Dynamics of open seat elections for the United States House of Representatives

MISSING IMAGE

Material Information

Title:
Dynamics of open seat elections for the United States House of Representatives
Physical Description:
ix, 194 leaves : ill. ; 29 cm.
Language:
English
Creator:
Green, Joanne Marie
Publication Date:

Subjects

Subjects / Keywords:
Political Science thesis Ph.D
Dissertations, Academic -- Political Science -- UF
Genre:
bibliography   ( marcgt )
non-fiction   ( marcgt )

Notes

Thesis:
Thesis (Ph. D.)--University of Florida, 1994.
Bibliography:
Includes bibliographical references (leaves 184-193).
Statement of Responsibility:
by Joanne Marie Green.
General Note:
Typescript.
General Note:
Vita.

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 001981100
oclc - 31913468
notis - AKF7987
System ID:
AA00002061:00001

Full Text
DYNAMICS OF OPEN SEAT ELECTIONS FOR THE UNITED
OF REPRESENTATIVES
By
JOANNE MARIE GREEN
A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY
UNIVERSITY OF FLORIDA
1994


Copyright 1994
by
Joanne Marie Green


The manuscript is dedicated to my mother, Mary Lou, to the memory of my father, Thomas, and to my loving family, Craig and Emma. Without all of their support, this work would
never had been possible.


ACKNOWLEDGMENTS
My debts of gratitude are many for this project. I will always be thankful for all of the assistance and support which I have received from the Department of Political Science in general, but also from the specific individuals listed here. I am in the debt of M. Margaret Conway for her unending assistance and support. Her influence in this work extends far from the material support which she so kindly provided me, but also to a more deeply felt emotional level. She serves as a role model for many individualsincluding myself. I am also very thankful to Wayne L. Francis who spent many hours working with me to refine my thinking and methodology. The amount of time he's spent with me goes far beyond what is required of an advisorI will always be grateful to him for that and his kindness. My thanks also extend to Richard K. Scher whose friendship and intellectual input has greatly affected this work and my thinking more generally. Special thanks also go out to Michael M. Martinez and David Rohde whose comments and feedback have assisted this project immensely. Thanks also go to Dean Colburn who graciously agreed to serve on my committee and assist in the development of the project. While I owe these individuals a great sense of gratitude, all errors or omissions are mine.
iv


TABLE OF CONTENTS
ACKNOWLEDGMENTS
iv
ABSTRACT
Vlll
CHAPTER 1: INTRODUCTION
1
Criteria for Evaluating Congress as an Institution........3
Key Values That Electoral Reform Should Promote...........6
Theories of Representation................................7
Unit of Analysis.........................................10
General Characteristics of Open Seat Elections from
1982 through 1992 ..................................... 11
Data Set.................................................14
Candidate Expenditures................................15
Candidate Occupational Background and Candidate
Gender.............................................15
National and State Unemployment Data..................16
District Presidential Vote and the Vote for the
Prior Incumbent....................................16
Party Monetary Assistance and Legislative Partisan
Behavioral Loyalty.................................16
The Percent of Vote for the Democratic Candidate......17
Discussion...............................................17
CHAPTER 2: THE RELATIVE ROLE OF CAMPAIGN EXPENDITURES AND CANDIDATE QUALITY: DIFFERENTIALS IN ELECTORAL ADVANTAGE S...............................................19
Introduction.............................................19
Disbursements............................................22
Candidate Quality........................................32
Discussion...............................................43
CHAPTER 3: NATIONAL, ECONOMIC AND LOCAL FACTORS IN OPEN SEAT ELECTIONS FOR THE UNITED STATES HOUSE OF REPRESENTATIVES FROM 1982 THROUGH 1992................... 48
Introduction.............................................48
National Factors.........................................50
Presidential Coat-tails...............................50
Surge and Decline.....................................55
Strategic Politicians.................................62
Economic Voting..........................................65
Traditional "Retrospective" Voting....................65
National Economic Voting..............................67
v


Local Factors............................................70
Incumbency Loss Potential.............................70
Volatility Of District................................72
Conclusion and Discussion................................73
CHAPTER 4: THE ROLE OF GENDER IN OPEN SEAT ELECTIONS FOR THE UNITED STATES HOUSE OF REPRESENTATIVES FROM
1982 -1992............................................... 78
Introduction.............................................78
Potential Advantages Accorded to Female Candidates.......82
Potential Disadvantages..................................84
Sexism................................................84
Political Resources...................................85
Occupational Bias.....................................86
Campaign Finance......................................89
Analysis.................................................92
Direct Effect of Gender on the Percent of Votes
Received...........................................92
Indirect Effects of Gender on the Percent of Votes
Received...........................................93
Descriptive Analysis.....................................97
General Findings......................................97
Comparison By Political Party........................101
General Comparisons Between Female and Male
Candidate Races ...................................102
Preliminary Conclusions.................................104
Primary Elections.......................................107
Average Margin of Victory............................108
Female Candidates' Primary Elections.................109
Female Primary Candidates Versus Male Primary
Candidates ........................................110
Conclusion and Discussion...............................113
CHAPTER 5: THE ROLE OF POLITICAL PARTIES: PATTERNS OF DEPLOYING RESOURCES AND POTENTIAL FOR SUBSEQUENT LEGISLATIVE BEHAVIORAL LOYALTY..........................116
Introduction............................................116
Weakened Parties.....................................117
The "Parties Just Begun"The Nationalization of
American Political Parties ........................120
Committee Structure and Operation....................121
Non-Monetary Assistance..............................123
Symbolic Nature of Party Donations...................124
Analysis................................................126
Distribution of Party Resources.........................128
Competitiveness of District and Race.................130
Gender of Candidates.................................133
Candidate Quality....................................135
Candidates With Financial Need.......................136
Direct Impact of Strategic Considerations on
Allocation of Party Resources........................138
vi


The Potential for Subsequent Behavioral Loyalty.........142
Conclusion..............................................147
CHAPTER 6: CONCLUSION....................................150
Campaign Expenditure Limits Coupled with Public
Financing............................................155
Opponents............................................156
Grants or Matching Funds?............................157
Determining the Proper Limit.........................160
Support For Spending Limits and Public Financing of
Elections .........................................161
Benefits of Public Financing and Expenditure Limits
System............................................163
Practicality of Instituting Reform...................165
Strong Parties..........................................166
Practicality of Instituting Reform...................172
Term Limits.............................................172
Practicality of Instituting Reform and Potential
Consequences ......................................174
Conclusion..............................................178
APPENDIX A: CHAPTER 2: SCATTER PLOTS....................179
APPENDIX B: CHAPTER 4: HISTOGRAMS OF FEMALE
CANDIDATE'S EXPENDITURES................................182
LIST OF REFERENCES........................................184
BIOGRAPHICAL SKETCH.......................................194
VII


Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy
DYNAMICS OF OPEN SEAT ELECTIONS FOR THE UNITED STATE HOUSE OF
REPRESENTATIVES
By
Joanne Connor Green
April 1994
Chair: M. Margaret Conway
Major Department: Department of Political Science
The central question which this research examines is how can electoral accountability and legislative responsiveness be maximized. Accountability and responsiveness are intimately tied to the level of competition; if competition is enhanced and electoral biases (for example, biases toward moneyed interests) minimized, the potential for electoral accountability and legislative responsiveness is increased. Although this is not a new question, this analysis tackles the research problem in a different manner than most past research. First, the universe of study is limited to open seat elections for the U.S. House of Representatives from 1982 through 1992, permiting a more lucid demonstration of the dynamics underlying elections.

Vlll


A second distinction of this research from prior research is a focus on differentials in electoral advantages rather than absolute resources. This means of envisioning campaign advantages extends to candidate expenditures as well as occupational background. Although benefits derive from examining the impact of each candidate's campaign expenditures, a more fruitful approach is to explore the relative advantages provided by differentials in spending.
This research will demonstrate the theoretical and statistical superiority of this measure. Candidate occupational background is also best envisioned and measured as an occupational advantage rather than an absolute resource; for example, that one candidate has more prior elected experience than the opponent is more important than if both candidates have electoral experience.
This research is motivated by a desire to make informed reforms to the current electoral system. Such reforms are only possible if we first understand what drives the system. The research examines factors affecting the elections, then examine proposed reforms in light of the research results.
ix


CHAPTER 1: INTRODUCTION
The distinguishing characteristic of democracies, as Harold Lasswell once observed, is the 'open interplay of opinion and policy. If people are not capable of expressing their choices through the electoral process, then elections become a methodology for concealing power rather than for sharing it. 1
The above statement depicts the purpose of elections in democratic societies and underlies a central concern evidenced throughout this research. Natchez states that "the meaning of the electoral process has rarely been considered on its own merits, that its functions always have been assumed rather than stated clearly. But without such a clear description of purpose, it is impossible to examine the capabilities of the electoral process or to evaluate its performance" (Natchez, 1985, p. 224). He goes on to point
out that the electoral process functions to provide political
choices; choices which extend over a full range of powers and
responsibilities. The electorate makes a variety of choices
in every election. These choices sustain the democratic
system in which we live. However, as the introduction
statement asserts, if individuals are not free to express
their choice, be it because of electoral fraud, corruption or
xPeter Natchez, 1985. Images of Voting / Visions of Democracy, p. 225.
1


2
inherent electoral biases, then the elections lose their capacity to reflect individual preferences and validation of power and become vehicles for misrepresentation.
When electoral politics in the United States are examined, normative theories and issues cannot be avoided. The point of view of this researcher and of this research is that electoral system in the United States is in terrible disarray and in desperate need of reform. The central goal of the research presented here is first understanding the underlying nature of the electoral system and then analyzing proposed reforms in light of the research presented here and other prominent literature in the field.
A major concern with the state of electoral politics stems from the fact that the costs of running for the House of Representatives is spiraling seemingly out of control. In 1992 the average candidate spent $311,902 in a bid for a House seat, up from an average of $228,060 in 1982 and $53,384 in 1974. Money is playing an ever increasing role in Congressional elections, and many are concerned over money's role in elections for the House of Representatives. Of special concern is the potential for corruption and the differential level of weight of opinions of big contributors versus the average constituent. In other words, many are concerned with the potential of legislators "paying back" large contributors with behavioral loyalty. If money does not "buy" Congressional roll call votes, it may increase access and influence at committee consideration stage of the


3
legislative process (Hall and Wayman, 1990). If this is the case, legislators are biased toward money and moneyed interests. Even if money does not buy votes or even access, it appears that it does. Such perceptions of impropriety are detrimental to the system. If the public believes that large contributors matter more to the legislator then fundamental conceptions of democracy are called into question, most notably the belief in "one person, one vote." According to this premise, every citizen should have equal input into the electoral system. If the system is biased toward money, the principle of equality is questioned.
Criteria for Evaluating Congress as an Institution
Leroy Rieselbach (1986) presents three means for evaluating Congress as an institution. These standards are responsiveness (representativeness,) responsibility (policymaking efficiency,) and accountability (opportunity for citizen control.) Responsibility relates to problem solving capacity. A responsible institution makes ''reasonably successful policies that resolve major issues confronting the nation" (Rieselbach, 1986, p. 10). The criteria for evaluating responsibility include speed, efficiency and success. Essentially, is the institution responding adequately, in a timely fashion, to the public needs?
The responsive nature of the institution places more importance on process and less on the content of the policy. A responsive government "takes into account ideas and


4
sentiments of those who will be affected by its actions: individual citizens, organized groups, local and state governments, and national executives [Government] must provide an open channel of communication..." (Rieselbach, 1986, p. 11). Responsiveness relates to the process of policy implementation. A responsive government is interested in promoting citizen involvement with legislation to ensure that the voices of those affected are heard.
Accountability functions after the fact as an evaluative tool for assessing governmental behavior. The concept of accountability asserts that the governing should be held to account for their behavior by the governed. If those who are governed are not satisfied with the legislator's behavior (or the institution's actions) then the represented can refuse to support the candidate in his or her bid for re-election. "They can use the ballot box to send new, presumably wiser and more honest individuals to Washington Accountability operates after the fact; decision-making failure may result in the loss of position and power should the voters conclude that new officeholders would perform more successfully" (Rieselbach, 1986, p. 12).
The key question of the research presented here is what role do elections play as a link between the represented and the representative? Much debate exists about legislators and legislators, but much less is present regarding electoral politics. How does the electoral link fit into these models? The three evaluative criteria are accountability,


5
responsibility and responsiveness. The means of selecting members of a legislature are intimately tied to legislative behavior, policy output and the ability to hold individuals independently and collectively accountable for their behavior. It is the contention of the author that although the three criteria for evaluation are not always compatible, they are all intimately tied to the means of member selection and negatively affected by electoral biases.
Biased elections decrease the ability to achieve responsible legislation because the elections are not necessarily producing quality representatives; they are simply reflecting the biases built into the electoral process. The same holds true for accountability. It is very difficult to hold individuals accountable for their behavior if the elections are not fair and competitive. If biases exist, for example, toward moneyed interests, the elections are not, by nature, fair and competitive. Fair and competitive elections can not be systematically biased toward one candidate or interest. If biases do exist, the ability of the governed to hold the governing accountable for policies or legislative behavior is diminished. If elections are reflecting biases rather than producing a legislature which is capable of being held accountable for behavior or one which is responsible for its actions or one which is responsive to public needs, the fundamental conception of democracy is called into question.


6
An evaluation of the extent to which biases are present is the central goal of the research presented here. The research will examine several lines of thought regarding potential factors and their electoral effects. Among these are the role of money, candidate quality ratings, national factors (such as presidential coat-tails and voting,) economic factors, candidate gender, and political parties. Their implications for legislative behavior will be analyzed. To conclude, the results of the research will be examined with special attention placed on prominent proposals for electoral reform including limiting the number of terms a legislator is permitted to serve, public financing of elections, campaign expenditure limits, and the potential for an increased role of political parties in the electoral process. The reforms will be examined in light of the conclusions reached in the research presented here as well as other prior research with special attention placed on potential for enactment and possible consequences, both intended and unintended, of the proposals.
Key Values That Electoral Reform Should Promote
Governmental effectiveness, legitimacy, representation,
accountability, participation and competition are key values of democratic governments that should be promoted by electoral reform (Nugent and Johannes, 1990; Magleby and Nelson, 1990). According to Nugent and Johannes, preserving the legitimacy of the national government is a fundamental


7
goal for a campaign finance system. "If elections legitimize the authority of government, certainly our system or financing elections must not undercut that legitimacy" (Nugent and Johannes, 1990, p. 7). Representation requires a correspondence between the characteristics, interests, political attitudes and policy preferences of the represented and their constituents. "An ideal system for financing campaigns should select and find candidates who represent the policy preferences, interests, partisanship and ideologies of their geographic constituency" (Nugent and Johannes, 1990, p. 8). "A campaign finance system that truly promotes representation would not select and fund candidates whose views correspond only with those of contributors, but also with those of the bulk of the citizenry. If contributors and non-contributors are essentially alike, no problem should arise. If they differ, some concerns about representation may be in order" (Nugent and Johannes, 1990, p. 8). Such concern is apparently warranted because the people who "give the resources ... are increasingly different from the people who vote in the election" (Sorauf, 1988 p. 348).
Theories of Representation
The examination of elections is also consequential because they are the means by which the members of the legislature are selected. One needs to examine the electoral mechanism because it has serious implications on representative theory. If elections are biased and are


8
therefore producing a biased legislature, crucial concepts of representation are violated.
Representation means making "present in some sense of something which is nevertheless not present literally or in fact" (Pitkin, 1967, p. 8). Representation occurs when one person is authorized to act in place for others (p. 41). The crucial criterion for assessing the representative nature of a democracy becomes elections. Elections are seen as "a grant of authority by the voters to the elected officials...In each election, voters grant authority anew, name representatives anew, though of course they may reauthorize the same individuals for another term...Elections are acts of 'vesting authority'" (Pitkin, 1967. p. 43). Representation is an ongoing relationship which requires trust and obligation on both sides (p. 128).
The "Accountability View" of representation sees the representative as someone who is to be "held to account, who will have to answer to another for what he does; those who he must eventually account are those whom he represents" (Pitkin, 1967, p. 55). According to this view, "representation, if it means anything, means that the representative must be responsible to the represented" (p. 55). Elections are the means to hold officials accountable for their actions.
"Descriptive", or "True", representation entails the concept of 'standing for' another. It requires that the legislature be selected so that its composition accurately


9
reflects that of the whole nation; only then can it really be a representative body. John Adams argued that it should "be an exact portrait, in miniature, of the people at large, as it should think, feel, reason and act like them" (Pitkin, 1967, p. 60). According to this line of thought, the representative does not act for those he or she represents; he or she stands for them. It is less important what the legislature does than how it is composed (Pitkin, 1967, p. 61). To represent means to be representative in the sense of having representative or typical characteristics. Elections are seen to be "a method of finding persons who possess this
representative quality" (Pitkin, 1967, p. 76). "Descriptive" representation entails a person standing for others "by being sufficiently like them" (p. 80). As seen in this vein, representation is less an action and more a correspondence; a pre-condition for governmental action (Pitkin, 1967, p. 82).
If one is to 'stand for' another, the method of selection is intimately tied to the concept of representation. This is not to state that individuals can only be represented by someone 'like them'. If this were the case, women could only accurately represent women; African-Americans could only truly represent African-Americans; homosexuals only represent homosexuals and so forth. Few would argue that this is the case, but the symbolic nature of the concept of representation should not be overlooked. If the legislature is to be seen as legitimate, it must be seen as representing the nation's needs, wishes and variety of


10
opinions. The legislature must be representative, sufficiently like, the population as a whole. To accomplish this, several requirements of a representative government must be met. First, regular elections which are free and genuine must be held. Second, a collegiate representative body must exist with more than advisory power. Finally, people of a nation must be "present in actions of its government in complex ways" (Pitkin, 1967. p. 234). If elections are not free and competitive, the government can not be truly representative because the elected officials were not elected to stand for the constituents and it is very difficult to hold them to account for their behavior.
Unit of Analysis
All regularly held open seat elections for the U.S. House of Representatives from 1982 to 1992 will be examined. Although open seat elections only represent a fraction of all elections, they are consequential and distinct. Most inquiries have been directed toward the effects of incumbency and, therefore, most research has focused on incumbent/ challenger Congressional races (i.e. Jacobson, 1986; Sorauf, 1988; and Stern, 1988). There is much less published research on the analysis of open seat elections. As evidenced by the extremely high rates of re-election (rates in excess of ninety or even ninety-five percent are not uncommon), incumbents benefit from institutional and habitual advantages not available to challengers. An examination of open seats,


11
however, provides the unique opportunity of examining the underlying nature of the elections and to achieve a clearer understanding of the variables which affect the electoral calculus. Open seat elections avoid the theoretically and statistically cumbersome incumbency advantage while providing a more clear understanding of the factors affecting elections for the U.S. House of Representatives and the subsequent implications for representative democracies and the prospects and potential for reformation.
General Characteristics of Open Seat Elections from 1982
through 1992
Open seat elections are best characterized as competitive races. Sixty-three percent of the races for open seat elections were won with sixty percent or less of the vote (a common standard for determining competitiveness.) In thirty-eight percent of the races, the winner received in excess of sixty percent of the vote (See Table 1-1). In the last twenty years, three-quarters of incumbents seeking reelection have received at least sixty percent of the vote.
The average amount of money spent, in real dollars, over the period under study has consistently increased every year except for 1992 (Table 1-2). The pattern is consistent for both Democrats and Republicans. The average percent of female candidates has increased from 1982 through 1992, with 1992 having the highest proportion of female candidates than any other year. Some have contended that the high proportion


12
of female candidates could possibly be indicating a trend which will continue throughout the decade (Table 1-3). Such a pattern will only be detectable via hindsight.
Table 1-1
Distribution of the Percent of Vote for the Democratic Candidate in All Open Seat Elections from 1982 through 1992
Percent of the Vote for the Democrat Less Than 40% Percent of the Vote for the Democrat Between 40% and 60% Percent of the Vote for the Democrat Greater Than 60%
41 (16%) 133 (63%) 56 (22%)
Table 1-2
Average Campaign Expenditures for the Democratic and Republican Candidates in All Open Seat Elections from 1982
through 1992
Year Averaqe Expenditure Democrat 1 Averaqe Expenditure Republican
1982 266,679 288,793 N = 110
1984 362,166 373,791 N = 50
1986 372,505 445,146 N = 84
1988 521,515 589,685 N = 50
1990 544,146 528,743 N = 58
1992 458,420 335,891 N 168
Candidates for open seat elections have electorally competitive occupational backgrounds. In every year, the most frequent candidate occupational background is prior experience in the state legislature. Such a pattern indicates first that open seat elections are seen as having potential risk factors which are deemed sufficiently low to warrant a state legislator to run for election to the higher office. For instance, state legislators can often run in years when they are not up for re-election without forfeiting


13
their state legislative seat in cases of defeat. The high proportion of state legislators running in open seat elections indicates the progressive nature of public office and the strategic tactics politicians employ since open seats represent the best chances of electoral victory.
Table 1-3
Number of Female Democratic and Republican Candidates in All Open Seat Elections from 1982 through 1992
Year Females Males
1982 13 97
1984 3 47
1986 9 75
1988 4 46
1990 8 50
1992 32 136
Democrats were more successful in winning open seat elections than Republicans. From 1982 through 1992, only one hundred fourteen Republicans won open seat races compared to one hundred forty-seven Democrats even though the party of the incumbent previously occupying the seats was nearly evenly divided between Republicans and Democrats. In addition, Democrats were more victorious than Republicans even though on average the Republicans consistently outspent the Democrats. Republicans were able to hold eighty of their vacated seats while the Democrats were able to hold one hundred and one of their vacated seats. The Democrats captured thirty-nine seats previously held by Republicans


14
while the Republicans snared thirty-three seats held by Democrats.
The general level of competition and the percent of seat-turnover in open seat elections are substantially higher than that evidenced in incumbent/challenger races. Open seat elections represent the best chance for qualified non-incumbents to achieve electoral victory. Given these two characteristics, open seat races are interesting and worthwhile units of study.
Data Set
All of the data were collected by the author. Most analyses examine only open seat general elections for the
United States House of Representatives from 1982 through 1992. The analysis of the role of candidate gender is also examined open seat primary elections in an effort to determine if biases for or against female candidates exist in the elections in open seat races both in the pre-nomination cycle and in the general election. The independent variables whose impact will be examined in the aggregate level analysis are candidate expenditures, candidate occupational background, presidential coat-tails, district vote for the presidential candidate, national factors, prior vote for the incumbent, party of the incumbent, national and state unemployment rates, political party monetary assistance, legislative partisan behavioral loyalty, and candidate gender. The dependent variable is, in nearly every case, the


15
percent of vote received by the Democratic candidate. If the circumstances dictate, the dependent variable will be changed, but it is primarily the percent of vote for the Democrat. In the instances where another dependent variable is examined, a clear notation of that will be provided.
The data were collected from a variety of sources by the author. In most instances when the variable is introduced the source for the data will be indicated, but for clarity, sources for the variables will also be provided in this introductory section.
Candidate Expenditures
Candidate campaign expenditures were collected directly from the Federal Election Commission. For 1982 through 1990, The Federal Election Commission's Final Report: U.S. Senate and House Campaigns was employed. In 1992 the Final Report was not released at the time of analysis, so the Federal Election Commission, Press Release was used. Candidate expenditures include all campaign related disbursements.
Candidate Occupational Background and Candidate Gender
Candidate occupational background data were compiled through reading candidate biographies published in Congressional Weekly Reports and The Almanac of American
Politics (Barone, Ujifusa and Matthews, 1982-1992.)


16
Candidate gender was ascertained through biographies in the above sources or, when available, candidate photographs.
National and State Unemployment Data
National and state unemployment data were collected for 1984 through 1992 from the Monthly Labor Review published by the United States Bureau of Labor Statistics. Data for 1982 were compiled from the Vital Statistics of the United States since the Monthly Labor Review did not release the data in the desired format in 1982.
District Presidential Vote and the Vote for the Prior Incumbent
The district vote for the presidential candidate was obtained from the Almanac of American Politics for all years except 1992. The National Review was used for that year since the Almanac of American Politics, 1994 had not been published at the time of the analysis. The vote for the prior incumbent was obtained from the Almanac of American Politics.
Party Monetary Assistance and Legislative Partisan Behavioral Loyalty
Data for political party candidate monetary assistance were collected from the Federal Election Commission Final Report: House and Senate Campaigns. Data are presented in two forms; direct contributions to


17
candidates and independent expenditures made on behalf of candidates by the party organizations. The data include all party direct and independent monetary assistance from all party organizations. Party unity scores were collected from Congressional Quarterly's Weekly Reports.
The Percent of Vote for the Democratic Candidate
The dependent variable, the percent of vote for the Democratic candidate, was obtained from the Almanac of American Politics and Congressional Quarterly's Weekly Report for 1982 through 1992.
Discussion
My assessment is premised on a belief in the importance of real electoral competition. For a democratic government to function, elections must at a minimum represent unbiased opportunity with regard to potential victory of qualified individuals. Inherent biases in the electoral system are intrinsically unequal. A limited government, such as that in the United States, derives its powers from the consent of the governed. When the electoral system is tainted with bias, consent is not truly given. In analyzing the consequences of proposed reforms in light of this research, a discussion of normative values is inevitable. Various interpretations of the results are possible; but in order to achieve significant insight, we must be willing to pull ourselves above the rhetoric and focus on the underlying key issues. To seek a


18
real understanding of the systemto seek answers to meaningful questions so informed reform is possible are the goals underlying the research presented in the coming pages.


CHAPTER 2:
THE RELATIVE ROLE OF CAMPAIGN EXPENDITURES AND CANDIDATE
QUALITY: DIFFERENTIALS IN ELECTORAL ADVANTAGES
Introduction
A notable amount of scholarly research has emerged examining the effects of money in Congressional elections following legislation in 1972 requiring the disclosure of campaign contributions and expenditures. A rich source of data became available to social scientists, data which potentially could shed some light onto the dynamics of campaign funding of Congressional elections. Much of the analysis has focused around the question "Can money buy elections?" If it does not "buy" victories, how significant of a role does it play in Congressional elections?
These questions are of utmost importance because of their critical implications on classic liberal democratic
theory. Key values of democratic governments such as legitimacy, accountability, and political equality are called into question. If campaign expenditures are of fundamental importance in electoral contests, an inherent bias is built into the systema bias toward moneyed interests and individuals. If such biases exist, competition will be stymied, which would directly impede democracy. A popular conception of democratic theory is the best person for the
19


20
job should be the one who elected; not the one with the most amount of money. If political elites do not have equality of opportunity in seeking political office then the legitimacy of the system is in question. How can elected officials be held accountable for actions if the system is biased by money? If accountability is impossible, how can a plural and
diverse nation be governed with any semblance of equality and legitimacy?
The central question here focuses on the actual role of money in congressional elections. The conventional wisdom holds that individuals with the most amount of money have an inherent advantage in electoral politics. Such opinions are espoused by many academics, journalists, practitioners, and laymen leading to a common call for reformreform which is needed if the conventional wisdom is correct, but is it correct? This is the fundamental question. Does the amount of money directly affect elections? Is its strength so powerful as to overshadow other variables? Can individuals simply buy their way into the national legislature? Or are other factors at work? Could it be that differences between candidates are more significant than previously thought? Could electoral advantages be less important as an absolute factor and more important in light of the level of competition?
These are the questions which I seek to answer. But, first it is important to amply understand the conventional approach to congressional elections and test its


21
applicability. Most research has been directed toward the effects of incumbency and has therefore focused on incumbent/ challenger Congressional races (i.e. Jacobson, 1986; Sorauf, 1988; and Stern, 1988). It is apparent by examining the extremely high rates of reelection that incumbents have benefits unavailable to challengers. It is difficult to ascertain whether this is a result of a rise in the availability of funds, higher name recognition, utility of staffs, franking privilege, travel allowances, or, more likely, some combination or interaction of these variables. In addition to these advantages, incumbents also are advantaged by the simple fact that they have won the same election for the same office; they have previously run, successfully, in the same district (barring re-districting.) However, the prospects for challengers is not entirely bleak. Challenger spending has a substantial and direct impact on the vote and has the potential of substantially altering the structure of the election (Green and Krasno, 1988). "As challengers spend more money, the importance of their party's previous performance decreases and candidate-specific attributes, typified by challenger political quality, come increasingly into play" (Green and Krasno, 1988, p. 900).
In incumbent races, the incumbent's record is often more important and salient than the challenger's background and qualifications. As Abramowitz clearly asserts, "by contrast a contest for an open seat generally involves a choice between two equally salient alternatives. As a result, the


22
candidate's qualifications and financial resources may have a stronger impact on the outcome of a contest for an open seat than the challenger's qualifications and finances have on the outcome of a contest involving an incumbent" (Abramowitz, 1988, p. 390). An examination of open seats avoids the worthwhile but cumbersome statistical problem of evaluating incumbency advantages and allows for a more lucid demonstration of the direct relationship between campaign expenditures and electoral results. Since incumbency effects are removed, one would expect that other factors would take a more significant role. One such factor is campaign expenditures.
Disbursements
Research has shown that expenditures have sharply different electoral effects for incumbents and challengers. The more non-incumbents spend, the greater their share of the vote, but the more incumbents spend, the smaller their portion of the vote (Jacobson, 1987, p. 180). Incumbent spending is seen as reactive; the stronger the perceived threat, the more money spent. Therefore, other things being equal, a given amount of money produces more votes if spent on non-incumbent rather than incumbent campaigns since challengers who spend more money win more often. Those non-incumbents who spend more than $250,000 have a one-in-four chance of winning. Those spending $150,000 have a one-in-ten chance of victory; the probability drops to one-in-fifty if


23
less than $50,000 is spent (Jacobson, 1987, p. 177).
The
more money the challenger spends, the more he or she is able to lessen the incumbent's relative advantages in familiarity
and popular regard among voters
Levels of familiarity and
regard are known to be strongly related to vote choice. Another significant advantage incumbents have over challengers is a "standing" campaign staff. Not only can their aides perform campaign services, the incumbents can call on prior volunteers and campaign staff for assistance in subsequent elections.
Table 2-1
Average Expenditures for Democratic and Republican Candidates
by the Average Percent of Vote Received in Open Seat
Elections for the U
S. House of Representatives
Averaqe Expenditures
Votes Received Democrats
Republicans
1982
< 40 %
46,193
218,201
40 60%
269,809
303,058
> 60%
312,806
519,419
1984
< 40%
116,233
136,669
40 60%
447,113
401,349
> 60%
290,694
422,366
1986
< 40%
204,662
141,076
40 -- 60%
440.115
451,521
> 60%
582,779
568,963
1988
< 40%
363,728
99,363
40 60%
615,557
676.320
> 60%
433,042
749,969
1990
< 40%
560,683
327,463
40 60%
559,810
510,675
> 60%
502,674
787,842
1992
< 40%
167,837
149,903
567,972
527,479
370,337
511,476


24
There is a direct and uniform relationship between the amount of money spent by candidates and the percent of votes received. As evidenced by the U.S. House data in Table 2-1, differences abound in the average amount of money spent by individuals winning open seat elections more than sixty percent, those who received between forty and sixty percent of the vote, and individuals receiving less than forty percent of the vote. These differences were consistent for both Democrat and Republican candidates indicating that expenditures have a direct and uniform influence on the vote for open seat elections for the U.S. House of
Representatives.
In presidential years, Democrats in competitive races consistently spend more money than Democrats in noncompetitive races. This could be the result of a self-perceived partisan disadvantage occurring in presidential election years (in most years under study, the Republicans won the presidency) or it could be the result of some other factor. The findings will be examined in depth in subsequent sections of the analysis.
Most scholars measure expenditures in sheer dollars spent. Gary Jacobson (1986) uses challenger expenditures and incumbent expenditures as separate variables because he expected different coefficients. A similar coding was employed in the research presented here. Campaign expenditures for the Democratic and Republican candidates were run as independent variables against the percent of


25
votes the Democratic candidate received (the dependent variable) using multiple regression (Table 2-2, Model A)1. As one can see, the amount of money spent by the Democratic and Republican candidates are highly significant, but the regression coefficients indicate that the money spent by the Republican candidate is more important. The negative relationship between the amount of money spent by the Republican candidate indicates that the more the Republican spends, the smaller the Democrat's margin of victory. Conversely, more votes will be received by the Democrat if he/she commits more funds. Table 2-2, Model A, indicates that for every $100,000 spent by the Republican candidate, the percent of vote received by the Democratic candidate decreases by approximately three percent. Conversely, $100,000 spent by the Democrat increases his or her percent of the vote by 1.4%. Possible explanations and implications of this will be discussed at length later in the analysis. As evidenced in Table 2-3, the Republicans benefit from an advantage in a differential return for campaign expenditures in every election cycle analyzed. This disparity is most prominent in 1992; in this election cycle, the money spent by the Democrat has an insignificant direct relationship with the amount of votes received by the Democratic candidate while the disbursements of the Republican candidate has a significant independent negative effect on Democratic
kXitliers excluded from analysis of campaign disbursements: 1992 CA district #22 and #36, NY district #12; 1990 NY #12; 1988 CA #12; 1986 MA #8, MY #8.


26
support
The relationship is similar in 1990, but to a
lesser degree
This would appear to be counter-intuitive;
either other factors are more important in this election
cycle or this is a poor measurement of
namic
relationship between campaign expenditures and the vote.
Table 2-2
Dependent Variable: Percent of Votes Received by Democratic
Candidate Open Seat Races for the U. S. House of
Representatives From 1982 Through 1992
Model A
Variable
Money of Rep
(in thousands)
Money of Dem
(in thousands)
B
T
Adjusted R-square Siq T | (Standard
Error)
-.031
12.44
0001
.014
5.8
0001
.38 (9.9)
Model B
Money of Rep.
(in thousands)
Money of Dem.
(in thousands)
Interaction ($Dem. $Rep.)
-.041
10.0
0001
.0035
.8
405
2.0 E-8
3.0
003
.40 (9.7)
Model C
Money of Rep.
(in thousands)
Money of Dem.
(in thousands)
Interaction
($Dem. $Rep.)
Prop. $ Dem.
.00014
-.018
8.3 E-9
46.6
.0027 .979
4.5 .0001
1.5 .14
11.0 .0001
17.44 .0001
.60 (8)
Model D
Proportion Money Dem
37 .99
N = 251
.55 (8.48)
Perhaps it is not how much money one spends that is of utmost importance, but how much one spends in comparison to one's opponent. Differentials in campaign expenditures are


27
especially important in open seat elections because both candidates may be relatively obscure and both can benefit from increased expenditures. Even if the candidates are familiar to the electorate, they can profit from high levels of disbursements, especially if there is a discrepancy in the amounts spent2. Also, the costs of running for Congress vary from district to district. Media costs will be cheaper or more expensive depending on the size of the district, the number of media markets that cover it, and the cost of exposure in the applicable market(s). Campaigning costs will vary in different areas. If the district is larger, it will cost more to travel through it to contact voters. If there is no tradition of volunteer activity in the district, workers will have to be hired.
As a base test of the hypothesis that differentials in
campaign expenditures are a significant predictor of the
vote, the interaction term (the amount of money spent by the
Democrat multiplied by the money spent by the Republican) was
included in a regression analysis (Table 2-2, Model B). The
interaction term is significant indicating that an indirect
effect of candidate expenditure is present. The amount that
each candidate spent is a significant predictor of the vote,
2Jacobson's research implies this in the sense that adding a specific amount to a poorer war chest closes the total proportion of money spent more than the same amount added to the richer chest. This relates to the nature of campaign expenditures; a differential in marginal returns for campaign expenditures exists. The differential in marginal returns for campaign expenditures is significant and must be
accounted for in a measure; by employing the proportion of money spent rather than sheer dollar amounts spent, this differential is incorporated.


28
but the interaction between the amounts spent is also significant. When the interaction term is included, the money spent by the Democrat becomes insignificant at the .05 level. Either the money spent by the Democrat is not an important factor in predicting the amount of votes received by the Democrat when the direct effect of the amount spent by the Republican and the indirect effect of the interaction of the amount spent by the two candidates are included, or another, more accurate measure is needed.
In an analysis of Senate elections, Alan Abramowitz (1988) uses the difference in the natural logarithm of the Democratic candidate's adjusted campaign spending and the natural log of the Republican candidate's adjusted campaign spending (adjusted for inflation and state population). Although his variable incorporates the theoretical logic of examining the differences between candidates' campaign expenditures, the interpretation is not straight-forward. Therefore, a simpler term will be employedthe proportion of money spent by the Democratic candidate. The proportion of money spent by the Democrat represents the percent of the Democrat's expenditures as compared to the total amount spent in the race (computed as money spent by Democrat divided by the total of the money spent by the Democrat plus that spent by the Republican.)
The proportion of money spent by the Democrat incorporates the idea of diminishing returns negating the need to resort to statistically cumbersome curvilinear


29
functions which are difficult to interpret. Furthermore, since the analysis incorporates several election cycles (six) it eliminates the need to control for inflation as the numerator and denominator are in same dollars (Campbell and Sumners, 1990). Also, such a coding approximates "reality", both in terms of practical campaign strategies, evaluations, and expectations as well as the electorate's impressions. Controlling for other variables, a disproportionate amount of spending would presumably benefit one candidate because levels of information and familiarity would subsequently increase.
The statistical robustness of the different measures is illustrated in Table 2-2. By using the proportion of dollars spent by the Democrat rather than the raw dollar amounts of each candidate as separate variables, the explanatory value
of the equation increases considerably (Adjusted R-square of .55 versus .38) (Table 2-2, Model D). There is a strong positive relationship between the proportion of money spent by the Democrat and the amount of votes received3. The strength of the variable is evidenced in its ability to hold its level of significance when included in a multiple regression model with other, highly correlated, variables (Table 2-2, Model C). In Model B, the money spent by the Republican candidate is highly significant while the money spent by the Democrat is not significant using standard
3For the reader's reference, scatter plots of these relationships are included int he Appendix A.


criteria when controlling for the interaction between the variables. When all four variables are run together, the money spent by the Democrat becomes significant and the money spent by the Republican becomes insignificant. The variables fluctuate considerably (changing from a significance level of .0001 to a level of .98) indicating the dynamic nature underlying the variables or a poor measurement. Since the proportion of money spent by the Democrat is highly significant in all models in which it is included, strong support is generated for its inclusion in subsequent analysis. The preceding statement is not meant to devalue the knowledge obtained from analyzing the amount of money spent by the individual candidates as separate variable, but to demonstrate the increase in knowledge obtained by looking at the differentials in expenditures. The proportion of money spent by the Democrat appears to statistically incorporate the direct and indirect effects of campaign expenditures while maintaining its theoretical importance.
An analysis was run for each election cycle to demonstrate that the results achieved were not biased by failure to adjust for inflation. As evidenced in Table 2-3, the model using the proportion of money spent rather than the more conventional variables is better with regard to every measure (overall fit, adjusted r-square, and standard error) for every year except 1990. Even in that year, however, the proportion of money remains a significant predictor. The variable does seem to capture the dynamics of the 1992


31
Dependent Variable
Table 2-3
Percent of Votes Receive by Democratic
Candidate Open Seat Elections for the U. S. House of
Representatives1982, 1984, 1986, 1988, 1990, and 1992
Variable
Model A
of Republican thousands) Money of Democrat
Money (in
(in thousands)
B
-.036
034
T 6.13
4.84
Siq. T .0001
.0001
Model B
Proportion of Money
41.62
11.53
.0001
Adjusted R-square (Model A) =.462 Adjusted R-square (Model B) =.715
Standard Error (A)= Standard Error (B)=
9.51 6.97
N = 55
1984
Model A
Model B
Money of Republican Money of Democrat
Proportion of Money
-.038 .02
36.82
4.03 2.75
6.72
.001 .01
.0001
Adjusted R-square Adjusted R-square
(Model A) (Model B)
=.397 Standard Error (A) =.648 Standard Error (B)
= 7.9 = 6.03
N = 25
1986
Model A
Money of Republican
Money of Democrat
-.03 .02
6.02 3.04
.0001 .004
Model B
Proportion of Money
41.29
8.50
.0001
Adjusted R-squ Adjusted R-squ
(Model A)=
(Model B)=
524 Standard Error
64
Standard Error
8.75 7.73
N = 39
1988
Model A
Money of Republican Money of Democrat
-.03 .02
3.05 2.23
.006 .037
Model B
Proportion of Money
39.58
3.65
.001
Adjusted R-square Adjusted R-square
(Model A) (Model B)
.26 .35
Standard E Standard E
12.58 11.82
N = 22
1990
Model A
Model B
Ad jus Ad jus
Money of Republican Money of Democrat
Proportion of Money
R-square (Model A)= R-square (Model B)=
.46 .258
-.03
.06 23.456
Standard Er Standard Er
4.71 1.62
3.06
(A) (B)
6.54 7.64
.0001 .12
.0055 N = 24
1992
Model A
Money of Republican Money of Democrat
-.03 .003
6.0 .6
.0001 .55
Model B
Proportion of Money
36.03
7.94
.0001
Adjusted R-square (Model A)= Adjusted R-square (Model B)=
344 .44
Standard Er Standard Er
ror (A)= ror (B)=
10.4 9.4
N = 82
election nicely. The variable is highly significant with a large t-value and unstandardized regression coefficient; as a
matter of fact, the beta in 1992 is the largest of all


32
electoral years. It appears that the money of the Democratic candidates is significant when considered in relation to the money of the Republican candidates.
The analysis presented here in combination with theoretical considerations provides strong support of the inclusion of this variable in the subsequent analysis. A significant amount can be learned by analyzing the independent affects of campaign expenditures for Democratic and Republican candidates, but more insight is provided by including the relative expenditures of the candidates.
Candidate Quality
Green and Krasno (1988) demonstrated that the strength
of the candidates, typically measured by prior elected
experience, is also a determinant in the outcome of
elections. They state that candidate quality is the sum of
two traitsattractiveness and skill. Attractiveness is
defined as qualifications for office in the form of political
experience and occupational background, fame or notoriety,
physical appearance and personality which assist in the
acquisition of electoral office. Political skill is a
candidate's ability to organize a campaign and present
himself or herself effectively (Green and Krasno, 1988, p.
886). The growth of the candidate-centered electoral
politics has increased the importance placed on candidate quality.


33
Candidates who have held prior elective office are much more likely to win elections; experienced challengers are
approximately four times more likely as an inexperienced challenger to beat incumbents (Jacobson, 1993, p. 246). Candidates running in open seat races also advantage from previously holding an elected office. "Better candidates attract more resources, just as the availability of resources attract better candidates" (Jacobson, 1993, p. 246). A person with prior elected experience may benefit from several advantages, including higher levels of name recognition and experience in campaigning. Campaign staffs and volunteers may be in place or at least readily accessible. Electoral districts may often completely or partially overlap; even if they are not the same, overlapping ensures some base of support. Candidates with prior elected experience are often seen as stronger candidates and are therefore able to raise more campaign contributions than inexperienced candidates. Experienced candidates have a higher probability of winning and gain far more votes even when national and local political circumstances are taken into account (Jacobson, 1989b, p. 781). Sabato states that attractive, well-financed challengers can provide a serious challenge to incumbents; the reason most incumbents are reelected is because they rarely face such challengers (Sabato, p. 81). Experienced challengers also benefit because they are better financed. The relationship between money and votes is reciprocal; campaign expenditures affect the vote, but the expected vote


34
affects campaign contributions. Open seat elections often are more competitive and therefore attract the largest number
of experienced candidates.
This research presumes that the essential characteristics which measures of candidate quality are trying to tap are levels of name recognition, electoral skills and viability of the candidacy. In other words, measures of candidate occupational background are surrogates for other measures which can not be directly measured in aggregate level analyses, most notably, name recognition, political attractiveness and skill. More electorally
experienced candidates are often advantaged because of the traditional, progressive nature of the career path, but other factors are also significant in influencing their electoral success. It may be less important that an individual held a particular office per se than that he or she acquired the accompanying name recognition and political skill associated with that position, which are potentially more significant.
Jacobson measures candidate strength by prior elected experience. It is coded as a dummy variable; if the candidate ever held elected office of any sort, he was considered experienced. Such coding presumes that name recognition and electoral skills are the driving force behind the relationship. Jacobson (1993) states that a dichotomous variable for candidate quality is used because it's objective, non-circular, and (most crucially,) available for entire period under study (p. 246). Model A, Table 2-4,


35
depicts the results of the regression equation using the proportion of money spent by the Democratic candidate and a dummy variable for candidate occupational strength. Although the variable reaches statistical significance, the validity of attributing any elected office as an advantage is
questionable. The coding asserts that name recognition and political skill come automatically with elected office elected office of any sort or level. Such broad categories fail to differentiate the subtle nature of the variable. Qualified candidates (candidates with some sort of prior elective experience) are over-concentrated in open seat races since they represent the best opportunity for victory (Jacobson, 1993, p. 260). Since there are so many high-quality candidates, one needs to differentiate further to adequately capture the underlying dynamic nature of the variable, especially when solely examining open seat elections. More refined distinctions need to be made.
Bond, Covington and Fleisher (1985) use a three point scale to reflect political experience. A score of three is given to individuals who were members of the state legislature, former members of Congress and candidates who won more than forty percent of the vote in a previous Congressional race. Elected city or county office (mayor, school board, city council,) or those who had some other politically useful experience (Congressional aide, members of politically prominent families in the district or party officials) were rated a score of two. All else were coded as


36
one. Those with scores of three were presumed to possess the highest levels of occupational advantagesname recognition and electoral skills. Those with ratings of two had either name recognition, political skill or both. Model B, Table 2-4, duplicates these classifications.
Table 2-4
Political Advantage of Republican and Democratic Candidates Dependent Variable: Percent of Votes Received by Democratic
Candidate
Variables
B
T
Siq. T
Model A*
Proportion of
Money Cand. Quality Dem
Cand. Quality Rep
34.6
2.5 -4.4
15.5
2.3 4.1
0001
02
0001
Model B**
Proportion of Money
Cand. Quality Dem Cand. Quality Rep
34
1.3 -2.5
14.4
1.8 3.7
0001
.07 .0003
Model C***
Proportion of
Money Cand. Quality Dem Cand. Quality Rep
33.7
1
-1.8
14.5
2.2 4.1
.0001
.03 .0001
N = 252
Adjusted R -square Model A Adjusted R-square Model B = Adjusted R-square Model C
.6 .6
= .6
Standard Error = 8.2 Standard Error = 8.3 Standard Error = 8.2
?Model A Dummy Variables for Occupation
Elected Experience or not. **Model B 3-point Ordinal Measure for Occupation. ***Model C 4-point Ordinal Measure for Occupation
The occupation of the Republican candidate was a significant independent factor explaining the percent of votes received by the Democratic candidate. The occupation
of the Democratic candidate did not succeed in meeting


37
statistical significance at the .05 level (however, it did meet the more liberal standard of .10). The results could simply be indicating that the occupation of the Democratic candidate did not have a significant direct affect on the percent of votes received by the Democratic candidate; or it could be indicating a need to further refine the measure.
Bond et al.'s coding mechanism is better than Jacobson's but it too has faults. The category of two encompasses many different types of individuals; all are politically advantaged, but not all are electorally advantaged. In other words, some have both political skills and name recognition, some have only name recognition, and others have only political skill. This distinction needs to be made. A further refinement would be to distinguish individuals which have name recognition and political skill (although presumably not as high of levels as those who were state legislators) from those individuals who possess only one or the other.
A four point ordinal scale measuring candidate strength was developed. Candidate quality was coded as follows:
4 = members of the state legislature, Governors,
and former Congress people. 3 = elected city or county office. 2 = non-elected political experience or
advantage.4 1 other.
including political activists, party officials, Congressional aides, White House Fellows, White House Council, and non-elected state officials, as well as candidates who formerly sought the office and received more than 35% of the vote.


38
This four point scale should be viewed as a candidate quality index; one that incorporates the underlying dimension of name recognition and political skill and advantage. The quality index is based on political experience and public exposure. This means of envisioning the variable encompasses the spirit of the dynamics underlying the effect of candidate strength. Model C in Table 2-4 depicts this coding mechanism and its direct affects on the percent of votes received by the Democratic candidates. Both candidate quality variables reach statistical significance and have high t-values. Given this analysis and theoretical considerations, the four-point ordinal index is the superior measure of candidate strength.
A further means to demonstrate the direct effect of candidate quality on the percent of the vote received by individual candidates is to examine the distribution of candidate quality ratings for candidates in races with Democratic and Republican winners (Table 2-5). In both cases, the candidates from the winning party were more politically advantaged than those from the losing party. Losers were consistently disadvantaged with regard to political and occupational experience.
The potential for indirect effects was also analyzed.
It has been hypothesized that political experience indirectly affects the vote via campaign expenditures; advantaged
candidates are able to raise more funds than less advantaged candidates. Little evidence is present to support this in
open seat elections for the U.S. House of Representatives


39
from 1982-1992. The low levels of the simple correlation between candidate disbursements and candidate occupation indicate little indirect effects are present (Table 2-6) Other factors must be affecting the amount of funds spent; those factors will be addressed in subsequent chapters of the analysis.
Table 2-5
Political Advantage by Electoral Outcome 1982-1992
Candidate Quality Index: 1 2 3 4
Democratic Winners Democrats 20 (14%) 21 (15%) 30 (21%) 15 (51%)
Republican 63 (43%) 23 (16%) 24 (16%) 36 (25%)
Republican Winners Democrats 42 (37%) 22 (19%) 23 (20%) 27 (24%)
Republican 21 (18%) 18 (16%) 22 (19%) 53 (47%)
Table 2-6
Correlation Matrix For Disbursements by Quality
Money Rep. Money Dem. Qual. Dem. Qual. Rep.
Money Rep. 1
Money Dem. .31 1
Quality Dem. -.12 .1 1
Quality Rep. .20 .03 -.11 1
In an analysis of Senate elections, Abramowitz (1988) finds that the outcomes of races for open seats are overwhelmingly based on the backgrounds and resources of the
5Note: separate correlation matrices were run for each election cycle separately with little change.


40
candidates The relative political experience of the candidates and the relative campaign expenditures of the candidates had the strongest effects. National political conditions and partisan and ideological makeup of the states
were much less important. To test the generalizability of
this in open seat elections for the U.S. House of
Representatives, variables were created to focus on the difference between candidates. The difference between
candidates was determined by using dummy variables to indicate the distribution of candidate quality ratings. For example, if the Democrat was rated a four and the Republican a one, the race was scored positively on the corresponding dummy variable. If the Democrat was coded as having a quality rating of four and the Republican two, the race was coded positively on the corresponding dummy variable. Since there were sixteen possible combinations of distributions of
occupational ratings, fifteen dummy variables were used. The results of the analysis are presented in Table 2-7. Six of the dummy variables meet statistical significance at the .10 level (only three at the .05 level). The findings could be indicating that the differences between the candidate's quality ratings are not significant, or that some other factor is at work.
By examining the correlation between the dummy variables and the proportion of money spent, a problem of multi-
6Abramowitz does, however, caution against generalizing from the results because of the relatively small number of open seat elections included.


41
collinearity is demonstrated. It appears that the differences between the candidate quality ratings are partly responsible for the differences in the amount of money spent (see Table 2-8). The average amount of money spent when the candidate quality ratings for the candidates are the same is nearly equal in almost every case. The average proportion of money spent by the Democrat also reflects the disparity between candidate quality ratings; when the Democrat has the highest rating and the Republican the lowest, the average proportion of money spent greatly benefits the Democrat. The same pattern holds true in the reverse situation. When the differentials between the candidate quality ratings are at a minimum, the differential between candidate expenditures is less dramatic.
Table 2-7
Difference in Candidate's Political Advantage Dependent Variable; Percent of Votes Received by Democratic
Variables
B
T
Siq T
Proportion of Money
Democrat Qual. Rating 4
Republican Qual Rating 1
36.15
2.6
16.15 .0001
1.1
29
Democrat Qual. Rating 4
Republican Qual Rating 2
-1.4
5
6
Democrat Qual. Rating 4
Republican Qual Rating 3
-3.5
1.2
25
Democrat Qual. Rating 4
Republican Qual Rating 4
-2.8
1.2
24
Democrat Qual. Rating 3
Republican Qual Rating 1
7.6
2.3
02
Democrat Qual. Rating 3
Republican Qual Rating 2
2.2
7
48
Democrat Qual. Rating 3
Republican Qual Rating 3
-4.2
1.4
15
Democrat Qual. Rating 3
Republican Qual Rating 4
-4.7
1.7
09
Democrat Qual. Rating 2
Republican Qual Rating 1
-1.5
6
6
Democrat Qual. Rating 2
Republican Qual Rating 2
-1.7
Democrat Qual. Rating 2
Republican Qual Rating 3
-6.3
4
1.9
7
06
Democrat Qual. Rating 2
Republican Qual Rating 4
-3.9
1.3
19
Democrat Qual. Rating 1
Republican Qual Rating 2
-.9
2.5
0001
Democrat Qual. Rating 1
Republican Qual Rating 3
-6.9
2
05
Democrat Qual. Rating 1
Republican Qual Rating 4
-4.4
1.8
08
N
252
Adjusted R-square Standard Error =
.6
8.1


42
Table 2-8
Difference in Candidate Quality Ratings by Proportion of
Money Spent in Open Seat Races from 1982 1992
Average Proportion of Money Spent by the Democrat
Same Candidate Dem. i; Rep. 1 .5
Quality Rating Dem. 2; Rep. 2 .5
1 Dem. 3; Rep. 3 .6
Dem. 4; Rep. 4 .5
Extreme Differences Dem. 1; Rep. 4 .4
in Quality Ratings Dem. 4; Rep. 1 .8 J
The evidence is mixed. It appears that a direct relationship exists between political advantage (evidenced by the differential in occupational ratings) and the percent of votes received by the candidate. Although the results are not as straight-forward as one would like, they are consistent with Abramowitz's findings as well as our previous analysis. Absolute resources are important factors, but differences between candidates are also significant. Disparities exist between candidates; the electorate views and acts on these disparities. Absolute political advantage is important, but the difference between two candidates must also be evaluated. Campaign expenditures are significant, but disparities in expenditures have a more significant impact on electoral results. These campaign advantages must be incorporated into any analysis of electoral behavior.


43
Discussion
Money does play a significant role in elections for open seat elections for the U.S. House of Representatives; in this respect the conventional wisdom is correct. But, its effect is neither all-encompassing nor as clear as previously envisioned. It is not necessarily the absolute amount of money which one spends that is of fundamental importance, but the relative differential in campaign expenditures of the two candidates which is crucial. The key seems to be the level of competition; if both candidates are competitive with regard to their campaign expenditures, the election is more apt to be competitive. The average percent of the vote for
the Democratic candidate of the races in which the proportion of money spent by the Democratic candidate was less than
four-tenths of that spent by the Republican was forty-one percent; the average vote for the Democratic candidate was fifty percent in the elections where the proportion of money spent was between .4 and .6. The average increased to sixty-four percent in those cases where the Democratic proportion was greater than six-tenths, indicating that the electoral competition is directly tied to the differential in campaign expenditures. If one wanted to increase the competitiveness of open seat elections, the means to do so would be to decrease the disparity in expenditures between candidates. If the differential between candidates were reduced, the role of money in elections would also be reduced.


44
Campaign expenditures are significant in congressional elections, but they are not so powerful as to overshadow other factors. Many are troubled over the large impact that disbursements have in electoral politics. Their concern should be tempered because the impact is not all encompassing. Candidate quality plays a significant direct role in electoral outcomes. Candidates who are more politically advantaged than their opponents are far more likely to win than those who are not. Politically and electorally advantaged candidates receive more campaign contributions than those who are not; a fact which is most evident when great disparities exist. Differences in electoral advantage partly account for the differentials in campaign contributions. Both prior political and electoral skills and name recognition prove to be significant factors in determining electoral success for open seat elections from 1982 to 1992; examination of differentials in these advantages is also enlightening, especially given the indirect effect via differentials in campaign expenditures.
It has been evidenced that Republican candidates benefit from a differential return in campaign expenditures compared to Democratic candidates. Money spent by Republican candidates has a larger impact on electoral results than that spent by their opponents. Several explanations could account for the differential return of Republican expenditures. It may be that campaign expenditures have larger returns for Republicans because they have to overcome their party's


45
minority status at the congressional level. A second possibility is that Republicans use their expenditures more efficiently.
The first seems contrary to what one would expect. Democrats should have a party advantage since they have controlled the House during the entire period under study. Given this, one would suspect that the Democratic party would have a significant electoral advantage given their incumbency advantage. Does the party's incumbency advantage transfer to a subsequent electoral advantage? In other words, is an incumbency advantage transformed into a party advantage? This question will be addressed in the following chapter. If Republicans benefit from being the minority party by having differential returns on campaign expenditures, can we anticipate that they will also benefit in other electoral factors as well?
An alternate explanation is that the Republicans use their campaign expenditures more efficiently. They may have superior methods of targeting potential voters either by direct communication or electronic means. Their campaign messages may be constructed to be more salient and easier to communicate than their opponents'. Republican candidates may benefit from their party's assistance in polling, technical assistance, get out the vote drives or other methods of assistance which could allow them to use their funds in other avenues or in a more efficient manner. At this point there is little we can do but speculate as to the causes of this


46
disparity, but there is no denying the fact that the differential is consistent and significant in almost every election under study.
One needs to examine both candidates separately in an analysis, but then it is the difference between the candidate that is more important. Campaigns are not conducted in a vacuum; opposition exists and is evaluated by the electorate. The evaluation between candidacies must be incorporated into the electoral model. It is not, for example, purely how strong one candidate is, but his or her strength relative to his or her opponent's. It is best to envision this
differentiation as relative campaign advantages; is one candidate more politically advantaged than his/her opponent? This question is of utmost importance. It is less important that one opponent, for example, spends a large amount of money, but that one opponent does and her opponent does not. A situation of this type produces a differential in campaign advantages that will prove to be significant in explaining votes in elections for the U.S. House of Representatives.
The key to understanding aggregate electoral behavior is to examine electoral advantages. Although a significant amount of insight can be obtained by examining absolute advantages, more results from analyzing differentials in electoral advantage. It is not so much that one candidate is electorally advantaged, but that one candidate is and his or her opponent is not. Counter to conventional wisdom, absolute amounts of campaign disbursements are less important


47
than differentials in expenditures. Competition would increase if the differentials between candidates are minimized. If nearly equal amounts of money were spent, than the role of money in elections would be minimized. The same holds true for occupation; to increase competition, candidates who are more politically advantaged should run against one another. If we seek to maximize competition, we need to encourage more quality candidates to enter races and also seek to equalize the amount of money spent.
Bias created by money does exist, but accountability and legitimacy are still possible to achieve. Since campaign expenditures are not all encompassing and overwhelming, other factors can affect the electionfactors such as candidate occupation, presidential coat tails, the incumbent party advantage, and economic factors. These and additional factors will be examined in the next two chapters, but at this point we can conclude with certainty that money is important, but that other factors are also significant in determining the electoral results.


CHAPTER 3:
NATIONAL, ECONOMIC AND LOCAL FACTORS IN OPEN SEAT ELECTIONS
FOR THE UNITED STATES HOUSE OF REPRESENTATIVES
FROM 1982 THROUGH 1992
Introduction
The role of money in Congressional elections has been amply demonstrated in the previous chapter. Candidates who spend more money than their opponents have a significant electoral advantage, but the advantage is not all encompassing. Other factors can, and do, also influence the electoral results. One such variable is the candidate's occupational background; more qualified candidates fare better in races than the less qualified. This is especially prevalent in cases where a disparity of occupational advantage is present; candidates with quality advantages over their opponents increase their likelihood of victory.
These results may ease some worry that many individuals have over the state of electoral politics in the United States. Fundamental underpinnings of democratic states rest on competitive elections; if money plays an overwhelming
role, true competition is biased and may be impossible to achieve. If elections are biased, electoral accountability may be jeopardized. Many hold accountability to be a fundamental goal of democracy; the elected officials must be
48


49
held accountable to those who elected them. By accountable, it is meant that the elected officials themselves, and others like them (for example, fellow partisans), must be held to account for their actions, policies and any subsequent implications of these policies. If the electorate does not support policies, they can "use the ballot box to send new, presumably wiser and more honest individuals to Washington ." (Rieselbach, 1986, p. 12). The government must be responsive (governmental policies must be congruent with public preferences) to the electorate to justify and legitimate its authority and existence. If elections are biased, it is nearly impossible for the government to be held accountable or responsive since competition is stymied. Biased elections decrease the ability for responsive legislation because elections are not necessarily producing quality representatives with constituency loyalty; the elections are simply reflecting the biases. If elections, for example, are solely or primarily influenced by money, then elections are biased toward the moneyed interests; this severely contradicts several basic tenants of democracy. If other factors influence the electoral process, for example presidential coat-tails, there is an increased likelihood of holding the institution accountable and subsequently producing more responsive legislation. When other factors influence the election, the bias toward money may be minimized; with the minimization of the bias, the potential of accountability is maximized.


50
National Factors
A significant amount of research exists in which the role of national factors in Congressional elections is examined (for example, Campbell et al., 1960; Kiewiet and Rivers, 1985; Fiorina, 1978, 1981; Jacobson and Kernell, 1983; Erikson, 1990). Such interest is warranted given the implications for policy making, electoral accountability, and inter-governmental relations. Do national factors such as Presidential coat-tails and the economy play a direct and significant role in congressional elections?
Presidential Coat-tails
The theory of Presidential coat-tails contends that presidential elections, in and of themselves, affect Congressional elections. The theory states that in presidential election years, candidates in the winning presidential candidate's party will have an advantage over their opponents; this advantage will be larger when the margin of victory of the presidential candidate is especially large. Presidential "coat-tails" are said to "carry" members into office. Candidates of the president's party have an inherent advantage (or disadvantage if the candidate is unpopular.) The strength of the coat-tail effect will be in relation to the popularity of the presidential candidate; the advantage will be larger with especially popular or victorious presidential candidates. In mid-term elections,


51
no such advantage is present. Presidential coat-tails may
serve to connect the president and members of Congress.
According to the theory, candidates benefit because they are in the same party as a popular presidential candidate.
Accordingly, a variable was coded to reflect the advantage
accorded to candidates of the winning presidential
candidate's party. In 1984 and 1988, Republicans were coded
as benefiting from Ronald Reagan's and George Bush's
respective victories; the Democrats were coded as suffering a
disadvantage. The same coding was used in 1992 to reflect
the Democrat's advantage incurred by Bill Clinton's victory
and the disadvantage suffered by the Republican's with Bush's
failed campaign for re-election. The theory states that the
coat-tail effects are removed in non-presidential elections.
Since this is the case, no advantage or disadvantage was
coded in off-years. The coding is as follows:
1 = Republican candidates in 1984 and 1988;
Democratic candidates in 1992.
0 = All candidates in 1982, 1986 and 1990.
-1 = Democratic candidates in 1984 and 1988;
Republican candidates in 1992.
Table 3-1, Model A, depicts the results of the coat-tails
variable along with the proportion of money spent by the
Democratic candidate and candidate quality indices1 to predict
1See Chapter 2 for a complete discussion of these variables. Briefly the proportion of money spent by the Democratic candidate is calculated by dividing the money spent by the Democrat by the total money spent (the money of the Democratic plus the money of the Republican.) The candidate quality index is based on political experience and public exposure. It is coded as a 4 point ordinal score; 4 = state


52
the percent of vote received by the Democratic candidate. The presidential coat-tails variable fails to attain statistical significance. The results could be indicating that presidential coat-tails do not directly effect elections for the House or it could be indicative of a need to further refine the measure. One means to accomplish this would be to include the vote for the Democratic presidential candidate. By including the district vote for the Democratic presidential candidate2, one is able to distinguish districts where the candidate does especially well or poorly. The refinement proves to be a significant improvement in
predicting the percent of vote the Democratic candidate for the House (Table 3-1, Models B and C). If the presidential vote for the Democratic candidate is included, the relative role of campaign expenditures is diminished; the t-values indicate that the percent of vote for the Democratic presidential candidate is a stronger variable in predicting the vote for Democratic candidates for the House than is the proportion of money spent by the Democratic candidate.
The problem, however is that we may be measuring either a presidential coat-tail effect or the underlying partisan distributions in the district. One means to settle the matter would be to separate presidential elections and mid-
legislators, former Congresspeople and Governors; 3= county or local elected office; 2 = non-elected political or electoral advantage; 1 = everyone else.
2In midterm elections, the vote from the prior presidential election was used.


53
term elections
Table 3-2 demonstrates the resulting
differences.
Table 3-1
Effect of Presidential Coat-tails in Open Seat Elections
for the U. S. House from 1982-1992 Variable:
Dependent
Percent of Votes Received by the Democratic Candidate
Model A*
Model B**
Model C***
= 252
Variables
Proportion of Money Candidate Quality
Index Dem. Candidate Quality
Index Rep. Coat-tail
Proportion of Money Candidate Quality
Index Dem. Candidate Quality
Index Rep. Coat-tail Percent Dem.
Vote
Proportion of Money Candidate Quality
Index Dem. Candidate Quality
Index Rep. Percent Dem.
Vote
20.7 1.1
-1.2
2.2
20.7 1.1
-1.2
16.12 2.1
1.4
9.5
3.4
11.6
9. 6 .0001
3. 1 .002
3. 5 .001
11. 7 .0001
Siq T
.0001 .04
.0001
.17
.0001 .002
.001
.75 .0001
Adjusted R-square Model A = Adjusted R-square Model Adjusted R-square Model
Standard Standard E Standard
= 8.2 = 6.6 = 6.6
?Model A Pooled Data 3-Point Scale for Coat-tails **Model B Pooled Data 3-Point Scale for
Presidential Coat-tails and Vote ***Model C Pooled Data Presidential Vote


54
Table 3-2
Presidential Coat-tails: Open Seat Races from 1982-1992 Dependent Variable: Percent of Vote Received by Democratic
Candidate
Presidential
Elections
(1984, 1988,
1992) N = 130
Mid-term Elections
(1982, 1986,
1990)
N = 122
Variable
Proportion of Money Candidate Quality
Index Dem. Candidate Quality
Index Percent of Dem.
Pres. Vote
Proportion of Money Candidate Quality
Index Dem. Candidate Quality
Index Rep. Percent of Dem.
Pres. Vote
26.4 1.2
-1.3
.4
8.7 2.2
2.6
6.3
Adjusted R-square Presidential Elections = .8
Standard Error = 6.4 Adjusted R-square Mid-term Elections = .7
Standard Error 6.3
.0001 .03
.01
.0001
Both models are good predictors of the vote cast for the Democratic candidate in open seat elections from 1982 to 1992, but the relative influence of the variables under consideration varies considerably. Elections in years when presidential elections were held, the relative impact of campaign expenditures decreased in importance. In fact, the vote for the Democratic presidential candidate had greater
impact on the vote for Democratic candidates than the proportion of money spent by the Democratic candidate. In mid-term elections, the relative impact of the two variables changes; the presidential vote is still a highly significant factor, but the proportion of money spent has a stronger


impact. Given this, it would seem that including the percent of votes cast for the Democratic presidential candidate in the district is measuring both a coat-tail effect and an underlying party composition in presidential election years. In mid-term elections, it most likely is measuring the remnants of the coat-tail effect and the underlying partisan composition of the district.
These results provide support for the concepts of accountability and responsibility. If presidential election results are influencing Congressional elections, several implications result. First, a connection between presidents and members of Congress follows; Congress-people may feel some gratitude to the president for an electoral advantage and may reciprocate, thereby contributing to the president's ability to enact key legislation. Secondly, if individuals are holding the candidates of the president's party to the same standards as the president, partisan accountability is maximized, potentially resulting in greater governmental responsiveness. Finally, it is widely accepted that the economy has a direct effect on presidential elections (Tufte, 1978; Erikson, 1989); if presidential coat-tail effects extend to congressional elections, the economy has a significant indirect effect on elections for the House.
Surge and Decline
The theory of "surge and decline" was developed to explain two events in American politicsthe consistent


56
pattern of congressional vote and seat loss by the party of the president in midterm elections and the lower levels of turnout in midterm elections. Agnus Campbell (1966) designed a theory which accounted for these two occurrences by using two fundamental distinctions; high versus low stimulus elections and core versus peripheral voters (Cover, 1985). The theory contends that presidential and mid-term elections are fundamentally different; in presidential elections, individuals surge to the polls; these individuals normally do not vote but are mobilized by the highly visible and salient presidential elections (Campbell, 1966). The party of the victorious presidential candidate benefits electorally from the surge since most of these individuals tend to "jump on the band-wagon" and vote for the leader. When the high stimulus presidential elections are removed in mid-term elections, the turnout returns to normal. The congressional vote will reflect the stable cues of partisanship. Since the voters mobilized by the presidential election are the most subject to coat-tail effects, "their withdrawal erases whatever advantage the president's party's congressional candidates enjoyed from his presence on the national ticket. Consequently, the president's party should normally lose votes and seats at the mid-term" (Jacobson and Kernell, 1983,
p. 61).
Cover's findings (1985) contradict part of Campbell's original theory. The theory adequately explains why voters surge to the polls during presidential elections, but it does


57
not explain well why the President's party regularly loses seats in the House of Representatives in midterm elections (Cover, 1985, p. 613). James Campbell devised a revised theory of surge and decline to better capture the later part of the occurrencethe 'decline'. Midterm referendum theories (Kernell, 1977; Tufte, 1975; Erikson, 1988) suggest that the midterm electorate evaluates presidential behavior and punishes the president's party (Campbell, 1991). The two theories of surge and decline and midterm referendum are compatible and as later demonstrated, make substantial statistical and theoretical contributions into the understanding of elections for the House of Representatives,
James Campbell's revised theory of surge and decline combines aspects of the original theory with those of the midterm referendum theory. He states that "In essence, midterm elections are, at least in part, repercussions from the previous presidential election" (Campbell, 1993, p. 222). In presidential election, short-term factors benefiting the president's party are translated into a surge of votes for that party in two ways. Independent voters are influenced by short-term factors which affect their vote choice (positively to benefit partisans in the president's party); partisans are stimulated to turnout in higher rates (Campbell, 1993, 1992). Partisans primarily contribute to the presidential surge via their turnout decisions; systematic differences exist in the patterns of turnout for the two parties. Partisans of the advantaged party (the winning president's party) turnout in


58
disproportionately high rates because of enthusiasm. Disadvantaged partisans will turnout in less than normal rates because they are presented with a difficult decision; should they vote for an unacceptable candidate (or simply one which will not win) or should they defect and vote for the other party's candidate. Many choose simply not to vote; depressing their turnout rates. High levels of defection are most common in elections which result in land-slides (Campbell, 1992, 1993).
The short-term factors which benefited the winning presidential candidate's party are diminished and less systematically advantageous in the midterm election. The extent of the decline depends partly on the degree of the prior surge for the president's party. The greater the surge, the greater the subsequent loss (Campbell, 1993). Although national factors are less important in midterm elections, they are nonetheless potentially consequential. Midterm elections are in part a referenda on the performance of a president. "A popular president can shorten his party's midterm fall and an unpopular president can lengthen it" (Campbell, 1993, p. 225). However, even a very popular president will be unable to turn the midterm into a gain (Campbell, 1993). Midterm losses for the president's party are in proportion to the party's prior vote margin (Campbell, 1991, 1992, 1993). "The president's party consistently loses in the midterm because of the prior surge; its losses vary a good deal because the public's midterm judgments vary, but


59
also, at least as importantly, because the magnitude of the prior presidential surge varies" (Campbell, 1993, p. 235). Although Campbell notes that the presidential surge and decline effects are weaker than in the past, presidential surge and decline effects on congressional elections remain substantial.
An initial examination of the data for open seat elections for the U.S. House of Representatives from 1982 through 1992 provide partial support for the theory of surge and decline (Table 3-2). As evidenced by the regression coefficients, the percent of vote cast for the presidential candidate in the district has a greater impact in presidential elections than it does in midterm elections. Also, the proportion of money spent by the Democrat has a larger impact in midterm elections than in presidential elections. In presidential election years, the short-term stimulus of presidential voting is the most consequential factor predicting the vote. The variable retains its significance in midterm election years, but it is no longer the variable with the greatest impactthe proportion of the money spent by the Democrat is. In midterm elections, money is of greater relative impact than in presidential election years indicating that other factors (short-term support for the winning presidential candidate's party) are more substantial in presidential years. When the short-term stimuli are removed, the proportion of money spent by the


60
Democrat becomes the most consequential variable predicting
the percent of
Democrat
Table 3-3 Revised Theory of Surge and Decline
Variables
Model A
Presidential
Proportion of Money of Dem.
Years ( 1984, Candidate Quality
1988, 1992)
Model B
Mid-term
(1982, 1986, 1990)
Model C
Pooled Data
Index Dem. Candidate Quality
Index Rep. Percent of Vote
Dem. President Presidential
Approval
Proportion of
Money of Dem. Candidate Quality
Index Dem. Candidate Quality
Index Rep. Percent of Vote
Dem. President Presidential
Approval_
Proportion of
Money of Dem. Candidate Quality
Index Dem. Candidate Quality
Index Rep. Percent of Vote
Dem. President Presidential
049
26.4
1.2
-1.2
-. 1
5.3
1.9
2.9
10.4
8.7
2.2
2.3
6.2
1.0
9.6
3.1
3.6
11.8
.049
1.1
Siq T
.0001
.06
.004
.0001
.61
0001
0001
0001
002
0004
0001
Model A Model B Model C
Adjusted R-square = Adjusted R-square -Adiusted R-sauare
8 7 7
Standard Standard
= 6.4
= 6.3
Standard Error = 6.6
In order to test the theory's contention that midterm loses are partly contingent upon the electorate's evaluation of the president, Gallop's measure of presidential popularity


61
is examined3. Table 3-3 depicts the results of the regression analysis incorporating presidential approval ratings4. The presidential approval ratings do not reach statistical significance in any of the models. According to Campbell's model, presidential approval ratings should have an effect in mid-term elections since short-term national factors (operationalized as presidential approval ratings) have a greater impact in mid-term rather than presidential elections. Such expectations are not reached in elections analyzing open seat elections for the U.S. House of Representatives from 1982 through 1992.
In presidential election years, the percent of the vote cast for the Democratic presidential candidate in the district has the most substantial impact on the percent of votes received by the Democratic candidate. However, in midterm election years, the proportion of money spent by the Democratic candidate has the greatest impact. The findings support the hypothesis that president's party benefits from the partisan surge to the polls with greater electoral benefits. Individuals in the president's party and independents supporting the president are probably turning out in larger numbers, as put forth by the theory, resulting in greater support of the party's candidates for open seat
3The average Presidential Approval rating in the ten months prior to the election was used. The approval ratings were obtained directly from data released by the Gallop Corporation.
4Campbell also included a midterm economic change variable in his analysis. Since the variable failed to met statistical significance, it will not be included in this analysis.


62
races. However, little evidence is present to support the second aspect of the theory regarding the subsequent mid-term decline. Such results are surprising, since open seat elections seem to present the greatest opportunity for the theory to reach fruition since the races involve more equal contests than typically evidenced in incumbent/challenger races. In open seats, one would expect short-term national factors to be substantial (if they are to be substantial at all) because the races to not involve more insulated
incumbents. Strategic Politicians
National politics may influence Congressional elections by affecting candidates' decisions to run, thereby indirectly influencing the electoral outcome distribution. Politicians' career plans and decisions are "strategically adapted to the political environment. National political forces which politicians expect to have some impact on voters shape their election plans. As a result, the relative quality of a party's candidates and the vitality of their campaignsthe things that have the strongest impact on individual votesare not at all independent of national events and conditions. Rather, they are a direct function of them" (Jacobson and Kernell, 1983, p. 19). Strategic politicians allow for meaningful elections because those
elections structure the vote choice so that electoral results


63
are consonant with national-level forces even if individual decisions are not (Jacobson and Kernell, 1983, p. 2).
One means to test the Jacobson-Kernell thesis is to determine if the quality of the candidates in different elections varies, and if so, can this variation be explained by changes in the national conditions that will affect the calculus for politicians to seek electoral office. It appears that the explanation may have some validity. Significant differences (as determined by a two-tailed, paired t-test) exist between candidate occupational ratings and candidate campaign expenditures for 1982, 1986 and 1992. The t-test probability that these measures were significantly different was greater than .999. There were no significant differences in candidate occupational ratings and/or expenditures for 1984, 1988 and 1990. These measures were chosen because candidate quality is often measured by occupational background and candidate expenditures. The theory states that stronger candidates will run when national partisan factors favor their political party. 1982 was a "bad" year for the Republicans because the country was in a deep recession and unemployment was high. The Democrats sported a better crop of candidates with regard to candidate quality and disbursements than the Republicans (Table 3-4). In 1992, the same holds true; the Democratic candidates were significantly advantaged both with regard to candidate quality and campaign financing.


64
Table 3-4: Strategic Politicians Open Seat Races from 1982-1992
II Averaqe Averaqe Averaqe 1
Quality Quality Averaqe Averaqe Proportion I
Year Ratinq Ratinq Expend. Expend. Money
Dem Rep. Dem. Rep. Dem.*
1982 2.7 2.5 266679 288792 .52
(1.2) (1.3) (192915) (229682) (.26)
1984 2.5 2.7 362166 373790 .49 |
(1.1) (1.3) (247775) (187369) (.22)
1986 2.6 2.8 363551 445146 .44 |
(1.3) (1.2) (291966) (246779) (.26)
1988 3.1 2.2 521515 589685 .48 |
(1.1) (1.1) (284345) (356885) (.23)
1990 2.7 3.2 544146 528743 .49 1
(1.2) (1.1) (342261) (246168) (.2)
1992 2.9 2.2 458419 335890 .62 1
(1.2) (1.3) (248056) (253545) (.23) |
(?Standard Deviation)
Since only open seats are examined, the calculus determining whether candidates would run may be slightly different than those in incumbent/challenger races. The results here however do provide definite support for the Jacobson-Kernell theory of strategic politicians. If the analysis examined all races in these years, one might find stronger support for the theory, possibly answering the question as to why there were not significant differences in the candidates in 1990 (presumably a time Republicans would deem a good time to run), 1988 (a year which should have benefited the Democrats since a lame-duck president occupied the White House,) and 1984 (the Republicans should have been advantaged since a strong, incumbent president sought and achieved re-election.) The analysis presented here provides support for the theory, but not a definitive endorsement.


65
Economic Voting
The role of the economy in electoral behavior has generated a significant amount of research (Downs, 1957; Fiorina, 1981, 1983; Erikson, 1990; Kinder and Kiewiet, 1981; Alesina and Rosenthal, 1989; Alesina, Londregan and Rosenthal, 1993). Economic voting is often seen as a referendum on presidential activity; if the president's economic policies have positive results, it is believed that the presidential party's candidates are electorally rewarded. Conversely, punishments are believed to be implemented if the president's economic policies result in negative consequences. The economy is seen as directly and indirectly affecting congressional elections. Indirect effects are evidenced via presidential coat-tails and evaluations. The two most prominent attempts to account for economic effects are traditional retrospective voting (also known as pocketbook voting,) and sociotropic voting. Although these theories are typically employed in individual level analyses, the concepts can be transferred to the aggregate level.
Traditional "Retrospective" Voting
Retrospective voting requires the assumption of rational citizen behavior. It is assumed that the rational person always acts to maximize his or her own greatest benefits. Downs contends that each citizen will vote for the party he or she believes will provide the greatest utility income


66
(Downs, 1957). He sees every election as a judgment passed upon the record of the incumbent party; the outcome calls for "change" or "no change" (p. 41). Fiorina states that elections may not signal the direction in which society should move, but may convey where it has been; rather than a prospective decision the voting decision can be more of a retrospective decision. Elections may "constitute a collective evaluation of actual trends from past to present as much as one of proposed trends from present to future" (Fiorina, 1981, p. 6). The electorate judges retrospectively and commands prospectively by approval or disapproval of past events. A retrospective voting electorate will enforce electoral accountability (in an ex post sense) (Fiorina, p. 11). Retrospective voting requires less information than other electoral models; it is a cost-cutting method for voting decisions. Traditional reward/punishment theory contends that elections have no policy implications other than a generalized acceptance or rejection of the status quo.
According to the theory of "pocketbook voting", individuals vote for the "in" party when their own economic situations improve or for the "out" party when their personal economics worsen. Little survey evidence of pocketbook voting in Congressional elections exists (Erikson, 1990; Fiorina, 1978; Kiewiet, 1983; Born, 1986). This analysis of aggregate data provides similar results. State unemployment


67
rates5 from the second quarter6 of the election year and that of the second quarter from the prior election year7 were used to test the concept of state economic factors affecting the vote, in the aggregate, in open seat elections for the U. S.. House of Representatives from 1982 to 1992. Table 3-5, Model A demonstrates the results of the regression analysis. The variable was developed in order to test hypotheses derived from the theory; if the state economic condition improved from the prior year, the party of the incumbent president was coded as benefiting. If the economic conditions worsened (unemployment increased from the prior year,) the variable was coded as benefiting the "out" party. If no change occurred, no reward or punishment was awarded. This coding was also used with the party of the incumbent as the reference group for attributing blame or reward rather than the party of the president with similar results (details not shown.)
National Economic Voting
It may be the case that the reference for these calculations is not the individual and his or her economic situation, but the nation and its economic state. "In
5State unemployment rates were used since this the most refined measure available.
6Voters appear to base their decisions on economic conditions of the recent past (see Kiewiet and Rivers, 1985 and Erikson, 1990).
7Kiewiet and Rivers (1985) state that unemployment or change in per capita real GNP appear to work as well as change in real per capita income (see also Fair, 1978).


68
reaching political preferences, the prototypic sociotropic voter is influenced most of all by the nation's economic condition. Purely sociotropic citizens vote according to the country's pocketbook, not their own. Citizens moved by sociotropic information support candidates that appear to have furthered the nation's economic well-being and oppose candidates and parties that seem to threaten it. Thus the party in power suffers at the polls during hard times because voters act on their negative assessments of national economic conditionsquite apart from the trials and tribulations of their own economic lives" (Kinder and Kiewiet, 1981, p.
132). This may relate to the traditional Protestant work ethic that individuals are responsible for their personal well-being. "If we work hard enough, we can succeed."
Personal failure can be attributed to personal short-comings; but if the nation is in economic straits, blame is often attributed to outsiders and officials. It is not so much that one individual is unemployed, that could be his or her own personal fault, but if many are unemployed or the unemployment has worsened, or greatly improved, some one else must be blamed or rewarded.
A coding similar to that used to test the existence of state economic voting was employed, except that the reference was changed from the state level of unemployment to the national level. The results are presented in Table 3-5, Model B. National factors affeting the vote also fail to meet statistical significance at the .05 level, but did reach the


69
.10 level, indicating that individuals may be voting for the "nation's pocketbook" in the aggregate.
Table 3-5: Economic Voting Open Seat Elections from 1982-1992
Dependent Variable: Percent of Votes Received by Democratic
Candidate
Model A
Model B
Variables
Proportion of Money Candidate Quality
Index Democrat Candidate Quality
Index Rep. State Econ. Voting
Proportion of Money Candidate Quality
Index Democrat Candidate Quality
Index Rep. National Economic
Voting
B
33.6 1
-1.7
.2
33 1
-1.8
.9
T
~4.7
2.2
4
.5
14 2.2
4.1
1.7
Siq T
0001 03
0001
65
0001 03
0001
09
N = 251
Adjusted R-squar e Model A = .6 Standard Erro r = 8.2
Adjusted R-sc[uar e Model B = .6 Standard Erro r = 8.2
The results are consistent with Erikson's (1990) conclusions that economic conditions matter little in Congressional elections if one employs a strict standard for the level of statistical significance accepted. However, if one uses a less strict standard for statistical significance, the .10 level, national economic conditions do affect open seat elections for the U.S. House. Indirect effects of economic conditions on congressional elections are also evidenced via presidential coat-tails (See Table 3-3). Congressional vote shares appear to be less sensitive to


70
economic conditions than is the Presidential vote. This research does provide some support for the assertion that national economic factors affect open seat congressional races, but no evidence is present of support that local economic conditions have a similar effect.
Local Factors
It may be appropriate to examine the potential influence of local factors. As the old saying goes, "All politics is local politics." Of special interest are the potential for a transfer of incumbency advantage to a party advantage and the nature of the district. One would expect different dynamics to occur in competitive versus non-competitive districts.
Incumbency Loss Potential
In open seat elections, the party that previously held the seat (and its candidate) would seem to have the edge, possibly an "incumbency advantage" that would transfer into party advantage. For instance, contributions to the candidate of the dominant party might be larger than those to the minority party. Even if one party is not dominant, party strength in the district may translate into significant campaign and electoral advantages for the candidate from the prior incumbent's party (for example readily available volunteers or benefits from habitual voting for the candidate of that party. ) The determination of strength is important because spending may matter more when a candidate suffers


71
from the electoral disadvantage of being from a minority party.
The strength of the party in the district is measured
by the percent of votes the party received in the last
election8. Analyses were run with presidential and mid-term
elections separately and also with pooled data. Table 3-6
depicts the results. District party strength, measured by the
prior vote for the Democratic candidate, was significant in
the pooled analysis as well as in the presidential election
years, but insignificant in mid-term years at the .05
statistical level. In order to demonstrate a loss of
incumbency advantage rather than simply the occurrence of an
"retirement slump", the party of the incumbent was
controlled. If the incumbent was a Democrat, the variable
was coded as benefiting the Democratic candidates. The
variable was not significant in all types of elections,
possibly indicating that the party of the incumbent is simply
insignificant. However, the results may be indicating an
alternative explanation is necessary. The later is probably
the correct interpretation.
In mid-term elections, one would expect that the vote
for the Democrat in the last election would be of most
importance since the presidential vote is less important
(Table 3-2). But, it is in these elections that the vote for
8A11 races from 1992 were excluded in this analysis since past projected voting was unavailable at time of writing. New districts and those severely changed by the 1982 election
were also excluded; the remaining were included. Please note: a separate analysis was run excluding all cases from 1982 with very similar results.


72
the prior Democratic candidate is not significant. Such an occurrence seems counter-intuitive; since national factors are less significant, one would expect local factors to be significant. In fact, the negative direction of the variable denoting the party of the incumbent (Table 3-6, Models A, B and C) indicates that Democratic candidates actually lose votes if the incumbent is from their party, indicating that the potential for an incumbency advantage is not fully realized. The vote for the incumbent in the prior race was perhaps largely a personal vote, not a party vote with the capacity of transmission. In other words, much of the incumbency advantage previously evident is lost; it is not transferred to the party's subsequent open seat candidate, consistent with the personal vote literature (for example, see Cain, FereJohn and Fiorina, 1987; also, Fenno, 1978 and Johannes and McAdams, 1981 for a complete discussion.)
Volatility Of District
In the analysis, competition was ascertained by determining whether the seat changed hands in previous elections. The determination of competition is important because spending may matter more when a candidate suffers from the electoral disadvantage of being from a minority party. One would expect that the dynamics of more competitive districts to be distinct from those in less competitive districts because the outcome may be affected by alternative influences. For example, party competition may


73
be greater or parties may matter little. The primary electoral force may be the candidates themselves. If one party has a habitual electoral advantage in the district, other factors may have little influence; it may take different strategies, qualifications and tactics for a minority party candidate to claim a victory.
A dummy variable was used to indicate volatility and party competitiveness. If the district had changed hands at least once in three prior elections it was coded as being competitive, indicating that it is possible for a candidate from either party to be elected in the district. Otherwise it was deemed non-competitive. The results are depicted in Table 3-7. The variable fails to meet statistical significance. It is concluded that the competitiveness of the district, as evidenced by shifts in party control, has little predictive power.
Conclusion and Discussion
National politics do have a significant and direct impact on elections for the U. S. House of Representatives; local factors seem to matter much less. This is not to imply that local circumstances are insignificant, but their significance is difficult to detect in an aggregate analysis. Individual case studies might be beneficial to examine the impact of local factors. The dynamics of open seat elections are in flux; in presidential election years, the proportion


74
Dependent
Table 3-6 Incumbency Loss Potential Variable: Percent of Votes Received by Democratic
Candidate in Open Seat Elections for the U.S. House from
1982-1992
Variables B_ T Siq T
Model A Proportion of 25.4 7.1 .0001
(Mid-term Money
Elections) Candidate Quality 1.4 2.3 .02
Index Democrat
N = 102 Candidate Quality -1.3 2.4 .02
Index Rep.
Percent Dem. .33 4.19 .0001
rlcS Incumbent Party -1.2 .6 .57
Last Percent Vote .1 1.8 .08
Dem
Model B Proportion of "l7 .0001
(Presidential Money
Elections) Candidate Quality .7 1.5 .15
Index Democrat
N = 49 Candidate Quality -1.3 2.6 .0001
Index Rep.
Percent Dem. T"} v* /-"\ c+ .6 9.5 .01
ries Incumbent Party -.4 .2 .84
Last Percent Vote 1 2 .01
Dem
Model C Proportion of 20.1 8.6 .0001
(Pooled Money
Data) Candidate Quality 1.1 2.8 .005
Index Democrat
Candidate Quality -1.1 3 .003
N = 151 Index Rep.
Percent Dem. .5 10 .0001
rlcS Incumbent Party -.7 .52 .63
Last Percent Vote 1 .7 .009
Dem
Adjusted R-square Model A = Adjusted R-square Model B =
.7 .8
Standard Error
6.4
Adjusted R-square Model C = .7
Standard Error = 6.3 Standard Error = 6.5


75
Table 3-7 Volatility of District
Dependent Variable: Percent of Votes Received by Democratic
Candidate Open Seat Elections From 1982 1992
Variables B T Siq T
Proportion of Money 32 9 .0001
Candidate Quality 1.3 1.9 .06
Index Democrat
Candidate Quality -1.8 2.7 .008
Index Rep.
Volatility of -1 .4 .4
District
N = 251
= .5
Adjusted R-square Standard Error = 8.1
of money spent by the Democratic candidate is less important than other factors; in mid-term elections, it is the strongest predictor of the vote. A significant amount of insight is available by analyzing the elections together as well as by conducting separate analyses of presidential elections and mid-term elections. The dynamics of these are significantly different and enlightening.
The potential for accountability is present, evidenced by the direct effect of economic policies on the vote and through the direct effect of presidential coat-tails. The impact of campaign expenditures in House races in presidential election years is a less important predictor of the vote than is the district's vote for the presidential candidate (Table 3-2). In mid-term elections, the impact of the vote for the president in the past election is still a


76
significant predictor of the vote, but the proportion of money spent has a greater impact.
Table 3-8: Final Model Dependent Variable: Percent of Votes Received by Democratic
Candidate in Open Seat Elections From 1982 -1992
Model A (Mid-term Elections) N = 121 Variables_ Proportion Money Candidate Quality Index Democrat Candidate Quality Index Rep. Percent Dem Pres. B 26.4 1.2 -1.3 4 1 8.7 2.2 2.6 6.3 Siq T | .0001 .02 .01 .0001
Model B Proportion Money 15.9 5.3 .0001
(Presidential Candidate Quality .9 1.9 .06
Elections) Index Democrat
N = 130 Candidate Quality -1.4 3 .004
Index Rep.
Percent Dem.Pres. .6 10.7 .0001
Model C (Pooled Data) N = 251 Proportion Money Candidate Quality Index Democrat Candidate Quality Index Rep. Percent Dem.Pres. 14.9 1 -1.3 .5 6.9 2.9 4.1 13.1 .0001 .005 .0001 .0001
Adjusted R-square Model A = .7 Standard Errc Dr = 6.3
Adjusted R-square Model B = .8 Standard Errc :>r = 6.4
Adjusted R-square Model C = .8 Standard Errc :>r = 6.1 J
Many politicians are si :ral begic. They base their
decision on the best time to s >ee> : office based on the
national tides. If one party is advantaged, more qualified
and viable candidates from that party will seek election.
Conversely, poor national times will yield less qualified and viable candidates. The strategic decision-making and self-


77
selection indirectly influences national events. The more qualified and viable a candidate is, the more likely he or she will achieve electoral success. Quality candidates make their strategic decision about when to seek political office based on expectations based on national and local political forecasts.
The electorate is directly exerting political accountability on political parties as well; be it via
presidential voting or mid-term penalty voting, partisan accountability is occurring. Money is still a significant factor in elections, in most cases the most significant variable, but its importance is not all-powerful. First, absolute resources matter less than relative resources. Other factors are able to influence the vote: factors such as candidate occupational background, presidential voting, presidential coat-tails, and strategic decisions about individual candidacies. If one wanted to further decrease the significance of campaign expenditures, the best means to do so would be to maximize competition and further increase the relative role of these additional factors. When their significance increases, the relative significance of expenditures will decrease.


CHAPTER 4:
THE ROLE OF GENDER IN OPEN SEAT ELECTIONS FOR THE UNITED
STATES HOUSE OF REPRESENTATIVES FROM 1982 -1992
Introduction
A question which has been plaguing academics, practitioners and laymen is why are women under-represented in the House of Representatives. Although the numbers of women legislators have been increasing in recent years (Table 4-1,) their total percentage is far below that seen in the general population. Women are also disproportionately under-represented in the ranks of candidates. While the percent of
female candidates increased throughout the time under study (a trend going back to the mid-1970's,) in 1992 (the year
with the highest percent of female candidates,) they still only encompassed eighteen percent of total candidates running for open seats (Table 4-1). That fact is especially troubling because open seat elections are seen as the most likely to be winnable; in these elections, under-represented groups would have the greatest potential to increase their numbers since they do not face entrenched incumbents. With such low numbers of candidacies, it is little wonder that women are so severely under-represented in Congress. The research analysis here will first examine female candidates in open seat general elections, then expand the scope of
78


79
analysis to include the primary nomination process to determine if biases exist in either the primary or general election process in the United States.
If bias does exist, questions regarding fundamental conceptions of democracy and representation are raised. The concept of representation is a complex one; many disagree over its basic nature. One form of representation, "virtual" representation, requires that the legislative body be selected in a manner so its composition mirrors that of the general society; only then it is really a representative body. John Adams argued that the legislature "should be an exact portrait, in miniature, of the people at large, as it should think, feel, reason and act like them" (Pitkin, p. 60). According to this line of reason, representative government must be "an accurate reflection" of the community, or of the general opinion of the nation, or of the variety of interests in society. The representative does not act for others, he or she "stands for" them; what is important is less what the legislature does than how it is composed (Pitkin, p. 61). One meaning of 'represent' is to be representative in the sense of having representative (typical) characteristics. Seen this way, elections "appear to be a method of finding persons who possess this representative quality" (Pitkin, p. 76). Descriptive representation mandates that a person stands for another by being "sufficiently like them" (p. 80).


80
Questions regarding the proper role of representatives have plagued theorists and laymen since the inception of democracy. The essential question revolves around the best means for a legislator to represent his or her constituency. Should the representative act as a free agent, otherwise known as a trustee, or is the representative bound as a delegate to mirror the views of those who are represented? The fundamental question is what do voters expect of their representatives? If they are expected to act as delegates, little differences exist regarding who actually serves. In other words, ideally if the representative is to "stand for" and act as an agent of the constituency, then it matters little if the representative is "Legislator A" or "Legislator B". But, if we wish our representative to serve in the trustee mold, it matters a great deal whether the representative is "A" or "B" since the individual is ultimately basing the voting calculus on personal opinions and views. If the latter point of view is taken, as it appears to be by legislators themselves (see Keefe and Ogul, 1993, p. 71 72), then personal characteristics of the legislator are of consequence.
Such characteristic statements are not to imply that
only women can represent issues important to women. However, a legislative composition reflective of the general society gives greater assurance that a variety of voices are heard as well as to assist in acceptance of enacted legislation. The symbolic nature of representation is an important facet.


81
Individuals need to feel that their interests are being represented; having "others like me" in the legislature is a good means to accomplish allow citizen affiliation to the legislature. If election rules are biased against individuals because of genetic characteristics, important issues are raised; most significant among them the fundamental nature of representation and the legitimacy of the legislative body and their subsequent actions.
Table 4-1
Number of Women in House of Representatives
1982-1992
Year 1982 1984 1986 1988 1990 1992
Total Number of Women In House 22 (4.7%) 22 (4.7%) 23 (4.9%) 25 (5.4%) 29 (6.2%) 48 (10.3%)
Total Number of Female Candidates for Open Seat Elections 13/110 (12%) 3/50 (6%) 9/84 (11%) 4/50 (8%) 8/58 (14%) 31/168 (18%)
Another facet underlying the significance of this inquiry is the emphasis placed on individual characteristics of candidates in elections in the United States. One important development driving the higher level of importance placed on individual characteristics of the candidates is the increased attention given on local television news coverage. For congressional elections, local news coverage increases have been accompanied by increased levels of advertising by candidates. Another factor leading to the increased importance placed on individual candidate characteristics is


82
the decreased importance of political parties, both in terms of voter identification and as a vote cue (Darcy and Schramm, p.2; Bond, Covington and Fleisher, 1985; Cain, Ferejohn and Fiorina, 1987; Krasno and Green, 1988).
First, the inquiry will examine relevant literature then examine the potential direct and indirect effects of gender on the percent of votes received by the candidate. Differences in political resources, campaign finance and candidate quality ratings, by gender of the candidates, political parties and the existence of patterns through time will be analyzed.
Potential Advantages Accorded to Female Candidates
Several prominent lines of reasoning exist which consider the possibility of female candidates benefiting from their gender. The first could be benefits incurred by a higher level of public recognition because of increased amounts of attention placed on female candidacies. Stokes and Miller (1962) and Tolchin and Tolchin (1973) contend that women candidates have an easier time gaining public recognition than men; they attract more attention and are remembered more easily than male candidates. Since name recognition is positively associated with the vote (Converse, 1962,) female candidates have an inherent electoral advantage. Darcy and Schramm (1977,) however, find little evidence to support the contention. They discovered no significant difference in the name recognition of male and


83
female incumbents; in fact women candidates gained no more name recognition than men candidates and women in the electorate did not react to women candidates differently than men did (p. 6).
A second potential advantage female candidates may possess is their ability to entice voters' sense of fair play; since they are an under-represented group, appeals can be made to elect them to diminish their group's under-represented position in the legislature. Such appeals are possible because Americans view Congress as representative of preponderant social groupings (Darcy and Schramm, 1977, p. 2). A third way female candidates may benefit is by their ability to politicize inactive females who are typically unstimulated by politics. Women (especially homemakers) are traditionally less interested and active politically, which may be due to the absence of female political leaders "able to articulate a woman's political viewpoint. Candidates of their own sex might increase the "stake" of women in the election, activate their interest and mobilize their vote" (Darcy and Schramm, 1977, p. 2-3). Finally, voters disenchanted with the established power structure may support females since they are often viewed as outsiders because their group is not traditionally viewed as part of the power structure.
Darcy and Schramm found no evidence to sustain these notions; there is no support of the claim that "women candidates act to politicize a politically passive female


84
population. Voting turnout of women in races with female candidates was not significantly greater than turnout in races with no women" (p. 6). As a matter of fact, Darcy and Schramm conclude that the "evidence indicates that the electorate is indifferent to the sex of congressional candidates" (p. 10).
Potential Disadvantages
Sexism
Some contend that female candidates may be at a disadvantage at the polls because of voter discrimination (Deber, 1982). A number of studies have examined the potential of a sexist voter bias. The existence of sexism in electoral politics has been tested using experimental data, survey results and analysis of election returns from a variety of electoral contests (local, state and national); overall, no clear advantage exists to candidates of either sex (Uhlaner and Schlozmann, 1986, p. 32).
Sexism can have an indirect impact via prejudice by political influentials. Considerable evidence exists that indicates that political influentials often actively discourageor at least fail to encouragewomen to run for elected office unless a race appears hopeless (Uhlaner and Schlozmann, 1986, p. 33). Women may be under-represented in Congress because they are over-represented in races that appear "hopeless", races in which their poor showing was


85
inevitable (Deber, 1982). Women are far more likely to face incumbents than to run for open seats. Once incumbency is controlled, women do as well as men in general election races (Bernstein, 1986). Bernstein states the reason the increase in the number of women nominated to run for Congress has not been followed by an increase in the number of seats held by women is because women are nominated as challengers to incumbents rather than in more electoral competitive races
(p. 155).
Political Resources
Structural and institutional explanations for the inequality of political resources have also been put forward. According to this line of thought, women are disadvantaged by structural biases in the political system; women do not have the advantages which are traditionally rewarded at the polls both with regards to candidate quality and campaign finance resources. Sexism on behalf of the voters or political influentials may be less important in explaining the under-representation of women in the House than the fact that females often lack the political resources which produce electoral success. Inequalities in political resources and situational responsibilities (for example, motherhood) potentially explain the under-representation of females in legislative bodies.


86
Occupational Bias
Socialization theory explains the under-representation of women in legislative bodies as a manifestation of the acceptance of traditional gender roles of women. Socialization theory contends that women learn early in life that politics is a man's game and, therefore they do not aspire to public office (Constantini and Craik, 1977). Women's socialization, career choices and domestic responsibilities delay entrance into political world or may completely deter it (Uhlaner and Schlozman, 1986, p.30). Politics is not seen as a traditional occupational role. Although many inroads have been made into breaking down the barriers imposed by traditional thought, many still believe that certain occupations are best "left to the men". If a woman runs for Congress, she may be seen as un-feminine and aggressive. It has also been held that occupations that women enter typically are not stepping stones for political office. Evidence exists that women lack the professional credentials necessary to be viewed as electorally advantaged (Uhlaner and Schlozman, 1986, p. 35).
Bernstein contends (in an analysis of races from 1964-1980) that the reason women are under-represented in Congress is because they are less politically ambitious than men. According to him, there is an "increasing number" of ambitious (by ambitious he means one who is driven by a desire for personal advancement) male opposition entering


87
open seat races. These ambitious men are utility maximizers; they see a greater opportunity to win in open seat elections,
so they are over-represented in these types of contests. He states that women are less ambitious than men because they delay their careers until after child-rearing years and possibly because of biological reasons1. "Contests for open-seat nominations have increasingly pitted 'young men in a hurry' against women without that kind of drive. Consequently, women have become less competitive in open-seat primaries" (Bernstein, 1986, p. 158).
There are several serious flaws in Bernstein's research. First, he uses age as a proxy for political ambition, severely biasing his results, subsequently allowing him to find what he is looking for. There is a link between age and electoral victory (p. 160), but this link is weak, especially when one fails to incorporate other intervening variables. Just because an individual is older, he or she is not necessarily less ambitious. Such a characterization is especially true regarding female candidates. Many women delay their entrance into politics because they often begin their families prior to entering the workforce. Bernstein also fails to control for other variables (such as strength of the party in the district or partisanship.) He concludes that men are more competitive in open seat races without controlling for anything except age. Problems also exist
AThis does not appear to be a serious reason, but it is briefly put forward.


88
with his data base. He has data for only forty-nine percent of the cases from one half of his sample (races in the 1960's). Although Bernstein acknowledges the limits of the data in his analysis, the results are still drawn into question.
When one examines the prior election experiences of females and males in open seat races for the House from 1982 through 1992, a very different picture is drawn. As evidenced in Table 4-2, the candidate quality ratings of male and female candidates are similar. It was a bit premature for Bernstein to state that females in open seat races are less occupationally (operationalized as electoral background) competitive than males.
Table 4-2
Percent of All Male and All Female Candidates In Open Seat Elections From 1982 1992 With Prior Elected Experience
Male Female
Proportion (and Percent)with prior elected experience 254/451 (56%) 36/69 (52%)
Carroll also challenges these and other results which contend that women are less politically ambitious than men. An "ambition gap" hypothesis is put forth as a possible explanation for the under-representation of women in Congress. An "ambition gap" may be the reason women are not occupying higher offices even though women are increasingly occupying lower offices. Such an occurrence is especially important since office holding is usually progressive


89
(Carroll, 1985,p. 1231). Prior research did find a gap in the level of ambition, but Carroll attributes the finding to problematic sampling techniques. Most research sampled delegates to national party conventions and other party activists rather than samples of male and female public office-holders; when the latter are sampled few differences in political ambition are evidenced. "Women and men serving in comparable elective offices in 1981 were very similar in their political ambitions" (p. 1235). She goes on to say that "...the findings presented here suggest that the ambitions of women office-holders at state legislative, county and local levels are not at fault. Rather, political scientists must begin to look not to political women's ambitions, but to possible patterns of discrimination and limitations in the structure of political opportunity in order to account for the paucity of women at the highest levels of office holdings" (p. 1242).
Campaign Finance
One such area of possible discrimination and limitation is campaign finance. In 1980, women candidates for the House received on an average of $25,000 less than men candidates (Uhlaner and Schlozman, 1986, p. 35). The authors put forth two possible explanations for this. First, discrimination could be occurring against female candidates, or second, women candidates might possess less of the characteristics that contributors reward (for example, incumbency and


90
chairmanships.) In 1980, female nominees were concentrated disproportionately in candidacies in which the chances of winning were lowest (challengers in districts safe for the incumbent). A much smaller proportion of women than men were incumbents; a significantly larger proportion were challengers. Only six of the eighty-four candidates for open seats were women. Five of six were from the "out" party; the pattern holds true for 1976 1978. The situation seems to be improving; from 1982-1992, 56% of all female candidates in open seat elections were from the "out" party (Table 4-3). The prospects appear to be better for Democratic women; in fact, their chances of being a member of the "out" party significantly decreased in 1992. Overall, in 1992, 46% of all female candidates were from the "out" party.
Table 4-3
Proportion of Women in Open Seat Elections from the "Out"
Party
The percent of female candidates with prior elected experience is also increasing. In 1980 more women non-incumbents had prior elected experience than their male counterparts (38% of women versus 24% of men) (Uhlaner and


91
Schlozman, 1986). Women non-incumbents were also slightly more likely to have previously run for Congress (16% versus 13%). Uhlaner and Schlozman's research provided mixed results; "women candidates are placed at a distinct disadvantage with respect to fund-raising by their disproportionate status as non-incumbents. However, women candidates command a disproportionate share of other attributes that are associated with higher receipts" (Uhlaner and Schlozman, 1986, p. 36). They concluded that "although women candidates for Congress in 1980 had, on average, smaller war chests than male candidates, they suffered no direct gender-based disadvantage with respect to campaign receipts, receipts relative to those of their opponents or contributions in large sums" (p. 46).
Table 4-4 indicates that females are not victims of discrimination with regard to campaign finance. As a matter of fact, both Democratic and Republican female candidates spend, on average, more than their male counterparts. The fact that the standard deviations are larger for female than male candidates indicates that the results may be slightly misleading since they may be victim of skewed distributions of campaign expenditures (for example a few candidates which spend a considerable amount of money could be skewing the results). Such a possibility will be explored in far more depth in the subsequent analysis. At this point in the research there is little evidence of female candidates suffering from electoral disadvantage.