction with the party system competitiveness measures allows us to prOperly specify the determinants for party organizational behavior. Parties do not simply base allocational strategies upon competitiveness. Parties also do not simply base allocational strategies on national conditions. Schlesinger's theory of party organization not only permits the consideration of the three main sets of determinants, but more importantly, how they interact at the theoretical level. This permits us to develop a model which includes the role of quality challengers, and party system characteristics such as district level competitiveness and national conditions. The party system is defined in terms of the competitive interaction between the two parties. This theoretical argument is extended to the parties acting and reacting to opposing party organizational actions. The evidence shows that a major determinant of monetary distribution is whether or not the opposing party is offering a quality challenger up against a party's incumbent. Likewise, a party places far more importance on money spent by the opposing party then on other monetary activity. This is further evidence of the advantage of the Schlesinger party model. The Jacobson and Kernell model does not allow for this type of activity. The main distinction in this thesis is on the difference between a symmetric and non-symmetric approach. Setting a goal of seat 121 maximization creates constraints on behavior and strategies. The symmetric position dictates the same strategies for both parties, whether in a minority or majority position. The non-symmetric goal of gaining and maintaining control of the House allows for the parties to play separate roles, and it allows for a variance in strategies. It also permits us to delineate the inconsistencies in the conventional wisdom of the offense and defense metaphor of resources strategies. The previous discussion demonstrates that the offense and defense metaphor as commonly understood is basically incongruous. It is a conventional wisdom that when deconstructed provides no theoretical or practical explanation of party responses to national conditions. The non-symmetric position accounts for these inconsistencies by allowing the parties to respond differently given favorable or unfavorable conditions. B. Congressional Election Literature When Jacobson and Kernell's Strategic Choice Theory of Congressional Elections first came out it was thought of as an interesting and clever way to resolve the contradiction between the individual level and the aggregate level studies. The paradox consisted of the fact that economic conditions and presidential popularity relate to movement in the House vote, but explain little in terms of individual voting variation. As the thesis came under greater scrutiny and more appropriate models were developed, limitations were discovered. Born's (1986) individual level study finds better evidence for direct effects voting than for indirect and he also finds that challenger quality has only a weak influence on individual voters. Uslaner and Conway (1985) analyze the 1974 election and propose that the conventional wisdom of it 122 being a referendum on Watergate and the economy is an adequate explanation and that evidence does not really exist for Jacobson and Kernell's explanation. Ostrom and Marra (1987) in an aggregate model of seat change find little evidence for national conditions having an indirect effect as specified by Jacobson and Kernell. 0n the question of the discrepancies between the macro- and micro- level studies, a far more relevant study is Kramer's analysis. (1983) He argues that the discrepancies are a ”statistical artifact" and represent another version of the "ecological fallacy”. Kramer also suggests that the aggregate level studies are more likely to yield valid inferences than the individual level studies. While the previous criticisms focus on the indirect effect argument of the Jacobson and Kernell thesis, the focus of this criticism is on the role of contributors in their thesis. This is a somewhat less developed aspect of their theory, but nonetheless is an integral part. The argument they present is that contributors -- individuals, parties, and interest groups -- serve to reinforce the indirect effect with their contributions. The contributors do this by reacting to the same set of forces that individual candidates react to within the structure of opportunities. The findings in this study severely question the primary linkage of national conditions having a dominant effect on the strategies of one specific group of contributors -- the parties. Party contributions do serve to reinforce individual decisions, but not as Jacobson and Kernell perceive it. The evidence shows that the main determinant for party contribution strategies comes as a result of the individual decision on whether a potential quality challenger decides to run. Parties in general support quality candidates, and they 123 do this no matter what the national conditions are perceived to be at the time. The relevant question is whether national conditions play an important role for the individual choice mechanism. The evidence for this, as the findings show, is very suspect. The role of national conditions and the effect they have on congressional elections is still somewhat amorphous. A final theoretical consequence relates to the offering of the Strategic Choice versus the Extrapolation position on the evaluation of performance. What makes this an interesting and significant proposal is the application of these principles to the strategic choice level of elite decision-making. Fiorina's application is at the individual voter decision level. The issue he suggests is one of performance evaluation or prospective evaluation. In this thesis we offer the prospective evaluation procedure as characteristic of individual and collective decision-making. The current performance evaluation procedure characterizes Jacobson and Kernell's Strategic Choice Model. The decisions by the individual candidate on whether to challenge and by the collective party unit on whether to contribute are based primarily on prospective evaluation along with various district level conditions. Fiorina's assertion at the individual voter level is that there is a continual interplay of past, present, and future evaluations. The decisions analyzed in this thesis are subject to a different characterization, and this thesis suggests that there exists a difference between party elite decision- making and individual voter decision-making. Party elite decisions and strategies are based on a perception of the conditions that will exist 124 when the voters make their decisions at the time of the general election. Further Research There are several important areas in which extensions on this study could be made. The first consideration is that more insight is needed on the determinants of relative strategies. Two factors are important, one is the structural limits which constrain organizational activity; and two, the extent to which scarcity of resources affects strategies. What effects does the limit on contributions have on party strategies, and do they vary between the Republicans and Democrats given the variance in the ability to raise money and the number of candidates in which to disperse it? How do the parties vary in their strategies in respect to the two main resources: money and candidates? Does the fact that quality candidates are more scarce for Republicans affect their recruitment strategies? Does the fact that Democrats have greater problems raising funds affect their reaction to district level or national level conditions? Given no limit on contributions to individual candidates, it is possible that the Republicans would be able to act in a more strategically effective manner given changes in national conditions? These previously mentioned questions and many other considerations could be clarified by a series of interviews with the participants making the decisions. Many of the relationships hypothesized about in this study could easily be empiricized by a series of direct questions 125 asked of the various members and staff of the campaign committees in question. What is also needed is further deconstruction of the offense/defense metaphor. This misapplication of a game theoretic term has led to generalizations and perceptions that are misguided. A further clarification of the specific strategies and what form they take in response to various environmental conditions would help to alleviate these concerns. The final key consideration is the analysis of non-party contributions, and more specifically the importance of national conditions as determinants for behavior. If Jacobson and Kernell are correct, short-term partisan forces should also play an important role in non-party contribution decisions. While there exists circumstantial evidence for this (Jacobson, 1985), no evidence exists that would allow the justification of a causal link. APPENDICES APPENDIX A Measurement Rules and Data Sources The variables are operationalized in the following manner; sources and the measurement rule are given. The data is specified for the 1976 - 1984 congressional elections. Y2 t d Non- rt end tur e su e Republican Party Expenditures Measurement Rule: summation of direct contributions and coordinated expenditures -- in hundreds of dollars Democratic Party Expenditures Measurement Rule: summation of direct contributions and coordinated expenditures -- in hundreds of dollars Total Non-Party Political Committee Contributions: Republican Candidate Campaign Measurement Rule: summation of corporation, labor, non-party political committees which have not reported a connected organization, trade/member/health, cooperative, and corporation without stock contributions -- in hundreds of dollars Total Non-Party Political Committee Contributions: Democratic Candidate Campaign Measurement Rule: summation of corporation, labor, non-party political committees which have not reported a connected organization, trade/member/health, cOOperative, and corporation without stock contributions -- in hundreds of dollars Sources for Expenditure and Contribution Data: EEC Qisgiosuze Segies, No. 9: 1976 House of Representatives Campaign Receipts and Expenditures September 1977 £29 Reporgs og Einancial Activity, 1977-1978 Interim Report No. 5, U.S. Senate and House Campaigns June 1979 EEC gepggts on Financial Aegivigy, 1979-1980 Final Report, U.S. Senate and House Campaigns January 1982 EEC Repgggs 9n Einanciai Aggivigy, 1981-1982 Interim Report No. 3, U.S. Senate and House Campaigns May 1983 126 127 EEC Reports on Fingncial Activity, 1983-1984 Interim Report No. 9, U.S. Senate and House Campaigns May 1985 Diggrigg anel Measures Y1 - Non-Quality Challenger/ Quality Challenger Measurement Rule: quality measured by previous electoral experience and positions of visibility (broadcast journalism) - - 0 if non-quality challenger, 1 if quality challenger Quality Challenger -- Republican Measurement Rule: 0 if non-quality or Democratic quality challenger, 1 if Republican quality challenger Quality Challenger -- Democrat Measurement Rule: 0 if non-quality or Republican quality challenger, 1 if Democratic quality challenger Number of terms held by incumbent Measurement Rule: number of terms Number of terms held by Republican incumbent Measurement Rule: number of terms, 0 otherwise Number of terms held by Democratic incumbent Measurement Rule: number of terms, 0 otherwise Sources: Congressional Quarterly Special Report, Part 1, Vol. 34, No. 41: 2771-2869, Oct. 9, 1976. ssiona uarterl S ecial Re ort, Supplement to Vol. 36, No. 41: 2793-2916, Oct. 14, 1978. Cpngressional Quarterly Special Report, Supplement to Vol. 38, No. 41: 2986-3086, Oct. 11, 1980. es ona ua e S ecial Re ort, Supplement to Vol. 40, No. 41: 2481-2610, Oct. 9, 1982. Cpngressional Quartegly Special Report, Supplement to Vol. 42, No. 41: 2503-2609, Oct. 13, 1984. Republican Incumbent Measurement Rule: 0 if Democratic incumbent, 1 if Republican incumbent Democratic Incumbent Measurement Rule: 0 if Republican incumbent, 1 if Democratic incumbent 128 X6 - Party of Incumbent Measurement Rule: 0 if Democrat, 1 if Republican Source: Congressional Quarteply WeeklynReport, Vol. 35, No. 44: 2305- 2311, Oct. 29, 1977. Congpgsgipnal Quarterly Almanac, Vol. 35, 5B-11B, 1979. Congzessional Quarterly Weekly Report, Vol. 38, No. 45: 3338- 3345, Nov. 8, 1980. Cpngressional Quarterly Weekly Report, Vol. 41, No. 7: 386-394, Feb. 19, 1983. Congressional Quarterly Weekly Report, Vol. 43, No. 15: 689- 695, April 13, 1985. X7 - Republican Party Strength Measurement Rule: Republican presidential vote of nearest preceding election for that particular district X8 - Democratic Party Strength Measurement Rule: Democratic presidential vote of nearest preceding election for that particular district Source: Cpngressional Districts in the 19805. Washinton, D.C.: Congressional Quarterly Press. Ehrenhalt, A. ed. Politics in America: Members of Cpngregs In Waghington and at Home. Washington, D.C.: Congressional Quarterly Press. a o a vel Measures X13 - Average Presidential Approval Rating for Democrat in Power Measurement Rule: average rating -- October of non-election year through March of election year X14 - Change in Presidential Approval Rating for Democrat in Power Measurement Rule: change in rating -- October of non-election year through March of election year; highest to most recent X15 - Average Presidential Approval Rating for Republican in Power Measurement Rule: average rating -- October of non-election year through March of election year 129 X16 - Change in Presidential Approval Rating for Republican in Power Measurement Rule: change in rating -- October of non-election year through March of election year; highest to most recent X17 - Change in Presidential Approval Rating for Republican in Power Measurement Rule: change in rating -- third quarter to third quarter; highest to most recent X18 - Change in Presidential Approval Rating for Democrat in Power Measurement Rule: change in rating -- third quarter to third quarter; highest to most recent Sources: For years 1974-1980 The Gallup Opinion Index, Oct.-Nov., 1980, Report No. 182: 13,14,17. For years 1981-1984 Inn Gallup Opinion Index, Dec. 1984, Report No. 231: 10. APPENDIX B McDonald and Moffitt’s "Uses of Tobit Analysis" McDonald and Moffitt (1980) point out that the coefficients obtained from using Tobit can be quantified in ways not previously specified. As Tobin showed, the expected value of y is BY ' XfiNZ) + 0f(2). (1) where Z - Xfi/a, f(z) is the unit normal density, and F(z) is the cumulative normal distribution function. The basic relationship consists of: Ey - FEy* (2) where Ey is the expected value of all observations; Ey* is the expected value conditional upon being above the limit; and F(z) is the probability of being above the limit. They therefore point out that the total change in y can be disaggregated into two parts: any/axi - F(z)(8Ey*/6X1) + Ey*(6F(z)/6Xi) (3) (l) The change in y of those above the limit, weighted by the probability of being above the limit; and (2) the change in the probability of being above the limit, weighted by the expected value of y if above. (1980, 318-319) The calculation for determining (1) consists of figuring out the fraction of the sample above the limit, or the percentage of nonzeroes [F(z)]. The value can be obtained directly from statistical tables. 130 LIST 01" REFERENCES LIST OF REFERENCES Adamany, D. 1984 Political Parties in the 19805 in Malbin, M.J. one 5' a e the 12§C§. Chatham, N.J.: Chatham House Publishers. Alexander. H.E. 1984 Einansina_22litiss;_usnsxl_fllsstionsi_sns Enlitipnl_Rg£_pn. 3rd. Ed. Washington, D.C.: Congressional Quarterly Press. Amemiya, T. 1978 The Estimation of a Simultaneous Equation Generalized Probit Model. Econometpica. 46:1193-1205. Amemiya, T. 1979 The Estimation of a Simultaneous Equation Tobit Model. Ingepnapional Eponomip Review. 20:169-181. Arterton, F.C. 1982 Political Money and Party Strength in Fleishman, J.L. e t er a t ' e C 1 n e of Gpvepnance. Englewood Cliffs, N.J.: Prentice-Hall, Inc. Bianco, W.T. 1984 Strategic Decisions on Candidacy in U.S. Congressional Districts. ngiglntive Studies Qnapperly. 9:351-364. Bibby, J. 1979 Political Parties and Federalism: The Republican National Committee Involvement in Gubernatorial and Legislative Elections. Rupliug. 9 Winter. Black, G. 1972 A Theory of Political Ambition: Career Choices and the Role of Structural Incentives. Anerican Rplipignl figience Review. 66:144-159. Born, R. 1986 Strategic Politicians and Unresponsive Voters. Anerican Rplipignl Ccienge Revigw. 80:599-612. Bullock, 0.8. and M.J. Scicchitano. 1982 Partisan Defection and Senate Reelection. Aasrisss_£21itissl_92er£srlx- 90:477-488- Burnham. W.D. 1974 Communications. Amsrisan_Bolitisal_§sisnss_Bsxisa. 68:210. Burnham, W.D. 1975 Insulation and Responsiveness in Congressional Elections. E21isisal_§sisnss_92ertsrlx- 90 411-435- 131 132 Calvert, R. and J. Ferejohn. 1983 Coattail Voting in Recent Presidential Elections. Angpipnn Rolitignl Spignce Review. 77:407- 419. Canon, D.T. 1985 Political Conditions and Experienced Challengers in Congressional Elections: 1972-1984. Prepared for delivery at the 1985 Annual Meeting of the American Political Science Association, Washington, D.C. Conway, M. 1983 Republican Political Party Nationalization, Campaign Activities, and Their Implication for the Party System. Rublius. 13:1-17. Cover, A.D. 1977 One Good Term Deserves Another: The Advantage of Incumbency in Congressional Elections. Anepican Joupnal pf Politisal_§sisnss. 21:523-542- Cover, A.D. and D.R. Mayhew. 1981 Congressional Dynamics and the Decline of Competitive Congressional Elections in Dodd, L.C. and B.J. Oppenheimer. Cpngpggg Reconsidered 2nd. Ed. Washington, 0.0.: Congressional Quarterly Press. Crotty, W. and G.C. Jacobson. 1980) er ca rties ec ine. Boston: Little, Brown. Downs, A. 1957 An Econonip Ineopy of Cemocragy. New York: Harper and Row. Duncan, R.B. 1972 Characteristics of Organizational Environments and Perceived Environmental Uncertainty. ngninistrative Science Quarteply. 17:313-327. Duverger, M. 1963 Rplipipnl_£nppip§. New York: John Wiley and Sons. Edwards. G.C. and S-J- Wayne. 1985 Prssidsntial_Lsadsrshisi_zsli£is§ nnd Rolicy MnRing. New York: St. Martin’s Press. Eismeier, T.J. and P.H. Pollack. 1986 Strategy and Choice in Congressional Elections: The Role of Political Action Committees. Amsrisan_lssrnal.sf.2211&isal_§sisnse. 30 197-213. Eubank, R.B. 1985 Incumbent Effects on Individual-level Voting Behavior in Congressional Elections: A Decade of Exaggeration. Journal pf Enligipg. 47:958-967. Eubank, R.B. and D.J. Gow. 1983 The Pro-incumbent Bias in the 1978 and 1980 National Election Studies. e t ca figipngg. 27:122-139. Fenno, R.F. 1978 H2ms_Stxls1_fl2sss_usmbszs_in_Ihsit_nistrists- Boston: Little, Brown. Ferejohn, J. 1977 On the Decline of Competition in Congressional Elections Anerignn Roliticnl Spienpe Revigw 71:166-176. 133 Fiorina, M.P. 1977 The Case of the Vanishing Marginals: The Bureaucracy Did It. Angzipan Rplitipal Sgignce Revigw. 71:177-181. Fiorina, M.P. 1980 The Decline of Collective Responsibility in American Politics. Dngnnlng, Summer, 25-45. Gow, D.J. and R.B. Eubank. 1984 The Pro-incumbent Bias in the 1982 National Election Study. n u a 0 ca Science. 28:224-230. Handler, E. and J.R. Mulkern. 1982 finging§§_in_fiplipip§. Lexington, Mass.: Lexington Books. Hanushek, E.A. and J.E. Jackson. 1977 Spaniggipnl Metnpdg to; §ocial Sgigngipgg. New York: Academic Press. Hausman, J.A. and W.E. Taylor. 1983 Identification in Linear Simultaneous Equations Models with Covariance Restrictions: An Instrumental Variables Interpretation. Eppnpngnnign, 51:1527-1549. Heckman, J. 1978 Dummy Endogenous Variables in a Simultaneous Equation System. Eppnpngpxipn. 46:931-951. Herrnson, P.S. 1986 Do Parties Make a Difference? The Role of Party Organization in Congressional Elections. Journal of Rolitics. 48:589-615. Herrnson, P.S. 1987 Party Strategies and Resource Distribution in the 1984 Congressional Elections. Paper presented at the 1987 Annual Meeting of the Midwestern Political Science Association, Chicago, Ill. Hibbs, D.A., Jr. 1982 The Dynamics of Political Support for American Presidents Among Occupational and Partisan Groups. Anericnn qupnal pf Rolitipnl Sgiencg 26:312-332. Jacobson, G.C. 1978 The Effects of Campaign Spending in Congressional Elections. WW. 72:469-491. Jacobson, G.C. 1980 flpngy_in_Cpngpg§§ipnnl_filgppipn§. New Haven: Yale University Press. Jacobson, G.C. 1981 Strategic Politicians and Congressional Elections, 1946-1980. Presented at the Annual Meeting of the American Political Science Association, New York. Jacobson, G.C. 1983 Th; Rolipicg pf Congpgsgipnal Elecgipns. Boston: Little, Brown. Jacobson, G.C. 1985-1986 Party Organization and Distribution of Campaign Resources: Republicans and Democrats in 1982. Rplipipnl WH- 100:603-625- 134 Jacobson, G.C. 1987 The Marginals Never Vanished: Incumbency and Competition in Elections to the U.S. House of Representatives, 1952-1982. er can urna o o t cal c ence. 31:126-141. Jacobson, G.C. and S. Kernell. 1983 Ctragegy and Choice in Cpngrggpipnnl_§lgptipn§. 2nd. Ed. New Haven: Yale University Press. Judge. G.C. et a1. 1985 W. 2nd. Ed. New York: John Wiley and Sons. Kayden, X. 1980 The Nationalizing of the Party System in Malbin, M.J. ed. MW. Washington, D.C.: American Enterprise Institute. Kernell, S. 1978 Explaining Presidential Popularity. Anggipnn Rolipipnl Spigncp Rgview. 72:506-522. Kirkpatrick, J. 1978 Digmangling the Rappies. Washington, D.C.: American Enterprise Institute. Kmenta, J. 1971 Elements of Economegzigs. New York: Macmillan Publishing Co., Inc. Kritzer, H.M. and R.B. Eubank. 1979 Presidential Coattails Revisited: Partisanship and Incumbency Effects. erica Jou nal olitical figignpg. 23:615-626. Luce, R.D. and H. Raiffa. 1983 Game§ nng Qecisipnp. New York: Wiley. MacKuen, M. 1983 Political Drama, Economic Conditions, and the Dynamics of Presidential Popularity. Anepipnn Journal pi Rplipipnl Science. 27:165-192. Maddala, G.S. 1983 Limited-dependenp nng Qnalipative Variables in Econpmetzics. Cambridge: Cambridge University Press. Maddala, G.S. and F. Nelson. 1975 Switching Regression Models with Exogenous and Endogenous Switching. e W fitn;ig§ipnl_figgpgin§ipn. (Business and Economics Section): 423-426. Mann. T.E. 1978 WWW- Elgggipnp. Washington, D.C.: American Enterprise Institute. Mann, T.E. and R. Wolfinger. 1980 Candidates and Parties in Congressional Elections. ie ce v w 74:617- 632. Marra, R. and C. Ostrom. 1987 Explaining Seat Change in the U.S. House of Representatives, 1950-1984. Unpublished Manuscript, Michigan State University. Matasar, A.B. 1986 Cpppppnge RACs nnd Reggral Cnnpnign Einanpe ans: Egg 9; Abuse of Rpwep?. New York: Quorum Books. 135 Mayhew, D. 1974 Congressional Elections: The Case of the Vanishing Marginals. Enligy. 6:295-317. McDonald, J. and R. Moffitt. 1980 The Uses of Tobit Analysis. Review pf chnonics nng Statisticg. 62:318-321. Michels, R. 1966 Rplipica 12npt ie eg; A Spgio lo ogic cnlS tugy pi the i a de . New York: Free Press. Nelson, F.O. and L. Olson. 1978 Specification and Estimation of a Simultaneous-Equation Model with Limited Dependent Variables. MW. 19:695-709. Olson, M. 1965 Inn Lpgip pf Collegpivg Agpipn. Cambridge: Harvard University Press. Ordeshook, P. 1986 Game Theopy and Politipnl Theopy; An Introduction. New York: Cambridge University Press. Ostrom, C. and D. Simon. 1985 Promise and Performance: A Dynamic Model of Presidential Popularity. ca it ca cience ev ew. 79:334-358. Pindyck, R.S. and D.L. Rubinfeld. 1981 Econometric Models and Economic Eorecasts. 2nd. Ed. New York: McGraw-Hill Book Company. Rapoport, A. 1983 Mngnematical Models in phe Social and Behavioral Sgiences. New York: Wiley. Ragsdale, L. and T.E. Cook. 1987 Representatives' Actions and Challengers’ Reactions: Limits to Candidate Connection in the House. Amepican Journal of Political Science. 31:45-81. Reichley, A.J. 1985 The Rise of National Parties in Chubb, J.E. and P.E. Peterson. e New irec o in an Po Washington, D.C.: The Brookings Institution. Rohde, D. 1979 Risk Bearing and Progressive Ambition: The Case of Members of the United States House of Representatives. can Spurnnl pf Rolitical Science. 23:1-26. Sabato, L. 1984 A ower' s d Wo d of P0 t cal Ac ion Connittees. New York: W.W. Norton and Company. Schattschneider, E.E. 1942 Rnppy_prnpnngn§. New York: Rinehart and Company, Inc. Schlesinger, J.A. 1965 Political Party Organization in March, J.G. ed. The Rnnfibopk pf Crganizations. Chicago: Rand McNally. Schlesinger, J.A. 1966 Ambition nnd Rplipics, Rplitical Caregrs in the Uniteg States. Chicago: Rand McNally. 136 Schlesinger, J.A. 1975 The Primary Goals of Political Parties: A Clarification of Positive Theory. Ang;ipnn_£pli§ipnl_§pignpp nginy,.69:840-849. Schlesinger, J.A. 1984 On the Theory of Party Organization. Joupnal of Enlipipg. 46:369-400. Schlesinger, J.A. 1985 The New American Political Party. e can WW- 79:1152-1169- Schumpeter, J.A. 1950 Cnpipnlign, Spcinlign nnd Qemoppncy. 3rd. Ed. New York: Harper and Row. Sickles, R. and P. Schmidt. 1978 Simultaneous Equation Models with Truncated Dependent Variables: A Simultaneous Tobit Model. Journal pf Eggnomipg nnd Bnginesp. 31:11-21. Stimson, J. 1976 Public Support for American Presidents. Public Qpinion Qnaptpply. 40:1-21. Sundquist, J. 1982 Party Decay and the Capacity to Govern in Fleishman, J.L. ed. of er can 0 l a e ' The len e pf_pr§;nnnng. Englewood Cliffs, N.J.: Prentice Hall, Inc. Thompson, J. 1967 ngnnizntions in Region. New York: McGraw Hill. Tobin, J. 1958 Estimation of Relationships for Limited Dependent Variables. Ecpnpnetpica. 26:24-36. Tufte, E. 1973 The Relationship Between Seats and Votes in Two-Party Systems. Anerigan Rpliticnl Science Rgview. 67:540-554. White, P.E. 1974 Resources as Determinants of Organizational Behavior. Agninigpznpivg Spignce Quarteply. 19:366-379. Wilson, J.Q. 1973 Rplitical Organizations. New York: Basic Books. "Illilfillllfillllllllllr