R.O. LEVY; Betty A. Etheredge, Plaintiffs-Appellees, v. LEXINGTON COUNTY, SOUTH CAROLINA, School District Three Board of Trustees, Defendant-Appellant, and Elton Wilson, in his official capacity as Chair of the Lexington County Registration and Election Commission, Defendant, 589 F.3d 708

Vacated and remanded by published opinion. Judge DUNCAN wrote the opinion, in which Judge KING and Judge KEELEY joined. OPINION DUNCAN, Circuit Judge: This appeal arises from a district court’s order holding that the school board election system of Lexington County, South Carolina, violates

Section 2 of the Voting Rights Act of 1965, 42 U.S.C. § 1973 (“Section 2”). Appellant Lexington County School District Three Board of Trustees (the “School Board”) argues that the district court erred in its analysis by considering only the elections that took place between 1994 and 2004, and by misapplying the factors set forth by the Supreme Court in Thornburg v. Gingles, 478 U.S. 30, 106 S.Ct. 2752, 92 L.Ed.2d 25 (1986). For the reasons that follow, we vacate the order and remand for further proceedings. I.

Lexington County School District Three (the “District”) is one of five school districts in Lexington County, South Carolina. It lies primarily within Lexington County, but also includes a small portion of neighboring Saluda County. The District is small: based on the 2000 Census, the total population of the District is only 12,807 persons. African-Americans constitute 28.5 percent of the population and approximately 22 percent of registered voters. The District educates roughly 2,100 students, about 40 percent of whom are minorities.

Before 1978, the School Board members were appointed by the Lexington County Board of Education. But pursuant to a 1978 referendum vote, the District established a seven-member School Board, elected in at-large, nonpartisan elections held in the last week of February of each year. School Board members were elected to staggered, four-year terms, with two seats elected in three of every four years, and the remaining seat elected in the fourth. Under this electoral system, three African-American individuals were elected to the School Board from 1978 through 1993.

In 1994, the District’s method of electing School Board members changed again. Under this new system, which continues to the present day, School Board members are elected in nonpartisan, at-large elections held during the November general election in even-numbered years. Four seats are elected in presidential years and three seats are elected in non-presidential even years. Since this change, voter turnout has quadrupled. Voter turnout among Whites, however, has outpaced voter turnout among African-Americans. On September 29, 2003, R.O. Levy, Betty A. Etheredge, and Shirley W.

Barr (collectively, “Plaintiffs”) filed an action against the School Board, its individual members, and Elton Wilson in his official capacity as Chair of the Lexington County Registration and Election Commission. Plaintiffs alleged that the current School Board election process dilutes the voting strength of African-American voters in violation of

Section 2. They explained that because of the new system, “[c]andidates for the [School Board] preferred by African Americans are usually defeated by the white majority voting as a bloc.” J.A. 18. This case went to a bench trial on December 5, 2005. After two days of testimony, the trial was recessed until March 8, 2006. At trial, Plaintiffs showed that from 1994 through 2003, nine African-American candidates ran for the School Board but none were elected. Although they conceded that one African-American candidate, Cora Lester, was elected to the School Board in 2004, Plaintiffs argued that the School Board encouraged her candidacy after this action was filed in order to thwart their

Section 2 challenge. The trial ended on March 10, 2006. The district court took this matter under advisement for three years. During that time, two School Board elections were held in 2006 and 2008. In the 2008 election, an African-American candidate named Leon Drafts was elected. Given his success, on January 21, 2009, the School Board filed a motion for leave to supplement the record with the 2008 election results. On February 19, 2009, and without first deciding the School Board’s motion to supplement, the district court issued its judgment on the merits (the “Order”).

The district court explained that, based on “information up to and including the 2004 election,” it determined that the School Board electoral system adopted in 1994 violates

Section 2. J.A. 731. This appealed followed. II.

Section 2(

a) of the Voting Rights Act of 1965 prohibits a state or its political subdivisions from imposing any electoral procedure or practice “in a manner which results in a denial or abridgement of the right of any citizen of the United States to vote on account of race or color.” 42 U.S.C. § 1973(a).

Section 2(

b) of the Act, as amended in 1982, further provides that a violation of

Section 2 occurs 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. 42 U.S.C. § 1973(b). This means that, where Whites have a majority, minority members have less opportunity than other members of the electorate to participate in the political process and to elect representatives of their choice if the White majority votes sufficiently as a bloc to defeat the minority’s preferred candidates. See Gingles, 478 U.S. at 48-51, 106 S.Ct. 2752. In Gingles, the Supreme Court explained that three preconditions must be established for any

Section 2 violation. First, “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.” Id. at 50,106 S.Ct. 2752. Second, “the minority group must be able to show that it is politically cohesive.” Id. at 51, 106 S.Ct. 2752. Third, “the minority must be able to demonstrate that the white majority votes sufficiently as a bloc to enable it ... usually to defeat the minority’s preferred candidate.” Id. (internal citation omitted). “If these preconditions are met, the court must then determine under the ‘totality of circumstances’ whether there has been a violation of

Section 2.” Lewis v. Alamance County, N.C., 99 F.3d 600, 604 (4th Cir.1996) (citing Johnson v. De Grandy, 512 U.S. 997, 1011, 114 S.Ct. 2647, 129 L.Ed.2d 775 (1994); Collins v. City of Norfolk, Va. (“Collins 7”), 816 F.2d 932, 938 (4th Cir.1987)). In this analysis, courts should consider the factors set forth in the Senate Judiciary Committee Majority Report (the “Senate Report”) accompanying the 1982 amendment to

Section 2 to determine whether, under the totality of the circumstances,

Section 2 has been violated. Collins I, 816 F.2d at 934. In the case before us, the School Board alleges several errors below. It argues that the district court erred by considering only the elections that took place between 1994 and 2004. It further argues that the district court’s application of the second and third Gingles factors was based on factual and legal errors. We discuss each argument in turn. A. We first review the School Board’s contention that the district court wrongly limited its analysis to elections that took place between 1994 and 2004.

The School Board insists that the district court should have allowed into the record the results from the 2006 and 2008 elections. As noted above, although the School Board moved to supplement the record with the 2008 election results after the case had been submitted for decision, the district court denied this motion in a paperless order almost two months after issuing the Order. In doing so, the court gave only this explanation: “[T]he interests of justice are best served by closing the record.” J.A. 14.

The School Board then moved for reconsideration, asking the court to consider both the 2006 and the 2008 elections. The court denied that motion as well. It is within the district court’s discretion to reopen a case to admit new evidence. See Dent v. Beazer Materials and Servs., Inc., 156 F.3d 523, 533 (4th Cir.1998).

When deciding whether a district court abused this discretion, we consider whether (1) the evidence sought to be introduced is especially important and probative; (2) the moving party’s explanation for failing to introduce the evidence earlier is bona fide; and (3) reopening will cause undue prejudice to the nonmov-ing party. Rivera-Flores v. Puerto Rico Tel. Co., 64 F.3d 742, 746 (1st Cir.1995); see also United States v. Abbas, 74 F.3d 506, 510-11 (4th Cir.1996) (considering similar factors in a criminal matter); Gibson v.

Mayor and Council of Wilmington, 355 F.3d 215, 229 (3d Cir.2004) (considering similar factors, “including the burden that will be placed on the parties and their witnesses, the undue prejudice that may result from admitting or not admitting the new evidence, and considerations of judicial economy”). “[I]t is generally understood that a trial court abuses its discretion if its refusal to reopen works an ‘injustice’ in the particular circumstances.” Rivera-Flores, 64 F.3d at 746 (citing Gas Ridge, Inc. v. Suburban Agric. Props., Inc., 150 F.2d 363, 366 (5th Cir.1945)).

We are persuaded that the district court’s failure to consider the 2006 and 2008 election results in the Order “work[ed] an ‘injustice’ in the particular circumstances.” Id. Those electoral results are important and probative given that an African-American candidate was elected to the School Board in 2008. In addition, the School Board had a bona fide explanation for failing to introduce this evidence earlier: such evidence was not available until after the bench trial concluded.

Finally, although we are sympathetic to Plaintiffs’ wish to have this case resolved sooner rather than later, we find that concern counterbalanced by the district court’s delay of almost three years in issuing the Order. Because of that delay, two more relevant elections occurred; on such facts, we cannot say Plaintiffs still retain the same interest in a timely disposition as they would have had before the 2006 election. Thus, reopening the case here will cause no undue prejudice to Plaintiffs, so long as they are afforded an opportunity to test and rebut the evidence presented by the School Board.

See Collins II, 883 F.2d at 1243 (although trial was originally held in 1984, subsequent elections were considered by both trial court and appellate court); Westwego Citizens for Better Gov’t v. City of Westwego, 906 F.2d 1042, 1045 (5th Cir.1990) (“[G]iv-en the long term nature and extreme costs necessarily associated with voting rights cases, it is appropriate to take into account elections occurring subsequent to trial.”). Accordingly, we vacate the district court’s Order and remand for further proceedings in which the parties will have the opportunity to present the 2006 and 2008 election results. B.

Because we are remanding the case, we will address other issues raised on appeal that are likely to recur. See Elm Grove Coal Co. v. Dir., O.W.C.P., 480 F.3d 278, 299 n. 20 (4th Cir.2007) (“We choose to address this discovery issue because it is likely to arise on remand.”). Specifically, we review the School Board’s arguments that the district court erred in its application of the second and third Gingles factors. 1. We first consider whether the district court erred in its analysis of the third Gingles factor.

Under this factor, the court must determine whether the majority votes as a bloc to enable it to usually defeat the minority’s preferred candidate. Gingles, 478 U.S. at 51, 106 S.Ct. 2752. To do so, the court must first identify those individuals who constitute minority-preferred candidates of choice, and then analyze whether those candidates are usually defeated by majority White bloc voting. Collins II, 883 F.2d at 1237; see Lewis, 99 F.3d at 611-14 (laying out the proper criteria for identifying minority-preferred candidates of choice).

The School Board argues that the district court made several errors in its identification of the minority-preferred candidates of choice. We address each alleged error in turn. a. First, the School Board challenges the method that the district court employed to determine whether a candidate who received less support from minority voters than an unsuccessful first choice may nevertheless be deemed a minority-preferred candidate of choice in a multi-seat election.

Pursuant to Lewis, if the unsuccessful candidate who was the first choice among minority voters did not receive a “significantly higher percentage” of the minority community’s support than did other candidates who also received a majority among minority voters, ... then the latter should also be viewed by the district court as minority-preferred candidates. 99 F.3d at 612. Lewis, however, left the term “significantly higher” undefined. In the Order, the district court defined this term based on Collins II, 883 F.2d at 1238.

The court read Collins II as finding that two successful candidates (with 58.2 percent and 56.5 percent of the African-American vote) were not minority-preferred candidates of choice because they received significantly less minority support than the unsuccessful top minority vote-getter, who received 73.4 percent of the African-American vote.

Noting that the differences between the unsuccessful candidate and the successful candidates were 15.2 and 16.9 percentage points respectively, the district court proceeded to classify any candidate in the present case 15 percentage points or more below the unsuccessful top vote-getter as having received significantly less minority support than the unsuccessful top minority vote-getter. We disagree with the district court’s reasoning. Although a 15-percent-age-point differential may be the appropriate standard, it is not necessarily so.

Interpreting analogous language, Lewis explained that “[t]he level of support that may properly be deemed ‘substantial’ will vary, of course, depending on the number of candidates on the ballot and the number of seats to be filled.” 99 F.3d at 614 n. 11. As such, to determine whether a candidate who received less support from minority voters than an unsuccessful first choice may be deemed a minority-preferred candidate of choice in a multi-seat election, a district court should first consider the number of candidates on the ballot and the number of seats to be filled.

Only then should the court make a determination of whether the unsuccessful top minority vote-getter received a significantly higher percentage of the minority community’s support than did other candidates. The district court, however, did not undertake this type of analysis. Thus, we agree with the School Board that the district court’s candidate-of-choice analysis rests on an improper understanding of the term “significantly higher”. b.

Next, the School Board argues that the district court, in applying the third Gingles factor, erred by assuming that no minority-preferred candidates of choice can come from an election in which no candidate received a majority of the minority vote. The district court reached this conclusion after determining that neither Lewis nor Collins II establishes a construct for analyzing elections in which no candidate received a majority of the minority vote.

Accordingly, the district court concluded that the 2002 election was devoid of minority-preferred candidates of choice because “no candidate received a majority of the black vote” during that election. J.A. 754-55. The district court’s reasoning assumes that a person without support from over 50 percent of minority voters cannot be deemed a minority-preferred candidate. Lewis, however, said the opposite: “[W]e do not believe that the mere failure to achieve a threshold of 50% in a multi-candidate election necessarily means that a candidate cannot be viewed as a black-preferred candidate.” 99 F.3d at 613 n. 10.

In that case, we considered the treatment of second- and third-place finishers behind a successful, minority-preferred candidate of choice who had received more than 50 percent of the minority vote. We ultimately found that in a multi-seat election, [candidates who receive less than 50% of the minority vote, but who would have been elected had the election been held only among black voters, are presumed also to be minority-preferred candidates, although an individualized assessment should be made in order to confirm that such a candidate may appropriately be so considered. Id. at 614.

If such candidates can be considered minority-preferred candidates of choice in an election in which the top vote-getter received more than 50 percent of the vote, we see no reason why a similarly popular candidate — one that would have been elected had the election been held only among African-American voters— should not also be considered a minority-preferred candidate of choice in an election in which no candidate received 50 percent or more of the minority vote. See, e.g., Ruiz v. City of Santa Maria, 160 F.3d 543, 552 (9th Cir.1998).

Thus, we find that in elections in which no candidate receives a majority of the minority vote, a candidate may still be labeled a minority-preferred candidate of choice if that candidate would have been elected had the election been held only among minority voters, so long as an individualized assessment of that candidate supports that conclusion. c. Finally, the School Board contends the district court erred in adopting Plaintiffs’ expert’s report over their own expert’s report to ascertain the minority-preferred candidates of choice.

To determine which candidates were minority preferred, Plaintiffs’ expert, Dr. John Ruoff, analyzed elections for the School Board from 1986 to 2004 using three statistical methods of analysis: homogenous precinct, ecological regression, and ecological inference. The School Board’s expert, Dr. David Epstein, analyzed the same data using the ecological inference method. Ruoff concluded that there were eight minority-preferred candidates of choice for the School Board between 1994 and 2004. Of those eight, one was elected to the School Board.

By contrast, Epstein concluded that there were seventeen minority-preferred candidates of choice during that period. Of those seventeen, six were elected to the School Board. The district court favored Ruoff s results over Epstein’s, but its reasons for this selection are unclear. The district court explained its choice, in total, as follows: The court finds more reliable Dr. Ruoff s use of three statistical methods of analysis — homogenous precinct (HP), ecological regression (ER), and ecological inference (ED — as compared to Dr. Epstein’s analysis utilizing only the El method of analysis.

The court also finds persuasive Dr. Ruoff s explanations regarding the Ridge Road precinct. The court [thus] adopts Dr. Ruoffs statistical analysis. J.A. 755-56 (footnote call number added). We recognize that our role in reviewing the district court’s reliance on Ruoffs expert testimony is limited. In re Corrugated Container Antitrust Litig., 659 F.2d 1322, 1326 (5th Cir.1981); see also Collins I, 816 F.2d at 936 (“It is within the trial court’s discretion to find that methodological flaws undercut the probative value of a study so deeply as to render it inadmissible.”); Ludlow Corp. v. Textile Rubber & Chem.

Co., Inc., 636 F.2d 1057, 1060 (5th Cir.1981) (“The competency and qualifications required of expert witnesses is a matter committed to the broad discretion of the trial judge.... When it sits without a jury, the trial court’s discretion is also the sole determinant of the weight to be given expert testimony.”) (internal citations omitted). Nevertheless, there are problems with Ruoffs data that do not appear to have been considered by the district court. For example, Ruoffs choice to exclude the Ridge Road precinct from his analysis appears suspect.

Ruoff used what he termed the “interocular trauma” or “eyeball” test to identify outliers, although other statistical methods of analysis of outliers were available. In essence, he looked at the scatter plots he prepared and assessed whether they offended his eyesight. Ruoff admitted that other precincts qualified under his “eyeball” test as outliers, but he explained that he chose not to take those other precincts out because it would be nearly impossible to “add [them] back in” as he did with Ridge Road. J.A. 386. This is not acceptable.

The district court failed to acknowledge or explain what appear, on review, to be the limitations of Ruoffs report. Although “absolute perfection on the base statistical data is not to be expected, a trial court should not ignore the imperfections of the data used nor the limitations of statistical analysis.” Overton v. City of Austin, 871 F.2d 529, 539 (5th Cir.1989). Here, the district court gave no explanation why one expert’s data was preferred over the other’s, aside from the fact that one expert used more methods of analysis than the other.

The district court’s failure to articulate why it preferred Ruoffs data over Epstein’s does not allow for a proper review of the district court’s findings. See Clay v. Bd. of Educ. of St. Louis, 90 F.3d 1357, 1362 (8th Cir.1996) (finding that the district court did not clearly err in accepting one expert’s analysis over the other where the court “cited several reasons for discounting the Plaintiffs’ expert analysis”).

For these reasons, while we do not necessarily agree with the School Board that the district court erred in adopting Ruoffs report, we also cannot say with certainty that the district court properly adopted Ruoffs expert report in its candidate-of-choice analysis. 2. We next consider the School Board’s contention that the district court’s approach to the second Gingles factor rests on error.

Under the second Gingles factor, to determine whether a minority group is politically cohesive, a court must ascertain whether “a significant number of minority group members usually vote for the same candidates.” Gingles, 478 U.S. at 56, 106 S.Ct. 2752. The School Board argues that the district court erred in its determination of political cohesiveness for two reasons. First, the School Board contends that the district court erred in its political cohesiveness analysis by only considering minority support for candidates which the court had already classified as minority-preferred candidates of choice. We agree.

Given the district court’s assumption that a minority-preferred candidate of choice needed to have over 50 percent of minority support, the court’s approach logically guaranteed a finding of cohesiveness. Second, the School Board contends that the district court wrongly equated political cohesiveness with racial polarization, i.e., the tendency for voters to prefer candidates of their same race, by emphasizing that “[e]ach of these candidates is black.” J.A. 739.

We note that, although the existence of racially polarized voting can establish that the minority group is politically cohesive, see Collins I, 816 F.2d at 935-36; Cottier v. City of Martin, 445 F.3d 1113, 1118 (8th Cir.2006); Cousin v. Sundquist, 145 F.3d 818, 826 (6th Cir.1998); Sanchez v. Colorado, 97 F.3d 1303, 1312 & n. 14 (10th Cir.1996); Westwego Citizens for Better Gov’t v. City of Westwego, 946 F.2d 1109, 1118 (5th Cir.1991), the two issues are quite distinct.

As the Fifth Circuit has keenly observed, “[t]hat a group’s voting behavior is racially polarized indicates that the group prefers candidates of a particular race. Political cohesion, on the other hand, implies that the group generally unites behind a single political ‘platform’ of common goals and common means by which to achieve them.” Monroe v. City of Woodville, Miss., 881 F.2d 1327, 1332 (5th Cir.1989). Thus, minority voters may be racially polarized but still lack political cohesion if their votes are split among several different minority candidates for the same office. Id.

The district court does not appear to have recognized this distinction. Both issues warrant consideration on remand. III. For the foregoing reasons, we vacate the district court’s February 19, 2009 order, and remand for further proceedings. The outstanding motions are denied as moot. VACATED AND REMANDED . Under at-large electoral systems, "a group of candidates is placed on the ballot to fill a designated number of seats, and voters from everywhere within the jurisdiction may vote to fill those seats.” Steven J.

Mulroy, The Way Out: A Legal Standard for Imposing Alternative Electoral Systems as Voting Rights Remedies, 33 Harv. C.R.-C.L. L.Rev. 333, 336 (1998). "The traditional winner-take-all form of at-large elections allows each voter to cast only one vote for each candidate, up to the number of available seats.” Id. . Barr was dismissed as a party on September 28, 2004. . Later, on April 17, 2009, the district court denied the School Board's motion to supplement.

The court also denied a subsequent motion for reconsideration. .Those factors are: 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; and 7. the extent to which members of the minority group have been elected to public office in the jurisdiction. [8.] whether there is a significant lack of responsiveness on the part of elected officials to the particularized needs of the members of the minority group[; and] [9.] 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.

S.Rep. No. 417, 97th Cong., 2d Sess. 28-29, reprinted in 1982 U.S.Code Cong. & Admin. News 177, 206-07 (footnote call numbers omitted) (cited in Collins v. City of Norfolk, Va. ("Collins II"), 883 F.2d 1232, 1236 n. 3 (4th Cir.1989)). . We recognize that Collins v. City of Norfolk, Va., 768 F.2d 572 (4th Cir.1985) was technically the first Collins case. Lewis, however, designated Collins v. City of Norfolk, Va., 816 F.2d 932 (4th Cir.1987) as “Collins I" and Collins v. City of Norfolk, Va., 883 F.2d 1232 (4th Cir.1989) as “Collins II” 99 F.3d at 604, 611.

Thus, we adopt those designations here for the sake of consistency and clarity. . At a hearing held on September 9, 2004, the district court held that Plaintiffs had established the first Gingles factor, so its analysis in the Order focused solely on the second and third factors. .

The School Board also contends that the district court, in its totality of the circumstances analysis, (1) failed to consider unre-butted evidence in the record showing that minority candidates lost for reasons other than White bloc voting; (2) improperly discounted the post-litigation election of Cora Lester in 2004 due to "special circumstances”; and (3) failed to consider substantial record evidence as to the Senate Report factors that supports the rejection of Plaintiffs’ claim.

Because we remand on other bases, we do not find it necessary to address these arguments. .The School Board did not seek to reopen the case but instead moved to supplement the record. To allow one side to supplement the record without allowing the opposing party the opportunity to contest the admissibility, reliability, and accuracy of the new evidence, and to offer rebuttal evidence, would implicate due process concerns. See, e.g., Greene v. McElroy, 360 U.S. 474, 497 & n. 25, 79 S.Ct. 1400, 3 L.Ed.2d 1377 (1959).

However, the district court could have granted the School Board’s request on the condition that Plaintiffs would be allowed to test and rebut the new evidence. Thus, the School Board’s request did not necessarily offend due process. . Although the School Board only moved to introduce the results of the 2008 election, we agree with the district court’s holding that “the 2008 school board election results would be of little benefit to the court absent the inclusion of the relevant information from both the 2006 and the 2008 elections in the expert analysis.” J.A. 14. .

The School Board also argues that the district court should have considered the results from the 1980-1993 elections. Although we remand for further consideration of the 2006 and 2008 elections, the district court is not required to consider the pre-1994 elections. The complaint challenged only the present voting system, and "[ejections held under a significantly different electoral structure” are less probative in a

Section 2 analysis. See NAACP v. City of Niagara Falls, N.Y., 65 F.3d 1002, 1012 (2d Cir.1995). The present electoral system here is significantly different from the pre-1994 system. Not only was the election moved from February to November, resulting in voter turnout quadrupling, but the number of candidates elected at each election also changed.

Still, the School Board makes much of the fact that the district court only considered six elections in its analysis, and thus argues that, pursuant to our holding in Lewis and the Second Circuit’s in Niagara Falls, the district court should have considered the results from the 1980-1993 elections. Neither Lewis nor Niagara Falls specifies how many elections a court must review to analyze a

Section 2 violation. Lewis declined to set a bright-line standard and chose to "leave to another day the question of precisely how many elections must be considered in order for a district court’s conclusions to be adequately supported.” 99 F.3d at 611. Admittedly, Lewis implied that a majority of elections should be considered, id. at 608, and Niagara Falls did characterize five elections as a slim record, 65 F.3d at 1012. However, the district court here considered all elections since 1994, in compliance with Lewis.

Also, although six elections may be a slim record, Niagara Falls does not require a district court to consider exogenous elections if they were held under significantly different electoral systems. Thus, we are not persuaded the district court committed legal error by excluding from its

Section 2 analysis the 1980-1993 elections. . We address the third Gingles factor before the second because doing so makes for ease of analysis in this case. The factors are conjunctive, but there is no requirement that we analyze them in alphanumerical order. See, e.g., Voinovich v. Quilter, 507 U.S. 146, 158, 113 S.Ct. 1149, 122 L.Ed.2d 500 (1993) (disposing of the plaintiffs'

Section 2 claim without analyzing the first Gingles precondition, since the Court found that the plaintiffs had failed to meet the third pre-condition). . We recognize that in Collins II, just as here, voters filled either three or four seats each election, depending on the year. 883 F.2d at 1239. However, the number of candidates that sought the open seats in Collins II differs from that in the case at hand.

Thus, while a 15-percentage-point differential was the appropriate standard in Collins II, and may be so here, we cannot conclude that without first considering the number of candidates on the ballot and the number of seats to be filled. . Under this construct, for example, William Berry and Stephen Padgett, the top minority vote-getters in 2002 with 48 percent of the African-American vote, could potentially be classified as minority-preferred candidates of choice if an individualized assessment supports that conclusion. .

In those circumstances, the district court should ensure that the candidate can be fairly considered a representative of the minority community. In addition to the bare statistics, the district court may consider testimony from political observers and the candidates themselves to determine whether those candidates may be labeled minority-preferred candidates of choice. See Collins II, 883 F.2d at 1238. . The School Board suggests that the district court preferred Ruoff's data over Epstein’s because Epstein “changed his definition of candidate of choice significantly during the course of litigation.” J.A. 752.

This argument is without merit. After finding Epstein had changed his definition of candidate of choice several times, the district court rejected Epstein’s definition for "minority-preferred candidate of choice,” not his data altogether. .Ruoff explained that Ridge Road, a particularly large, overwhelmingly White precinct, was an outlier and thus required special treatment. At the time of the 2004 election, Ridge Road contained 1,087 voters, only 22 (2 percent) of whom were African-American.

Ruoff opined that, since the Ridge Road precinct overwhelmingly consists of White voters, including this precinct in the regression analysis would have overestimated the percent of African-American votes in favor of minority-preferred candidates. Rather than simply discarding the votes from the Ridge Road precinct, Ruoff added them back into the analysis after separately completing the ecological regression and ecological inference estimates for the remaining precincts. .

This is of particular concern given that had the district court preferred Epstein's data over Ruoffs, the results of this case would likely have been very different. . For example, the district court determined that no minority-preferred candidates of choice existed in the 2002 election because no candidate during that election achieved more than 50 percent of the minority vote. Thus, in its political cohesiveness analysis, the district court did not consider the 2002 electoral results.

Yet, the 2002 results presumably demonstrate a lack of political cohesiveness— the very factor the district court sought to establish under the second Gingles factor. . It is not necessary to prove racial polarization in order to find political cohesiveness. As this court noted in Lewis, "to our knowledge, no court has held that a white candidate cannot, as a matter of law, be a minority-preferred candidate.” 99 F.3d at 609. .

In addition, we note that in reviewing the level of support received by minority-preferred candidates of choice in the various elections, the district court failed to note the level of support African-Americans gave to other candidates during those elections. See Monroe, 881 F.2d at 1332 ("Crossover voting by a minority group does not preclude a finding of political cohesiveness per se, but its presence is relevant.”).