William L. Clay, Jr. and John F. Bass (Plaintiffs) brought suit against the Board of Education of the City of St. Louis (Board of Education), alleging violation of § 2 of the Voting Rights Act. 42 U.S.C. §§ 1973-1973p. Plaintiffs contend that the at-large voting system used to elect members to the Board of Education operates to dilute African-American voting power. The district court,1 finding that they failed to show that the white majority votes sufficiently as a bloc to enable it to usually defeat the AfricanAmerican preferred candidate, dismissed their suit. The Plaintiffs appeal and we affirm, holding that they failed to establish one of the necessary preconditions to a § 2 claim.

The St. Louis School District (School District) is the largest in Missouri. In 1991, the seventy-two public schools within the School District educated over 40,000 students. School Data Section, Department of Elementary & Secondary Education, Missouri School Directory 1991-92, 163 (1992).

The School District is governed by the Board of Education, which consists of twelve members elected for staggered six-year terms. In every odd-numbered year, four seats on the Board are contested in at-large elections. See Mo.Rev.Stat. § 162.581 (1991). Each eligible city resident has four votes which can be allocated, one to a candidate, to four different candidates. The voter also has the option to cast fewer than four votes (the "bullet voting" option), thereby marginally enhancing the weight of the votes that the voter does cast. The four candidates receiving the most votes from throughout the city are elected.

Since 1967, African-American candidates have consistently held seats on the Board of Education.2 African-American candidates have won twenty-one of the sixty-six (31%) Board seats available in elections between 1967 and 1995. Of the thirty-eight seats contested from 1977 to 1995, eleven (28.9%) were filled by African-American candidates and another eleven (28.9%) were filled by white candidates who received enough African-American votes to have won if only African-American voters participated. Resp't Br. at A-6.3

Currently, the Board of Education consists of five African-American members and seven white members, a ratio that corresponds closely with the actual percentage of African-American and white voters in the city. According to the 1990 Census, African-Americans comprise 42.7% of St. Louis's voting-age population of 210,000.

On April 1, 1991, Plaintiffs brought suit against the Board of Education, claiming that the at-large electoral system used to elect Board members violates § 2 of the Voting Rights Act, 42 U.S.C. § 1973,4 by denying African-American voters an equal opportunity to effectively participate in the political process. Plaintiffs claimed that the at-large electoral system, in combination with bloc voting patterns and election practices, operates to dilute the voting strength of African-Americans.5 Plaintiffs sought declaratory relief and an injunction requiring that "districts be fairly drawn for each of the twelve positions on the Board of Education for the City of St. Louis." Compl. at 7.

At the bench trial, both parties offered expert testimony analyzing past Board of Education elections. The Plaintiffs relied on Dr. Kenneth Warren, who used a hybrid homogenous analysis to explain the Board of Education and exogenous election results.6 Under this approach, Warren assumed that ward clusters with at least 90% African-American populations were entirely African-American and ward clusters with 90% white populations were entirely white.7 From the election results of these largely homogenous areas, he sought to extrapolate a racial voting pattern. Warren's analysis emphasized the ability of African-American candidates to be elected to the Board of Education, relying on the implicit assumption that African-American candidates were the preferred candidates of African-American voters.8

Dr. Ronald Weber testified as the expert for the Board of Education. At the outset, Weber defined the African-American preferred candidates to be the four candidates who received the most African-American votes in each contested election. Weber employed two different statistical methods to study Board of Education election voting patterns. First, he used homogeneous precinct analysis, which differed from hybrid homogenous analysis only in that the voting areas studied were smaller. Second, he applied bivariate regression analysis, plotting the percentage of the vote garnered by a particular candidate against the racial composition of the precinct to determine if a pattern of political support emerges across a cross section of different racial compositions. Mem. Op. at 7-8.

Based on the results of both methods, Weber testified that the white majority had not voted sufficiently as a bloc to enable it to usually defeat the minority preferred candidate. Rather, according to Weber's results, the minority preferred candidate was elected in most instances. He demonstrated that, overall, St. Louis voters elected African-American preferred candidates 57.9% of the time.

In light of the evidence presented, the district court concluded that the Plaintiffs had failed to prove that the majority voted sufficiently as a bloc to usually defeat the minority preferred candidate. Specifically, the court found that the Plaintiffs had failed to identify the minority preferred candidate or offer a legitimate method for making such an identification. In the absence of a reasonable alternative, the court accepted the School District's definition of minority preferred candidate. The district court also found that, due to flaws in Warren's statistical approach, Weber's analysis provided a sounder explanation of the Board of Education elections. Based on these two crucial findings, the district court found that minority preferred candidates were elected 57.9% of the time and, therefore, the white voting bloc did not tend to thwart the minority preferred candidate.

Plaintiffs appeal, raising six challenges to the district court's findings of fact. The central argument asserted by Plaintiffs is that the district court erred in defining the minority preferred candidates to be those candidates who receive the most minority votes. In addition, Plaintiffs argue that the district court erred when it found that the School District's bivariate regression analysis of election results provided a more accurate description of racial voting patterns.

To establish a § 2 violation, the minority group must demonstrate that, based on the totality of circumstances, they "have less of an opportunity to participate in the political process and to elect representatives of their choice." 42 U.S.C. § 1973(b). The minority group must initially show that three preconditions exist. See Thornburg v. Gingles, 478 U.S. 30, 50, 106 S. Ct. 2752, 2766, 92 L. Ed. 2d 25 (1986). 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. Second, the minority group must be able to show that it is politically cohesive. Third, the minority group must be able to demonstrate that the majority votes sufficiently as a bloc to enable it--in the absence of special circumstances--to defeat the minority preferred candidate. Id. at 50-51, 106 S.Ct. at 2766-67.

We review the district court's factual findings for clear error and the legal conclusions it draws from these factual findings de novo. Harvell v. Blytheville Sch. Dist. # 5, 71 F.3d 1382, 1386 (8th Cir.1995), cert. denied, --- U.S. ----, 116 S. Ct. 1876, 135 L. Ed. 2d 172 (1996). In order to prove that the third Gingles precondition exists, the plaintiffs must identify the minority preferred candidates and show that, due to majority bloc voting, they usually are not elected. See Gingles, 478 U.S. at 55-56, 106 S.Ct. at 2768-69. Plaintiffs did neither.

The Plaintiffs offered, by implication, a definition of "minority preferred candidate" based solely on the candidate's race. As a matter of law, such a definition is untenable and must be rejected in favor of the alternative offered by the Board of Education.

There is no blanket definition of "minority preferred candidate." Rather, the plaintiffs must prove, on an election-by-election basis, which candidates are minority-preferred. Harvell, 71 F.3d at 1386 (8th Cir.1995) (citing Jenkins v. Red Clay Consol. Sch. Dist. Bd. of Educ., 4 F.3d 1103, 1126 (3d Cir.1993), cert. denied, --- U.S. ----, 114 S. Ct. 2779, 129 L. Ed. 2d 891 (1994)).

In explaining his statistical analysis, the Plaintiffs' expert did not explicitly identify who the minority's candidates of choice were or what methodology should be used to make such a determination. Mem. Op. at 21. Nor did the Plaintiffs otherwise offer evidence on who the minority preferred candidates are. Rather, the Plaintiffs relied on the presumption that African-American voters preferred African-American candidates.

In Harvell, supra, we specifically rejected the presumption suggested by Plaintiffs, that only African-American candidates can be preferred by African-American voters. Inferences based solely on race are insufficient to establish which candidate is minority-preferred.9 The notion that a minority candidate is the minority preferred candidate simply because of that candidate's race offends the principles of equal protection. Harvell, 71 F.3d at 1386. As Justice Brennan stated in Gingles, "under § 2, it is the status of the candidate as the chosen representative of a particular racial group, not the race of the candidate, that matters." 478 U.S. at 68, 106 S.Ct. at 2775 (plurality opinion).

In contrast, the Board of Education's expert offered a definition of the minority preferred candidate and identified, in each election, who those candidates were. Since four seats on the Board of Education are contested in every Board election, he designated the four candidates receiving the highest number of African-American votes as the "minority preferred candidates."10

The district court properly adopted the School District's definition of minority preferred candidate. Absent a showing that minority preferred candidates are, for some reason, excluded from the ballot, it is a near tautological principle that the minority preferred candidate "should generally be one able to receive [minority] votes." Harvell, 71 F.3d at 1387.11

The fact that Plaintiffs failed to provide an adequate description of the minority preferred candidates does not end the analysis. It may be that the Board of Education's definition, when applied to the Plaintiffs' statistical analysis, demonstrates the conditions necessary to satisfy the third Gingles precondition. On appeal, the Plaintiffs argue that the court erred in accepting the analysis of the Board of Education over the analysis propounded by their expert, Warren. In light of the more thorough analysis presented by the Board of Education, we conclude that the district court was not clearly erroneous in accepting the Board of Education's statistical analysis.

The district court cited several reasons for discounting the Plaintiffs' expert analysis. First, in performing his hybrid homogeneous analysis, Warren relied on ward clusters, but failed to define the term beyond the fact that a ward cluster is larger than both precincts and wards. Given its relatively large size, ward clusters are not as conducive to homogeneous analysis as the smaller precincts used by Weber in his homogeneous analysis. Second, much of the Plaintiffs' expert's analysis relies on exogenous elections, which should be used only to supplement the analysis of the specific election at issue. Third, the Plaintiffs' expert was retained on a contingency basis, and would be compensated only if Clay prevailed. We find the rationale of the district court compelling and perceive no reason for reversing its finding.

Based on the School District's definition of minority preferred candidate and its statistical analysis, minority preferred candidates realized a substantial degree of election success. Overall, 57.9% of minority preferred candidates were elected to the Board of Education. In addition, the minority preferred candidate was elected 80% of the time when African-American voters voted cohesively. Therefore, Plaintiffs failed to establish the third Gingles precondition because they did not show that the white voting bloc did tend to thwart the minority preferred candidate. In light of our conclusion that Plaintiffs failed to establish a necessary precondition to their § 2 claim, we do not reach the other issues raised by Plaintiffs on appeal.

For the above stated reason, we hold that the Plaintiffs failed to establish the existence of the third Gingles precondition and, therefore, cannot prevail in their § 2 vote dilution claim. We affirm the judgment of the district court.