TABLE OF CONTENTS
PAGE
I.FACTUAL BACKGROUND..........................:................. 1112
II. SECTION 2 OF THE VOTING RIGHTS ACT.................. 1113
III. STANDARD OF REVIEW............................................. 1116
IV. THE DISTRICT COURT’S OPINION................................... 1117
V. THE MERITS .................,....................................... 1118
A. INTRODUCTION ................................•..!.............. 1118
B. WHITE BLOC VOTING ....!...................................... 1118
1. The Supreme Court Standard................................... 1118
2. The Plaintiffs Evidence Of White Bloc Voting......:............. 1119
3. The District Court’s Evaluation of the Plaintiffs’ Evidence......... 1121
4. Validity of The District Court’s Plurality-Win Theory............. 1122
5. Black Voters’ Candidates of Choice.............................. 1124
a. Introduction................................................ 1124
b. Were the Black Candidates the Candidates of Choice Among the Black Voters of Red Clay?.................................. 1126
c. Were Any of the White Candidates in White Versus White Elections the Minority-Preferred Candidates?..................... 1128
d. Can the Court Determine a Pattern Based on the Limited Number of Elections Addressed by the Plaintiffs?..................... 1130
C. SUSTAINED PROPORTIONAL REPRESENTATION............... 1131
D. POLITICAL COHESIVENESS OF THE BLACK RED CLAY VOTERS 1133
E. TOTALITY OF THE CIRCUMSTANCES........................... 1135
VI.CONCLUSION..................................... 1136
Before: BECKER, MANSMANN and NYGAARD, Circuit Judges.OPINION OF THE COURT
BECKER, Circuit Judge.This appeal arises from a class action challenge brought pursuant to § 2 of the Voting Rights Act of 1965, 79 Stat. 437, as amended, 42 U.S.C. § 1973 (1988) (“the Act”), to the at-large scheme for electing members of the school board for the Red Clay Consolidated School District in Delaware. The three named plaintiffs, Alden Jenkins, Harlan Roberts, and Gwendolyn Neal, brought the action on behalf of all eligible black voters in the district. They allege that the current method of electing the Red Clay Board of Education unlawfully dilutes the voting strength of the black citizens of Red Clay, thereby *1112depriving them of an equal opportunity “to participate in the political process and to elect representatives of their choice,” 42 U.S.C. § 1973(b). The plaintiffs seek declaratory and injunctive relief barring the use of the present at-large voting system, and mandatory injunctive relief establishing an alternative, non-discriminatory system for future elections. The defendants are the Red Clay Board of Education and the individual members of the Board in their official capacities. Following a bench trial, the district court found that the plaintiffs had failed to prove a § 2 violation, and therefore entered judgment for the defendants. See Jenkins v. Red Clay Consolidated School Dist. Bd. of Educ., 780 F.Supp. 221 (D.Del.1991). Plaintiffs’ appeal followed. We have jurisdiction pursuant to 28 U.S.C. § 1291.
In Thornburg v. Gingles, 478 U.S. 30, 106 S.Ct. 2752, 92 L.Ed.2d 25 (1986), the Supreme Court established three preconditions (the “Gingles factors”) to a finding that the dilutive effects of a multimember district violated § 2. We focus here predominately on the third Gingles factor, “that the white majority votes sufficiently as a bloc to enable it — in the absence of special circumstances ... — usually to defeat the minority’s preferred candidate,” id. at 50-51, 106 S.Ct. at 2766-67, because we have determined that it is dispositive of this appeal.
We conclude that the district court committed reversible error with respect to the third Gingles factor by relying on the potential for black voters to elect their representatives of choice with a plurality of the vote (as is permitted under the Red Clay voting scheme), when the record demonstrated that no Red Clay candidate had won with a mere plurality since 1981. The district court thus erred in failing to ground its analysis in the actual, rather than the potential, effect that the plurality voting scheme had on the ability of black Red Clay voters fully to participate in the political process and to elect their representatives of choice. Concluding also that the judgment cannot be affirmed on alternative grounds, we will reverse and remand for further proceedings, and suggest that the district court receive additional evidence on remand.
I.. FACTUAL BACKGROUND
The Red Clay Consolidated School District was established by the Delaware State Board of Education in November of 1980, pursuant to the authority of 62 Del.Law Ch. 351. The creation of the district was approved by the United States District Court for the District of Delaware, Evans v. Buchanan, 512 F.Supp. 839 (D.Del.1981), as part of its ongoing oversight of the desegregation of the Delaware public school system. Red Clay combines a portion of the city of Wilmington with certain of its suburbs in New Castle County. While Wilmington as a whole has a black majority population, the portion of Wilmington included in the Red Clay School District has a white majority, as do the suburban portions of the district. In sum, according to 1990 census figures, the Red Clay School District has a total population of 132,-674, with a total black population of 19,252 or 14.51%. The total voting age population is 102,196, of whom 13,257, or 12.97%, are black.
The School Board has seven members, each of whom normally serves a term of five years.1 Each Board Member must be a resident of a different nominating district (denoted as districts A-G), and must run for the seat associated with that nominating district. Elections are staggered so that only one or two seats are up for election each May. Beyond the residency requirement, the only requirement for becoming a candidate is to submit a petition signed by twenty eligible district voters. Elections are conducted on an at-large, non-partisan basis. Voters may vote at any polling place in the district, and they may vote for a single candidate running for each of the seats up for election in that year. There is no voter registration; any resident of the district who is at least eighteen years of age may vote. The candidate *1113who receives at least a plurality of the votes for a particular seat is elected.
In 1981, when the first elections were conducted under the reformulated Red Clay School District scheme, elections were held for six of the seven School Board seats.2 For the district B seat, a black candidate, Harlan Roberts, was elected over five white candidates. Two other black candidates, William Anderson from district A and Gregory Con-nor from district E, also sought seats on the Board, but were defeated. All other candidates were white.
In 1982, two seats were up for election. Two white candidates ran for the seat in district C, and the white incumbent ran unopposed for the, seat in district G.
In 1983, a black and a white candidate vied for the seat in district A. Similarly, the seat from district D was contested by a black and a white candidate. Both of these elections were won by the white candidates.
In 1984, three white candidates ran for the seat in district D, and two white candidates ran for the seat in district E.
In 1985, incumbent black candidate Roberts was defeated by a white candidate for the seat in district B. Additionally, the incumbent white candidate ran unopposed for the seat in district G.
In 1986, a white candidate was appointed to fill the seat in district C. Three white candidates ran for the seat in district F.
In 1987, the only election was for the seat in district C, in which there were four white candidates.
In 1988, a. black candidate ran against two white candidates for the seat in district A. The seat was won by one of the white candidates.. Additionally, an incumbent white candidate ran unopposed for the seat in district D.
In 1989, two white candidates ran for the seat in district E, the sole seat up for election.
In 1990, two black candidates, Carol Scot-ton and Rita Shockley, ran against each other for the seat in district B. Scotton was elected. Two white candidates ran against each other for the seat in district G.
In 1991, a black and a white candidate ran for the seat in district C, with the black candidate, Ronald Greene, winning. Two white candidates ran for the seat in district F. The 1991 election occurred after the trial. Neither party attempted to reopen the record to introduce evidence about the election, but the district court took judicial notice of the results.
In sum, in the period between 1981 and 1991, twenty-five vacant seats on the Red Clay School Board were filled. Twenty of these were filled through contested elections; three through uncontested races; and two were appointed. Ten black candidates ran in nine separate contested elections (the 1990 election for the seat in district B involved two black candidates running against each other).3 Black candidates were elected in three of those races: Roberts in 1981, Scotton in 1990, and Greene in 1991. In contrast, sixty-five white candidates have sought twenty-four separate seats, including two seats that were appointed and three instances in which a white candidate ran unopposed. A white candidate ended up filling those seats on twenty-two of the twenty-four occasions.
II. SECTION 2 OF THE VOTING RIGHTS ACT
Section 2 of the Voting Rights Act, as amended, establishes that:
No voting qualification or prereqüisite 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 abridgment of the right of any citizen of the United States to vote on account of race or color....
42 U.S.C. § 1973(a).4
In 1980, the Supreme Court interpreted § 2 to require a plaintiff to prove discrimina*1114tory intent in the design or maintenance of the challenged scheme in order to establish a violation. City of Mobile v. Bolden, 446 U.S. 55, 100 S.Ct. 1490, 64 L.Ed.2d 47 (1980) (plurality opinion). Responding to Bolden, Congress in 1982 amended § 2 to make clear that a violation could be proved by showing a discriminatory effect and to establish that the “results test” that had been applied by the courts before Bolden was the appropriate legal standard. See Thornburg v. Gingles, 478 U.S. 30, 35, 106 S.Ct. 2752, 2758, 92 L.Ed.2d 25 (1986).
In particular, Congress added the following provision setting out the general framework for proving a violation of § 2:
A violation of subsection (a) of this section is established 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) of this section 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.
42 U.S.C. § 1973(b).
Under this “totality of circumstances” analysis, the courts are to consider a wide variety of factors. To aid the analysis, the Senate Report accompanying amended § 2 set out a number of typical factors, which we refer to as “the Senate Factors.” See S.Rep. No. 417, 97th Cong., 2d Sess. 28-29 (1982), reprinted in 1982 U.S.C.C.A.N. 177, 206-07.5 *1115“[T]here is no requirement that any particular number of factors be proved, or that a majority of them point one way or the other.” Id. at 29, reprinted in 1982 U.S.C.C.A.N. at 207. “Rather, the Committee determined that the question whether the political processes are equally open depends upon a searching practical evaluation of the past and present reality and on a functional view of the political process.” Gingles, 478 U.S. at 45, 106 S.Ct. at 2768-64 (internal quotation and citation omitted).
In Gingles, the Court was presented with its first post-amendment § 2 challenge to a voting scheme. There, a class of all registered black citizens of North Carolina challenged the composition of seven state legislative districts established by a 1982 act of the North Carolina General Assembly that redis-trieted the state for elections of North Carolina’s bicameral legislature. See Gingles v. Edmisten, 590 F.Supp. 345, 349-50 (E.D.N.C.1984) (three judge court).- The plaintiffs alleged that in six of these seven districts the redistricting “plan impermissibly dilute[d] the voting strength of the state’s registered black voters by submerging black voting minorities in multi-member” districts. Id. at 349.
In evaluating the plaintiffs’ challenge, the Supreme Court explained: “The essence of a § 2 claim is that a certain electoral law, practice, or structure interacts with social and historical conditions to cause an inequality in the opportunities enjoyed by black and white voters to elect their preferred representatives.” Gingles, 478 U.S. at 47, 106 S.Ct. at 2764. As a gloss on the totality of the circumstances analysis of § 2, the Court set out three factors which it held to be “necessary preconditions” for- a finding that multimember districts “operate to impair minority voters’ ability to elect representatives of their choice”: (1) “the minority group must be able to demonstrate that it is sufficiently large and geographically compact to constitute a majority in a single-member district”; (2) “the minority group must be able to show that it is politically cohesive”; and (3) “the minority must be able to demonstrate that the white majority votes sufficiently as a bloc to enable it — in the absence of special -circumstances ... — usually to defeat the minority’s preferred candidate.” Id. at 50-51, 106 S.Ct. at 2766-67.
.The Gingles factors, however, are merely threshold considerations. The ultimate determination under § 2 remains whether, under the totality of the circumstances, “the multimember districting scheme at issue in this case deprived , black voters of an equal opportunity to participate in the political process and to elect representatives of their choice.” Id. at 77, 106 S.Ct. at 2780; see also S.Rep. at 30, reprinted in 1982 U.S.C.G.A.N. at 207. Thereforé, even after a- court has determined that the plaintiffs proved each of the Gingles factors, it must go on to consider whether the totality of circumstances, evaluated under “a searching practical evaluation of the ‘past and present reality’ ” and a “functional view of ‘political process,’ ” establishes that the particular voting -scheme diminishes the minority group’s opportunity fully to participate in the political process. S.Rep. at 30 & n. 120, *1116reprinted in 1982 U.S.C.C.A.N. at 208 & n. 120 (citation omitted); see Westwego Citizens for Better Gov’t v. City of Westwego, 872 F.2d 1201, 1206 (5th Cir.1989) (’Westwego I) (“The final determination, however, must be made by an evaluation of the ‘totality of the circumstances,’ including the factors listed in the Senate Report.”); see also Gingles, 478 U.S. at 79, 106 S.Ct. at 2781.6
We will elaborate further on the Gingles factors in our discussion of the merits.
III. STANDARD OF REVIEW
The Supreme Court’s opinion in Gin-gles includes specific guidance on the proper standard of appellate review in eases under the Voting Rights Act. The parties in Gin-gles argued that the district court’s ultimate determination as to vote dilution constituted a mixed question of law and fact subject to de novo review. The Court, however, referencing the particularly local and fact-intensive nature-of the inquiry, reaffirmed its view in prior voting cases “that the clearly-erroneous test of Rule 52(a) is the appropriate standard for appellate review of a finding of vote dilution.” Gingles, 478 U.S. at 77-79, 106 S.Ct. at 2780-81; see also id. at 78, 106 5.Ct. at 2781 (“ ‘we are not inclined to overturn these findings [that minority voters had less opportunity to participate in the political process], representing as they do a blend of history and an intensely local appraisal of the design and impact of the ... multimember district in the light of past and present reality, political and otherwise.’ ”) (quoting White v. Regester, 412 U.S. 755, 769-70, 93 S.Ct. 2332, 2341, 37 L.Ed.2d 314 (1973)). The Court further explained:
As both amended § 2 and its legislative history make clear, in evaluating a statutory claim of vote dilution through district-ing, 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. “ ‘This determination is peculiarly dependent upon the facts of each case,’ ” and requires “an intensely local appraisal of the design and impact” of the contested electoral mechanisms.
Id. 478 U.S. at 79, 106 S.Ct. at 2781 (citation omitted).
In light of the Court’s attention to the superior vantage point of the district courts, we conclude that the various subsidiary findings in a voting rights case, such as the court’s findings on each of the three Gingles factors and on the various Senate Factors, are, also factual determinations subject to clearly erroneous review. See, e.g., Westwego Citizens for Better Gov’t v. City of Westwego, 946 F.2d 1109, 1118 n. 13 (5th Cir.1991) (Westwego III).
The Rule 52(a) standard does not, however, give a district court unfettered discretion in evaluating a claim. While the Court in Gingles emphasized the deference due district court fact findings, it also recognized that
Rule 52(a) “does not inhibit an appellate court’s power to correct errors of law, including those that may infect a so-called mixed finding of law and fact, or a finding of'fact that is predicated on a misunderstanding of the governing rule of law.”
*1117Gingles, 478 U.S. at 79, 106 S.Ct. at 2781 (quoting Bose Corp. v. Consumers Union, 466 U.S. 485, 501, 104 S.Ct. 1949, 1960, 8 L.Ed.2d 502 (1984)). This result is clearly necessary since both the ultimate determination of vote dilution and the accompanying-subsidiary determinations often depend on important legal questions. Accordingly, we must subject the district court’s underlying legal analysis to plenary review to ensure that the appropriate legal standards are applied. As the Court concluded, it is this combination of factual deference and legal review that best “preserves the benefit of the trial court’s particular familiarity with the indigenous political reality without endangering the rule of law.” Id. 478 U.S. at 79, 106 S.Ct. at 2781.
IV. THE DISTRICT COURT’S OPINION
Following a six day bench trial, the district' court filed an extensive opinion denying the plaintiffs’ claims under- § 2 of the Voting Rights Act, 42 U.S.C. § 1973. Jenkins v. Red Clay Consolidated School Dist. Bd. of Educ., 780 F.Supp. 221 (D.Del.1991). The court began its analysis by considering the extent of black electoral success.7 Id. at 224-26. Recognizing that “black citizens have achieved some electoral success in the District,” the court nevertheless concluded that because there also had been some under-representation and because some of the success may have resulted from unusual circumstances, it would “proceed to consider the Gingles threshold factors.” Id. at 226.
The district court concluded that the first Gingles factor was satisfied. In particular, the court found that if the school district was divided into single member districts, a majority black district could be created. Id. As to the second factor, after discussing the evidence on the issue, the court determined that the black voters in the Red Clay School District usually were politically cohesive. Id. at 231-. The court stressed, however, that while it believed that the plaintiffs had managed to prove political eohesiveness, the evidence supporting this finding was “generally weak.” Id.
With regard to the third Gingles factor, the district court concluded that the plaintiffs had failed to establish the existence of white bloc voting because their statistical evidence either was not probative or failed to show a pattern of white bloc voting that usually defeated the black voters’ candidate of choice. Id. at 231-32. In the court’s view, the most relevant evidence, because it covered a wider number of elections than the plaintiffs’ other statistical -evidence, arguably was the plaintiffs’ calculation of the theoretical maximum white vote that the various black candidates could have' received. These calculations demonstrated that in the overwhelming number of elections a substantial majority of white voters voted- against the minority-preferred candidate. The court found this evidence unpersuasive, however, concluding that the fact that white voters usually vote sufficiently as a bloc- against the black candidate, thereby preventing that candidate from receiving a majority of the votes, is insufficient to prove the existence of legally significant white bloc voting because minority-preferred candidates could still be elected by winning a plurality of the votes in multi-candidate elections. Id. at 233. We refer to this as the district court’s “plurality-win” theory.
The court also found that, because the theoretical maximum white vote for.black candidates ranged from nine to fifty-nine percent, there was not “a consistent pattern of whites voting against the black candidate.” Id. Finally, the court expressed doubt “that all of these black candidates were the preferred candidates .of the black community.” Id.
Having found that the plaintiffs established only two of the- three Gingles factors, the district court concluded that the plaintiffs’ claim must fail. Id. Nevertheless, for the sake of providing a comprehensive opinion,- the court went on to address the Senate Factors, and found, albeit in a conclusory fashion, that the plaintiffs had failed to estab*1118lish a violation of § 2 under the totality of the circumstances. Id. at 233 — 41. We will discuss the district court’s opinion in more detail in the course of the merits discussion that follows.
V. THE MERITS
A. INTRODUCTION
It is undisputed that voting age blacks in the Red Clay School District constitute a group “sufficiently large and geographically compact to constitute a majority in a single-member district.” See Gingles, 478 U.S. at 50, 106 S.Ct. at 2766. This is, however, essentially the only issue about which the parties agree.
The plaintiffs’ primary argument on appeal is that the district court committed reversible legal error in its consideration of whether the plaintiffs proved that white voters vote sufficiently as a bloc usually to defeat the preferred candidate of the- minority community. The plaintiffs -argue that the district court’s reliance on the potential for minority-preferred candidates to win with a plurality, thereby overcoming the majority opposition of white voters, was contrary to the standard for legally significant white bloc voting announced by the Supreme Court in Gingles.
In response, the defendants argue that the district court was correct that the potential to win with a plurality offsets the effect of showing that black voters do not receive substantial support from white voters. In the alternative, the defendants contend that, even if the district court did err, its ultimate judgment was nevertheless correct. The defendants’ primary alternative argument for upholding the judgment is that the plaintiffs’ evidence was insufficient to establish which candidates were minority-preferred. This is fatal to the plaintiffs’ claim, the defendants argue, because to show legally significant white bloc voting the plaintiffs must establish that there is a pattern of whites voting sufficiently as a bloc usually to defeat the minority-preferred candidate, which can only be determined by the court if it knows who the minority-preferred candidates were in a sufficient number of elections to establish a pattern.
The defendants offer three additional arguments for upholding the judgment: (1) that the black voters of Red Clay have achieved sufficient proportional representation to bar a § 2 claim; (2) that the plaintiffs’ evidence on the second Gingles factor — that the black voters of Red Clay are politically cohesive — was too infirm to prove the factor’s existence; and (3) that the court’s ultimate conclusion that the plaintiffs’ claim failed under the totality of the circumstances was proper and should be affirmed regardless of any other errors.
B. WHITE BLOC VOTING
Under the ' third Gingles factor, plaintiffs must show “that the white majority votes sufficiently as a bloc to enable it — in the absence of special circumstances ...-— usually to defeat the minority’s preferred candidate.” Gingles, 478 U.S. at 51, 106 S.Ct. at 2766-67. With regard to white bloc voting, we will treat two primary issues. First, -we will discuss whether the district court’s plurality-win theory is consistent with the Supreme Court’s standard for legally significant white bloc voting. Second, having concluded that the district court erred on that point, we will discuss which candidates were minority-preferred and whether the plaintiffs have carried -their burden of analyzing a sufficient number of elections to permit the court to determine whether there was legally significant white bloc voting.
1. The Supreme Court Standard
In elaborating on the degree of bloc voting that plaintiffs must show to establish the third Gingles factor, the Supreme Court concluded that “there is no - simple doctrinal test for the existence of legally significant racial bloc voting.” Id. at 58, 106 S.Ct. at 2770. The Court, however, did set out the following general principles for determining whether legally significant white bloc voting exists:
[I]n general, a white bloc vote that normally will defeat the combined strength of minority support plus white “crossover” votes rises to the level of legally significant white bloc voting. The amount of white bloc voting that can generally “minimize or cancel” black voters’ ability to elect repre*1119sentatives of their choice, however, will vary from district to district according to a number -of factors____
Because loss of political power through vote dilution is distinct from the mere inability to win a particular election, a pattern of racial bloc voting that extends over a period of time is more probative of a claim that a district experiences legally significant polarization than are the results of a single election. Also for this reason, in a district where elections are shown usually to be polarized, the fact that racially polarized voting is not present in one or a few individual elections does not necessarily negate the conclusion'that the district experiences legally significant bloc voting.
Id. at 56-57, 106 S.Ct. at 2769-70 (citations and footnotes omitted).
2. The Plaintiffs’ Evidence Of White Bloc Voting
In attempting to establish the existence of legally significant white bloc voting, the plaintiffs introduced expert testimony, which included statistical analyses of a number of both Red Clay elections and exogenous elections (i.e., elections other than those in the Red Clay School District), lay testimony, and evidence of the results of past Red Clay School Board elections. The evidence established that black candidates were involved in nine of the twenty-three elections occurring between 1981 and 1991,8 including the post-trial 1991 election.
A black candidate won three of the nine elections in which black candidates were involved. These were the 1981 district B race in which Roberts was elected; the 1990 district B race in which Seotton was elected; and the 1991 district C race in which Greene was elected. However, as to the two black candidates elected before 1991, the plaintiffs claim that the court should have disregarded the results of those elections because the candidates were elected under “special circumstances,” namely, that Roberts was able to win in 1981 only because the five white candidates against whom he ran split the white vote, and that Seotton was elected in 1990 because of the pendency of the present litigation.9 Moreover, with regard to the 1991 election, the plaintiffs claim that because there was no evidence introduced about the circumstances surrounding that election, it cannot be considered.
The plaintiffs also introduced lay testimony and expert statistical evidence on the voting patterns of both black and white Red Clay voters.' The plaintiffs’ expert, Dr. Allan Lichtman, performed bivariate regression analysis10 on the 1988 Red Clay School *1120Board election between Mitchell, a black candidate, and two white candidates. According to Dr. Liehtman’s regression analysis, only approximately 1% of the white voters voted for Mitchell, whereas virtually 100% of the black voters supported him. Additionally, Dr. Lichtman’s analysis found a correlation coefficient for this election of 0.91, which represents a highly consistent association between the number of votes received by a candidate of a particular race and the racial composition of the polling place.11 See Jenkins, 780 F.Supp. at 227. 'The plaintiffs also introduced the results of the extreme case analysis12 done by Dr. Lichtman on the 1983, 1985, and 1988 elections involving black and white candidates.13 Additionally, the plaintiffs provided extreme case, regression, and scatter point14 analyses for a number of exogenous elections.
Finally, Dr. Lichtman calculated the maximum percentage of the white vote that the various black candidates theoretically could have received in the elections through 1988. Using the actual vote totals received by the black candidates, and assuming, for the sake of analysis, that all of the votes they received were cast by white voters, Lichtman calculated the maximum possible percentage of white voters who could have voted for the black candidates under either equal white and black turnout, or, alternatively, a black voter turnout rate twice that of white voters.15 Assuming equal turnout rates, the theoretical maximum white vote for black candidates ranged from a low of 9% to a high of 53%. The average theoretical maximum white vote for the black candidates in the seven elections was approximately 28%.
The calculation of the theoretical maximum white vote for black candidates took on importance in the present case because of the plaintiffs’ inability to obtain the underlying data necessary to determine actual voting patterns, see supra n. 11. While the plaintiffs could not establish the actual voting pattern of white voters, the theoretical maximum white vote calculations allowed the plaintiffs to demonstrate the level of white bloc voting that must have existed. The analysis shows that none of the black candi*1121dates through 1988 received a majority of the white vote.
While the analysis suggests that it was theoretically possible for Roberts to have received 53% of the white vote, realistically, because Roberts received a substantial portion of the black vote, the actual percentage of white voters who voted for Roberts was well below 53%. More generally, because of the very conservative assumptions used to make the calculations — that every black voted for the white candidate and that there was an equal turnout16 — the percentage of white voters who voted for particular black candidates was, on average, substantially lower than the 28% suggested by the analysis. The calculations therefore served to demonstrate that generally a black candidate could prevail only if the white vote was split among a number of white candidates, thereby allowing the black candidate to prevail with a plurality of the vote.
3. The District Court’s Evaluation of the Plaintiffs’ Evidence
In evaluating whether-there was white bloc voting, the court first addressed the plaintiffs’ evidence on the exogenous elections. The court rejected this evidence on the ground that the plaintiffs had failed to prove that the exogenous elections were “representative of the voting patterns of the white Red Clay electorate” because “less than 21 percent of the white Red Clay electorate were eligible to vote in those [exogenous] elections” and because “the exogenous elections generally covered districts within the city of Wilmington [whereas] most of the white Red Clay electorate lives in the suburban areas of the District.” Jenkins, 780 F.Supp. at 231.
One exogenous election did include a majority of the white voters from the Red Clay School District, but the court also found the evidence of that election unhelpful because the predominately black districts in that election were outside of the Red Clay School District. The court concluded that there was no indication that the candidate supported by the majority black districts outside Red Clay would have been the “minority-preferred” candidate of the black Red Clay voters had the election taken place in-Red Clay. Id. at 232.
As to the expert evidence on the Red Clay elections, -the court first addressed the regression analysis of the 1988 Mitchell-Nut-tal-Manning election. While the court credited the plaintiffs’ regression analysis, it concluded that that evidence, standing alone, was insufficient because the results of a single election could not establish a pattern of white bloc voting. Id. at 231. Turning to the plaintiffs’ extreme case analysis, the district court again credited their analysis of the 1988 election. Because of the lack of data, however, Lichtman had been forced to assume that the proportion of white voters at the various polling places was the same in 1983 and 1985 as it had been in 1988 and 1989. The district court found that this assumption was unsupportable because there was “no substantial evidence that the racial composition of those districts had not changed ... [or] that the turnout by race was the same.” Id. at 232.
Finally, the district court addressed what was the plaintiffs’ broadest statistical evidence, their calculations of the theoretical maximum white votes that each of the black candidates could have received:
The Court notes that a.candidate need not receive 50 percent of the white vote to win an election in Red Clay when more than two candidates are running for a seat. Therefore, a finding of white bloc voting cannot be based upon the observation that usually less than 50 percent of the white [voters] favor the black candidate. Assuming equal turnout of black and white voters, the white vote for the black candidate ranged from a low of 9 percent to a high of 53 percent. The Court finds that these varied results do not indicate a consistent *1122pattern of whites voting against the black candidate. Furthermore, the Court is not convinced that all of these black candidates were the preferred candidates of the black community. The Voting Rights Act is designed to protect the minority voter not the minority candidate. Only the 1988 election has received extensive analysis by the experts. The lay testimony regarding the 1983 elections is contradictory.
Id. at 233.
In sum, the district court concluded that the plaintiffs had failed to produce relevant proof sufficient to establish, by a preponderance of the evidence, a pattern of white bloc voting that usually would defeat the black Red Clay voters’ candidate of choice.
4. Validity of the District Court’s Plurality-Win Theory
The plaintiffs’ principal argument on appeal is that the district court committed legal error in reasoning that, because candidates can win elections in Red Clay with a plurality of the votes, proof that minority-preferred candidates routinely receive less than a majority of the votés does not tend to establish the third.Gingles factor — legally significant white bloc voting. The plaintiffs argue that the district court’s plurality-win theory, because it focuses on the theoretical potential to elect minority-preferred candidates rather than on the functional reality that the plurality voting scheme has not generally had the effect of facilitating the election of minority-preferred candidates, contradicts the Gingles definition of what is legally significant white block voting. We agree.
We do not gainsay that the ability to elect a candidate with a plurality of the vote may be relevant to the court’s determination of white bloc voting. In Gingles, the Supreme Court adopted a practical, results oriented approach, specifically noting that “[t]he amount of white bloc voting that can generally ‘minimize or cancel’ black voters’ ability to elect representatives of their choice ... will vary from district to district according to a number of factors, including ... the presence or absence of other potentially dilu-tive electoral devices, such as majority vote requirements____” Id. 478 U.S. at 56, 106 S.Ct. at 2769 (citation omitted). If minority-preferred candidates are consistently able to prevail in representative numbers because of the existence of a plurality rule, then it cannot be said that white voters vote sufficiently as a bloc usually to prevent the election of the minority preferred candidates. See id. at 51, 106 S.Ct. at 2767 (“In establishing [the third Gingles factor], the minority group demonstrates that submergence in a white multimember district impedes its ability to elect its chosen representatives.”). The focus of the third Gingles factor is not on the size of the bloc, per se, but the effect of bloc voting on the ability of minority voters fully to participate in the political process and to elect their representatives of choice.
The district court, therefore, did not err in considering the existence of a plurality rule. Rather, the court’s error occurred in analyzing the effect of the rule in terms of whether it created any set of circumstances whereby black candidates might theoretically prevail. The Senate Report repeatedly emphasizes that the court must evaluate plaintiffs’ claims under “a searching practical evaluation of the past and present reality and on a functional view of the political process.” Id. at 45, 106 S.Ct. at 2764 (internal quotation and citation omitted) (quoting Senate Report at 30 & n. 120, reprinted in 1982 U.S.C.C.A.N. at 208 & n. 120). We believe, therefore, that the proper question is not whether the existence of a plurality regime makes it possible for black candidates to prevail, but whether it in fact has had the practical effect of allowing more than sporadic victories by minority-preferred candidates.
The district court’s reasoning in regard to its plurality-win theory is not unambiguous, and it is possible that the court in fact understood the proper legal standard. However, if the court applied the correct legal standard, its conclusion that Red Clay’s plurality rule made it such that white bloc voting did not usually cause minority-preferred candidates to lose was clearly erroneous. While it is true that one black candidate, Roberts, was able to prevail in 1981 because of the existence of the plurality rule (running against five white candidates), as a practical matter, this was a rare occurrence in Red Clay elections. The record demonstrates that no candidate, black or white, was elected to the Red *1123Clay Board of Education with a mere plurality of the votes since 1981; moreover, since 1981, rarely have more than three candidates vied for a single seat,17 and only one post-1981 election involving a black candidate also involved more than one other candidate.
Whether the error here was factual or legal makes no practical difference. The basic concern addressed by the third Gingles factor is whether or not white bloc voting usually defeats the minority-preferred candidate. The existence of a single instance in which a black candidate was elected with a plurality of the votes would not be sufficient to alter an otherwise supportable conclusion that white voters vote sufficiently as a bloc usually to defeat the minority-preferred candidate. While the district court’s findings are due significant deference, see supra Part III., where, as here,, they reflect a misunderstanding of. an essential legal component of the inquiry or a misapprehension of the relevant legal facts, the findings cannot stand. See Gingles, 478 U.S. at 79, 106 S.Ct. at 2781.
The district court also pointed to the variations in the theoretical maximum level of white votes black candidates could have received and found “that.these varied results do not indicate a consistent pattern of whites voting against the black candidate,” Jenkins, 780 F.Supp. at 233, apparently assuming that Gingles requires a consistent level of hostility towards minority-preferred candidates. We believe that this conclusion suffers from a similar legal infirmity.
Gingles does not require a showing that white voters vote as an unbending monolithic block against whoever happens to be the minority’s preferred candidate. To the contrary, the Gingles standard presup-, poses the existence of crossover voting. See Gingles, 478 U.S. at 56, 106 S.Ct. at 2769 (defining legally significant white bloc voting as “a white bloc vote that normally will defeat the combined strength of minority support plus white ‘crossover’ votes”).. Moreover, the standard in Gingles necessarily presupposes variations in the levels of crossover voting from election to. election. The Court specifically stated:
[I]n a district where elections are shown usually to be polarized, the fact that racially polarized voting is not present in one or a few individual elections does not necessarily negate the conclusion that the district experiences legally significant bloc voting.
Id. at 57, 106 S.Ct. at 2770. The correct question is not whether white voters demonstrate an unbending or unalterable hostility to whoever may be the minority group’s representative of choice, but whether, as a practical matter, the usual result of the bloc voting that exists is the defeat of the minority-preferred candidate.
In short, we. conclude that the district court erred in focusing on the potential for Red Clay’s plurality scheme to aid minority-preferred candidates rather than on the practical effects which that scheme actually has had. Since no candidate, black or white, has. won with a mere plurality since 1981, and only one.black candidate has run in an election against more than a single opposing candidate since 1981, the court could not have concluded on this record that the plurality scheme had the actual effect of allowing the black voters of Red Clay to elect their representatives of choice even though the white voters consistently voted against the minority-preferred candidates in numbers sufficient to prevent those candidates from winning a majority of the overall vote.
We have no way', of knowing what conclusions the district court would have arrived at in regards to white bloc voting had it not been laboring under its misperception as to the significance of Red Clay’s plurality voting scheme. Thus, unless the defendants can prevail in their contentions as to alternate bases for affirming the judgment, we must reverse and remand the case to allow the district court to address white bloc voting in light of this opinion.18
*11245. Black Voters’ Candidates of Choice
a. Introduction
The defendants’ principal argument on appeal is that the plaintiffs’ failed to come forward with sufficient proof to establish which candidates, over all the eléctions, were the candidates of choice among the black voters of Red Clay. This is fatal, the defendants submit, because to establish the third Gingles factor a plaintiff must show a pattern of white bloc voting such that the minority voters’ candidate of choice usually will be defeated. According to the defendants, in order to evaluate whether there is such a pattern, i.e., whether the minority-preferred candidate is usually defeated, the court must know who the minority-preferred- candidates were across the whole range of elections. Here, the defendants submit, because the 'plaintiffs analyzed only a small subset of the elections — the seven pre-1991 elections involving a choice between a black and a white candidate — the court had- no indication whether the minority-preferred candidate was defeated in the other 'eleven pre-1991 elections. Further, the - defendants argue, even within the subset of elections that the plaintiffs did analyze, the plaintiffs failed to establish that the black candidates were, all in fact the Candidates of choice among black Red Clay voters.
. This issue has severa^ layers. First, there is the question of which of the candidates in the black versus white elections analyzed by the plaintiffs were in fact the minority-preferred candidates. Second, we must consider whether the plaintiffs were required to introduce evidence of who was the minority-preferred ' candidate in all elections, including those elections involving only white candidates. Ultimately, we must determine whether the plaintiffs’ limited evidence would have been sufficient to support a finding of legally significant white bloc voting. We will address these issues in turn.19 First, however, it is important to provide some background.
The defendants’ specific argument is subsumed within a broader question — the significance of á candidate’s race to the court’s analysis of white bloc voting. The conflict arises in part because Gingles produced a split of opinions on this issue. While Justice Brennan garnered a majority for most of his opinion, that majority was lost in his discussion of the relevance of a candidate’s race. Justice Brennan, joined by Justices Marshall, Blackmun, and Stevens, took the position 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.” Id. at 68, 106 - S.Ct. at 2775 (Brennan, J., plurality opinion) (emphasis in original);. see also id. at 62,106 S.Ct. at 2772 (Brennan, J., plurality opinion) (“For purposes of § 2, the legal concept of racially polarized voting incorporates neither causation nor intent. It means simply that the race of voters correlates with the selection of a certain candidate or candidates----”).
Justice White, who joined Justice Brennan’s opinion in all other aspects, wrote sepa*1125rately to state that, in his view, courts could not ignore the race of the candidate. • To do so, warned-Justice White, would lead to violations based simply on differing interests, not racial discrimination. Id. at 83, 106 S.Ct. at 2783 (White, J., concurring) (“This is interest-group politics rather than a rule hedging against racial discrimination.”).
Justice O’Connor, in a separate opinion joined by Chief Justice Burger and Justices Powell and Rehnquist, joined Justice White in rejecting Justice Brennan’s position that the race of the candidate could never be relevant. Id. at 100-01, 106 S.Ct. at 2792 (O’Connor, J., concurring in the judgment). Justice O’Connor, taking issue with such an absolutist view, concluded that “[t]he overall vote dilution inquiry neither requires nor permits an arbitrary rule against consideration of all evidence concerning voting preferences other than statistical evidence of racial voting patterns.” Id. at 101,106 S.Ct. at 2792 (O’Connor, J., concurring in the judgment).
The courts of appeals, in the absence of explicit direction from the Supreme Court, have adopted a variety of approaches to determining which candidates are minority-preferred. The Seventh Circuit in Baird v. Consolidated City of Indianapolis, 976 F.2d 357, 361 (7th Cir.1992), cert. denied, — U.S. -, 113 S.Ct. 2334, 124 L.Ed.2d 246 (1993), for instance, focused on the race of the candidate, refusing to discount the election of a black Republican, even though the majority of black voters were Democrats who generally refused to support a Republican candidate. The Eleventh Circuit, in contrast, has held that “[ujnder 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.” Carrollton Branch of the NAACP v. Stallings, 829 F.2d 1547, 1557 (11th Cir.1987), cert. denied, 485 U.S. 936, 108 S.Ct. 1111, 99 L.Ed.2d 272 (1988). The Fifth Circuit has adopted yet another position:
[W]e conclude that 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____ Justice Brennan’s plurality opinion is careful not to state that a black candidate is tantamount to . the black preference; but implicit in the Gingles holding is the notion that black preference is determined from elections which offer the choice of a black candidate. The various Gingles concurring and dissenting opinions do not- consider evidence, of elections in which only whites were candidates. Hence, neither do we.
Citizens for a Better Gretna v. City of Gretna, 834 F.2d 496, 503-04 (5th Cir.1987), cert. denied, 492 U.S. 905, 109 S.Ct. 3213, 106 L.Ed.2d 564 (1989).
While we think these standards have much to recommend them, we believe that they are also inadequate standing alone because each adopts too rigid a view as to which candidates may be considered minority preferred. In so doing, each adopts a standard which is over- or under-inclusive. For instance, we conclude that the Seventh Circuit, by calling any same-race candidate minority-preferred, has established an over-inclusive standard. There may well be minority candidates who cannot be considered the minority community’s representative of choice, and § 2 specifically provides that the section is violated if “members [of a protected] minority have less opportunity than other members of'the electorate to participate in the political process and to elect representatives of their choice,” 42 U.S.C. § 1973(b) (emphasis added). Conversely, a rule permitting only minority candidates to be considered minority-preferred may well be under-inclusive; there may be majority candidates who truly may be the minority community’s representative of choice.
Instead, we believe that the goals and underlying principles of the Voting Rights Act are best promoted by a more flexible approach, as appears to have been adopted by the Tenth Circuit in Sanchez v. Bond, 875 F.2d 1488, 1494-95, 1496 (10th Cir.1989), cert. denied, 498 U.S. 937, 111 S.Ct. 340, 112 L.Ed.2d 305 (1990). There, the court rejected a per se rule that elections involving only white candidates could never be relevant. Id. at 1495. Instead, the court concluded:
*1126Such 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. As with other types of relevant evidence, the district court has a right to consider these elections and to give them such wéight- as the circumstances warrant.
Id. While the court in Sanchez provided some guidance for determining which candidates may be considered minority-preferred,20 we believe it is important to embellish slightly on its standard, as described below, to emphasize two points: (1) that not all minority .candidates will be minority-preferred; and (2) that it is important to look beyond whether a white candidate received a majority of the minority community’s vote and determine whether that white candidate is truly the minority community’s representative of ehpice.
b. Were the Black Candidates the Candidates of Choice Among the Black Voters of Red Clay?'
While it may be tempting to assume that a minority candidate is always the candidate of choice among minority voters, this is not always true. A minority candidate may have very different interests than the majority of minority voters. Or, the minority candidate may be viewed as outside the mainstream with no possible hope of success, and may therefore be unable to garner minority support. Accordingly, the plaintiffs must establish on an election-by-election basis, by a preponderance of the evidence, which black candidates are minority-preferred. Cf. Gingles, 478 U.S. at 46, 106 S.Ct. at 2764 (“[T]he results test does not assume the existence of racial bloc voting; plaintiffs must prove it.”).
Nevertheless, experience does demonstrate that minority candidates will tend to be candidates of choice among the minority community. This practical experience leads to the inference that any particular minority candidate is minority preferred, and thereby takes plaintiffs a considerable distance towards meeting their production burden on this issue. This inference, however, is not enough on its own to meet this burden. Rather, .plaintiffs must introduce some additional evidence showing that the particular minority candidate is minority-preferred. The additional evidence required to meet the threshold, however, is not very substantial, and the burden may be satisfied with a variety of evidence, including lay testimony or statistical analyses of voting patterns.
The defendants argue that the plaintiffs failed to establish that the black candidates on which they rely were the preferred candidates of Red Clay’s black voters. While the district court failed to make explicit findings as to which candidates were or were not minority-preferred, it generally noted that it was “not convinced that all of these black candidates were the preferred candidates of the black community.” Jenkins, 780 F.Supp. at 233.
The defendants’ arguments on this point are insubstantial. Of the seven pre-trial elections in which there was a black and a white candidate, the defendants specifically challenge the status of only three of the black candidates as minority-preferred. In particular, they submit that the lay testimony establishes that in two of the elections, the 1983 Anderson-Cavanaugh race and the 1988 Mitehell-Manning-Nuttal race, a white candidate was the preferred candidate of black voters. The defendants also argue from a third election, the 1985 Eckman-Roberts race, where, according to the plaintiffs’ own theoretical maximum white vote analysis, up to 53% of white voters theoretically could have voted for Roberts, the black candidate. While the argument is not well developed, *1127the defendants seem to contend that Roberts was not shown to have been minority-preferred because, based on the theoretical maximum white vote analysis, he may have received a substantial portion of the white vote and yet still lost. Thus, the argument would follow, he could not have been supported by a majority of black voters as well, or he would have won.
As to the other four candidates, the defendants admitted in their Proposed Findings of Fact and Conclusions of Law that the three black candidates in the 1981 elections and Miller in the 1983 election were preferred by black voters. The defendants unambiguously conceded that “the uncontroverted lay testimony tends to show that [these candidates] for the Red’ Clay Board were preferred by black voters.”21 The defendants have pointed to no evidence that would tend to contradict this lay testimony, and they are, at all events, bound by their admissions.
Returning to the three candidates that the defendants do challenge, their arguments as to two are without merit. With regard to the 1988 Mitchell-Manning-Nuttal race, the plaintiffs introduced a regression analysis which indicated that approximately 100% of black Red Clay voters supported Mitchell in the 1988 election. While the defendants expressed a number of concerns. about the methodology employed in that analysis, the district court extensively considered those arguments and ultimately rejected them, crediting the results of the analysis. Jenkins, 780 F.Supp. at 228-29, 230, 231. This finding was supported by the plaintiffs’ evidence, and was not clearly erroneous. Since approximately 100% of the black voters of Red Clay supported Mitchell, he was clearly their candidate of choice.
The defendants point to the testimony of two plaintiffs that they supported one of the white candidates over Mitchell in 1988 to argue that the evidence demonstrated that Mitchell was not the minority’s preferred candidate. The testimony of two individual plaintiffs, however, does not negate the analysis of the plaintiffs’ expert, Dr. Liehtman. Sporadic lay testimony, while not irrelevant, is necessarily less indicative of black voters’ voting patterns generally than the regression analysis. See Gomez v. City of Watsonville, 863 F.2d 1407, 1416 (9th Cir.1988), cert. denied, 489 U.S. 1080, 109 S.Ct. 1534, 103 L.Ed.2d 839 (1989).
The defendants’ argument with regard to.the 1985 Eekman-Roberts race is similarly unavailing. The theoretical maximum white vote analysis does not, and cannot, address the issue of which candidates are minority-preferred. The analysis takes as its starting point the artificial assumption that no black voters supported the black candidate. Thus, it reveals nothing about how many black voters actually supported a particular candidate. At a minimum, therefore, at least six out of the seven black candidates who ran' against white candidates in Red Clay elections prior to 1991 were minority-preferred.
The defendants further assert that the district court’s finding of political cohesiveness among the black citizens of Red Clay does not support an inference that the black candidates in Red Clay elections were minority-preferred. According to the defendants, because the'cohesiveness finding was based primarily on statistical evidence from exogenous elections, it tells us nothing about which candidates the black voters supported in Red Clay elections. The defendants’ argument, however, ignores both the practical realities of the voting behavior and the nature of the proof used to establish cohesiveness.
The plaintiffs’ proof of cohesiveness was premised on Dr. Lichtman’s analysis showing “that the black citizens of Red Clay consistently vote in greater than landslide proportions for the black candidate in elections between black and white candidates.” Jenkins, 780 F.Supp. at 228. The district court credited this testimony, id. at 230, and concluded that the expert testimony in conjune*1128tion with the “mixed” lay testimony, while not overwhelming, was sufficient to establish that the black voters of Red Clay usually were politically cohesive, id. at 230-31. According to the district court’s finding, there was a strong correlation between the race of the candidate and, the preference of black voters. This raises an inference that any particular black candidate was the minority voters’ candidate of choice. And while that inference alone might not be sufficient to establish that a particular candidate was minority-preferred, it takes the plaintiffs á considerable distance toward establishing that proposition.
In light of the foregoing analysis, we conclude that, in at least six of the seven pre-1991 elections in which there was both a white and a black candidate, the plaintiffs’ evidence would have required a finding that the black candidate was the candidate of choice among the black voters of Red Clay. With regard to the seventh candidate, Anderson, the evidence was mixed. Thus, it will be up to the district court on remand to determine whether Anderson was a minority-preferred candidate.
The question remains, however, whether in analyzing only these seven elections the plaintiffs provided a sufficient basis for a finding of legally significant white bloc voting. To answer that question, however, we need consider whether there is necessarily a minority-preferred candidate in every election and whether plaintiffs must identify all of the minority-preferred candidates.
c. Were Any of the White Candidates in White Versus White Elections the Minority-Preferred Candidates?
The defendants also argue that the plaintiffs may not selectively choose which elections to analyze, but rather must analyze all the elections, including those involving only white candidates. It is only on the basis of such a comprehensive analysis, the defendants submit, that the court is able to evaluate whether or not there is a pattern of white bloc voting that usually defeats the minority voters’ candidate of choice. We disagree.
As a general matter, we believe that elections involving white candidates only are much less probative of racially polarized voting than elections involving both black and white candidates. As was explained by the Fifth Circuit in Westwego III, 946 F.2d 1109 (5th Cir.1991):
“[T]he evidence most probative of racially polarized voting must be drawn from elections including both black and white candidates.” “[W]hen 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.’ ” Thus, it is not particularly surprising — and not particularly probative — that analysis of elections that included only white candidates did not reveal any racial polarization.
Id. at 1119 n. 15 (citation omitted; quoting Westwego I, 872 F.2d 1201, 1208 n. 7 (5th Cir.1989)); see also Campos v. City of Baytown, 840 F.2d 1240, 1245 (5th Cir.1988) (rejecting defendants’ argument that “any time a candidate gets a majority of the minority votes he is the ‘chosen representative’ of the minority group”), cert. denied, 492 U.S. 905, 109 S.Ct. 3213, 106 L.Ed.2d 564 (1989).22
We therefore do not believe that plaintiffs are required to present evidence on white versus white elections if they do not believe that those elections are probative. If the defendants want to introduce evidence on those elections in an attempt to rebut the plaintiffs’ evidence of cohesiveness or white bloc voting, they may do so. Nevertheless, because white versus white elections tend to be less probative, there may still be cases in which “[t]he evidence of polarized voting ... *1129is so strong ... that it cannot be overcome even when all reasonable inferences are accorded to the evidence of elections involving only white candidates.” Smith v. Clinton, 687 F.Supp. at 1317.
Because the district court may, in its discretion, decide to allow the defendants to introduce evidence on which, if any, white candidates were minority-preferred,23 see infra Part VI, we will provide guidance on this inquiry. In deciding which, if any, of the white candidates were minority-preferred, the court must engage in a detailed, practical evaluation of the extent to which any particular white candidate was, as a realistic matter, the minority voters’ representative of choice. We believe that there are a number of factors to which a court can look to determine whether a white candidate may properly be considered the minority’s preferred candidate.24 One relevant consideration is the extent to which the minority community can be said to have sponsored the candidate. See Evelyn E. Shockley, Note, Voting Rights Act Section 2: Racially Polarized Voting and the Minority Community’s Representative of Choice, 89 Mich.L.Rev. 1038, 1061 (1991). In determining whether the minority community sponsored the candi-' date, the court should look to the level of minority involvement in initially advancing the particular candidate and in conducting or financing that candidate’s campaign. See id. at 1062.
Additionally, the attention which the candidate gave to the particular needs and interests of the minority community, including the extent to which the candidate campaigned in predominately minority areas or addressed predominately minority crowds and interests, may be relevant factors. Cf. S.Rep. at 29, reprinted in 1982 U.S.C.C.A.N. at 207 (setting out as one of the additional Senate Factors to be considered in evaluating the totality of the. circumstances “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”); Hall v. Holder, 955 F.2d 1563, 1568 (11th Cir.1992) (“[T]he totality of the circumstances surrounding a § 2 claim may properly be considered when determining whether plaintiffs have established the three Gingles preconditions.”), cert. granted on a different question, — U.S. -, 113 S.Ct. 758, 122 L.Ed.2d 1382 (1993).
Another relevant consideration would be the rates at which black voters turned out when a minority candidate sought office as compared to elections involving only white candidates. See Campos, 840 F.2d at 1245. Finally, bearing in mind the disincentives that may exist for minority candidates to seek office, the extent to which minority candidates have run for office and the ease or difficulty with which a minority candidate can qualify to run for office may be relevant considerations.
This inquiry, while complex, is nevertheless flexible, and we believe that "it best furthers the purposes of the Voting Rights Act. Courts, however, should be especially careful, to explain fully their reasons for finding that a particular non-minority candidate was minority-preferred and their conclusions as to the impact of that election on the overall analysis. Cf. Westwego I, 872 F.2d 1201, 1203 (5th Cir.1989).25
*1130d. Can the Court Determine a Pattern Based on the Limited Number of Elections Addressed by the Plaintiffs?
The defendants’ ultimate argument is that the plaintiffs, in addressing only seven of the twenty contested Red Clay elections, introduced insufficient evidence as to which candidates were minority-preferred to permit a finding of legally significant white bloc voting. In order for white bloc voting to be legally significant, the defendants argue, the evidence must demonstrate a pattern of white bloc voting sufficient usually to defeat the minority-preferred- candidate. The defendants submit that seven out of the twenty contested elections since 1981 is too small a sample to allow a court to' determine whether a pattern of white bloc voting exists. We disagree.
First, seven elections over a period of ten years is not such a small sample that a court would be unable to discern the presence of a pattern of white bloc voting usually defeating the minority voters’ candidate of choice. As the plaintiffs point out, the Supreme Court in Gingles upheld a finding of vote dilution based on the analysis of only three elections involving black candidates. See Gingles, 478 U.S. at 58-59, 80-82, 106 S.Ct. at 2770, 2782-83. Similarly, the Fifth Circuit has rejected a defendant’s argument that analysis of only two elections within the relevant political unit cannot adequately reflect past and present reality. Citizens for a Better Gretna v. City of Gretna, 834 F.2d 496, 502-03 (5th Cir.1987), cert. denied, 492 U.S. 905, 109 S.Ct. 3213, 106 L.Ed.2d 564 (1989).
Moreover, we believe that the defendants put undue emphasis on the need to prove the existence of a pattern. We conclude that the Supreme Court took a much more flexible approach to establishing white bloc voting than the defendants’ argument indicates. While the Court held in Gingles that “a pattern of racial bloc voting that extends over a period of time is more probative of a claim that a district experiences legally significant polarization than are the results of a single election,” 478 U.S. at 57, 106 S.Ct. at 2769 (emphasis added), the Court specifically went on to state:
The number of elections that must be studied in order to determine whether voting is polarized will vary according to pertinent circumstances. One important circumstance is the number of elections in which the minority group has sponsored candidates. Where a minority group has never been able to sponsor a candidate, courts must rely on other factors that tend to prove unequal access to the electoral process. Similarly, where a minority group has begun to sponsor candidates just recently, the" fact that statistics from only one or a few elections are available for examination does not foreclose a vote dilution claim.
Id. at 57 n. 25, 106 S.Ct. at 2769 n. 25; see also Westwego I, 872 F.2d at 1208-10; Citizens for a Better Gretna, 834 F.2d at 502 (“Gingles ... suggests flexibility in the face of sparse data.”). The Supreme Court’s language clearly indicates that an inability to offer statistical evidence on a large number of elections should not necessarily foreclose a voting rights claim.
In the present case, the plaintiffs established that the black candidate in at least six of the seven pre-1991 Red Clay elections involving a choice between a black and a white candidate was minority-preferred. Further, they produced sufficient evidence on- the seventh candidate to permit the district court’s consideration of whether that candidate was minority preferred. The plaintiffs are not, however, required to introduce evidence on elections involving white candidates only, and the defendants did not show that any white candidates in elections offering only a choice among white candidates were minority-preferred in an attempt to counter the plaintiffs’ proof. We conclude therefore that the district court had a sufficient basis from which to find the existence of legally significant white bloc voting; i.e., *1131white bloc voting sufficient usually to defeat the minority’s representative of choice.26 Accordingly, we cannot affirm the district court’s judgment based on the defendants’ alternative argument.
C. SUSTAINED PROPORTIONAL REPRESENTATION
In Gingles, a majority of the Supreme Court held that persistent proportional representation by minority-preferred candidates was presumptively inconsistent with proving a § 2 violation. Justice Brennan, in a portion of his opinion joined only by Justice White, stated:
This persistent proportional representation is inconsistent with appellees’ allegation that the ability of black voters in District 23 to elect representatives of their choice is not equal to that enjoyed by the white majority. ■
478 U.S. at 77, 106 S.Ct. at 2780 (opinion by Brennan, J.). This view was endorsed by Justice O’Connor in her concurring opinion, joined by Chief Justice Burger and Justices Powell and Rehnquist:
I agree with Justice Brennan that consistent and sustained success by candidates preferred by minority voters is presumptively inconsistent with the existence of a § 2 violation:
Id. at 102, 106 S.Ct. at 2793 (O’Connor, J., concurring in the judgment).27 The majority of the Court, however, also made clear that “[w]here multimember districting generally works to dilute the minority vote, it cannot be defended on the ground that it sporadical-: ly and serendipitously benefits minority voters.” Id. at 76, 106 S.Ct. at 2780.
The defendants argue that the election of three black candidates from 1981 through 1991 demonstrates that the black voters of Red Clay have achieved what is essentially proportional representation and that they therefore cannot prevail on a § 2 claim. Specifically, the defendants point out that black candidates won three out of the twenty contested elections between 1981 and 1991, or roughly 15%, a success rate which, given that 12.97% of the total voting age population of the Red Clay School District is black, exqeeds proportional representation.28 The defendants’ argument fails, however, in light of the district court’s finding on this issue.
The district court'thoroughly considered the extent to which' black candidates have been successful. Jenkins, 780 F.Supp. at 224-26. It noted that three black candidates — Roberts in 1981, Scotton in 1990, and *1132Greene in 1991 — had won seats on the Red Clay School Board. Id. at 224. The court further concluded, however, that the weight given to at least two of these results should be discounted (though not entirely disregarded) because of the special circumstances under which the candidates were able to prevail. Id. at 225. In particular, with regard to Roberts’ victory in 1981, the court noted that Roberts ran against five white candidates. While the court refused to disregard the result entirely, because multiple candidate races were not that unusual, it found that the court should not “weigh [the result] as heavily as the Court would weigh a black candidate’s victory in- a smaller field of candidates.”Id.29
Ultimately, the court concluded that, while the black voters of Red Clay had achieved some success, because there had also “been some under-representation of black citizens on the Red Clay Board and because some of the black candidates’ success may have occurred in unusual circumstances, the Court [would] proceed to consider the Gingles threshold factors.” Id. at 226. We read this as a finding that the black voters of Red Clay had not achieved the persistent proportional representation discussed in Gingles that would defeat a § 2 claim at the threshold and thereby obviate the need to consider the Gingles factors. We conclude that the court’s finding in this regard was not clearly erroneous as the record amply supports a finding that the black voters of Red Clay have not achieved persistent proportional representation.
Prior to the commencement of this lawsuit, only a single black had been elected to the Red Clay School Board. Although Scotton’s victory in the 1990 election was not shown to have been caused by the pendency of the lawsuit,30 see Jenkins, 780 F.Supp. at 225, the minority successes in 1990 and 1991 are sufficiently recent and, potentially, transitory to limit their probative value. Both Justices Brennan and O’Connor stated clearly that the minority’s electoral success must be consistent and sustained. See Gingles, 478 U.S. at 77, 106 S.Ct. at 2780 (opinion by Brennan, J.); id. at 102, 106 S.Ct. at 2793 (O’Connor, J., concurring in the judgment). While the success of black candidates in recent elections may presage a change in voting patterns that will lead to consistent proportional representation, on the present record there is simply no way to conclude that the results are evidence of such a change.
Additionally, the court found that both the election of Roberts in 1981 and Scotton in 1990 were accompanied by special circumstances and refused, based on those findings, to accord full weight to either election. See Jenkins, 780 F.Supp. at 225. Based on the record, we conclude that the district court’s finding in this respect was not clearly erroneous.31 Accordingly, we con-*1133elude that the judgment here cannot be affirmed on the defendants! alternative argument that the black voters of Red Clay have achieved persistent proportional representation.
D. POLITICAL COHESIVENESS OF THE BLACK RED CLAY VOTERS
As the second Gingles factor, the Supreme Court held that in order to establish a violation of § 2, “the minority group must be able to show that it is politically cohesive.” Gingles, 478 U.S. at 51, 106 S.Ct. at 2766.32 After reviewing the parties’ evidence on the political cohesiveness of the black Red Clay voters, the district court concluded:
[T]he evidence of minority voter cohesion in Red Clay has not been established by clear and convincing evidence but it was sufficient to establish-by a preponderance of the evidence that black voters of Red Clay usually are politically cohesive. The' Court, therefore, concludes that the Plaintiffs "have established the second factor of the Gingles analysis.
Jenkins, 780 F.Supp. at 231.
While the plaintiffs clearly welcome the district court’s ultimate finding of political cohesiveness, the defendants’ position on this finding, is less clear. Although the defendants, never directly challenge the district court’s finding of cohesiveness, they forcefully attack the sufficiency of all the evidence, especially the statistical evidence, upon which that finding is based. Since the defendants have challenged the foundations of the court’s finding, we believe that -they have effectively - challenged the finding itself. Therefore, we will. review the district court’s finding of political eohesiveness.
First, the district court thoroughly considered the plaintiffs’ expert analysis and the defendants! attacks on the reliability of that analysis. See Jenkins, 780 F.Supp. at 227-30. Our review satisfies us that the district court’s consideration of the statistical evidence was supported by a reasonable analysis of the expert witnesses’ conflicting opinions, id. at 230, and by reference to the existing literature in the field, id. at 229; see also Trial Transcript, Vol. F, at 48-52 (Apr. 23, 1991).
Additionally, we conclude that the district court took a properly"' flexible approach to the evidence and recognized that, in addition to examining the validity of particular pieces of evidence, it must consider the evidence as a whole. Because voters in Red Clay may vote at any polling place within the district, the plaintiffs were only able to acquire the underlying data necessary for their expert to do a statistical analysis of the voting patterns on a single Red Clay election — the 1988 election among Manning, Nuttal, and Mitchell. See supra nn. 11 & 12; Jenkins, 780 F.Supp; at 227. While the court found the plaintiffs’ regression analysis of the 1988 election probative, it also recognized that that evidence alone could not carry the plaintiffs’ burden of proof. Jenkins, 780 *1134F.Supp. at 230. The court, however, correctly found that the lack of additional evidence specific to Red Clay elections did not necessarily prevent the plaintiffs from proving their case. See id. (noting that the regression analysis pertained only to a single Red Clay election, but going on to consider other evidence and concluding that that evidence supported the results of the regression analysis of the 1988 Red Clay election).
The Supreme Court’s discussion in Gingles on the closely related issue of proving white bloc voting is instructive.33 The, Court noted that “a pattern of racial bloc, voting that extends over a period of time is more probative of a claim that a district experiences legally significant polarization than are the results of a single election.” Gingles, 478 U.S. at 57, 106 S.Ct. at 2769 (emphasis added). The Court, however, explicitly qualified this statement, explaining:
The number of elections that must be studied in order to determine whether voting is polarized will vary according to pertinent circumstances____ Where a minority group has never been able to sponsor a candidate, courts must rely on other factors that tend to prove unequal access to the electoral process. Similarly, where a minority group has begun to sponsor candidates just recently, the fact that statistics from only one or a few elections are available for examination does not foreclose a vote dilution claim.
Id. at 57 n. 25, 106 S.Ct. at 2769 n. 25. Thus the Court concluded that, even where there is no evidence of actual voting patterns from relevant elections within' the district, by adopting a flexible approach to the evidence, plaintiffs may still be able to prove the existence of a violation.
The specific reason for the páucity of data in the present case is slightly different. The underlying concern, however, remains the same, namely, that minority voters should not be forced to suffer a violation of their rights under the Act because of external circumstances that limit the availability of data specific to the challenged district if other evidence supports their claim. We therefore follow Gingles’ mandate that courts adopt a flexible approach in the face of these types of evidentiary problems. See id.; Citizens for a Better Gretna, 834 F.2d at 502 (“Gingles ... suggests flexibility in the face of sparse data.”); cf. Hall v. Holder, 955 F.2d 1563, 1568 & n. 8 (11th Cir.1992) (holding that plaintiffs may rely on the totality of the circumstances .to establish the Gingles factors because otherwise, “[e]ven where the totality of the circumstances established that a challenged system violated § 2, plaintiffs would be unable to prevail if the method of polling or the electorate’s -size and composition were such that the gathering of mathematically ‘certain’ statistical data would be impossible”), cert. granted on a different question, - U.S. -, 113 S.Ct. 1382, 122 L.Ed.2d 758 (1993).
In the present case, the district court appropriately considered whether the plaintiffs’ evidence reinforced or contradicted the direct statistical evidence of eohesiveness in the 1988 Red Clay election. Specifically, the court found that the statistical analyses of exogenous elections also tended to show political cohesiveness, even while recognizing that, by definition, exogenous elections could not perfectly mirror the population, issues, or election procedures found in Red Clay elections. See Jenkins, 780 F.Supp. at 229, 230; see also Citizens for a Better Gretna, 834 F.2d at 502 (“Although exogenous elections alone could not prove racially polarized voting in Gretna aldermanic elections, the district court properly considered them as additional evidence of bloc voting — particularly in light of the sparsity of available data.”). Similarly, while the district court recognized the short-comings of the plaintiffs’ “homogenous” precinct analysis,34 Jenkins, 780 F.Supp. at 230, rather than rejecting the *1135analysis entirely, the court concluded that even those less convincing results supported the results of the regression analysis. Id.
Finally, while the lay testimony was mixed, the court found that “in conjunction with the expert testimony it at least tends to support a finding that the black voters of Red Clay are politically cohesive.” Id. Thus, the court concluded that the evidence as a whole, while far from overwhelming, “was sufficient to establish by a preponderance of the evidence that black voters of Red Clay usually are politically cohesive.” Id. at 231. Although the defendants have made a number of cogent arguments against specific pieces of the plaintiffs’ expert evidence on political cohesiveness, their arguments are not sufficient to warrant reversing the district court’s finding as clearly erroneous. •
E. TOTALITY OF THE CIRCUMSTANCES
Finally, the defendants argue that even if we conclude that the district court’s finding as to white bloc voting was erroneous, we should affirm based on the court’s alternate finding on the totality of the circumstances:
Even if the Court had concluded that the Plaintiffs had satisfied the three Gingles factors, the Plaintiffs have not established a violation when the totality of the circumstances are considered.
Jenkins, 780 F.Supp. at 241. Notwithstanding the district court’s assertion that it would have found for the defendants even if the plaintiffs had established the three Gingles factors, we cannot affirm on this basis for three reasons.
First, the district court’s good intentions notwithstanding, merely declaring that it would assume that the plaintiffs proved the existence of all three Gingles factors did not eliminate the fact that it was evaluating the totality of the circumstances in light of the flawed plurality-win theory, i.e., under the misapprehension that, even if white voters usually voted as a bloc to submerge minority preferences in an at-large election scheme, the voting scheme as a whole was not problematic because it allowed for the possibility (though rarely realized) for minority voters to elect a representative of choice with a plurality of the vote.'
Second, the Voting Rights Act’s requirement that courts consider the totality of circumstances based “upon a searching practical evaluation of the past and present reality and on a functional view of the political process,” see Gingles, 478 U.S. at 45, 106 S.Ct. at 2764 (internal citation and quotation omitted), seems incompatible with the conclu-sory (“even if’) analysis engaged in by the district court; see also Westwego I, 872 F.2d at 1203-04 (“‘Because the resolution of a voting dilution claim requires close analysis of unusually complex factual patterns, and because the decision of such a case has the potential for serious interference with state functions, we have strictly adhered to the rule 52(a) requirements in voting dilution cases and have required district courts t.o explain with particularity their reasoning and the subsidiary factual conclusions underlying their reasoning.’ ” (quoting Velasquez v. City of Abilene, 725 F.2d 1017, 1020 (5th Cir.1984)).35
Third, in a related vein, even if we thought it was permissible for a court to engage in this type of analysis, we would be obliged to remand for a fuller explanation. As discussed, supra n. 6, it will be only the very unusual case in which the plaintiffs can establish the existence of the three Gingles factors but still have failed to establish a violation of § 2 under the totality of circumstances. In such cases, , the district court must explain -with particularity why it has concluded, under the particular facts of that case, that an electoral system that routinely results in white voters voting as a bloc to defeat the candidate of choice of a politically cohesive minority group is not violative of § 2 of the Voting Rights Act. Cf. supra n. 25 and accompanying text. While we do not *1136rule out that possibility on remand, such a full explanation was not heretofore provided by the district court.36
VI. CONCLUSION
We conclude that the district court committed reversible error in relying on the potential for the black voters of Red Clay to elect their candidates of choice even in the face of persistent white bloc voting when the record demonstrated that this potential was rarely realized. The third Gingles factor, as well as the Voting Rights Act generally, is concerned with the practical effects of a particular voting scheme on the rights of minority voters. Accordingly, whether the scheme might theoretically allow a victory for a minority-preferred candidate if factors were to coalesce properly is not the correct question. The correct question is whether the scheme as a whole has the actual effect of providing minority voters the same “opportunity [as] other members of the electorate to participate in the political process and to elect representative of their choice.” 42 U.S.C. § 1973(b).
The record in this case did not establish that the ability to elect a candidate with a plurality of the votes in fact had the effect of allowing the black voters of Red Clay to elect their candidates of choice in the face of white bloc voting. To the contrary, the record demonstrated that no candidate, black or white, had been elected with a plurality since 1981.
We also conclude that the defendants’ alternative arguments for affirming the judgment are unpersuasive. Contrary to the defendants’ arguments, there was a sufficient basis to have allowed the district court to find that there was legally significant white bloc voting. We also reject the defendants’ arguments that the plaintiffs have clearly attained persistent proportional representation, that the plaintiffs’ failed to prove political cohesiveness, and that the district court could be affirmed on the basis of its consideration of the totality of the circumstances.
Voting Rights challenges, by their nature, are often complex and difficult. The problem is greatly exacerbated in the present case because of the paucity of data available from which to analyze actual voting behavior in Red Clay elections. Given the importance of the issues at stake and the likelihood that the proper resolution of this case could be aided substantially by the introduction of additional evidence, we believe that the district court should consider the receipt of such evidence on remand.
At a minimum, we suggest that the district court on remand receive evidence on the 1991 election. We leave it to the district court to determine what if any weight to give the results of this election. In doing so, the district court may reevaluate the entire record in light of the additional evidence and may want to give particular consideration to the existence of special circumstances, in this or any prior elections, as it sees fit. While we recognize that such a procedure is not common, it is also not unique to this case, see Westwego Citizens for Better Gov’t v. City of Westwego, 906 F.2d 1042, 1043-45 (5th Cir.1990) (Westwego II) (remanding for the district court to accept evidence of a post-trial election); see also Pittsburgh Press Club v. United States, 579 F.2d 751, 755 (3d Cir.1978) (“Whether a trial court will reopen the proceedings on remand is, of course, a matter for its sound discretion.”). More importantly, we believe that it is appropriate in light of the evidentiary problems presented here and the guidance we have provided in this opinion.
The judgment of the district court will be reversed and the case remanded for further proceedings consistent with this opinion. The parties will bear their own costs.
. Although the normal term is five years, for a variety of reasons not all Board Members have actually served for five years. As a relevant example, Ronald Greene, who is one of the three black candidates elected to the Red Clay School Board, was elected only to complete the final year of a regular five year term.
. The Board member from the seventh nominating district (district G) was an appointed candidate who was white.
. A black candidate has never run unopposed or been appointed, and the 1990 election for the seat in district B was the only election in which there was no white candidate.
.In full, § 2 provides:
*1114Denial or abridgement of right to vote on account of race or color through voting qualifications or prerequisites; establishment of violation
(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 abridgment 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 1973b(f)(2) of this title, as provided in subsection (b) of this section.
(b) A violation of subsection (a) of this section is established 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) of this section 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.
42 U.S.C. § 1973 (emphasis in original).
.The Senate Report states:
If as a result of the challenged practice or structure plaintiffs do not have an equal opportunity to participate in the political processes and to elect candidates of their choice, there is a violation of this section. To establish a violation, plaintiffs could show a variety of factors, depending upon the kind of rule, practice, or procedure called into question.
Typical factors include:
1. the extent of any history of official discrimination in the state or political subdivision 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 the elections of the state or political subdivision 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 in the state or political subdivision 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.
Additional factors that in some cases have had probative value as part of plaintiffs' evidence to establish a violation are:
whether there is a significant lack of responsiveness on the part of elected officials to the *1115particularized needs of the- members of the minority group.
whether the policy underlying the state or political subdivision's use of such voting qualification, prerequisite to voting, or standard, practice or procedure is tenuous.
While these enumerated factors will often be the most relevant ones, in some cases other factors will be indicative of the alleged dilution.
The cases demonstrate, and the Committee intends that there is no requirement that any particular number of factors be proved, or that a majority of them point one way or the other.
S.Rep. No. 417, 97th Cong., 2d Sess. 28-29 (1982), reprinted in 1982 U.S.C.G.A.N. 177, 206-07 (footnotes omitted). The report further states:
The courts ordinarily have not used these factors, nor does the Committee intend them to be used, as a mechanical "point counting" device. The failure of plaintiff to establish any particular factor, is not rebuttal evidence of non-dilution. Ráther, the provision requires the court’s overall judgment, based on the totality of circumstances and guided by those relevant factors in the particular case, of whether the voting strength of minority voters is, in- the language of Fortson [v. Dorsey, 379 U.S. 433, 85 S.Ct. 498, 13 L.Ed.2d 401 (1965),] and Burns [v. Richardson, 384 U.S. 73, 86 S.Ct. 1286, 16 L.Ed.2d 376 (1966)]; "minimized or canceled out.”
Id. at 29 n. 118, reprinted in 1982 U.S.C.C.A.N. at 207 n. 118.
. This is not to suggest that there is some specific, additional proof necessary for a plaintiff to prevail. Although a court’s evaluation of the totality of the circumstances requires consideration of all the Senate Factors, the Court stated:
Under a "functional’ view of the political process mandated by § 2, the most important Senate Report factors bearing on § 2 challenges to multimember districts are the "extent to which minority group members have been elected to public office in the jurisdiction” and the "extent to which voting in the elections of the state or political subdivision is ■ racially polarized" [issues directly addressed .by Gin-gles factors 2 and 3], If present, the other factors ... are supportive of, but not essential to, a minority voter’s claim.
Gingles, 478 U.S. at 48 n. 15, 106 S.Ct. at 2765 n. 15 (emphasis in original; citation omitted). The ultimate determination that a plaintiff has proved a violation of § 2, nevertheless, must be made by the district court in light of all the circumstances. While it would be a highly unusual case in which a plaintiff successfully proved the existence of the three Gingles factors and still,failed to establish a violation, we cannot rule out that possibility entirely. See Baird v. Consolidated City of Indianapolis, 976 F.2d 357, 359 (7th Cir.1992), cert. denied, - U.S. -, 113 S.Ct. 2334, 124 L.Ed.2d 246 (1993).
. The Supreme Court has said that persistent proportional representation is presumptively inconsistent with the existence of a § 2 violation. Gingles, 478 U.S. at 77, 106 S.Ct. at 2780 (opinion by Brennan, J.); id, at 102, 106 S.Ct. at 2793 (O’Connor, J., concurring in the judgment); see also infra Part V.C.
. In two additional instances the board members were appointed rather than elected. Both of these board members were white.
. The district court considered the existence of' special circumstances in its initial evaluation of whether the black voters of Red Clay had achieved sustained proportional representation. See infra Part V. The court, however, never applied these findings, or addressed the existence or effect of special circumstances, in its analysis of white bloc voting, even though consideration-of special circumstances is specifically mandated within the definition of the third Gingles factor: "Third, the minority must be able to demonstrate that the white majority votes sufficiently as a bloc to enable it- — in the absence of special circumstances ... — usually to defeat the minority's preferred candidate.” Gingles, 478 U.S. at 51, 106 S.Ct. at 2766-67 (emphasis added).
Because the district court never addressed the existence of special circumstances in the context of white bloc voting and because we conclude that the plaintiffs' evidence would have been sufficient to support a finding in their favor on white bloc voting irrespective of the existence of special circumstances, see infra at 1131, we express no opinion on the relationship between the consideration of special circumstances in the two contexts (sustained proportional representation and white bloc voting)-or on the legal effect of a finding of special circumstances -on a court's analysis of white bloc voting. On remand, however, the district court will need to address these issues in order properly to evaluate the effect of whatever success minority-preferred candidates have achieved on the existence of legally significant white bloc voting.
. Bivariate regression analysis is used to determine the degree of relationship between two variables — here the relationship between the racial composition in each political unit (the independent variable) and the support provided a particular candidate within that political unit (the dependent variable). See Richard L. Engstrom & Michael D. McDonald, Qualitative Evidence in Vote Dilution Litigation: Political Participation and Polarized Voting, 17 Urb.Law. 369, 374 (Summ.1985). From an analysis of the regression equation (generally Y = a + bx), the plaintiffs' expert was able to estimate the per*1120centage of blacks and whites who voted for the particular candidates. .Additionally, Dr. Licht-man derived a correlation coefficient, here denoted as R, which measures "how consistently” the dependent variable varies in correspondence with the independent variable. The correlation coefficient’s values “can range from +1.0 (a perfectly consistent positive relationship) through 0.0 (no relationship) to - 1.0 (a perfectly consistent negative relationship).” Id. at 375.
. The 1988 election was the only Red Clay election involving black and white candidates that Lichtman was able to analyze through regression analysis. Because Red Clay voters may vote at any polling place within the district, the only means to determine the racial composition of the electorate at a particular polling place is by reference to the voters' place of residence, which can be obtained only from the sign-in lists for those polling places. The sign-in lists, however, are destroyed after two or three years. Since the present suit was initiated in May of 1989, the only election involving white and black candidates for which the sign-in lists were available was the 1988 election (no black candidates ran in either 1986 or 1987). See Jenkins, 780 F.Supp. at 227.
. Extreme case analysis examines the actual voting percentages received by candidates in the most heavily white or most heavily black polling places. Because actual information on the racial composition of the polling places was unavailable for. the 1983 and 1985 elections, the plaintiffs' expert had to assume that the polling places that were "homogenous” (those with at least 90% white" voters) in 1988 were also "homogenous” in 1983 and 1985. See Jenkins, 780 F.Supp. at 232.
. Dr: Lichtman's analysis revealed that the black candidate received the following percentages of the vote in homogenous (90+ %) white polling places: 1988 — 4%; 1985 — 38%; 1983 District A — 16%; and 1983 District D — 15%. See Jenkins, 780 F.Supp. at 232.
. Scatter point analysis consists of plotting the percentage of votes received by a.candidate as a function of the percentage of voters in a district that are of a particular race. Plotting the results on a graph is a simple method for determining whether the percentage of votes received by a candidate consistently increases or decreases as the percentage -of white voters in the district increases. See Jenkins, 780 F.Supp. at 229-30.
. In fact, the turnout rate for black voters in Delaware tends to be below that of white voters, as it was in the 1988 Red Clay election. See Jenkins, 780 F.Supp. at 234. The actual turnout by race in prior elections, however, could not be determined for the reasons explained, supra n. 11. ■
. These assumptions are not just conservative but are counter-factual. The record demonstrated that most, if not all of these candidates, were the preferred candidates, of the black voters, see infra Part V.B.5.b., and rather than equal turnout, the turnout rate for black voters tends in Delaware to be lower than the turnout rate for white voters, see supra n. 15. Accounting for a lower black turnout in the maximum theoretical white crossover vote calculation reduces the maximum percentage of whites who could have voted for the black candidate because while the number of votes received by the black candidate remains constant (which by assumption were all cast by white votes), the number of white votes cast increases.
. Only one post-1981 election has involved more than three candidates. •
. Because we believe that the district court's error in evaluating the significance of a candidate’s ability to win an election in Red Clay with a plurality of. the vote necessitates our reversing the court's finding on white bloc vo.ting and remanding for further consideration of the issue, we do not need to address the court's evaluation *1124of the remaining.evidence on white bloc voting, such as its evaluation of the exogenous elections or its evaluation of the extreme case analysis of the Red Clay elections. We urge the court on remand, however, to take the same flexible, realistic approach in evaluating the plaintiffs' evidence on white bloc voting that it employed in evaluating the evidence of political cohesiveness. See infra Part V.D.
. We do not reach the related, but potentially more difficult question of what evidentiary weight, if any, should be given the election of minority candidates who are not the minority voters’ candidate of choice. This issue produced a shatp split between the majority and the dissent in a recent opinion by the Fifth Circuit. See League of United Latin American Citizens, Council No. 4434 v. Clements, 986 F)2d 728 (5th Cir.1993) (addressing, among other issues, the effect of the election of black Republican judges tb the vote dilution analysis), reh’g en banc, 999 F.2d 831 (5th Cir.1993).
Two elections here could, at least, potentially, pose this issue — the 1990 election of Scotton over another black candidate and the 1991 election of Greene over a white candidate. Although no evidence was introduced on the 1991 election, the court took judicial notice of the results. Similarly, there was little evidence on whether or not Scotton was the minority voters’ representative of choice. See Jenkins, 780 F.Supp. at 224-25. Accordingly, we do not know whether the court will have to address this issue as to either election. Since the issue is not directly presented by the instant appeal and has not been briefed by the parties, we will not address it.
. The court in Sanchez affirmed the district court's finding that particular Anglo candidates were minority-preferred, in part because there was substantial, evidence that the Hispanics constituted a majority of the Democratic party, were very active in politics, and that no candidate could secure the Democratic nomination without the support of Hispanics. Thus, the court concluded, not only did the relevant Anglo "candidates receive substantial Hispanic support in terms of votes, but the evidence suggests that they were chosen by Hispanics in the first place.” Id. at 1496.
. Similarly, in their appellate brief, the defendants state:
Plaintiffs' lay evidence did tend to show some black cohesion for black candidates Roberts, Anderson and Connor in 1981, candidate Miller in 1983 (District A) and Roberts again in 1985.
. While the Voting Rights Act encompasses a broader focus than simply the election of minority candidates, if the minority community would prefer a minority candidate or believes that a minorily candidate would best represent its interests, then § 2 establishes a right on the part of those citizens for their preferred minority candidates to have an equal opportunity to be elected. As the court stated in Smith v. Clinton, the requirements of the Voting Rights Act are not met if “[c]andidates favored by blacks can win, but only if the candidates are white.” 687 F.Supp. 1310, 1318 (E.D.Ark.) (three-judge court), aff'd mem., 488 U.S. 988, 109 S.Ct. 548, 102 L.Ed.2d 576 (1988).
.It does not appear that the defendants have argued that particular white candidates in white versus white elections were minority-preferred. They do, however, make such an argument in connection with their claim that the plaintiffs have attained proportional representation. See Appellee Brief at 26-27. Contrary to their assertion therein, however, there is no indication that the court considered the success of white candidates in evaluating any portion of the plaintiffs' claim. Moreover, a review of the defendants’ argument demonstrates that their evidence amounted to little more than an attempt to show that particular white candidates may have received the majority of the black vote. As demonstrated infra, we do not think that this standing alone is sufficient to establish that a white candidate in an election offering only a choice between white candidates was minority-preferred.
. We of course do not intend this list to be exhaustive.
. In Westwego I, the Fifth Circuit said of Voting Rights cases generally:
"Because the resolution of a voting dilution claim requires close analysis of unusually complex factual patterns, and because the decision of such a case has the potential for serious *1130interference with state functions, we have strictly adhered to the rule 52(a) requirements in voting dilution cases and have required district courts to explain with particularity their reasoning and the subsidiary factual conclusions underlying their reasoning.... Perhaps in no other area of the law is as much specificity in reasoning and fact finding required....”
Id. (quoting Velasquez v. City of Abilene, 725 F.2d 1017, 1020 (5th Cir.1984)).
.There are two additional elections as to which the plaintiffs arguably had an obligation to introduce evidence concerning the minority community's preferred candidate, hut failed to do so: Scotton's victory in 1990 and Greene’s post-trial victory in 1991. As noted elsewhere, however, there are a number of outstanding questions as to the weight that could be given these elections even if the winning, candidates were deemed to have been minority-preferred. See supra n. 9 and accompanying text (noting the plaintiffs’ argument that because the court found some special circumstánces in connection with Scotton’s election it could not consider that election in evaluating white bloc voting); supra at 1119 (noting the plaintiffs’ argument that no weight may be accorded post-trial elections because the court has no evidence concerning the circumstances surrounding that election). At all events, we need not address whether the plaintiffs had an obligation to introduce evidence on these elections or the weight that should be accorded these victories. Even assuming that both Scotton and Greene were minority-preferred, there was still a sufficient basis for the district court to have found legally significant white bloc voting.
. Both opinions also recognized the possibility that plaintiffs could rebut this presumption, but stated that Gingles gave the Court no opportunity to consider what, if any, circumstances would constitute such proof. Id. at 77, 106 S.Ct. at 2780 (opinion by Brennan, J.) ("In some situations, it may he possible for § 2 plaintiffs to demonstrate that such sustained success does not accurately reflect the minority group's ability to elect its preferred representatives, but appellees have not done so here.”); id. at 102, 106 S.Ct. at 2793 (O'Connor, J., concurring in the judgment) ("Moreover, I agree that this case presents no occasion for determining what would constitute proof that such success did not accurately reflect the minority group’s actual voting strength in a challenged district 'or districts.”).
. By addressing only the contested elections, the defendants exclude the two seats that were appointed and the three instances in which a candidate ran unopposed. Including these "uncontested” elections, however, would still give black candidates a success rate of 12%, just under their percentage in the electorate.
. With regard to Scotton’s election in' 1990, the district court specifically rejected the plaintiffs’ arguments that the failure of a white candidate to run for the seat, the decision of a white candidate to run on a slate with Scotton, and Scotton’s ultimate victory were all the result of the pending lawsuit; i.e., that the majority acquiesced in (and promoted) a minority victory because they knew it would support, and was possibly essential, to their position in litigation to have a second black candidate elected. Id. However, the court temporized, also finding that Scotton’s election failed to ’’provide a measure of a black candidate’s ability to be elected when running against a white candidate.” Id. at 224-25.
Finally, it is not entirely clear what weight the district court gave Greene's post-trial victory in 1991. The court recognized that there was no evidence about the circumstances surrounding the election, but it nevertheless took judicial notice of the result as showing that “a black candidate did win an election when running against one white candidate." Id. at 225.
. The plaintiffs argue, in essence, that certain factors associated with Scotton’s victory would not have occurred but for concern over the lawsuit. In particular, they argue that the fact that this is the only election in which a white candidate did not challenge a black candidate (here there were two black candidates) and the fact that Scotton was included on a slate with a white candidate were directly attributable to concern over the lawsuit.
. The plaintiffs argue on appeal that once the district court made a finding of special circumstances, as a matter of law it had to entirely disregard the results of those elections, rather than just diminish the weight that they would be accorded. Because we affirm the district court’s finding that there was no persistent proportional representation, we need not address whether a court may discount, rather than entirely disregard, the results of elections accompanied by special circumstances.
. The Supreme Court explained the rationale behind the requirement as follows: “If the minority group is not politically cohesive, it cannot . be said that the selection of a multimember electoral structure thwarts distinctive minority group interests.” Id.
There is a relationship between political cohesiveness and the inquiry, relevant to the third Gingles factor, as to which candidates are minority-preferred. Unless the minority community is politically cohesive, with the similarity of interests and goals that that embodies, it will be impossible to identify a candidate who can be .said to be that community’s representative of choice. For these same reasons, plaintiffs' proof on political cohesiveness and which candidates are minority-preferred is often similar; both factors are commonly shown through evidence that the minority Community tends to support particular candidates. The correspondence also makes sense within the general framework of Gingles, which is directed towards determining whether racially polarized voting in conjunction with a particular electoral structure thwarts the minority community’s ability to have its interests fairly represented in the political process:
The essence of a § 2 claim is that a certain electoral law, practice, or structure interacts with social and historical conditions to cause an inequality in the opportunities enjoyed by black and white voters to elect their preferred representatives.... The theoretical basis for this type of impairment is that where minority and majority voters consistently prefer different candidates, the majority, by virtue of its numerical superiority, will regularly defeat the choices of minority voters.
Gingles, 478 U.S. at 47-48, 106 S.Ct. at 2764-65 (citation omitted).
. See supra n. 32 (discussing the close relationship between political cohesiveness and the inquiry into which candidates are. minority preferred in proving white bloc voting).
. The analyzed elections did not involve any precincts that were 90% black, Jenkins, 780 F.Supp. at 230, which is the normal standard used for homogenous precinct analysis, see Eng-strom & McDonald, supra, at 371. Rather, each of the analyzed precincts had populations that were over 70% black. Id. at 230.
. We endorse the Fifth Circuit's position on this point. Because of the complexity of the proof and the importance of the issues involved, the district courts must be particularly thorough in explaining their findings on both the ultimate issue of vote dilution and on their subsidiary findings as well.
. Because we remand based on the district court’s error in evaluating the existence of white bloc voting, we need not address the particulars of the court's analysis of the Senate Factors.