Minority Representation and the Quest for Voting Equality

MINORITY REPRESENTATION AND THE QUEST FOR VOTING EQUALITY

Minority Representation and the Quest for Voting Equality BERNARD GROFMAN School of Social Sciences University of California, Irvine LISA HANDLEY Election Data Services Washington, DC RICHARD G. NIEMI Department of Political Science University of Rochester

CAMBRIDGE UNIVERSITY PRESS

Published by the Press Syndicate of the University of Cambridge The Pitt Building, Trumpington Street, Cambridge CB2 1RP 40 West 20th Street, New York, NY 10011-4211, USA 10 Stamford Road, Oakleigh, Victoria 3166, Australia © Cambridge University Press 1992 First published 1992 First paperback edition 1994 Library of Congress Cataloging-in-Publication Data Grofman, Bernard. Minority representation and the quest for voting equality / Bernard Grofman, Lisa Handley, Richard G. Niemi. p. cm. Includes index. 1. Afro-Americans - Suffrage. 2. Election districts - United states. 3. Representative government and representation - United States. I. Handley, Lisa. II. Niemi, Richard G. HI. Title. KF4893.G76 1992 324.6'2l08996073-dc20 92 -12041

cn> ISBN 0-521-39128-8 hardback ISBN 0-521-47764-6 paperback Transferred to digital printing 2003

Contents

Acknowledgments List of Tables Introduction 1. The Right to Vote and the Right to Representation The Two-Hundred-Year Struggle for Minority Voting Rights The 1965 Voting Rights Act Vote Dilution: The Right to Vote Versus the Right to Representation 2. The Evolution of a Vote Dilution Standard Through 1986 Section 5 Protection Against Vote Dilution Constitutional Protection Against Vote Dilution Congress Dictates a Vote Dilution Standard: The 1982 Amendments to the Voting Rights Act The Lower Courts Consider Amended Section 2 The Supreme Court Addresses Section 2: Thornburg v. Gingles 3. The Vote Dilution Standard in the Post-Gingles Era: Clarifications and Complications in the Lower Courts What Constitutes a "Sufficiently Large and Geographically Compact" Minority Group? When Is a Minority Group "Politically Cohesive"? How Does One Determine When the White Majority Votes Sufficiently as a Bloc to Enable It . . . Usually to Defeat the Minority's Preferred Candidate? 4. Defining and Measuring Racially Polarized Voting and Other Elements of the Totality of the Circumstances Defining Racial Bloc Voting Measuring Individual Voting Behavior

page vii viii 1 4 4 15 23 29 30 31 38 42 47 61 62 67 73 82 83 84

vi

Contents Commonsense Checks on Estimates of Bloc Voting Problems of Data Availability Special Issues Problems in Interpreting Racial Polarization How Reliable Are Estimates of Racial Bloc Voting? Other Elements of the Totality of the Circumstances

5. Vote Dilution in Single-Member Districts and Other Issues of the 1990s Considerations of Size Identifying Racial Gerrymanders Influence Districts An Equal Opportunity to Elect Candidates of Choice Runoff Elections Standards Other Than the Single-Member District

88 93 96 98 103 105 109 110 113 117 118 122 124

6. The Voting Rights Act and the Realistic Politics of Second Best: An Optimistic Look to the Future Legislative Intent and the Right to Vote Versus the Right to Representation A Color-blind Society Versus the Institutionalization of Racial/ Ethnic Divisions The Specter of Proportional Representation The Voting Rights Act and the Future of American Politics

131 132 134

Notes

138

References

154

Index of cases

163

Index

165

129 129

Acknowledgments

This book grows out of our concerns as citizens about the representation of minority groups in public office and out of our concerns as political scientists about the manner and degree of political representation in the United States. It draws on our involvement in court cases relating to the Voting Rights Act as well as on our broader and more theoretical writings about representation generally. We are indebted to many who helped along the way: to attorneys and other expert witnesses too numerous to name, to coauthors (especially Mike Migalski, Nick Noviello, and Kim Brace), and to readers of the manuscript (especially Chandler Davidson and Robert Berman), and to Harold Stanley for helpful conversations. We have learned much from them; however, we sometimes rejected their advice, so any blame for errors or misinterpretations must rest squarely on our shoulders. We also are happy to acknowledge the bibliographic assistance of Dorothy Gormick. Chapters 1 through 3 draw heavily on Lisa Handley's doctoral dissertation. Chapter 4 and parts of Chapter 5 draw on Brace, Grofman, Handley, and Niemi (1988); Grofman (1991b, 1992); Grofman and Handley (1989, 1991, 1992); Grofman and Migalski (1988); and Grofman, Migalski, and Noviello (1985). Some of this research was supported by National Science Foundation Grant SES #88-09392 to Chandler Davidson and Bernard Grofman and by NSF Grant SES #81-07754 to Grofman.

vn

List of Tables

Table 1. Registration by race, before and after 1965, in southern states covered by the Voting Rights Act (percent)

23

Table 2. Registration by race since 1970 in southern states covered by the Voting Rights Act (percent)

24

Table 3. Turnout in the 1964 and 1968 presidential elections in southern states covered by the Voting Rights Act (percent)

25

Table 4. Turnout of blacks and whites in presidential elections since 1964 for the United States and the South (percent)

26

vin

Introduction

Even as late as 1960, voter registration rates among whites in the South were more than twice those of blacks, and in Mississippi it is estimated that less than 5 percent of black adults were registered to vote. In the three decades since then, legal barriers to minority electoral participation have largely fallen, and voting rates of blacks have risen substantially. Yet representation of minorities remains problematic, with black and Hispanic officeholding at all levels of government only a fraction of their percentages in the population.1 There are two very different explanations for this situation, leading to diametrically opposed policy recommendations. Minorities themselves, and many civil rights attorneys, see the principal barriers to greater representation as electoral arrangements (including at-large elections and gerrymandered districts) and the continued reluctance of white/Anglo voters to vote for minority candidates. They attribute much of the recent success of minorities to voting rights legislation and to ceaseless litigation to protect minority voting rights and end the use of dilutive voting systems (e.g., Parker, 1990). They sometimes call for nongeographically based interpretations of existing election laws (e.g., Karlan, 1989) and a fundamental restructuring of the mechanisms of representation (Guinier, 1991). On the other hand, a number of social scientists, and some white and minority politicians, point to the failure of minorities themselves and of politicians generally to seek aggressively the cross-racial coalitions necessary for minority success. They emphasize socioeconomic differences that might account for disparities in political power as well as the liabilities inherent in "affirmative action in the electoral sphere" (Thernstrom, 1987, p. 242). They also raise the specter of proportional representation and the prospect of an institutionalization of present racial divisiveness (e.g., Butler, 1982; Graham, 1992; O'Rourke, 1992; Thernstrom, 1985, 1987). Accordingly, they advocate a less expansive interpretation of voting rights legislation. What can be said with certainty is that the subject of minority representation will remain on the forefront of electoral politics for at least the next decade. During the 1990s round of districting, most redistricting litigation will center 1

2

Minority Representation and Voting Equality

on issues of racial fairness, perhaps in combination with new concerns such as the compactness of minority and majority districts. We see no reversal of the pattern of the 1980s, in which there were far more cases involving racial and linguistic minorities than cases dealing with one person, one vote, or other issues. Moreover, as we work our way through the 1990s, an increasing number of cases involving Hispanics, Asian-Americans, and Native Americans will be added to cases concerning the voting rights of African-Americans. A major concern in this book is the role of social scientists and social science testimony in voting rights cases. As we look to the voting rights litigation of the 1990s, we can anticipate that social science testimony will continue to play a critical role, as it did in the 1980s. Whether establishing the elements of the Gingles test or the various factors of the "totality-of-circumstances" test, the testimony of social scientists will continue to be indispensable. One purpose of this book, therefore, is to explain the methods that have come to be accepted by most courts as the appropriate tool for determining major aspects of minority vote dilution, especially with respect to the determination of racial bloc voting. Our aim in this book goes beyond explicating the technical issues of social science testimony in voting rights litigation. We describe in considerable detail the evolution of voting rights case law, focusing on the confusion and conflict in the federal district and circuit courts in the years since the Supreme Court's 1986 decision in Thornburg v. Gingles.2 We also discuss the issues that are just beginning to confront the courts (and social scientists) as attention turns to alleged vote dilution in the single-member district context. And in our concluding chapter we consider the implications of developments in voting rights case law and the outcomes of voting rights cases for the future of representation in America. In Chapter 1 we provide an overview of the struggle for minority voting rights and the history of the Voting Rights Act of 1965 and its subsequent amendments and extensions through 1975. We also distinguish between the right to vote and the right to representation and sketch alternative mechanisms and measures of vote dilution. In Chapter 2 we carry forward the evolution of a vote dilution standard as it developed in the courts in the 1960s and 1970s - in Congress with the adoption of the 1982 amendments to the Voting Rights Act and again in the courts through the Gingles decision. We also discuss the ambiguity surrounding the totality-of-circumstances test in the wake of Gingles. In Chapter 3 we examine the operationalization of the Gingles three-pronged test that is at the heart of the current Section 2 case law. This requires a review of recent, sometimes contradictory decisions of the federal courts, as well as a discussion of some of the issues addressed in these cases.3 Chapter 4 is largely technical, and readers more interested in the law per se

Introduction

3

may wish to skip the most detailed parts. The entire chapter, however, is indispensable reading for social scientists who might consider serving as an expert witness in a voting rights case. The majority of the chapter is devoted to defining and measuring racially polarized voting, but we also discuss other elements of the totality-of-circumstances test. In Chapter 5 we look at how the concept of vote dilution might be defined and measured in single-member district plans as well as at other issues that are likely to dominate voting rights litigation in the 1990s. This covers such current topics as influence districts, runoffs, and population growth, as well as more "radical" alternatives to the single-member district system. In the concluding chapter we take a broader look at representation in the United States and suggest that we are dealing with the "politics of the second best." Nonetheless, in looking at the direction in which the Voting Rights Act is taking us, we conclude on an optimistic note about the future of minority representation and of American politics.

The right to vote and the right to representation

Our book is chiefly about the post-1965 era and about representation, as opposed to registration or turnout. However, the rationale for, as well as the nature of, recent and current efforts to protect minority voting rights cannot be understood without an awareness of the history of minority disfranchisement, especially black disfranchisement in the South. The Voting Rights Act of 1965 and the long line of litigation that forms its lineage can be understood only in the context of the failure of the previous hundred years to establish a secure basis for the right to vote. Thus we begin with a brief overview of that troubling history. We then describe the passage of the act itself and its initial extensions and expansion. Finally, we distinguish between voting and representation and raise a number of questions about the current and future operationalization of the concept of nondilution of minority votes. This distinction will serve as a backdrop for the remainder of the book. The two-hundred-year struggle for minority voting rights Before the Civil War, black men could vote in only six northern states; black women, of course, were denied the franchise in all states.1 After the war, the right to vote was extended to most adult males in the seceding states. Southern states were required by the Military Reconstruction Act of 1867 to adopt new constitutions granting universal male suffrage, regardless of race, as a condition for readmission to the Union. Congress was to enshrine the promise of the franchise for blacks in a constitutional amendment three years later. Following the Civil War, three constitutional amendments designed to eradicate the remaining vestiges of slavery in the South were passed, beginning with the Thirteenth Amendment (1865), which guaranteed freedom from slavery. The Fourteenth Amendment, considered by Congress in 1866 and adopted in 1868, granted citizenship to "all persons born or naturalized in the United States" along with its concomitant privileges and immunities, due process of law, and equal protection of the laws. This amendment dealt with the franchise only

The right to vote and the right to representation

5

indirectly, by providing in Section 2 that if a state barred any adult males from the vote, the state's representation in Congress "shall be reduced in the proportion which the number of such male citizens shall bear to the whole number of male citizens twenty-one years of age in such State." This provision was adopted in order to induce states to repeal state laws excluding blacks from the right of suffrage, but it was in fact never utilized. Enfranchisement of black citizens was granted nationally with the ratification of the Fifteenth Amendment in 1870. This amendment provided that "the right of citizens of the United States to vote shall not be denied or abridged by the United States or by any State on account of race, color, or previous condition of servitude." Section 2 (like Section 2 of the Thirteenth Amendment and Section 5 of the Fourteenth Amendment) gave Congress the right to enforce the amendment through appropriate legislation. Shortly after the Fifteenth Amendment became effective on March 30, 1870, Congress exercised its authority to pass appropriate legislation by enacting two measures to protect black suffrage: the Enforcement Act in 1870 and the Force Act of 1871. The Enforcement Act of 1870 "provided for criminal sanctions against those who interfered with the constitutionally guaranteed right to vote," and the Force Act of 1871 "was to supplement the 1870 statute by supplying independent enforcement machinery . . . to be provided by the appointment of federal officials to supervise the election process in each election district" (Schwartz, 1970, p. 547). These two acts prohibited state officials from applying election laws in a racially discriminatory manner, outlawed physical threats and economic intimidation against voters, and made it illegal for individuals to conspire to interfere with a person's right to cast a ballot. To implement the Force Act, Congress authorized the appointment of supervisors to observe registration and election proceedings in large towns and cities (Kousser, 1984, p. 28). Before ratification of the Fifteenth Amendment, the military leaders assigned to replace the Confederate governments had already launched a campaign to register blacks to vote.2 By 1868, more than 700,000 blacks had been registered to vote under the supervision of federal troops (Franklin, 1961, p. 80). Blacks not only participated in politics by voting; they also held a variety of public offices in each of the southern states and even occupied a majority of the seats in the lower house in South Carolina. Although no black ever served as governor of a southern state, three states elected black lieutenant governors (South Carolina, Mississippi, and Louisiana), and blacks were chosen to fill a number of other statewide offices as well. Furthermore, between 1869 and 1901, the South sent twenty blacks to the U.S. House of Representatives and two blacks to the U.S. Senate (Franklin, 1967, pp. 319-320). In the North, whites came to accept black voting, but in the South, from the outset, whites resisted enforcement of the Fifteenth Amendment. Violence, in-

6

Minority Representation and Voting Equality

timidation, and fraud were the primary means used initially to discourage blacks from voting. According to information gathered by a House subcommittee in 1868: Over 2,000 persons were killed, wounded and otherwise injured in [Louisiana] within a few weeks prior to the presidential election; half the State was overrun by violence; midnight raids, secret murders, and open riot kept the people in constant terror until the Republicans surrendered all claims, and then the election was carried by the democracy. (U.S. Commission on Civil Rights, 1968, pp. 3-4) Blacks who managed to overcome these obstacles and cast a ballot were faced with elections fraught with corruption: Candidates' names were scratched off the ballots; votes were stolen from the boxes; polls were not open at all in heavily Republican areas; and police were stationed at polling places to allow only those favoring the Democratic party to enter (U.S. Commission on Civil Rights, 1968, p. 4). By using this array of techniques, whites in the South were able to oust radical Republican administrations and replace them-with white supremacist Democratic regimes, thus "redeeming" the South. The formal end of Reconstruction, however, was marked by the Compromise of 1877, in which Southern Democrats helped resolve the corruption-plagued presidential election of 1876 by throwing their votes to the Republican candidate Rutherford B. Hayes in return for the removal of the remaining military troops from southern soil and an understanding that white southerners were to manage their affairs without as much supervision as in the past. Left to its own devices, the South showed no inclination to protect the right of blacks to vote and, in fact, began to institute a series of measures designed to prevent blacks from casting ballots. The prospect of active federal enforcement of the Fifteenth Amendment also ended at approximately this time with two Supreme Court decisions that greatly weakened the effectiveness of the Enforcement and Force acts. The Court, in two separate cases decided on the same day (March 27, 1876) - United States v. Reese and United States v. Cruikshank - threw out indictments based on various provisions of the Enforcement Act of 1870 and the Force Act of 1871, thereby nullifying critical sections of the two acts. The Court declared that in order to secure convictions under these acts, it must be proved that the accused operated under the authority of state law and intended to discriminate for reasons of race. In Reese, the Court overturned the indictments of two election officials who had refused to receive the vote of a black citizen in Kentucky and were charged with violating two sections of the Enforcement Act. The Court held that the provisions on which the indictments were based (Sections 3 and 4 of the act) were too broad, as they were not confined in their operation to

The right to vote and the right to representation

1

unlawful discrimination on the basis of race alone and therefore exceeded congressional power under the Fifteenth Amendment. In Cruikshank, eight men who had participated in a mob that had massacred sixty blacks in 1873 in Coifax, Louisiana, appealed convictions under another section of the Enforcement Act (Section 6), which held that conspiring to hinder citizens in the enjoyment of rights or privileges guaranteed by the federal Constitution was illegal. The Supreme Court held that each of the rights at which the conspiracy was aimed was in fact not a federal right but was, rather, a right derived from the states and therefore not subject to federal protection. The Court held that the only voting rights that Congress had the authority to protect were the right to vote in a federal election and the right to vote free of racial discrimination. Because denial of neither of these rights was alleged, the indictment was held not to state an offense. "We may suspect that race was the cause of the hostility; but it is not so averred," according to Chief Justice Morrison R. Waite (p. 556). Cruikshank was released from prison, and although the Court did not actually strike down Section 6 of the Enforcement Act, it was construed so narrowly that it became virtually useless. Thus, on the first occasion on which the Supreme Court was asked to consider the acts designed to implement the Fourteenth and Fifteenth amendments, the Court succeeded in crippling the efforts of Congress to protect the right to vote against both official and private interference (Derfner, 1973, p. 529). Reese and Cruikshank paralleled a shift in national attitudes to support noninterference in the South and marked a general retreat from federal enforcement of the Fifteenth Amendment:3 The phase that began in 1877 was inaugurated by the withdrawal of federal troops from the South, the abandonment of the Negro as a ward of the nation, the giving up of the attempt to guarantee the freedman his civil and political equality, and the acquiescence of the rest of the country in the South's demand that the whole problem be left to the disposition of the dominant Southern white people. (Woodward, 1966, p. 6) At the same time, events in the South dramatically accelerated this regressive trend toward black disfranchisement. The Compromise of 1877 not only symbolized the end of Reconstruction; it also signaled the beginning of the movement to exclude blacks totally from the southern electorate. The newly redeemed governments of the South managed to reduce by half the number of black voters over the next fifteen years (Lawson, 1976, p. 6). State statutes passed in an effort to discourage black political participation included such legally sanctioned devices as long residency requirements and very short registration periods. Perhaps the most inventive technique was one adopted by South Carolina in 1882, the "Eight Box Law." This statute established eight categories of elections with separate ballot boxes for each category. Ballots were not to be counted if placed in the incorrect box, making it impos-

8

Minority Representation and Voting Equality

sible for illiterate blacks to cast ballots (election officials were instructed to aid white voters but not black voters in placing their ballots in the correct boxes). A less well known and more subtle means of minimizing black political participation without actually denying the franchise to blacks was to employ one or a number of "dilutive" techniques; the method(s) chosen depended on the proportion of blacks in the population and their geographic concentration. Such techniques included racial gerrymandering, at-large election systems, annexations or deannexations, or abolishing local elections for certain offices altogether and making such positions appointive. Each of these measures had as its intended purpose minimization of the number of black officeholders or white officeholders supported by blacks (see Kousser, 1984, pp. 31, 36). In addition to all of these measures, local white supremacists used intimidation and violence against blacks wishing to vote and outright fraud at the polls when blacks did manage to cast a ballot. Despite these efforts, some blacks continued to exercise the franchise and to hold public office, primarily through the operation of the ''fusion principle." Under this practice some whites (usually conservatives) would appoint blacks to a small share of minor public offices in return for black support at the polls. However, even this limited amount of black participation and influence was soon to be eliminated. Black disfranchisement solidified in southern state conventions Although most southern states had enacted statutes designed to reduce the number of blacks eligible to vote, it was not until the 1890s that states began to convene constitutional conventions to eliminate what was left of the black vote. Between 1890 and 1910, most southern states rewrote their constitutions in ways that were intended mainly to exclude blacks from the electorate without obviously violating the Fifteenth Amendment.4 Mississippi initiated the movement by convening a state constitutional convention in 1890. Of the franchise qualifications adopted to ensure that blacks would no longer exercise any political power, the "crowning achievement" (Key, 1949, p. 537) of the Mississippi Constitutional Convention was the "understanding" clause, which required a potential elector to read any section of the state constitution or provide a "reasonable" interpretation of any section read to him. Recognizing that a literacy test fairly administered would undoubtedly prevent large numbers of whites as well as blacks from voting, the "reasonable" interpretation clause gave white registrars enough discretion in evaluating applicants' performance on the test to pass most whites but reject most blacks. The Mississippi convention marked the beginning of the movement to dis-

The right to vote and the right to representation

9

franchise blacks in the South by means of a state constitutional convention. South Carolina followed suit with a convention in 1895, Louisiana in 1898, North Carolina in 1900, Alabama in 1901, Virginia in 1902, and Georgia in 1908. These constitutions outlined new qualifications for voting that included not only the "understanding" clause passed by Mississippi and other variants on the literacy test but also such measures as the "good character" test and the payment of a poll tax.5 Throughout the South, literacy tests were instituted; residency requirements were lengthened; property qualifications for registration were established; and the list of disfranchising crimes was expanded to include offenses believed to be committed more frequently by blacks.6 Registration and election officials were given broad discretion to determine whether or not a potential voter met the conditions for voting. The appointment of these officials was placed in the hands of state officials rather than local officials in order to ensure white control of the election process even in predominantly black areas. The Louisiana constitution adopted in 1898 contributed the "grandfather clause" to the list of disfranchising devices. The "grandfather clause" was not actually a disenfranchising technique but, rather, a method of exempting illiterate whites from having to pass the state-imposed literacy test. The Louisiana constitution provided that no male - or son or grandson of such male - who was entitled to vote on January 1, 1867, was to be denied the right to vote. "But if the Negroes did learn to read, or acquire sufficient property, and remember to pay the poll tax and to keep the receipt on file, they could even then be tripped by the final hurdle devised for them - the white primary" (Woodward, 1966, p. 84). Democratic party leaders across the South declared that only whites were eligible for membership or permitted a voice in the nomination of party candidates. Because the Democratic party was viewed as a private organization, and therefore perceived to be outside the purview of the federal constitution, it was believed that the party could legitimately discriminate if it so chose. As victory in the Democratic primary was tantamount to victory in the general election in the South, being denied access to the Democratic nominating process was the equivalent of disfranchisement. Every state in the South adopted the policy of white primaries, with the exception of Tennessee and selected counties in North Carolina (Lewinson, 1932, p. 153). The disfranchisement of blacks in the South occurred with the acquiescence, if not the explicit approval, of the federal government. In 1894, for example, Congress - by then under Democratic control - repealed significant portions of the Enforcement Act of 1870 and the Force Act of 1871. Congressional withdrawal from voting rights enforcement was followed four years later by a favorable review from the Supreme Court for the Mississippi state constitution suffrage provisions. In Williams v. Mississippi (1898), the Court was asked to consider the con-

10

Minority Representation and Voting Equality

stitutionality of some of the suffrage provisions of the Mississippi Constitution of 1890. Williams, who had been indicted for murder by an all-white grand jury, challenged his conviction on the grounds that the jury panel had been drawn from the voting rolls, from which blacks had been effectively excluded by the 1890 Mississippi Constitution. Rejecting the assertion that the various requirements (the "understanding" test, poll tax, lengthy residency requirements, etc.) discriminated against black citizens in violation of the due process clause of the Fourteenth Amendment, the Court concluded that "they do not on their face discriminate between the races, and it has not been shown that their actual administration was evil, only that evil was possible under them" (p. 225). Six years later, the next challenge to the suffrage provisions of a southern disenfranchising constitution met with the same response. In Giles v. Harris (1903) and Giles v. Teasley (1904), a black citizen attacked several suffrage provisions (such as the "fighting grandfather" clause and the "good character" clause) of the 1901 Alabama Constitution, contending that they violated the Fifteenth Amendment. The Supreme Court disposed of both of these cases on procedural grounds. These two decisions (as well as the earlier decision in 1898) no doubt provided those few southern states that had not yet acted to disfranchise blacks by means of the state constitution the incentive to imitate their fellow southern states. By the turn of the century, virtually all blacks had been disfranchised in the South.7 Abandoned by the federal government, thwarted by the Supreme Court, and faced with a multitude of state laws designed expressly for the purpose of disenfranchising them, the majority of blacks in the South would not be permitted to exercise the franchise until 1965. A gradual rekindling of the Fifteenth Amendment in the courts For an extended period of time, the Supreme Court did very little to alter the level of black disfranchisement. In fact, the white primary was declared legal as late as 1935 in Grovey v. Townsend; the poll tax was upheld in 1937 (Breedlove v. Suttles); and the literacy test was declared constitutional as recently as 1959 (Lassiter v. Northampton County Board of Elections). But the decades between 1900 and 1960 were not completely void of progress on the suffrage front. For instance, the ratification of the Nineteenth Amendment in 1920 enfranchised women, though black women in the South remained unable, for the most part, to exercise that right. The Supreme Court also began to carve out some exceptions to the doctrine of black disfranchisement that it had sustained in earlier decisions. In 1915 in Guinn v. United States, the Supreme Court declared Oklahoma's

The right to vote and the right to representation

11

"grandfather clause" unconstitutional. This 1910 amendment to the state constitution exempted persons who had voted before 1866 and their lineal descendants from taking the literacy test mandated by the state. The Court held that this statute constituted a transparent effort on the part of the state to limit the administration of the literacy test to blacks only, which was a violation of the Fifteenth Amendment. Although the provision did not overtly single out blacks for unequal treatment, the Court wrote that "the standard itself inherently brings that result into existence since it is based purely upon a period of time before the enactment of the Fifteenth Amendment" (pp. 364-365). The decision, though perhaps significant in a symbolic sense because it represented the first time the Court expressed a willingness to look beyond "nondiscriminatory form to discriminatory substance" (Elliott, 1974, p. 71), had little effect. The Oklahoma state legislature responded to Guinn by passing a decree in 1916 that established a once-in-a-lifetime period of twelve days for those who were previously not on the registration rolls to register to vote. This scheme was not struck down by the Court until 1939, when a black who had failed to register within the stipulated twelve-day period challenged the statute and won in Lane v. Wilson. In delivering the opinion of the Court, JusLce Felix Frankfurter wrote that the Fifteenth Amendment "nullifies sophisticated as well as simple-minded modes of discrimination. It hits onerous procedural requirements which effectively handicap exercise of the franchise by the colored race although the abstract right to vote may remain unrestricted as to race" (p. 275). The demise of the white primary constituted the Supreme Court's most important contribution to black enfranchisement during the first part of the twentieth century. But the collapse of the white primary came only after the Court considered a series of four Texas cases over a seventeen-year period (19271944).8 It was not until the fourth decision, Smith v. Allwright (1944), that the Court recognized the Democratic primary as an "integral part of the election process" and determined that any attempt to limit the primary to white voters was unconstitutional. Smith, the black plaintiff in Smith, sued Texas election officials for refusing to permit him to cast a ballot in the primary election of 1940 for Democratic congressional candidates. Smith lost his case in both the district court and the Fifth Circuit Court of Appeals, but the Supreme Court reversed and held that the exclusion of blacks from voting in a Democratic primary was a state action in violation of the Fifteenth Amendment.9 The Smith decision produced a variety of responses in the South: Some states, after expressing dissent "chiefly for the record," made no changes to offset the decision (Key, 1949, p. 626); other states, most notably South Carolina, sought to bypass the decision by divorcing the process of selecting party candidates

12

Minority Representation and Voting Equality

from governmental action, by erasing all the primary election statutes from the books.10 Attempts at circumventing the decision were voided by the courts, however, and the white primary eventually vanished. The demise of the white primary was not the end of black disfranchisement in the South. Although there was a rise in black political participation in many areas,11 prompting some scholars to refer to this period as "the initial political mobilization of Negroes in the twentieth-century South" (Garrow, 1978, p. 6), states that wished to continue excluding blacks from the electoral system simply fell back on literacy tests and other discretionary registration laws. Alabama, Louisiana, and Mississippi were particularly effective in utilizing alternative methods to restrict black registration. Black registration rates three years after the Smith decision were still below 3 percent in each of these three states, compared with an average of over 17 percent in the other eight states of the Old Confederacy (Garrow, 1978, p. 7). The civil rights acts of 1957, 1960, and 1964 Reforms in black enfranchisement were not confined solely to the results of court decisions during the first six decades of the twentieth century. Events at the national level served to increase the demand for reforms in the southern electoral system, and these events eventually led Congress to reassert its interest in protecting black voting rights. Two important early events came during the New Deal and World War II. Although President Franklin D. Roosevelt did not assign civil rights a high priority on his legislative agenda, he did appoint several blacks to high positions within the federal government and also granted blacks direct economic aid for the first time (Lawson, 1976, pp. 20-21). During World War II, black veterans, permitted to cast a vote during the war, returned to the South and found, in many instances, their right of franchise revoked. Pressure against existing conditions increased as a result. As it became more and more evident that blacks constituted an important voting bloc in the North (and because blacks had showed signs of deserting the party of Lincoln ever since Roosevelt's bid for reelection in 1936), the Eisenhower administration began to consider ways of recapturing the vote of northern blacks. The strategy used to solicit black support was the introduction of civil rights legislation in Congress. Hence, the Justice Department drafted what was eventually to become the Civil Rights Act of 1957. Indeed, a new era in congressional involvement in protecting the franchise began in 1957 when Congress defeated a filibuster and passed the first civil rights law since 1875. The 1957 act, as well as the 1960 Civil Rights Act that followed, "employed a litigative strategy of enforcement which placed much faith in the federal courts' ability to rectify racial discrimination in the electoral process" (Garrow,

The right to vote and the right to representation

13

1978, p. 5). Reliance on litigation proved to be a mistake, but given a reluctant Republican president (despite Eisenhower's earlier endorsement of the idea) and a Congress in which southern Democrats hostile to federal interference held a great deal of power, a bill that authorized nothing stronger than a litigative approach was the only one that would have made it into law. What became available for blacks in 1957 was a fairly weak civil rights statute. The act authorized the appointment of an assistant attorney general for civil rights, upgraded the Civil Rights Section to a division of the Justice Department, and created a Commission on Civil Rights to monitor voting activities in the nation, especially in the South. The third section of the act granted the Justice Department the right to intervene in civil suits brought by private individuals alleging a denial of their civil rights. And the fourth section - probably the most important provision of the 1957 act - empowered the attorney general to file suits in federal district court seeking injunctive relief against violations of the Fifteenth Amendment. Another provision in the act gave jurisdiction of civil rights suits to special three-judge federal district courts convened to hear cases taken out of state courts by the Department of Justice without the complainant's having to exhaust all state remedies. In the three years following the 1957 legislation, the attorney general filed only four suits. The constitutionality of the act was upheld in the first case brought under the act, United States v. Raines (I960),12 but this case, as well as the other three suits initiated under the act, exposed several weaknesses in the act. The most obvious problems encountered by the Justice Department were "the difficulties in obtaining access to the necessary registration records, and most important, the almost limitless opportunities for delay in the judicial process" (Garrow, 1978, p. 14). Moreover, not a single black who was not already registered to vote qualified to vote in the three-year period following the passage of the act (Hawk and Kirby, 1965, p. 1060). As a consequence, new legislation - the 1960 Civil Rights Act - was passed in an effort to close some of the loopholes in the 1957 legislation. The most significant feature of the 1960 act was that it authorized a federal district court judge, after finding a "pattern or practice" of voting discrimination, to appoint federal referees to replace state registration officials. The referee was to register any applicants who demonstrated that they were qualified to vote under prevailing state law or standards. This provision had the effect of "freezing" state voter requirements at the level at which they had been applied to whites (regardless of any state laws formally prescribing higher standards) (Derfner, 1973, p. 546). District courts were directed to use the registrars' actual standards for registering white voters if those standards were less stringent than what was specified in the state statutes (and those applied to potential black voters). This section of the act was not as innovative or as effective as might

14

Minority Representation and Voting Equality

be supposed; first, the doctrine of "freezing" had already been established in earlier southern voting rights litigation, and second, the referee provision "went virtually unused by the courts" (Garrow, 1978, p. 29). Because the Justice Department had experienced difficulty procuring registration records from state officials (some registrars even destroyed the records rather than allow them to be used in litigation by the Justice Department), a provision was added to the 1960 act that declared that local voting records had to be preserved for a period of twenty-two months following an election and had to be made available to the attorney general upon request. The 1960 act also provided that a registrar's discriminatory acts be deemed those of the state, which might be joined as a party defendant in any suit. This provision was inserted to resolve the problem encountered in United States v. Alabama (1960), in which the registrars resigned and left no one against whom an injunction might be issued. Despite the inclusion of a number of measures designed to close the loopholes in the 1957 act, the 1960 Civil Rights Act proved no more successful than the earlier act. After four years and dozens of frustrating cases under the 1960 act, the only possible conclusion was that these acts failed in their purpose. Litigation under the acts was protracted and limited to only a small percentage of the counties in which black registration was being limited. Many legal observers attributed the numerous delays and obstructions to the personal sympathies of some of the federal district court judges themselves. The five judges most notorious for their recalcitrance (William Harold Cox, Daniel H. Thomas, Claude F. Clayton, E. Gordon West, and Benjamin Dawson) sat in districts in Mississippi, Alabama, and Louisiana, the three states in which the great majority of the voting rights suits were filed and those that had the poorest records in black enfranchisement. Eventually, most of the cases were resolved in favor of the Justice Department, but not until reaching a higher court. When decisions were handed down outlawing certain disenfranchising practices, however, states merely switched to other discriminatory devices, or alternatively, certain local officials simply defied or evaded court orders (South Carolina v. Katzenbach, 1966, p. 314). Meanwhile, dramatic events were drawing public attention to the plight of blacks in the South. The outbreak of violence - for example, the bombings of black churches in Birmingham, Alabama, during the freedom summer of 1963 and the killing of three civil rights workers in Philadelphia, Mississippi, in the summer of 1964 - galvanized civil rights advocates and created conditions sympathetic to action on behalf of blacks in the South. Even conservative Republicans were prompted to join the coalition that resulted in the passage of the 1964 Civil Rights Act. Although the major focus of the Civil Rights Act of 1964 was in the area of

The right to vote and the right to representation

15

public accommodations, federal funding of schools, and equal employment opportunities, the legislation did devote some attention to the issue of voting discrimination problems in the South. For example, one section of the act provided for expedition of the three-judge federal district courts designated to hear the vote discrimination cases. In addition, several sections were directed at the most flagrant methods of abusing literacy tests; these sections required that all literacy tests be given entirely in writing, barred the use of immaterial errors (e.g., misspellings) to deny registration, and established a sixth-grade education as presumptive proof of literacy in any voting discrimination suit brought by the Justice Department. However, like its predecessors, this act relied on a litigative strategy, and thus no real improvement was likely to result from its enactment. In any case, these voting provisions were outdated less than a year after being signed into law by passage of the Voting Rights Act of 1965. The three civil rights acts passed in 1957, 1960, and 1964 were remedial measures that did little to change black disfranchisement in the South. The record of eight years of litigation on behalf of minority voting rights reveals a great deal of effort generating few concrete results. According to Derfner, "the United States filed 71 suits . . . including actions against discriminatory registration practices, suits directed at private or official intimidation, and suits simply to obtain registrars' records . . . [as well as] omnibus actions against Alabama, Louisiana and Mississippi" (1973, p. 548). The outcome of this litigative effort was an increase of just 36,000 blacks on the registration rolls in the counties where suits had been brought (Hawk and Kirby, 1965, p. 1212). According to Chief Justice Earl Warren, writing in South Carolina v. Katzenbach (1966), registration of voting-age blacks "in Alabama rose only from 14.2% to 19.4% between 1958 and 1964; in Louisiana it barely inched ahead from 31.7% to 31.8% between 1956 and 1965; and in Mississippi it increased only from 4.4% to 6.4% between 1954 and 1964" (p. 313). Eventually the barriers to black registration would be broken down, but not by litigation under these three acts. The civil rights acts of 1957, 1960, and 1964 were nevertheless important because the progressively more stringent remedies adopted in the acts were subsequently upheld by the courts. This reassured Congress that the Supreme Court would no longer impede Congress in its exercise of power to enforce the Fourteenth and Fifteenth amendments through '' appropriate legislation.'' The 1965 Voting Rights Act The series of civil rights acts passed by Congress in the late 1950s and early 1960s did little to open the political system to African-Americans in the South. It was not until Congress adopted an entirely different tack that large numbers

16

Minority Representation and Voting Equality

of blacks began to gain access to the political process. The Voting Rights Act of 1965 was a revolutionary measure that went far beyond the previous civil rights acts, and it went a long way toward enforcing the promises of the Fourteenth and Fifteenth amendments. We shall not recount the civil strife or the legislative battles that led to the passage of the act. The story of the civil rights movement has been told in numerous books and films (e.g., Branch, 1988; Brooks, 1974; Kluger, 1975; Morris, 1984; Williams, 1987); the legislative history has also been discussed in some detail (Garrow, 1968; Lawson, 1976, 1985; Thernstrom, 1987). It is important, however, to give a general overview of the act's original provisions, the 1970 and 1975 extensions, and the expansion of its coverage to language minorities, as well as to note its immediate effects on minority registration and turnout. The content of the 1965 act The Voting Rights Act passed in 1965 contained a number of permanent provisions that affected the entire nation. These were found in Sections 2 and 3. Section 2 was essentially a restatement of the Fifteenth Amendment. It prohibited voting qualifications or prerequisites to voting or any standard, practice, or procedure by any state or political subdivision that would "deny or abridge the right of any citizen of the United States to vote on account of race or color." Section 3 strengthened the procedures by which the attorney general filed suit to enforce the voting guarantees of the Fifteenth Amendment. Under this section, a court could authorize the appointment of federal examiners and observers, suspend the use of tests and devices, or require preclearance in any jurisdiction in the United States - regardless of whether or not the jurisdiction was covered by the "triggering" formula of special provisions - if the court deemed it appropriate to enforce the Fifteenth Amendment. (The use of federal examiners and observers, the suspension of literacy tests, and the preclearance provision will be described in greater detail later.) Other general provisions provided for civil and criminal penalties for violations of the act. The heart of the act was the special provisions listed in Sections 4 through 9. These special provisions authorized the Justice Department to take direct administrative action on behalf of disfranchised blacks, bypassing the southern judiciary. Unlike the general provisions, the special provisions were to be temporary and applied only to jurisdictions that met the criteria specified in Section 4(b). According to Section 4(b), a jurisdiction was subject to the act's special provisions if it met the following test: (1) The jurisdiction maintained a test or device as a precondition for registering or voting as of November 1, 1964, and (2) less than 50 percent of the voting-age population was registered to vote on

The right to vote and the right to representation

17

November 1, 1964, or less than 50 percent of the voting-age population voted in the November 1964 presidential election. If a state as a whole did not meet these criteria, the standard was applied to individual counties within the state, so that in some instances entire states were "covered" and in other cases only certain counties within a state were "covered." This coverage formula was devised by Congress to target southern states with a history of racial discrimination in the election process. Although the formula did not capture every southern state that had disfranchised blacks, it did reach the worst offenders. The areas initially captured by this "triggering mechanism" were Alabama, Georgia, Louisiana, Mississippi, South Carolina, Virginia, and parts of North Carolina.13 Under Section 4(a) of the act, a state or political subdivision could escape from coverage by filing a declaratory judgment action in the U.S. District Court for the District of Columbia. If that court found that no test or device had been used to deny or abridge the right to vote on account of race during the five years preceding the filing of the action, the state or subdivision would be removed from the operation of the act's special provisions. Section 4(a) of the act also suspended for five years the use of all "tests and devices" used as a prerequisite for voting in all "covered" jurisdictions.14 It had long been held that literacy tests were the primary deterrent to black efforts to vote in many areas of the Deep South, and this measure was designed to eradicate this obstacle, at least temporarily. The Voting Rights Act also attempted to prevent jurisdictions from passing new legislation that would deny or abridge the right of African-Americans to vote. Section 5 of the 1965 legislation required that covered states submit changes in "any voting qualification or prerequisite to voting, or standard, practice or procedure with respect to voting" different from that in effect on November 1, 1964, to either the attorney general or the U.S. District Court for the District of Columbia for "preclearance" before any such change could be enforced. This section froze all voting regulations in the covered jurisdictions as of November 1964, unless the attorney general or the District Court of the District of Columbia determined that the proposed voting change did not have the purpose or the effect of denying or abridging the right to vote on account of race or color. If the attorney general denied preclearance to a submission, the change could not be implemented. However, even if the attorney general did object to a proposed change, the jurisdiction could still ask for a declaratory judgment, from a three-judge district court in the District of Columbia, that the change was not discriminatory. According to the act, decisions as to whether a jurisdiction could change a voting procedure rested only with the Justice Department or the District Court for the District of Columbia. A Washington environment was designated to

18

Minority Representation and Voting Equality

resolve these issues, because past experience indicated that local officials and district courts in the South frequently hindered efforts to ensure blacks' voting rights. Because the broad discretion vested in local registrars was often abused (exactly as the original framers of the state laws had intended), the act provided, in Sections 6 through 8, for the assignment of federal registration examiners and election observers. The Civil Service Commission was directed to appoint federal examiners for any covered jurisdiction when the attorney general certified either (1) that he had received twenty meritorious complaints from that jurisdiction alleging denial of the right to vote on racial grounds or (2) that based on his own judgment, he determined that the appointment of examiners was necessary to enforce the guarantees of the Fifteenth Amendment. The primary duty of federal examiners was to determine whether an applicant was qualified to vote under state laws that were "not inconsistent with the Constitution and the laws of the United States." All who qualified on these grounds were to be placed by the federal examiner on a list of eligible voters. This list was then to be transferred to the appropriate state or local registrar for placement on the jurisdiction's official voting list. Section 8 provided for the assignment of federal election observers. Federal observers were to be appointed by the Civil Service Commission, at the request of the attorney general, to serve in jurisdictions that had been designated for the appointment of federal examiners. The duty of the observers was to act as poll watchers to observe whether all eligible persons were allowed to vote and whether all ballots were accurately counted. Another provision of the act (Section 10) expressed the opinion of Congress that "the constitutional right of citizens to vote is denied or abridged in some areas by the requirement of the payment of a poll tax as a precondition for voting." This section directed the attorney general to institute suits against states that retained the poll tax in order to determine whether such a tax violated the constitution.15 The final draft of this section represented a compromise between the advocates of a total ban on the poll tax and those who feared such a ban with regard to state elections was unconstitutional and might therefore endanger the constitutional position of the entire act (Hawk and Kirby, 1965, p. 1203). President Lyndon B. Johnson, shortly after his landslide election in November 1964, instructed his attorney general, Nicholas Katzenbach, to "write the goddamnedest toughest voting rights act that you can devise" (Raines, 1977, p. 337). Thus, it is not surprising that the 1965 act was immediately challenged in court. The case of South Carolina v. Katzenbach was heard by the Supreme Court under its original jurisdiction and was decided on March 7, 1966, just seven months after its enactment. Writing for eight members of the Court, Chief Justice Warren recognized that the act was "an uncommon exercise of congres-

The right to vote and the right to representation

19

sional power" but that "exceptional conditions can justify legislative measures not otherwise appropriate" (p. 334). All facets of the act presented to the justices, including the power to suspend literacy tests and other devices of Section 4 and the preclearance provisions of Section 5, were determined to be constitutional.

Extensions of the act and expansion of its coverage The special provisions of the Voting Rights Act of 1965 were initially conceived of as temporary measures to be terminated once blacks had been successfully integrated into the political system in the South. Originally enacted for a fiveyear period, the scheduled expiration date was August 6, 1970. At that time a jurisdiction could "bail out" from coverage by convincing the District Court for the District of Columbia that it had not used a discriminatory "test or device" for the preceding five years. Because the use of such tests had been suspended for five years, these jurisdictions would presumably have no difficulty meeting this requirement unless the act were extended for an additional period of time. Despite efforts by the Nixon administration to pass a severely weakened version, the legislation that was adopted eventually proved to be as stringent as the original act. Congress suspended literacy tests in all states until August 1975. The special provisions of the act were extended for another five years, and the bailout criterion was amended so that a jurisdiction had to have gone at least ten years - rather than five - without the use of a discriminatory test or device. The coverage also was expanded to include states or political subdivisions that, as of November 7, 1968, were using one of the specified tests or devices and in which less than half of the voting-age population was either registered to vote or had actually voted in the 1968 presidential election. No additional southern jurisdictions were captured by this revision of the coverage formula, but three boroughs of New York City - Manhattan, Brooklyn, and the Bronx - were brought under the special provisions.16 The 1970 legislation also lowered the voting-age requirement to eighteen in all elections, and it standardized residency requirements by placing a ban on state residency requirements for federal elections of more than thirty days. The first case in which the Supreme Court was asked to consider the 1970 amendments to the Voting Rights Act was Oregon v. Mitchell (1970). In a decision that had no clear-cut majority, the Court ruled that Congress did not have the power to lower the voting age for state elections.17 However, all nine justices, in four separate opinions, upheld Congress's power to impose a national ban on literacy tests. Even though literacy tests were not always discriminatory,

20

Minority Representation and Voting Equality

each of the justices agreed that Congress should be given a great deal of leeway in creating a remedy, given the prevalence of the problem. Because the extension of the Voting Rights Act was for only five years, its special provisions were due to expire again in August 1975. When Congress began to consider possible renewal of the act early in 1975, there was a broad consensus for extending the act; even the Republican administration of President Gerald R. Ford favored a straightforward extension. The general feeling at the time was that despite a great deal of progress, black electoral participation still relied heavily on federal surveillance and any curtailment of federal involvement would jeopardize these gains (Lawson, 1985, pp. 225-227). Black civil rights groups were not the only groups lobbying for an extension of the act; other minority groups, particularly Hispanics, argued that they too should have federal protection of their voting rights. Although Congress had previously given little thought to voting discrimination against minority groups other than African-Americans, the development of the Chicano political movement and continuing voting rights litigation in Texas (which was not a covered jurisdiction), convinced many that protection should be extended to other minority groups as well (Hunter, 1976, pp. 254-255). Some black leaders, such as Clarence Mitchell of the NAACP (National Association for the Advancement of Colored People), were initially hostile to the idea of expanding coverage to other minority groups, arguing that this would dilute the bill or that the inclusion of these groups might threaten the constitutionality of the measure as a whole (Lawson, 1985, p. 229). Eventually, however, such a provision was included when President Ford signed the bill into law on the day that the 1970 legislation was due to lapse (August 6, 1975). The most important of the 1975 revisions to the Voting Rights Act was the expansion of its protection to language minorities. The coverage formula of Section 4 was amended to include those states and counties with a substantial language-minority population and English-only election materials. More specifically, Congress added to the list of tests and devices the conduct of registration and elections in only the English language in those states or local subdivisions where more than 5 percent of the voting-age population belonged to a single language minority of the following specified groups: Alaskan natives, Native Americans, Asian-Americans, or people of Spanish heritage.18 Furthermore, the act required the use of bilingual election materials and assistance if 5 percent of the area's voting-age citizens were of a single language minority and the illiteracy rate in English of that language minority was greater than the national illiteracy rate. (An exception would be made if the language historically had been oral or unwritten; if this were the case, then the jurisdiction would be required only to furnish oral instructions, assistance, and voting information.) The 1975 bill also extended the act's special provisions, this time for an

The right to vote and the right to representation

21

additional seven years. The coverage formula was revised to include jurisdictions that maintained any test or device and had less than half of its voting-age population either registered on November 1, 1972, or casting votes in the 1972 presidential election. Literacy tests, which had been temporarily suspended in 1970, were eliminated permanently under the 1975 legislation. Other amendments to the act codified the effects of the Twenty-fourth and Twenty-sixth amendments and the Supreme Court's decision in Harper v. Virginia Board of Elections (1966).19 By expanding the protection of the act to encompass not only racial minorities but a number of language-minority groups as well, the 1975 legislation extended the Voting Rights Act beyond the confines of the Fifteenth Amendment and beyond its original geographic target of the South. The provisions directed at language minorities did not rely on Congress's power to enforce the Fifteenth Amendment's prohibition against voting discrimination on the basis of race. Instead, Congress invoked its power to pass "appropriate legislation" to secure the Fourteenth Amendment's guarantee of equal protection. The geographic focus of the act was broadened by extending protection to language minorities; the coverage formula captured far more than the original six states in the South. Areas in the Southwest and West were now subject to the special provisions of the act; for example, the entire state of Arizona was covered, as well as jurisdictions in Colorado, South Dakota, and California. Additional areas in the South captured by this language-minority trigger were the entire state of Texas, five counties in Florida, and another county in North Carolina. In addition, all of Alaska was now covered. The impact of the Voting Rights Act on black participation The Voting Rights Act of 1965 and its extensions in 1970 and 1975 had a profound effect on black political participation in the South. The percentage of voting-age blacks registered in the South in March 1965 was only 35.5 percent, compared with 73.4 percent of voting-age whites (U.S. Commission on Civil Rights, 1968, pp. 222-223). The percentage of blacks registered was particularly low in those states targeted by the special provisions of the act (29.3 percent in March 1965), and it was in this area of the South that the act had the most direct impact. By the end of the year, federal examiners working in thirty-two counties in the covered states had listed the names of over 79,000 AfricanAmericans to be added to the registration lists (Lawson, 1985, p. 15). By the end of 1967, more than half a million new black voters had been listed in the seven covered states: Civil Service Commission records indicate that federal examiners assigned to fifty-eight counties in the covered southern states had registered 158,094 persons, and officials of the Justice Department estimated

22

Minority Representation and Voting Equality

that an additional 416,000 blacks had been registered by local registrars (U.S. Commission on Civil Rights, 1968, p. 12). Survey data indicate that the percentage of eligible blacks registered rose sharply in the seven covered states, from an average of 29 percent before passage of the act to over 52 percent two years later. By 1967 no covered state had less than 50 percent of the blacks of voting-age population registered (see Table 1). The increase in registration was particularly dramatic in Alabama and Mississippi. Since 1970, changes in black registration rates have been more erratic but have generally moved slightly upward. White registration rates, on the other hand, have climbed more slowly since 1967 or have even declined (see Table 2). The disparity between black and white registration rates has fallen dramatically as a result of these trends. In no covered state is there a difference of more than 7.5 percent, and in Louisiana the percentage of blacks registered is greater than the percentage of whites. This is not to suggest that there is no longer a discrepancy in black and white registration rates; in most southern states, blacks are still not registering at the same rates as whites are.20 The substantial increase in the numbers of blacks registered has been accompanied by a significant rise in the number of blacks actually voting. The growth in the turnout of blacks was most evident immediately following passage of the act. Table 3, for example, shows the turnout in the presidential elections of 1964 and 1968 in each of the covered states as well as the United States as a whole. As the table shows, although the national turnout dropped slightly in 1968, it increased in all seven covered states. The increase was particularly marked in Mississippi and Alabama - 19.3 and 16.8 percentage points, respectively. Moreover, since passage of the Voting Rights Act, the disparity between the turnout of blacks and whites has often been smaller in the South than in the entire United States (see Table 4, p. 26). The increases in black participation since 1960 are remarkable, perhaps even more so when one considers that as late as 1940, 95 percent of the adult blacks residing in the South were deterred from registering. Only an estimated 250,000 blacks were listed on the registration rolls in the eleven states of the Old Confederacy in 1940. The barriers to black participation at that time were numerous: the all-white Democratic primary, poll taxes, literacy tests, the biased administration of registration procedures, and white brutality and economic intimidation. Within a generation, these barriers were largely dismantled, and the number of black registrants rose thirteenfold. By far the biggest increases in black registration occurred in the late 1960s in the seven southern states covered by the Voting Rights Act. Although not all of this rise can be directly attributed to the passage of the act, it must certainly be in large measure responsible: The suspension of literacy tests and the use of federal examiners unquestionably played

The right to vote and the right to representation

23

Table 1. Registration by race, before and after 1965, in southern states covered by the Voting Rights Act (percent) Pre-act registration

Post-act registration

Black

White

Gap

Black

White

Gap

Alabama Georgia Louisiana Mississippi North Carolina South Carolina Virginia

19.3 27.4 31.6 6.7 46.8 37.3 38.3

69.2 62.6 80.5 69.9 96.8 75.7 61.1

49.9 35.2 48.9 63.2 50.0 38.4 22.8

51.6 52.6 58.9 59.8 51.3 51.2 55.6

89.6 80.3 93.1 91.5 83.0 81.7 63.4

38.0 27.7 34.2 31.7 31.7 30.5 7.8

Total

29.3

73.4

44.1

52.1

79.5

27.4

Note: Percentages are of the voting-age population. Pre-act registration is from March 1965; post-act registration is from September 1967. Both figures are estimates. Source: U.S. Commission on Civil Rights (1975, p. 43).

a major part in clearing the path to the polling places for nearly three-fifths of the voting-age blacks in this area of the South. In this respect, the Voting Rights Act - and the civil rights movement of which it was an outgrowth - was an unquestioned success. Vote dilution: the right to vote versus the right to representation At one level, then, the right of minorities to register and to vote has been largely achieved. To be sure, there is some continued discrimination involving even this basic right. It may be more subtle than previously, and the individuals who act on it may each operate on a fairly narrow scope, but it still exists.21 And one can argue that the government should take an active role in encouraging minority registration and voting rather than simply removing the barriers to participation (Jones, 1985). Nevertheless, vast strides have been taken; the enormous barriers to registration and voting that existed up to the 1960s have been removed, and by comparison with earlier periods, electoral participation by African-Americans is very high. Yet even before the Voting Rights Act was passed, it was apparent that voting was only the first hurdle to effective participation. The resurgence of black electoral participation brought to light a second-order obstacle that had remained obscure as long as blacks were completely excluded from the political process. Jurisdictions that wished to continue to discriminate against blacks simply moved from denying them access to the ballot to more sophisticated schemes

Minority Representation and Voting Equality

24

Table 2. Registration by race since 1970 in southern states covered by the Voting Rights Act (percent) 1970 Registration

Alabama Georgia Louisiana Mississippi North Carolina South Carolina Virginia

1980 Registration

1988 Registration

Black

White

Gap

Black

White

Gap

Black

White

Gap

69.7 58.3 59.7 75.5 53.6 58.6 58.7

82.5 70.0 75.0 78.9 66.3 60.2 63.3

12.8 11.7 15.3 3.4 12.7 1.6 4.6

62.2 59.8 69.0 72.2 49.9 61.4 49.7

73.3 67.0 74.5 85.2 63.7 57.2 65.4

11.1 7.2 5.5 13.0 13.8 -4.2 5.7

68.4 56.8 77.1 74.2 58.2 56.7 63.8

75.0 63.9 75.1 80.5 65.6 61.8 68.5

6.6 7.1 -2.0 6.3 7.4 5.1 4.7

Note: Percentages are of the voting-age population. Source: 1970: Statistical Abstract of the United States (1974, Tables 701, 705); 1980: U.S. Department of Commerce, Bureau of the Census, (1982, Table 5); 1988: U.S. Department of Commerce, Bureau of the Census, (1989, Table 4).

developed to dilute the impact of their new voting strength. Devices such as atlarge elections, anti-single-shot laws, a decrease in the size of legislative bodies, racial gerrymandering, exclusive slating, and changing offices from elective to appointive ensured that although blacks might vote, they would often be unable to elect candidates of their choice.22 In many instances these devices were already in place, a legacy from the post-Civil War "Redeemer" days and later municipal "reforms," and had only to be dusted off or expanded in their coverage (Kousser, 1984, pp. 30-40). In other instances, changes were adopted for the express purpose of excluding blacks from office, often without the "preclearance" required by Section 5 of the Voting Rights Act.23 There was little doubt that these efforts had to be met head-on if the Voting Rights Act was to have any meaning at all. They were such bald-faced attempts to prevent blacks from achieving positions of power that even Abigail Thernstrom, a strong opponent of later interpretations of the Voting Rights Act, concedes that "the [Supreme] Court could hardly refuse to act" (1987, p. 24) and that the decision in Allen v. State Board of Elections (1969) to invalidate certain changes in electoral laws in Mississippi was both "correct and inevitable" (p. 30). With the Allen decision, it was clear that Section 5 of the Voting Rights Act, requiring preclearance by the attorney general or the D.C. district court of changes in any "standard, practice, or procedure," would be applied broadly to all manner of election laws rather than narrowly to laws affecting only registra-

The right to vote and the right to representation

25

Table 3. Turnout in the 1964 and 1968 presidential elections in southern states covered by the Voting Rights Act (percent) 1964

1968

Alabama Georgia Louisiana Mississippi North Carolina South Carolina Virginia

35.9 43.3 47.3 33.9 52.3 39.4 41.1

52.7 43.4 54.8 53.2 54.3 46.7 50.1

United States

61.8

60.7

Note: Percentages are of the voting-age population. Source: Abstract of the United States, 1974, Table 704.

tion and voting. Straightforward, obvious efforts to nullify, in effect, the votes of large numbers of blacks - "a dilution of voting power" in Chief Justice Warren's words (Allen v. State Board of Education, 1969, p. 569) - were ruled out. It turns out, however, that the question of what constitutes vote dilution is not easily answered. Indeed, efforts to define, operationalize, and eradicate vote dilution have been the largest source of "districting" litigation since the 1960s, surpassing even that involving the equal population requirement. At its most general level, vote dilution refers to "the practice of reducing the potential effectiveness of a group's voting strength by limiting its ability to translate that strength into the control of (or at least influence with) elected public officials" (Engstrom, 1980, p. 197). In the context of minority voting rights, the operative definition comes from the revised language of Section 2 of the Voting Rights Act:24 that members of racial or language minorities "have less opportunity than other members of the electorate to participate in the political process and to elect representatives of their choice.' ' 25 It is clear from these definitions that an essential ingredient for identifying vote dilution is the rate of election of candidates chosen by minority voters. Indeed, the Voting Rights Act makes that point (immediately after the preceding quotation): ' 'The extent to which members of a protected class have been elected to office in the State or political subdivision is one circumstance which may be considered." Beyond that general point, however, lies considerable ambiguity, and it is in response to this ambiguity that courts have been heavily involved

26

Minority Representation and Voting Equality

Table 4. Turnout of blacks and whites in presidential elections since 1964 for the United States and the South (percent) United States

1964 1968 1972 1976 1980 1984 1988

South

Black

White

Gap

Black

White

Gap

58.5 57.6 52.1 48.7 50.5 55.8 51.5

70.7 69.1 64.5 60.9 60.9 61.4 59.1

12.2 11.5 12.4 12.2 10.4 5.6 7.6

44.0 51.6 47.8 45.7 48.2 53.2 48.0

59.5 61.9 57.1 57.1 57.4 58.1 56.4

15.5 10.3 9.2 11.4 9.2 4.9 8.4

Note: Percentages are of the voting-age population. Source: U.S. Department of Commerce, Bureau of the Census (1989, Table A).

in determining when minorities have less opportunity to elect candidates of their choice. How many candidates must win for there to be nondilution? At one extreme is a notion embodied in Section 5 of the Voting Rights Act. A new or altered electoral law or device is dilutionary if it results in "retrogression" - that is, if it results in a decline in minority representation. Thus, under the retrogression criterion, minority representation is judged according to what currently exists rather than what might exist under alternative circumstances. The problem, of course, is that what currently exists may be patently unfair or clearly the product of past discrimination.26 Although nonretrogression is thus a useful tool to prevent backsliding, it is certainly not sufficient to eliminate vote dilution. At the other extreme is proportional representation, which would require that minorities be represented in the same proportion as they are in the overall population.27 Although such a requirement would surely eliminate dilution (at least for the groups that were proportionately represented), there are two major problems. First, a system of geographically based single-member districts, which is our predominant electoral arrangement, yields proportional representation only by chance; most often, the majority group (whether a party, racial, or other group) gains proportionately more representation than votes.28 Second, and more to the point, the Voting Rights Act specifically states that "nothing in this section [of the act] establishes a right to have members of a protected class elected in numbers equal to their proportion in the population."29 Obviously, there is a vast ground between these extremes, and there is nothing in political theory or the law that tells us unambiguously when a voting system

The right to vote and the right to representation

27

is nondilutionary. It has been left to the courts, with some help from the Senate Report accompanying the most recent (1982) revisions of the Voting Rights Act, to determine what evidence is sufficient to determine when minorities have been given less opportunity to elect candidates of their choice. Given the difficulties of this task, the courts have typically focused on mechanisms that tend to dilute the vote, requiring specific information about their applicability in the jurisdiction in question.30 But even focusing on specific electoral devices leaves important questions unanswered: (1) What does it mean to be a candidate "of choice"? Surely it cannot simply mean the candidate who obtains the highest number of minority votes; if only whites/Anglos were allowed to run, it could hardly be said (at least as a general rule) that the one with the most minority votes was the candidate of choice for minority voters. But does this mean that only minorities can represent minorities? (2) Does an election law have to have been adopted for discriminatory purposes? Certainly some election laws - including at-large elections in some cities - were adopted for "good government" reasons. Is it illegal vote dilution if under such circumstances these laws make it extemely difficult for minorities to be elected? (3) How does one design an electoral system, or a specific district, so that it does not dilute the minority vote (a question that usually arises at the remedy stage of a court case)? Just how certain does one have to be that a minority-preferred candidate will win the election? (4) And going back to the most fundamental point, how many minority victories are enough, or what is the standard against which to compare the number of minority victories? Should the standard be the number of minorities who are likely to win with "neutral" districting? The number likely to win with raceconscious districting? The number who could win if a "semiproportional" system were used? These are the sorts of questions with which Congress, the courts, minority groups, civil rights lawyers, and political scientists have been wrestling for the past two decades. Some answers have been found, particularly in regard to atlarge elections and multimember districts. We will describe these in some detail as we relate the evolution of a vote dilution standard in Congress and the courts. But numerous questions remain, especially pertaining to the single-member district context, which we see as the battleground for voting rights litigation in the 1990s. On these matters, we will offer perspectives that we hope will shape the ongoing arguments and, ultimately, lead to fair and equitable arrangements. We are aware, however, that we provide more in the way of problems than solutions. We do this in order to avoid the strident tone exhibited in many writings on the subject, even some that we have contributed. If political science has taught us anything, it is that there are no perfect solutions, even in the abstract. Knowing that, the task is to arrive at compromises that offer a reason-

28

Minority Representation and Voting Equality

able degree of equity to all involved. Even that may be an impossible task. But if we are to get that far, it will come only from a full understanding of the problems we face. Illuminating those problems will fill the remainder of this book. In this chapter we first provided an overview of the struggle for minority voting rights through the passage of the Voting Rights Act of 1965 and the 1970 and 1975 amendments. We then noted that securing the right to vote did not guarantee minorities the power to elect representatives of their choosing, which led us to the notion of vote dilution and a discussion of some of the problems in defining and measuring the concept, which serves as an introduction to the chapters to follow. In Chapter 2 we trace the evolution of a vote dilution standard as it developed in the courts during the 1970s and early 1980s, in Congress with the adoption of the 1982 amendments to the Voting Rights Act, and again in the courts through the Supreme Court's decision in Thornburg v. Gingles (1986).

The evolution of a vote dilution standard through 1986

Dilutive electoral devices such as at-large election systems and racial gerrymandering have been challenged on at least two separate grounds. First, plaintiffs have cited constitutional issues, alleging that such devices violate the Equal Protection Clause of the Fourteenth Amendment and the guarantee in the Fifteenth Amendment that the right to vote shall not be denied on racial grounds. Second, and more recently, plaintiffs have relied on the Voting Rights Act. Especially since 1982, plaintiffs have argued that certain practices violate Section 2 of the act; for jurisdictions captured by the trigger provision of Section 4, Section 5 offers another alternative. Section 5 cases were especially important after the 1970s round of redistricting because they could be brought against those jurisdictions considered the worst offenders and because the burden of proof rests with the jurisdiction attempting to alter the electoral practice or procedure. These cases significantly reduced the use of multimember districts in southern state legislatures (Grofman and Handley, 1991), and they remain, according to one civil rights lawyer, the "first line of defense" in the covered areas.1 Nonetheless, the applicability of Section 5 to only covered jurisdictions and to only changes in electoral practices is a severe limitation. Consequently, cases involving constitutional arguments were also brought throughout the 1970s and early 1980s. In 1982 the Voting Rights Act was again extended, and in the process important amendments were made to Section 2; most of the litigation since then has been based on that revision. Lower courts began interpreting the revised Section 2 immediately after it was approved, and eventually, in June 1986, the Supreme Court decided Thornburg v. Gingles, the current authoritative legal interpretation for most vote dilution cases.2 In this chapter we review these developments, first summarizing the most significant cases brought under Section 5 and then covering in detail the development of constitutional arguments, the amendment of the Voting Rights Act, the series of cases culminating in Thornburg v. Gingles, and the Gingles decision itself. 29

30

Minority Representation and Voting Equality

Section 5 protection against vote dilution Section 5 was included in the Voting Rights Act to prevent covered jurisdictions from replacing the tests and devices suspended by Section 4 of the act with new legislation that would deny or abridge the right of blacks to vote. As noted in Chapter 1, this section froze all voting regulations in covered jurisdictions as of November 1964, requiring these jurisdictions to submit changes in "any voting qualification or prerequisite to voting, or standard, practice or procedure with respect to voting" to either the attorney general or the U.S. District Court for the District of Columbia for "preclearance." The proposed changes would be precleared only if the jurisdiction could convince the Justice Department or the district court that they did not have a racial purpose or effect. The significance of Section 5 did not become evident until the decision in Allen v. State Board of Elections (1969), in which the Supreme Court applied this section to changes that diluted black citizens' votes as well as to devices that disfranchised blacks. In Allen the majority held that "Congress intended to reach any state enactment which altered the election law of a covered State in even a minor way" (p. 566). Consequently, the Court determined that a Mississippi statute authorizing a switch from district to at-large elections for county supervisors was a change in voting practices covered by Section 5:3 The right to vote can be affected by a dilution of voting power as well as by an absolute prohibition on casting a ballot. .. . Voters who are members of a racial minority might well be in the majority in one district, but a decided minority in the county as a whole. This type of change could therefore nullify their ability to elect the candidate of their choice just as would prohibiting some of them from voting, (p. 569) The Court in Allen also ruled that private parties could challenge unsubmitted changes and that local federal district courts could consider whether Section 5's requirements were pertinent to the changes at issue, but that only the District Court of the District of Columbia or the attorney general could rule on whether or not a "covered" change had a discriminatory purpose or effect on black voters. Two years later, the Supreme Court reaffirmed its broad construction of Section 5 in Perkins v. Matthews (1971) by holding that jurisdictions covered by Section 5 could not annex territory or relocate polling places without preclearance. Forcing Section 5 compliance on municipal boundary alterations was important because the annexation of an adjacent area with large numbers of white voters was a common ploy for diluting the voting power of blacks (e.g., the cities of Richmond and Petersburg, Virginia, both used annexations to transform cities that had become majority black into cities in which whites once again comprised a majority of the population).4 Following Allen and Perkins, an increasing number of electoral changes were

The evolution of a vote dilution standard

31

submitted by covered jurisdictions for review and approval by the Justice Department. In the wake of the 1970 census and the ensuing redistricting, a substantial number of important districting proposals were submitted to the Justice Department, and in Georgia v. United States (1973), the Court determined that redistricting plans were subject to Section 5. The Court went on to state that "had Congress disagreed with the interpretations of Section 5 in Allen, it had ample opportunity to amend the statute," and given that Congress chose not do so when it extended the Voting Rights Act in 1970, "we can only conclude, then, that Allen correctly interpreted" Section 5 as encompassing a broad spectrum of voting practices and procedures (p. 534). Despite the Court's broad interpretation of Section 5 to include a wide variety of dilutive techniques, there are serious limitations to Section 5 as a weapon for attacking minority vote dilution. First, Section 5 does not apply to the entire United States but, rather, only to jurisdictions captured by the Section 4 trigger.5 Second, Section 5 applies only to proposed changes in districting. In particular, voting laws and practices adopted before 1964, when Section 5 of the act went into effect (or before a jurisdiction's inclusion under Section 5 of the act), are not subject to challenge. Third, until the mid-1980s, the only test for a Section 5 violation other than intentional discrimination was a test for retrogression, in which a change was held to be violative if, and only if, it led to a deterioration (retrogression) in minority representation. This is an extremely weak test, as it does not address the possible dilutive effects of the original laws.6 Because of these limitations, many discriminatory practices had to be challenged through alternative routes. Thus, at the same time that Section 5 cases were being adjudicated, plaintiffs mounted constitutional attacks on vote dilution outside the framework of Section 5 of the Voting Rights Act. Constitutional protection against vote dilution Beginning in 1964, with its decision in Reynolds v. Sims, the Supreme Court recognized vote dilution as an actionable wrong under the Constitution. In Reynolds, the landmark legislative reapportionment case arising out of Alabama, the Court held that voting districts that varied significantly in population violated the Equal Protection Clause of the Fourteenth Amendment. The Court did so by determining that the right to vote could be abridged unconstitutionally by a dilution of one's voting strength as well as by an outright denial of the ballot. This decision, of course, led to a long series of "one person, one vote" cases, most of which are not directly relevant here. However, there was another series of cases leading directly to the matter of minority voting rights. These began with an assault on the constitutionality of multimember districts.

32

Minority Representation and Voting Equality

The Court was first asked to consider the possible dilutive effect of a multimember districting plan only six months after Reynolds was decided. In Fortson v. Dorsey (1965), residents of the seven multimember senatorial districts in Georgia alleged that multimember districts were a violation of the Equal Protection Clause because specific areas of a district could be outvoted by other areas of the district, thereby preventing the former areas from electing candidates of their own choice. The Court held that multimember legislative districts were not unconstitutional per se, but Justice William J. Brennan, Jr.'s majority opinion warned: "It might well be that, designedly or otherwise, a multi-member constituency apportionment scheme, under the circumstances of a particular case, would operate to minimize or cancel out the voting strength of racial or political elements of the voting population" (p. 439). In 1966, in Burns v. Richardson, the Court further elaborated on the Fortson formulation. In Burns, the Court reviewed the constitutionality of an interim apportionment plan adopted by the Hawaii legislature. The Court again declined to mandate single-member districts, but Justice Brennan, delivering the opinion of the Court, repeated his admonition that apportionment schemes that used multimember districts would constitute an "invidious discrimination" if they met the rule established in Fortson. The Court argued that an "invidious effect" or an "invidious result" could be more easily demonstrated if the multimember districts were large in relation to the total number of legislators, if there were no residential subdistricts within the districts, or if multimember districts were employed in the legislative apportionment plans of both houses (p. 88). The Court's first venture into the issue of "racial" vote dilution came five years later, in Whitcomb v. Chavis (1971).7 In Whitcomb, the Court held that the validity of multimember districts was justiciable on minority vote dilution grounds but rejected a claim that the county wide election of a multimember state legislative delegation from Marion County (Indianapolis), Indiana, "illegally minimizes and cancels out the voting power of a cognizable racial minority" (p. 144). Black residents of Indianapolis had challenged the county wide multimember districting plan, alleging that the multimember legislative district plan discriminated against blacks living in the "ghetto area" by "invidiously diluting] the force and effect of [their] vote" (p. 128). As proof of the dilutive effect of the multimember district, the challengers offered evidence that the number of black legislators was not in proportion to the entire minority population. The Court found no evidence that "Marion County's multi-member district . . . [was] conceived or operated as [a] purposeful devic[e] to further racial or economic discrimination" (p. 149). In rejecting this challenge, the Court held that a claim alleging a lack of proportional representation alone was not sufficient to support a Fourteenth Amendment vote dilution challenge. The Court

The evolution of a vote dilution standard

33

added that in order to sustain their claim, plaintiffs would also have to produce sufficient evidence that minority "residents had less opportunity than did other . . . residents to participate in the political process and to elect legislators of their choice" (p. 149). However, the Court neglected to state explicitly what evidence would have been necessary to meet this standard. Although the Court declined to find multimember districts dilutive of minority voting rights in its initial exposure to the issue, its reasoning in subsequent cases indicated that the constitutionality of such schemes was more suspect in the South. Indeed, a series of rulings in the early 1970s, at both the lower court and Supreme Court levels, established that multimember districts in the South, given that area's long history of racial disfranchisement, had less chance of withstanding a legal assault from blacks. White v. Regester (1973), decided two years after Whitcomb, was the first case in which the Supreme Court sustained an attack on the use of multimember districts.8 In this case, the Supreme Court unanimously held that multimember state legislative districts in Dallas and Bexar (San Antonio) counties, Texas, diluted the voting power of African-American and Mexican-American voters in violation of the Equal Protection Clause. After reiterating the Whitcomb holding that multimember districts are not unconstitutional per se and that the plaintiff must show more than a mere lack of proportional representation, the Court indicated that a lack of equal access to the political process is necessary to establish a constitutional violation and, for the first time, established relatively formal guidelines for the evaluation of equal opportunity. In attempting to ascertain the political access of the two minority groups, the Court deferred to the district court's "intensely local appraisal" (p. 769) of the situation. In doing so, the Court reviewed factors that supported plaintiffs in their claim that multimember districts were being used to unconstitutionally dilute minority voting power: the history of official racial discrimination in Texas's election process (i.e., the white primary system and the use of a poll tax), multimember districts combined with a majority vote requirement and a place system not tied to residency, slating organizations that were controlled by whites, election campaigns that employed racial tactics, a lack of minority elected officials in the community, and the election of candidates who were "insufficiently responsive" (p. 769) to minority group interests. Although the Court noted that these characteristics viewed by themselves were not necessarily invidious or improper, they were enough to sustain the judgment based on the "totality of the circumstances" (p. 769). In Zimmer v. McKeithen (1973), the Fifth Circuit Court relied on the "totality of circumstances" standard handed down by the Court in White to rule that atlarge elections for police jurors and school board members in East Carroll Parish, Louisiana, diluted the voting strength of black residents in violation of the

34

Minority Representation and Voting Equality

Fourteenth and Fifteenth amendments. In establishing the standards according to which minority vote dilution might be judged, the Fifth Circuit relied heavily on a list of factors gleaned from the Supreme Court's opinion. The circuit court, however, in attempting to bring some order to the "panoply of factors" identified in White, offered a more systematic and comprehensive set of dilution guidelines than Justice Byron R. White had presented. According to Zimmer, unconstitutional dilution exists where a minority can demonstrate a lack of access to the process of slating candidates, the unresponsiveness of legislators to their particularized interest, a tenuous state policy underlying the preference for multimember or at-large districting, or the existence of past discrimination in general precludes the effective participation in the election system. . . . Such proof is enhanced by a showing of the existence of large districts, majority vote requirements, anti-single shot voting provisions and the lack of provision for at-large candidates running from particular geographic subdistricts. (p. 1305) The Zimmer Court went on to explain that not all of the criteria need be satisfied in a successful dilution claim; "the fact of dilution is established upon proof of the existence of an aggregate of these factors" (p. 1305). This formula for testing multimember districts for their possible dilutive effect became known as the Zimmer test, or the "totality of the circumstances doctrine." 9 From 1973 until 1980, the growing number of legal challenges to multimember or at-large election systems in the South were decided primarily on the basis of this dilution standard, particularly in the Fifth Circuit, where the vast majority of such cases were tried. As minority vote dilution litigation progressed during the decade, the court refined its standards of proof under the Zimmer analysis. For example, evidence of racial bloc voting became necessary to a successful dilution challenge in the Fifth Circuit because of the logical presumption that black voters were not disadvantaged if whites were not voting against the minority's candidate of choice (see Nevett v. Sides, 1978). As thus refined, the Whitcomb-White-Zimmer approach continued to be utilized by the courts, with its supporters asserting that it provided a "flexible, fact-specific, precise and workable" standard (Parker, 1983, p. 725). However, in 1980 this approach to vote dilution was dramatically altered by the Supreme Court in City of Mobile v. Bolden. In this decision, the Court rejected the evidentiary standards developed in Zimmer and declared that proof of discriminatory intent was required for plaintiffs to prevail in a constitutional vote dilution claim. The Supreme Court establishes the need to prove discriminatory purpose in Mobile v. Bolden City of Mobile v. Bolden (1980) was a classic Zimmer type of suit. Mobile operated under a commission form of government in which three city commis-

The evolution of a vote dilution standard

35

sioners were elected at-large within the city. Even though the population of Mobile was 35 percent black, no black candidate had been elected to a commission seat since the system had been adopted in 1911. Employing a Zimmer analysis, the district court found: (1) a history of racial discrimination, (2) city commissioners who were unresponsive to minority interests, and (3) a number of suspect voting practices in addition to at-large voting (e.g., the absence of any district or ward residency requirements, a place system, and a majority vote requirement). The district court concluded that the plaintiffs "met the burden cast in White and Whitcomb by showing an aggregate of the factors catalogued in Zimmer" and therefore held that the at-large election scheme "results in an unconstitutional dilution of black voting strength" (p. 402). The Fifth Circuit sustained the lower court's decision that black voting strength was diluted in violation of blacks' right to vote under the Fifteenth Amendment, their right to equal protection under the Fourteenth Amendment, and also the prohibition against any denial or abridgment of the right to vote contained in Section 2 of the Voting Rights Act. The Supreme Court reversed and rejected the claim that the at-large election of the city commission in Mobile impermissibly diluted black voting strength. The Supreme Court's decision in Bolden emerged from a sharply divided court, with six justices writing separately. The six-person majority overturning the lower courts' decisions consisted of four justices - Justices Potter Stewart, Lewis F. Powell, Jr., and William H. Rehnquist and Chief Justice Warren E. Burger who joined in a plurality opinion; Justice John Paul Stevens, who concurred in the judgment; and Justice Harry A. Blackmun, who concurred only in the result. Justice White wrote a dissenting opinion, and Justices Thurgood Marshall and William Brennan joined in a separate dissent. The plurality in Bolden held that a voting rights suit grounded on the Fourteenth Amendment must demonstrate that a questioned practice or law was established or maintained with racially discriminatory intent.10 Justice Stewart, who delivered the plurality opinion, asserted that "only if there is purposeful discrimination can there be a violation of the Equal Protection Clause of the Fourteenth Amendment" (p. 66). For plaintiffs to prevail in a challenge to an at-large voting scheme, "an illicit purpose must be proved" (p. 67). In addressing the Fifteenth Amendment challenge, Justice Stewart argued that the Fifteenth Amendment protects only against discriminatory voting practices that directly prohibit minorities from gaining access to the ballot, and because "Negroes in Mobile 'register and vote without hindrance,' " no Fifteenth Amendment violation existed (p. 65). (A majority of five justices, however, disagreed with the plurality on this issue and concluded that the Fifteenth Amendment also protects against vote dilution.) In any case, according to the plurality, proof of "racially discriminatory motivation is a necessary ingredient of a Fifteenth Amendment violation" (p. 62).

36

Minority Representation and Voting Equality

As for the statutory challenge, the plurality contended that Section 2 of the Voting Rights Act "was intended to have an effect no different from that of the Fifteenth Amendment itself" and therefore "adds nothing to the appellees' Fifteenth Amendment claim" (p. 61). There was no clear majority on the issue of whether Section 2 required proof of discriminatory purpose. Justices Marshall and Brennan, although agreeing with Justice Stewart that the statutory prohibition in Section 2 "contains the same standard as the Fifteenth Amendment" (p. 105), thought that proof of discriminatory intent was not required to demonstrate Fifteenth Amendment unconstitutionality. Justices Blackmun, Stevens, and White, in their separate opinions, did not discuss the Section 2 question. However, "because six Justices agreed that the scope of section 2 was the same as that of the fifteenth amendment, the lower courts generally considered themselves bound by the Stewart plurality's ruling that section 2 requires proof of discriminatory purpose" (Parker, 1983, p. 733). n Having restricted the dilution issue to Fourteenth Amendment grounds, the plurality proceeded to launch an assault on the Zimmer factors that had governed the lower courts' consideration of the case. The Court held that although these factors "may afford some evidence of a discriminatory purpose," they "were most assuredly insufficient to prove an unconstitutionally discriminatory purpose in the present case" (p. 73). Drawing inferences of discriminatory intent based on such circumstantial evidence was clearly not enough; minority voters had to prove that a challenged voting law was adopted " 'because of,' not merely 'in spite of,' its adverse effects" (p. 72). By sharply limiting the inferences that the lower courts were allowed to draw from the evidence and by treating each of the facts in isolation rather than as part of a whole, the Mobile court rendered discriminatory impact largely irrelevant and essentially overturned the WhiteZimmer "totality of the circumstances" test. Justice White, author of the White decision, dissented from the plurality, calling the plurality's opinion "flatly inconsistent" with White. He adopted the Fifth Circuit's view that the Zimmer factors provided inferences of discriminatory purpose. Justices Marshall and Brennan, in an opinion delivered by Justice Marshall, issued a vociferous dissent in which they asserted that the Fourteenth and Fifteenth amendments outlawed all laws that, intentionally or otherwise, lessened the impact of every citizen's vote; thus, according to Marshall, "vote-dilution decisions require only a showing of discriminatory impact to justify the invalidation of a multimember districting scheme" (p. 104). Rather than having judges take "an unguided, tortuous look into the minds of officials in the hope of guessing why certain policies were adopted and others rejected," he proposed a "disproportionate-impact test" that would "reach only those decisions having a discriminatory effect upon the minority's vote" (p. 134). Even accepting the plurality's premise, however, the two jus-

The evolution of a vote dilution standard

37

tices found that the plaintiffs had met the burden of proving discriminatory purpose. Essentially, the Supreme Court plurality established in Bolden a standard of proof that required a "smoking gun" - that is, specific evidence that a particular election practice was rooted in racist intent, as revealed in the statements or actions of public officials. This standard is very difficult to meet, and though not foreclosing vote dilution suits altogether, constituted a "severe setback," by making success in such actions much less likely (Avila, 1982). Indeed, the decision had an "immediate and devastating impact" on litigation challenging discriminatory dilution of minority voting strength (Parker, 1983, p. 735); cases before the courts were "overturned and dismissed," and "plans for new cases were abandoned" (Derfner, 1984, p. 149). The lower courts frequently rejected challenges to at-large voting systems for insufficient evidence of discriminatory intent, even in instances in which minority voters were completely shut out of the political processes (Parker, 1983, p. 736). In those cases in which constitutional challenges were sustained, direct "smoking gun" evidence of discriminatory intent had been produced (Parker, 1983, p. 737).12 The Mobile decision aroused a storm of protest, particularly within the civil rights community. A number of criticisms were leveled: The decision broke with precedent in previous vote dilution cases; the Court ignored the fact that voting was considered a "fundamental" right and therefore that any curtailment of that right must be subject to "strict scrutiny"; and in determining that proof of discriminatory intent is necessary, the Court asked "the wrong question" and established a burden of proof that was almost impossible for plaintiffs to meet. Perhaps the most thorough critique of the decision was presented by Marshall in his dissent. He argued that the plurality decision failed to reflect the traditional solicitude for the right to vote: The right to vote is a fundamental right "because it is 'preservative of all rights'" (quoting Yick Wo v. Hopkins, 1886, p. 370), and it is therefore subject to "strict judicial scrutiny," according to Marshall (p. 113). Proof of discriminatory purpose is clearly not necessary (p. 114). Furthermore, the Bolden plurality opinion failed to advance a reason for its differential treatment of a Reynolds type of vote dilution and minority vote dilution in making discriminatory intent the keystone of the latter and not the former. (Population malapportionment does not require proof of invidious purpose; mathematical inequality alone is deemed sufficient to support a claim of dilution.) Even if the Court had been successful in distinguishing the malapportionment dilution suits from racial vote dilution, argued Marshall, Bolden still broke with precedent in past vote dilution cases: "The plurality's approach requiring proof of discriminatory purpose in the present cas[e] is . . . squarely contrary to White and its predecessors" (p. 112). Although the plurality contended that White required a showing of discriminatory intent, their argument

38

Minority Representation and Voting Equality

was unconvincing, according to Marshall. He argued that not a single word in the White decision is devoted to the possible motivations of the Texas legislature and that the Court examined only "the discriminatory effects of the combination of an electoral structure and historical and social factors" (p. I l l , n. 7). Critics of the decision also contended that Bolden established an exceedingly burdensome and irrelevant evidentiary standard that would make it nearly impossible to win suits against racially discriminatory electoral systems. Proving discriminatory intent would be impossible because "legislators cannot be subpoenaed from their graves for testimony about the motives behind their actions" (U.S. Senate, 1982, p. 36) in the case of a law passed more than a decade or so ago.13 A legislative record that would reveal the reasons for adopting the election practice is unlikely to exist in most communities. Even if such a record does exist, legislators may well have sought to offer a nonracial rationalization for a law that in fact purposefully discriminates. Furthermore, the motives of officials acting "100 years ago is of the most limited relevance" with regard to whether the electoral system is operating to exclude minority groups today (U.S. Senate, 1982, p. 36).14 In an effort to minimize the damage to the voting rights movement inflicted by Bolden, civil rights advocates focused on Section 2 as a medium for "reversing" the decision. Although the Supreme Court's decisions on the meaning of the Constitution are final, Congress is free to supersede the Supreme Court's interpretation of a statute by passing an amendment or a new statute spelling out in clearer language how that statute is to be interpreted. By amending Section 2 to make evidence of discriminatory effects sufficient to establish a statutory violation of voting rights, a statutory bypass around the Bolden decision would be created. The civil rights community began to lobby Congress to adopt a "results" test in Section 2, so as to provide a separate statutory standard for weighing dilution claims - as opposed to the constitutional standard set out in Mobile. It was hoped that Congress could be persuaded to modify the evidentiary standard for vote dilution set out by the Court and "restor[e] the legal standards . . . which applied in voting discrimination claims prior to . . . Bolden" (U.S. Senate, 1982, p. 2). Congress dictates a vote dilution standard: the 1982 amendments to the Voting Rights Act In one sense the Supreme Court's decision in Mobile v. Bolden (1980) was well timed. The special provisions of the Voting Rights Act, having been extended for seven years in 1975, were due to expire in August 1982. The need to consider extending the Voting Rights Act by that date offered critics of the decision an opportunity to formulate a statutory bypass around it. Civil

The evolution of a vote dilution standard

39

rights advocates began an extensive lobbying campaign to persuade Congress to replace the intent standard established by the Court in Bolden with a "results test." Although Section 2 was not a special provision of the act and was not due to expire, it provided a handy vehicle for legislative modification of the Bolden intent standard. A simple rewording would eliminate the need to prove purposeful discrimination in a vote dilution case, provided that the plaintiffs alleged a Section 2 violation rather than (or in addition to) a constitutional violation. The legislative history of the Section 2 revision, which began in May 1981, with hearings before the House Subcommittee on Civil and Constitutional Rights, is detailed elsewhere (Handley, 1991, chap. 4). After lengthy and sometimes acrimonious debate in both chambers, revisions to the act were passed by Congress on June 23, 1982, and signed by President Ronald W. Reagan on June 29. Section 2 now reads: (a) No voting qualification or prerequisite to voting, or standard, practice, or procedure shall be imposed or applied by any State or political subdivision in a manner which results in a denial or abridgement of the right of any citizen of the United States to vote on account of race or color, or in contravention of the guarantees set forth in section 4(f)(2). (b) A violation of subsection (a) if, based on the totality of circumstances, it is shown that the political processes leading to nomination or election in the State or political subdivision are not equally open to participation by members of a class of citizens protected by subsection (a) in that its members have less opportunity than other members of the electorate to participate in the political process and to elect representatives of their choice. The extent to which members of a protected class have been elected to office in the State or political subdivision is one circumstance which may be considered; Provided, that nothing in this section establishes a right to have members of a protected class elected in numbers equal to their proportion in the population. In addition to the revision made to Section 2, the new law extended the special provisions of the act for an additional twenty-five years and provided a new procedure under which jurisdictions subject to special coverage can win exemption or "bail out" from the act.15 Other provisions of the 1982 amendments included the extension until 1992 of bilingual assistance provisions and a new provision that established the right of a blind, disabled, or illiterate voter to "assistance by a person of the voter's choice," other than the voter's employer or union agent. The legislative history of the Section 2 amendment provided extensive guidance as to the intent of Congress concerning how the courts were to apply the new results standard. A report of the Senate Judiciary Committee states that this amendment is designed to make clear that proof of discriminatory intent is not required to establish a violation of Section 2. It thereby restores the legal standards,

40

Minority Representation and Voting Equality

based on the controlling Supreme Court precedents, which applied in voting discrimination claims prior to the litigation involved in Mobile v. Bolden. (U.S. Senate, 1982, p. 2) According to the report, requiring proof of purposeful discrimination is inconsistent with the original legislative intent of Section 2 as well as court precedent prior to the Bolden decision. Courts applying the statutory results test need not examine the decision makers' intent in creating or maintaining challenged electoral schemes, and they cannot rely solely on the degree to which minority groups are represented on legislative bodies in making their determination. Instead, the courts are to consider such typical factors, derived from "the analytic framework used by the Supreme Court in White, as articulated in Zimmer" (U.S. Senate, 1982, p. 28), as are listed in the Senate report. Proof of a violation of Section 2 in a given jurisdiction might include "a variety" of the following factors: 1. the extent of any history of official discrimination . .. that touched the right of the members of the minority group to register, to vote, or otherwise to participate in the democratic process; 2. the extent to which voting in elections . . . is racially polarized; 3. the extent to which the state or political subdivision has used unusually large election districts, majority vote requirements, anti-single shot provisions, or other voting practices or procedures that may enhance the opportunity for discrimination against the minority group; 4. if there is a candidate slating process, whether the members of the minority group have been denied access to that process; 5. the extent to which members of the minority group . . . bear the effects of discrimination in such areas as education, employment and health, which hinder their ability to participate effectively in the political process; 6. whether political campaigns have been characterized by overt or subtle racial appeals; 7. the extent to which members of the minority group have been elected to public office in the jurisdiction. (U.S. Senate, 1982, pp. 28-29) The report indicates two additional factors that may have some probative value: "whether there is a significant lack of responsiveness on the part of elected officials to the particularized needs of the members of the minority group" and whether the policy underlying the challenged electoral practice is "tenuous." The Senate committee report makes clear that although the factors it lists are often the most relevant considerations, they are not exclusive and therefore not controlling. A litigant need not prove all, or even a majority, of the factors to establish a Section 2 violation: "There is no requirement that any particular number of factors be proved, or that a majority of them point one way or another" (U.S. Senate, 1982, p. 29). The totality of the circumstances approach derived from White and applied by the Fifth Circuit Court in Zimmer is the

The evolution of a vote dilution standard

41

approach that should be employed by the courts to determine whether "the voting strength of minority voters is . . . 'minimized or canceled out'." The Supreme Court revisits the intent test of Bolden On July 1, 1982, two days after the Voting Rights Act extension bill was signed into law by President Reagan, the Supreme Court retreated somewhat from its Bolden ruling, holding that "circumstantial evidence" could be used to determine whether an electoral system was created or maintained for the purpose of racial discrimination. In Rogers v. Lodge, the Court affirmed the Fifth Circuit Court of Appeals decision that the at-large county commission elections in Burke County, Georgia, diluted the voting strength of blacks in violation of the Fourteenth Amendment's guarantee of equal protection. In this six-to-three decision, the Court backed away from a strict "intent" test promulgated less than eighteen months earlier in Bolden. Specifically, the Court held that "discriminatory intent need not be proven by direct evidence" (p. 618) and that proof of the White-Zimmer factors was sufficient to show discriminatory purpose. The Court adhered, however, to its position that proof of discriminatory purpose was required. A holding of unconstitutional vote dilution in Rogers is difficult to reconcile with Bolden because the evidence in the two cases is so similar. (For details, see Handley, 1991, pp. 146-147.) As it happened, most of the justices thought the two cases did, in fact, require the same decisions. The four justices who voted to strike down the at-large election system in Bolden also voted to strike it down in Rogers, whereas three of the justices who voted to uphold the Mobile City Commission also voted to uphold the county commission in Burke County. The reasons for the contrary ruling rest with Chief Justice Burger, who "switched" his vote from "constitutional" for Mobile, to "unconstitutional" for Burke County, and Justice Sandra Day O'Connor, who voted to hold the Burke County system unconstitutional, whereas her predecessor, Justice Stewart, had voted to uphold Mobile's system. Although the majority opinion implicitly cast doubt on the continuing validity of the Bolden ruling, it did not overrule Bolden. Justice White, the author of the Court's 1973 decision in White and a dissenter in Bolden, wrote for the majority in Rogers. In his opinion he attempted to distinguish the two cases; the difference between Rogers and Bolden, according to Justice White, was the standard of proof used by the lower court to determine whether a constitutional violation of voting rights was present. The district court in Bolden had relied on the White-Zimmer factors without considering whether they established intent, but the trial court in Rogers had used such White-Zimmer factors as the lingering

42

Minority Representation and Voting Equality

effects of past discrimination on black political participation, the depressed socioeconomic status of blacks, blacks' historical defeat at the polls, and the unresponsiveness of public officials to black political interests to infer discriminatory purpose.16 The majority's acceptance of the probative value of evidence organized around the Zimmer factors substantially eases the burden of proving intent by allowing the courts to draw inferences of discriminatory intent from circumstantial evidence. However, an ultimate finding of intentional discrimination, even if based only on the circumstantial and inferential factors of White-Zimmer, is still required in order to prove unconstitutional vote dilution. Thus, the Rogers ruling does not do away with the requirement of proving intent to discriminate in order to establish a constitutional violation. Rogers is now the controlling constitutional precedent, but what the courts will make the intent standard after Rogers is difficult to foresee. Although it is clear that the decision signals a retreat from the Bolden plurality's minority vote dilution analysis and a revitalization of the Zimmer factors, the courts have had no further opportunity to develop Fourteenth (or Fifteenth) Amendment theories of minority vote dilution. Since the Rogers venture into deciding a dilution case under the Constitution, the Supreme Court has insisted that other vote dilution cases be decided under Section 2, and it has sent several cases back to the lower courts for that purpose (Derfner, 1984, p. 161). The standard for determining the presence of unconstitutional vote dilution may therefore be a moot issue.17 What is important is what the courts have made of the results standard set by Section 2. The lower courts consider amended Section 2 After the passage of the 1982 amendments to the Voting Rights Act, it was left to the courts to grapple with the problem of developing appropriate guidelines for judging a vote dilution claim based on the results test of Section 2. Beyond endorsing the totality of the circumstances test announced in White v. Regester and outlining a number of "typical" factors to be considered indicative of vote dilution, Congress left future interpretation of amended Section 2 to the courts. The Fifth Circuit was the first court of appeals to consider a vote dilution claim on the basis of the newly amended Section 2. In Jones v. City ofLubbock (1984), African-American and Hispanic residents of Lubbock, Texas, challenged the city's at-large system for electing city council representatives on both constitutional and Section 2 grounds. The Fifth Circuit in Jones upheld the constitutionality of newly amended Section 2, affirmed the district court's decision that the city's election system violated Section 2, and outlined a standard for determining the presence of illegal vote dilution based on the nine factors out-

The evolution of a vote dilution standard

43

lined in the Senate Report. The court reviewed the evidence it found persuasive in concluding that Lubbock's electoral system violated Section 2: the history of official discrimination against blacks and Hispanics and evidence of its lingering effects in Lubbock in the socioeconomic disparities between Anglos and minorities and in depressed minority political participation; an electoral system that featured every structural impediment "that courts have identified as aggravating the impact of an at-large election system" (p. 385); and evidence of persistent racial polarization, such that it was unnecessary for candidates and elected public officials to seek minority political support. Indicating that there was no requirement that any specific number of factors be proven in order to establish a violation of Section 2, the Fifth Circuit nevertheless stressed the importance of a finding of racial polarization, stating that "the legislative discussion of polarized voting requires that we weigh more carefully the effect that polarization has on the political scheme challenged" (p. 385). Furthermore, the appellate court rejected the city's argument that the responsiveness of city officials to the needs of minorities undercuts a finding of racially polarized voting: "The absence of unresponsiveness does not negate other inferences that flow from polarization. Wheth r or not city officials do ignore minority interests, polarization nevertheless frees them of political reprisal for disadvantaging the minority community" (p. 381). In United States v. Marengo County Commission (1984), the Eleventh Circuit Court of Appeals devised a Section 2 vote dilution standard very similar to the one outlined by the Fifth Circuit in Jones, emphasizing the importance of racial polarization even more than the Fifth Circuit had in Jones. In Marengo County, the court reviewed a vote dilution challenge to an at-large scheme for electing county commissioners and school board members in Marengo County, Alabama. The district court, ruling before Congress amended Section 2, upheld the atlarge scheme because plaintiffs had failed to prove discriminatory purpose. On appeal, the Eleventh Circuit held that the newly revised Section 2 applied to the case, found compelling evidence that the at-large election system violated Section 2, and remanded the case to the district court for an update of the record. Accompanying the remand was an explanation of how the Section 2 results test should be applied to an allegation that an at-large election system unlawfully dilutes minority votes. Judge John Minor Wisdom, writing for a unanimous three-judge panel of the Eleventh Circuit, reviewed the legislative history of Section 2 and determined that racially polarized voting "will ordinarily be the keystone of a dilution case" (p. 1566). He reasoned that Section 2 is intended "to remedy it [race-conscious politics] where it already exists" and that "the surest indication of raceconscious politics is a pattern of racially polarized voting" (p. 1567). Although stressing the importance of racial bloc voting, the Eleventh Circuit also discussed

44

Minority Representation and Voting Equality

at some length the evidence presented with regard to each of the other "typical" factors identified in the Senate Report: an undisputed record of past discrimination; considerable social and economic differences between blacks and whites; structural elements, in addition to at-large elections, that were suggestive of vote dilution (staggered terms, numbered posts, and a majority vote requirement in the Democratic primary); and the fact that no black had ever been elected to the county commission or the school board, even though approximately half of the county's population was black. The Eleventh Circuit also briefly discussed the two "additional" factors listed in the Senate Report. After finding that the state policy underlying the at-large requirement was tenuous (the state enacted the at-large system in 1955 "in direct response to the prospect of increased black political participation" [p. 1571]), the court explained that the "tenuousness" factor, although less important under the results test than under the intent standard, is still relevant because "evidence that a voting device was intended to discriminate is circumstantial evidence that the device has a discriminatory result" (p. 1571). The second "additional" factor - the issue of responsiveness - was also deemed less important. "Unresponsiveness would be relevant only if the plaintiff chose to make it so," because "although a showing of unresponsiveness might have some probative value a showing of responsiveness would have very little" (p. 1572). The court found that although the Marengo County Commission may be responsive to the minority group's material needs (roads, sewers, etc.), the evidence indicated that it was not responsive to the group's political needs - the desire to elect candidates of its choice - which is, of course, the focus of the Voting Rights Act. Reiterating the reasoning articulated in Marengo County, the Eleventh Circuit Court in United States v. Dallas County Commission (1984), and later the Fifth Circuit Court in McMillan v. Escambia County (1984), utilized a hierarchy of factors in which racially polarized voting was preeminent. In Dallas County, a Section 2 challenge to the at-large election system for the county commission and the county school board in Dallas County, Alabama, the Eleventh Circuit discussed all nine factors listed in the Senate Report but stressed the importance of racial polarization as the keystone of a dilution case. The appellate court determined that the district court had erred in its assessment of racially polarized voting in Dallas County. Although the district court had found evidence of racial polarization,18 it discounted its effect, emphasizing several factors it contended counteracted a finding of racial bloc voting: Some of the blacks who ran were fringe-party candidates; blacks did not actively seek the votes of whites; whites often supported white incumbents over black candidates; and there was apathy among black voters. The appeals court, however, concluded that none of these factors was sufficient to override a finding of racial polarization. The case was remanded to the district court for further consideration. In McMillan, plaintiffs challenged the at-large system for electing county com-

The evolution of a vote dilution standard

45

missioners in Escambia County, Florida. The Fifth Circuit Court of Appeals, in an opinion that relies heavily on Marengo County, affirmed the district court's finding of a violation of Section 2. This conclusion was based on a finding of racial polarization combined with evidence of a history of past discrimination and indications of its lingering effects as well as such structural impediments as staggered terms, numbered posts, and a majority vote requirement in the Democratic primary. According to the Fifth Circuit, racial polarization occurs "whenever a black challenges a white for county wide office, a consistent majority of the whites who vote . . . consistently vote for the black's opponent" (p. 1043). Statistical evidence pointing to a "consistent pattern of racially polarized voting" (p. 1040) and the fact that no black had ever been elected to office - despite the fact that blacks comprised 17 percent of the registered voters in the county - led the court to conclude that voting in Escambia County was polarized. The Eleventh Circuit adopted a somewhat different approach to racial bloc voting in Lee County Branch of NAACP v. City of Opelika (1984), a challenge to the at-large scheme for electing the city commission in Opelika, Alabama. The appellate court remanded the case to the district court for additional evidence regarding the question of racially polarized voting. Noting that under the Marengo County standard, racial bloc voting is a key consideration, the court found the evidence presented at trial insufficient to make a determination as to the existence of racially polarized voting in the Opelika city elections. The Opelika decision parts company with the reasoning set forth in Marengo County, because the majority in Opelika argued that evidence of polarized voting requires more than a showing of divergent voting patterns between blacks and whites. Evidence that race is the motivating factor in disparate voting patterns is necessary, according to the Opelika court: It will often be necessary to examine factors other than race that may also correlate highly with election outcomes - campaign expenditures, party identification, income, media advertising, religion, name recognition, position on key issues, and so forth. Wellestablished statistical methods, such as step-wise multiple regressions, can test the relative importance of multiple factors. Such analysis can assist in the determination of whether race is the dominating factor in political outcomes (p. 1482). Thus, the majority of the Eleventh Circuit in this case rejected the use of a bivariate statistical analysis that simply regresses the racial composition of the precincts against the precinct votes for a given candidate in favor of a multivariate analysis that measures the relative importance of a number of factors in addition to the race of the voters.19 Other appeals court decisions emulated the Opelika approach to the question of racially polarized voting. The Court of Appeals for the Fourth Circuit in Collins v. City of Norfolk (1985) not only condoned the use of a multivariate statistical analysis to determine whether disparate voting patterns were "racially

46

Minority Representation and Voting Equality

motivated," but it also let stand a unique definition of racial bloc voting - one that made it virtually impossible for plaintiffs to prevail. In Collins, a Section 2 challenge to the at-large method of electing the Norfolk City Council, the district court rejected as "overly broad" the definition offered by the plaintiffs' expert witness, Dr. Richard Engstrom, that "racially polarized voting occurs when the majority of one racial group votes contrary to the majority of another racial group" (1984, p. 386) and accepted instead the definition offered by the defendants' expert witness, Dr. Timothy O'Rourke. According to Dr. O'Rourke, three indicators must be examined before racial polarization can be said to exist: First, the presence or absence of "white backlash," that is, whether white voters turn out in greater numbers than usual in response to the potential election of black candidates. Second, the voting patterns of black and white voters over a period of years. Finally, whether whites attempt to limit the field of candidates, (p. 386) The district court adopted the defendant's approach to defining and measuring racially polarized voting as "more sensible" and "more applicable to the multiple-candidate, multiple-office" elections in Norfolk (p. 386) and, using this evidentiary test for racial polarization, found that the Norfolk City Council elections did not suffer from racial polarization. Another distinctive feature of the district court in Collins was its approach to another "typical" factor listed in the Senate Report: the extent of any history of official discrimination. The district court discounted the egregious history of official racial discrimination in voting practices in Norfolk and in the state of Virginia, by finding that no lingering effects of that discrimination now inhibited blacks from fully participating in the electoral process. This conclusion was based on the district court's observation that since the Voting Rights Act was enacted in 1965, black political participation had increased dramatically, to the point that the registration and turnout of blacks slightly exceeded that of whites in the 1984 elections. The lower court in Collins held that black plaintiffs in Norfolk had failed to prove a Section 2 violation: With neither racially polarized voting nor the lingering effects of discrimination depressing political activity, blacks in Norfolk were simply not entitled to any relief. The Fourth Circuit affirmed the decision of the lower court with essentially no discussion of the controversial approach taken by the district court.20 The Eleventh Circuit in McCord v. City of Fort Lauderdale (1986) closely paralleled the approach adopted by the Fourth Circuit in Collins. In McCord, the Eleventh Circuit was asked to consider a vote dilution claim by blacks in Fort Lauderdale, Florida, alleging that the at-large election system violated Section 2. The appeals court upheld the district court finding that there was no Section 2 violation. In reaching this decision, the Eleventh Circuit argued that

The evolution of a vote dilution standard

47

it was relying on the guidelines it had previously outlined in Marengo County, but what distinguished McCord from Marengo County was the absence of any lingering effects of past discrimination. Citing Collins, the court observed that although the city had a history of discrimination and plaintiffs offered evidence that racial discrimination still existed, voting rights were unaffected because "blacks in Fort Lauderdale since 1970 have turned out to vote in higher numbers on a percentage basis than white voters in every election but one" (p. 1531). In addition, the Eleventh Circuit was not persuaded that the district court's finding concerning a lack of racially polarized voting was clearly erroneous. The district court had relied on a multivariate analysis presented by defendants that showed that "when race was added to other independent variables . . . the race factor explained only .6 of one percent of the dependent variables [sic] of candidate success" (p. 1532). The appeals court, citing Opelika, recognized the "potential usefulness of a multiple regression analysis" in determining whether voting is racially polarized. As is evident by a comparison of such early Section 2 cases as Jones, Marengo County, and McMillan with the later cases of Collins and McCord, there was a decided lack of consensus among the courts on the standards to be applied when weighing an allegation of vote dilution, especially with respect to the definition of racial polarization and how it should be measured. The Supreme Court addressed this issue, as well as other important Section 2 issues, for the first and, to date, the only time in Thornburg v. Gingles (1986). The Supreme Court addresses Section 2: Thornburg v. Gingles When the Supreme Court granted review of Gingles, observers anticipated a clarification of the vote dilution standard outlined in the legislative history of Section 2. And in its decision, the Court did shed light on all of the major issues, including the role of racial bloc voting in a vote dilution suit. The Court outlined the specific factors that must be proved in order to demonstrate that the use of multimember districts impairs the ability of minority voters to elect representatives of their choice, thereby creating a three-part test for judging vote dilution cases. Even though several concurring opinions were filed in the case, the Court provided reasonably definitive answers to many of the questions raised in the Fourth, Fifth, and Eleventh circuits, particularly with regard to the proper evidentiary standard to be used in assessing the degree of racial bloc voting. However, its opinion also raised new issues. The suit originated as Gingles v. Edmisten (1984) before a three-judge district court in North Carolina. In the spring of 1982, North Carolina enacted a legislative redistricting plan for the state's senate and house seats. Black registered voters brought suit, challenging one single-member and six multimember dis-

48

Minority Representation and Voting Equality

tricts under Section 2 of the Voting Rights Act. Plaintiffs contended that the plan diluted minority voting strength by submerging concentrations of black voters sufficient in size to form a majority of the voters in a single-member district in majority-white multimember districts and by fragmenting into more than one state senate district "a concentration of black voters sufficient in numbers and contiguity to constitute a voting majority in at least one single-member district" (p. 350). The district court relied heavily on the legislative history of the amendment, carefully examining each of the Senate Report factors as they applied to North Carolina and determining that most of them were present to one degree or another. The court recounted North Carolina's history of official racial discrimination and the lingering effects this history had on black registration and voting. Discrimination in other areas, such as education and employment, resulted in a lower socioeconomic status for blacks as a group, which in turn " operate [d] to hinder the group's ability to participate effectively in the political process" (p. 363). The majority vote requirement was viewed by the court as a "general, ongoing impediment" to the election of black candidates (p. 363). Racial appeals in election campaigns in North Carolina were found to be widespread, persistent, and as recent as the 1984 campaign for a U.S. Senate seat. Although blacks had been elected to office at most levels of government in North Carolina (with the exception of any statewide office, or to the U.S. Congress), the court determined that the overall results were minimal in relation to the percentage of blacks in the total population. Finally, and most importantly, the court found evidence of "persistent and severe" racial polarization based on a statistical analysis of fifty-three elections involving black candidates in all of the challenged districts (p. 367). The court asserted that of the Senate Report factors, the "demonstrable unwillingness of substantial numbers of the racial majority to vote for any minority race candidate or any candidate identified with minority race interests is the linchpin of vote dilution by districting" (p. 355). The district court concluded on the basis of the totality of the circumstances that blacks had less opportunity than others to participate in the political process and elect representatives of their choice in the challenged districts and unanimously upheld the plaintiffs' Section 2 challenge. The state of North Carolina appealed the district court's holding in regard to five of the multimember districts.21 On direct appeal to the Supreme Court, the state of North Carolina and the United States as amicus maintained that the lower court erred in concluding that there was a violation of Section 2. According to North Carolina, the lower court (1) had incorrectly defined legally significant racially polarized voting and had

The evolution of a vote dilution standard

49

adopted an erroneous standard for measuring the degree of polarization and (2) did not give sufficient weight to the electoral success of some black candidates. On June 30, 1986, the Supreme Court unanimously upheld the findings of the district court for four of the five contested multimember state legislative districts involved in the suit. However, the Court was not in unanimous agreement on the legal analysis to be applied in vote dilution cases or in the evidentiary standard to be used by the court in measuring the degree of racially polarized voting, the "linchpin" of a vote dilution claim. There were actually four separate opinions filed in Gingles. Justice Brennan wrote the five-part opinion of the Court, joined by Justices Blackmun, Marshall, Stevens, and White. There were three concurring opinions. Justice White disagreed with Justice Brennan with regard to a specific section of Part III-C of the opinion and filed a separate opinion on the point of contention. Justice O'Connor, joined by Chief Justice Burger and Justices Powell and Rehnquist, filed an opinion concurring only with the judgment of the Court and disagreeing with almost the entirety of Justice Brennan's opinion. Justice Stevens, joined by Justices Marshall and Blackmun, dissented from the part of the opinion that reversed the district court's finding of vote dilution in House District 23. The majority opinion: a three-pronged test for vote dilution The majority opinion affirmed the decision of the district court with respect to four of the five contested districts and established a three-pronged test for analyzing vote dilution claims involving multimember districts. The opinion also provided a definition of legally significant racially polarized voting. Justice Brennan reviewed the nine factors outlined in the Senate Report accompanying the amendments to Section 2 and reasoned that although all of the factors listed may be relevant to a vote dilution claim challenging a multimember districting scheme, they will not be dispositive unless plaintiffs can first show a conjunction of three circumstances. The Court indicated: These circumstances are necessary preconditions for [a violation]. . . . First, the minority group must be . . . sufficiently large and geographically compact to constitute a majority of a single-member district. . . . Second, the minority group must be . . . politically cohesive. . . . Third, . . . the white majority votes sufficiently as a bloc to enable it . . . usually to defeat the minority's preferred candidate, (pp. 50-51) If the minority group is not sufficiently large enough or geographically compact enough to form a majority in a single-member district, then the fact that the district is structured as multimember is irrelevant to the minority's lack of ability to elect candidates of choice. According to the Court: "Unless minority voters possess the potential to elect representatives in the absence of the chal-

50

Minority Representation and Voting Equality

lenged structure or practice, they cannot claim to have been injured by that structure or practice" (p. 50). If the minority group is not politically cohesive, then there is no distinct minority interest to protect, according to the Court. Finally, the white majority must vote as a bloc to enable it usually to defeat the minority's preferred candidate. The Court recognized that it is the fact that the minority's preferred candidates usually lose that "distinguishes structural dilution from the mere loss of an occasional election" due to nondilutive factors (p. 51). The majority's three-part test establishes that the degree of racial bloc voting is the key element of a vote dilution claim. In emphasizing the preeminence of this factor, Justice Brennan echoed the reasoning first offered by the Fifth and Eleventh circuits in Jones, Marengo County, Dallas, and McMillan: Unless there is an initial showing of significant racial bloc voting, the other factors will not demonstrate that the plaintiffs have suffered a substantial inability to elect their preferred candidates. The majority accepted the definition of racially polarized voting adopted by the district court and presented by plaintiffs' expert (Grofman) that racially polarized voting exists "where there is 'a consistent relationship between race of the voter and the way in which the voter votes' . . . or to put it differently, where 'black voters and white voters vote differently' " (p. 53, n. 21). By accepting the district court's definition of racial bloc voting, the Court rejected the more stringent definition provided by the defendants and accepted by the lower courts in Collins and McCord. The purpose of investigating whether or not voting is racially polarized is twofold, according to the Court: "to ascertain whether the minority group members constitute a politically cohesive unit and to determine whether whites vote sufficiently as a bloc usually to defeat the minority's preferred candidates" (p. 56). According to the majority, legally significant racial bloc voting (i.e., the degree of bloc voting necessary to prove a claim) occurs when a white bloc voting majority is usually able to defeat candidates supported by a politically cohesive, geographically insular minority group. By differentiating between racial polarization per se and legally significant racial polarization, the Court recognized that polarized voting does not necessarily lead to electoral loss, and so the critical consideration is whether members of the minority group are able to elect candidates of their choice.22 The Court stated that "there is no simple doctrinal test for the existence of legally significant racial bloc voting" and noted that the degree of polarized voting necessary to show dilution varies by jurisdiction; an intensely local appraisal is necessary to measure whether polarized voting impairs a minority group's ability to elect candidates of their choice (p. 58). The majority, however, outlined a general approach to determining the extent of racial bloc voting,

The evolution of a vote dilution standard

51

noting first that the court must ascertain whether there is usually enough bloc voting by whites to defeat candidates preferred by the minority. This differs depending on the presence or absence of other potentially dilutive devices such as a majority-vote requirement or numbered posts, the size and racial composition of the challenged districts, and the percentage of the voters who are members of the minority group. Second, bloc voting is generally more significant if it occurs over a period of time. Thus, a pattern of voting along racial lines that has existed over several elections is more probative than are the results of a single election. On the other hand, the results of one election in which significant racial bloc voting has not occurred are insufficient to sustain the contention that the jurisdiction is now free from legally significant racial bloc voting. In Part IV of the decision, the majority specifically rejected the argument raised by appellants and the United States that proportional or near proportional minority success in a single election precludes, as a matter of law, a finding of a Section 2 violation. The Court found that both the language of Section 2 and its legislative history indicated that the electoral success of some minority candidates does not foreclose a Section 2 claim. However, in the case of House District 23, in which blacks had achieved proportional representation in each of the last six elections, the majority did conclude that minority success had been substantial enough to reverse the holding of the district court. The plurality opinion: an evidentiary standard for racial bloc voting The extent of racial bloc voting present in a challenged jurisdiction is the "linchpin" of a vote dilution claim based on Section 2 according to the majority opinion. In Part III-C of his opinion, Justice Brennan outlined the standard of statistical evidence necessary to determine the degree of racially polarized voting. Because Justice White did not join this section, this part of the opinion does not command a majority (Part III-C was joined only by Justices Blackmun, Marshall, and Stevens). However, in his separate concurring opinion, Justice White expressed disagreement with only a small portion of Part III-C (Subpart four, which discusses the relevance of the candidate's race); he did not specifically object to the remainder of Part III-C. Justice Brennan advanced his evidentiary standard regarding racially polarized voting by repudiating certain arguments made by the state of North Carolina (and the United States as amicus). The state argued that statistical evidence must demonstrate not only that there is a correlation between race of the voters and their choice of candidates but also that race (as opposed to other factors such as socioeconomic status or party affiliation) is the principal reason for the voters' selections. According to the plurality, however, the proper inquiry under Section 2 is to ask whether voters of different races favor different candidates, not why

52

Minority Representation and Voting Equality

they do so. Exploring the reasons for the relationship between race and votes cast interjects intent into the analysis, and "the legal concept of racially polarized voting incorporates neither causation nor intent," according to Justice Brennan (p. 62). Central to this debate about the reasons that voters cast the votes that they do is whether a court should use a bivariate or multivariate statistical analysis to determine the presence of racially polarized voting. In Jones, Marengo County, and Escambia County, a bivariate analysis was the preferred method; but in Opelika, Collins, and McCord, the courts accepted a multivariate analysis. Justice Brennan held that a bivariate statistical analysis is the proper method by which to determine racial bloc voting, because under Section 2 only the fact that the race of the voters correlates with the selection of certain candidates matters: "It is the difference between the choices made by black and white voters and not the reasons for the differences, that leads to blacks having less opportunity to elect their candidates of choice" (p. 63). In addition, Justice Brennan recognized that many of the other factors that North Carolina sought to introduce into the analysis as nonracial explanations are in fact highly correlated with race, and therefore polarized voting would virtually never be found "whenever the black and white populations could be described in terms of other socioeconomic characteristics" (p. 65).23 Moreover, according to Justice Brennan: We can find no support in either logic or the legislative history for the anomalous conclusion to which the appellants' position leads - that Congress intended, on the one hand, that proof that a minority group is predominately poor, uneducated, and unhealthy should be considered a factor tending to prove a section 2 violation; but that Congress intended, on the other hand, that proof that the same socioeconomic characteristics greatly influence black voters' choice of candidates should destroy these voters' ability to establish one of the most important elements of a vote dilution claim, (p. 67) Similarly, the plurality refused to accept North Carolina's argument that racially polarized voting should be defined as existing only when the white bloc voting is fueled by racial hostility. Plaintiffs do not have to demonstrate that racial animosity is the cause of differences in voting patterns, according to Justice Brennan, but only that the electorate does, as a matter of fact, divide along racial lines. The plurality recognized that the principal reason Congress amended Section 2 was because intent was so difficult to prove (not to mention irrelevant to the issue) - and proving racism on the part of the voters is likely to be an even more burdensome and racially divisive task than is proving racism on the part of legislators. In Part III-C, the plurality also rejected North Carolina's contention that bloc

The evolution of a vote dilution standard

53

voting must be defined with reference to the candidate's race. Justice Brennan argued that the race of the candidate per se is irrelevant to the analysis: The fact that race of voter and race of candidate is often correlated is not directly pertinent to a Section 2 inquiry. Under Section 2, it is the status of the candidate as the chosen representative of a particular racial group, not the race of the candidate, that is important, (p. 68) Thus, according to Justice Brennan, the appropriate inquiry is whether the preferred candidates of the minority group, be they black or white, are usually defeated.24

Justice O'Connors concurrence: the inevitability of proportional representation as the guiding standard Acquiescing only to the final judgment of the Court in Gingles, Justice O'Connor vehemently objected to the majority opinion - particularly the three-pronged test for vote dilution adopted by the Court. In a concurrence joined by the chief justice and Justices Powell and Rehnquist, she challenged the three-pronged test adopted by the Court as going beyond the legislature's intent in its enactment of Section 2 by, in effect, providing minorities with proportional representation. Justice O'Connor argued that the Court has, in effect, created "the right to a form of proportional representation" for certain minority groups with its combination of an erroneous definition of minority voting strength and its threepronged test. According to Justice O'Connor, the Court defined undiluted minority voting strength as "maximum feasible minority voting strength," calculating it as the maximum number of districts in which the minority group could constitute a majority in the most favorable single-member district plan (p. 90). The Court then proceeded to measure the degree of dilution by comparing the maximum feasible electoral success, given the number of minoritycontrolled districts possible, with the actual degree of minority electoral success. This definition is erroneous, she contended, both because it leads inevitably to proportional representation for minority groups and because it emphasizes only the ability of minorities to elect their preferred candidates and ignores other possible avenues by which minority groups might participate in the political process. Adopting this definition, Justice O'Connor reasoned, would result in a mandatory finding of vote dilution in violation of Section 2 unless the minority's preferred candidates are elected in rough proportion to the minority population percentage. Thus Justice O'Connor challenged the Court's test for vote dilution as leaning too far in the direction of proportional representation, ignoring the

54

Minority Representation and Voting Equality

carefully negotiated compromise inserted in the amendment of Section 2 to curtail just such a development.

The impact of Gingles on the ' 'totality-of-circumstances'' test In her concurrence in Thornburg v. Gingles, Justice O'Connor criticized the Court for adopting a standard that she perceived as essentially replacing the statutory test outlined in the legislative history of Section 2: "As shaped by the Court today, the basic contours of a vote dilution claim require no reference to most of the 'Zimmer factors' . . . which were highlighted in the Senate Report" (p. 92). She stated that she would adhere more to the approach outlined in Whitcomb and White, considering all of the relevant factors and making the decision as to whether the minority group has less opportunity than do other members of the electorate to participate in the political process and to elect representatives of their choice based on the totality of the circumstances. Is the three-pronged Gingles test designed to replace the totality-ofcircumstances test? Or is the three-pronged test a threshold requirement for further evaluation under the totality-of-circumstances test? Or does it simply offer a set of factors to be considered in addition to the list provided in the Senate Report? This uncertainty regarding the role of the three-pronged test relative to the totality-of-circumstances test is reflected in recent court opinions, which span the continuum from simply considering the three prongs as additional factors to be considered along with the Senate Report factors (e.g., see Buckanaga v. Sisseton Independent School District, 1986), to treating the three factors as preconditions to be met before considering the totality of the circumstances (e.g., see Monroe v. City ofWoodville, 1989, and McNeil v. Springfield Park District, 1988), to approaching the three-pronged Gingles test as if it were the sole standard to be met in a vote dilution challenge (see, e.g., Gomez v. City of Watsonville, 1988). The opinion produced by the Court in Gingles provides conflicting signals as to the role of the totality-of-circumstances test: On the one hand, the Court gives little attention to any of the Senate Report factors other than the degree of racial bloc voting and the extent of minority electoral success; in fact, in a footnote, the plurality explicitly states that the most important Senate Report factors to consider in a vote dilution challenge are the extent to which voting is racially polarized and the extent to which minority group members have been elected to office (pp. 4 8 ^ 9 , n. 15). The other Senate Report factors are to play a role "supportive of, but not essential to, a minority voter's claim" (p. 49, n. 15). However, at several points in the decision the Court specifically refers to the three parts of the Gingles test as "necessary preconditions," suggesting that

The evolution of a vote dilution standard

55

meeting the three Gingles prongs is only the first step in a vote dilution inquiry. Finally, the Court does not completely neglect the totality-of-circumstances test: As both amended Section 2 and its legislative history make clear, in evaluating a statutory claim of vote dilution through districting, the trial court is to consider the "totality of the circumstances" and to determine, based "upon a searching practical evaluation of the 'past and present reality,' " . . . whether the political process is equally open to minority voters, (p. 79) Furthermore, because the district court made specific findings with regard to each of the Senate Report factors, the fact that the Supreme Court did not discuss the totality of the circumstances in each of the challenged districts does not necessarily mean that the three Gingles factors completely replace the totalityof-circumstances test or that the three factors alone are sufficient to establish unlawful vote dilution. Thus, in cases subsequent to Gingles, the courts have had to decide the following: Are the three Gingles factors sufficient, in themselves, to demonstrate vote dilution, or must the courts also consider evidence pertaining to the other Senate Report factors and base their decision on the "totality of the circumstances?" Is satisfying the three Gingles factors necessary to a showing of vote dilution, or can a totality-of-circumstances inquiry be substituted and a conclusion drawn from factors other than the requirements established in Gingles? As the post-Gingles case law has evolved, most courts continue to apply some sort of totality-of-circumstances inquiry, reviewing all of the evidence offered with respect to each of the Senate Report factors and making findings of fact and conclusions of law with respect to the totality of circumstances in each case (Butler, 1990; Fontana, 1990; McDonald, 1989). However, courts have almost always found for plaintiffs in cases in which plaintiffs were able to satisfy the three prongs of Gingles. Of course, this still does not answer the question of whether proof of the Gingles factors is necessary or sufficient. Nor have the circuits been uniform in their approach to these considerations. At one end of the continuum is the Eighth Circuit, which has essentially treated the Gingles factors as three additional factors to be considered along with the other Senate Report factors. In Buckanaga v. Sisseton Independent School District (1986), a Section 2 suit brought by Native Americans challenging the at-large method for the election of the Sisseton School Board, the Eighth Circuit remanded the case with instructions to the district court to make detailed findings of fact with regard to each of the three Gingles factors (the district court had reached its decision before Gingles and therefore had not discussed the three-part test), as well as considering other Senate Report factors. No indication of the relative importance of the two sets of factors was given by the Eighth Circuit; nor did the court indicate what findings were necessary or sufficient for a vote dilution challenge to succeed.25

56

Minority Representation and Voting Equality

In the only Fourth Circuit case to address the issue directly, Collins v. City of Norfolk (1987), the appellate court stated that Gingles essentially offered a "gloss" on the Senate Report factors, and the "implication of this gloss on section 2 is that of the seven primary factors on the Senate Report list, two are typically the most important: the existence of racially polarized voting . . . and the actual results of minority-preferred candidates in winning elections" (p. 935). However, the court noted that although the presence of these two "cardinal factors" would weigh heavily in the final decision, the "ultimate determination still must be made on the basis of the 'totality of the circumstances''' (p. 938).26 The Fifth Circuit has generally discussed both the Gingles factors and the Senate Report factors in its opinions (see, e.g., Campos v. City of Bay town, 1988; LULAC v. Midland Independent School District, 1987; Overton v. City of Austin, 1989; and Westwego Citizens for a Better Government v. City ofWestwego, 1989, 1990) and stated in Citizens for a Better Gretna v. City of Gretna (1987) that it did not agree with the defendants' contention in that case that "the Supreme Court in Gingles made the Zimmer analysis obsolete" (p. 498). For the most part, the Fifth Circuit appears to have rested its decisions on the plaintiffs' ability to demonstrate that they have satisfied the three-part Gingles test, but most of the opinions issued by the Fifth Circuit have avoided spelling out the court's view of the relative importance of the two sets of factors. However, Judge Edith Jones, writing for the majority in two recent Fifth Circuit opinions, explicitly stated that the Gingles factors, although necessary to a showing of vote dilution, are not sufficient to prove a violation of Section 2. In Brewer v. Ham (1989), a challenge to the at-large system for electing the school board of the Killeen (Texas) Independent School District, Judge Jones found for the defendant school district on the grounds that the minority plaintiffs had failed to meet the first two prongs of Gingles, and she indicated the necessity of meeting the "Gingles threshold" before any other factors were to be considered. In Monroe v. City ofWoodville (1989), Judge Jones wrote that "satisfying the [Gingles] threshold test . . . does not prove a plaintiff's Section 2 claim; the district court must then proceed to the totality of the circumstances inquiry" (p. 1330). Hence, the current trend in the Fifth Circuit appears to be that the Gingles test is to be considered a necessary threshold to further consideration of the totality of the circumstances. The Seventh Circuit has had the opportunity to consider only the question of whether the Gingles test is necessary, as the plaintiffs failed to meet the first prong of Gingles in the only Section 2 case to reach the appellate court. In McNeil v. Springfield Park District (1988), the Seventh Circuit upheld the atlarge method of electing park board and school board members in the city of Springfield, Illinois, because the plaintiffs could not prove that they were able to constitute a majority in a single-member district. However, the Seventh Cir-

The evolution of a vote dilution standard

57

cuit suggested that the totality of the circumstances was still relevant to a Section 2 challenge, but only after the "three necessary preconditions" established by Gingles had been met: "Only upon satisfaction of these threshold criteria should a court conduct its totality-of-the-circumstances analysis and consider other relevant factors set forth in White" (p. 942).27 The Eleventh Circuit is divided on the role of a totality-of-circumstance inquiry in a vote dilution challenge and, more specifically, on the issue of the sufficiency of the Gingles factors in proving a Section 2 violation. Earlier Eleventh Circuit panels, in such cases as Carrollton Branch of the NAACP v. Stallings (1987) and Dillard v. Crenshaw County (1987), devoted some attention to both the Senate Report factors and the Gingles factors, but the court held in Carrollton that Gingles established a "new, three-part test" that, though not designed to replace the Senate Report factors completely, clearly designated some of them as more relevant to a finding of vote dilution than others were. The court in this case also indicated that proof the plaintiffs satisfied the Gingles test was sufficient to a finding of vote dilution: "The district court was incorrect when it held that even assuming racial polarization was found, the plaintiffs could not prevail because the other Senate Report factors were not proven . . . proof of these [Senate Report] factors is not required to show a vote dilution" (pp. 1555, 1561). In Solomon v. Liberty County (1990), however, the Eleventh Circuit altered its approach, with one segment of the evenly divided (five to five) en bane court indicating that although it is necessary for the plaintiffs to demonstrate that they meet the Gingles requirements, this evidence is not sufficient to conclude that vote dilution has occurred. In Solomon, black citizens challenged the at-large method of electing the county commission and the school board of Liberty County, Florida. The district court entered judgment in favor of the defendants, and the plaintiffs appealed. The Eleventh Circuit, in an opinion written by Judge Gerald Bard Tjoflat, outlined a new four-part test of vote dilution (which included the need to prove invidious discrimination) for the district court to follow on remand (Solomon v. Liberty County, 1989). The plaintiffs' request for a rehearing en bane was granted, and the three-judge panel decision was vacated. The en bane court (Solomon v. Liberty County, 1990) produced two concurring opinions, one written by Judge Phyllis A. Kravitch (joined by four judges) and the other written by Judge Tjoflat (joined by four judges). Both of the opinions acknowledged that the plaintiffs had met the three Gingles requirements; the Eleventh Circuit panel was divided, however, on whether or not this evidence was sufficient to conclude that a violation of Section 2 had occurred. Judge Kravitch held that proof that the minority group satisfied the Gingles criteria was sufficient to make out a claim of vote dilution and that "although a district court may consider the totality of the circumstances, those

58

Minority Representation and Voting Equality

circumstances must be examined for the light they shed on the existence of the three core Gingles factors" (p. 1017). Judge Tjoflat asserted that Judge Kravitch had "redefined the totality-of-thecircumstances test to include only the three, mechanical factors articulated in Gingles" (p. 1022). In his view, the totality of the circumstances should be interpreted such that not only is it necessary to meet each of the three Gingles requirements but also that proof of racial bias on the part of voters is required to succeed in a vote dilution claim: Section 2 prohibits those voting systems that have the effect of allowing a community motivated by racial bias to exclude a minority group from participation in the political process. Therefore, if a section 2 defendant can affirmatively show, under the totality of the circumstances, that the community is not motivated by racial bias in its voting, a case of vote dilution has not been made out (p. 1022).28 Subsequent Eleventh Circuit decisions have avoided resolving this issue. In Meek v. Metropolitan Dade County (1990), a suit brought by black and Hispanic registered voters challenging the at-large system for electing Dade County's (Florida) board of commissioners, the Eleventh Circuit panel noted that "this circuit is divided on . . . the question of whether proof of the three core Gingles factors was sufficient and whether a defendant could raise as a defense the lack of racial bias in the community" and left it to the district court to decide the issue on remand (p. 1544). The Ninth Circuit has specifically held that proof of the Gingles factors alone is sufficient to establish a Section 2 violation. In Gomez v. City of Watsonville (1988), the Ninth Circuit held that the at-large scheme for electing the Watsonville (California) City Council violated Section 2 based on proof of the three Gingles factors alone: "Factors other than the three elements discussed [in Gingles] ... while supportive of a Section 2 claim, are 'not essential to a minority voter's claim'" (p. 1419). In Romero v. City of Pomona (1989), the Ninth Circuit again pointed to the sufficiency of the Gingles factors in establishing vote dilution, holding that although the district court - "perhaps out of an abundance of caution" (p. 1421) - had applied the Senate Report factors as well as the three prongs of Gingles, this was unnecessary. The Ninth Circuit has been less clear with regard to the necessity of satisfying the three prongs of Gingles. In Romero, the appellate court rejected the plaintiffs' Section 2 claim owing to the failure of the combined minority group (blacks and Hispanics) to meet the first two prongs of Gingles. But in a recently decided case - Garza v. Los Angeles County of Supervisors (1990) - the Ninth Circuit argued that the plaintiffs need not show that they could constitute a majority in a single-member district if there were evidence of intentional discrimination.30 To date, only the Sixth Circuit has specifically held that proof of the three

The evolution of a vote dilution standard

59

Gingles prongs is not necessary to a showing of vote dilution - and the appellate court relied on the totality-of-circumstances test as outlined in the Senate Report to support this contention. In Armour v. State of Ohio (1990), black citizens residing in two Ohio state legislative districts brought action alleging unlawful vote dilution. Although not large enough in size to constitute a majority in a redrawn single-member district, the plaintiffs asserted that the current state legislative lines fragmented the black population in such a way that they were being denied the ability to "influence" elections. The district court, citing McNeil v. Springfield Park, dismissed the complaint on the grounds that the plaintiffs had failed to meet the first prong of Gingles. The Sixth Circuit reversed, holding that a minority group is not required to demonstrate that it would constitute a majority in a redrawn district to make out a successful vote dilution claim.31 The Sixth Circuit argued that in endorsing the totality-of-circumstances approach, the Senate Report "clearly indicates that [Section 2] . . . requires a case by case analysis by the Court, thus precluding a bright line test applicable under all circumstances" (slip opinion, p. 10). The district court ignored this directive, according to the Sixth Circuit,and employed a "mechanical 'point counting' requirement that the minority constitute a majority in a district before the court would exercise its overall judgment based upon the totality of the circumstances of the case" (slip opinion, p. 10). The Sixth Circuit accepted the plaintiffs' "influence" argument, favorably citing Karlan's reasoning (1989, p. 202): "To the extent that courts have read Gingles to elevate the ability to create a district with a majority black electorate into a threshold requirement for establishing liability in all vote dilution litigation, they have improperly applied one particular theory of liability to other distinct types of vote dilution" (slip opinion, p. 13, n. 7). In short, there is considerable disagreement about the proper role of the totality-of-circumstances test and the necessity and sufficiency of the three Gingles factors. An approach that perhaps comes close to the intent of Congress in amending Section 2 and the Supreme Court's decision in Gingles is one that allows the courts to view at least the first prong of the Gingles test as germane only in certain instances and that allows the three prongs to be sufficient in that circumstance. Surely there are instances in which one would wish to relax the first prong, such as in cases that do not involve districting (McDonald, 1989). As its legislative history makes clear, Section 2 applies to any racially discriminatory voting practice or procedure, not simply to dilutive districting plans. Therefore, such potentially dilutive devices as majority vote requirements are subject to challenge under Section 2 - and, in fact, the Justice Department has recently filed suit in Georgia challenging the state's majority vote requirement under Section 2. The evidence presented in this case is unlikely to rely solely on the

60

Minority Representation and Voting Equality

factors outlined in Gingles, and the requirement that the minority group be sufficiently large and geographically compact is obviously the least relevant of the three prongs. In these sorts of cases, the totality of the circumstances would seem to remain particularly relevant and the first prong of Gingles irrelevant.32 Nonetheless, when plaintiffs are in fact challenging at-large/multimember election systems and requesting single-member districting as a remedy, we think that all three prongs of Gingles are necessary for showing a violation of Section 2. In taking this position, we are persuaded by Justice Brennan's reasoning, cited earlier, that "unless minority voters possess the potential to elect representatives in the absence of the challenged structure or practice, they cannot claim to have been injured by that structure or practice" (Thornburg v. Gingles, 1986, p. 50). However, as we shall see in the next chapter, even the conclusion that all three prongs are necessary does not resolve all of the potential issues. For one thing, there remain some questions of interpretation - for example, about what constitutes a majority. For another, the first prong may not, in fact, be necessary if the Court is willing to require remedies that involve alternatives to the single-member district or if there has been intentional discrimination (see Chapter 5). As for sufficiency, it is our feeling that the Gingles three-pronged test places the focus on a set of relatively clear, objective criteria, creating a manageable standard with a list of critical factors that is both small and closed ended. By designating racially polarized voting as the keystone of a vote dilution challenge, the Court has constructed a standard that contains a "core" value. Thus, in the context of multimember districts, the three-pronged test sets out sufficient conditions for finding a violation of Section 2. Although the Gingles decision effectively resolved a number of the questions that the courts had confronted when considering a vote dilution claim, not all the conflict was settled by this decision (such as the discussion about the role of the totality-of-circumstances test in Section 2 claims makes clear). Subsequent litigation based on Section 2 has both further refined the standard outlined in Gingles and given rise to divergent opinions on some issues. This is the subject of the next chapter.

The vote dilution standard in the post-Gingles era: clarifications and complications in the lower courts

The Supreme Court's decision in Thornburg v. Gingles (1986) was an important victory for minority voting rights. Not only did the decision confirm that amended Section 2 eliminates the need to prove discriminatory intent in statutory vote dilution claims, but the three-part Gingles test developed by the Court also appears to simplify considerably the standard of proof to be applied in Section 2 districting challenges. As noted in Chapter 2, the Court created a three-pronged test for vote dilution in an at-large or multimember district challenge. To establish a violation, "a bloc voting majority must usually be able to defeat candidates supported by a politically cohesive, geographically insular minority group" (1986, p. 49). This requires that plaintiffs satisfy three conditions: First, the minority group must be . . . sufficiently large and geographically compact to constitute a majority of a single-member district. . . . Second, the minority group must be . . . politically cohesive. . . . Third, . . . the white majority votes sufficiently as a bloc to enable it . . . usually to defeat the minority's preferred candidate, (pp. 50-51) These three elements seemingly establish a straightforward, objective test for proving unlawful vote dilution. However, at least partially because the opinion rendered in Gingles included a complex web of pluralities, concurrences, and dissents, not all Section 2 issues were resolved. The confusion in Gingles has also spawned considerable confusion in subsequent interpretations; a few points of contention have been resolved, but there are numerous conflicting views and inconsistent lower court decisions. In this chapter we discuss the continuing evolution of the vote dilution standard encompassed in the amended Section 2 of the Voting Rights Act, examining how the lower courts have applied the three Gingles factors (and pointing out when the courts have disagreed on the correct standard to use). Each of the three Gingles factors will be considered separately.1

61

62

Minority Representation and Voting Equality

What constitutes a "sufficiently large and geographically compact" minority group? The first prong of Gingles has led the lower courts to grapple with several questions. First, what constitutes a "sufficiently large" minority population that is, what is the percentage of minority population needed in a single-member district? Second, what does "geographically compact" mean? Third, can two or more minority groups be combined to meet this first prong? What does it mean to constitute the ' 'majority'' of a single-member district? The Supreme Court in Gingles required plaintiffs to prove that they were large enough to constitute a "majority" in a single-member district.2 However, the Court did not specify what was meant by the term majority: Are plaintiffs required to constitute a majority of the proposed district's total population or voting-age population? Or perhaps plaintiffs must show that they are sufficiently large to be able to comprise a majority of the registered voters or even a majority of the actual voters at the polls on election day? Recent court opinions have been relatively uniform in assuming that majority used in this context means that the minority group must demonstrate that it can form a majority of the voting-age population in a single-member district. For instance, in McNeil v. Springfield Park District (1988), the Seventh Circuit stated that the first prong of Gingles "roughly measures minority voters' potential to elect candidates of their choice. Because only minorities of voting age can affect this potential, it is logical to assume that the Court intended the majority requirement to mean a voting age majority" (p. 945). Therefore, although the plaintiffs in McNeil were able to draw a single-member district that was 50.4 percent black in total population, because the proposed district was only 43.7 percent black in voting-age population, the court held that the plaintiffs had failed to meet the first prong of Gingles. The other circuits have tended to follow the Seventh Circuit's lead on this issue. For example, although the Fourth Circuit has issued no appellate court opinions that directly confront this issue, a recent district court opinion (McDaniels v. Mehfoud, 1988) was explicit in its requirement that plaintiffs demonstrate that they could comprise a majority of the voting-age population in a single-member district (and the court cited McNeil v. Springfield Park District to support this contention). Earlier decisions rendered in the Fifth Circuit discuss only the total minority populations of the plaintiff-proposed districts (Campos v. City of Baytown, 1988; LULAC v. Midland Independent School District, 1987), but because combined minority populations in excess of 70 percent were

The vote dilution standard in the /?<9,sr-Gingles era

63

possible in both of these cases, later Fifth Circuit decisions did not have to override precedent in explicitly requiring plaintiffs to prove that they would be able to constitute a majority of the voting-age population in a single-member district (Brewer v. Ham, 1989; Overton v. City of Austin, 1989; Westwego Citizens for a Better Government v. City of Westwego, 1990). In the Eleventh Circuit, the first case to address the specific question of how "majority" should be measured was Solomon v. Liberty County (1990). In this en bane decision in which the court was evenly divided (five to five) over the standard of proof necessary to establish a Section 2 violation, both divisions of the court held that because a district could be drawn in which blacks comprised 51 percent of the voting-age population, plaintiffs had satisfied the first prong of Gingles. However, Judge Phyllis A. Kravitch, in her concurring opinion (joined by four judges), added the following caveat: This holding should not be read to imply an opposite result where blacks do not constitute an outright majority of the voting age population in any district. So long as the potential exists that a minority group could elect its own representative in spite of racially polarized voting, that group has standing to raise a vote dilution challenge under the Voting Rights Act. (p. 1018, n. 7) Hence, the possibility that minorities need not prove that they could constitute a voting-age majority in a single-member district has been left open in the Eleventh Circuit.3 Defendants have also argued - and courts have on occasion accepted the contention - that proof that a district can be drawn in which the minority group constitutes an "effective voting majority" is needed. For example, in Romero v. City of Pomona (1989), the Ninth Circuit stated that "Thornburg repeatedly makes reference to effective voting majorities, rather than raw population totals" (p. 1425) and therefore required the plaintiffs in that case to show that the Hispanic community was numerous enough to constitute a citizen voting-age majority in at least one district (which the plaintiffs were unable to do). Subsequent district court opinions in the Ninth Circuit have reiterated the requirement that citizenship as well as voting-age population must be taken into account in determining whether plaintiffs have met the first prong of Gingles (see, e.g., Skorepa v. City of Chula Vista, 1989; but also see Garza v. Los Angeles County Board of Supervisors, 1990, discussed later). Generally, however, the courts have been reluctant to accept this argument (see, e.g., McDaniels v. Mehfoud, 1988; Neal v. Coleburn, 1988; Solomon v. Liberty County, 1990).4 Even requiring plaintiffs to prove that they can constitute a majority of the voting-age population in a district ignores the possibility of special circumstances that may give the minority community the potential to elect candidates of its choice with less than 50 percent of the voting-age population. Jurisdictions

64

Minority Representation and Voting Equality

where the minority turnout exceeds the white turnout and jurisdictions where there is a fair degree of white crossover voting are two such instances.5 In recent litigation focusing on single-member districts, still more questions have begun to arise about the size part of the first Gingles prong - for example, whether so-called influence districts have any constitutional standing. We will take up these matters in Chapter 5. What does "geographically compact" mean? Despite the Supreme Court's use of the term compact, lower courts have, almost without exception, interpreted this part of the first prong to mean only contiguity - that is, that all parts of the district are joined together. Thus, the courts have tended not to separate the question of geographic compactness from the question of whether the minority group is numerous enough to constitute a majority; if the plaintiffs are able to draw a (contiguous) plan in which they comprise a majority in at least one district, then they have met the first prong, regardless of the shape of the district. On those occasions in which a separate meaning of compactness has been proposed (inevitably by the defendants), the courts have typically, though not always, rejected these efforts. In a few cases the arguments have had little to do with conventional notions of compactness. For example, in Campos v. City of Bay town (1988), a challenge brought by blacks and Hispanics against the atlarge method of electing Baytown, Texas, city council representatives, the defendants argued that because 51 percent of the blacks and 44 percent of the Hispanics would not reside in the minority district, the plaintiffs had failed to meet the requirement of geographic compactness. The Fifth Circuit disagreed, holding that "the fact that there are members of the minority group outside the minority district is immaterial" (p. 1244).6 Still, most of the discussion of compactness has revolved around the actual shape of plaintiff-proposed districts, and it has occurred only in district court opinions. For example, in Dillard v. Baldwin County Board of Education (1988), a suit brought by black voters challenging the at-large method for electing members of the board of education in Baldwin County, Alabama, the court rejected the defendants' argument that the plaintiffs' proposed district failed to satisfy the first prong of Gingles "because it is too elongated and curvaceous" (p. 1465). The Eleventh Circuit district court held that "by compactness, Thornburg does not mean that a proposed district must meet, or attempt to achieve, some aesthetic absolute, such as symmetry or attractiveness" (p. 1465). In Neal v. Coleburn (1988) - a Section 2 challenge to a single-member district plan for electing the Nottoway County (Virginia) Board of Supervisors - a district court in the Fourth Circuit rejected the defendants' claim that the

The vote dilution standard in the p^-Gingles era

65

proposed remedy districts were "too tortuous and irregular in shape to pass muster" (p. 1437). The court decreed that the proposed minority districts were "not unreasonably irregular in shape, given the population dispersal within the County and the need to create majority black districts" (p. 1437). In Jeffers v. Clinton (1989), a three-judge panel in the Eighth Circuit found that although some of the plaintiffs' proposed districts "look rather strange," they did not believe this to be "fatal to plaintiffs' position" (p. 207). The court pointed out that the plaintiffs' alternative districts were "not materially stranger in shape than at least some of the districts contained in the present apportionment plan" (p. 207) and hence concluded that the plaintiffs (black voters challenging the legislative districting plan for the state of Arkansas) were sufficiently large and geographically compact to satisfy the first prong of Gingles. In East Jefferson Coalition for Leadership and Development v. Parish of Jefferson (1988), a district court in the Fifth Circuit held that the method of electing the Jefferson Parish Council violated Section 2, but it did reject the plaintiffs' proposed remedy as a "totally unacceptable gerrymander" (p. 1008). According to the court: A proposed district is sufficiently compact if it retains a natural sense of community. To retain that sense of community, a district should not be so convoluted that its representatives could not easily tell who actually lives within the district. . . . The plaintiffs' plan stretches along the river and reaches around the airport to include a concentration of black residents living above the airport. This court cannot accept a plan which contains a district which is drawn with the acknowledged intent to include minorities, but which does not meet the minimal requirements of reapportionment. (p. 1007) The district court subsequently accepted (in its proposed order dated January 19, 1989) a redistricting plan offered by the defendants that contained no majority black districts but did include a district in which blacks constituted 41 percent of the voting-age population (a district that the court concluded provided blacks "with a realistic opportunity to influence the outcome of councilmanic elections" (p. 5, emphasis added). However, the plan was denied Section 5 preclearance by the Justice Department on the grounds that the "submitted changes . . . may well have been motivated by an invidious purpose to minimize black voting strength" (letter from James Turner, acting assistant attorney general, Civil Rights Division, to Harry Rosenburg, attorney representing Jefferson Parish, November 17, 1989). At least in this case, the Justice Department's objection has effectively repudiated the district court's holding that a lack of geographic compactness (evaluated in terms of a district's actual shape) can prevent the establishment of a minority district if it is possible to create such a district.7 Thus, court decisions so far suggest that the creation of minority-controlled districts is, after equal population, the central goal in drawing proposed redis-

66

Minority Representation and Voting Equality

tricting plans.8 If this position is maintained, compactness - whatever its value and however defined - would give way completely to the drawing of majorityminority districts. The same is likely to be true of requirements to adhere, to the extent possible, to existing jurisdictional boundaries, though the courts have not addressed this matter specifically. Hence, this deceptively small part of the Gingles decision could lead to an important principle governing districting decisions. On the other hand, three developments suggest that compactness could yet play a larger role in districting. First, approximately half of the states have some sort of compactness requirement in their state constitutions or statutes (Eig and Seitzinger, 1981), and state courts have, on occasion, found a violation of these requirements.9 Second, quantitative measures of compactness have been codified, regularized, and thoroughly explicated.10 Although this has not led to the development of a single test with a precise cutoff point, it does indicate that the concept can be measured more rigorously than with an "eyeball" test. Third, the extraordinary ingenuity of mapmakers, aided by computers and geographicbased census files, has already produced new districts for the 1990s that surpass almost any earlier district in degree of contortion.11 With these developments, especially the test cases provided by some of the new districts, federal courts are likely to have to rule soon on whether there are limits to the geographic sprawl of districts designed to be winnable by minorities. Can minority groups be combined to meet the first prong? Several appellate courts have recently been asked to consider cases in which there is more than one protected minority group present in a jurisdiction and these groups have joined forces to sue the jurisdiction on Section 2 grounds. In a number of these suits (arising mainly in Texas, Florida, and California), the question has been posed in regard to whether the minority groups (usually blacks and Hispanics) can be combined for purposes of meeting the first prong of Gingles.12 Thus far, courts have resolved this issue by permitting the groups to be treated as a combined group only if they can demonstrate that they are cohesive; if the groups are unable to meet this standard, then the courts have been reluctant to allow aggregation (see Brewer v. Ham, 1989; Campos v. City ofBaytown, 1988; LULAC v. Midland Independent School District, 1987; and Overton v. City of Austin, 1989, in the Fifth Circuit; Romero v. City of Pomona, 1989, in the Ninth Circuit; and Concerned Citizens of Hardee County v. Hardee County Board of Commissioners, 1990, in the Eleventh Circuit).13 The issue of combining minority groups for purposes of meeting the Gingles test is discussed at greater length later when the issue of political cohesiveness is considered.

The vote dilution standard in the post-Gingles era

67

When is a minority group "politically cohesive"? Of the three requirements put forth in Gingles, the requirement of politically cohesiveness was the least anticipated by interested observers - it was, in the words of one commentator, a "wild card" (Samuel Issacharoff, speech presented at the Lawyers Committee for Civil Rights Under Law, Conference on Voting Rights, New Orleans, April 20-21, 1990). There had been no previous case law indicating the possible importance of this factor (Lichtman and Hebert, 1986); in fact, the only basis for including it as a condition for establishing a Section 2 violation indicated in Gingles was a reference to two previously published law review articles: Blacksher and Menefee, 1982, and Carpeneti, 1972.14 Nevertheless, this factor has emerged as one of critical importance in a number of cases, particularly those suits in which the minority community is composed of more than one racial or ethnic group. How should "political cohesiveness" be measured? The term political cohesiveness is not expressly defined by the Court in Gingles. The Court's comments concerning the requirement, however, indicate that plaintiffs must demonstrate that members vote sufficiently as a group to establish that (1) there is a distinct group interest and (2) enough members of the group can be expected to vote together to be able to elect candidates of their choice in a single-member district. Therefore an investigation into the question of political cohesiveness entails an examination of the minority community's voting patterns. In fact, according to the Gingles Court, the purpose of inquiring into the existence of racially polarized voting is not only to determine white voting patterns but also to "ascertain whether minority group members constitute a politically cohesive unit" (p. 56).15 Not all courts have taken political cohesiveness to be limited to an inquiry into voting patterns. For example, a district court found that the Hispanic community was not cohesive in Gomez v. City of Watsonville, despite the fact that according to the Ninth Circuit, "the district court found that 95% of the Hispanic voters in heavily Hispanic precincts support Hispanic candidates" (1988, p. 1414). In the district court's view, the issue of whether or not the minority community is politically cohesive must be analyzed in terms of "the community as a whole" and should not be based simply on voting patterns. Evidence cited by the court included the fact that there were some socioeconomic differences among Hispanics (e.g., not all Hispanics shared the same level of income or education) and many eligible Hispanics failed to register or vote, leading the lower court to conclude, in the words of the Ninth Circuit, that "the Hispanic community as a whole was too apathetic to be politically cohesive" (p. 1415).

68

Minority Representation and Voting Equality

However, the Ninth Circuit reversed the district court's ruling, indicating that the district court's adverse ruling was based on "a misunderstanding of what is meant by 'political cohesiveness' " (p. 1415). The appellate court stated that an inquiry into the issue of political cohesiveness should be directed at "whether the minority group has expressed clear political preferences that are distinct from those of the majority" and should be "judged primarily on the basis of the voting preferences expressed in actual elections" (p. 1415). The Ninth Circuit is not the only circuit to have concluded that cohesiveness is to be measured with reference to voting patterns: The Fourth, Eleventh, and, for the most part, the Fifth circuits all have agreed that political cohesiveness is to be determined by examining the voting preferences of the minority community. Only the Tenth Circuit has clearly indicated that other types of evidence can be considered in making a determination regarding the degree of political cohesiveness.16 In Collins v. City of Norfolk (1987), the Fourth Circuit indicated that where the court found the existence of racially polarized voting, "both cohesiveness of the minority group and the power of white bloc voting to defeat the minority's candidates" have been established (p. 935). This approach to defining political cohesion was reiterated by the Fourth Circuit two years later, when the court was asked once again to consider the Norfolk case. In this opinion (Collins v. City of Norfolk, 1989) the Fourth Circuit found that Norfolk's black citizens "vote sufficiently as a bloc to constitute a politically cohesive unit" (p. 1237 n. 4). The Eleventh Circuit has also relied on an examination of voting patterns to determine whether the minority group is sufficiently cohesive. In Carrollton Branch of the NAACP v. Stallings, the appellate court devoted little attention to this issue but, in a footnote, indicated that "proof of racial polarization may, of course, be found sufficient... to make the necessary finding of cohesiveness" (1987, p. 1563, n. 15). In Solomon v. Liberty County (1990), both contingents of the evenly split court used statistical evidence indicating the existence of racial bloc voting to hold that blacks in Liberty County, Florida, were politically cohesive. In cases decided after Solomon, the Eleventh Circuit continued to emphasize the importance of voting patterns - holding that both blacks and Hispanics, considered separately, were politically cohesive groups in Dade County, Florida (Meek v. Metropolitan Dade County, 1990), and that blacks and Hispanics, as a combined group, were not politically cohesive in Hardee County, Florida (Concerned Citizens of Hardee County v. Hardee County Board of Commissioners), because no evidence had been presented to demonstrate that the two groups "have ever voted together" (1990, p. 527). The vast majority of decisions in the Fifth Circuit have also relied exclusively on vote polarization (e.g., Brewer v. Ham, 1989; Campos v. City of

The vote dilution standard in the po^^-Gingles era

69

Baytown, 1988; LULAC v. Midland Independent School District, 1987; Overton v. City of Austin, 1989). However, the Fifth Circuit has also cited factors other than minority voting patterns as the basis for concluding that plaintiffs were not sufficiently cohesive to meet the second prong of Gingles. In Monroe v. City of Woodville (1989), Judge Edith Jones stated that a finding of racial polarization is not "synonymous with a group's political cohesiveness"; instead, political cohesiveness "implies that the group generally unites behind a single political 'platform' of common goals and common means by which to achieve them" (p. 1331). Judge Jones's opinion goes even further and suggests that the failure of blacks to elect a black to office, in a city in which they constitute 60 percent of the population, also signals a lack of political cohesiveness.17 In another Fifth Circuit decision, Houston v. Haley (1988), the court determined that blacks were not cohesive in Oxford, Mississippi, in part because no black had run for office: "Despite a 53.8% black district created by the 1984 electoral plan, no black person qualified to run for the board of aldermen. This failure despite the presence of a potential majority calls into question whether the minority group members constitute a politically cohesive unit" (p. 346).18 The only circuit in which precedence has clearly established that evidence other than a statistical analysis of racial bloc voting is relevant to a finding of political cohesiveness is the Tenth Circuit. In Sanchez v. Bond (1989), the Tenth Circuit affirmed the district court's finding that Hispanics were not politically cohesive in Saguache County, Colorado. Although statistical evidence presented at trial by the plaintiffs' expert indicated that voting was racially polarized, the district court pointed to anecdotal testimony offered by a single lay witness (an Anglo-supported Hispanic school board member), who reported that Hispanics had "differing political objectives." The Tenth Circuit held that the district court was not clearly erroneous because it could "find nothing in Gingles . . . to suggest that a trial court is prohibited from considering lay testimony in deciding whether a minority group is politically cohesive" (p. 1493, emphasis in original). In the end, what is most important is the Court's explicit assertion about the need to use racial bloc voting inquiries. Whatever else is considered, courts cannot ignore evidence relating to the degree of racially polarized voting in arriving at a conclusion about the cohesiveness of a particular minority group. It is also important, however, and not definitively determined, whether Gingles prohibits courts from considering additional evidence in deciding whether a group is politically cohesive. Indeed, one can imagine that this question will become more important - in the short run as defendants learn to argue that minorities are not monolithic and in the long run as minorities make economic gains and thereby become more differentiated.

70

Minority Representation and Voting Equality

Can separate and distinct minority groups be combined to form a single, politically cohesive group? An issue not addressed by the Gingles Court that has proved to be of considerable importance in subsequent cases is whether two or more distinct minority groups can be said to constitute a single, cohesive minority group for Section 2 purposes. In those instances when faced with plaintiffs from two or more minority groups who collectively assert that a voting scheme unlawfully dilutes their votes, courts have usually held that the minority groups may aggregate themselves for the purposes of meeting the first prong of Gingles if they can prove that they meet the second prong - that is, if they can prove that they are politically cohesive. However, several scholars have argued for a more rigorous standard than simply examining voting patterns to determine cohesiveness if the group seeking relief is composed of more than one minority group (see, e.g., Butler, 1990; Strange, 1989), and at least one appellate judge has expressed the belief that simply demonstrating that the minority groups vote for the same candidates is insufficient grounds for allowing aggregation. One early case to consider the issue of combining minority groups was LULAC v. Midland Independent School District (1987) - a suit filed by AfricanAmerican and Mexican-American citizens challenging the at-large system of electing the Midland County (Texas) School Board. It was important that the two minority groups be considered together in this case because blacks alone could not satisfy the first of the Gingles requirements.19 The trial court allowed the aggregation, finding that the two minority groups were politically cohesive because the two groups had a shared history of past discrimination and inseparable political goals that led to the formation of coalitions on occasion (including the joint filing of the Section 2 lawsuit). The Fifth Circuit, in an opinion written by Judge John Minor Wisdom, affirmed the lower court's finding of political cohesion, noting the two factors mentioned by the district court and adding that the minority groups consistently voted for minority candidates over Anglo candidates.20 Judge Patrick E. Higginbotham filed a dissenting opinion in LULAC arguing that the two minority groups should not have been considered in tandem. It was his contention that simply examining the voting patterns of the two minority groups to determine whether the groups are voting for the same candidates is insufficient to ascertain political cohesiveness: "If a minority group lacks a common race or ethnicity, cohesion must rely principally on shared values, socio-economic factors, and coalition formation, making the group almost indistinguishable from political minorities as opposed to racial minorities" (p. 1504). Judge Higginbotham expressed concern that if blacks and Hispanics should be found politically cohesive solely on the grounds of similar voting

The vote dilution standard in the post-Gingles era

71

patterns, then the Voting Rights Act might become a refuge for every losing political coalition: "Diluting the requirement of cohesion expands the mission of the Act beyond the treatment of present-day manifestations of chronic bigotry to a more general device for accommodating majority government and plural constituents" (p. 1504). Judge Higginbotham repeated his warning in Campos v. City of Baytown (1988). The district court in Campos found that blacks and Hispanics in Baytown were politically cohesive based on a statistical analysis of voting patterns; the Fifth Circuit affirmed this holding, although vacating the lower court's opinion because the remedy plan imposed by the lower court had not been submitted to the Justice Department for Section 5 preclearance.21 In its opinion, the Fifth Circuit specifically rejected the defendants' argument that the cohesiveness of two distinct minority groups be measured "prior to and apart from a study of polarized voting" and indicated that "voting patterns" should be the "central focus" of a determination of cohesiveness (p. 1244). Judge Higginbotham filed a dissent, expressing alarm that the protection of the Voting Rights Act had been extended "by fiat" (p. 945) to a "newly defined minority - a coalition of Blacks and Browns" (p. 944). It was his contention that the Voting Rights Act was never intended to cover a group composed of two distinct minority groups, and by "stretching the concept of cohesiveness" to include such a coalition, the court was delving into the political arena. Some scholars have indicated that they share Higginbotham's fears. According to Strange, "the danger of simply accepting the notion that different minority groups may be aggregated . . . is that the scope of the Voting Rights Act may be expanded beyond congressional intent... tak[ing] the Voting Rights Act into the realm of interest group or coalition politics" (1989, p. 112). He proposed three factors that the courts should consider before allowing minority groups to aggregate for Section 2 purposes: Not only should the component groups vote consistently for the same candidates, but they also should have similar socioeconomic backgrounds and similar attitudes toward significant issues. Butler went one step further and suggested that "minority coalition dilution" suits be permitted only in the "most unusual of circumstances" (1990, pp. 623624). She believes that minority groups should be allowed to combine forces in a vote dilution claim only when the two groups can prove that they consider themselves as "one"; it is only after this has been established to the court's satisfaction that the three-pronged Gingles test should be applied (including the question of whether the combined minority group is politically cohesive).22 The courts, however, have not found persuasive the arguments by Judge Higginbotham and by commentators such as Strange and Butler. The fact that the minority community is less than monolithic in its socioeconomic status and attitudes has been seen as irrelevant to determining whether the votes of the

72

Minority Representation and Voting Equality

minority community count less than do the votes cast by the white/Anglo community. What is relevant is an empirical determination of whether blacks and Hispanics consistently vote together in elections in which minority and white candidates compete for office. The courts have not often allowed minority groups to be combined. LULAC and Campos are the only two cases in which aggregation for Section 2 purposes has been permitted. There have been several cases where aggregation of blacks and Hispanics was sought but not granted. In each instance, the plaintiffs were denied relief because the two groups were found not to be politically cohesive {Brewer v. Ham, 1989; Concerned Citizens ofHardee County v. Hardee County Board of Commissioners, 1990; Overton v. City of Austin, 1989; Romero v. City of Pomona, 1989). Political cohesiveness, in all of these cases, was measured with reference to minority voting patterns. For example, in the Fifth Circuit decisions of Overton v. City of Austin and Brewer v. Ham, an examination of the voting patterns of the minority groups indicated that the groups did not support the same candidates; hence the court concluded that the two groups were not cohesive. In Overton, a challenge by blacks and Mexican-Americans to Austin's at-large city council elections, the Fifth Circuit upheld the lower court's finding that the plaintiffs had failed to show political cohesiveness because "Anglos voted with one or the other minority groups more frequently than the two groups voted together" (p. 536).23 In Brewer, the Fifth Circuit indicated that because the statistical analysis showed that "blacks and Hispanics were in direct opposition in most . . . elections" (p. 454), the plaintiffs (black, Hispanic, and Asian citizens challenging the Killeen, Texas, Independent School District's at-large method of electing school board members) had failed to satisfy the second prong of Gingles. In Romero v. City of Pomona (1989), involving a challenge to the city of Pomona, California's at-large system for electing its city council, the Ninth Circuit agreed with the district court that blacks and Hispanics were not politically cohesive based on the results of an exit poll that indicated that the two groups did not vote for each other's candidates (in the 1985 city council elections). Because the plaintiffs could not satisfy the second requirement, the Ninth Circuit held that blacks and Hispanics could not be combined to meet the first prong and therefore concluded that the challenged at-large system did not violate Section 2. In Concerned Citizens of Hardee County v. Hardee County Board of Commissioners (1990), the Eleventh Circuit upheld the district court's finding that blacks and Hispanics were not politically cohesive, noting that' 'the class failed to demonstrate that blacks and hispanics in Hardee County have ever voted together" (p. 527). Thus, it seems clear from the opinions generated by those courts that have

The vote dilution standard in the post-Ging\e$> era

73

considered this question that aggregation is permissible if (and only if) it can be demonstrated that groups support each other's candidates. Hence, the issue of whether two or more minority groups can be combined for Section 2 purposes appears to have been answered: Minority groups may be combined if they can demonstrate that they are politically cohesive. And minority groups are politically cohesive if they vote together for minority candidates.24 How does one determine when the white majority votes sufficiently as a bloc to enable i t . . . usually to defeat the minority's preferred candidate? The third condition for proving a Section 2 violation is to demonstrate ' 'legally significant" white bloc voting. As outlined by the Court, an inquiry into polarization25 has two steps. The first is determining how minority and white voters have cast their ballots - polarization occurs "where black voters and white voters vote differently" (Thornburg v. Gingles, 1986, p. 53, n. 21). The second step is determining whether the polarization is, in fact, legally significant. This requires establishing who the minority-preferred candidates are and how often these candidates are successful in their bids for office. The Gingles Court declined to establish any definitive point at which white bloc voting becomes legally significant. Instead, it adopted a "functional view," indicating that "there is no simple doctrinal test for the existence of legally significant racial bloc voting" (p. 58). The Court stated that "because . . . the extent of bloc voting necessary to demonstrate that a minority's ability to elect its preferred representatives is impaired varies according to several factual circumstances, the degree of bloc voting which constitutes the threshold of legal significance will vary from district to district" (pp. 55-56). 26 What is clearly established by Gingles is that white bloc voting is legally significant, regardless of the actual percentages of whites voting against minority-preferred candidates, when it usually results in the defeat of the minority-preferred candidates. This is, of course, a requirement of proof that the challenged electoral practice actually has a discriminatory impact that causes injury to the voting rights of minority voters; if minority-preferred candidates usually win, despite racially polarized voting, then no injury can be said to have occurred. Most of the issues relating to the first step of a polarization inquiry have been resolved with Gingles; it is the second step - determining who the minoritypreferred candidates are and what elections are appropriate to include in an analysis of legally significant racially polarized voting - that still generates conflict in the lower courts. Another point that has been disputed is whether it is ever appropriate to attempt to find out whether disparate voting patterns can be explained by factors other than race.

74

Minority Representation and Voting Equality

How should we measure racial bloc voting? Proof of racially polarized voting depends on statistical evidence of voting patterns by race. There are two standard methods employed by expert witnesses to estimate the voting behavior of minority voters and white voters: homogeneous precinct analysis and bivariate ecological regression analysis.27 The Court in Gingles sanctioned the use of both of these methods when it accepted the statistical analyses conducted by the plaintiffs' expert witness (Grofman) in that case. Thus any controversy that may have existed over the use of these two methods for measuring racial bloc voting has essentially been laid to rest.28 The Court was unanimous in indicating that the reasons that blacks vote differently than whites cannot affect judgments as to whether or not the minority community is politically cohesive or whether voting is racially polarized.29 One issue that remains unclear, given the divided opinions in Gingles, however, is whether a multivariate analysis (i.e., an analysis that includes variables other than simply the race of the voters and the way in which they voted) can ever provide any useful information in a vote dilution suit. On one side of the controversy is Justice William J. Brennan, Jr., joined by Justices Harry A. Blackmun, Thurgood Marshall, and John Paul Stevens. In Part III-C of his opinion, Justice Brennan concludes that the reasons minority and white voters vote differently are of no relevance to a Section 2 inquiry and that therefore there should be no consideration of any factors other than the race of the voter and the selection of certain candidates in a racial bloc voting analysis. According to Justice Brennan, then, the only appropriate method of measuring the extent of racially polarized voting is a bivariate statistical analysis. But this section of the opinion does not command a majority (Justice Byron R. White did not join in this section of Justice Brennan's opinion). On the other side is Justice Sandra Day O'Connor - joined by Chief Justice Warren E. Burger and Justices Lewis F. Powell, Jr., and William H. Rehnquist - who in her concurrence held that evidence that a minority-preferred candidate was rejected by white voters for reasons other than race would be relevant to answering questions pertinent to vote dilution inquiries in general, even though the evidence could not be used as a rebuttal against a finding of racially divergent voting patterns or the political cohesiveness of the minority community.30 According to Justice O'Connor: Evidence that a candidate preferred by the minority group in a particular election was rejected by white voters for reasons other than those which made that candidate the preferred choice of the minority group would seem clearly relevant in answering the question whether bloc voting by white voters will consistently defeat minority candidates. Such evidence would suggest that another candidate, equally preferred by the minority group, might be able to attract greater white support in future elections, (p. 100)

The vote dilution standard in the post-Ginglzs era

75

Given the lack of a majority on this issue, the question occasionally arises as to whether a multivariate analysis is ever appropriate to determining whether a challenged election scheme violates Section 2. It is inevitably expert witnesses for the defendants who seek to enter this type of statistical analysis into evidence. For example, the experts for the defendant jurisdictions in Citizens for a Better Gretna v. City of Gretna (1987) and Romero v. City of Pomona (1989) sought to interject a multivariate analysis into evidence. Neither district court accepted the argument that a multivariate analysis yielded relevant information, however. It therefore appears that the lower courts have chosen to follow the plurality's approach to the use of multivariate analysis; that is, the lower courts have thus far concluded that a regression analysis that considers explanatory factors other than race is irrelevant to a Section 2 vote dilution claim. What are appropriate elections to study in a racial bloc voting analysis? A particularly important issue that remains unresolved has to do with the appropriate elections to examine in order to test for a pattern of racially polarized voting. For example, should the analysis be limited to elections that include minority candidates, or can so-called white-versus-white elections also be considered by the court in making its determination? Are the only pertinent elections those that were for the office under litigation, or can "exogenous" elections also be examined? To prove that minority-preferred candidates are usually defeated, it is necessary to know who the minority-preferred candidates are; as one court noted, "ascertaining whether legally significant white bloc voting exists begins with the identification of the minority members' 'preferred candidates' or 'representatives of their choice' " (Collins v. Norfolk, 1989, p. 1237). Does the term minority-preferred candidate refer simply to the candidate who received the majority (or the plurality) of the minority community's votes in an election, regardless of that candidate's race? Or is the race of the candidate relevant to whether a given candidate should be deemed the minority-preferred candidate? Essentially, the issue is whether or not the success of white candidates with more than 50 percent minority support in white-versus-white elections demonstrates that minority voters have the opportunity to elect candidates of their choice. The question of what type of elections to include in an analysis of racial bloc voting is important because the choice of elections often dictates the conclusions reached. In Buchanan v. City of Jackson (1988), for example, experts for both sides analyzed Jackson (Tennessee) City Commission elections from 1967 through 1983. The defendants' expert, Dr. Charles Bullock, conducted a statistical analysis of all fourteen elections since 1967; the plaintiffs' expert, Dr.

76

Minority Representation and Voting Equality

James Loewen, analyzed only the five contests that included a black candidate. Dr. Bullock concluded that black voters had been able to elect candidates of their choice in half of the elections and thus that the plaintiffs had failed to prove that minority-preferred candidates usually lose. Dr. Loewen concluded that all of the elections (i.e., all five) were racially polarized and that black voters in Jackson "have had essentially no opportunity to elect the candidates of their choice" (p. 1529). The district court determined that although "the court may consider all elections in determining whether members of a minority have less opportunity than other members of the electorate to elect representatives of their choice" (p. 1529), the court is not obligated to give all elections equal weight in its final determination. The district court found that "elections in which there was a black candidate are more probative in this case upon the issue of whether voting is racially polarized and whether the candidates favored by blacks are usually defeated" (p. 1531) and held that the at-large city commission system in Jackson had violated Section 2.31 In Part III-C of Gingles, Justice Brennan advocates a race-neutral approach to determining the minority-preferred candidate, arguing that "it is the status of the candidate as the chosen representative of a particular racial group, not the race of the candidate that is important" (p. 68). Justice Brennan's plurality opinion provoked strong disagreement on the Court; both Justice White, in his separate concurring opinion, and Justice O'Connor, writing on behalf of three of the other justices, rejected Justice Brennan's approach and took the position that the race of the candidate could (or perhaps, should) be considered relevant. Although most circuits have rejected Justice Brennan's approach, one court has accepted his race-neutral approach and considered all election contests including contests in which only white candidates competed for office.32 The Fifth Circuit has been the most adamant in its rejection of the stance that whiteversus-white contests are relevant. There has been no case in which the Fifth Circuit has consented to consider white-versus-white contests, and in several decisions the appellate court has specifically held that such contests are irrelevant. In Citizens for a Better Gretna v. City of Gretna (1987), the Fifth Circuit held that white-versus-white contests are not probative because when there are only white candidates to choose from, it is "virtually unavoidable that certain white candidates would be supported by a large percentage of... black voters" (p. 502).33 In so holding, the Fifth Circuit acknowledged the problem with designating a white candidate as "minority preferred" - the fear that courts will fail to recognize situations in which neither candidate was truly a candidate of choice. For instance, certain circumstances (such as the fact that a black has never been elected to office) may dissuade minority candidates from running for office. When no viable minority candidates compete for office, should the fact that

The vote dilution standard in the /?6>.s£-Gingles era

11

blacks vote anyway, and therefore some candidate will, simply by default, receive more of the black vote than any other candidate, automatically render that candidate the "minority-preferred" candidate? According to the Fifth Circuit in Gretna: "Gingles is properly interpreted to hold that the race of the candidate is in general of less significance than the race of the voter - but only within the context of an election that offers voters the choice of supporting a viable minority candidate" (p. 504). In Collins v. Norfolk (1989), the Fourth Circuit avoided specifically rejecting the argument that white candidates can be the minority community's candidates of choice but held that in an at-large contest in which voters can cast more than one vote, the fact that some white candidates received more than 50 percent of the black vote did not make these candidates the minority-preferred candidates when there were black candidates running who received a much higher percentage of the black community's votes: "Support for some successful candidates by a majority of minority voters in multimember district races does not prove that the successful candidates were the chosen representatives of the minority when a candidate who received much greater minority support was defeated" (p. 1240). Although the Ninth and Eleventh circuits have been less explicit than the Fifth Circuit in eschewing the argument that white-versus-white contests are relevant to a finding of racial bloc voting, no panel on either circuit has thus far considered white-versus-white contests in arriving at a decision regarding the second or third Gingles factor. In both Gomez v. City of Watsonville (1988) and Romero v. City of Pomona (1989), the Ninth Circuit focused only on election contests that included minority candidates to determine that voting was racially polarized (Gomez) and to conclude that blacks and Hispanics, combined, were not politically cohesive (Romero). The Eleventh Circuit, in Carrollton Branch of the NAACP v. Stallings (1988), reversed the judgment of the lower court and specifically held that the plaintiffs' expert's statistical analysis was sufficient to establish racial bloc voting; the expert in that instance examined only contests in which black candidates ran for office. However, at one point in the opinion the appellate court repeated Justice Brennan's contention that the race of the candidate per se was irrelevant. Meek v. Dade Metropolitan County (1990) elicited a similar response; although the focus of the court was on the voting patterns of minorities and whites when blacks and Hispanics ran for office, the court stated that "whether the candidate elected is the preferred candidate of a particular minority group turns on more than the candidate's race" (p. 1548). The only Eleventh Circuit decision to suggest directly that the race of the candidate is an important consideration is the opinion written by Judge Kravitch in Solomon v. Liberty County (1990): "On average, nearly eighty percent of whites in Liberty County have voted as

78

Minority Representation and Voting Equality

a bloc in elections involving black candidates for county office. . . . Thus, black voters have never had an opportunity to elect a black representative despite their manifest preference for those black candidates that have presented themselves" (p. 1021). The Tenth Circuit is the only circuit to express clearly its willingness to consider white-versus-white contests. In Sanchez v. Bond (1989), the district court found that Hispanics controlled the Democratic party in the county and had a strong say as to which candidates could run on the Democratic ticket; therefore, because three Democrats had been elected to the county commission, Hispanics were not being denied the opportunity to elect candidates of their choice. The court ignored the fact that each of the three successful candidates labeled minority preferred were Anglos competing in elections in which only Anglos ran. More importantly, the court also ignored the fact that no Hispanic had ever won a contested countywide election - including the three Hispanics who had run for county commissioner and received a majority of the Hispanic voters' support. The appellate court affirmed the lower court's conclusions, stating: There is no rule of law prohibiting the district court from examining those elections having only Anglo candidates. Such elections may be relevant and can be used to discern whether racially polarized voting exists and to measure the success of minority preferred candidates, so long as one of the Anglo candidates can be considered a preferred candidate of the minority group, (p. 1495) The Tenth Circuit has thus given its approval to an at-large election system that gives Hispanic voters an opportunity to elect candidates of their choice, but only as long as those candidates are Anglo candidates. The problem, of course, is that if the race of a candidate is not considered, then whenever minority voters cast ballots in white-versus-white contests, the winners in those contests can be designated minority preferred even if the minority community was not presented with a ' 'true'' candidate of choice. To focus on white-versus-white contests ignores the possible inability of the minority community to elect minority candidates despite its strong support. If the minority community's opportunity to elect is restricted to only those elections in which no minority candidates are competing, minorities are not being given the same opportunities as whites to elect their preferred representatives. This does not accord with the mandate of Section 2.34 We should note, however, that despite a plurality in Gingles emphasizing the race of the voter over the race of the candidate, the Supreme Court upheld the district court's finding of legally significant racial bloc voting based only on an analysis of elections in which blacks ran (neither expert in Gingles analyzed contests in which only whites competed for office). Justice Brennan never mentions the need to consider a broader range of elections than those contemplated

The vote dilution standard in the /?os£-Gingles era

79

by the lower court. Furthermore, none of the various concurring and dissenting opinions required information on any elections other than those reviewed by the lower court to reach conclusions about whether or not voting was polarized. Hence, implicit in the Gingles holding is the notion that black voters' preferences are determined from elections that offer the choice of a black candidate.35 What happens when there are only a limited number of minority candidates competing for office in the challenged jurisdiction? Can the court consider election contests other than those for the office under challenge? Generally, courts have opted for case-specific flexibility in judging the relevance of what the Fifth Circuit has termed "exogenous elections" {Citizens for a Better Gretna, 1987, p. 502) as long as the elections involve minority candidates (Grofman, 1990). In Gingles, for example, although most of the elections examined were for the office at issue, in several districts elections other than those for the state house were analyzed because there were not even as many as three house elections involving black candidates. In Citizens for a Better Gretna, the Fifth Circuit held that the district court was justified in considering exogenous elections, given the sparsity of relevant statistical data on aldermanic elections including black candidates: "Because the district court had statistical data of only two Gretna aldermanic elections to consider, it properly looked to voting patterns in two additional elections [Jesse Jackson's bid for nomination in the 1984 presidential primary and the 1979 election for Louisiana's secretary of state] in which Gretna voters had the opportunity to vote for a black candidate" (p. 503). In Westwego Citizens for a Better Government v. City of Westwego (1990), the Fifth Circuit reiterated its holding that exogenous elections were relevant in certain instances. In Westwego, no black had ever run for alderman, but the Fifth Circuit held that this should not preclude plaintiffs from making out a vote dilution claim, especially given that "blacks often do not sponsor a black candidate because of the very barriers to political participation that Congress sought to remove" (p. 1044).36 The only other circuit courts thus far to address the question of the relevance of exogenous elections, either explicitly or implicitly, are the Tenth and Eleventh circuits.37 In the Tenth Circuit, the district court in Sanchez v. Bond found that the success of an unopposed Hispanic candidate in an exogenous county election served as evidence that minority-preferred candidates were not usually defeated. The Tenth Circuit Court of Appeals affirmed this conclusion. The Eleventh Circuit has issued what appear to be contradictory signals with regard to the relevance of exogenous elections. In Carrollton Branch of the NAACP v. Stallings (1987), a challenge to Carroll County's single county commissioner form of government, the Eleventh Circuit held that the district court erred in relying on municipal elections as evidence of the possible dilutive effects of the county election scheme: "The [district court's] reliance on other

80

Minority Representation and Voting Equality

elections here is clearly misplaced" (p. 1558). Because the inquiry in this instance was directed at the county commission form of government, the focus of the court should be on "the ability of minority voters to elect candidates of their choice to political office on a county wide basis" (p. 1558). Other appellate decisions in the Eleventh Circuit have not directly confronted this issue;38 however, subsequent district court decisions in this circuit have indicated a willingness to consider exogenous elections. In Bradford County NAACP v. City of Starke (1989), for example, the district court held that atlarge elections for the city commission violated Section 2, even though only exogenous elections could be analyzed: "Because there have been only three elections involving black candidates in the City of Starke and because it was not possible to perform a statistical analysis of those elections [all city voters in city commission elections vote at a single location], the Court has relied, in part, on statistical evidence from non-city elections in reaching its conclusion" (p. 1539). In our view, case-specific flexibility appears to be the only sensible route to take. Surely the results of elections for the office at issue are the most relevant to a Section 2 suit; just as certainly, however, given such special circumstances as the absence of viable minority candidates or a jurisdiction with a very limited number of precincts, exogenous elections ought to be considered. If the courts were to fail to consider exogenous elections in such circumstances, plaintiffs could be precluded from relief simply because no minority candidates had run for office or the jurisdiction had too few voting precincts to conduct a statistical analysis. This is hardly in keeping with the spirit of Section 2 or the holding of the Supreme Court in Gingles. Our review of the vote dilution standard in the post-Gingles era leaves us with two somewhat contradictory conclusions. On the one hand, Gingles has provided clear guidelines for the decision in many subsequent cases and for the settlement of many other potential cases. In the context of multimember or at-large elections, Gingles has proved a resounding success. That there remain a few areas of ambiguity should not detract from this major accomplishment of the Court. Nor should it obscure the reality that a very large number of discriminatory systems have been dismantled, leading to the election of a considerably greater number of minority officials (Grofman and Davidson, 1992). On the other hand, Gingles speaks less clearly to a new class of cases that has begun to arise and that is likely to dominate court agendas in the 1990s. As the more obvious discrimination of the multimember type of district declines precisely because of Gingles - the more difficult issues of vote dilution in singlemember districts will come to the fore, matters less clearly addressed by Gingles. Consequently, we devote Chapter 5 to a discussion of the relevance and appli-

The vote dilution standard in the posJ-Gingles era

81

cability of Gingles to the single-member district context and to some of the additional issues that will arise in this new venue. We should also note, of course, that although Gingles has provided the means for determination and resolution of discriminatory effects in the multimember context, this does not mean that matters have been resolved in all such jurisdictions.39 Cases and out-of-court discussions will continue for the foreseeable future about possible discrimination in at-large and multimember elections. Moreover, despite important legal differences in the single-member context, many of the same technical considerations will occur as in multimember districts. Therefore, before looking to the future, we provide a detailed presentation in Chapter 4 of the standard social science methods used in vote dilution cases, especially for the determination of racially polarized voting.

Defining and measuring racially polarized voting and other elements of the totality of the circumstances

As early as the mid-1970s, racially polarized voting was recognized as an important evidentiary element in challenges to at-large/multimember districting systems as well as in some cases involving retrogression and annexation (e.g., Beer v. United States, 1976; Wallace v. House, 1975). By the mid-1980s, it was identified by the Eleventh Circuit Court of Appeals as "the keystone of a dilution case" {United States v. Marengo County Commission, 1984, p. 1566), and as discussed in Chapter 2, the Supreme Court in 1986 made it the foundation for two of the three prongs of the Gingles test (Thornburg v. Gingles, 1986). Yet racially polarized voting is not self-evident. Because individual voting records are secret, one cannot immediately determine whether minorities and whites vote for the same or for different candidates. Some analytical techniques are required. Moreover, once the data are in hand, they must be interpreted. Just what analytical techniques should be used and how the data are to be interpreted have sometimes been major bones of contention between plaintiff and defendant attorneys and among the social scientists testifying on opposite sides in voting rights cases. With the ruling in Gingles that bivariate analysis is an appropriate tool and that data obtained by "ecological" analyses are acceptable, many of these disputes have been settled. Consequently, what we present here are now generally regarded as the standard methods to be used for determining the presence of racial polarization. It would be incorrect, of course, to say that all disputes about racial bloc voting have been settled. As we noted in Chapter 3, recent cases have involved arguments about which elections to analyze, whether multiple minority groups can be combined, and so on. Nevertheless, the approach to the basic problem determining how whites and minorities vote in the absence of information about individual voting behavior - is now relatively well agreed upon. It is this analytical process that we describe here. We then consider potential problems that may arise, such as the absence of racially homogeneous or nearly homogeneous geographic areas or the presence in the jurisdiction of more than one minority 82

Defining and measuring racially polarized voting

83

group of significant size. We also consider arguments that the standard ecological regression methods are unreliable. We emphasize that the methodology we discuss is intended for general use. It is not a methodology for plaintiffs' experts or a methodology for defendants' experts. It is simply the standard methodology. Defining racial bloc voting Of the two problems - obtaining data on white and minority voting and interpreting those data - the latter was handled with dispatch by the Court in Gingles, and we deal with it first. There were three obvious approaches open to the courts. First, they could focus on the correlation coefficient obtained when the support rate for minority candidates was regressed on the minority percentage in the voting precinct.1 Experts in some cases treated correlation levels of .7 or above as prima facie evidence of bloc voting. Some experts who used the correlational approach also testified about the "statistical significance" of the evidence for polarization, with "f" statistics and similar measures being offered in evidence.2 Second, courts could focus on percentages showing the degree of "own race" voting - for example, the extent to which black voters vote for black candidates and white voters vote for white candidates. This approach has a great deal of intuitive appeal. An obvious question, however, is what level of difference in support proved polarization. Many experts proposed to look at the sum of ownrace voting, with the defendants' witnesses sometimes contending that voting should not be considered polarized unless the sum exceeded 180 percent, and the plaintiffs' experts arguing for lower thresholds.3 Third, courts could require that experts explain the vote by demonstrating that racial animus was driving the choices of voters, especially nonminority voters. Because this required something more than a showing of polarized voting, it was the defendants who were especially interested in this argument. Their expert witnesses proposed two ways of approaching this question. The first was to require evidence of a racial backlash, such as racial campaign appeals or especially high levels of white/Anglo turnout in elections involving minority candidates.4 The second approach was to require the use of "multivariate regression" - a statistical method that helps determine whether one variable here race - makes an "independent" contribution to voting decisions once other factors such as newspaper endorsements, incumbency, campaign spending, and the socioeconomic characteristics of the voters are taken into account.5 In Gingles the Court opted in favor of the commonsense notion of racially polarized voting that had characterized the early cases, including Beer v. United

84

Minority Representation and Voting Equality

States (1976). As described in Chapter 2, the Court accepted the definition offered by the plaintiffs' expert witness (Grofman) in this case that voting was polarized when there is " 'a consistent relationship between race of the voter and the way in which the voter votes,' or to put it differently, where 'black voters and white voters vote differently' " (Thornburg v. Gingles, 1986, p. 53, n. 21). As also noted there, the Court made a distinction between racial polarization per se and polarization that was of legal significance, leading to the second and third prongs of Gingles. A final relevant point noted earlier is that Justice William J. Brennan, Jr., also decisively rejected the claim that race must be shown to be the primary determinant of voting behavior. More generally, he rejected claims like those that had been accepted by lower courts in the Norfolk and Fort Lauderdale cases that it was necessary to explain voter behavior or to show that voters possessed racial animus. The Gingles Court also accepted Grofman's judgment about the statistical significance of the correlation coefficients obtained in ecological regressions, citing testimony that the data "reflected positive relationships and that the correlations did not happen by chance" (p. 53, n. 22). Thus, the thresholds for statistical significance that were accepted at trial became incorporated, in effect, into the voting rights case law. The Gingles ruling therefore settled many of the key legal questions about how to define and measure racially polarized voting. The remainder of this chapter is devoted to an explanation of the standard procedures for determining the voting behavior of minorities and whites in the absence of racially separate vote counts, a discussion of some common errors made in using these procedures, an examination of special problems that are sometimes encountered, and some comments on recent attempts to undermine the standard approach. Measuring individual voting behavior Because voting is secret and relevant survey data are rarely available,6 differences in group voting patterns can usually be established by either of two techniques: homogeneous precinct analysis (also called extreme case analysis) and ecological regression. The former analyzes precincts (or other small units) that are racially homogeneous; the latter compares, graphically and statistically, the votes for minority candidates in each precinct with the racial composition of that precinct. The Supreme Court's decision in Gingles provided legal legitimacy to the use of these techniques. In fact, in cases since Gingles, the findings of expert witnesses using these procedures have only rarely been rejected by federal courts, and rejection is almost unheard of when proper cautions (described later) have

Defining and measuring racially polarized voting

85

been taken. Indeed, it is now commonplace for expert witnesses for both defendants and plaintiffs to present nearly identical calculations, though of course they disagree on what significance should be attached to the findings. Homogeneous precinct analysis In a voting unit, such as a precinct, that is entirely composed of one race or ethnic group, one can obviously make inferences about how members of that race or group voted - despite the lack of direct access to the ballot. Thus, for example, if the voters in a precinct with only black residents cast 80 percent of their ballots for a black candidate and 20 percent for a white candidate, it is obvious that that is how black voters in that precinct voted. Often, of course, precincts are not composed entirely of one racial or ethnic group. Nonetheless, if there are precincts that are overwhelmingly (say, 90 or 95 percent) composed of members of the same race, one can be extremely confident of the voting behavior of members of that group in those precincts. By using precincts (or other small units), an analyst can usually find at least a few areas that are nearly homogeneous. If precincts are somewhat less homogeneous (say 80 to 90 percent of one race or ethnicity), one can generally use election outcomes to derive appropriate upper or lower bounds. With nearly homogeneous minority precincts, one can use the outcomes as a lower bound on the support given by minority voters to the minority candidate; in doing so, one uses the almost certain expectation that levels of nonminority support for the minority candidate will not exceed that of minority voters. Similarly, in the nearly homogeneous white/Anglo precincts, one can take the level of support for the minority candidate as an upper bound on support from nonminority voters; one assumes that these precincts probably contain some minority voters whose support of the minority candidate is being attributed to the nonminority voters. Of course it may be argued that voters who live in racially homogeneous precincts vote differently from those who live in more heterogeneous areas; blacks in overwhelmingly black precincts, for example, may have lower income and education levels than do blacks living in more nearly integrated areas, and those differences may affect their voting behavior. One way to check the plausibility of this assumption is by comparing results from a homogeneous precinct analysis with those from an ecological regression (which, of course, includes data from the racially mixed precincts). Ecological regression There are a number of variants of ecological regression. We shall show for the case of a single-member district contest involving at least one black candidate

86

Minority Representation and Voting Equality

and at least one white candidate - for which registration, voting-age population (VAP), or total population data by race by precinct are available - the most common method of ecological regression. This method generates not only an estimate of the voting behavior of the electorate by race but also an estimate of voter turnout by race.7 The following illustration is for the case in which registration data by race are available, but the methodology will be identical if the only available racial data on a precinct level are population or voting-age population; one simply substitutes population (or VAP) for registration in the definitions given.8 We illustrate the data with two mutually exclusive and jointly exhaustive groups, to which we refer as black and white. The analysis for Spanish origin/surname versus non-Spanish origin/surname is parallel to that for black versus white.9 Later we address the complications created when there are multiple groups covered by the Voting Rights Act that must have their voting behavior disaggregated for the purposes of a particular lawsuit. Let x 1 - x Pw

= the proportion of registered voters who are white = the proportion of registered voters who are black = the proportion of registered voters who vote for the white candidate = the proportion of registered voters who vote for the black candidate.

PB

All these four variables can be determined directly from election returns, registration data from local election boards, or U.S. census data matched to the precinct level; that is, their values are, in principle, directly observed. Let Pww PWB PBW PBB

=

me

proportion candidate = the proportion candidate = the proportion candidate = the proportion candidate.

of white registered voters who vote for the white of white registered voters who vote for the black of black registered voters who vote for the white of black registered voters who vote for the black

These four variables are unobservable. To ascertain their values, it is necessary to make inferences from the variables that we can directly observe. It is true by definition, for the electorate as a whole, that Pw

= xiPww) + (1 -

x)(PBW)\

Defining and measuring racially polarized voting

87

that is, Pw

=

(Pww

PBW)X

~

+

PBW

Let the superscript (/) denote the value of a variable in the /th precinct; thus, for example, Pw{i) is the value of Pw in the /th precinct.10 If we regress Pw(l) on x(l\ then the slope and intercept of the regression line will have a straightforward "natural" interpretation in terms of the proportion of white (black) registered voters who vote for the white candidate. The regression line can be stated as

Then, because of the identity dictating that the vote for a candidate is simply the sum of that candidate's support from black and white voters, it is necessary that m \ = P\vw ~ PEW and

We may rewrite these equations to solve for PBW and Pww> PBW

t0

obtain

~ bi

and Pww = ml + bv Thus, we can use the precinct-level data on observed voting behavior and known minority percentages to estimate the unobservable variables PBW and F w . Similarly, if we regress P B (l) on x(l\ the result will be PWB

=

m 2 + b2

and PBB

=

b 2.

(We could alternatively do a regression using the proportion of registered voters who are black.11) Let us specify a new notation such that P ww

=

Pww'\Pww ~*~ PWB)

P BB

=

PBB'^PBB

P WB ~ P BW

=

PWBI\PWW PBWI\PBB

~*~ PBW) ~^~ *WB) ~^~ PBW)

These primed variables, which we obtain by combining the information derived from the two regression equations (PW(i) o n x(l) a n d PB({) o n x(u)t0 obtain values

88

Minority Representation and Voting Equality

of Pww, PBB, and so forth, now give us the estimated white vote for white candidates as a proportion of the total white vote, the estimated black vote for black candidates as a proportion of the total black vote, and so on. The unprimed variables express support for candidates as a proportion of the registrants; the primed variables express support for candidates as a proportion of the voters. It is the latter that are of direct interest; we estimate the former in order to obtain the latter. Of the four primed variables, it is P'ww and P'BB in which the analyst is most interested, as these denote the proportion of votes that go to candidates of the voters' own race. The following additional notation will be of use later: P'w

= PW/(PW + PB) and P'B = PB/(PW + PB).

These "primed" variables give us the white (black) share of the actual vote. Of course, P'w = 1 - P'B. Estimating turnout Let Tw TB

= proportion of white registered voters who vote (i.e., white turnout as a proportion of white registered voters) = proportion of black registered voters who vote (i.e., black turnout as a proportion of black registered voters)

and T

= the proportion of registered voters who vote.

The two-equation estimating procedure allows us to estimate (differential) turnout among white and black registered voters,12 as *w

=

-*S

=

*WB

+ Pww

and PBB +

PB\V-

Commonsense checks on estimates of bloc voting Comparing the results of homogeneous precinct analysis and ecological regression The methods of homogeneous case analysis and ecological regression each have advantages and disadvantages. The principal advantages of homogeneous case analysis are its simplicity and the fact that it is based directly on the behavior of voters of a given race, thus requiring no statistical inference. The principal

Defining and measuring racially polarized voting

89

disadvantages are that it may not always be possible to use it because of the absence of sufficiently homogeneous precincts and it does not use all the data that are available. The principal advantage of ecological regression is that it uses the information from all the precincts, both those that are racially homogeneous and those that are racially mixed. The principal disadvantages are the possibility of errors owing to ecological inference and the fact that trial judges may not be familiar with it other than from the discussion in Gingles and other cases. Because each of these methods has pluses and minuses, a standard procedure is to use them both. By comparing the actual results in homogeneous precincts with the estimates of voting derived from the double-equation ecological regression method, one can customarily ascertain the extent (if any) to which voters of a given race as a whole (including voters of that race living in racially mixed precincts) vote differently from voters of that same race in racially homogeneous precincts. If voters of a given race in nonhomogeneous precincts vote differently from voters of that race in homogeneous precincts, the estimates derived from ecological regression will differ from the voting behavior observed in the homogeneous precincts.13 The homogeneous case estimates often appear to reflect less polarization than the ecological regression estimates do because the precincts used are not actually 100 percent racially homogeneous. In practice, however, if the homogeneous precincts are close to 100 percent of one race, then in our experience, the differences between the estimates derived from ecological regression and the observed results in homogeneous precincts are likely to be minimal (see, e.g., Grofman, Migalski, and Noviello, 1985). Both homogeneous case analysis and ecological regression methods should be used if possible, as each compensates for the potential flaws in the other. Most important, when combined with homogeneous case analysis, ecological regression analysis normally allows us to avoid the "ecological fallacy" (Robinson, 1950), one common form of which is the error of attributing the average behavior of voters in a given geographic area (ecological unit) to all voters in that area - here, assuming that all voters in a given precinct vote alike. Note that even though one uses the value of the regression coefficient in a precinct that is racially homogeneous to estimate the voting behavior of that racial grouping, the value in question is actually an estimate of the average voting behavior of all members of that group, as it is based on information from all of the precincts. Even so, we do not regard it as necessary to do ecological regressions if a very clear pattern of polarization can be shown by comparing votes in precincts that are overwhelmingly composed of minority voters with votes in precincts that are preponderantly white/Anglo, especially if a substantial number of voters reside in homogeneous or nearly homogeneous precincts. Homogeneous precinct analysis is considerably less costly and less time-consuming than using regres-

90

Minority Representation

and Voting

Equality

sion methods, so to require the latter to prove vote dilution might place an undue and unnecessary burden on minority plaintiffs. We briefly illustrate the application of our methods for determining levels of racial polarization to data used in Gingles. In a 1982 Senate election in Mecklenburg County, North Carolina, the values for the sole black candidate (Mr. Polk) were mx bY m2 b2

= .28 = .03 = —.31 = .45

Hence, PBW p

= b, = .03

ww = m\

+ b\ = -31

= m2 + b2 = .14

PWB

PBB

Hence

= t>2 = -45

06 = whs p.

J*

WB

31

~ .14 + .31 " = -94

The regression estimate for the proportion of the black vote that went to the black candidate is .94; the homogeneous precinct analysis generated a value of .86. The regression estimate for the proportion of white vote that went to the black candidate was .31; the homogeneous precinct analysis was .30.14 It is apparent that homogeneous and regression estimates are very similar - indeed identical in their first significant digit. As expected, the regression estimate was marginally higher than the homogeneous precinct estimate, as the "homogeneous" precincts (90 percent or more black registered voters) contain some nonblack voters.

Defining and measuring racially polarized voting

91

Measures of goodness of fit for ecological regression: the coefficient correlation Correlation coefficient. When ecological regression is used, statistical measures can be generated for the likelihood that an observed linear relationship between candidate support and the racial composition of precincts is due to chance. The single most important measure is the correlation coefficient (Pearson's r). It tells us the extent to which data points fall on a straight line. As noted previously, both the trial court in Gingles and the Supreme Court accepted the view of the plaintiffs' expert (Grofman) that "correlations above an absolute value of .5 are relatively rare and correlations above .9 extremely rare" {Gingles v. Edmisten, 1984, p. 368, n. 30). Statistical significance. An analyst can also look at the statistical significance of the correlation coefficient. This measure tells us the likelihood that a pattern that looks as much like a straight line as that found in the data could have occurred by chance alone. If, for a given election, the observed value of r has a statistical significance of .01 or less, that is, a likelihood of occurring by chance alone of one in one hundred or less, courts have taken this to be evidence that race was a factor in that election.15 The commonsense (interocular) test. There is no substitute for careful visual inspection of the data in graphical form, that is, a plot of votes for the minority candidates against percentage minority in the precinct. In a number of recent cases the standard methods of ecological regression and homogeneous precinct analysis have been subject to attack on a variety of statistical grounds, by expert witnesses testifying for defendant jurisdictions. In several of those cases the courts have responded to these statistical criticisms by taking the commonsense view that the "interocular" test is often the best one - that is, if when one looks at tables or graphs showing the relevant election data, the pattern of polarization jumps up and hits one between the eyes, then there is polarization, esoteric quibbles to the contrary notwithstanding. This view was perhaps expressed most directly by a federal district court in Jeffers v. Clinton (1989).16 In inspecting the data visually one can also look at which precincts (if any) give rise to "outliers" from the best-fitting regression line. Sometimes knowing where the predictions of ecological regression fit the data least well allows a locally knowledgeable informant to suggest reasons for cross-precinct variations in predictive fit. (For example, if census data are being used to define the independent variable, there may be particular precincts that have

92

Minority Representation and Voting Equality

changed dramatically in their racial composition since the time of the census.) The fact that some precincts are outliers to a general pattern of polarization does not invalidate the finding that there is a pattern of polarization, as it is the "average" behavior of groups (in the relevant elections) that is most relevant.17 Comparison with single-equation results. The single-equation ecological regression involves regression percent of the vote for the (minority) candidate(s) on minority registration or population. It is useful to generate single-equation estimates of P'ww and P'BB as checks on the estimates of these quantities from the double-equation method, but one must recognize that the single-equation approach is less reliable than the double-equation approach. Nonetheless, the single-equation approach allows us to generate a useful correlation to use in showing the degree to which the data can be characterized by a linear relationship between support for minority candidates and minority percentages in the electorate. Checks against known data. One of the most important commonsense checks of estimates derived from homogeneous precinct or ecological regression analysis is a comparison of estimates with known jurisdictionwide quantities such as total turnout or the actual vote share of the minority candidate(s). One must keep in mind, of course, that the match between estimated values and actual values is never perfect and that the degree of fit is not uniform across all geographic areas. Nonetheless, if at the jurisdictionwide level, estimates of known values show a discrepancy of more than, say, 10 percent, there is reason to scrutinize the results with special care. Existence of a pattern The existence of some crossover voting does not in itself show the absence of legally significant polarization. A few instances of even majority-white support for some black candidates cannot outweigh a general pattern of white unwillingness to support black candidates who are the clear choice of black voters, nor can the fact that some black voters have voted for white candidates (especially incumbents). As the court stated in Gingles: It is "the usual predictability of the majority's success [that] distinguishes structural dilution from the mere loss of an occasional election" (1986, p. 51). Normally one would like to see evidence of polarization/cohesion in at least three elections for the office under challenge in which there were viable, minority-supported candidates. However, if the nature of the election scheme tends to discourage minority candidacies, this may be impossible. In such instances,

Defining and measuring racially polarized voting

93

it is necessary to look at elections exogenous to those at issue in the litigation. Just which elections to consider has been at issue in recent federal court cases (see Chapter 3). Problems of data availability Turnout versus registration versus CVAP versus VAP versus population as the independent variable Homogeneous precinct analyses must be interpreted with some care if the data used to define homogeneity are not the percentage of minority voters in a precinct. The percentage of minority voters in the precinct can be much lower than the minority population if eligibility rates, determined by voting age and citizenship, or registration or turnout rates are significantly different for minority and nonminority populations. If the percentage of minority population in an ostensibly overwhelmingly minority precinct overstates the percentage of minority voters in that precinct, then - if one assumes that minority voters vote at least as heavily for the minority candidate as do nonminority voters - it must also be true that the level of support for minority candidates in such precincts understates the true level of minority support for minority candidates. For example, if black voter support for the black candidate is at the 90 percent level and white voter support for the black candidate is only at the 20 percent level, and if only 50 percent of black registrants but 90 percent of white registrants vote, then a "homogeneous" precinct that is 90 percent black in registration will cast only (.45 X .9 + .09 X .2) / (.45 + .09) = 78.3% of its votes for the black candidate. Thus, in this instance, simply assuming that the vote in the homogeneous precinct tells us the level of black support for the black candidate would significantly underestimate the true level of black support for the black candidate (78.3 percent compared with 90 percent). Special data problems in dealing with Hispanic populations Because of the high proportion of Hispanic noncitizens in some areas, in general, one would prefer to do regression analyses using Spanish-surname registration data,18 or if they are not available, with (estimated) citizen voting-age population data broken down by race and ethnicity rather than only population or even voting-age population data. If only voting-age population data are available and if it is known that the area in question has a high proportion of noncitizens, then one must be especially careful in interpreting the results of ecological

94

Minority Representation and Voting Equality

regressions, as the independent variable may be a poor indicator of the eligible electorate. In such situations, homogeneous precinct analysis may be especially useful - if there are precincts that are very highly homogeneous. (A precinct that is almost 100 percent minority in terms of voting-age population is still very heavily minority in registration and turnout.) In this special case, homogeneous precinct analysis may be more informative than ecological regression is, although even in this situation we still recommend using and comparing the results of both methods. Moreover, as noted previously, to the extent that homogeneous minority precincts contain an admixture of nonminority population, the overall support in such precincts given to minority candidates understates the support being given to such candidates by minority voters.

Problems in matching census data to precinct boundaries As a result of extensive efforts by the Bureau of the Census to obtain geographically based descriptions of voting precincts from the states, the 1990 TIGER Files normally allow analysts to meld census data (based on census units such as census blocks or what are called VTDs) and electoral data (based on voting precincts) into one database. Previously, analysts often had to spend a great deal of time and effort looking at detailed maps in order to match census and electoral geography. Unfortunately, even in the 1990s, the problem of census-to-precinct matchings is unlikely to disappear completely. First, in large cities there may be voting precincts that split census blocks (the census block is the smallest unit of census geography). This requires the analyst to combine voting precincts to create new "superprecincts" in order to mesh data at the two levels or somehow to estimate the census characteristics of portions of a block. Second, analysis is sometimes done at the tract level, either for convenience in a jurisdiction with a large number of census blocks or because there are some data available only at the tract level. But there may be some tracts that split precinct lines. Here, unless one is able to solve the problem by going back to block-level information, one must estimate the census characteristics of a precinct by combining information about each of the tracts in which it is located or use only those precincts that are coterminous with tracts. Moreover, even if neither of these problems is present in 1991, as precinct lines change over the course of the decade (or if one needs to match current precinct lines with those of earlier periods), it is likely that such problems will emerge. Still another type of matching problem is simply knowing how to translate old precinct geography into current districts. Sometimes tracking down old precinct boundaries is not easy. In short, some map-reading skills are still necessary.

Defining and measuring racially polarized voting

95

A limited number of homogeneous precincts Although it is desirable to have as many homogeneous precincts as possible in order to observe directly the voting patterns of different groups, there is no minimum number of such precincts. In our own expert-witness testimony, we have reported homogeneous precinct data even in situations in which there was only one homogeneous precinct. The purpose of homogeneous case analysis is to provide direct and irrefutable evidence regarding how members of a given race in a given precinct have voted. As previously noted, the validity of the claim that the voting behavior in such homogeneous precincts is not "typical" of the group can be directly examined by comparing the behavior in the homogeneous precincts with the estimate of the "average" behavior of the group derived from ecological regression and by other of the checks described in this chapter. The absence of homogeneous precincts for one group Sometimes there are no precincts with over 90 percent minority voters, and there may even be no precincts with over 80 percent minority voters; but it is rare that there are no homogeneous white/Anglo precincts. In instances in which there are homogeneous precincts for only one group, an analyst may use what one of us calls mixed-case analysis (Grofman and Noviello, 1983). The basic idea is simple. One can use the homogeneous precincts of the one group to derive an estimate of how the members of that group vote and then substitute that estimate into the most homogeneous precincts of the other group, apportioning the still unexplained votes among the members of the other group. For example, if we observe that in very homogeneous white precincts, only 20 percent of the votes went to the black candidate, we can take P'WB to be no more than .20. If the most heavily black precinct is 60 percent black and it gives 50 percent of its vote to the black candidate, we can assume that no more than 8 percent of the vote in the precinct (.20 X 40) is from whites voting for the black candidate. That leaves forty-two percentage points of support for the black candidate that must be coming from the 60 percent of the precinct that is black. Hence, black support for the black candidate is estimated to be at least 70 percent (42/60). No homogeneous precincts for either group In jurisdictions with only a few precincts, there may be no precincts that are overwhelmingly minority or nonminority. In such instances, one might wish to use what one of us (in testimony in Gomez v. City ofWatsonville, 1988) referred

96

Minority Representation and Voting Equality

to as a "most different case analysis," comparing the level of support for minority candidates in the most heavily minority and the most heavily nonminority precincts. Though one must recognize that the nonhomogeneity of the precincts masks the true magnitude of racial differences in voting, such a comparison may nonetheless yield usable inferences about differences in voting between groups. In Gomez, the court accepted testimony about racial bloc voting based in part on such a comparison.19

Special issues Estimates of Spanish-origin population Estimating Spanish-origin registration. As previously noted, in doing ecological regressions to estimate Hispanic voting behavior, one would prefer to have the percentage of Spanish-surnamed registrants/voters as the independent variable, rather than simply the percentage Hispanic of the voting-age population. To determine the percentage of Spanish-surnamed registrants/voters, one can use the Census Bureau's list of Spanish surnames, as the accuracy of that list has been accepted by numerous courts (e.g., in Garza). Of course, the match between Spanish surnames and Spanish origin (i.e., persons who self-identify themselves as such on the census forms) is not perfect. There can be both "Type I " and "Type II" errors - that is, individuals with Spanish surnames who are not of Spanish origin and individuals without Spanish surnames who are of Spanish origin. The most detailed analysis of the correspondence between Spanish surname and Spanish origin was done by the demographer William O'Hare in connection with his testimony in Garza. With the help of the Justice Department, O'Hare obtained a special census run that matched, at the census tract level, the Spanishsurname list with the names of those known from their census returns to be Spanish-origin citizens of voting age living in Los Angeles County. In this fashion, he was able to specify tract-specific Type I and Type II error rates. He then used corrections in the Spanish-surname percentages obtained by using the census identification of the names of voting-age citizens as the basis for an estimate of Spanish-origin registration. Regressions run with Spanish-surname registration as the independent variable and those run with O'Hare's estimate of Spanish-origin registration differed minimally. Differences in the derived estimates of parameters such as F ' w were rarely above three percentage points and usually much smaller. Differences in correlations were virtually nonexistent. Based on O'Hare's findings, it is almost certainly unnecessary to attempt to develop an estimated Spanish-origin registration figure for purposes of regres-

Defining and measuring racially polarized voting

97

sion analysis; for that purpose Spanish-surname registration data usually are sufficient. We should note, however, that O'Hare did find that in heavily Hispanic areas, the Spanish-surname percentage marginally underestimated the Spanish-origin percentage. Estimating Spanish-origin citizen voting-age population. If Spanish-surname registration data are not available, the next-best independent variable for purposes of bloc-voting analysis is Spanish-origin citizen voting-age population (CVAP) data. Unfortunately, at the levels of census geography needed to construct precinct-level data files, such data have not been released until several years after the census is conducted. Thus, during the period immediately after a decennial census, if Spanish-surname registration data are not available, it may be necessary to estimate Spanish-origin CVAP by combining current census data on Spanish-origin voting-age population with information on rates of citizenship within Spanish-origin and non-Spanish-origin voting-age populations obtained from earlier periods, or by applying jurisdictionwide citizenship rates to lowerlevel units. Because citizenship rates among Hispanics may vary considerably within a given jurisdiction, if census data on Spanish-origin CVAP are not available, estimating Spanish-origin CVAP must be done with care. Party primaries If there are primaries, it is important to look at patterns of polarization in them. In many states the victor in the primary of the preponderant party in the state is all but assured victory in the general election.20 For Hispanics, Spanishsurname matching usually enables one to determine the Hispanic proportion of the registrants of a given party in those states that maintain party registration lists. If registration data by race and party are not available for blacks (as is usually the case, except in the six states where such data are kept separately for blacks and whites21), then one must estimate the black proportion of the party registrants/primary-voting electorate if one wishes to analyze election returns from partisan primaries to determine racial voting patterns. Given known patterns of party identification, blacks are overrepresented in Democratic primaries relative to their overall numbers and are virtually unrepresented in Republican primaries. Hispanics, too, are overrepresented in Democratic primaries relative to their overall numbers of eligible voters, although usually not to the same extent as blacks are. Multimember districts without numbered places In multimember elections without a numbered place system there may be more than one candidate who is a minority candidate of choice. Here one looks to

98

Minority Representation and Voting Equality

see whether minority-supported candidates would be elected if only nonminority voters were voting. Thus, for example, in a situation with only black and white voters and a single black candidate, if white voters (on average) ranked the black candidate fifth and black voters (on average) ranked that candidate first (based on the number of votes for the various candidates), that would give rise to racially polarized voting if there were four or fewer candidates being elected in the multimember district, but not otherwise. Racial polarization can occur even if white voters vote for some minority candidates, as long as they are not as willing to elect as many minority candidates as would be the minority community.22 Problems in interpreting racial polarization Overstating or understating the importance of the correlation coefficient It is important not to confuse a high correlation with a high magnitude of political cohesion or with a lack of nonminority support for minority candidates. A high r value merely indicates that the observed relationship between race and voting patterns is sufficiently consistent across precincts (and linear in nature) that it is not likely to have been a matter of chance. For example, in a singlemember district election with two candidates, one black and one white, if in each precinct 65 percent of black voters and 63 percent of the white voters voted for the black candidate, one would obtain a perfect linear fit (r = 1.00), even though there was almost no difference in the voting patterns of whites and blacks. The magnitude of differences is best judged from the slope and intercept of the regression line. Thus, looking at correlations is not a substitute for looking at estimates of P'ww an<^ P'BBOn the other hand, some expert witnesses have wrongly claimed that the correlation coefficient is irrelevant. In Garza, in situations in which exit poll or other data provided an independent estimate, experts testifying on behalf of the county claimed that erroneous predictions resulting from regression equations in which the correlations were miniscule demonstrated the failure of the ecological regression methodology. A better interpretation is that the miniscule correlation coefficient indicated that evidence for polarization was absent or inconclusive in these particular elections. Reporting the wrong correlation The correlation that best captures the relationship between race of candidate and race of voter is that obtained when we regress the estimated values of P'B (see n. 12) on the actual values of that variable. In our rather extensive experience,

Defining and measuring racially polarized voting

99

however, if one simply regresses P'B on x, that is, if one performs a singleequation ecological regression, the correlation one obtains is likely to be very close to the correlation obtained from regressing the double equation-derived values of P'^ on the actual values of that variable. Thus, many experts who report correlations derived from analyses of racial polarization simply present the correlation from the single-equation ecological regression of P'B on x. Indeed, all of the correlations reported in Gingles were obtained in that fashion. That correlation, of course, tells us the relationship between the minority candidate's share of the votes in a precinct and the minority percentage in that precinct. It is, however, inappropriate to use the correlation from either of the two equations used in the double-equation approach as the reported correlation for racial bloc voting. Those correlations generally understate (often considerably) the extent to which there is a consistent pattern of polarization. The reason is that the two-equation approach is sensitive to variations in turnout across precincts as well as to variations in support for the minority candidates across precincts; the presence of turnout variations typically lowers the correlations in each of the equations used in the double-equation approach from the correlation in the preceding single-equation approach or from the regression of actual on predicted values of P'B (see Loewen and Grofman, 1989). Characterizing polarization or minority cohesion incorrectly in multicandidate elections When there are candidates from more than one racial or ethnic group and at least two candidates are of the same race or ethnicity, then determining racial polarization or minority cohesion requires the analyst, in general, to look at both the combined votes for the set of minority candidates and the combined votes for the set of all nonminority candidates as well as the estimates of minority and nonminority votes for each candidate. If one does not do a candidate-specific analysis, the plurality choice may not be identified. On the other hand, when there is more than one minority candidate, these candidates are competing with one another as well as with nonminority candidates. In such situations, failure to look at the level of minority support for the combined set of minority candidates may give a very misleading picture of minority political cohesion. For example, no single minority candidate may be given more than plurality support from minority voters, even though minority voters give majority support to the set of minority candidates. In addition, some seemingly idiosyncratic variations in the support levels of any given minority candidate may vanish when one looks at the combined level of support for the set of minority candidates in that precinct. In such situations, looking at the

100

Minority Representation and Voting Equality

combined vote for all minority candidates sometimes provides a more accurate picture of minority cohesion than what is derived from candidate-specific analysis does.23 Using multivariate methods inappropriately At least in those situations involving only two groups, for example blacks and nonblacks or Hispanics and non-Hispanics, the methods that are appropriate for estimating the extent of racial bloc voting are bivariate methods intended to identify differences in the two groups' voting behavior, rather than multivariate methods that attempt to explain voting differences (see Chapter 3). Some scholars, however, have recently proposed using multivariate methods not to explain polarization but to estimate it better by looking at differences in voter behavior across subgroups. In Garza, for example, expert witnesses for the county attempted to estimate by means of ecological regression methods the voting behavior of Hispanics who are Mexican-Americans separately from that of Hispanics who are not of Mexican origin. But in a jurisdiction like Los Angeles, in which most Latinos are Mexican-American, the correlation between the location of Mexican-Americans and that of other Hispanics is very high, and there are virtually no sizable pockets of Latino strength that are not heavily Mexican-American. In such circumstances, trying to disentangle the voting behavior of subgroups within the Hispanic community through multivariate ecological regression is probably just not feasible. In general, of course, great care must be taken in multigroup multivariate analysis, especially if some of the groups are very small or if the correlations in the groups' residential location introduce multicollinearity (highly correlated independent variables) into the model. We do not wish to imply that it is never possible to use multivariate methods to identify separately the behavior of groups such as blacks and Hispanics in jurisdictions in which each is found in significant numbers. We would, however, emphasize the need for caution in interpreting the results of multigroup multivariate ecological regression, and we would, in general, eschew such analyses unless their accuracy can be crosschecked.24 If a multigroup multivariate methodology is used, we emphasize the need to compare carefully the results from such analyses with those obtained from bivariate equations and homogeneous precinct analysis.25 Being inattentive to the limits of the one-equation model Many social scientists doing racial voting analysis have used the percentage of the votes for the minority candidate(s) as their dependent variable and the percentage minority among the registrants or the total voting-age population as the

Defining and measuring racially polarized voting

101

independent variable to estimate parameters such as P'BB and P'ww. Thus, when a regression line is plotted, the analyst interprets the y value of that regression line at 100 percent on the x axis as the percentage of minority votes that went to the minority candidate(s) and interprets a reading at 0 percent on the x axis as the percentage of votes that went to nonminority candidates. This single-equation model assumes that the rates of turnout as well as the proportion of voters at the polls who actually vote for the office in question are identical for both racial groups (see Loewen and Grofman, 1989). Historically, whites, blacks, and Hispanics have differed in turnout and in the extent of votes cast, although the differences are now much narrowed. If fewer minority than white registrants turn out on election day and still fewer cast votes for a given office, this one-equation method will often find somewhat more white bloc voting and somewhat less black bloc voting than actually occurred, although errors in the reverse direction can also occur because the single-equation method is also more likely than the double-equation method is to give rise to estimates that lie below 0 percent or above 100 percent. This problem may or may not be serious. In the example discussed in Loewen and Grofman (1989), the single-equation estimates for P'ww and P'BB were 100 percent (109.8 percent rounded down) and 84.0 percent; these were quite close to their best estimates for those data, namely, 98.4 percent and 89.1 percent. Misinterpreting estimates above 100 percent or below 0 percent Ecological regression analyses sometimes yield impossible results, indicating that more than 100 percent of minorities or fewer than 0 percent of nonminorities voted for the minority candidate. When this happens, it is important to remember that the results of ecological regression are statistical estimates. The fact that some of these are slightly outside the feasible range may not be of any particular importance if one pays attention to the types of checks just outlined. Moreover, based on our experience, one rarely obtains estimated minority support for minority candidates above 100 percent except when such support is truly overwhelming (as verified, e.g., by homogeneous precinct analysis); similarly, one rarely obtains estimated nonminority support for minority candidates below 0 percent except when such support is truly miniscule. Nonetheless, rarely is not never. An estimate outside the feasible range is a signal to the analyst to pay particular attention to that contest, and estimates substantially outside the 0-to100 range suggest considerable caution in interpreting ecological regression results (especially when there is more than one covered minority group present in substantial numbers).26 There are two ways to deal with estimates outside the feasible range. The first

102

Minority Representation and Voting Equality

and simplest is just to set those estimates to 0 percent or 100 percent, respectively. A technically preferred alternative (suggested by Alan Lichtman, personal communication, January 1990) is to set those estimates to 0 percent or 100 percent, respectively, and then to transform all the remaining estimates to reflect the shift. For example, if one initially estimates 104 percent black support for the black candidate and then lowers that estimate to 100 percent, there are votes for the black candidate assumed to have come from black voters that should now be assigned to whites. If, say, blacks made up 20 percent of the electorate, one would assign 0.8 percentage points of additional support for the black candidate (.2 X [104 - 100]) to the 80 percent of the electorate that is white, thus raising their estimated support for black candidates by one percentage point (.8/ 80) while lowering their estimated level of support for the white candidate by 1 percent. In practice, these additional technical adjustments are often so trivial that they can safely be disregarded. Misspecifying the test for political cohesion in the case of combined groups To show cohesiveness in a combined group, say blacks and Hispanics, requires that in elections in which there are viable minority candidates, both black and Hispanic voters usually support minority candidates in preference to nonminority candidates. Some expert witnesses have, we believe, misinterpreted this standard by failing to distinguish properly between situations in which the only minority candidates in the contest are of the same race/ethnicity and those in which there are both black and Hispanic candidates. If the only minority candidates in the contest are of the same race/ethnicity, it is appropriate to estimate separately how blacks and Hispanics are voting. Otherwise one might wrongly infer that each group is supporting the minority candidates when in fact only one group is doing so. For example, if blacks and Hispanics are present among the electorate in equal numbers and if the sole minority candidate receives 60 percent support from the combined group, it could well be that 90 percent of the blacks but only 30 percent of the Hispanics support that candidate (.90 X .50 + .30 X .50 = .60).27 Failing to estimate subgroup voting behavior properly when there is a voting rights claim from a ' 'combined'' minority group Without homogeneous precincts for each group, it may not be easy to estimate their voting behavior separately. However, even if there are no homogeneous precincts for the separate groups, if there are homogeneous precincts for the combined group and these precincts vote overwhelmingly for minority candidates, one may be able to establish that a majority of each of the subgroups

Defining and measuring racially polarized voting

103

voted for minority candidates. To take an obvious extreme case, if the combined minority cast 100 percent of its votes for minority candidates in precincts that were overwhelmingly minority, then each subgroup within that combined minority must also have cast 100 percent of its votes for minority candidates. A similar argument can be made in cases far less extreme. Consider, for example, precincts whose electorate is 40 percent Hispanic, 40 percent black, and 20 percent nonblack non-Hispanic, and in which the minority candidates receive 80 percent of the vote. Even under the highly implausible assumption that all nonblack, non-Hispanic voters in the heavily minority precincts voted for the minority candidates, at least 75 percent (60/80) of the minority voters must have voted for minority candidates. But then both black and Hispanic voters must have given at least 50 percent of their votes to the minority candidates; that is, even if one of the two groups gave 100 percent of its support to the minority candidates, for the combined support to be 75 percent, the other group would have to have voted for minority candidates at the 50 percent level. More realistically, if one estimated (say, on the basis of results in overwhelmingly homogeneous nonblack, non-Hispanic precincts) that nonminority voters cast only 5 percent of their votes for the minority candidate, minority voters in the heavily minority precincts had to have cast an average of almost 99 percent (79/80) of their votes for the minority candidates; that is, both blacks and Hispanics would have had to vote for the minority candidate at nearly unanimous levels. How reliable are estimates of racial bloc voting? In several recent cases, expert witnesses for defendant jurisdictions have argued that the methodology typically used to detect bloc voting (ecological regression and homogeneous precinct analyses) is unreliable. In the extreme, Dr. Jerome Sacks stated during one deposition that none of the statistical analyses of racial voting patterns that he had ever seen (including that accepted by the Supreme Court in Gingles) was statistically valid (Sacks deposition, 1987), and although Sacks tempered his remarks about the accuracy of at least the Gingles analysis in testimony in later cases, he continues to attack the standard methods for proving polarization as unreliable except in circumstances in which housing patterns are very heavily segregated.28 Garza witnessed the most extensive attack on the reliability of bloc-voting methodology of any court case to date. Both statisticians (Dr. Jerome Sacks and Dr. David Freedman) and social scientists testified that it was impossible to estimate reliably the voting behavior of blacks or Hispanics in Los Angeles County elections without exit poll or other survey data. According to these experts, no conclusions could be drawn about whether or not voting in the

104

Minority Representation and Voting Equality

county was polarized along Hispanic-versus-non-Hispanic lines in a fashion that would affect the ability of Hispanics to elect a candidate of choice to the County Board of Supervisors.29 The district court found otherwise: "While in theory there exists a possibility that ecological regression could overestimate the [degree of polarization], experts for defendants have failed to demonstrate there is in fact any substantial bias" (slip opinion at p. 86). Of course, that a court said it does, does not necessarily make it so. But we believe the data fully support Judge David V. Kenyon's findings in Garza.30 At the countywide level, for both Hispanics and non-Hispanics, in the two elections that were most directly relevant to determining polarization (a reconfirmation vote for a controversial Hispanic judicial candidate who was among the subjects of a recall effort and a referendum on English as an official language), exit poll data were available, and ecological regression perfectly mirrored the findings of the exit polls at the level of the county as a whole. The only situations of which we are aware in which federal courts have failed to find the results of standard racial bloc analysis methods to be reliable were situations in which (1) there was more than one covered minority of substantial size and (2) minority populations were heavily intermingled without any singleminority homogeneous precincts. Moreover, even in these special circumstances, there are likely to be indications that clearly signal the problem, such as low correlations or estimates that cannot be reconciled with observable aggregate data. As soon as such severe problems are detected, the analyst should report them and either indicate that it was impossible to estimate the behavior of the group or try to put bounds on what was likely to be true, rather than trying to specify a point estimate.31 However, we should also note that when there is a consistent pattern in which the vote for the minority candidate increases as the minority proportion in the precinct increases and in which heavily minority precincts can be expected to give a majority of their vote to viable minority candidates identified with the particularized interests of the minority community, if we are going to draw a remedy plan, it does not matter that we know with certainty who in those precincts is voting for the minority candidates. What matters for redistricting is that we know the areas in which minority-preferred candidates are likely to receive strong support, and these are the heavily minority precincts. Second, to the extent that there are questions about minority political cohesion, they are almost invariably about cohesion between minority members who live in minority areas and those who do not. But as a matter of practical reality, the minority district(s) that will be drawn are districts centered in the heavily minority areas of the jurisdiction. These are exactly the areas where there normally is the least dispute about minority political cohesion.

Defining and measuring racially polarized voting

105

Other elements of the totality of the circumstances As discussed in Chapter 3, the totality of the circumstances may still play an important role in civil rights litigation, and to that extent, social scientists will continue to be called on to provide expert witness testimony about the seven elements. One of the elements - racially polarized voting (Factor 2) - has been discussed in detail in this chapter as well as in Chapter 3. Another - the extent to which minorities have been elected (Factor 7) - was also dealt with extensively in Chapter 3. In the rest of this chapter, we provide a brief discussion of the important elements of social science testimony about the remaining factors. Factor 1: History of official discrimination. Ascertaining the existence or nonexistence of this factor for the period before 1965 is unproblematic in most relevant jurisdictions. In southern states, a historian using standard sources can readily establish a statewide history of official discrimination against blacks (see, e.g., Kousser, 1974); the same is true in the Southwest for Hispanics or American Indians.32 However, de jure segregation is increasingly "old" history. Courts have generally been interested in whether or not there is evidence of more contemporary patterns of discrimination. A showing of recent discrimination has usually been achieved by tracing official positions, including legal resistance, to various racerelated issues that have resulted in court-imposed solutions such as affirmative action hiring for police and fire departments, simplified procedures for voter registration in the minority community, school busing, denial of preclearance under Section 5, and location of public housing. Evidence of statewide discrimination, though often sufficient, has frequently been supplemented with evidence specific to a given polity, for example, regarding segregated private clubs and other forms of social segregation in which local political officials participate, segregated housing patterns buttressed by redlining, or historic use of restrictive covenants. Of particular relevance has been evidence regarding barriers to registration and voting (e.g., few bilingual registrars, few minority poll watchers and election officials, polling precincts located disproportionately in white/Anglo areas). Some of these data are gathered from official archives (e.g., polling maps, property deeds, litigation dockets, and court records) and from census data, but some are introduced only via the testimony of knowledgeable local citizens.33 Factor 3: Election practices that increase the likelihood of minority vote dilution. In the context of multimember districts, federal courts have repeatedly accepted the claim that majority vote requirements, anti-single-shot voting rules,

106

Minority Representation and Voting Equality

numbered places, staggered elections, and unusually large districts make it more likely that minority voting strength will be diluted. In fact, it has not been necessary to establish a causal link between such practices and minority vote dilution in the particular jurisdiction involved. Moreover, in the case of majority runoff requirements, numbered places, or staggered elections, their existence or past use is clearly a matter of record, so there are no questions of interpretation. The meaning of the phrase "unusually large election district" is not so clear.34 In Garza, a single-member district case, district population was a significant issue. The population in each of the districts, roughly 1.7 million, make them as populous as any of the sixteen smallest states, and they are the largest county supervisor districts in the country (by a factor of two). The district court specifically acknowledged that the small size of the board (five), which led to the creation of such highly populous districts, was a factor that could make it harder for minority candidates to compete successfully, given the limited financial resources of the minority community.35 A related concern is the number of districts in a jurisdiction (see Chapter 5). Factor 4: Candidate slating process. In many jurisdictions there is no formal slating process (especially for nonpartisan elections), although groups of candidates may band together to run as an informal slate. In other jurisdictions, party organizations may designate certain candidates as having official endorsement, even though the actual nomination may take place in a party primary. In other jurisdictions (most commonly ones with a nonpartisan ballot), business and civic groups may endorse slates. In expert witness testimony in United States v. City of Augusta (1987), a case settled out of court, the endorsement of the major newspaper in the city was argued to be a type of slating process. If there is a slating process and endorsed candidates are more likely to win than are nonendorsed candidates, the number of minority candidates who are interviewed or nominated is, of course, the single most important piece of information bearing on the fairness of the process. Potentially of almost equal importance is information on the composition of the slating group itself. For example, in Alonzo v. Jones (1983), a successful Section 2 challenge to at-large city council elections in Corpus Christi, Texas, the district court judge noted in his finding of facts that although Mexican-American candidates had been slated and successful in their election campaigns, all who had won did so "by being members of a slate basically assembled by Anglo . . . leaders." In fact, the slateendorsed Hispanic candidates defeated non-slate-endorsed Hispanic candidates who had greater support from the Hispanic community. Factor 5: Lingering effects of discrimination. Lower levels of education, employment, income, health and longevity, and the like have been taken by courts

Defining and measuring racially polarized voting

107

to be indicators of the extent to which minority group members bear the burden of previous discrimination. Such data are published at the county level and for at least the larger cities in each state. They can be obtained from census tapes for any desired unit of aggregation. Sociologists or demographers customarily present such data as a routine part of Section 2 litigation. They are rarely subject to dispute. Lower levels of minority registration and turnout relative to eligible population have also been taken as indicators of lingering effects of discrimination. For the handful of states where registration (or sign-in) data are available by race, the basic facts on comparative levels of minority and nonminority registration (or turnout) can be readily found.36 For states, and for other jurisdictions large enough to be reasonably coterminous with SMSAs, the Census Bureau's Current Population Survey (CPS) permits an estimate of registration rates (and turnout in federal elections) of eligible voters by race and by Spanish origin. In Garza, CPS data for Los Angeles County were accepted by the court as a check on the accuracy of the county's own population projections.37 For cases involving Hispanics, as noted previously, a matching of registrar lists with the Bureau of the Census's list of Spanish surnames has been accepted by courts as a proxy for Spanish-origin registration (e.g., in Gomez v. City of Watsonville, 1988). For jurisdictions for which data of the preceding sort are not available, estimates of registration and turnout data by race or Spanish origin can be generated by combining census, registration, and election data and using the ecological and homogeneous precinct analysis techniques described earlier in this chapter. Factor 6: Racial campaign appeals. Racial appeals in campaigns come close to being in the category of "know it when I see it." The only definitions of racial appeal of which we are aware are those offered by sociologists Dr. Paul Luebke and Dr. Jerry Himelstein. Luebke's definition, offered as testimony in Gingles v. Edmisten, is as follows: Racial appeals are present in a campaign if one candidate calls attention to the race of his or her opponent or opponent's supporters or if the media covering a campaign disproportionately call attention to the race of one candidate or of that candidate's supporters. Luebke's definition is probably as clear and unambiguous as one can hope for, given the inherent fuzziness of a term such as racial appeal. An example of a racial campaign appeal that fits the definition perfectly is a campaign in which a white candidate uses a picture of his black opponent in his own campaign material. Except as a notice to white voters of his opponent's race, no candidate would give free photographic publicity to his opponent. Luebke provided such an example (for the 1982 Michaux-Valentine Democratic congressional primary in North Carolina) in his testimony in Gingles.

108

Minority Representation and Voting Equality

Often, however, racial appeals are less blatant and therefore more difficult to identify. They often involve the use of "code words." According to Himelstein (1983, p. 156), a code word is a word or phrase that communicates a wellunderstood but implicit meaning to part of a public audience while preserving for the speaker a deniability of that meaning by reference to its explicit meaning. For example, in 1968 presidential candidate Hubert Humphrey accused his Republican opponents of using the phrase "law and order" as a code word for repression of blacks in reaction to riots in black ghettos. Similarly, in Jimmy Carter's 1976 awkward attempt to pacify residents in a Polish neighborhood who were worried about desegregation in housing, he endorsed the preservation of "ethnic purity" in residential patterns. Blacks quickly identified this phrase as a code word for segregation, and Carter spent some time trying to explain it away. Code words are intended as rhetorical winks; if they are too easily detected, they can lose their deniability and thus their effectiveness. Luebke's definition of racial appeal is easier to measure objectively than is Himelstein's definition of code words. As Himelstein himself notes, "Identification of code words is an enterprise akin to the interpretation of symbols in literary criticism. For code words, such identification requires careful sociological and historical analysis" (1983, p. 157). Nonetheless, because overt racial appeals may be absent, testimony about more subtle and covert forms of racial appeal such as the use of code words and of themes associated with white supremacy or antiblack sentiment may be required if this element of the totality of-circumstances test is to be proved. It does not appear that anything beyond the mere existence of racial appeals needs to be demonstrated. In particular, it does not appear that the effectiveness of the racial appeals must be shown. Of course, if voting is racially polarized, that fact provides indirect support for the probable impact of any observed racial appeals. Methods for defining and measuring the key elements in vote dilution cases have become relatively standard. They are subject to continuing debate by defendant jurisdictions, but the objections have been rather routinely discarded by the courts in favor of established procedures. The procedures may need some expansion and refinement as they are increasingly applied to situations with more than one covered minority group, but in general, the methods we have outlined in this chapter are likely to remain the basic methods.

Vote dilution in single-member districts and other issues of the 1990s

Legal standards to determine when a single-member district plan constitutes a racial gerrymander in violation of the Voting Rights Act (or the U.S. Constitution) are not as well developed as are the standards for dilution involving multimember districts and at-large elections. This is in large part because there have been far fewer cases involving challenges to single-member district plans. Moreover, except for the legislative districts in New York whose redrawing was the subject of a challenge to the Justice Department's discretionary authority under Section 5 of the Voting Rights Act1 and for the congressional seats in the Dallas area that were redrawn by a federal district court in remedying a Fourteenth Amendment violation,2 no single-member district plan challenged as a racial gerrymander has been the subject of other than a per curiam opinion of the Supreme Court, and neither of these cases postdates Gingles. Indeed, since Gingles was decided in 1986, as of mid-1991 only a handful of Section 2 cases involving challenges to single-member districts had been decided,3 and only four of these had been reviewed at the appellate level.4 Despite this lack of activity to date, single-member district issues and related matters will rise to the fore in the 1990s. For one thing, the number of challenges to at-large or multimember district systems is likely to fall, as a large number of the relevant jurisdictions have already been successfully challenged (see the state-by-state summaries in Davidson and Grofman, 1992), and many of the remaining jurisdictions are likely to adopt single-member districts rather than incur the costs of a voting rights lawsuit in which they are unlikely to prevail.5 In addition, the redrawing of lines in light of the 1990s census will generate a considerable body of new litigation, and much of it will be in jurisdictions that have previously shifted to single-member district (or mixed) plans. As the single-member district becomes the focus, new litigation will give rise to more serious questions than heretofore about just how expansively courts should interpret the Voting Rights Act. The questions will be as narrowly bounded as how to balance the equal population requirement with the principle 109

110

Minority Representation and Voting Equality

of nondilution of the minority vote, and as broad and fundamental as to require rethinking our entire system of representation. The central purpose of this chapter is to identify the issues and viewpoints that will govern the debate about minority representation in the next decade beyond those dealing with specific interpretations of the Gingles prongs, which we dealt with in Chapter 3 - and to suggest some ways in which these issues might be resolved. We shall, however, raise some questions for which there are no certain answers and on which even we are not fully agreed.6 Considerations of size Allowable deviations from population equality In one sense, standards for applying the equal population requirement are now well established. Following the decision in Karcher v. Daggett (1983), congressional districts are to be held to a standard of strict equality; in line with Mahan v. Howell (1973), state legislative districts may vary in size up to 10 percent if no justification is provided and to as much as perhaps half again with proper reasons. Yet the question arises: How much deviation from equality is justified in pursuing the goal of racial fairness? With respect to population deviations in general, the Court has provided some guidance by indicating the kinds of factors that constitute adequate explanation and justification for existing size differences. In Karcher, for example, Justice William J. Brennan, Jr., noted that "any number of consistently applied legislative policies might justify some variance [from precise numerical equality], including, for instance, making districts compact, respecting municipal boundaries, preserving the cores of prior districts, and avoiding contests between incumbent Representatives" (p. 740). Even in this general case, however, there is little guidance as to the permissible size of such deviations. Justice Brennan refers to "minor" population deviations (p. 740) but adds that "the showing required to justify population deviations is flexible, depending on the size of the deviations, the importance of the State's interests, the consistency with which the plan as a whole reflects those interests" (p. 741). He also notes that "by necessity, whether deviations are justified requires case-by-case attention to those factors" (p. 741). Thus, it is unclear just how much deviation from exact (or de minimis) equality of district size is permissible if thorough justification is provided. If the size of permissible deviations is unclear in the general case, the situation is more complicated when the justification is prevention of minority vote dilution. Nondiscriminatory effects of districting are required by Section 2 of the Voting Rights Act. Does this federal requirement justify somewhat greater de-

Vote dilution in single-member districts

111

viations from equality than other factors?7 One answer, of course, follows from Justice Brennan's statement just quoted: It depends on an analysis of the case at hand and on the specificity and consistency with which that rationale is applied. But that still leaves a good deal of ambiguity.8 Alternatively, in the remedy phase of a case in which a violation of Section 2 has been found, how much deviation from equality can be tolerated to achieve an equal opportunity for minority groups to elect a candidate of their choice? For example, if a 2 percent deviation allows creation of a district that is majority minority in total population but a 3 (or 4 or 5) percent deviation allows for a majority minority in voting-age population, is the larger deviation allowable? To say that the answer requires a case-by-case analysis is simply to invite still more litigation in an area already overflowing with court involvement. A possible solution to this dilemma is for the Supreme Court to establish firmer guidelines regarding population equality - for example, that no more than a given percentage deviation will be permitted "except under extreme circumstances." In order to avoid litigating every case with a deviation falling within the firmer guidelines, this approach would make the most sense if combined with a recognition that some small degree of departure from equality (say 5 percent) would be regarded as de minimis in congressional districting cases.9 Population projections and census questions Localities often make projections of population growth by area in order to plan better for future service needs. Often these projections include estimates of relative growth of minority populations, at least in areas with high concentrations of minorities. Should it be necessary to use such population projections in order to predict ability to elect over the course of a decade? The arguments on either side are obvious. On the one hand, population projections are just that; there is no way to guarantee accuracy. On the other hand, when major changes in demography are widely anticipated, they should be taken into account so that representation is relatively fair beyond the immediate moment. This argument is especially forceful when population changes are expected to be rapid, so that failing to take them into account means that elected officials will not be representative of the population throughout much of the ensuing decade.10 From a legal perspective, it appears as though the Supreme Court would be willing to consider anticipated population shifts if projections were applied consistently to all districts (Karcher v. Daggett, 1983, pp. 740-741; Kirkpatrick v. Preisler, 1969, p. 535). It seems doubtful, however, that consideration of this factor would be required of jurisdictions.11 Other issues related to the use of population projections and census data revolve around the question of when and how often districting needs to be done.

112

Minority Representation and Voting Equality

In Garza v. Los Angeles County Board of Supervisors (1990), it was held that a jurisdiction whose initial plan was found to be dilutive could be required to implement a new plan even quite late in the decade; part of the evidence concerning dilution involved anticipated consequences of the plan over the decade.12 It was also held in Garza that population estimates that a court found to be more reliable than "stale" census data could be used in fashioning a remedy plan near the end of a decade, though generally courts have held that census data are to be used (see n. 11). Still other questions remain to be addressed. Can a jurisdiction whose initial plan is arguably not dilutive based on the evidence available at the time of its passage be forced to redistrict later in the decade when evidence becomes available about subsequent population changes? How reliable must the evidence for population changes be before it can replace "stale" census data?13 Must a jurisdiction redistrict in time for the first election after the availability of federal census data, or is there a certain grace period? And of course, there are issues concerning the appropriateness of and the methodology for statistical adjustment of the federal census data to deal with the minority undercount.14 The number of districts Another "size" problem has to do with the number of districts into which a jurisdiction has been divided. In some situations it is impossible to create a minority district (or an additional minority district) given the existing number of seats, but it would be easy to create such a district for a slightly larger body. Consider, for example, a county with an 8 percent minority population and a five-member county board. There is no way that a majority-minority district could be created, even if the entire minority population lived in a single area and even if considerable population deviation is allowed.15 However, if the size of the board were increased to seven members, one might well be able to create a district that was at least majority minority in its total population. Thus, the question arises, Is there a violation of Section 2 if a jurisdiction with a substantial minority population has a legislative body of a size such that minority voters cannot be given a reasonable chance to elect a candidate of their choice? Or in such a situation, is it reasonable, as part of a remedy phase, to require increasing the size of the board? Put somewhat differently, in judging whether or not a district can be drawn that meets the first prong of Gingles, does one have to rely on the existing number of districts, or can one argue that the number of districts is in itself unfair? Traditionally, courts have looked only at legislative bodies as they currently exist. If a minority population is large enough to meet the Gingles size test only by increasing the number of districts, it would normally mean that plaintiffs

Vote dilution in single-member districts

113

would be unable to prevail (e.g., McNeil v. Springfield Park District, 1988). However, there have been two exceptions to this norm of which we are aware, and the question has been raised elsewhere, though it was not part of the decision (e.g., McDaniels v. Mehfoud, 1988).16 The problem will not occur with respect to Congress, of course, because each state is assigned a specific number of House seats. Nor is it likely to arise with respect to state legislatures, because the number of legislators is usually fixed for long periods of time and bears no particular relationship to the size or distribution of the minority population. In addition, the effect on minority representation of moving from, say, 99 legislators to 101 is not likely to be large. In small jurisdictions, however, such as cities and counties, sizes of boards and commissions are probably less often fixed, and the effects of changes in size can be dramatic. It may also be easier to show that the relationship between size and minority representation was at least known at the time the current size was fixed, if not necessarily a criterion used in creating a districting plan. Thus, in local areas, the size of legislative bodies may occasionally arise again in the 1990s. Identifying racial gerrymanders Because Gingles dealt exclusively with the submergence of a minority population in an at-large or multimember district setting, it is impossible simply to use the three prongs of Gingles as the test for a liability finding in the context of a racial gerrymandering claim.17 There are a number of alternatives the courts might pursue in order to specify a test for vote dilution in single-member districts. The first three avoid Gingles altogether. A totality-of-circumstances test One possible approach to the single-member district situation is to use a totalityof-circumstances test, referring directly back to the factors identified in the Senate Report accompanying the 1982 extension of the Voting Rights Act. However, there are two key problems with this approach. First, the Senate Report itself is almost entirely oriented to dilution occurring in the at-large or multimember district context, because that was the setting for the key vote dilution cases on which it relies, for example, White v. Regester and Zimmer v. McKeithen. Second, and relatedly, the Senate Report factors are not always the most appropriate ones when one is looking at a single-member district plan, regardless of what we may think of their relevance in the at-large or multimember district context. Some of the seven factors identified in the Senate Report are of only tangential relevance in a single-member district setting (e.g., unusually large election dis-

114

Minority Representation and Voting Equality

tricts, the presence of anti-single-shot provisions) and others (e.g., usual minority loss) need to be reinterpreted in terms of some appropriately defined baseline. In racial gerrymandering cases, factors not among the seven identified in the Senate Report (e.g., fragmenting or packing of the minority population, deviation from standard districting criteria) have often been the central focus of legal inquiry. Thus, one could hardly rely exclusively on the Senate's rendering of the totality of the circumstances; minimally, one would have to add one or more factors, and these would almost certainly turn out to be the most important factors in single-member district cases.

Lowered standards when intentional discrimination is found A second approach was used in Garza v. Los Angeles County Board of Supervisors (1990). As evidence of injury, the circuit court drew on a factor not in the Senate list of seven: "The supervisors' intentional splitting of the Hispanic core . . . violated both the Voting Rights Act and the Equal Protection Clause" (p. 771, emphasis added). This fragmentation is the heart of the finding. The court does not even discuss any of the factors listed in the Senate Report, even racial polarization, despite the fact that the defendants claimed that the district court was wrong as a matter of law in the evidence it accepted as proof of polarization.18 The court held that once intentional discrimination had been found, the Gingles standards for group size and geographic concentration necessary for a liability claim did not apply, and it argued against a "bright line" test (such as a 50 percent minority citizen voting-age population) in such a setting: The plaintiffs' claim is not, as in Gingles, merely one alleging disparate impact of a seemingly neutral electoral scheme. Rather, it is one in which the plaintiffs have made out a claim of intentional dilution of their voting strength, (p. 770) To impose the requirement the County urges would prevent any redress for districting which was deliberately designed to prevent minorities from electing representatives in future elections governed by that districting. This appears to us to be a result wholly contrary to Congress' intent in enacting Section 2 of the Voting Rights Act and contrary to the equal protection principles in the fourteenth amendment, (p. 771, internal citations omitted) The circuit court went on to say that some showing of injury was necessary but this showing "need not be as rigorous as in effects cases" (p. 771). Although the approach used by the circuit court in Garza makes considerable sense, it is obviously restricted to cases in which there is a finding of intentional discrimination and therefore cannot serve as a general guide to the identification of racial gerrymanders.

Vote dilution in single-member districts

115

Nonretrogression test A third approach to racial gerrymandering claims in single-member district plans, found so far only in decisions antedating Gingles such as Ketchum v. Byrne (1984), is to apply the nonretrogression test developed in Section 5 jurisprudence to cases arising under Section 2. Under this approach, the presence of retrogression would be a sufficient but not a necessary condition for a Section 2 violation. After Gingles, however, the legal status of this approach is in doubt. In Badillo v. City of Stockton, a case still on appeal (December 1991), a federal district court rejected the claim that a retrogression test was applicable in a Section 2 case.19 Of course, even if a nonretrogression test were to be adapted to Section 2 litigation, it would resolve only a few of the racial gerrymandering claims that might be brought, as the continuation of an existing pattern of gerrymandering would be nonretrogressive, even though it could still be held to violate Section 2. In short, a variety of approaches that circumvent the Gingles conditions are either inappropriate or applicable only to a fraction of the cases that are likely to arise in the 1990s. Thus, another approach is to modify Gingles in such a way as to make it applicable to single-member districts. As we shall see, this is not a simple task. Modifying Gingles to apply to single-member districts A three-judge panel in Jeffers v. Clinton (1989) opened the door to this approach: Thornburg and Smith cannot be automatically applied to the single-member context. . . . But the basic principle is the same. If lines are drawn that limit the number of majority black single-member districts, and reasonably compact and contiguous majority black districts could have been drawn, and if racial cohesiveness in voting is so great, that, as a practical matter, black voters' preferences for black candidates are frustrated by this system of apportionment, the outlines of a Section 2 theory are made out. (p. 205) Indeed, despite the caveat about automatic application, Judge Richard Sheppard Arnold's opinion suggests that the extension of Gingles to the single-member district context is quite straightforward. We agree that two of the Gingles prongs can probably be applied with little or no modification. The second prong, minority political cohesion, would seem to require no changes. The third prong - lack of white/Anglo support for minority candidates leading to ''usual" minority loss — could also remain essentially unchanged, though it would need to be understood in terms of some appropriate baseline of what success could reasonably have been anticipated under an alternative fairly drawn districting configuration.20

116

Minority Representation and Voting Equality

The first prong, however, is more difficult to modify in a suitable way. One modification may be to change the first prong to read that there is a Section 2 violation if one additional minority district can be created. Indeed, the reasoning behind such a requirement would seem to be exactly the same as in the multimember situation: To paraphrase Justice Brennan's words (Thornburg v. Gingles, 1986, p. 50), if a minority group is not sufficiently large and geographically compact to constitute a majority in an additional single-member district, the single-member district plan in question cannot be responsible for the minority voters' inability to elect more of their candidates. Moreover, this is exactly the tack taken by a number of district courts in cases arising in the Fourth Circuit. In particular, in White v. Daniel (no. 88-0568-R, E.D.Va. 1989), the district court required the plaintiffs to demonstrate that blacks could constitute effective majorities in three of the proposed districts rather than in only the two that were already represented by black officeholders.21 Although an "additional district" requirement is unproblematic in many circumstances, it is not without difficulties. For one thing, the modified requirement is subject to the same problems of interpretation as is the original prong (see Chapter 3). More important, it leads to the interpretation that one must create minority districts whenever possible - that is, that the standard for comparison is the maximum number of minority districts that can be created in the jurisdiction.22 There are at least three potential problems with this interpretation. First is a technical one: Is the standard the number of bare majority-minority districts, when a bare majority (of the total population) often fails to give minorities a realistic opportunity to elect (see our later discussion of effective voting majorities and realistic opportunity to elect)?23 Second is the legal question of what other criteria a district must satisfy. Here it is important to distinguish between the liability and the remedy phases of a trial. What may be a district with a sufficient minority population to satisfy the size requirement of the first prong of Gingles at the liability stage may not be a suitable remedy. In particular, if the Gingles conditions are met, must one draw additional minority seats whenever possible, even if these are extremely torturously shaped districts that combine discontiguous (or tenuously connected) pockets of minority concentration and show a blatent disregard of both neighborhoods and existing political boundaries? Third is also a legal question: Is it constitutional to guarantee a group the maximum representation possible with single-member districts? We emphasize, of course, that guaranteeing maximum representation under single-member districts is not the same as proportional representation, which both Section 2 and the Supreme Court have noted is not an entitlement. Nonetheless, the subsidiary

Vote dilution in single-member districts

111

question of whether guaranteeing maximum possible representation within the single-member district context is sure to be debated in future litigation. In short, modification of the Gingles first prong and acceptance of all three prongs as sufficient to establish a violation of Section 2 in single-member districting cases seem a promising approach, though one that is not without problems or controversy. Influence districts Gingles left open the question whether minority groups that cannot demonstrate a potential to elect candidates of choice under an alternative configuration can nonetheless make out a showing of liability if they can demonstrate that their opportunity to influence election outcomes has been reduced.24 This question takes on great importance in the single-member district context, especially in light of a recent decision by a three-judge court supporting such a claim.25 However, most courts that have looked at the question have concluded that to permit "influence" claims would embroil courts in an area in which there are no clear standards, as well as make it likely that the courts would be overburdened with marginal cases.26 We believe that if the views of the now vacated Sixth Circuit opinion in Armour were to be accepted widely, minorities might be harmed more than helped. Though this is perhaps counterintuitive, it follows directly from language in the opinion that seems to invite claims that minorities would be better off with their population dispersed among multiple districts. Until Gingles clarified the standards for Section 2, it was not uncommon for defendant jurisdictions to argue that an existing at-large plan provided minorities with jurisdictionwide influence that would be lost were they to be concentrated in majority-minority districts. Such protestations ring hollow. If minorities have no opportunity to elect one of their own and they would be apt to do so if given the choice, it is paternalistic to say that they should be content with their supposed opportunity to influence elections of white representatives. But ironically, that is precisely what could result if the influence concept is embraced. We also share the concern that the concept of influence is murky. We are sympathetic to the claim advanced by the dissenting judge in the vacated opinion in Armour that "under the type of influence claim asserted here, the court is sailing a chartless [i.e., uncharted] sea" (slip opinion, p. 30). Attempts by social scientists to define "pivotal" voting power in a realistic fashion are problematic (Grofman, 1981), and mathematically derived "power" indices have been rejected by federal courts (see, e.g., Morris v. Board of Estimates, 1989; cf. Grofman and Scarrow, 1981a, b). Moreover, as the dissenting judge in Armour

118

Minority Representation and Voting Equality

pointed out (slip opinion, pp. 29-32), if one claims that increasing the number of black voters in a district ipso facto increases their influence, then it would seem to follow that the influence of black voters is reduced in those districts from which black voters are removed. When there are "electability" claims at issue, there is a natural threshold. Without such a threshold, how does one decide whether shifting minorities from one district to another increases or decreases their overall influence?27 An equal opportunity to elect candidates of choice At the remedy stage of a lawsuit, the focus shifts from trying to show that minorities do not have an equal opportunity to elect candidates of their choice to constructing districts in which they do have an equal opportunity. The effort to construct fair districts - usually single-member districts - raises a number of important questions.28 Fortunately, some relatively simple procedures have been developed to help us answer those questions. We will describe the general procedures here, but we emphasize that they must be applied to data from the jurisdiction in question in a particular case. One of the lessons of past analyses is that no single answer is suitable for all times and places. The initial problem is that minority populations most often have lower levels of eligibility, registration, and turnout than do majority populations. Moreover, the effects of incumbency usually favor majority candidates, and minorities often have less access to campaign finances. Owing to these factors, it has been known for some time that creating districts in which a minority group has a bare majority of the population is unlikely to provide a realistic chance of electing a candidate preferred by the minority community. Thus, one is first led to ask: How large a population majority is required to equalize opportunities for majority and minority candidates? And then, how does one create districts that have a sufficient minority population - whether the goal is to create a competitive district or a safe minority district - and yet avoid the dilution that comes from "packing" large numbers of minorities into a small set of districts? The analysis begins with the calculation of what is called effective voting quality - the size of a minority population large enough to yield a bare majority of voters on election day. Effective voting equality Calculation of effective voting equality begins with four key numbers determined separately for minority and majority populations: (1) the proportion of noncitizens, (2) the proportion of the citizen population that is age eligible to vote (eighteen or older), (3) the proportion of the eligible population registered

Vote dilution in single-member districts

119

to vote, and (4) the proportion of those registered who actually vote. The four components are combined, using a simple algebraic model, to determine the percentage of the overall population required to equalize minority and nonmi nority turnout.29 Consider a simple example that uses only (1) to determine overall minority and nonminority populations that would equalize citizen populations. Suppose that only 1 percent of the nonminority population is noncitizen but that 10 percent of the minority population is noncitizen. Then, if the total population in a district is 50 percent nonminority and 50 percent minority, the nonminority group will constitute, in actuality, 52.4 percent of the potentially eligible voters. That is, .99(50)/[.99(50) + .90(.50)] = 52.4 If we wish to equalize nonminority and minority citizen populations, we would have to create a constituency with 52.4 percent minorities and 47.6 percent nonminorities. That is, .99(47.6) = 47.1 and .90(52.4) = 47.1 indicating that both minority and nonminority populations would have 47.1 percent of the total district population as their potentially eligible voters. The four components can be considered separately, allowing one to calculate separate ''corrections" for the effects of citizenship, voting age, registration, and turnout.30 More commonly, one simply wishes to look at the end result the size of the overall minority population required to equalize the turnout. Empirical estimates (Brace et al., 1988, p. 51) indicate that effective voting equality for blacks may at times require little more than 50 percent of the total population, though the usual requirement is for between 55 and 60 percent, and the figure is sometimes as high as 65 percent. Because blacks tend to be Democratic in far greater proportions than nonblacks are, effective voting equality in primary elections usually requires lower proportions of the overall population - often less than 50 percent - though even in these cases proportions of over 60 percent are sometimes necessary. Because of citizenship and other factors, percentages as high as or higher than 65 percent are sometimes needed to achieve effective voting equality for Hispanics.31 As a final methodological point, we should note that one can often validate the proportion needed for effective voting equality by comparing turnout percentages in homogeneous minority and nonminority areas. Otherwise, it is likely

120

Minority Representation and Voting Equality

that one can provide some validation using the two-equation ecological regression of the sort described in Chapter 4. The so-called 65 percent rule Some of the earliest cases dealing with a realistic potential to elect, such as Kirksey v. Board of Supervisors of Hinds County (1977) and United Jewish Organizations of Williamsburgh v. Carey (UJO) (1977), proposed a 65 percent minority population as the basis for effective voting equality. This so-called 65 percent rule has been claimed both to have the support of the Justice Department and to have been given the imprimatur of the U.S. Supreme Court in UJO. Both claims are wrong. Neither Kirksey nor UJO stands for the proposition that the 65 percent rule is right for all times and all jurisdictions. In Kirksey, a review of the district court opinion makes it clear that the expert witness in that case, Dr. James Loewen, based the 65 percent figure on an analysis of data specifically for the Mississippi county whose election practices were being challenged; the 65 percent figure used in UJO for some legislative districts in Brooklyn could best be characterized as a "rule of thumb" that rested on shaky empirical grounds.32 Senior officials in the Voting Rights Section of the Justice Department have repeatedly made clear that the department does not regard the 65 percent figure as having any special significance - rather, each case is to be investigated in terms of the case facts special to it (Paul Hancock, personal communication, 1987; Gerald Hebert, personal communication, 1989). Nonetheless, a mischaracterization of the Justice Department's views appears even in some of the most recent work on redistricting (e.g., Anderson and Dahlstrom, 1990, p. 70). Through repetition (and misreading of the early cases), errors regarding the legal and social science status of the 65 percent rule had, by the late 1970s, become enshrined in court lore and were even perpetuated by some expert witnesses. For example, the district court in Gingles v. Edmisten (1984), even though it specifically refused to attempt to "define the exact population level at which blacks would constitute an effective (non-diluted) voting majority, either generally or in this area" nonetheless went on to say: "Defendant's expert witness testified that a general 'rule of thumb' for insuring an effective voting majority is 65%. This is the percentage used as a 'benchmark' by the Justice Department in administering Section 5 " (p. 358, n. 21).33 Quite simply, however, there is nothing special about the 65 percent figure: It is sometimes too high and sometimes too low as an estimate of what is needed for effective minority voting equality. Fortunately, a number of recent courts have recognized this fact. For example, in Martin v. Mabus (1988), the court found a 60 percent figure right for Mississippi judicial elections.

Vote dilution in single-member districts

121

An appraisal of local political reality Even if a district is created in which it is estimated that half of the actual voters will be minority members, it is not certain that this will effectively equalize the opportunity of the majority and minority communities to elect candidates of their choice. Further appraisal of the local situation is still necessary. On the one hand, even an "effective majority" may not be adequate to provide a realistic opportunity to elect candidates of choice, as such a purely numerical threshold does not take into account financial and other disadvantages faced by minority (or minority-backed) candidates and does not take into account the advantages of incumbency, which are often held by nonminorities. More generally, by focusing entirely on one type of number, equalizing (only) turnout may fail to take into account other subtler but important features of the situation. On the other hand, a district may provide a minority group with considerably less than an effective majority but still give them a realistic opportunity to elect candidates of their choice. The most common context in which this might occur is in the partisan elections in which the minority group has an effective majority in the primary but not in the general election, but in which its candidates may still win the general election as a result of crossover voting.34 Presence of incumbents Particularly in the remedy phase, other aspects of the local situation may also be relevant. Perhaps the most salient is the incumbency status of districts designed to give minorities an equal opportunity to elect a candidate of their choice. For example, in the remedy phase of Garza v. Los Angeles County Board of Supervisors (1990), the district court, in rejecting a remedy plan with a 74 percent Hispanic population and a near majority Hispanic registration, accepted expert witness testimony that the district had been gerrymandered to include the home of a nonminority incumbent who had already raised nearly a million-dollar war chest and that an open seat in which a greater number of Hispanic voters resided was the more obvious district to convert into a Hispanic majority district. These arguments were also accepted by the Eleventh Circuit majority in affirming the district court's opinion. Thresholds of representation and exclusion When there have been some minority victories, one way to determine the conditions that lead to minority success is to look at minority population or registration levels in both districts that do elect a minority candidate and districts

122

Minority Representation and Voting Equality

that do not. Such an analysis uses two analytical concepts: the threshold of representation and the threshold of exclusion. The threshold of representation is the smallest minority proportion in districts that have elected a minority candidate; the threshold of exclusion is the greatest minority proportion in districts that have failed to elect a minority candidate.35 As of 1990, for blacks, the threshold of exclusion in the U.S. House of Representatives was 45.2 percent (in Mississippi's Fourth District). Every district above 45.2 percent black elected a black representative, including sixteen districts with a black population majority and three districts with a black population in the very high forty thousands. However, a few blacks also were elected to Congress from districts with a much smaller black population. Indeed, the congressional threshold of representation for black Democrats in 1990 was a mere 22.9 percent, whereas a black Republican was elected to Congress in 1990 from a district with a miniscule (4 percent) black population.36 Thresholds of representation and exclusion for blacks in southern state legislatures show a great deal of variation across states and over time, with Mississippi having the highest threshold of representation (over 70 percent), whereas in states such as Alabama, Florida, and North Carolina, all districts that are majority black in population elect black representatives. There is even greater variation across jurisdictions at the levels of city, county, or school board elections. Runoff elections Runoff, or double-ballot, elections require that when no candidate receives a certain percentage of the vote (usually a majority) in the first election, the two candidates with the greatest number of votes in that election must run against each other in a second election. In most local jurisdictions, runoff primaries are found in conjunction with nonpartisan elections, but in a number of southern states, runoffs are required for all or most partisan primaries. The rationale behind the runoff is twofold: First, because the second round is limited to two candidates, one will necessarily win a majority, thus preventing a candidate from winning with the backing of only a small percentage of the electorate (which could happen if, say, five or six candidates split the vote rather evenly);37 second, runoff primaries may strengthen the party system by encouraging majority coalitions within the parties rather than factionalism. These arguments in favor of runoffs are unconnected with the issue of representation of racial or ethnic minorities. But it is often argued that runoffs have the effect of reducing the number of minorities elected38 and that in some instances, they were adopted for the express purpose of thwarting black candidates (Parker, 1990, p. 62; U.S. Commission on Civil Rights, 1981, pp. 49-50). When

Vote dilution in single-member districts

123

runoffs have been cited in litigation, however, it has most often been as an incidental part of a challenge to multimember or at-large elections (or to a combination of multimember and single-member districts). The remedies have involved a shift to single-member districts (or the addition of more majorityminority districts), and the matter of runoffs has not specifically addressed (e.g., Jeffers v. Clinton, 1989; Thornburg v. Gingles, 1986; Westwego Citizens for a Better Government v. City of Westwego, 1989). Only a handful of cases have challenged the use of runoffs in single-seat elections, though we expect the number of such cases to increase in the 1990s. One of the few single-member district cases is Butts v. City of New York (1985), in which the court rejected a claim that the 40 percent vote requirement in New York City's mayoral elections had been adopted for a racially discriminatory purpose. Litigation was also unsuccessful in a case involving the use of runoff primaries in one county of Arkansas. Although that decision, based on an effects rather than a purpose test, was initially vacated (Whitfteld v. Democratic Party of State of Arkansas, 1989), a divided en bane panel (no. 88-1953, 8th Cir., May 1990) affirmed the lower court decision, but with no written opinion issued. The Supreme Court denied certiorari on the case (1991).39 For a number of reasons, we find Section 2 challenges to the use of runoff elections for single-member districts to be problematic.40 First, just as there are instances in which runoffs help white candidates, there are circumstances in which runoffs help minority candidates. As Stanley (1987, p. 263) has pointed out, in majority-minority areas, a runoff requirement makes it less likely that a white plurality winner will be elected when the minority is dividing its vote among several minority candidates. Of course there are far fewer majorityminority districts than white majority districts, but it is still important to recognize the symmetry in these situations. Second, even if it is true that minority electoral success is reduced under a runoff primary in a particular jurisdiction, it can be argued that as long as the overall plan is racially fair (i.e., in all other respects), the outcomes of elections that minorities lost should perhaps not be identified as minority vote dilution, but as instances in which a group that was a voting minority was simply outvoted. Third, a sophisticated understanding of the way that runoff primaries work suggests that their elimination would not have the projected consequences for increased minority representation. In particular, the existence of runoff primaries affects the incentives for candidates to contest the election, with more candidates on average entering elections when there are runoff elections (Canon, 1978; Wright and Riker, 1989). Thus, the elimination of runoffs might simply mean that minority candidates would routinely face a single (white) opponent anyway (just as they would in a runoff).41

124

Minority Representation and Voting Equality

In any event, the mere fact that some minority members who had won a plurality victory have then lost in a subsequent runoff to a nonminority candidate ought not be sufficient to establish a Section 2 violation. Rather, we believe that it would, at a minimum, also be necessary to show that a higher proportion of minority members who have won a plurality subsequently lose the runoff than is true for nonminority members in the same situation, at least when there had not been proof that the runoff had been adopted with a discriminatory purpose.42 Standards other than the single-member district Americans are not accustomed to thinking in terms of anything other than the single-member district. In Gingles, for example, Justice Brennan asserted: The single-member district is generally the appropriate standard against which to measure minority group potential to elect because it is the smallest political unity from which representatives are elected. Thus, if the minority group is so small in relation to the surrounding white population that it could not constitute a majority in a single-member district, these minority voters cannot maintain that they would have been able to elect representatives of their choice. (1986, p. 50, n. 17) Because of this predominance of the single-member district system in the United States and because the language of Section 2 states that "nothing in this section establishes a right to have members of a protected class elected in numbers equal to their proportion in the population," proportional representation (PR) systems are usually considered inappropriate as baselines for comparing potential to elect or as actual remedies when voting rights violations have been found.43 There is, however, an entire range of proportional systems against which to compare minority potential to elect, and there are even stronger devices that guarantee real sharing of power. Because that entire range of systems and devices has been introduced into discussions of minority representation, we shall briefly discuss a number of these ideas and procedures. We begin with two voting systems (limited voting and cumulative voting) that are generally classed as "semiproportional," in that if voting is polarized along racial lines, these systems generally operate to provide at least a minimal level of minority representation but not proportionality as such. We then move on to systems that explicitly (single, nontransferable vote) or potentially (weighted voting) provide for PR. Finally, we mention procedures that call for the explicit sharing of power in decision-making bodies. Semiproportional systems Proponents of semiproportional systems argue that they offer an ideal compromise between the interests of fair and effective representation for minorities and

Vote dilution in single-member districts

125

the concerns of those who wish to maintain some form of at-large representation. As in an at-large system, cumulative voting and limited voting permit both minority and nonminority voters the option of voting for candidates of either their own race or ethnicity or for candidates of a race or ethnicity different from their own. But these methods also permit a voter to structure ballot choices more finely on the basis of whatever criteria (racial, partisan, ideological, or otherwise) the voter might wish. Opponents, of course, contend that despite any disclaimers, semiproportional systems usually yield more nearly proportional representation than does singlemember districting and therefore move us closer to what is explicitly not guaranteed by the Voting Rights Act or the Constitution. More important, opponents argue, such systems would represent a fundamental shift in our representation system away from geographically based districts to a system designed to foster group representation and thus would reinforce a racial and ethnic basis for representation at a time when the nation is trying to reduce an emphasis on racially based politics.44 In any event, our reading of the case law is that courts have been unwilling to require use of semi-PR methods (e.g., McGhee v. Granville County, 1988) but that when jurisdictions propose alternative election mechanisms intended to achieve some forms of minority representation, they would be held constitutional. As one federal district court said in considering the constitutionality of a Connecticut statute providing for both limited votes and limited nominations (from each political party), "it is hard to fault minority representation as nondemocratic or impermissible as a legislative goal: . . . it is not anti-majoritarian to limit the power of the majority to command more power than its actual strength at the polls" (Lo Frisco v. Schaffer, 1971, p. 750).45 Limited voting. With the so-called limited vote, voters have fewer ballots to cast than there are seats to be filled.46 Variants of the limited vote are used in countries such as Spain and Japan. Since 1870, over half a dozen cities, including New York, Indianapolis, and Boston, have at one time or another used the limited vote for primary or general elections. Philadelphia has used this method since 1951 for its seven at-large council seats. Pennsylvania counties have used limited voting since 1871. According to current law, all counties in Pennsylvania except for Philadelphia and a few others under home rule charters elect county commissioners under a limited voting system in which individuals can vote for only two candidates out of the three to be elected at large. A Connecticut statute adopted in the early 1960s provided for limited voting for all local school board elections in the state. Hartford and other Connecticut cities and counties currently use limited voting.47 The Democratic party in Conecuh County, Alabama, under pressure to

126

Minority Representation and Voting Equality

achieve some representation for racial minorities, adopted limited voting in 1982 for governing bodies internal to the party (Still, 1984). As a consequence of settlements reached in the massive Dillard v. Crenshaw County (1987) litigation, limited voting was adopted as a remedy in several local Alabama governments (Still, 1989, 1990). There it has largely worked as anticipated to provide greater minority representation. Limited voting was also adopted for use in the at-large component of a mixed election scheme accepted as a settlement to a U.S. Department of Justice lawsuit against Augusta, Georgia (United States v. City of Augusta, 1987). Cumulative voting. Under cumulative voting, voters may choose to express their intensity of preference by casting multiple votes (up to a fixed total). The cumulative vote was used from 1880 to 1980 in elections to the lower chamber of the Illinois legislature. There, voters had three votes; they could give all three to a single candidate, divide them between two candidates (two votes to one and one vote to the other, or one and one-half votes to each), or give one vote to each of three different candidates. The use of cumulative voting in Illinois was ended largely as an incidental consequence of a Republican-backed referendum to reduce the size of the legislature. Rockford, Illinois, used cumulative voting for a brief period in the 1880s. Cumulative voting is also used in many U.S. states for electing corporate boards of directors (Glazer, Glazer, and Grofman, 1984). Cumulative voting was recently adopted in Alamagordo, New Mexico, as part of a settlement to a voting rights lawsuit and in Sisseton, South Dakota, in a voting rights case involving at-large school board elections. In the first elections held under the new system, many minority voters used their opportunity to cumulate their vote - 64.2 percent of the Hispanic voters in Alamagordo and 93.4 percent of the Native American voters in Sisseton cast all their ballots for a single candidate - and minorities were elected roughly in proportion to their numbers in the population (Engstrom and Barrilleaux, 1991; Engstrom, Taebel, and Cole, 1989). Cumulative voting has also been adopted as part of a settlement in a number of the jurisdictions sued in Dillard v. Crenshaw County (Still, 1989, 1990). Proportional representation The arguments for and against proportional representation are similar to those about semiproportional systems. Proponents of strict PR argue that no "lesser" methods provide minorities with the full amount of representation they are due. Opponents, of course, counter even more vociferously than over semiproportional systems that PR is nowhere required and that adopting such methods

Vote dilution in single-member districts

127

would fundamentally alter our system to one of group representation. By and large, the courts have not yet been asked to rule on such systems, aside from a very limited excursion into weighted voting. The single-nontransferable vote and mixed systems. Among the strict PR systems, the single transferable vote (STV) is the only type ever to have been used in the United States. Since 1915, about two dozen U.S. cities have used STV (the Hare System) for city council and school board elections (Weaver, 1986, p. 140). Most cities, however, used STV for only a short period. One wave of adoptions occurred in the early 1920s, and another wave of adoptions came in the 1940s. By 1954, there were only six U.S. cities still using STV for city council elections, and by 1982, only Cambridge, Massachusetts.48 Cambridge also uses STV for its citywide school board. New York City has used STV for its community school board elections since the early 1970s, though a proposal to eliminate its use is currently (December 1991) pending in the New York state legislature as part of a general package of reform of school administration. Weighted voting. In weighted-voting schemes, representatives are not equal in the number of votes they cast. Frequently, the number of votes that a representative is entitled to cast is a function (sometimes linear, sometimes simply monotonic) of the number of persons he or she received. The best-known example of weighted voting is in corporate settings, in which proxy holders cast a number of votes in proportion to the number of proxies they hold. Weighted voting has occasionally been suggested as a possible voting rights remedy, in that minority groups might obtain representation but have their representative^) be given a smaller weight, reflecting the size of the groups. Although it continues to be used in some county governments in New York (Grofman and Scarrow, 1981b) and has been accepted in a per curiam decision of the Supreme Court (WMCA, Inc. v. Lomenzo, 1964), the most recent federal court decision regarding its use was quite negative (Morris v. Board of Estimates, 1989).49 Power-sharing devices Whatever the election method used, a minority in the electorate can be expected to remain a minority in the legislature. Although there is evidence that the presence of minority officeholders changes the dynamics of majority decision making (Fraga, 1991), several scholars skeptical of this claim have been searching for ways to increase minority influence in systems that are fundamentally majoritarian in tone. Karlan (1989), Still (1990), and Guinier (1991, 1992), for example, have argued in favor of formal mechanisms to give "proportionate

128

Minority Representation and Voting Equality

influence," such as those found in some consociational democracies, in which, for example, groups may be allocated control of shares of certain government budgets (Lijphart, 1977). Such mechanisms include rotation of the presiding officer in local councils or imposition of provisions requiring more than a simple majority vote for passage of key legislative items. As part of the Dillard v. Crenshaw litigation, Still (1990) was successful in obtaining rotating chairs in a handful of jurisdictions. The problems that will be encountered in the 1990s have a familiar ring because most were raised in the multimember district cases that formed the heart of voting rights litigation in the 1980s. Despite their familiar sound, however, many of these problems are very new, as they were rarely the focus of the 1980s cases. As a consequence, we foresee in the first part of the 1990s a large number of cases dealing with single-member elections in which courts will enunciate conflicting standards of judgment. Perhaps by the middle of the decade, a case will arise that will allow the Supreme Court to arrive at a new set of standards uniquely applicable to the single-member district context.

6 The Voting Rights Act and the realistic politics of second best: an optimistic look to the future

In the previous chapters we have largely limited ourselves to descriptions of existing case law and closely related, unresolved issues. Yet the development of this case law - especially the expansive interpretation of Section 2 of the amended Voting Rights Act - has been highly criticized (Butler, 1990; Graham, 1992; O'Rourke, 1992; Thernstrom, 1987), and the continuing stream of decisions, dealing with less obviously racially motivated and racially dilutive election systems and laws, is likely to be even more controversial. If there has been less public discussion of representation than of other areas of race relations, such as the use and interpretation of test scores or access to jobs and promotions, it is because districting is much less visible to the average person.1 But matters of representation are arguably more fundamental to the future of our country than are issues that have grabbed the most attention, for the resolution of those policy issues, as well as the legitimacy of the government itself, hinge in part on the composition of our representative bodies. With this in mind, we turn our attention in this chapter to normative aspects of the Voting Rights Act and its interpretation in the courts.

Legislative intent and the right to vote versus the right to representation The application of the Voting Rights Act to matters of representation has sometimes been challenged as inappropriate, on the grounds that its use against vote dilution was never fully debated and never sanctioned by Congress. Judge Garnett Thomas Eisele, in a dissenting opinion in Jeffers v. Clinton (1989, p. 227), forcefully makes this point: It is my view that many Voting Rights cases, such as this one, are changing the political landscape of America in fundamental ways without legislative mandate and without the benefit of scholarly legal and political discourse. In so doing these cases are, in an almost 129

130

Minority Representation and Voting Equality

inadvertent manner, redefining the nature of our democratic form of government, contrary, I believe, to the Constitution. Judge Eisele then goes on to assert that "when the Voting Rights Act was passed in 1965, its single aim was 'black enfranchisement in the South. Obstacles to registration and voting, that is, were the sole concern of those who framed the statute' " (quoting Thernstrom, 1987, pp. 3-4). We believe that Judge Eisele was wrong in accepting Thernstrom's characterization of the legislative history of the Voting Rights Act. We are highly skeptical of the claim that Congress intended to do no more in the Voting Rights Act than to guarantee to minorities the right to vote.2 First, the concept of vote dilution as minimizing or canceling out minority voting strength antedates the Voting Rights Act, as is clear from court decisions just before the Voting Rights legislation and in much older race-related actions such as passage of white primary laws (Kousser, 1992). The redrawing of Tuskegee, Alabama's boundaries that was overturned in Gomillion v. Lightfoot (1960) did not deny blacks an opportunity to vote; it changed where they could vote and thus changed their ability to affect outcomes.3 The legal challenge to multimember districts that was being carried out at the very time the Voting Rights Act was being considered was brought by whites, but the Supreme Court referred specifically to possible racial as well as political discrimination (Fortson v. Dorsey, 1965, p. 439).4 Thus, when the Voting Rights Act was passed in 1965, members of Congress surely recognized that white domination could be maintained through a variety of electoral devices and that equality of the right to vote as such would not necessarily ensure equality of black participation in any larger sense of the term. Second, to the degree that there was ignorance of or doubt about the possibility of vote dilution at the time the Voting Rights Act was passed, there was ample evidence in the years following that southern states were adopting a host of strategies expressly to reduce the practical effects of black enfranchisement (see Chapter 2, n. 22). Had Congress wished to overturn court interpretations of the act and mandate a narrow interpretation, it could have done so by passing clarifying language. Instead, in 1970 - after the Supreme Court decision in Allen - Congress extended the act. Five years later Congress again extended the act, expanding the number of groups covered. In 1982 Congress once again extended the act. This time, in effect, it overruled a Supreme Court decision; rather than narrowing the impact of the act, it explicitly reversed the impact of Mobile v. Bolden, making it clear that the relevant statutory legal standard was an effects test and that electing representatives, as opposed to merely casting ballots, was a crucial component of the act. Thus, however one views Congress's initial intent, legislative approval of prior court decisions and a mandate for future actions can be found in the revised language of Section 2.

The Voting Rights Act and the future

131

A color-blind society versus the institutionalization of racial/ethnic divisions Judge Eisele was also concerned with the effects of the Voting Rights Act: "Do we really believe in the idea of one political society or should this be a nation of separate racial, ethnic, and language political enclaves?" (Jeffers v. Clinton, 1989, p. 227). Again, Eisele was echoing Thernstrom's argument, in this case that the act is ' desegregating" or at least institutionalizing existing racial/ethnic divisions in American society. The underlying argument is that recent interpretations of the Voting Rights Act are rooted in a group-based theory of politics that is fundamentally at odds with the "color-blind" spirit of the U.S. Constitution and with our historical struggle for political equality. We reject this view of the Voting Rights Act as well, at least as it has been interpreted up to now. We would emphasize, first, that the rights provided by the act are contingent, appropriate only when a significant liability threshold has been met. Only when African-Americans or Hispanics are made a "permanent minority" as a result of racial bloc voting by the majority or by various practices and procedures is there intervention under the Voting Rights Act. Intervention in such circumstances is, we believe, in accord with the Madisonian tradition (as articulated in numbers 10 and 51 of the Federalist Papers), which condemns factions, even majority factions, and seeks to design constitutional rules that serve as safeguards against the most pernicious consequences of such factionalism. Because the rights are contingent, the applicability of Section 2 of the Voting Rights Act, like Section 5, is, in principle, "self-liquidating." For Section 5 the obvious self-liquidating feature is the bailout provision. For Section 2, the relevant features are those used to determine a violation, chiefly the three prongs of the Gingles test.5 Moreover, the three conditions - residential segregation sufficient to allow the drawing of districts in which minority group members are a majority, racially polarized voting, and a "usual" lack of minority electoral success - are conditions that few people wish to see perpetuated. Thus, if minority assimilation proceeds in such a fashion that residential segregation becomes a thing of the past, minority groups will be unable to launch successful voting rights suits. Or if voting in a jurisdiction is no longer polarized along racial or linguistic/ethnic lines - or even if it is, but the level of white crossover voting permits significant and repeated minority success - the Voting Rights Act will become a dead letter in that jurisdiction. Second, though it is true that most other voting rights violations (e.g., of the one-person, one-vote standard) are customarily defined in terms of the violation of individual rights, clearly there are types of discrimination directed against individuals as a function of their status as members of a minority community.

132

Minority Representation and Voting Equality

In such situations it seems foolish to think that liability and remedies cannot be race conscious. In the present context, because racially polarized voting is a prerequisite for a voting rights violation and residential segregation a prerequisite for submergence, it is not plausible to attempt only "color-blind" tests and solutions. Furthermore, to argue that policies in the voting rights area must be free of all considerations of race and ethnicity is like blaming the messenger for the message. Enforcement of the Voting Rights Act cannot be blamed for causing the conditions (such as segregated housing and racially polarized voting) that litigation involving it brings to the light of day. Moreover, there seems to be no factual basis for asserting that enforcement of the Voting Rights Act has led to an increase in racial polarization by making race a more salient feature of politics than it had been previously. Certainly Thernstrom (1987) provides no empirical evidence to support this claim. In terms of voting behavior, the best evidence suggests relative constancy of polarization in elections in majoritywhite districts in the deep South (Grofman and Handley, 1991; cf. Loewen, 1990) and diminished polarization on the part of whites in majority-black areas as voters accept the fact that the winner will be black and as they experience the reality that black elected officials are not as undesirable as they might have previously feared. The specter of proportional representation It is sometimes argued that requirements of racial fairness are pushing us headlong toward a system of proportional representation, despite the congressional disclaimer of such a requirement (for racial minorities) in the revised Section 2 of the Voting Rights Act and despite repeated statements by the Supreme Court that the Constitution does not guarantee proportional representation for any group. In a strict sense, of course, the claim that fairness to minorities is a proportional representation test in thin disguise is simply wrong, at least insofar as the basis for representation remains the single-member district. Singlemember district systems cannot be expected to yield proportional representation (Taagepera and Shugart, 1989; Tufte, 1973). Strict interpretation aside, however, the question is an important one. Surely the test for vote dilution as embodied in Gingles is not explicitly a form of proportional representation. Not only is a finding of a violation conditional on residential segregation, racially polarized voting patterns, and a consistent lack of minority electoral success, but the remedy usually involves winner-take-all elections in single-member districts. Moreover, if we look at empirical evidence from state legislatures or the U.S. Congress, minorities remain very much underrepresented.6 And it is likely that they will continue to

The Voting Rights Act and the future

133

be underrepresented (by any single-member district plan) as long as racially polarized voting persists, because there are not enough sufficiently large, geographically concentrated minority populations to make possible the creation of many more heavily minority single-member districts (Grofman and Handley, 1991).7 Moreover, from a normative perspective, fair representation would seem to require that numbers of representatives bear some relationship to the kinds of individuals represented. In fact, precisely this point was the basis for the preBaker controversy over rural malapportionment. Existing schemes were judged to be unfair because rural areas were represented in numbers all out of proportion to their populations. That the groups that are severely underrepresented at present are not exclusively geographic and are not a majority does not undercut this point. Blacks and Hispanics cannot expect to have a number of representatives exactly equal to their numbers in the population, both for the technical reasons alluded to earlier and for the more important reason that ours is not a system of proportional representation. But that does not mean that any disproportionate results are acceptable or even that moderately disproportionate results are acceptable, whatever the cause. Blacks cannot be guaranteed 11 or 12 percent and Hispanics 8 percent of the representatives on each elective body (or comparable amounts based on their proportion in a particular area), but neither minorities nor whites/Anglos should be content with minority percentages far below their proportions of the population, let alone with no representatives whatsoever.8 Legal questions aside, reasonable individuals might differ on what degree of proportionality precludes corrective action. And in an ideal world, voting would not be sharply polarized, so that the relationship between representation and race, ethnicity, gender, religion, and other such characteristics would be far less important (and, in any event, degrees of nonproportionality would presumably be quite small). But when disproportionality is great and when attitudes and interests differ radically across groups - as with rural/urban differences of the 1950s or current racial/ethnic differences - corrective steps must be taken if our system is to be regarded as fair. Moreover, if a lack of proportionality or a reasonable approximation thereof comes about because of discrimination, it surely cannot be tolerated. Yet the burden of proof should not rest entirely on the matter of intent. Even in the 1960s it was often difficult to prove intent to discriminate. Some electoral laws are of long standing, so the historical record is not well preserved. Few laws can be justified by only one rationale - whatever their true intent in the minds of legislators - so it often is possible, at least, to present alternative reasons for adopting electoral reforms. With contemporary reforms it is perhaps even more difficult to prove discriminatory intent. For one thing, supporters are not so

134

Minority Representation and Voting Equality

likely as in the pre- 1960s South to claim such purpose. For another, the types of discrimination are likely to be more subtle than in earlier years and are especially likely to be cloaked in a disguise of good government. There is no doubt that litigation surrounding the Voting Rights Act has brought the matter of proportionality to the fore. But what it has most properly called to our attention are the gross inequities in representation - a level of unequal treatment that in our opinion deserves corrective action. Given the remaining level of inequities - more than twenty-five years after passage of the Voting Rights Act - our concern remains much more with the considerable ^^proportionalities than with a theoretical possibility that is most unlikely anyway in single-member, winner-take-all systems. The Voting Rights Act and the future of American politics We have already rejected the view that the Voting Rights Act was never intended to deal with problems of representation. A closely related view is that the act has outlived its usefulness and should be ended. Self-serving reasons aside, some would agree with the statement in 1981 by Craig Washington, then a state senator (black) in Texas: "We are sophisticated enough that we have arrived at the point where black people can represent white people and white people can represent black people and brown people can represent either or both" (Texas Observer, June 12,1981, p. 7, quoted in Anderson and Dahlstrom, 1990, p. 71). We agree with the sentiment expressed by Washington (now a U.S. congressman) but are skeptical about its empirical reality. Unpalatable as it may be, the simple truth is that at the congressional and state legislative level, at least in the South, blacks are very unlikely to be elected from any districts that are not majority minority, and most majority-black legislative districts and all majorityblack congressional districts now elect black officeholders (Grofman and Handley, 1991). To be sure, blacks have sometimes been elected in majority-white areas, but such situations are not that common, despite various well-publicized cases. And when they do occur, they can often be attributed to overwhelming support from black voters combined with limited support from Hispanics, Asians, or whites (e.g., the mayoral elections of David Dinkins in New York City and Harold Washington in Chicago, the gubernatorial election of Douglas Wilder in Virginia), or to plurality victories against a divided opposition (Wilder's initial legislative victory against a field of white candidates). Moreover, it is easy to forget how little support winners such as Dinkins or Wilder actually received from whites in their successful races (27 percent and 26 percent, respectively).9 Furthermore, although patterns of polarization may not be as stark for Hispanics

The Voting Rights Act and the future

135

as for blacks, with one important exception (New Mexico), they are uncomfortably similar (Grofman and Handley, 1989). In a similar vein, we agree with Swain (1989) that in the long run, major additional gains in black (or Hispanic) representation can come only by building coalitions that make possible the election of minority candidates from districts that are not majority or near majority minority, but we do not see this as a reason to stop trying to create black or Hispanic majority seats where there is evidence of vote dilution. We also differ with Swain and other scholars who look to multiracial coalition building as the primary direction for future black (or Hispanic) politics, so severe, in our perception, is the level of present-day racial polarization. At the same time, we share with Swain and others the concern that there is only a limited prospect for further gains in African-American representation in Congress or the state legislatures from the creation of additional majority-black districts. Patterns of black geographic distribution are such that only a relatively small number of additional black majority legislative and congressional seats can be created. Moreover, there are also limits to the foreseeable gains in black representation from further shifts from at-large to single-member district representation at the municipal level, although there are greater prospects for gains in black representation at the level of school boards. For Hispanics, however, there are still great gains in representation possible at the local level with the elimination of at-large city council and school board elections (e.g., in Texas and California; see Brischetto et al., 1990), and further gains to be made in Congress and the state legislatures because of the dramatic increase in the Hispanic population over the past decade. A variant of the argument in favor of biracial coalitions is the suggestion that a pursuit of black majority districts may ultimately be counterproductive because it reduces the possibility of distributing blacks' votes in such a fashion that they can become pivotal members in cross-racial coalitions that elect white liberals (see, e.g., Schuck, 1983; Swain, 1989; Thernstrom, 1985). Here, although we recognize that white liberal legislators may vote similarly to their black counterparts on roll call votes, this does not mean that they have the same commitment to a leadership role on civil rights or on economic issues of concern to the black community. Also, the view that black votes should be used to (re)elect white liberals is a variant of a very old style of paternalism. The rejection of black claims to self-representation in favor of the interests of white incumbents has been a factor retarding black representation.10 Although revisionist critics of the Voting Rights Act thus see it as accentuating the importance of racial and linguistic cleavages, we see it as forcing an assessment, on the basis of case-specific evidence, of the reality of those cleavages and an effort to give minorities a full opportunity to be part of the political

136

Minority Representation and Voting Equality

process when those divisions are especially strong. Rather than fearing the election of minority candidates from largely minority constituencies, we ask whether we would really prefer the most likely alternative - the lack of minority electoral success still characteristic of majority-white jurisdictions in the South, especially those at the state legislative and congressional level. Whereas some revisionist critics see the success of the Voting Rights Act threatening our aspirations for a color-blind society, we see it as a necessary evil in a color-conscious world admittedly the "politics of second best." Yet our view of the "politics of second best" is ultimately an optimistic one. Though we may agree with Guinier (1992, in press) that the increase in the number of black elected officials "has not visibly altered the disadvantaged socioeconomic condition or social isolation of black voters" and that sustained black mobilization has not emerged despite some black electoral success (1990, pp. 1109-1112), we would also say that too much is being expected of the franchise. Some suffragettes, it is said, thought that achieving the vote would bring an end to war. That they were wrong does not make us want to repeal the Nineteenth Amendment. Similarly, the fact that the rhetoric concerning what was to be expected from black enfranchisement now seems dramatically overstated ought not to lead us to dismiss the real gains that have resulted from that enfranchisement - the change, for example, from fewer than 1,500 elected officials in 1970 to over 7,300 in 1990 (Stanley and Niemi, 1992, p. 393). The strong likelihood that minority elected officials will be outvoted by nonminority legislators once in office, or the possibility that some minority legislators will be diverted from serving the interests of the minority community (especially if elected from a majority-white constituency), does not vitiate the need for their election. Minority politicians may fail - indeed, will fail - to deliver fully on a substantive agenda serving their constituency's interests, but that is the nature of politicians of all colors and does not cause us to give up on the processes of representative democracy. Moreover, if we are ever to dispel the various myths surrounding minority participation - that minorities are unqualified to serve in leadership positions, that they will put forward only their own agendas, that minorities cannot represent whites - it is virtually required that minority members be elected and that they share all of the various experiences, including both triumphs and failures, representation of both broad and narrow interests, that are the lot of majority politicians. Still another reason for optimism over a broad interpretation of the Voting Rights Act is that majority-minority districts are serving as a necessary "port of entry" for minorities into pluralist politics. The opportunity to hold office and to make a record while in office is perhaps the most important means whereby minority candidates can establish a reputation that will earn them considerable crossover support. Even Governor Wilder, one of the most celebrated

The Voting Rights Act and the future

137

examples of black success in a majority-white constituency, was initially elected to the state legislature as a plurality victor in a district with a significant black population and was reelected to the legislature from a district that had been reconfigured to have a clear black majority. As with whites/Anglos, success of minority politicians at one level can be parleyed into higher office. But the important difference is that minority politicians are often able to get started because of majority-minority districts.11 Finally, we would emphasize that cross-racial coalitions may be easier to achieve at the level of political elites than at the mass level. Political scientists have shown repeatedly that greater educational attainment and elite status itself are highly correlated with a fuller understanding of democratic norms and a greater understanding of and willingness to tolerate diversity (McClosky and Zaller, 1984, chap. 2). But of course, multiracial/ethnic elite interaction demands that minorities as well as majorities be elected in the first place. And as we noted, that has been greatly facilitated by the implementation of the Voting Rights Act. We do not presume, of course, that racial prejudice vanishes within legislative halls. There are surely instances in which the influence of minority elected officials on outcomes appears to have been minimal. Guinier (1991, p. 1117), for example, cites testimony from four black Arkansas state legislators elected in the 1980s who saw themselves as "an ineffective minority" unable to get whites to cosponsor legislation of concern to blacks. Or one might cite the experience of Mayor Harold Washington in the early part of his first term in Chicago, in which whites loyal to old-style machine politics controlled the majority of seats on the city council and effectively excluded the mayor and his black (and Hispanic and liberal white) supporters from any influence. But there is considerable evidence for political change and accomplishment, as well, some of it anecdotal and limited in scope but increasingly representing larger-scale, systematic investigations by social scientists.12 Of course we have not tried to hide the fact that there remain significant differences of opinion about implementation of the Voting Rights Act; we expect that litigation surrounding the act will remain with us well past the next decennial census, at least. And at some point, we hope the argument will be correct that the act has outlived its usefulness and is no longer necessary. For now, however, we are convinced that by ensuring that the right to vote is not an empty ritual but that minorities will be involved in decision-making bodies as well as in campaigns, the Voting Rights Act serves to integrate minorities into the American political process and helps ameliorate the alienation that came from their previous exclusion. If that is so, the act will be seen as one of the most important and successful pieces of legislation of this century, and its broad interpretation one of the most important achievements of the courts.

Notes

Notes—Introduction 1. In 1990, approximately 5.4 percent of state legislators and 1.5 percent of all elected officials were black, compared with 11.1 percent of the voting age population. Hispanics made up 1.7 percent of the state legislators and 7.3 percent of the voting-age population. Figures are from Stanley and Niemi, 1992, pp. 395-397. 2. Following the conventions of social science, but not legal, writings, we give full citations in the list of references at the end of the book and identify sources in the text or notes with only the author's last name (or the case name) and the year of publication. When specific pages need to be cited and the source has already been referenced in the immediately preceding sentences, we give only the page number. 3. Our review is carried through December 1, 1991, though insofar as possible, we have updated references beyond that date.

Notes—Chapter 1 1. Blacks could vote freely in Maine, Massachusetts, New Hampshire, Rhode Island, and Vermont, and they could vote in New York if they could satisfy a property requirement of $250 (a condition that did not have to be met by whites wishing to vote). 2. An argument has been made that the Fifteenth Amendment was actually designed to enfranchise blacks in the North as blacks in the South had already been enfranchised by the reconstruction Act of 1867 (Gillette, 1969, pp. 46-49). 3. Ex Parte Yarbrough (1884) was the one notable deviation from the path forged by Reese. In Yarbrough, the Supreme Court affirmed the convictions of eight white men who were charged with conspiring to beat a black man in the exercise of his right to vote in a federal election - which was a violation of Section 6 of the Enforcement Act. The Court upheld the constitutionality of the statute in this case, recognizing that Congress had the power to protect the right to vote in congressional elections. 4. Although southern legislators did furnish loopholes such as the "understanding clause" and the "grandfather clause" for poor or illiterate whites, literacy tests and poll taxes were also meant to, and did, disfranchise large numbers of poor whites (Kousser, 1974, passim; Woodward, 1971, chap. 12). 5. Alabama, Georgia, Louisiana, Mississippi, North Carolina, South Carolina, and Virginia passed statutes requiring a literacy test in one form or another; all eleven ex-Confederate states made the poll tax a prerequisite for voting (Key, 1949, pp. 557, 578). 6. For example, the list of disenfranchising crimes listed by the 1895 South Carolina Constitutional Convention included such acts as fornication, sodomy, assault with intent to ravish, miscegenation, incest, and petty larceny. Not included in the list were theft and murder (Kousser, 1984, p. 35). 7. Lewinson reports that reliable black registration statistics are available for only one state, Louisiana, during this period. In the 1896 national election 130,344 blacks were registered to

138

Notes to pages 11-24

8.

9. 10.

11.

12.

13. 14.

15. 16.

17. 18.

19. 20.

21.

22.

139

vote, and by 1900 (two years after the disenfranchising convention) only 5,320 remained on the rolls - a 96 percent decrease (Lewinson, 1932, p. 81). The first two cases declared that exclusion of blacks from primary elections was unconstitutional when dictated by state statute (Nixon v. Herndon, 1927) or when mandated by the state executive committee in the exercise of the power delegated to it by the state legislature (Nixon v. Condon, 1932). But in Grovey v. Townsend (1935), the Supreme Court upheld the exclusion of blacks from party primaries when required by a resolution of the state party convention acting on its own. Smith v. Allwright (1944) overturned Grovey nine years later. The three earlier cases had been decided on the basis of the Equal Protection Clause of the Fourteenth Amendment. It was hoped that repealing all state laws relating to primary elections would mean that Smith v. Allwright would no longer apply. The major decisions that voided these efforts were Elmore v. Rice (1947), and Brown v. Baskin (1948). See Key, 1949, chap. 29, for a discussion of these cases, as well as other attempts to resuscitate the white primary. Before this decision it was estimated that approximately 151,000 blacks were registered to vote in the South (less than 5 percent of the adult black population). By 1947, the estimated number of blacks registered to vote had increased to 595,000 (Garrow, 1978, p. 7). United States v. Raines involved the Board of Registrars of Terrell County, Georgia. When the complaint was filed in 1958, only forty-eight blacks (1 percent of the blacks in the county) were registered to vote. The district court dismissed the case on the grounds that the 1957 Civil Rights Act was unconstitutional, but the Supreme Court reversed this holding. The entire state of Alaska and four counties in Arizona, as well as single counties in Hawaii and Idaho also were originally included under the "trigger" formula, but most of these jurisdictions successfully sued to remove themselves from coverage. The phrase "test or device" was defined in Section 4(c) to mean any requirement that a potential voter (1) demonstrate the ability to read, write, understand, or interpret any matter, (2) demonstrate any educational achievement or his knowledge of any particular subject, (3) possess good moral character, or (4) prove his qualifications by the voucher of registered voters. The poll tax had already been banned in federal elections with the ratification of the Twentyfourth Amendment in 1964. The 1970 formula also captured for the first time a county in Wyoming, two counties in California, five counties in Arizona, and a number of local political subdivisions in Connecticut, New Hampshire, Maine, and Massachusetts. In addition, some counties that had been exempted after 1965 were re-covered in 1970: four election districts in Alaska, a county in Idaho, and several counties in Arizona (U.S. Commission on Civil Rights, 1975, pp. 14-15). Congress adopted the Twenty-sixth Amendment to provide for the right of eighteen-year-olds to vote in all elections shortly after the decision was rendered. This amendment was ratified in 1971. Congress amended the act such that if any "registration or voting notices, forms, instructions, assistance or other materials or information relating to the electoral process, including ballots," were only in English and that jurisdiction had a substantial language-minority population, then that jurisdiction would be considered to have employed a "test or device." The Twenty-fourth Amendment prohibits the use of a poll tax as a prerequisite for voting in federal elections, and Harper v. Virginia State Board of Elections prohibits its use in state elections. The registration figures presented in Tables 1 and 2 are based on surveys, and the reported registration is known to overestimate actual registration. Actual registration data by race are available for a few covered jurisdictions, and this information supports the general trends discussed earlier. Parker (1990, pp. 205-206), for example, cites restrictive registration laws in Mississippi that impinge more heavily on blacks. With respect to another minority, Henderson and Wolfley (1990) relate the story of a registrar who told Native Americans that they could be given only a few forms for remote registration, while handing out dozens of such forms to inquiring whites. The situation of Hispanic-Americans is complicated when questions are raised about citizenship. (1) At-large elections (or multimember districts) mean that a group that is a minority of the

140

23. 24. 25. 26.

27. 28. 29. 30.

Notes to pages 24-32 population may be outvoted for all legislative seats (or all seats in a given district); (2) antisingle-shot laws, which prohibit voting for just a single candidate (in multicandidate situations), prevent minority groups from concentrating their votes on one candidate; (3) decreasing the size of legislative bodies (thereby increasing the size of districts) means that minorities may have enough voters to control very few or no districts; (4) racial gerrymandering splits concentrations of minority voters among districts or, in some cases, packs them into a small number of districts; (5) exclusive slating practices give minorities little voice in candidate selection; and (6) changing offices from elective to appointive has obvious effects. Numerous examples of these devices, and additional sources, are given in Davidson, 1984; Davidson and Fraga, 1984; Parker, 1984; and U.S. Commission on Civil Rights, 1975, chaps. 8-9; 1981, chaps. 5—6. Parker (1990, chap. 2) provides a detailed account of the devices used in Mississippi. See also the state-specific chapters in Davidson and Grofman, 1992. Ball, Krane, and Lauth, 1984; Parker, 1990, chap. 2; U.S. Commission on Civil Rights, 1975, chaps. 8-9; 1981, pp. 70-73. This wording was adopted in 1982, as discussed in Chapter 2. Nondilution would then be defined as an "equal opportunity to elect representatives of their choice." In Beer v. United States (1976), for example, the Supreme Court ruled that a New Orleans City Council reapportionment plan that increased minority representation from zero to one did not violate Section 5 of the act, even though New Orleans had a black population of about 45 percent. Actually, Guinier (1991) goes even further, suggesting what she calls "proportionate influence." It would be achieved by devices such as supermajority voting requirements and rotation of legislative presiding officers. See, for example, Rae, 1971. In addition, the Supreme Court has frequently pointed out that the Constitution does not mandate proportional representation. See, for example, City of Mobile v. Bolden, 1980, pp. 75-76, 79. These are the mechanisms described in note 22.

Notes—Chapter 2 1. Laughlin McDonald, personal communication. 2. As of this writing (December 1991), the Supreme Court has not reviewed other Section 2 cases except in the form of denial of certiorari or summary affirmance and in its decision in Chisom v. Roemer (1991) and Houston Lawyers Association v. Attorney General of Texas (1991) that judges are representatives in the context of the Voting Rights Act. 3. Actually, Allen was a consolidation of four cases, three challenging various provisions from Mississippi and one from Virginia (involving how illiterate voters could vote for write-in candidates). 4. Changes in municipal or other political boundaries, like changes in the lines of individual districts, may now be subject to Section 2 challenge, although until now virtually all annexation and deannexation cases have been dealt with under the Section 5 nonretrogression test (see n. 6) or under the Fourteenth and Fifteenth amendments directly, as in Gomillion v. Lightfoot (1960). In Gomillion, Tuskegee, Alabama, removed from the city's perimeter all of the heavily black areas, creating "an uncouth twenty-eight-sided figure" {Gomillion v. Lightfoot, p. 340). There is no way to understand this deannexation except as a device to reduce the importance of black voters by "fencing them out" of the city. The Supreme Court rejected that attempt as a violation of the Fifteenth Amendment. 5. Section 5 currently applies to nine states in their entirety (Alabama, Alaska, Arizona, Georgia, Louisiana, Mississippi, South Carolina, Texas, and Virginia) and to portions of seven other states. 6. Since at least 1986 (formalized in new Section 5 regulations adopted in 1987), the Justice Department has withheld preclearance to plans that result in a clear Section 2 violation. 7. Many of the cases litigated during the 1970s made claims as well under Section 2 of the Voting Rights Act, but most vote dilution cases handled by the courts during that period were decided on constitutional grounds (Parker, 1983, p. 727).

Notes to pages 33-^6 8.

9. 10.

11.

12.

13. 14. 15. 16.

17. 18.

19.

20.

141

Although the multimember district issue was the focus of much attention, other procedures that blacks believed were being used to dilute their vote were also being challenged in the early 1970s. The most prevalent of these dilutive practices, such as numbered-place laws and full-slate requirements, were often used in conjunction with multimember or at-large districts. In a number of instances, such as in the Carolinas, such election law changes were challenged in the courts and struck down (see, e.g., Dunston v. Scott, 1972, and Stevenson v. West, 1973). Another popular dilutive technique was annexing white areas in order to lower the ratio of black to white citizens. The Supreme Court affirmed the court of appeals' order mandating single-member districts in East Carroll Parish, Louisiana, but expressly "without approval of the constitutional views expressed by the Court of Appeals," in East Carroll Parish School Board v. Marshall (1976). There were clear indications before Bolden that the Court might adopt an intent standard for vote dilution cases. In Washington v. Davis (1976), a case involving racial discrimination in employment, the Supreme Court held that a showing of intent to discriminate was necessary to substantiate a Fourteenth Amendment equal protection claim. The Court explicitly rejected the position that "a law or other official act . . . is unconstitutional solely because it has a racially disproportionate impact" (p. 239) and held that "the invidious quality of a law claimed to be racially discriminatory must ultimately be traced to a racially discriminatory purpose" (p. 240). The district court in Bolden concluded, however, that the failure of the Supreme Court to overrule or comment on the vote dilution cases, including White, indicated that Washington did not overrule those cases or establish a new Supreme Court intent test in vote dilution cases. The Fifth Circuit, on the other hand, had attempted to reconcile Washington with previous vote dilution cases by holding that a showing of racially motivated discrimination was a necessary element in a vote dilution claim, but that the Zimmer factors, if proved, provided a factual basis from which the necessary intent may be inferred (Nevett v. Sides, 1978). The Supreme Court plurality in Bolden flatly rejected this approach. The legislative history of Section 2 is sparse, but there is some evidence to indicate that the plurality's interpretation of Section 2 was incorrect. Although the language of Section 2 made no mention of such terms as purpose, effects, or results, Attorney General Katzenbach, testifying during the Senate hearings regarding the Voting Rights Act, said that Section 2 would ban "any kind of practice . . . if its purpose or effect was to deny or abridge the right to vote on account of race or color." The Mobile case was remanded to the district court for a rehearing in light of the Supreme Court's ruling that discriminatory purpose must be established in order for plaintiffs to prevail. An expert witness (Dr. Peyton McCrary) for the plaintiffs challenging Mobile's commission system was able to uncover evidence that city officials had adopted the at-large election system for consciously racial reasons, and thus the case was eventually won. However, see note 12. See criticisms of the Bolden decision offered by Hartman, 1982, and Parker, 1983. The revised bailout requirements are quite stringent, and congressional critics argued that few jurisdictions could satisfy them. For details, see O'Rourke, 1983. Justice Powell, joined in dissent by Justice Rehnquist, argued that the evidence in Rogers was no different in kind from that found insufficient by the Court to prove invidious intent in the Bolden case. He therefore felt that the decision in Rogers "cannot be reconciled persuasively" with Bolden (p. 629). However, an intent finding may still be of importance. See the discussion of Garza v. Los Angeles County Board of Supervisors (1991) in Chapter 5. The district court found statistical evidence that "in those races in which whites have opposed blacks, there has been evidence of polarization in the voting of both black and white voters" (p. 1535). Furthermore, in a county in which blacks made up almost 50 percent of the votingage population, no blacks had ever been elected to office. The Opelika majority was not the first to deem the use of a multivariate regression analysis to be more acceptable because it takes into account and tests the relative importance of a variety of factors in addition to race that could conceivably explain disparate voting patterns; Judge Higginbotham in a concurrence in Jones made the same argument. This was not to be the end of the Norfolk litigation. The Supreme Court vacated and remanded the Fourth Circuit decision in light of Thornburg v. Gingles, and the court of appeals in turn

142

21. 22.

23.

24.

25.

26.

27.

Notes to pages 46-57 remanded the case back to the district court. On remand, the district court again found for the defendants. However, on the second appeal, in August 1989, the Fourth Circuit reversed and held that the Norfolk at-large election system violated Section 2 of the Voting Rights Act (Collins v. City of Norfolk, 1989). The Supreme Court let the appeals court decision stand by denying certiorari in October 1990. The state did not appeal the district court's finding of dilution with respect to two of the challenged districts: the single-member senate district that was held to fragment black voters and a house multimember district in which a black had never won state legislative office. Actually, the Court distinguished among statistically significant polarization, substantively significant polarization, and legally significant polarization. Statistically significant bloc voting occurs when the correlation between the racial composition of the precincts and the votes cast is high enough (given the number of precincts) that it could not have happened by chance. Substantively significant polarization refers to estimates of black and white voting patterns that indicate that the outcome of the election would have been different "depending upon whether it had been held among only the white voters or only the black voters" (p. 54). Legally significant polarization is present only if white bloc voting usually is sufficient to defeat the candidate preferred by blacks—that is, except in the presence of special circumstances. Justice O'Connor agreed, to a limited extent, with the plurality on this issue. She stated that defendants should not be able to rebut a finding of racially polarized voting with evidence that the divergent voting patterns may be explained by causes other than race. On the other hand, she argued, possible alternative explanations for disparate voting patterns may be relevant to the overall vote dilution inquiry, such as when attempting to predict the possibility of a minority candidate's future success. Justice Brennan's position on this issue does not have the support of the majority of the Court; five other justices argued that the race of the candidate could very well be a relevant consideration. Justice White wrote a concurring opinion disagreeing with precisely this point. He objected to Justice Brennan's reasoning on the grounds that ignoring the race of the candidate could lead to the use of Section 2 as a rule protecting "interest groups" rather than a rule designed to protect against racial discrimination. Justice O'Connor, writing for herself, the chief justice, and Justices Powell and Rehnquist, expressed her agreement with Justice White on this subject. We return to this issue in Chapter 3. See also Grofman, 1991a, c. The other Section 2 cases considered by the Eighth Circuit Court of Appeals did not address the role of the three-pronged test relative to the totality-of-the-circumstances test (McGruder v. Phillips County Election Commission, 1988; Roberts v. Wamser, 1989). However, a threejudge district court panel in the Eighth Circuit held in Jejfers v. Clinton (1989) that satisfying the three Gingles factors was merely an "essential predicate" for proving a Section 2 violation. The court must then go on to "examine all of the other relevant factors and decide whether, on balance, a diminution of black political opportunity, in violation of Section 2, has been shown" (p. 209). Although the Fourth Circuit has heard a number of Section 2 challenges other than Collins (see, e.g., Irby v. Virginia State Board of Elections, 1989; Neal v. Harris, 1987; United States v. City of Cambridge, 1986; and White v. Daniel, 1990), the appellate court has not had to assess directly the applicability of either set of factors, usually because the issue contested was not a districting question, at least not by the time the Fourth Circuit heard the case (e.g., special elections, appointive positions, and the question of latches were at issue in the aforementioned cases). However, the district courts in the Fourth Circuit continue to consider evidence relevant to the Senate Report factors (McDaniels v. Mehfoud, 1989; Neal v. Coleburn, 1988), and a Fourth Circuit panel in McGhee v. Granville County (1988) specifically referred to the Gingles factors as "threshold proof requirements" (p. 117), suggesting a need for the court to go on to consider other Senate Report factors once the plaintiffs have demonstrated that they have satisfied the three Gingles requirements. The Tenth Circuit has essentially reached the same conclusion as the Seventh Circuit—that it is necessary to prove the three "preconditions" established in Gingles if the plaintiffs are to prevail in a Section 2 challenge. In the only Section 2 case considered by the Tenth Circuit, Sanchez v. Bond (1989), Hispanic voters from Sagauche County, Colorado, challenged the atlarge method of electing county commissioners. The district court entered judgment in favor

Notes to pages 58-64

28. 29.

30.

31. 32.

143

of the defendants on the grounds that plaintiffs failed to meet the second and third prongs of Gingles, and the Tenth Circuit affirmed. The appellate court did not discuss the issue further, leaving unclear the role of the totality of the circumstances. Judge Tjoflat simply ignored the fact that the Senate Report makes it clear that Congress intended to eliminate any requirement that racial purpose be shown before a Section 2 violation can be found. In the only other decision handed down by the Eleventh Circuit since Solomon, the appellate court affirmed the lower court's holding that the plaintiffs had failed to prove vote dilution because they were unable to satisfy the Gingles requirements (Concerned Citizens of Hardee County v. Hardee County Board of Commissioners, 1990). Hence it appears that the Eleventh Circuit has deemed proof of the three Gingles elements to be necessary, if not sufficient, to a showing of vote dilution. In Garza, a challenge brought by Hispanics against the single-member district plan used to elect the Los Angeles County Board of Supervisors, the appellate court held that when plaintiffs are able to demonstrate intentional discrimination, the "effects test" outlined in Gingles need not be met: "The showing of injury in cases involving discriminatory intent need not be as rigorous as in effects cases . . . [although] some showing of injury must be made" (p. 771). This decision was subsequently vacated (Armour v. State of Ohio, 1990), and the case was tried de novo before a three-judge panel. For further discussion, see the section "Influence Districts" in Chapter 5. One might also wish to follow the lead of the Ninth Circuit in relaxing the requirements for any of the three prongs when there is a showing of intentional discrimination.

Notes—Chapter 3 1. We limit our discussion in this chapter to the three Gingles prongs. As discussed in the last section of Chapter 2, there is ambiguity about the role of the totality-of-circumstances test as well. However, that is about whether the totality-of-circumstances test should be used at all, not about the components of the test. In contrast, confusion about the Gingles prongs is about the interpretation of the prongs themselves. 2. To prove the first prong, plaintiffs in Section 2 cases usually provide the court with an alternative plan(s) to the current election scheme that includes at least one predominantly minority district. This plaintiff-proposed plan is necessarily only illustrative, for if the court ultimately finds a violation of Section 2, the defendant jurisdiction will retain the right to develop a remedy. 3. Eleventh Circuit decisions rendered subsequent to Solomon have avoided directly addressing this issue (Concerned Citizens of Hardee County v. Hardee County Board of Commissioners, 1990; Meek v. Metropolitan Dade County, 1990). 4. The courts have thus put the emphasis on the potential to elect rather than on the actual voters, both of which are referred to a footnote intended to clarify the Supreme Court's interpretation (Thornburg v. Gingles, 1986, n.17). 5. Justice O'Connor recognized the latter point in her concurrence (joined by three other justices) in Gingles: If a minority group that is not large enough to constitute a voting majority in a singlemember district can show that white support would probably be forthcoming in some such district to an extent that would enable the election of candidates its members prefer, that minority group would appear to have demonstrated t h a t . . . it would be able to elect some candidates of its choice, (pp. 89-90, n. 1) Judge Kravitch may also have had such circumstances in mind. 6. Similarly, defendants in Gomez v. City of Watsonville (1988) contended, and the district court accepted the argument, that the geographic compactness prong of Gingles was not satisfied because 60 percent of the city's Hispanics would reside outside either of the two majorityHispanic districts. The Ninth Circuit Court of Appeals reversed the lower court's decision, holding that because "Hispanics in Watsonville are capable of constituting a majority in at least one district" (p. 1414), the district court's finding of insufficient geographic compactness was clearly erroneous.

144 7. 8. 9. 10.

11. 12.

13. 14. 15. 16. 17. 18. 19.

20. 21. 22.

Notes to pages 65-71 The action of the Justice Department is limited to only those jurisdictions covered by Section 5; hence it is still possible for a court to hold that the lack of geographic compactness can prevent a plaintiff group from succeeding on a Section 2 claim. In United Jewish Organizations of Williamsburgh v. Carey (1977), the Supreme Court held that a districting plan that deliberately attempts to create effective black voting-majority districts violates neither the Fourteenth nor the Fifteenth amendment. Hellar v. Cenarrusa, 1984; In Re Legislative Districting of General Assembly, 1972; Martin v. Soucie, 1982; Schrage v. State Board of Elections, 1981. Niemi, Grofman, Carlucci, and Hofeller (1990) summarize and evaluate about two dozen measures of compactness. Most vary between 0 and 1 and are based on one of three explicit "components" of compactness - geographic dispersion, perimeter length, and population dispersion. See, for example, the proposed Texas House District 30 and North Carolina Congressional districts 1 and 2 {Congressional Quarterly Weekly Report, August 17, 1991, pp. 2293-2395). If the groups do not claim that they should be considered as a single protected group, then each group's ability to satisfy the Gingles test will evaluated separately. See, for example, Meek v. Metropolitan Dade County, 1990, in which both black and Hispanic voters challenged the Dade County, Florida, at-large county commission system as violating Section 2, but neither group claimed it ought to be combined for purposes of meeting the Gingles test. In Meek, black voting was compared with nonblack voting (including Hispanics), and Hispanic voting was compared with non-Hispanic voting (including blacks). The courts have not been consistent in the standards used to measure the cohesiveness of blacks and Hispanics, however. Cohesiveness is perhaps implicit, however, in the Zimmer factor of racially polarized voting. The statistical proof required to demonstrate the political cohesiveness of the minority community is therefore the same as the statistical evidence employed to demonstrate racial bloc voting. To date, decisions rendered by appellate courts in the Sixth, Seventh, and Eighth circuits have not addressed the issue of political cohesiveness. Both of these assertions were later removed in a corrected opinion issued seven months later (Monroe v. City ofWoodville, 1989). This decision was later vacated and a rehearing granted when it was discovered that critical evidence, unrelated to the second prong of Gingles, may not have been considered by the court (Houston v. Haley, 1989). Black and Hispanic plaintiffs proposed to the Midland court a single-member district plan that contained two minority seats. However, one of the two minority seats - the seat containing the majority of the black population - had only a 46 percent black population. When Hispanics were considered, the combined minority population of the district was 72 percent. The Fifth Circuit, sitting en bane, later granted a motion for rehearing and vacated Judge Wisdom's opinion on grounds unrelated to the Voting Rights Act (LULAC v. Midland Independent School District, 1987). Legislative plans, even those drawn in response to a court order, must receive preclearance before they can be implemented if the jurisdiction is covered under Section 5. Butler never clearly defines what she means by "one"; the closest she comes to describing what she has in mind is the following: To be within the protection afforded by the act, these groups should be required to establish that they have interests that transcend politics - interests roughly equivalent to those shared by members of a single minority group. Perhaps the litmus test should be whether the two groups consider themselves "one" when each group could have benefited separately. Did blacks consider Mexican-Americans to be "black" in evaluating the need for affirmative action? In other apportionment situations in which Mexican-Americans had the opportunity to be a majority of a district, were they just as happy to be thirty percent of a district that was also thirty percent black? (1990, p. 688) Butler offers no clue as to how one might go about measuring this, nor does she offer any citations — in the legislative history of the Voting Rights Act, in previous court decisions, or in the social science literature - to support her contention that this is the proper test to use in determining whether minority groups should be considered jointly in a Section 2 suit.

Notes to pages 72-9 23.

24. 25. 26.

27. 28. 29.

30. 31.

32. 33.

34.

35.

36.

145

Of the nineteen election contests examined, there appeared to be no racial polarization in ten contests (Anglos, blacks, and Hispanics all supported the same candidates); in the other nine contests, blacks and Hispanics supported the same candidate in five contests and different candidates in the other four contests. The Fifth Circuit arrived at the conclusion the ' 'Anglos voted with one or the other minority groups more frequently than the two groups voted together," by ignoring the nonpolarized contests. If all of the contests were considered, blacks and Hispanics voted cohesively in fifteen of the nineteen contests. On the other hand, if the previous question - about what constitutes political cohesion - is answered by saying that cohesion involves more than voting together, it would follow that combining groups also requires more than demonstrating common voting patterns. Polarization addresses the voting patterns of minorities as well as of whites. But as we noted, it makes little sense to consider white voting behavior alone. It is only when whites and blacks vote in opposition that the whole matter of vote dilution arises. The Court pointed to such factors as the presence or absence of majority vote requirements, designated posts and other inhibitions on single-shot voting, the percentage of minority voters, the size of the district, and the number of open seats and the number of candidates (in multimember districts) as relevant to the degree of white bloc voting necessary to defeat minoritypreferred candidates (p. 56). These methods are explained in detail in Chapter 4. Despite the Supreme Court's specific endorsement of the statistical proof offered by the plaintiffs' expert in Gingles, some cases have evolved into a "battle of the experts" over the statistical methods used to generate estimates of voting patterns by race. See Chapter 4. Justice O'Connor, joined by Justices Burger, Powell, and Rehnquist, asserts: Insofar as statistical evidence of divergent racial voting patterns is admitted solely to establish that the minority is politically cohesive and to assess its prospects for electoral success, . . . defendants cannot rebut this showing by offering evidence that the divergent racial voting patterns may be explained in part by causes other than race, such as an underlying divergence in the interests of minorities and white voters, (p. 100) Presumably the fact that there may be other reasons for divergent voting patterns is relevant to the issue of whether racial bloc voting is "legally significant." This example also illustrates why it is usually the defendants who analyze all election contests (the vast majority of which tend to be white-versus-white contests) and the plaintiffs who focus only on those elections in which minority candidates ran for office: White candidates who receive a majority of the black votes fare better in elections than do black candidates who receive a majority of the black votes. Other courts, such as the Sixth Circuit District Court in Buchanan v. City of Jackson (1988), have attempted to hold a middle ground, arguing that courts may consider all elections but should give added weight to those elections that include minority candidates. Other Fifth Circuit decisions have been equally explicit in their rejection of white-versus-white elections as probative. In Campos v. City of Baytown (1988), for example, the Fifth Circuit rejected the defendants' argument that the court should have considered all elections, and it noted that "minority voter turnout increased dramatically when there was a candidate who was a member of the minority group" (p. 1245). The district court in Gretna was prompted to quote Henry Ford's famous line that "any customer can have a car painted any color he wants so long as it is black" to illustrate the constrained choice offered to the minority community in white-versus-white elections {Citizens for a Better Gretna v. City of Gretna, 1986, p. 1121). The fact that the Gingles Court examined only elections in which black candidates ran has not escaped the notice of other courts. For instance, in Gretna, the Fifth Circuit noted: "The various Gingles concurring and dissenting opinions do not consider evidence of elections in which only whites were candidates. Hence, neither do we" (p. 504). In another Fifth Circuit suit, Monroe v. City of Woodville (1989), the plaintiffs' expert examined elections that encompassed all of Wilkinson County, even though the challenged atlarge election scheme was located within a single city (Woodville). However, Woodville had only one election precinct, making impossible any statistical analysis of voting patterns by race. Although the Fifth Circuit indicated that there were problems with the plaintiffs' expert's analysis, the court reached conclusions based on the results of that analysis anyway.

146 37.

38.

39.

Notes to pages 79-87 Although the district court in Garza v. Los Angeles County Board of Supervisors concluded that it was necessary to examine exogenous elections, owing to the lack of viable Hispanic candidates competing for the office of county supervisor, the Ninth Circuit did not address this question (the appellate court held that because there was a finding of intentional discrimination, the Gingles factors did not apply). Although the plaintiffs' expert in Solomon v. Liberty County, Florida, conducted a three-tiered analysis that included a number of exogenous elections, both divisions of the Eleventh Circuit en bane court appear to have focused solely on county wide elections to reach their conclusions. However, neither opinion clearly rejected the use of exogenous elections in a Section 2 vote dilution suit. Moreover, as populations shift, some jurisdictions with homogeneous populations become much less so, raising questions about representation that are not currently relevant.

Notes—Chapter 4 1. This assumes that precinct data are used. We shall return to this point later. 2. Tests of statistical significance are statistical procedures designed to indicate whether observed results could have occurred by chance. After flipping a coin five times and getting four "heads," for example, one might wonder how likely this was to happen if the coin were in fact a fair one (.50 probability of heads on any single flip). Statistical theories can, in principle, answer such questions (although questions about just when and how to apply these theories occur in many applications). 3. Note that for "own race" voting to sum to over 180 percent, the percentage of blacks voting for black candidates and whites voting for white candidates would have to be around 90 percent. Indeed, even if one percentage were 100 percent, the other would have to exceed 80 percent for the sum to be over 180. Thus, 180 represents a very high level of polarization. 4. Essentially this approach was accepted by a district court in Collins v. City of Norfolk. That suit was originally filed in 1983; it was remanded for consideration consistent with Gingles, and the original judgment for the city was reversed on appeal in 1989. See Jacobs and O'Rourke, 1987; note that Dr. O'Rourke testified for the city of Norfolk. 5. This approach was accepted by a district court in McCord v. City of Fort Lauderdale (1985). It was also advocated by a U.S. Court of Appeals judge in an orbiter dicta in his concurring opinion in Jones v. City of Lubbock (1984). 6. Exit poll data are usually available, if at all, for presidential or statewide elections, and these rarely involve minority candidates. Hence, for elections most relevant to a finding of racial polarization - those that involve minority candidates, those of the same type as are at issue in the lawsuit, and those that are relatively recent - inferences about the voting behavior of individuals must normally be based on aggregate (e.g., precinct level) data. Moreover, there may be errors of reporting in the survey data, especially if voting is polarized along racial lines. 7. This method has come to be known as the double-equation method. It was independently developed by a number of scholars (including Laura Langbein and Alan Lichtman, and Bernard Grofman) and has come to be used by many others. It is a modification of the single-equation approach of Goodman (1956). Although referred to as a double-equation method, it consists of two bivariate equations and should not be confused with so-called simultaneous equation techniques. 8. In a handful of southern states, precinct-level voter registration data by race are available. If, as is almost universally the case, the only available racial data are population (or VAP) data from the U.S. census, they must be matched to electoral precincts, a task aided by the fact that most precincts include whole census blocks. 9. In a number of states, the Hispanic composition of the actual electorate may be determined (quite tediously, unless the information is computerized) by using polling booth sign-ins to identify Spanish-surnamed voters. In some states, computer tapes of the names of registered voters are available. We shall discuss later the problem of going from Spanish surname to Spanish origin. 10. Normally PBB + PBW < PWB + Pww < 1 (i.e., fewer black than white registrants vote), and P\v + PB < 1 (some registered voters do not vote).

Notes to pages 87-97

HI

That is, if we regress Pw(l) on 1 — x(l), the intercept will be an estimate of P^w a n d if we regress PB(i) on 1 - x(l), the intercept will be an estimate of PWB. Similarly, we can use the sum of the slope and the intercept of each of these equations to estimate PBW and PBB. Mathematically, because of symmetry, it must be the case that the four values we so obtain are identical to those that we obtained from our earlier pair of equations. 12. We could also directly estimate turnout by race (as a proportion of registered voters) by using the identity

11.

T = Twx + TB{\ - x)

13.

14.

15. 16.

17.

18. 19. 20.

21.

and regressing Tw(i) on x(l). The intercept of this regression will give us an estimate of TB, and the sum of the slope plus the intercept will give us Tw. If we estimated turnout in this fashion, we could then use the Tw and TB parameters so obtained to estimate the number of voters of each race in each precinct and then regress precinct-level minority turnout percentages on P'w(l) to obtain an alternative estimate of P'WB, and the like. Although this approach might seem to be more straightforward than the one described in the text - at least for obtaining an estimate of turnout by race - our experience in analyzing racial bloc voting in hundreds of different elections suggests that the method described in the text for estimating differential rates of turnout by race is usually the more reliable. The reason is that the regression equation, with turnout as the dependent variable and race as the independent variable, will have a very low correlation if average differences in turnout between the races are slight or if turnout across precincts has a high variance. For example, Dr. Jerome Sacks, a statistician generally skeptical of ecological regression, nonetheless believes that homogeneous precincts will yield reliable estimates of a group's voting behavior if over 40 percent of that group's voters are located in homogeneous precincts. See the discussion of Sacks's views in Judge Kenyon's trial decision in Garza. The 1982 Senate election in Mecklenburg was in a multimember district in which four candidates were to be selected. The analysis for such multimember districts parallels that for single-member district competition, with a few important exceptions. For a more detailed technical discussion of this issue see Grofman and Migalski, 1988. The .01 level was used in Gingles. The somewhat weaker .05 level is often used in the social sciences and might also be acceptable to the courts. It is also possible to use confidence intervals. For details, see Grofman and Migalski, 1988. "Exhibit after exhibit was introduced in the form of 'scattergrams,' in which the votes for black and white candidates are plotted for various precincts. To our untrained eye the cumulative effect of these exhibits is overwhelming, whatever the technical merits or demerits of the various statistical theories" (1989, p. 208). Indeed, in Campos v. City of Baytown (1988, pp. 1247-1248), the court accepted testimony from a political scientist that one black-majority precinct that was a severe outlier was controlled by a particular political boss who had agreed to throw his support to various white candidates. In that case, the trial judge minimized the importance of voting patterns in that outlying precinct for a judgment about whether the minority community was politically cohesive, and the Court of Appeals deferred to this judgment. As we discuss later, one may be able to generate estimated Spanish-origin registration percentages based on the Spanish-surname data, but they are normally not needed. Even if there are no homogeneous precincts, if there are more than, say, six precincts, it may still be possible to ascertain the presence or absence of a pattern of racial polarization from ecological regression alone. Of course, even in jurisdictions in which the victor in the primary of the preponderant party is all but assured victory in the general election, minority primary victors may face a more uncertain fate. In some of the multimember legislative elections in North Carolina analyzed in Gingles, the only Democratic candidates defeated in the general election were black. Comparing levels of white defections from the Democratic party when the Democratic candidate is black rather than white can be informative in regard to polarization, but even evidence that white Democrats support the black primary victor cannot negate a finding that voting is polarized in the primary itself. Florida, Georgia, Kentucky, Louisiana, North Carolina, and South Carolina.

148 22. 23.

24.

25.

26. 27.

28. 29.

30. 31.

32. 33. 34.

35. 36.

Notes to pages 98-107 For more detailed discussion of the technical issues raised by multimember districts without numbered places, see Grofman and Migalski, 1988. For example, in Garza, Dr. Jerome Sacks, one of the statisticians testifying for the county, asserted that a dropoff in support for a particular Hispanic candidate as precincts became more heavily Hispanic demonstrated the unreliability of ecological regression estimates because there was a nonlinear pattern in this candidate's support. However, as another of the county's experts admitted in deposition, if one looked at the combined votes for the two Hispanic candidates in this contest, the nonlinearity essentially vanished. One exception might be those situations in which it was legally necessary to disentangle the voting behavior of more than one distinct group covered under the Voting Rights Act, because of a legal claim that these groups should be combined for purposes of polarization analysis and line drawing owing to a commonality in their voting patterns. When two groups - each of which is covered under the Voting Rights Act - wish to be treated as a combined group for purposes of voting rights litigation, if the voting behavior of a combined group is sufficiently lopsided, we may be able to infer mathematically that each of the subgroups was casting at least a majority of its vote for particular candidates. Failure to heed this warning flag led one expert witness (Race Davies) to errors that led a federal district court to dismiss his testimony as noncredible. See Badillo v. City of Stockton, 1989. A related, legal point is whether the fact that blacks prefer black to Hispanic candidates and Hispanics prefer Hispanic to black candidates negates a finding of political cohesion for the combined group. In our opinion, as long as (1) each group usually supports viable candidates of its own race or ethnicity and (2) each group usually supports viable candidates of the other group in preference to nonminority candidates in situations in which no candidate of the group's own race/ethnicity is in the contest, then the combined group ought to be regarded as politically cohesive for purposes of voting-rights litigation. Thus, the fact that minority voters split their vote among several minority candidates is not, by itself, an indication of lack of minority cohesion as long as, on balance, minority voters usually prefer minority candidates to nonminority candidates. Of course, to show both points (1) and (2) requires a greater variety of election contests (in terms of the candidates running) than when only one group is involved. Sacks is prepared to admit that voting patterns in homogeneous (90+ percent) precincts that contain at least 35 to 40 percent of the electorate of a given group are a reliable indicator of how that group votes. The statisticians asserted that failure to satisfy, perfectly, certain assumptions rules out use of the methods. However, those assumptions are never perfectly satisfied except in the pages of statistical textbooks (see Freedman et al., 1991; Grofman, 1981b; Lichtman, 1981; and Lupia and McCue, 1990 for further debate on this point). One of us (Grofman) was an expert witness in Garza for the U.S. Department of Justice. In Latino Political Action Committee v. Boston (1986), the absence of any precincts that were more than 25 percent Hispanic made it impossible to estimate Hispanic voting behavior reliably, as one of us (Grofman) testified. Commingled black and Hispanic populations presented a severe problem for analysis in Badillo v. City of Stockton (1989). Extensive evidence about discrimination against Hispanics in California is reviewed in Castro v. State of California (1970). On discrimination against Native American voting rights, see, for example, Garcia and Hain, 1981, p. 171. Hard data are usually more persuasive than personal reports are, which can too easily be dismissed as idiosyncratic or biased. In testimony in Gingles, the number of representatives elected from a particular multimember district was compared with the average number of legislators per district in the North Carolina legislature, as well as with the average number of legislators per district in the state houses and senates in each state as of 1980. However, that testimony was not cited in the district court's opinion. The data are reproduced in Grofman, Migalski, and Noviello 1985, Table 2. Also presumably relevant to a claim that a particular district was "unusually large" would be data on the physical size of the district and the transportation difficulties in gaining access to its remotest parts, or on the number of different media that serve the district. However, because of inadequacies in record keeping or systematic biases in purging the rolls

Notes to pages 107-12

149

of "deadwood," such data may be suspect. In PUSH v. Allain (1989), Mississippi registration information was held to provide an inaccurate picture of relative rates of black and white registration in the state. 37. However, in PUSH v. Allain (1989) the accuracy of CPS data for Mississippi was challenged because of claimed differential misreporting of registration by whites and blacks. These may be factors idiosyncratic to Mississippi. Notes—Chapter 5 1. United Jewish Organizations of Williamsburgh v. Carey, 1977. 2. Seamon v. Upham, 1982. 3. Of course, hundreds of remedial plans have been approved by courts that involve singlemember districts (sometimes in conjunction with continuing at-large elections - so-called mixed plans). Reviewing this considerable body of data and court precedent is beyond our scope. 4. Of these four cases, only two address substance, Garza v. Los Angeles County Board of Supervisors (1990) and Washington v. Tensas Parish (1987), and the latter decision dealt only with the remedy phase. The Tensas School Board district was nearly evenly balanced in terms of its racial composition, with a miniscule black voting-age majority and a white registration majority. The circuit court approved a plan with three clearly black majority districts, three clearly white majority districts, and a seventh "swing" district with a 52 percent black votingage population. The two appellate decisions that do not reach substantive issues are Armour v. Ohio (1990) and White v. Daniel (1990). Armour was put on remand to a three-judge district court for a trial de novo; the district court found for the plaintiffs (1991), and the case is likely to be appealed again. In White v. Daniel the Fourth Circuit reversed a lower court finding of a Section 2 violation on the basis of laches. 5. Challenges to at-large systems are most likely to continue in states such as Texas and California, where suits will be brought primarily by Hispanic s. 6. Opinions of the first two authors are found in Grofman and Handley, 1992. See also Grofman and Davidson, 1992. 7. Gelfand and Allbritton point out that under certain circumstances "a plan might be able to include a majority-black district only by creating a district that is much less populous than the other, majority-white districts. Such a plan might be challenged on one-person, one-vote grounds" (1989, p. 120). 8. Even more to the point, perhaps, is the following question: Is it permissible for a jurisdiction to argue that it cannot create a majority-minority congressional district because to do so would require, say, a 5 percent deviation from exact equality? 9. A de minimis standard for congressional districting was supported by Justice Byron R. White - joined by Chief Justice Warren E. Berger and Justices Lewis F. Powell, Jr., and William H. Rehnquist - in his dissent in Karcher v. Daggett (1983). 10. Of course, if there are configurations that provide a realistic opportunity for minorities to elect a candidate of their choice at the beginning of a decade, it should not be permissible to draw districts that initially elect a nonminority in the hope that population gains later in the decade will give minorities an opportunity. 11. "Estimates based on past trends are generally not sufficient to override 'hard' decennial data" {McNeil v. Springfield Park District, 1988, p. 946). 12. It was generally conceded that Los Angeles County officials anticipated significant Hispanic population growth over the decade of the 1980s. Moreover, the county of Los Angeles had prepared intracensus population estimates that supported the existence of continuing Hispanic growth, especially in the Hispanic core area in which the proposed Hispanic majority district was to have been located. 13. Certainly it is not likely that any hypothetical, remote potential of minority success would be adequate to satisfy the first prong of the Gingles test. In Garza, the proof of electability accepted by the district court included a wide range of evidence, for example, evidence that a district with over 65 percent Hispanic population could have been drawn, evidence that this district would have had a Hispanic registration percentage in excess of 40 percent as early as 1981 and at the time of trial would have contained a Hispanic registration majority, and evi-

150

14. 15.

16.

17.

18.

19. 20.

21. 22. 23.

24.

25.

26.

Notes to pages 112-17 dence that districts in partisan contests within the county of Los Angeles elected Hispanic candidates with near certainty once they were above 40 percent Hispanic in registration. See Grofman, 1992, for further discussion of census issues. Permitting a total deviation of 18.75 percent would allow the creation of a district with only 17 percent of the overall population. (Each of the other districts would have to have 20.75 percent of the population; a negative deviation of 15 percent and a positive deviation of 3.75 percent yield the total deviation of 18.75). If the entire minority group were in the undersized district, it would still make up only 8/17 or 47.1 percent of the district population. In the first, McNeil v. City of Springfield (1987), there was a state ordinance that held that the minimum size for a districted body was to be ten members; therefore, the court held that the Gingles threshold could be met based on a ten-member body, even though the existing atlarge council had fewer members. In Gingles, Justice Brennan asserted: We have no occasion to consider whether the standards we apply to respondents' claim that multimember districts operate to dilute the vote of geographically cohesive minority groups that are large enough to constitute majorities in single-member districts and that are contained within the boundaries of the challenged multimember districts, are fully pertinent to other sorts of vote dilution claims, such as a claim alleging that the splitting of a large and geographically cohesive minority between two or more multimember or single-member districts resulted in the dilution of the minority vote. (1986, p. 47, n. 12) We find it hard to believe that even given intentional discrimination, a liability claim could be sustained without evidence of polarized voting. Given the need to demonstrate injury, an obvious choice for that showing would seem to be racial bloc voting that results in the usual defeat of minority candidates. If a nonretrogression test were held to be applicable, plaintiffs would still have to show racially polarized voting patterns so as to demonstrate that districting choices affected their opportunity to elect candidates of choice (see Beer v. United States, 1976). Certainly the fact that some minorities were being elected would not by itself ward off a liability finding, inasmuch as there might be other plans in which greater minority success was near certain. See also McDaniels v. Mehfoud, 1988, and Neal v. Coleburn, 1988. This would seem to be a natural interpretation of what Judge Arnold had in mind when he wrote: "if lines are drawn that limit the number of majority black single-member districts . . . " {Jeffers v. Clinton, 1989, p. 205). Conceivably, the number could also include districts in which minorities are able to elect candidates of choice because they constitute a majority in a party primary (though not in the overall population) and partisanship is sufficiently strong that a minority person who is the party's nominee will typically be elected. ' 'The claim we address in this opinion is one in which the plaintiffs alleged and attempted to prove that their ability to elect the representatives of their choice was impaired by the selection of a multimember electoral structure. We have no occasion to consider whether Section 2 permits, and if it does, what standards should pertain to, a claim brought by a minority group, that is not sufficiently large and compact to constitute a majority in a single-member district, alleging that the use of a multimember district impairs its ability to influence elections" (Thornburg v. Gingles, 1986, pp. 46-47, n. 12, emphasis in original). The appellate court in this case, Armour v. Ohio (1990), initially accepted an influence claim, but the Sixth Circuit subsequently vacated that decision, and the case was remanded to a threejudge district panel for a de novo trial in which the majority found evidence of intentional discrimination in the drawing of lines so as to fragment a black population concentration in Youngstown. In East Jefferson Coalition for Leadership and Development v. Parish of Jefferson (1988,1989), a district court sanctioned a 45.9 percent black "influence" district created by a defendant jurisdiction in a situation in which one of the plaintiffs' proposed alternatives included a district with a 53 percent black population majority (one district out of seven). The court asserted that the proposed black majority district was not compact and that, in the circumstances, creating a black majority district was creating a right to proportional representation for minorities. However, the plan required preclearance, and after preclearance was denied, a majority-black district was accepted as part of a remedial plan. For example, in McNeil v. Springfield Park District, the Seventh Circuit expressly rejected

Notes to pages 118-23

27.

28. 29.

30. 31.

32.

33. 34.

35. 36.

37.

38. 39.

40.

41.

151

influence claims because the court "would be flooded by the most marginal section 2 claims if plaintiffs had to show only that an electoral practice or procedure weakened their ability to influence elections" (1988, p. 947). It might be appropriate to create a district with as high a minority population as possible if such a district were one in which the minority would have a realistic possibility of electing a candidate of choice over the course of a decade. But what if even this is unattainable? What if the best one can do is create a 30 percent minority district? Is dividing the minority population 30 to 0 percent better or worse from the standpoint of influence than dividing it 20 to 10 or 15 to 15 percent? Of course, these questions are raised in the liability phase of a case as well, because a violation will not be considered to have occurred if no remedy is feasible. The algebraic model is described in Brace, Grofman, Handley, and Niemi, 1988, and is based on the same ideas as James Loewen's expert witness testimony in Kirksey v. Board of Supervisors of Hinds County (1977). That testimony was accepted by the Kirksey court and became the basis of much of the subsequent discussion by courts and expert witnesses of effective voting equality. For blacks, differences in voting-age population are likely to be of greatest significance in calculations of effective voting equality; for Hispanics (other than, of course, Puerto Ricans), differences in citizenship levels are almost certain to be the most important. In Garza v. Los Angeles County Board of Supervisors (1990), the district court recognized that because of the lower proportions of voting-age citizens among Hispanics than among nonHispanics, even a 65 percent Hispanic population might not be sufficient to provide an "effective Hispanic majority" district, although it might give rise to a district in which Hispanics had a realistic potential to elect a candidate of choice. Legend has it (at least as told to one of us by an attorney who worked on UJO for minority plaintiffs) that someone in the Justice Department took 50 percent and added 5 percent to compensate for the higher proportion of Hispanic noncitizens, 5 percent for lower Hispanic registration, and 5 percent for lower Hispanic turnout. See also Ketchum v. Byrne (1984) and Mississippi v. United States (1979). Outside the South, blacks may sometimes be elected to Congress or a state legislature in districts with substantial but not majority-black voting strength, because of black voters' dominance in the primary and Hispanic (or non-Hispanic white) crossover in the general election (see Handley and Grofman, 1992). These terms are adapted from the theoretical literature on comparative election systems (Taagepera and Shugart, 1989, pp. 36-37), in which threshold of representation has a different but related meaning. Most of the districts without a black population majority that elect a black House member are ones in which there is at least a substantial black plurality (over 40 percent) and a combined black-plus-Hispanic population over 55 percent (Grofman and Handley, 1989). In these instances it is likely that blacks make up a majority of the Democratic primary electorate. Thresholds for Hispanics are more problematic because of population changes since the 1980 census. However, it is now well understood that winning a runoff does not necessarily mean that the winning candidate is preferred to all other candidates. In fact, a candidate defeated in the first round might defeat all other candidates if there were a series of paired competitions. For a simple introduction to this and similar topics of what has come to be known as the social choice literature on voting, see Riker, 1982 (esp. chap. 4). The argument is that minority candidates win a plurality in a three-or-more-person race but then are defeated in a runoff in which whites coalesce behind the single white candidate. The solicitor general's brief for the Justice Department urged denial of certiorari on the grounds that the legal issues in the case were not yet ripe for review, although it expressed "significant doubts" (p. 9) as to whether, on the facts before it, the circuit court had properly applied the Section 2 standard. In the list of "Senate factors," "majority vote requirements" (runoffs) are cited as one of those "other voting practices or procedures" that enhance the possibility of discrimination against minorities. However, that does not mean that it violates Section 2 when considered in isolation. Further, if large numbers of candidates were to continue to run after the elimination of runoffs

152

Notes to pages 123-31

- leading to a greater number of black plurality winners - this might weaken the Democratic party and thus weaken the election prospects of blacks associated with that party. As Stanley (1987, quoted in Dillard v. Baldwin County Board of Education, 1988, p. 1377) puts it: Making southern whites choose between race and party when their party ties have considerably loosened might accelerate the decline of the Democrats - and promote Republican prospects to an extent that Republicans themselves have not yet managed. More generally, the nomination of splinter candidates [as plurality winners might be in cases of large fields of candidates] could make the Republican alternative more attractive in the eyes of many. Stanley goes on to quote U.S. Congressman Wyche Fowler (D. Ga.): "No question, the best thing that could happen to the Republicans is the abolition of the runoff primary. It would build the Republican party overnight." 42. The data presented in Whitfield do not allow us to be certain whether that case would meet this test, but it is likely that it would. We know that four black candidates who were plurality winners subsequently lost the general election and that no black had ever been elected to office in Phillips County (p. 1430), despite the fact that blacks made up nearly 50 percent of the county population. Thus, every black plurality winner subsequently lost; it is almost inconceivable that the same shutout was true for white plurality winners. 43. And as we pointed out in Chapter 1 (n. 29), the Supreme Court has noted that the Constitution does not mandate proportional representation. 44. Proponents would counter that these systems are no more "racial" in their effects than are plurality, at-large systems, and single-member systems in which lines are drawn in a raceconscious fashion. 45. We would therefore expect greater use of these remedies as negotiated settlements in voting rights cases. We also expect attempts to relitigate the question of whether the probable outcome of an election under one of these methods can be used instead of a single-member district plan as the baseline against which to judge vote dilution. 46. Limited voting is, in a sense, the opposite of anti-single-shot requirements. The latter prevents a voter from casting fewer votes than the number of positions to be filled, whereas the former requires it. 47. For references and more details on the use of limited voting, see Grofman, 1985. 48. At least in Cincinnati, "racial antagonism is widely regarded to have been the critical or decisive factor in the abandonment of PR" (Engstrom, 1990, p. 219) and in the failure to readopt STV in a 1988 referendum. 49. Whether the use of weighted voting is legally dead remains an open question. The factual circumstances in the New York City Board of Estimates case were quite unusual (Gelfand and Allbritton, 1989). Notes—Chapter 6 1. In addition, the effects of changes in election systems are felt most directly and immediately by officeholders and aspirants, a relatively small population. And it is easier for, say, a prospective firefighter who has met the physical requirements and scored very high on a civil service examination to complain that he, personally, has been denied a job because of reverse discrimination than it is for a candidate for political office to argue that he is more qualified than the others are and that the office "rightfully" belongs to him. 2. For a detailed critique of Thernstrom's account, see Karlan and McCrary, 1989; also Kousser, 1992. On whether or not such a broad construction of Section 5 was the original intent of the framers of the statute, see Handley, 1991, pp. 94-97. 3. Justice Felix Frankfurter's opinion in Gomillion treated the case as one of denial of the right to vote on account of race, but as Rush (1990, p. 694) notes, Frankfurter's opinion "does not explain how denial of the opportunity to vote in Tuskegee constituted denial of the franchise." And two years later, in writing the majority opinion in Baker v. Carr (1962), Justice William J. Brennan, Jr., cited Gomillion to the effect that "vote impairment" was justiciable under the Equal Protection Clause of the Fourteenth Amendment (pp. 229-231). 4. Parker (1990, p. 34) notes that in the late 1950s and early 1960s the leading newspaper in Mississippi discussed the possibility of changing the boundaries of the state's congressional districts so that the effects of votes by blacks would be minimized.

Notes to pages 131-7 5. 6. 7. 8. 9. 10.

11. 12.

153

We say "chiefly" because the role of the totality-of-circumstances test, even in multimember situations, and the appropriate test for dilution in single-member districts has not been fully clarified; see Chapters 3 and 5. See Introduction, n. 1. Only when the minority population is relatively large and highly ghettoized (the latter, unfortunately, all too common for blacks in the United States) will the number of districts in which minorities can elect candidates of choice be nearly proportional to minority voting strength. There is also no reason that minorities cannot have more than a proportionate number of representatives. In a color-blind society, minorities would sometimes have less and sometimes more than their "share" of elected representatives. The figure for Dinkins is reported in the New York Times, November 9, 1989, p. B l l . The figure for Wilder is from a CBS/NYT exit poll reported in Time, November 20, 1989, p. 60. In South Carolina, for example, a white liberal Democrat testified that without a "leavening" of roughly 30 percent black population, his urban district might be won by a Republican (deposition testimony of State Senator Ravenel in South Carolina v. United States, 1984). Almost invariably this argument has been made by incumbents who are white. It buttresses concerns that the outcome of a rainbow coalition that does not emphasize black representation may well be that "black folks will get all the rain and white folks will get all the bows" (Thomas N. Todd, cited by Guinier, 1989, p. 415). For example, many of the black and Hispanic members of the U.S. Congress began as state legislators. In virtually all such cases the legislative district from which their career was launched was a majority-minority district. See, for example, Browning, Marshall, and Tabb, 1984, 1990; Fraga, 1991; Meier and England, 1982; and Welch and Bledsoe, 1988.

References

Anderson, Arthur J., and William S. Dahlstrom. 1990. 'Technological Gerrymandering: How Computers Can Be Used in the Redistricting Process to Comply with Judicial Criteria." Urban Lawyer 22: 59-77. Avila, Joaquin. 1982. "Mobile Evidentiary Requirements." In The Right to Vote. New York: Rockefeller Foundation. Ball, Howard, Dale Krane, and Thomas P. Lauth. 1984. "The View from Georgia and Mississippi." In Chandler Davidson, ed., Minority Vote Dilution. Washington, DC: Howard University Press. Blacksher, James, and Lawrence Menefee. 1982. "From Reynolds v. Sims to City of Mobile v. Bolden." Hastings Law Journal 34: 1-64. Brace, Kimball, Bernard Grofman, Lisa Handley, and Richard G. Niemi. 1988. "Minority Voting Equality: The 65 Percent Rule in Theory and Practice." Law and Policy 10: 43-62. Branch, Taylor. 1988. Parting the Waters: America in the King Years, 1954-1963. New York: Simon & Schuster. Brischetto, Robert, David Richards, Bernard Grofman, and Chandler Davidson. 1990. "Minority Voting Rights in Texas, 1965-1990." Paper presented at the conference, The Impact of Minority Rights Voting in Texas, 1965-1990. Rice University, Houston, May 11-12. Brooks, Thomas R. 1974. Walls Come Tumbling Down: A History of the Civil Rights Movement, 1940-1970. Englewood Cliffs, NJ: Prentice-Hall. Browning, Rufus P., Dale R. Marshall, and David H. Tabb. 1984. Protest Is Not Enough: The Struggle of Blacks and Hispanics for Equality in Urban Politics. Berkeley and Los Angeles: University of California Press. Browning, Rufus P., Dale R. Marshall, and David H. Tabb, eds. 1990. "Racial Politics in American Cities: Blacks and Hispanics in the U.S." In Political Mobilization, Power and Prospects. Reading, MA: Addison-Wesley. Butler, Katharine I. 1982. "Challenges to Election Structures: Dilution and the Value of the Right to Vote." Louisiana Law Review 42: 851-950. Butler, Katharine I. 1985. "Reapportionment, the Courts, and the Voting Rights Act: A Resegregation of the Political Process?" University of Colorado Law Review 56: 1-97. Butler, Katharine I. 1990. "Minority Vote Dilution Suits and the Problem of Two Minority Groups: Can a 'Rainbow Coalition' Claim the Protection of the Voting Rights Act?" Pacific Law Review 21: 619-689. Canon, Bradley. 1978. "Factionalism in the South: A Test of Theory and a Revisitation of V.O. Key." American Journal of Political Science 22: 831-848. 154

References

155

Carpeneti, Walter L. 1972. "Legislative Apportionment: Multimember Districts and Fair Representation." University of Pennsylvania Law Review 120: 666-700. Davidson, Chandler. 1984. "Minority Vote Dilution: An Overview." In Chandler Davidson, ed., Minority Vote Dilution. Washington, DC: Howard University Press. Davidson, Chandler, and Luis Ricardo Fraga. 1984. "Nonpartisan Slating Groups in an At-Large Setting." In Chandler Davidson, ed., Minority Vote Dilution. Washington, DC: Howard University Press. Davidson, Chandler, and Bernard Grofman, 1992. "Minority Representation and Voting Rights in the South." Unpublished paper, University of California, Irvine. Derfner, Armand. 1973. "Racial Discrimination and the Right to Vote." Vanderbilt Law Review 26: 523-584. Derfner, Armand. 1984. "Vote Dilution and the Voting Rights Act Amendments of 1982." In Chandler Davidson, ed., Minority Vote Dilution. Washington, DC: Howard University Press. Eig, Larry M., and Michael V. Seitzinger. 1981. "State Constitutional and Statutory Provisions Concerning Congressional and State Legislative Redistricting." Congressional Reference Service, Report no. 81-143A. Elliot, Ward E. 1974. The Rise of Guardian Democracy: The Supreme Court's Role in Voting Rights Disputes, 1845-1969. Cambridge, MA: Harvard University Press. Engstrom, Richard L. 1980. "Racial Discrimination in the Electoral Process: The Voting Rights Act and the Vote Dilution Issue." In Robert Steed, Laurence Moreland, and Ted Baker, eds., Party Politics in the South. New York: Praeger. Engstrom, Richard L. 1990. "Cincinnati's 1988 Proportional Representation Initiative." Electoral Studies 9: 217-225. Engstrom, Richard L., and Charles J. Barrilleaux. 1991. "Native American and Cumulative Voting: The Sisseton-Wahpeton Sioux." Social Science Quarterly 72: 388-393. Engstrom, Richard L., Delbert Taebel, and Richard L. Cole. 1989. "Cumulative Voting as a Remedy for Minority Vote Dilution: The Case of Alamagordo, New Mexico.'' Journal of Law and Politics 5: 469^97. Fontana, Vincent. 1990. Municipal Liability: Law and Practice. New York: Wiley. Fraga, Luis. 1991. "Policy Consequences and the Change from At-Large Elections to Single Member Districts." Paper presented at the annual meeting of the Western Political Science Association, Seattle. Franklin, John Hope. 1961. Reconstruction: After the Civil War. Chicago: University of Chicago Press. Franklin, John Hope. 1967. From Slavery to Freedom: A History of Negro Americans. 3rd ed. New York: Vintage Books. Freedman, David A., Stephen P. Klein, Jerome Sacks, Charles A. Smyth, and Charles G. Everett. 1991. "Rejoinder." Evaluation Review 15: 800-816. Garrow, David J. 1978. Protest at Selma: Martin Luther King, Jr., and the Voting Rights Act of 1965. New Haven, CT: Yale University Press. Gelfand, M. David, and Terry E. Allbritton. 1989. "Conflict and Congruence in OnePerson, One-Vote and Racial Vote Dilution Litigation: Issues Resolved and Unresolved by Board of Estimate v. Morris." Journal of Law and Politics 6: 93-123. Gillette, William. 1969. The Right to Vote: Politics and Passage of the Fifteenth Amendment. Baltimore: Johns Hopkins University Press. Glazer, Amihai, Deborah Glazer, and Bernard Grofman. 1984. "Cumulative Voting in Corporate Elections: Introducing Strategy into the Equations." South Carolina Law Review 35: 295-309. Goodman, Leo. 1956. "Ecological Regression and the Behavior of Individuals." American Sociological Review 18: 663-664.

156

References

Graham, Hugh Davis. 1992. "Voting Rights and the American Regulatory State." In Bernard Grofman and Chandler Davidson, eds., Controversies in Minority Voting: A 25 Year Perspective on the Voting Rights Act of 1965. Washington, DC: Brookings Institution. Grofman, Bernard. 1981. "Fair and Equal Representation." Ethics 91:477-485. Grofman, Bernard. 1985. "Criteria for Districting: A Social Science Perspective." UCLA Law Review 33: 77-84. Grofman, Bernard. 1991a. "Multivariate Methods and the Analysis of Racially Polarized Voting: Pitfalls in the Use of Social Science by the Courts." Social Science Quarterly 72: 826-833. Grofman, Bernard. 1991b. "Statistics Without Substance: A Critique of Freedman et al. and Clark and Morrison." Evaluation Review 15: 746-769. Grofman, Bernard. 1991c. "Straw Men and Stray Bullets: A Reply to Bullock." Social Science Quarterly 72: 840-843. Grofman, Bernard. 1992. "Continuing and Emerging Controversies in Voting Rights: From One Person, One Vote to Partisan Gerrymandering." Stetson Law Review forthcoming. Grofman, Bernard, and Chandler Davidson, eds. 1992. Controversies in Minority Voting: A 25 Year Perspective on the Voting Rights Act of 1965. Washington, DC: Brookings Institution. Grofman, Bernard, and Lisa Handley. 1989. "Minority Population Proportion and Black and Hispanic Congressional Success in the 1970s and 1980s." American Politics Quarterly 17: 436-445. Grofman, Bernard, and Lisa Handley. 1991. "The Impact of the Voting Rights Act on Black Representation in Southern State Legislatures." Legislative Studies Quarterly 16: 111-127. Grofman, Bernard, and Lisa Handley. 1992. Identifying and Remedying Racial Gerrymandering. Journal of Law and Politics 8: 345^04. Grofman, Bernard, and Michael Migalski. 1988. "Estimating the Extent of Racially Polarized Voting in Multicandidate Contests." Sociological Methods & Research 16: 427-453. Grofman, Bernard, Michael Migalski, and Nicholas Noviello. 1985. "The 'Totality of Circumstances Test' in Section 2 of the 1982 Extension of the Voting Rights Act: A Social Science Perspective." Law and Policy 7: 209-223. Grofman, Bernard, and Nicholas Noviello. 1983. "An Analysis of Racial Bloc Voting. Paper presented at the annual meeting of the Southwestern Social Science Association, Houston. Grofman, Bernard, and Howard Scarrow. 1981a. "The Riddle of Apportionment: Equality of What?" National Civic Review 70: 242-254. Grofman, Bernard, and Howard Scarrow. 1981b. "Weighted Voting in New York." Legislative Studies Quarterly 6: 287-304. Guinier, Lani. 1989. "Keeping the Faith: Black Voters in the Post-Reagan Era." Harvard Civil Rights-Civil Liberties Law Review 24: 393-435. Guinier, Lani. 1991. "The Triumph of Tokenism: The Voting Rights Act and the Theory of Black Electoral Success." Michigan Law Review 90: 1077-1154. Guinier, Lani. 1992. "Voting Rights and Democratic Theory: Where Do We Go from Here?" In Bernard Grofman and Chandler Davidson, eds., Controversies in Minority Voting: A 25 Year Perspective on the Voting Rights Act of 1965. Washington, DC: Brookings Institution. Handley, Lisa. 1991. "The Quest for Minority Voting Rights." Ph.D. diss., George Washington University.

References

157

Handley, Lisa, and Bernard Grofman. 1992. 'Thresholds of Representation and Exclusion for Black Legislative Candidates Outside the South." Paper presented at the annual meeting of the Western Political Science Association, San Francisco. Hartman, Joan F. 1982. "Racial Vote Dilution and Separation of Powers: An Exploration of the Conflict Between the Judicial 'Intent' and the Legislative 'Results' Standards." George Washington Law Review 50: 689-753. Hawk, Barry, and J. J. Kirby. 1965. "Federal Protection of Negro Voting Rights." Virginia Law Review 51: 1051-1213. Henderson, Gordon, and Jeanette Wolfley. 1990. "Indian Success in Different Election Systems." Paper presented at the Conference on the Voting Rights Act, Rice University, Houston. Himelstein, Jerry. 1983. "Rhetorical Continuities in the Politics of Race: The Closed Society Revisited." Southern Speech Communication Journal 48: 153-166. Hunter, David. 1976. "The 1975 Voting Rights Act and Language Minorities." Catholic University Law Review 25: 250-270. Jacobs, Paul W., and Timothy G. O'Rourke. 1987. "Racial Polarization and Vote Dilution Cases under Section 2 of the Voting Rights Act: The Impact of Thornburg v. Ginglesr Journal of Law and Politics 3: 295-353. Jones, Mack H. 1985. "The Voting Rights Act as an Intervention Strategy for Social Change: Symbolism or Substance?" In Lorn S. Foster, ed., The Voting Rights Act: Consequences and Implications. New York: Praeger. Karlan, Pamela S. 1989. "Maps and Misreadings: The Role of Geographic Compactness in Racial Vote Dilution Litigation.'' Harvard Civil Rights-Civil Liberties Law Review 24: 173-248. Karlan, Pamela S., and Peyton McCrary. 1988. "Without Fear and Without Research: Abigail Thernstrom on the Voting Rights Act." Journal of Law and Politics 4: 751777. Key, V. O., Jr. 1949. Southern Politics in State and Nation. New York: Vintage Books. Kluger, Richard. 1975. Simple Justice: The History of Brown v. Board of Education and Black America's Struggle for Equality. New York: Knopf. Kousser, J. Morgan. 1974. The Shaping of Southern Politics. New Haven, CT: Yale University Press. Kousser, J. Morgan. 1984. "The Undermining of the First Reconstruction: Lessons for the Second." In Chandler Davidson, ed., Minority Vote Dilution. Washington, DC: Howard University Press. Kousser, J. Morgan. 1992. "The Voting Rights Act and the Two Reconstructions." In Bernard Grofman and Chandler Davidson, eds., Controversies in Minority Voting: A 25 Year Perspective on the Voting Rights Act of 1965. Washington, DC: Brookings Institution. Lawson, Steven R. 1976. Black Ballots: Voting Rights in the South, 1944-1969. New York: Columbia University Press. Lawson, Steven R. 1985. In Pursuit of Power: Southern Blacks and Electoral Politics 1965-1982. New York: Columbia University Press. Lewinson, Paul. 1932. Race, Class, and Party: A History of Negro Suffrage and White Politics in the South. Oxford: Oxford University Press. Lichtman, Alan J. 1991. "Passing the Test: Ecological Regression Analysis in the Los Angeles County Case and Beyond." Evaluation Review 15: 770-779. Lichtman, Alan J., and J. Gerald Hebert. 1986. "Civil Rights Law: High Court Decision of Voting Act Helps to Remove Minority Barriers." National Law Journal, November 10.

158

References

Lijphart, Arend. 1977. Democracy in Plural Societies. New Haven, CT: Yale University Press. Loewen, James W. 1990. "Racial Bloc Voting and Political Mobilization in South Carolina." Review of Black Political Economy 19: 503-513. Loewen, James W., and Bernard Grofman. 1989. "Recent Developments in Methods Used in Voting Rights Litigation." Urban Lawyer 21: 589-604. Lupia, Arthur, and Kenneth McCue. 1990. "Why the 1980s Measures of Racially Polarized Voting Are Inadequate for the 1990s." Law and Policy 12: 353-387. McClosky, Herbert, and John Zaller. 1984. The American Ethos. Cambridge, MA: Harvard University Press. McDonald, Laughlin. 1989. "The Quiet Revolution in Minority Voting Rights." Vanderbilt Law Review 42: 1249-1297. Meier, Kenneth J., and Robert E. England. 1982. "Black Representation and Educational Policy: Does It Make a Difference?" Paper presented at the annual meeting of the American Political Science Association, Denver. Morris, Aldon. 1984. The Origins of the Civil Rights Movement. New York: Free Press. Niemi, Richard G., Bernard Grofman, Carl Carlucci, and Thomas Hofeller. 1990. "Measuring Compactness and the Role of a Compactness Standard in a Test for Partisan and Racial Gerrymandering." Journal of Politics 52: 1155-1181. O'Rourke, Timothy. 1983. "Voting Rights Amendments of 1982: The New Bailout Provision and Virginia." Virginia Law Review 69: 765-804. O'Rourke, Timothy. 1992. "The Voting Rights Paradox and the 1982 Voting Rights Act Amendments." In Bernard Grofman and Chandler Davidson, eds., Controversies in Minority Voting: A 25 Year Perspective on the Voting Rights Act of 1965. Washington, DC: Brookings Institution. Parker, Frank R. 1983. "The 'Results' Test of Section 2 of the Voting Rights Act: Abandoning the Intent Standard." Virginia Law Review 69: 715-764. Parker, Frank R. 1984. "Racial Gerrymadering and Legislative Reapportionment." In Chandler Davidson, ed., Minority Vote Dilution. Washington, DC: Howard University Press. Parker, Frank R. 1990. Black Votes Count. Chapel Hill: University of North Carolina Press. Rae, Douglas W. 1971. The Political Consequences of Electoral Laws. 2nd ed. New Haven, CT: Yale University Press. Raines, Howell. 1977. My Soul Is Rested. New York: Putnam. Robinson, Theodore P., and Thomas R. Dye. 1978. "Reformism and Black Representation on City Councils." Social Science Quarterly 59: 133-141. Robinson, W. E. 1950. "Ecological Correlations and the Behavior of Individuals." American Sociological Review 15: 351-357. Rush, Mark E. 1990. "The Supreme Court's Approach to Gerrymandering and Representation." Journal of Law and Politics 6: 683-722. Schuck, Peter. 1983. "What Went Wrong with the Voting Rights Act?" Washington Monthly 15: 51-56. Schwartz, Bernard, ed. 1970. Statutory History of the United States: Civil Rights, Parts I and II. New York: Chelsea House. Stanley, Harold W. 1987. "Runoff Primaries and Black Political Influence." In Laurence W. Moreland, Robert P. Steed, and Tod A. Baker, eds., Blacks in Southern Politics. New York: Praeger. Stanley, Harold W., and Richard G. Niemi. 1992. Vital Statistics on American Politics. 3rd ed. Washington, DC: CQ Press. Still, Edward. 1984. "Alternatives to Single-Member Districts." In Chandler Davidson, ed., Minority Vote Dilution. Washington, DC: Howard University Press.

References

159

Still, Edward. 1989. "Limited Voting and Cumulative Voting as Remedies for Minority Vote Dilution in Judicial Elections." Paper presented at the annual meeting of the American Political Science Association, Atlanta. Still, Edward. 1990. "Cumulative and Limited Voting in Alabama." Paper presented at the Conference on Representation, Reapportionment, and Minority Empowerment, Pomona College, Claremont, CA, March 29. Strange, Rick. 1989. "Application of Voting Rights Act to Communities Containing Two or More Minority Groups-When Is the Whole Greater Than the Sum of the Parts?" Texas Tech Law Review 20: 95-154. Swain, Carol. 1989. "The Politics of Black Representation in U.S. Congressional Districts." Ph.D. diss., University of North Carolina. Taagepera, Rein, and Matthew Soberg Shugart. 1989. Seats and Votes: The Effects and Determinants of Electoral Systems. New Haven, CT: Yale University Press. Thernstrom, Abigail M. 1985. "Voting Rights Trap'." New Republic 193: 21-23. Thernstrom, Abigail M. 1987. Whose Votes Count? Cambridge, MA: Harvard University Press. Tufte, Edward R. 1973. "The Relationship Between Seats and Votes in Two-Party Systems." American Political Science Review 67: 540-547. United States Commission on Civil Rights. 1968. Political Participation. Washington, DC: U.S. Government Printing Office. United States Commission on Civil Rights. 1975. The Voting Rights Act: Ten Years After. Washington, DC: U.S. Government Printing Office. United States Commission on Civil Rights. 1981. The Voting Rights Act: Unfulfilled Promises. Washington, DC: U.S. Government Printing Office. United States Department of Commerce, Bureau of the Census. 1982. Voting and Registration in the Election of November 1980, Series P-20, no. 370. Washington, DC: U.S. Government Printing Office. United States Department of Commerce, Bureau of the Census. 1989. Voting and Registration in the Election of November 1988, Series P-20, no. 440. Washington, DC: U.S. Government Printing Office. United States Senate. 1982. Senate Report no. 97-^17. Weaver, Leon. 1986. "The Rise, Decline, and Resurrection of Proportional Representation in Local Governments in the United States." In Bernard Grofman and Arend Lijphart, eds., Electoral Laws and Their Political Consequences. New York: Agathon. Welch, Susan, and Timothy Bledsoe. 1988. Urban Reform and Its Consequences: A Study in Representation. Chicago: University of Chicago Press, 1988. Williams, Juan. 1987. Eyes on the Prize: America's Civil Rights Years, 1954-1965. New York: Penguin. This book is a companion volume to the six-hour PBS television series of the same title. Woodward, C. Vann. 1966. The Strange Career of Jim Crow. 2nd rev. ed. Oxford: Oxford University Press. Woodward, C. Vann. 1971. Origins of the New South, 1877-1913. Baton Rouge: Louisiana State University Press. Wright, Stephen G., and William H. Riker. 1989. "Plurality and Runoff Systems and Number of Candidates." Public Choice 60: 155-175. Cases cited Allen v. State Board of Education, 393 U.S. 544 (1969). Alonzo v. Jones, no. C-81-227 (S.D. Texas 1983). Armour v. State of Ohio, 895 F.2d 1078 (6th Cir. 1990), vacated and rehearing en bane granted 58 U.S.L.W. 2484 (1990), no. 88-CV1104Y (N.D. Ohio 1991).

160

References

Badillo v. City of Stockton, no. S-87-1726 EJG (E.D. Cal. 1989), appeal pending. Baker v. Carr, 369 U.S. 186 (1962). Beer v. United States, 425 U.S. 130 (1976). Bradford County NAACP v. City of Starke, 712 F.Supp. 1523 (M.D. Fla. 1989). Breedlove v. Suttles, 302 U.S. 277 (1937). Brewer v. Ham, 876 F.2d 448 (5th Cir. 1989). Brown v. Baskin, 78 F.Supp. 933 (1948). Buchanan v. City of Jackson, 683 F.Supp. 1515 (W.D. Tenn. 1988). Buckanaga v. Sisseton Independent School District, 804 F.2d 469 (8th Cir. 1986). Burns v. Richardson, 384 U.S. 73 (1966). Butts v. City of New York, 614 F.Supp. 1527 (S.D. N.Y. 1985), 779 F.2d 141 (2nd Cir. 1985). Campos v. City of Bay town, 840 F.2d 1240 (5th Cir. 1988); 849 F.2d 943 (5th Cir. 1988), cert, denied 492 U.S. 905 (1989). Carrollton Branch of the NAACP v. Stallings, 829 F.2d 1547 (11th Cir. 1987), cert. denied 485 U.S. 936 (1988). Castro v. State, 2 Cal. 3d 223; 466 P.2d 244. Chisom v. Roemer, 111 S.Ct. 2354 (1991). Citizens for a Better Gretna v. City of Gretna, 834 F.2d 496 (1987). City ofLockhart v. United States, 460 U.S. 125 (1983). City of Mobile v. Bolden, 446 U.S. 55 (1980). Collins v. City of Norfolk, 605 F.Supp. 377 (E.D. Va. 1984); off d 768 F.2d 572 (4th Cir. 1985); rev'd 816 F.2d 932 (4th Cir. 1987); 679 F.Supp. 557 (E.D. Va. 1988); 883 F.2d 1232 (4th Cir. 1989). Concerned Citizens of Hardee County v. Hardee County Board of Commissioners, 906 F.2d 524 (11th Cir. 1990). Dillard v. Baldwin County Board of Education, 686 F.Supp. 1459 (M.D. Ala. 1988). Dillard v. Crenshaw County, 831 F.2d 246 (11th Cir. 1987). Dunston v. Scott, 336 F. Supp. 206 (1972). East Carroll Parish School Board v. Marshall, 424 U.S. 636 (1976). East Jefferson Coalition for Leadership and Development v. Parish of Jefferson, 691 F.Supp. 991 (E.D. La. 1988), 703 F.Supp. 28 (E.D. La. 1989). Ex Parte Yarbrough, 110 U.S. 651 (1884). Fortson v. Dorsey, 379 U.S. 433 (1965). Garza v. Los Angeles County Board of Supervisors, Slip opinion no. CV 88-5143 KN, 918 F.2d 763 (9th Cir. 1990). Georgia v. United States, 411 U.S. 526 (1973). Giles v. Harris, 189 U.S. 475 (1903). Giles v. Teasley, 193 U.S. 146 (1904). Gingles v. Edmisten, 590 F.Supp. 345 (E.D. N.C. 1984), off d sub nom Thornburg v. Gingles, 478 U.S. 30 (1986). Gomez v. City of Watsonville, 863 F.2d 1407 (9th Cir. 1988), cert, denied 489 U.S. 1080 (1989). Gomillion v. Lightfoot, 364 U.S. 339 (1960). Grovey v. Townsend, 295 U.S. 45 (1935). Guinn v. United States, 238 U.S. 347 (1915). Harper v. Virginia State Board of Elections, 383 U.S. 663 (1966). Hellar v. Cenarrusa, 106 Idaho 586, 682 P.2d 539 (1984). Houston v. Haley, 859 F.2d 341 (5th Cir. 1988), 869 F.2d 807 per curium (5th Cir. 1989). Houston Lawyers Association v. Attorney General of Texas, 111 S.Ct. 2376 (1991).

References

161

In Re Legislative Districting of General Assembly, 193 N.W.2d 784 (1972). Irby v. Virginia State Board of Elections, 889 F.2d 1352 (4th Cir. 1989). Jeffers v. Clinton, 730 F.Supp. 196 (E.D. Ark. 1989). Jones v. City of Lubbock, 727 F.2d 364 (5th Cir. 1984). Karcher v. Daggett, 462 U.S. 725 (1983). Ketchum v. Byrne, 740 F.2d 1398 (7th Cir. 1984), cert, denied 471 U.S. 1135 (1985). Kirkpatrick v. Preisler, 394 U.S. 526 (1969). Kirksey v. Board of Supervisors of Hinds County, 554 F.2d 139 (1977). Lane v. W/ttwi, 307 U.S. 268 (1939). Lassiter v. Northampton County Board of Elections, 360 U.S. 45 (1959). Latino Political Action Committee v. City of Boston, 784 F.2d 409 (1st Cir. 1986). Lee County Branch of the NAACP v. City of Opelika, 748 F.2d 1473 (11th Cir. 1984). Lo Frisco v. Schaffer, 341 F.Supp. 743 (D. Conn. 1971), aff d, 409 U.S. 972 (1972). LULAC v. Midland Independent School District, 812 F.2d 1494 (5th Cir. 1987), vacated; 829 F.2d 546 (5th Cir., en bane, 1987). Mahan v. Howell, 330 F.Supp. 1138 (E.D. Va. 1971), aff d in part, 410 U.S. 315 (1973). Major v. Treen, 574 F.Supp. 325 (E.D. La. 1983). Martin v. Mabus, 700 F.Supp. 327 (S.D. Miss. 1988). Martin v. Soucie, 109 111. App.3d 731, 441 N.E.2d 131 (1982). McCord v. City of Fort Lauderdale, 787 F.2d 1528 (11th Cir. 1986). McDaniels v. Mehfoud, 702 F.Supp. 588 (E.D. Va. 1988); 708 F.Supp. 754 (E.D. Va. 1989). McGhee v. Granville County, 860 F.2d 110 (1988). McGruder v. Phillips County Election Commission, 850 F.2d 406 (8th Cir. 1988). McMillan v. Escambia County, 748 F.2d 1037 (1984). McNeil v. City of Springfield, 658 F.Supp. 1015 (CD. 111. 1987). McNeil v. Springfield Park District, 851 F.2d 937 (7th Cir. 1988). Meek v. Metropolitan Dade County, 908 F.2d 1540 (11th Cir. 1990). Mississippi v. United States, 490 F.Supp. 569 (D.D.C. 1979), aff d 451 U.S. 934 (1981). Monroe v. City of Woodville, 819 F.2d 507 (5th Cir. 1987); 881 F.2d 1327 (5th Cir. 1989); rehearing denied, 897 F.2d 763 (5th Cir. 1989). Morris v. Board of Estimates, 489 U.S. 688 (1989). Neal v. Coleburn, 689 F.Supp. 1426 (E.D. Va. 1988). Neal v. Harris, 837 F.2d 632 (4th Cir. 1987). Nevett v. Sides, 571 F.2d 209 (5th Cir. 1978), cert, denied 446 U.S. 951 (1980). Nixon v. Condon, 286 U.S. 73 (1932). Nixon v. Herndon, 273 U.S. 536 (1927). Oregon v. Mitchell, 400 U.S. 112 (1970). Overton v. City of Austin, 871 F.2d 529 (5th Cir. 1989). Perkins v. Matthews, 400 U.S. 379 (1971). PUSH v. Allain, 111 F.Supp. 1189 (N.D. Miss. 1989). Reynolds v. Sims, 311 U.S. 533 (1964). Roberts v. Wamser, 883 F.2d 617 (8th Cir. 1989). Rogers v. Lodge, 458 U.S. 613 (1982). Romero v. City of Pomona, 883 F.2d 1418 (9th Cir. 1989). Rybicki v. State Board of Elections, 574 F.Supp. 1082 (N.D. 111. 1982). Sanchez v. Bond, 875 F.2d 1488 (10th Cir. 1989). Schrage v. State Board of Elections, 88 I11.2d 87, 430 N.E.2d 483 (1981). Seamon v. Upham, 536 F. Supp. 931 (E.D. Tex. 1982); 456 U.S. 37 (1982). Skorepa v. City of Chula Vista, 723 F.Supp. 1384 (S.D. Cal. 1989). Smith v. Allwright, 321 U.S. 649 (1944).

162

References

Solomon v. Liberty County, 865 F.2d 1566 (11th Cir. 1989), vacated; 899 F.2d 1012 (11th Cir., en bane, 1990). South Carolina v. Katzenbach, 383 U.S. 301 (1966). South Carolina v. United States, 589 F.Supp. 757 (D.D.C. 1984). Stevenson v. West, 413 U.S. 902 (1973). Thornburg v. Gingles, 478 U.S. 30 (1986). United Jewish Organizations of Williamsburgh v. Carey, 430 U.S. 144 (1977). United States v. Alabama, 362 U.S. 602 (1960). United States v. City of Augusta (S.D. Ga., 1987), settled out of court. United States v. City of Cambridge, 799 F.2d 137 (4th Cir. 1987). United States v. Cruikshank, 92 U.S. 542 (1876). United States v. Dallas County Commission, 739 F.2d 1529 (1984). United States v. Marengo County Commission, 731 F.2d 1546 (11th Cir. 1984). United States v. Raines, 362 U.S. 17 (1960). United States v. Reese, 92 U.S. 214 (1876). Wallace v. House, 515 F.2d 619 (5th Cir. 1975). Washington v. Davis, 426 U.S. 229 (1976). Washington v. Tensas Parish, 819 F.2d (5th Cir. 1987). Westwego Citizens for a Better Government v. City of Westwego, 872 F.2d 1201 (5th Cir. 1989); 906 F.2d 1042 (5th Cir. 1990). Whitcomb v. Chavis, 403 U.S. 124 (1971). White v. Daniel, no. 88-0568 (E.D. Va. Aug. 31. 1989); 909 F.2d 99 (4th Cir. 1990). White v. Regester, 412 U.S. 755 (1973). Whitfield v. Democratic Party of State of Arkansas, 890 F.2d 1423 (8th Cir. 1989), cert. denied, 111 S.Ct. 1089 (1991). Williams v. Mississippi, 170 U.S. 213 (1898). WMCA, Inc. v. Lomenzo, 371 U.S. 633 (1964). Yick Wo v. Hopkins, 118 U.S. 356 (1886). Zimmer v. McKeithen, 485 F.2d 1297 (5th Cir. 1973), off d on other grounds sub nom East Carroll Parish School Board v. Marshall, 424 U.S. 636 (1976).

Index of cases

Allen v. State Board of Elections, 24-5, 30, 31, 130 Alonzo v. Jones, 106 Armour v. State of Ohio, 59, 117-18, 143n31, 149n4, 150n25 Baker v. Carr, 133, 152n3 Badillo v. City of Stockton, 115, 148n26, n31 Beer v. United States, 82, 83-4, 140n26,

East Carroll Parish School Board v. Marshall, 141n9 East Jefferson Coalition for Leadership and Development v. Parish of Jefferson, 65, 160n25 Elmore v. Rice, 139nlO Ex Parte Yarbrough, 138n3 Fortson v. Dorsey, 32, 130

150nl9 Bradford County NAACP v. City of Starke, 80 Breedlove v. Suttles, 10 Brewer v. Ham, 56, 63, 66, 68, 72 Brown v. Baskin, 139nl0 Buchanan v. City of Jackson, 75—6, 145n32 Buckanaga v. Sisseton Independent School District, 54, 55 Burns v. Richardson, 32 v. C/fy of New York, 123

Garza v. Los Angeles County Board of Supervisors, 58, 63, 96, 98, 100, 1 0 3 ^ , 106, 107, 112, 114, 121, 141nl7, 146n37, 147nl3, 148n23, n30, 149n4, 149-50nl3, 151n30 Georgia v. United States, 31 Giles v. Harris, 10 Giles v. Teas ley, 10 Gingles v. Edmisten, 47-8, 91, 107, 120 Gomez v. City of Watsonville, 54, 58, 67-8, 77, 94, 96, 143n6 Gomillion v. Lightfoot, 130, 140n7, 152n3 Grovey v. Townsend, 10, 139n8 Guinn v. United States, 10-11

v. C/fy ofBaytown, 56, 62, 64, 66, 6 8 9, 71, 72, 145n33, 147nl7 Carrollton Branch of the NAACP v. Stallings, 57, 68, 77-8, 79-80 Castro v. State, 148n32 Chison v. Roemer, 140n2 Citizens for a Better Gretna v. City of Gretna, 56, 75, 76, 77, 79, 145n34, n35 City of Mobile v. Bolden, 34-8, 39, 40, 41-2, 130, 140n29 Collins v. City of Norfolk, 45-6, 50, 52, 56, 68, 75, 77, 142n20, 146n4 Concerned Citizens of Hardee County v. Hardee County Board of Commissioners, 66, 68, 72, 143n29, n3

In Re Legislative Districting of General Assembly, 144n9 Irby v. Virginia State Board of Elections, 142n26

Dillard v. Baldwin County Board of Education, 64, 152n41 Dillard v. Crenshaw County, 57, 126, 128 Dunston v. Scott, 141n8

Jeffers v. Clinton, 65, 91, 115, 123, 129-30, 131, 142n25 Jones v. City of Lubbock, 42-3, 47, 50, 52, 146n5

Harper v. Virginia Board of Elections, 21 Hellar v. Cenarrusa, 144n9 Houston Lawyers Association v. Attorney General of Texas, 140n2 Houston v. Haley, 69, 144nl8

163

Index of cases

164 Karcher v. Daggett, 110, 111, 149n9 Ketchum v. Byrne, 115, 151n33 Kirkpatrick v. Preisler, 111 Kirksey v. Board of Supervisors of Hinds County, 120, 151n29 Lane v. Wilson, 11 Lassiter v. Northampton County Board of Elections, 10 Latino Political Action Committee v. Boston, 148n31 Lee County Branch of NAACP v. C7fy of Opelika, 45, 47, 52 Lo Frisco v. Schaffer, 125 LULAC v. Midland Independent School District, 56, 62, 66, 69, 70, 72, 144n20 McCord v. City of Fort Lauderdale, 46-7, 50, 52, 146n5 McDaniels v. Mehfoud, 62, 63, 113, 142n26, 150n21 McGhee v. Granville County, 125, 142n26 McGruder v. Phillips County Election Commission, 142n25 McMillan v. Escambia County, 44, 45, 47, 50, 52 McNeil v. City of Springfield, 150nl6 McNeil v. Springfield Park District, 54, 56-7,

59, 62, 113, 149nll, 150-ln26 Mahan v. Howell, 110 Martin v. Mabus, 120 Martin v. Soucie, 144n9 Mee£ v. Metropolitan Dade County, 58, 68,

77, 143n3, 144nl2 Mississippi v. United States, 151n33 Monroe v. City of Woodville, 54, 56, 69,

144nl7, 145n36 Morris v. Board of Estimates, 117, 127 Neal v. Coleburn, 63, 64-5, 142n26, 150n21 Neal v. Harris, 142n26 Nevett v. Sides, 34, 14In 10 Nixon v. Condon, 139n8 Nixon v. Herndon, 139n8 Oregon v. Mitchell, 19-20 Overton v. City of Austin, 56, 63, 66, 69, 72 Perkins v. Matthews, 30 PC/5// v. A/tom, 149n36, n37 Reynolds v. S/ms, 31, 32, 37 Roberts v. Wamser, 142n25

Rogers v. Lodge, 41-2 Romero v. City of Pomona, 58, 63, 66, 72, 75, 77 Sanchez Schrage Seamon Skorepa

v. Bond, 69, 78, 79, 142n27 v. State Board of Elections, 144n9 v. Upham, 149n2 v. City of Chula Vista, 63

SmzY/z v. Allwright,

11-12, 139n8, nlO

Solomon v. Liberty County, 57, 63, 68, 77-8, 146n38 StfMf/z Carolina v. Katzenbach, 14, 15, 18-19 South Carolina v. United States, 153nlO Stevenson v. West, 141n8 Thomas v. Todd, 153nlO Thornburg v. Gingles, 2, 28-9, 47-60, 73, 7 8 9, 80, 82, 8 3 ^ , 89, 90, 92, 99, 103, 109, 113, 115, 116, 117, 123, 124, 132, 1412n20, 143n4, n5, 148n34, 150n24; United Jewish Organizations of Williamsburgh v. Carey (UJO), 120, 144n8, 149nl United States v. Alabama, 14 United States v. City of Augusta, 106, 126 United States v. City of Cambridge, 142n26 United States v. Cruikshank, 6, 1 United States v. Dallas County Commission, 44, 50 United States v. Marengo County Commission, 4 3 ^ , 45, 47, 50, 52, 82 United States v. Raines, 13 United States v. Reese, 6-7 Wallace v. House, 82 Washington v. Davis, 14In 10 Washington v. Tensas Parish, 149n4 Westwego Citizens for a Better Government v. City of Westwego, 56, 63, 79, 123 Whitcomb v. Chavis, 32-3, 34, 35, 54 White v. Daniel, 116, 142n26, 149n4 White v. Regester, 33, 34, 35, 36, 37-8, 40, 41,42, 54, 57, 113 Whitfield v. Democratic Party of State of Arkansas, 123, 152n42 Williams v. Mississippi, 9-10 WMCA, Inc. v. Lomenzo, 127 Yick Wo v. Hopkins, 37 Zimmer v. McKeithen, 33-4, 40, 41, 42, 113

Index

affirmative action, 1, 105 African-Americans; see blacks Alabama, 12, 17, 22, 31, 122, 126, 138n5, 140n5; Constitution, 10; constitutional convention, 9; voting rights cases, 14, 15 Alamagordo, N. Mex., 126 Alaska, 21, 139nl3, nl6, 140n5 Alaskan natives, 20 Allbritton, Terry E., 149n7, 152n49 American Indians, see Native Americans analytical techniques, 82; see also statistical analysis Anderson, Arthur J., 120, 134 anecdotal testimony, 69 annexation, 8, 30, 82, 141n8 anti-single-shot laws, 24, 40, 105-6, 114, 140n22, 152n46 Arizona, 139nl3, nl6, 140n5 Arkansas, 65, 123 Arnold, Richard Sheppard, 115 Asian-Americans, 2, 20, 134 at-large elections, 1, 8, 24, 27, 29, 30, 3 3 ^ , 35, 77, 82, 109, 113, 117, 123, 125, 135, 139-40n22, 149n5; vote dilution in, 37, 41, 42-6, 55, 56, 57, 60, 61, 78, 80, 81 at-large representation, 135 attorney general, 13, 16, 17, 18, 24, 30 Augusta, Ga., 126 Austin, Tx., 72 Avila, Joaquin, 37 backlash, 83 Baldwin County, Ala., 64 Ball, Howard, 140n22 Barrilleaux, Charles J., 126 Bay town, Tex., 64 Bexar County (San Antonio), Tex., 33 bilingual assistance, 20, 39, 105 Birmingham, Ala., 14 bivariate regression analysis, 74, 75, 82, 100

Blackmun, Harry A., 35, 36, 51, 74 blacks, 64, 66, 100, 101, 131, 148n27; in Democratic party, 119, 152n41; disfranchisement of, in South, 8-10, 1415; effective voting equality for, 119-20; electoral success of, 134—7; enfranchisement of, 5, 12-15, 130, 136; history of official discrimination against, 105; impact of Voting Rights Act on participation of, 21-3; and influence districts, 118; legislation ensuring voting rights of, 16-23; and party primaries, 97; political cohesiveness of, 68, 69, 70-3, 77; political participation of, 46; and proportional representation, 133; and racial appeals, 107-8; racial polarization of, 132; registration of, 1, 21-2, 23t, 24t; and runoff elections, 122-3; thresholds of exclusion/representation of, 122, 151n36; turnout of, in presidential elections, 26t; voting rights of, 2, 4—15 see also combined groups Blacksher, James, 67 Bledsoe, Timothy, 153nl2 blind, disabled, or illiterate voters, assistance for, 39 bloc voting, 34, 43-5, 47, 49, 50, 54, 61, 96, 131; appropriate elections to study in analysis of, 75-80; commonsense checks on estimates of, 88-93; defining, 46, 834; evidentiary standard for, 51-3; key element of vote dilution claims and, 50; legally significant, 50-1, 145n30; measuring, 74-5; reliability of estimates of, 1 0 3 ^ ; statistical analysis of, 69; white, 68, 73-81 Boston, 125 Brace, Kimball, 119, 151n29 Branch, Taylor, 16 Brennan, William J., Jr., 32, 35, 36-7, 49, 50,

165

166 51-3, 60, 74, 76, 77, 78-9, 84, 110-11, 116, 124, 150nl7, 152n3 Brischetto, Robert, 135 Brooklyn, N.Y., 120 Brooks, Thomas R., 16 Browning, Rufus P., 153nl2 Bullock, Charles, 75-6 Burger, Warren E., 35, 41, 74, 145n29, 149n9 Burke County, Ga., 41 Butler, Katharine L, 1, 55, 70, 71, 129 California, 21, 66, 135, 139nl6, 149n5 Cambridge, Mass., 127 candidate of choice, 25-6, 27, 44, 51-2, 97-8; of blacks, 149n27; circumstances affecting election of, 63^-; determining, 75-6; equal opportunity to elect, 111, 118-22; of Hispanics, 148n27; minority potential to elect, 124—8; in minority precincts, 104; in multimember districts, 47; political cohesiveness in selecting, 67; and proportional representation, 53-4; usually defeated, 53, 61, 73-81, 115; and vote dilution, 48, 49-50, 51-2 Canon, Bradley, 123 Carlucci, Carl, 144nlO Carpeneti, Walter L., 67 Carter, Jimmy, 108 census, 31, 109 census data, 66, 86, 105, 107, 111-12, 146n8; matching, to precinct boundaries, 94; statistical adjustment of, 112 Chicago, 134, 137 Chicano political movement, 20 Cincinnati, 152n48 circumstantial evidence, 41, 42 citizen voting-age population (CVAP), 93, 97 citizenship, 97, 118, 119 Civil Rights Act of 1957, 12-13 Civil Rights Act of 1960, 12, 13, 14, 15 Civil Rights Act of 1964, 14-15 civil rights advocates, 14, 20; and vote dilution cases, 37, 38-9 civil rights attorneys, 1, 27 civil rights movement, 16, 23 Civil Service Commission, 18 civil suits, 13 Civil War, 4 Clayton, Claude R, 14 coalitions, 1, 71, 135, 137 code words, 108 Cole, Richard L., 126 Colfax, La., 7 color-blind society vs. institutionalization of racial/ethnic divisions, 131-2, 136 Colorado, 21

Index combined groups, 66, 70-3, 102-3, 148n24, n25, n27 Commission on Civil Rights, 13 commission form of government, 34-5 commonsense (interocular) test, 91 compactness of district, 61, 62-6, 116-17, 133; components of, 144nlO; meaning of, 64-6; measuring, 66; and vote dilution, 49-50, 60 Compromise of 1877, 6, 7 Conecuh County, Ala., 125-6 confederate governments, 5 congressional interest, see Voting Rights Act of 1965, congressional intent of Connecticut, 125, 139nl6 consociational democracies, 128 Constitution, 7, 109, 125, 131, 132; interpretation of, 38; no mandate for proportional representation in, 140n29, 152n43 constitutional amendments, 4—5; see also individual amendments constitutional conventions (states) and disfranchisement of blacks, 8-10 constitutional protection against vote dilution, 31-8 Corpus Christi, Tex., 106 corrective action, 1 3 3 ^ correlation coefficients, 83, 84, 91-2; overunderstating importance of, 98; reporting wrong, 98-9 "covered" states/jurisdictions, 17, 18, 19, 201, 22, 23t, 24t, 25t, 29, 30, 31 Cox, William Harold, 14 crimes, disfranchising, 9, 138n6 crossover voting, 64, 92, 121, 131, 136 cumulative voting, 124, 125, 126 Dade County, Fla., 58, 68, 144nl2 Dahlstrom, William S., 120, 134 Dallas, Tex., 109 Dallas County, Ala., 44 Dallas County, Tex., 33 data problems: availability, 82, 84, 93-6; Hispanic populations, 93-^, 107, 146n6; matching census data to precinct boundaries, 94; see also census data Davidson, Chandler, 80, 109, 140n22, 149n6 Davies, Race, 148n26 Dawson, Benjamin, 14 deannexation, 8, 140n4 Democratic party, 6, 9, 152n41; blacks in, 119, 152n41; primaries, 22, 97; and white primary, 9, 11, 22 demographic change, 111; see also population Derfner, Armand, 7, 13, 15, 37, 42

Index dilutive devices, see tests and devices Dinkins, David, 134 discrimination, 17, 23, 26, 130; disproportionality resulting from, 1 3 3 ^ ; effects of past, 36, 38, 40, 42, 46, 47; history of, 40, 43, 44, 45, 46, 47, 48, 70, 105; intentional, 60, 114, 143n32; lingering effects of, 106-7; recent, 105; Texas, 33 discriminatory intent, 39-40, 61, 133-4 discriminatory purpose, 34-8, 39 disfranchisement, 4, 33; courts and, 10-12; devices in {see also tests and devices), 7 8, 9; nineteenth-century, 4-8; in the South, 8-10, 12-15; special provisions of Voting Rights Act and, 16-17 district size, 40, 106, 112, 113-14, 140n22, 148n35; considerations of, 110-13 districting, 1-2, 25, 29, 31, 120, 129, 131; census data in, 112; compactness in, 66; frequency of, 111-12; "neutral," 27; requirement of nondiscriminatory effects, 110-11; see also multimember districts districting criteria, deviation from standard, 114 districts: number of, 106, 112-13; shape of, 64-5; see also gerrymandering double-ballot elections, 122 double-equation method, see ecological regression due process clause, see Fourteenth Amendment, due process clause East Carroll Parish, La., 33-4, 141n9 ecological fallacy, 89 ecological regression, 82, 83, 84, 85-8, 94, 95, 96-7, 98, 103, 104, 107, 120; homogeneous precinct analysis compared with, 88-90; measures of goodness of fit, 91-2; misinterpreting estimates of, 101-2; single-equation, 92, 99, 100-1 effective voting equality, 118-20, 65 percent rule of, 120 effective voting majority, 63-4, 121 effects test, 130, 143n30 Eig, Larry M., 66 "Eight Box Law," 7-8 Eighth Circuit Court of Appeals, 55, 65, 144nl6 Eisele, Garnett Thomas, 129-30, 131 Eisenhower, Dwight D., 13 Eisenhower administration, 12 election campaigns, racial appeals in, 48, 83, 107-8 election data, checks against known, 92

167 election laws, 1; and vote dilution, 27; see also state laws election officials, 5, 9, 105 election practices: and likelihood of vote dilution, 105-6; see also tests and devices elections: examined in study of racial bloc voting analysis, 75-80, 93; see also exogenous elections electoral devices, see tests and devices electoral successes, 105; proportional representation in, 53, 54, 56; "usual" lack of, 53, 61, 73-81, 131, 132 Eleventh Circuit Court of Appeals, 4 3 ^ , 45, 46-7, 50, 57-8, 63, 64, 66, 121; and candidate of choice, 77-8; and exogenous elections, 79-80; and political cohesiveness, 68; and racially polarized voting, 82 Elliot, Ward E., 11 Enforcement Act (1870), 5, 6-7, 9; Section 6 of, 138n3 enfranchisement of blacks, 5, 12-15, 130, 136 England, Robert E., 153nl2 Engstrom, Richard L., 25, 46, 126 equal population requirement, 25, 31, 109—11, 131 Equal Protection Clause, see Fourteenth Amendment, Equal Protection Clause of Escambia County, Fla., 45 ethnicity and voting rights policy, 131-2 evidentiary standards; see standard of proof (evidentiary standard) exclusion thresholds, see threshold(s) exit poll data, 98, 146n6 expert witnesses, 3, 46, 50, 75-6, 95, 98, 100, 102, 105, 120, 121; "battle" of, 145n28; and reliability of estimates of racial bloc voting, 103-4; statistical analysis by, 74, 75, 77, 83, 84-5, 91 "eyeball" test, 66 factionalism, 131 federal election observers, 5, 16, 18, 21, 22 federal government and disfranchisement of blacks in the South, 9-10 Federalist Papers, 131 Fifteenth Amendment, 8, 10, 11, 13, 21, 140n4, 144n8; enforcement of, 6, 7, 15, 16, 18; provisions of, 5; rekindled in courts, 10-12; Section 2, 5; violations of, 34, 35-6; vote dilution and, 29, 42 Fifth Circuit Court of Appeals, 11, 3 3 ^ , 35, 36, 41, 44, 45, 47, 50, 56, 62-3, 64, 65, 66; and candidate of choice, 76-7; and exogenous elections, 79; and political

168 cohesiveness, 68-9, 70-1, 72; and vote dilution, 42-3 Florida, 21, 66, 122, 147n21 Fontana, Vincent, 55 Force Act (1871), 5, 6-7, 9 Ford, Henry, 145n34 Ford, Gerald R., 20 Fort Lauderdale, Fla., 46-7 Fort Lauderdale case, 84 Fourteenth Amendment, 7, 140n4, 144n8; due process clause of, 10; enforcement of, 15, 16; Equal Protection Clause of, 29, 31, 32, 33, 35, 36, 41, 139n9, 141nlO, 152n3; guarantee of equal protection, 21; provisions of, 4-5; Section 2, 5; Section 5, 5; violations of, 34; vote dilution and, 42, 109 Fourth Circuit Court of Appeals, 45-6, 47, 56, 62, 64-5; and candidate of choice, 77; minority districts, 116; and political cohesiveness, 68 Fowler, Wyche, 152n41 Fraga, Luis, 127, 153nl2 Frankfurter, Felix, 11, 152n3 Franklin, John, 5 fraud, 6, 8 Freedman, David A., 103, 148n29 fusion principle, 8 Garrow, David J., 12, 13, 14, 16, 139nll Gelfand, M. David, 149n7, 152n49 geographic areas: racially homogenous, 82-3; see also compactness of district Georgia, 17, 59-60, 138n5, 140n5, 147n21; constitutional convention, 9; multimember districts, 32 gerrymandering, 1, 65, 121; see also racial gerrymandering Gillette, William, 138n2 Gingles factors/test, 2, 77, 82, 110, 131, 1323, 142n26, 149nl3; application of, 61-81; applied to single-member districts, 81, 115-17; and gerrymandering, 113, 114, 115; necessary/sufficient, 55, 56-8, 59, 60; and number of districts, 112-13; political cohesiveness of, 67-9; sufficiently large/geographically compact, 61-6; and totality-of-circumstances test, 54-60 Glazer, Amihai, 126 Glazer, Deborah, 126 "good character" test, 9, 10 Goodman, Leo, 146n7 goodness of fit, measures of, 91-2 government role in encouraging participation, 23

Index Graham, Hugh Davis, 1, 129 grandfather clause, 9, 10, 11, 138n4 Grofman, Bernard, 29, 50, 74, 79, 80, 84, 89, 91, 95, 99, 101, 109, 117, 126, 127, 132, 133, 134, 135, 140n22, 142n24, 144nlO, 146n7, 147nl4, 148n22, n29, n30, n31, n34, 149n6, 150nl4, 151n29, n33, n36 group representation, 125, 127 groups, 67; average behavior of, 82; and violation of voting rights, 131-2 Guinier, Lani, 1, 127, 136, 137, 140n27, 153nl0 Hancock, Paul, 120 Handley, Lisa, 29, 39, 41, 132, 133, 134, 135, 149n6, 151n29, n34, n36, 152n2 Hardee County, Fla., 68 Hare system, 127 Hartman, Joan F., 141nl4 Hawaii, 32, 139nl3 Hawk, Barry, 13, 15, 18 Hayes, Rutherford B., 6 Hebert, J. Gerald, 67, 120 Henderson, Gordon, 139n21 Higginbotham, Patrick E., 70-1 Himelstein, Jerry, 107, 108 Hispanic populations data problems of, 9 3 ^ , 107, 146n6 Hispanics, 2, 33, 63, 64, 66, 70, 78, 100, 101, 106, 114, 126, 131, 139n21, 148n27, 149n5, 149-50nl3; bloc voting and, 1034; effective voting equality for, 119; electoral success of, 134-5; history of official discrimination against, 105; majority districts, 121; office holding of, 1; political cohesiveness of, 67-8, 70-3, 77; and proportional representation, 133; thresholds for, 151n36; and Voting Rights Act, 20; see also combined groups Hofeller, Thomas, 144nlO homogeneous precinct analysis (extreme case analysis), 74, 84, 85, 91, 94, 95, 100, 103, 107; compared with ecological regression, 88-90; independent variable of, 93 homogeneous precincts: absence of, 95-6, 102, 104; limited number of, 95 housing segregation, 103, 105, 131, 132 Humphrey, Hubert, 108 Hunter, David, 20 Idaho, 139nl3, nl6 Illinois, 126 illiteracy, 8, 9 incumbency, 118, 121 Indianapolis, Ind., 125

Index individual rights, violation of, 131 influence districts, 3, 64, 117-18 intent test/standard, 41-2, 14In 10; see also discriminatory intent interpretation: expansive, of Section 2, 129; of Voting Rights Act, 129-37 intimidation, 5-6, 8 Issacharoff, Samuel, 67 Jackson, Jesse, 79 Jackson (Tenn.) City Commission, 75-6 Jacobs, Paul W., 146n4 Jefferson Parish (La.) Council, 65 Johnson, Lyndon B., 18 Jones, Edith, 56 Jones, Mack H., 23 Justice Department, see U.S. Department of Justice Karlan, Pamela A., 1, 59, 127, 152n2 Katzenbach, Nicholas, 18, 141nll Kentucky, 6, 147n21 Kenyon, David V., 104, 147nl3 Key, V. O.^Jr., 8, 11, 138n5, 139nl0 Killeen (Tex.) Independent School District, 56, 72 Kirby, J.J., 13, 15, 18 Kluger, Richard, 16 Kousser, J. Morgan, 5, 8, 24, 105, 130, 138n4, 152n2 Krane, Dale, 140n22 Kravitch, Phyllis A., 57-8, 63, 77-8, 143n5 Langbein, Laura, 146n7 language minorities, 20, 21; Voting Rights Act extended to, 16, 25 Lauth, Thomas P., 140n22 Lawson, Steven R., 7, 12, 16, 20, 21 lay testimony, 69, 105 legislative body size, 24 legislative intent: and right to vote versus right to representation, 129—30; see also Voting Rights Act of 1965, congressional intent of legislative reapportionment, 31 Lewinson, Paul, 9, 138-9n7 liability, 132 liability phase, 116 Liberty County, Fla., 57, 68 Lichtman, Alan J., 67, 102, 146n7, 148n29 Lijphart, Arend, 128 limited voting, 124, 125-6 literacy tests, 8, 9, 11, 12, 15, 16, 22, 138n4, n5; declared unconstitutional, 10; elimination of, 19-20, 21 litigation, 1, 4, 14, 27, 132; civil rights, 14,

169 105; districting, 25; minority vote dilution, 34, 37; new, 109-10, 117, 128; population equality, 111; proportionality, 134; redistricting, 1-2; runoffs, 123; Section 2, 60; social science testimony in, 2; as strategy for enforcement, 12-13, 15; Voting Rights Act, 3, 29, 137 local elections, abolishing, 8 local political reality, appraisal of, 121 Loewen, James W., 76, 99, 101, 120, 132, 151n29 Los Angeles County, 96, 107, 147nl2, 150nl3 Los Angeles County Board of Supervisors, 143n30 Louisiana, 12, 17, 22, 138n5, 140n5, 147n21; black registration in, 138-9n7; blacks holding office in, 5; constitutional convention, 9; disfranchisement of blacks, 6; voting rights cases, 14, 15 lower courts, 36; and amended Section 2, 427; clarifications/complications of vote dilution standard in, 61-81; racially polarized voting in, 73; use of multivariate analysis in, 75; vote dilution cases, 37 Lubbock, Tex., 42-3 Luebke, Paul, 107, 108 Lupia, Arthur, 148n29 McClosky, Herbert, 137 McCrary, Peyton, 141nl2, 152n2 McCue, Kenneth, 148n29 McDonald, Laughlin, 55, 59, 140nl Madisonian tradition, 131 Maine, 138nl, 139nl6 majority, effective, 63-4; meaning of, 62-4; question of what constitutes, 60 majority minority districts, 66, 111, 116, 117, 123, 134, 136-7, 149n8 majority vote requirement, 33, 40, 48, 51, 59, 105-6, 151n40 Marengo County, Ala., 43^4 Marion County (Indianapolis), Ind., 32-3 Marshall, Dale R., 153nl2 Marshall, Thurgood, 35, 36-8, 51, 74 Massachusetts, 138nl, 139nl6 maximum representation, guaranteeing, 116-17 Mecklenburg County, N.C., 90 Meier, Kenneth J., 153nl2 Menefee, Lawrence, 67 Mexican-Americans, see Hispanics Midland County (Tex.) School Board, 70-1 Migalski, Michael, 89, 147nl4, 148n22, n34 Military Reconstruction Act of 1867, 4 minority-controlled districts, goal of redistricting, 65-6

170 minority districts, 2, 112, 116, 133, 135, 144n8, 151n27 minority groups/minorities, 23, 27; combining for political cohesiveness, 66, 70-3, 102— 3, 148n24, n25, n27; elected to office, 105; fragmenting/packing, 114, 118; ghettoized, 153n7; multiple, 82-3; politically cohesive, 61, 66, 67-73; proportional success of, in a single election, 51; responsiveness to interests of, 35, 40, 42, 43, 44; sufficiently large/ geographically compact, 59-60, 61, 62-3; vote dilution of, 23-8; and Voting Rights Act, 20, 21 minority populations, 121-2; estimates of growth of, 111 minority potential to elect: standards other than single-member district, 124—8 minority-preferred candidate, see candidate of choice minority representation, 1-2, 110, 132-3; future of, 3; see also representation minority voting rights, see voting rights Mississippi, 12, 17, 22, 30, 120, 122, 138n5, 140n5; black voter registration rates, 1; blacks holding office in, 5, 10; constitutional convention, 8-9; registration data, 149n36, n37; suffrage provisions, 9 10, 24-5, 139n21; voting rights suits, 14, 15 Mitchell, Clarence, 20 mixed election scheme(s), 126 Mobile, Ala., 35 Mobile City Commission, 41 Morris, Alson, 16 multicandidate elections, 99-100 multicollinearity, 100 multimember districts, 27, 113, 147nl4; challenges to, 82, 109, 123; constitutionality of, 3 1 ^ ; litigation regarding, 128; without numbered places, 97-8; practices of, increasing likelihood of vote dilution, 105-6; vote dilution in, 29, 34, 47, 49-51, 60, 80, 81 multivariate statistical analysis, 45-6, 47, 83, 100; inappropriate use of, 100; in vote dilution suits, 74—5 municipal boundary alterations, 30, 140n4 National Association for the Advancement of Colored People (NAACP), 20 Native Americans, 2, 20, 55, 126, 139n21; history of official discrimination against, 105 New Deal, 12

Index New Hampshire, 138nl, 139nl6 New Mexico, 135 New Orleans City Council, 140n26 New York (state), 109, 127, 138nl New York City, 19, 123, 125, 127, 134 New York City Board of Estimate, 152n49 Niemi, Richard G., 138nl, 144nlO, 151n29 1990s: cases likely in, 80-1; issues of, 109-28 Nineteenth Amendment, 10, 136 Ninth Circuit Court of Appeals, 58, 63, 66, 143n6, n32; and candidate of choice, 77; and political cohesiveness, 67-8, 72 Nixon administration, 19 noncitizens, 118; Hispanic, 93-4 nondilution, 4 nondilution principle, 110 nonretrogression, 26, 31, 82, 115-17, 140n4 Norfolk case, 84 Norfolk City Council, 46 North (the), blacks voting in, 5, 12 North Carolina, 17, 21, 51-3, 107, 122, 138n5, 141n8, 147n20, n21; Congressional districts 1 and 2, 144nll; constitutional convention, 9; redistricting, 47-9 Nottoway County (Va.) Board of Supervisors, 64-5 Noviello, Nicholas, 89, 95, 148n34 numbered-place laws, 51, 106, 141n8 O'Connor, Sandra Day, 41, 5 3 ^ , 74, 76, 142n23, n24, 143n5, 145n29 office holding: by blacks, 5, 8, 12; by minorities, 1 O'Hare, William, 96, 97 Oklahoma, 10-11 Opelika, Ala., 45 O'Rourke, Timothy, 1, 46, 129, 141nl5, 146n4 outliers, 91-2 "own race" voting, 83 Oxford, Miss., 69 packing (minority), 114, 118 Parker, Frank R., 1, 34, 36, 37, 122, 139n21, 140n7, n22, 141nl4, 152n4 participation: barriers to, of minorities, 1, 22, 23; of blacks, 12, 20, 46, 130; impact of Voting Rights Act on black, 21-3; minority, 136-7; see also turnout paternalism, 135 Pennsylvania, 125 Petersburg, Va., 30 Philadelphia, Miss., 14 Philadelphia, Pa., 125 Phillips County, 152n42

Index "pivotal" voting power, 117 political campaigns, 40; see also racial campaign appeals political cohesion/cohesiveness, 61, 66, 67—73, 77, 98, 104; combining minority groups to achieve, 66, 70-3, 102-3; implicit in racially polarized voting, 144nl4; measuring, 67-9; misspecifying test for, 102; in multicandidate elections, 99-100; in single-member districts, 115; and vote dilution, 49, 50; voting patterns in, 74 political elites, 137 political equality, 131 political process, equal access to, 33 politics: group-based theory of, 131; raceconscious, 43-4; racially based, 125; Voting Rights Act and future of, 134-7 politics of second best, 3; Voting Rights Act and, 129-37 poll tax, 9, 10, 22, 33, 138n4, n5, 139nl5; in Voting Rights Act, 18 population, 106; change in, 3, 149n39; malapportionment, 37 population equality, allowable deviations from, 110-11 population majority(ies) required to elect candidate of choice, 118-22 population projections, 107, 111-12 Powell, Lewis R, Jr., 35, 53, 74, 141nl6, 142n24, 145n29, 149n9 power-sharing devices, 124, 127-8 preclearance, 16, 17-18, 19, 24-5, 30, 65, 71, 140n6, 144n21; denial of, 105; see also Voting Rights Act of 1965, Section 5 presidential elections, 146n6; black turnout in, 22, 26t; voter turnout in (1964, 1968), 25t primary elections, 97, 121; effective voting equality in, 119; runoff, 122, 123; see also white primary system property qualifications, 9 proportional minority success in a single election, 51 proportional representation, 1, 26, 51, 116, 124, 125, 126-7, 132-4; as guiding standard, 53-4; lack of, 32-3; no mandate for, 140n29, 152n43 public housing, 105 race: of candidate of choice, 75, 76, 77-8; of candidate/of voter, 98-9; in definition of bloc voting, 53-4; as motivating factor in disparate voting patterns, 45; registration by, 23t; in voter behavior, 51-2, 84; and voting rights policy, 131-2 racial bloc voting, see bloc voting

171 racial campaign appeals, 48, 83, 107-8; defined, 107 racial/ethnic divisions, institutionalization of, 131-2, 135-6 racial fairness, 2, 123; population equality in, 110; requirement of, 132 racial gerrymandering, 8, 24, 29, 109, 140n22; identifying, 113-18 racial polarization, 114, 132, 134-5; in multicandidate elections, 99-100; patterns of, 92-3; problems in interpreting, 98103; statistically significant, 142n22 racially polarized voting, 40, 44-6, 47, 48-9, 54, 56, 60, 105, 131, 132, 133; defining and measuring, 3, 46, 49-51, 52, 82-108; elections to study in analysis of, 75-80; evidentiary standard for, 51-3; indications of, 46; keystone of dilution cases in, 4 3 4, 45, 48, 49, 51, 82; legally significant, 73-81, 84, 142n22; and political cohesiveness, 67, 68-9, 144nl4; social science methods in voter dilution cases in, 81; special issues in, 96-8 racism: court-imposed solutions to, 105; proving, 52; in racially polarized voting, 83, 84 Rae, Douglas W., 140n28 rainbow coalition, 153nl0 Raines, Howell, 18 Reagan, Ronald W., 39, 41 Reconstruction, 6, 7 redistricting, see districting redistricting plans: goal of, 65-6; North Carolina, 47-8 redlining, 105 referee provision, 13-14 registration: barriers to, 12, 105; of blacks, 5, 15, 21-2; estimating Spanish-origin, 96-7; in homogeneous precinct analysis, 86, 93; levels of minority, 107, 118-19, 121-2; qualifications/prerequisites to {see also tests and devices), 16-17, 19, 20; by race, 23t, 24t; simplified procedures for, 105 registration data (records), 13, 14, 86, 105, 107 registration laws, 12 registration officials, 13, 18 registration rates, 1, 107; of blacks, 12 Rehnquist, William H., 35, 53, 74, 141nl6, 142n24, 145n29, 149n9 remedy(ies), 15, 60; race-conscious, 132; for vote dilution cases, 132 remedy phase, 111, 112, 116, 118; of Garza, 121 representation, 1, 3, 129; distinct from voting, 4, 23-8; inequities in, 134; fair, 133;

172 maximum, 116-17; system of, 110, 125; thresholds of, 121-2; vote dilution and, 26; see also proportional representation; right to representation, right to vote and representative democracy, 136 Republican party, 6; primaries, 97 residency requirements, 7, 9, 10, 11; standardized, 19 responsiveness of government, 40, 42, 43, 44 restrictive covenants, 105 results test/standard, 38, 39, 40, 42, 43 retrogression, see nonretrogression Rhode Island, 138nl Richmond, Va., 30 right to representation, right to vote and, 2, 4 28, 130-1 right to vote, 137; as fundamental right, 37; and right to representation, 2, 4-28, 129— 30 Riker, William H., 123 Robinson, W. E., 89 Rockford, 111., 126 Roosevelt, Franklin D., 12 Rosenburg, Harry, 65 runoff elections, 3, 106, 122-4 rural malapportionment, 134 Rush, Mark E., 152n3

Sacks, Jerome, 103, 147nl3, 148n23 Saguache County, Col., 69, 142n27 Scarrow, Howard, 117, 127 scattergrams, 147nl6 school board elections, 127, 135 school busing, 105 Schuck, Peter, 135 Schwartz, Bernard, 5 segregation: de jure, 105; residential, 103, 105, 131, 132; social, 105 Seitzinger, Michael V., 66 semiproportional systems, 124-6 Senate factors, see totality of the circumstances Seventh Circuit Court of Appeals, 56-7, 62, 144nl6, 150-ln26 Shugart, Matthew Soberg, 132, 151n35 simultaneous equation techniques, 146n7 single-equation ecological regression, 92 single-member districts, 3, 26, 106, 135, 147nl4; applicability of Gingles to, 81; minority voting in, 67; majority of, 62-A; modifying Gingles to apply to, 115-17; and proportional representation, 132-3, 134; as remedy, 60; runoffs in, 123; standards other than, 124-8; vote dilution

Index in, 2, 109-28; vote dilution standard for, 80-1, 113-18 single-nontransferable vote, 124, 127 single transferable vote (STV), 127 Sisseton, S.D., 55, 126 Sixth Circuit Court of Appeals, 58-9, 117, 144nl6, 145n32, 150n25 65 percent rule, 120 slating, 40, 106, 140n22, 141n8; exclusive, 24 social science methods in vote dilution cases, 81-108 social scientists, 1, 2, 3; see also expert witnesses society, color-blind versus racial/ethnic divided, 131-2, 135-6 socioeconomic status of minorities, 1, 42, 43, 44, 48, 52, 136; in political cohesiveness, 71-2; previous discrimination in, 106-7 South (the), 136; constitutionality of multimember districts in, 33; disfranchisement in, 4, 14—15; political participation of blacks in, 12; racial polarization in, 132; struggle for minority voting rights in, 4-15; voter registration rates in, 1 South Carolina, 11-12, 17, 138n5, 140n5, 141n8, 147n21, 153nlO; blacks holding office in, 5; Constitutional Convention, 9, 138n6; disfranchisement of blacks in, 7-8 South Dakota, 21 southern state conventions, black disfranchisement in, 8-10 southern states/jurisdictions, 29; covered, in Voting Rights Act, 17, 19, 21, 22, 23t, 24t, 25t; history of official discrimination in, 105; thresholds of representation and exclusion for blacks in, 122; vote dilution in, 130 Southwest (the), 21; history of official discrimination in, 105 Spanish-origin population, 20; estimates of, 93-4, 96-7, 107, 146n6; see also Hispanics Spanish-surname registration, 9 3 ^ , 96-7 Springfield, 111., 56-7 staggered elections, 106 standard of proof (evidentiary standard), 37, 38, 41, 42; districting challenges, 61; racial bloc voting, 51-3, 60; vote dilution cases, 47, 49, 63 Stanley, Harold W., 123, 138nl, 152n41 state laws, 18; compactness requirement in, 66; disfranchising blacks, 5, 7-8, 10, 11 statistical analysis, 52, 69; of racially polarized voting, 83-108; reliability of, 103-4

Index statistical evidence: standard for, 51-3; of voting patterns by race, 74-5, 77 statistical significance, 83, 84, 91; tests of, 146n2 Stevens, John Paul, 35, 36, 51, 74 Stewart, Potter, 35, 41 Still, Edward, 126, 127 Strange, Rick, 70, 71 suffrage provisions, state constitutions as, 8-10 Supreme Court, see U.S. Supreme Court Swain, Carol, 135 Taagepera, Rein, 132, 151n35 Tabb, David H, 153nl2 Taebel, Richard L., 126 Tennessee, 9 Tensas School Board, 149n4 Tenth Circuit Court of Appeals, 69, 143n27; and candidate of choice, 78; and exogenous elections, 79; and political cohesiveness, 68 tenuousness factor, 40, 44 Terrell County, Ga., 139nl2 test scores, 129 tests and devices, 16-17, 19, 20, 27, 139nl8; defined, 139nl4; dilutive, 1, 8, 24, 29, 30, 31, 51,59, 141n8 Texas, 11, 21, 38, 66, 135, 140n5, 149n5; House District 30, 144nll; racial discrimination in election process, 33; voting rights litigation, 20 Thernstrom, Abigail M., 1, 16, 24, 129, 130, 131, 132, 135 Thirteenth Amendment, 4, 5 Thomas, Daniel H., 14 Thornberg, see Gingles threshold(s): Gingles, 56, 59; liability, 131; of representation and exclusion, 118, 121-2, 142n26 TIGER Files, 94 Tjoflat, Gerald Bard, 57, 58, 143n28 totality of the circumstances, 33, 34, 40-1, 423, 44, 46, 48, 142n26; cases using, 35, 36; defining and measuring elements of, 82-108; in Gingles, 49-51; other elements of, 105-8; in revision of Section 2, 39; in Zimmer v. McKeithen, 33-4 totality-of-circumstances test, 3, 143nl, 151n40; ambiguities in, 2; impact of Gingles on, 54-60; for single-member districts, 113-14 "triggering" formula, 16, 17, 29, 31, 139nl3 Tufte, Edward R., 132 Turner, James, 65 turnout: black, 22; equalizing, 121; estimation

173 of, 88; in homogenous precinct analysis, 93, 101; minority, 64, 107, 118, 119; for presidential elections of 1964 and 1968, 25t; for presidential elections since 1964, 26t; by race, 86; white/Anglo, 83 Tuskegee, Ala., 130, 140n4 Twenty-fourth Amendment, 21, 139nl5 Twenty-sixth Amendment, 21, 139nl7 "understanding clause," 8, 9, 10, 138n4 United States as amicus, 48-9, 51 U.S. Bureau of the Census, 94, 96; Current Population Survey (CPS), 107 U.S. Congress, 2, 113; and franchise for blacks, 4—5; and interpretation of Constitution, 38; vote dilution standard, 27, 28, 38-42; Voting Rights Act, 15-21, 31; and voting rights enforcement, 7, 910, 12-15 U.S. Congress, House of Representatives, 5, 122; Subcommittee on Civil and Constitutional Rights, 39 U.S. Congress, Senate, 5; Judiciary Committee, 39^0 U.S. Department of Justice, 12, 13, 15, 16, 17, 21-2, 30, 31, 59, 65, 71, 96, 109, 126, 140n5; Voting Rights Section, 120 U.S. District Court for the District of Columbia, 17, 19, 24, 30 U.S. Supreme Court, 2, 15, 128; and acceptance of statistical analysis, 74-5; and black disfranchisement, 10-11; and cases involving single-member districts, 109; decisions regarding Enforcement and Force acts, 6-7; and determination of candidate of choice, 76; and Mississippi state constitution suffrage provisions, 9— 10; and population projections, 111-12; and proportional representation, 132; and racial bloc voting, 78-9, 83-4; and racially polarized voting, 82, 91; and right to vote, 130; Section 2 issues and, 47-60; and 65 percent rule, 120; and size of districts, 110-11; and vote dilution, 30-8, 41-2; and Voting Rights Act, 18-21, 29; and weighted voting, 127 Vermont, 138nl violence, 5—6, 8, 14 Virginia, 9, 17, 46, 134, 138n5, 140n3, n5 vote dilution, 2, 23-8, 135; concept of, 130; constitutional protection against, 31-8; definition and measurement of, 3, 25; election practices increasing likelihood of, 105—6; racial, 32-3; Section 5 protection

174 against, 30-1; in single-member districts, 109-28; three-pronged test for, 49-51; use of Voting Rights Act in, 129-30 vote dilution cases: legal analysis in, 49; social science methods in, 81; three-part test for, 47, 53^4, 61; relative to totality-ofcircumstances test, 54—5 vote dilution claims, statutory, 54, 61 vote dilution standard/test, 2, 132-3; dictated by Congress, 3 8 ^ 2 ; evolution of, 28, 2960; post-Gingles, 61-81; in singlemember districts, 113-18 voter requirements/qualifications, 9, 13-14, 16, 39 voter turnout, see turnout voting: barriers to, 105; distinct from representation, 4, 23-8; federal monitoring of, 13, 16, 18, 21, 22 voting age, 19, 118-19 voting-age population (VAP), 62-4, 86, 93, 100-1; estimating Spanish-origin, 97 voting behavior, 132; differences in, across subgroups, 100; inferences about, 146n6; measuring individual, 84-8; subgroup, 102-3 voting patterns, 77; disparate, 45-6, 73, 84; of minority communities, 67-9; polarization in, 92-3; in political cohesiveness, 67-9, 70-1, 72-3; by race, 74-5, 97; and redistricting, 104 voting rates: of blacks, 1; see also participation voting rights, 1, 2; effect of Gingles on, 61; enforcement of, 5; protection against vote dilution, 30-8; protection of, 7; statutory violation of, 38, 40; struggle for, 2, 4-15, 28; see also right to vote Voting Rights Act of 1965, 2, 3, 4, 15-23, 28, 71, 86; challenges to, 18-19, 123, 140n4; congressional intent of, 39^40, 52, 53, 71, 129-30; content of, 16-19; enforcement/ implementation of, 132, 137; extensions/ expansion of, 19-21, 130; and future of American politics, 134-7; impact on black participation, 21-3; interpretation of, 109-10; 1970 amendments, 19-20; 1975 revisions, 20-1; 1982 amendments, 2, 28, 38-42; normative aspects and interpretation of, 129-37; proportional representation not guaranteed in, 125; Section 2, 2, 16, 25, 29, 35, 36, 38, 61, 70, 73, 75, 78, 80, 107, 109, 110-11, 112, 115, 116, 117, 123, 124, 130, 131, 140n4, n7; Section 2, amended, 42-7, 61;

Index Section 2, expansionist interpretation of, 129; Section 2 and proportional representation, 132; Section 2, revision of, 39—41; Section 2, Supreme Court review of, 47-60; Section 3, 16; Section 4, 1617, 19, 20, 29, 30; Section 5, 17-18, 19, 24-5, 26, 29, 30-1, 65, 71, 109, 115, 120, 131, 140n4, n5, 144n7; Section 6, 18; Section 7, 18; Section 8, 18; Section 10, 18; Senate Report (1982), 27; special provisions, 16-17, 19, 20-1, 38-9; and vote dilution, 24—8; and vote dilution standard, 29 voting rights cases, 2; in the South, 14, 15; testimony of social scientists in {see also expert witnesses), 2 voting systems in minority potential to elect, 124-6 Waite, Morrison R., 7 Warren, Earl, 15, 18-19, 25 Washington, Craig, 134 Washington, Harold, 134, 137 Watsonville (Calif.) City Council, 58 Weaver, Leon, 127 weighted voting, 124, 127 Welch, Susan, 153nl2 West (the), 21 West, E. Gordon, 14 White, Byron R., 34, 35, 36, 41, 51, 74, 76, 142n24, 149n9 white bloc voting, 73-81 white primary system, 9, 10, 11-12, 33, 130 white supremacists, 6, 8 white-versus-white elections, 75, 76, 77, 78 white(s), 8, 101, 137; disfranchisement of poor, 138n4; illiterate, 8, 9; registration rates, 1, 22, 23t, 24t; voter turnout for presidential elections, 26t; voting for blacks {see also crossover voting), 1, 134; voting patterns, 67, 68 Wilder, Douglas, 134, 136-7 Williams, Juan, 16 winner-take-all elections, 132 Wisdom, John Minor, 43-4, 70 Wolfley, Gordon, 139n21 women, black, 4, 10 Woodward, C. Vann, 7, 9, 138n4 World War II, 12 Wright, Stephen G., 123 Wyoming, 139nl6 Zaller, John, 137 Zimmer factors, 34, 141nlO; see also totality of the circumstances