Plaintiffs Rose Johnson, et al., a group of African-American citizens of Gainesville, Georgia, appeal the district court's judgment in favor of Defendants Robert Hamrick, et al., members of the Gainesville City Commission, on the plaintiffs' civil rights challenge to Gainesville's atlarge method of electing city council members. See Johnson v. Hamrick, 155 F. Supp. 2d 1355 (N.D.Ga.2001). This appeal marks the third time that this case has come before us. See Johnson v. Hamrick, No. 94-9203 (11th Cir. Jan. 25, 1996) (unpublished opinion) ("Hamrick I"); Johnson v. Hamrick, 196 F.3d 1216 (11th Cir.1999) ("Hamrick II"). The central question since the beginning of the lawsuit has been whether the plaintiffs can show vote dilution in violation of Section 2 of the Voting Rights Act, 42 U.S.C. § 1973, under the test established by the Supreme Court in Thornburg v. Gingles, 478 U.S. 30, 106 S. Ct. 2752, 92 L. Ed. 2d 25 (1986).

In its most recent order, on remand from Hamrick II, the district court employed the Gingles test and specifically found that Gainesville's electoral system does not dilute minority votes. The district court also concluded that the plaintiffs could not show the discriminatory intent necessary to support their claim that Gainesville's system violates the Fourteenth and Fifteenth Amendments of the United States Constitution. On appeal, the plaintiffs do not pursue their constitutional claim, but contend that the district court erred in finding no Section 2 violation under Gingles. After thorough review, we conclude that the district court did not clearly err in finding no vote dilution, and we therefore affirm.

The City of Gainesville, Georgia is governed by a city council comprised of five members. Each member resides in one of the city's five geographically designated wards, which serve as electoral districts. Although an individual must reside in a ward in order to occupy that ward's council seat, voting is conducted on an at-large basis, meaning that all registered voters in the city can vote for every council seat. The office of the mayor rotates among the city council members and is not filled directly by popular election. Administratively, the city is run by a manager who is appointed by and serves at the pleasure of the city council.

Figures from the 1990 census show that Gainesville had a total population of 17,885, 23.5 percent of which was African-American and 68.8 percent of which was white. Other racial groups comprised 8 percent of the population. Of the city's 13,575 votingage residents in 1990, 20.24 percent were African-American and 74.42 percent were white. By 2000, Gainesville's population had grown to 25,578. According to the 2000 census, only 15.7 percent of the city's residents are African-American, while 65.2 percent are white and 19.1 percent identify with another racial group. Census results show that 14.2 percent of Gainesville's voting-age residents are African-American, while 69.5 percent are white. Currently, one of Gainesville's five city council members, Myrtle Figueras of Ward 3, is African-American. Ward 3 has been represented by an African-American council member since 1978.

This lawsuit began its long life in 1991, when the plaintiffs filed a claim alleging that Gainesville's at-large electoral system violates Section 2 and the Fourteenth and Fifteenth Amendments by diluting minority voting power. Since that time, the district court has issued three separate opinions and this Court has considered two appeals. Throughout the litigation, the plaintiffs have sought to show the four factors essential to proving a claim of vote dilution. As discussed in more detail infra, these factors are (1) a sufficiently compact and numerous minority community, (2) the existence of minority political cohesion or bloc voting, (3) majority bloc voting, and (4) that the totality of the circumstances indicates vote dilution. See Gingles, 478 U.S. at 49-51, 79, 106 S. Ct. at 2765-67, 2781.

The district court issued its first decision in 1994, when it held that the plaintiffs had not satisfied the third prong of Gingles because they were unable to demonstrate that the white majority in Gainesville voted as a bloc to defeat the African-American minority's preferred candidates. Specifically, the district court found that candidates preferred by African-Americans had achieved considerable success in the eight city council, or "endogenous," elections reviewed, as well as in various "exogenous" elections, in which Gainesville citizens voted for county, state, and national officials. The district court did not address the constitutional claims in its 1994 order. After the plaintiffs appealed, however, this Court remanded the matter for consideration of the constitutional questions. See Hamrick I.

On remand, the district court reopened the case to receive evidence regarding elections that had taken place since the 1994 order. Based on the new evidence, in 1998 the district court reversed its earlier determination and concluded that the plaintiffs had indeed proven a Section 2 violation under the Gingles test. Taking the recent elections into account, the court found that the white preferred candidate had prevailed over the African-American candidate of choice in five of nine endogenous elections between 1990 and 1998. The district court explained that white voters were able to elect the candidate of their choice in all eight of the elections in which they had expressed a preference, while African-American voters were able to do so in only three campaigns. In addition to holding that the three-part Gingles test had been satisfied, the district court concluded that the totality of the circumstances weighed in favor of finding a Section 2 violation. Finally, the district court rejected the defendants' argument that Section 2 was unconstitutional.

In Hamrick II, this Court vacated the district court's 1998 order, concluding that the district court had made "insufficient findings" to allow proper review of the defendants' arguments. Stating that it lacked "a sufficiently detailed explanation of the district court's basis for weighing the elections" under the third prong of Gingles, id. at 1222-23, the panel questioned the basis for the district court's conclusion that the two races consistently preferred different candidates merely because they did so in five of nine endogenous elections. As a result, this Court remanded the case to the district court for "additional and more specific findings of fact and conclusions of law." Id. at 1224.

On remand, the district court reopened its evidentiary inquiry in order to comply with the Gingles requirement that it examine "the totality of the circumstances" and conduct "a searching practical evaluation of the past and present reality" of the voting process. Hamrick, 155 F.Supp.2d at 1359 (quoting Gingles, 478 U.S. at 79, 106 S. Ct. at 2781). The new evidence presented to the district court in 2001 included data from the 2000 census as well as election statistics from special city council elections in March and November 2000. The court also considered reports from election and demographic experts and Gainesville city officials.

Based on the evidence, the district court concluded that the plaintiffs could not prove a Section 2 violation because they failed to show consistent white bloc voting under the third prong of Gingles. Before reaching that issue, the district court analyzed the first two prongs of Gingles in some detail, stating under the first prong that the compactness and numerosity of the African-American community in Gainesville presented a difficult question, and that, under the second prong, the African-American community was politically cohesive internally even though there was insufficient evidence of cohesiveness between the African-American and Hispanic communities. Ultimately, however, the district court did not base its decision on either of the first two prongs. Instead, the court held that the plaintiffs could not prevail because the evidence did not show the existence of a white majority voting bloc. Focusing entirely on endogenous elections for city council, the district court concluded that the evidence did not support a finding that the candidate preferred by the white majority usually defeats the candidate supported by African-Americans. To the contrary, based on the extremely close percentage figures for white and African-American preferences in various campaigns, the court suggested that the interests of the white and African-American communities might in fact be merging in Gainesville. Because it held that the plaintiffs could not satisfy the third Gingles prong, the district court did not address the totality of the circumstances argument in any detail, but noted a lack of evidence of racial appeals in Gainesville's elections and the fact that the city council is 20 percent African-American while the city itself is only 15.7 percent African-American. After rejecting the plaintiffs' Section 2 argument, the district court held that the Gainesville electoral system does not violate the Fourteenth and Fifteenth Amendments.

In appealing the district court's 2001 order, the plaintiffs challenge only the determination that the Gainesville electoral system does not violate Section 2. This provision of the Voting Rights Act states, in relevant part:

(a) No voting qualification or prerequisite to voting or standard, practice, or procedure shall be imposed or applied by any State or political subdivision in a manner which results in a denial or abridgement of the right of any citizen of the United States to vote on account of race or color, or in contravention of the guarantees set forth in section 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.

It has long been recognized that "multimember districts and at-large voting schemes may `operate to minimize or cancel out the voting strength of racial [minorities in] the voting population.'" Gingles, 478 U.S. at 47, 106 S. Ct. at 2764 (quoting Burns v. Richardson, 384 U.S. 73, 88, 86 S. Ct. 1286, 1294, 16 L. Ed. 2d 376 (1966)) (internal quotations omitted). This is because "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." Id. at 48, 106 S. Ct. at 2765. However, "[m]ultimember districts and at-large election schemes ... are not per se violative of minority voters' rights." Id. To the contrary, "[m]inority voters who contend that the multimember form of districting violates § 2 must prove that the use of a multimember electoral structure operates to minimize or cancel out their ability to elect their preferred candidates." Id. (punctuation altered).

In Gingles, the Supreme Court held that, in order to find the minimization or cancellation of minority power in an electoral system using multimember districts or at-large voting, a court must determine that "a bloc voting majority [is] usually ... able to defeat candidates supported by a politically cohesive, geographically insular minority group." Id. at 49, 106 S. Ct. at 2765-66 (emphasis in original). Plaintiffs seeking to make this showing must meet a three-part test. First, the minority group must prove that "it is sufficiently large and geographically compact to constitute a majority in a single-member district." Id. at 50, 106 S. Ct. at 2766. If the group cannot make this showing, it is apparent that the multimember or at-large nature of the electoral system is not responsible for the minority-preferred candidates' lack of success. Second, "the minority group must be able to show that it is politically cohesive." Id. at 51, 106 S. Ct. at 2766. 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, such as the minority candidate running unopposed — usually to defeat the minority's preferred candidate." Id., 106 S.Ct. at 2766-67 (citation omitted).

The Supreme Court has explained that "[i]n establishing this last circumstance, the minority group demonstrates that submergence in a white multimember district impedes its ability to elect its chosen representative." Id., 106 S.Ct. at 2767. The three factors articulated in Gingles have come to be known as (1) compactness or numerosity, (2) minority cohesion or bloc voting, and (3) majority bloc voting. See, e.g., Johnson v. De Grandy, 512 U.S. 997, 1011, 114 S. Ct. 2647, 2657, 129 L. Ed. 2d 775 (1994). Ultimately, proving all three Gingles factors "is not the end of the story." Negron v. City of Miami Beach, 113 F.3d 1563, 1566 (11th Cir.1997). Even if the plaintiffs prove that the three conditions are met, a court cannot find a Section 2 violation unless it looks at the "totality of the circumstances" and determines that there has been impermissible vote dilution. See id.1

Significantly, a court cannot find vote dilution unless the plaintiffs prove all of the necessary factors. If any one of the Gingles prongs is not established, there is no vote dilution. See, e.g., Gingles, 478 U.S. at 50, 106 S. Ct. at 2766 (describing the three factors as "necessary preconditions" for finding vote dilution); Negron, 113 F.3d at 1567 ("Because we hold that the district court correctly determined that plaintiffs had failed to establish the first Gingles precondition, ... it is unnecessary for us to review its other determinations."). In this case, the district court based its decision that there was no vote dilution primarily on its finding that the plaintiffs could not show white bloc voting under the third Gingles factor. See 155 F.Supp.2d at 1368, 1376-77. Because, as discussed below, the district court's findings of fact and analysis of the third Gingles prong were not erroneous, we affirm the district court's judgment without addressing the first two prongs or the totality of the circumstances. Our analysis focuses entirely on the third prong.

As we explained in Hamrick II, plaintiffs seeking to establish the third Gingles factor "must show not only that whites vote as a bloc, but also that white bloc voting regularly causes the candidate preferred by black voters to lose; in addition, plaintiffs must show not only that blacks and whites sometimes prefer different candidates, but that blacks and whites consistently prefer different candidates." 196 F.3d at 1221 (emphases in original). Because the Section 2 analysis is concerned with more than the results of any individual 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." Gingles, 478 U.S. at 57, 106 S. Ct. at 2769 (emphasis added). Similarly, the fact that racial polarization is not present in one or a small number of campaigns or the fact that a minority candidate achieves success in a given election "does not necessarily negate the conclusion that the district experiences legally significant bloc voting." Id., 106 S.Ct. at 2770. What matters under the third prong of Gingles are larger trends, and it is important to bear in mind that such factors as incumbency, the absence of an opponent, or the use of bullet voting might explain divergences in individual elections. See id.

A district court's determination regarding one of the Gingles prongs is entitled to considerable deference. As the Supreme Court and this Court have established, we review the district court's findings on a Section 2 vote dilution claim for clear error. See Hamrick II, 196 F.3d at 1219 (citing Fed.R.Civ.P. 52(a); Gingles, 478 U.S. at 79, 106 S. Ct. at 2781). Under this standard, a finding of fact is clearly erroneous only "if the record lacks substantial evidence to support it." Lightning v. Roadway Express, Inc., 60 F.3d 1551, 1558 (11th Cir.1995) (quotations omitted). The Supreme Court has explained in the Section 2 context 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." Gingles, 478 U.S. at 79, 106 S. Ct. at 2780 (quoting Bose Corp. v. Consumers Union of U.S., Inc., 466 U.S. 485, 501, 104 S. Ct. 1949, 1960, 80 L. Ed. 2d 502 (1984)).

When we examine a district court's findings in a Section 2 case, it is essential to remember that the clearly erroneous standard "preserves the benefit of the trial court's particular familiarity with the indigenous political reality" of the state or local government under review. Id. "Deference is afforded the district court's findings due to its special vantage point and ability to conduct an intensely local appraisal of the design and impact of a voting system." Negron, 113 F.3d at 1565 (quotations omitted). Unless we are compelled to conclude that the district court's findings are not supported by substantial evidence, we must affirm.

In this case, the district court's conclusions regarding the third Gingles prong were not clearly erroneous. As part of the most recent proceedings, which were held on remand from Hamrick II, the district court considered extensive evidence concerning this prong, including information already in the record about elections between 1985 and 1995, as well as information presented about two special city council elections that took place in 2000. The district court also considered reports and testimony from the parties' election experts, Dr. Michael Maggiotto for the defendants and Dr. Richard Engstrom for the plaintiffs. Only Dr. Maggiotto prepared estimates of racial voting in the two endogenous 2000 elections. The district court noted these results along with results from the nine other contested endogenous elections in the record from the previous proceedings in the case:

Date/Ward         Candidate            White Votes     African-American
                                                       Votes

2000              Ellard               28.9%           13.4%
Ward 1            Miller               12.5%           25.4%
                  Musselwhite(win)     58.6%           61.2%

2000              Geyer                50.5%           32.0%
Ward 5            Lawson(win)          49.5%           68.0%

1995              Johnson               5.4%           69.7%
Ward 3            Morrow(win)          94.6%           30.3%

1990              Hamrick(win)         73.5%           71.8%
Ward 2            Will.             26.5%           28.2%

1990              Johnson              13.2%           66.2%
Ward 3            Morrow(win)          86.8%           33.8%

1989              Baker                 8.4%           10.3%
Ward 1            West(win)            68.7%           85.1%
                  Wiginton             22.9%            4.6%

1989              Hayes                31.4%           68.8%
Ward 4            Wangemann(win)       68.6%           31.2%

1986              Bearden              16.9%           25.8%
Ward 1            Dobbs(runoff)        44.3%           37.1%
                  Moore(runoff)        38.7%           37.1%

1986              Dobbs                49.6%           54.4%
Ward 1 Runoff     Moore(win)           50.4%           45.6%

1986              Wangemann(win)       52.8%           47.7%
Ward 4            Waters               47.2%           52.2%

1985              Allison              37.8%           20.8%
Ward 5            Canupp               12.1%            6.6%
                  Lawson(win)          50.1%           72.7%

In his report and testimony, Dr. Maggiotto explained that racial preferences could be determined in nine of these eleven elections. He could discern no preference in the 1995 Ward 3 election because he could find no precinct with a sufficiently large African-American community,2 and there was no preference in the 1986 Ward 1 election because no candidate received a majority of the vote. For the remaining nine elections, Dr. Maggiotto explained that the candidate preferred by a majority of African-American voters prevailed five times, or 55.6 percent of the time (2000 Ward 1, 2000 Ward 5, 1985 Ward 5, 1989 Ward 1, 1990 Ward 2). He also noted that in two of the remaining four elections (1986 Ward 1 runoff and 1986 Ward 4), the winning candidate received only 50.4 percent and 52.8 percent of the white vote respectively, indicating that the white community may not have voted as a bloc. Removing these two elections from the equation, Dr. Maggiotto determined that the candidate favored by the African-American community won in five of seven relevant elections, or 71.4 percent of the time.

Testifying for the plaintiffs, Dr. Engstrom did not offer any alternative voting pattern statistics. Instead, he argued that Dr. Maggiotto had erred in his analysis of the two 2000 elections. First, Dr. Engstrom claimed that the 2000 Ward 5 election, in which the candidate preferred by African-American voters defeated the candidate preferred by a slim majority of white voters, should not be treated as a so-called "split-preference election" — one in which the majority of African-American voters and the majority of white voters preferred different candidates. Second, Dr. Engstrom argued that Dr. Maggiotto's analysis of the 2000 Ward 5 election was flawed because the analysis should have been based on voter turnout rather than voter registration. Finally, Dr. Engstrom argued that the 2000 elections were not probative because they did not involve any African-American candidates.

In its findings of fact, the district court generally accepted Dr. Maggiotto's analysis. The court explained that, excluding the 1986 Ward 1 election in which no preference could be discerned, candidates preferred by African-American voters prevailed in five of eleven campaigns, or 45.5 percent, and lost in another five, or 45.5 percent. Although the African-American preferred candidate also received the majority of white votes in four of these five elections, the African-American preferred candidate won without a majority of the white votes in the 2000 Ward 5 election. In two of the five elections in which the white preferred candidate prevailed over the candidate preferred by African-Americans, the margin between the races was narrow. In the 1986 Ward 1 runoff, the winning candidate received 50.4 percent of the white vote and 45.6 percent of the African-American vote. In the 1986 Ward 4 campaign, the winner earned 52.8 percent of the white vote and 47.7 percent of the African-American vote. Of the six elections in which the races expressed different preferences, three were clearly split-preference elections in which the white candidate won (1995 Ward 3, 1990 Ward 3, and 1989 Ward 4), while three were closer, "arguably" split-preference elections. The white preferred candidate won two of the arguably split-preference elections (1986 Ward 1 runoff and 1986 Ward 4), while the African-American preferred candidate won one (2000 Ward 5). The district court noted that the significance of the one election in which the African-American preferred candidate won was diminished by the fact that the candidate was an incumbent. See 155 F.Supp.2d at 1375.

The elections that raised the greatest concern for the district court were the Ward 3 campaigns in 1990 and 1995, which were the only endogenous elections involving any African-American candidates. Although both candidates, incumbent John Morrow and challenger Rose Johnson, were African-American, the results show a clear white preference for Morrow, the winning candidate, and a clear African-American preference for Johnson, who was defeated. Despite recognizing that these elections "present the most polarized voting data," 155 F. Supp. 2d at 1373, the district court concluded that they did not support a finding of white bloc voting. While it is possible that the results show white bloc voting to defeat an African-American candidate preferred by African-American voters, the district court suggested that the results could be explained equally by Morrow's long incumbency. The court also noted that, because the two campaigns involved the same two candidates, their results "may be merely duplicative." Id. at 1376. Because it found "no evidence, testimonial or otherwise," supporting either the bloc voting explanation or the incumbency explanation, the court determined that "these two elections ultimately are not more probative than the other endogenous elections." Id. In addition, the district court observed that "[w]hen viewed in conjunction with the most recent election evidence and the election evidence as a whole, these two instances of polarized voting appear isolated. There is nothing to indicate that these elections were part of or started a trend." Id.

Based on these findings of fact, the district court concluded that "it does not appear that the white majority usually defeats the candidate preferred by black voters. Indeed, the candidate preferred by black voters wins as often as he or she loses, 45.5% of the time." Id. (emphasis in original). Furthermore, the court held that the evidence indicated "that the interests of the black and white communities may be merging," id., as evidenced by the fact that African-American and white voters preferred the same candidates in four elections and that in another three elections white preference was just over 50 percent while the African-American preference was just under 50 percent. In light of these findings, the district court determined that the plaintiffs could not show white bloc voting and could therefore not meet the third prong of Gingles.3

The plaintiffs argue that this determination was erroneous because the district court (1) failed to give adequate weight to elections involving African-American candidates, (2) improperly evaluated splitpreference elections, and (3) placed too much emphasis on a "simple mathematical approach." We are unpersuaded.

First, as a matter of fact, the district court did not clearly err in declining to afford special weight to the two endogenous elections involving African-American candidates. Although we have on various occasions held that district courts deciding African-American vote dilution claims may give more weight to elections involving African-American candidates than those involving all white contestants, see, e.g., Davis v. Chiles, 139 F.3d 1414, 1417 n. 5 (11th Cir.1998); Southern Christian Leadership Conference v. Sessions, 56 F.3d 1281, 1293 (11th Cir.1995), there is no requirement that a district court must do so. In fact, the panel in Hamrick II expressly stated that "[w]e do not mean to imply that district courts should give elections involving black candidates more weight; rather, we merely note that in light of existing case law district courts may do so without committing clear error." 196 F.3d at 1221-22 (emphasis in original).

Although a district court is free to accord extra weight to campaigns involving minority candidates, plainly it is not required to do so, especially when, as in this case, there are compelling reasons to question the probative value of such elections. While it is true that one could infer from the results of the 1990 and 1995 elections between Morrow and Johnson that white citizens voted as a bloc to defeat the candidate of choice of the African-American community, there is an equally plausible explanation that white voters supported the incumbent Morrow in 1990 and maintained their support for him in 1995. As the district court found in its thorough analysis of these campaigns, nothing in the record compelled one conclusion over the other.

The plaintiffs argue, however, that the district court overemphasized Morrow's incumbency and thus violated our admonition that incumbency "should not be viewed as a talisman by courts, sufficient in and of itself to deem an election involving an incumbent irrelevant to a plaintiff's vote dilution claim." Nipper v. Smith, 39 F.3d 1494, 1539 (11th Cir.1994) (en banc). Contrary to the plaintiffs' characterizations, however, the district court did not improperly rely on Morrow's incumbency. Rather than deeming the 1990 and 1995 Ward 3 elections irrelevant or discounting them on this basis, the court merely cited the incumbency as one part of its rationale for not affording extra weight to those campaigns. In so doing, the district court complied with the command of Gingles, which cautions courts to take one candidate's incumbency into account when deciding whether an election shows the existence of white bloc voting. See Gingles, 478 U.S. at 57, 106 S. Ct. at 2770; see also Abrams v. Johnson, 521 U.S. 74, 93, 117 S. Ct. 1925, 1937, 138 L. Ed. 2d 285 (1997) (stating that, under Gingles, "incumbency is a special circumstance to be taken into account in evaluating racial bloc voting") (quotations and punctuation omitted). Furthermore, the district court did not merely cite Morrow's incumbency without explanation. Instead, the court reflected on the substantial length of Morrow's service, which began when he joined the city council in 1978 and included two terms as mayor under Gainesville's mayoral rotation system.

In addition to the incumbency issue, there is no reason to conclude that the district court erred by emphasizing that the 1990 and 1995 elections involved the same candidates. If white voters preferred Morrow over Johnson in 1990, it is understandable that they could do so, and did so, again in 1995. Once more, it bears emphasizing that the district court did not discount or disregard the 1990 and 1995 Ward 3 elections when it conducted its overall analysis. To the contrary, the court recognized that these elections provided "the strongest evidence supporting a finding of white bloc voting," 155 F. Supp. 2d at 1375, and merely declined to give them extra weight. The Johnson-Morrow elections were instead assigned value equal to the eight other split-preference elections, including many that strongly supported the defendants' argument. In short, we can find nothing in the district court's fact-finding process or in its analytical approach that was clearly erroneous or amounted to legal error.

Next, the plaintiffs contend that the district court erred by discounting evidence that the African-American candidate of choice lost all five clear split-preference elections. According to the plaintiffs, the district court should have found white bloc voting because the candidate preferred by whites prevailed over the candidate preferred by African-Americans in the 1995 and 1990 Morrow-Johnson races, as well as in the 1989 Ward 4 campaign, the 1986 Ward 1 runoff, and the 1986 Ward 4 election. The plaintiffs also claim that the district court should not have considered the 2000 Ward 5 election, in which the candidate preferred by African-Americans defeated the candidate preferred by a small majority of whites (50.5 percent), a split-preference election. The plaintiffs say that proper evaluation of these split-preference elections would have compelled the district court to find that the white and African-American communities regularly support different candidates and that the white community's choice usually prevails.

The plaintiffs' arguments do not support a conclusion that the district court clearly erred. As an initial matter, two of the five split-preference elections emphasized by the plaintiffs are the Morrow-Johnson elections in Ward 3. As discussed above, the district court did not discount these races, but merely held that they were not entitled to special additional weight because of Morrow's incumbency and the fact that the 1995 results may have been merely duplicative of the 1990 outcome. Two of the other elections involved extremely close disparities in white and African-American voting for the two candidates. In the 1986 Ward 1 election, African-American voters preferred Dobbs over Moore by a margin of 54.4 percent to 45.6 percent, while white voters preferred Moore over Dobbs by 50.4 percent to 49.6 percent. Similarly, in the 1986 Ward 4 election, African-American voters preferred Waters over Wangemann by a margin of 52.2 percent to 47.7 percent, while whites preferred Wangemann over Waters by a margin of 52.8 percent to 47.2 percent. As the district court recognized in its 1994 opinion and as a panel of this Court indicated in Hamrick II, these two elections are not compelling evidence of a white voting bloc uniting to defeat the clear African-American candidate of choice. See Hamrick II, 196 F.3d at 1223 (suggesting that, in these two elections "at least arguably, ... white voting did not constitute a bloc and black voting did not demonstrate a preference").

Simply put, we cannot say as a matter of fact that the district court clearly erred in downplaying the significance of the splitpreference 1986 Ward 1 runoff and Ward 4 election. Based on the narrow margins in the voting, substantial evidence undoubtedly supported the district court's suggestion that "the fact that the percentages of `preference' in these cases were very close could indicate that the interests of the black and white communities may be merging instead of indicating white bloc voting." 155 F. Supp. 2d at 1376. This conclusion is strengthened by the fact that white and African-American voters unambiguously shared preferences in four other elections, and it is not undermined by the fact that the white candidate of choice prevailed over the African-American candidate of choice in the one other clearly split-preference election, the 1989 Ward 4 race. It bears repeating that the clearly erroneous standard "preserves the benefit of the trial court's particular familiarity with the indigenous political reality" of the state or local government under review, Gingles, 478 U.S. at 79, 106 S. Ct. at 2780, and, accordingly, we are required by Supreme Court precedent to afford considerable deference to its findings of fact.

Similarly, we can discern no clear error in the district court's treatment of the 2000 Ward 5 election, in which the African-American candidate of choice prevailed. Contrary to the plaintiffs' characterization of the analysis, the district court afforded this election only minor weight, treating it as nothing more than an arguably splitpreference election like the 1986 Ward 1 runoff and Ward 4 election. Labeling the election as only arguably split-preference was altogether consistent with the district court's approach to other elections with narrow margins. Although African-Americans voted for Lawson by a margin of 68 percent to 32 percent, whites preferred Geyer by a very narrow margin of only 50.5 percent to 49.5 percent. Also consistent with Supreme Court precedent, see, e.g., Abrams, 521 U.S. at 93, 117 S. Ct. at 1937, and with its own treatment of Morrow's incumbency in the Ward 3 elections, the district court recognized that the "special circumstance" of Lawson's incumbency "diminishes the relevance of that black preference win." 155 F. Supp. 2d at 1376.

We also note that the district court did not clearly err in accepting Dr. Maggiotto's statistical analysis that labeled the 2000 Ward 5 election split-preference. Although Dr. Maggiotto himself recognized that, due to the close margin in white voting, the election could not with certainty be labeled split-preference, the district court accounted for this doubt by treating the election as only arguably split-preference. Furthermore, while Dr. Engstrom, the plaintiffs' expert, argued that 2000 Ward 5 election would not have been splitpreference if Dr. Maggiotto had based his analysis of the voters' racial composition on voter turnout rather than voter registration figures, the district court correctly noted that the plaintiffs presented no evidence supporting this hypothesis. Without such evidence, we cannot conclude that the district court clearly erred in treating the election as arguably split-preference.4 In any event, it seems clear that the plaintiffs would not be able to prevail under the third prong of Gingles even if the 2000 Ward 5 election were deemed a same-preference race. The plaintiffs would still be able to show no more than one clearly split-preference election in which the white candidate of choice defeated the African-American candidate of choice without any special factor such as incumbency.

Finally, we are unpersuaded by the plaintiffs' claim that the district court clearly erred by relying on a "simple mathematical approach" instead of conducting a "searching practical evaluation of the past and present reality" with "a functional view of the political process," as required by Gingles, 478 U.S. at 45, 106 S. Ct. at 2763-64 (citations and quotations omitted). As we have already observed, it is apparent that the district court conducted a thorough analysis of the endogenous elections in Gainesville and expressly rejected the plaintiffs' claim that the majority had repeatedly vetoed African-American preferences through white bloc voting. The district court comprehensively analyzed all of the relevant elections, observed long-term trends, and explained how it reached its conclusions.

Simply put, we can find no error in the district court's finding that the plaintiffs cannot prevail under the third prong of Gingles because they failed to show (1) that whites and African-Americans "consistently prefer different candidates" and (2) that "white bloc voting regularly causes the candidate preferred by black voters to lose." Hamrick II, 196 F.3d at 1221 (emphasis in original). The evidential foundation allows for the conclusion that the preferences of the white and African-American communities have often been the same, or extremely close to one another, and that the African-Americans' candidates of choice have prevailed as frequently as they have lost, some 45.5 percent of the time. In light of this evidence, there is no basis for concluding that the district court committed clear error.

Because the district court did not clearly err in determining that the plaintiffs have not shown white bloc voting under the third prong of Gingles, the Section 2 claim fails.5 Accordingly, we affirm the judgment of the district court.

AFFIRMED.