Appellants Gregory Solomon and two other black registered voters of Liberty County, Florida, appeal the district court’s judgment that the county’s at-large system of electing its commissioners and school board members does not run afoul of section 2 of the Voting Rights Act of 1965, as amended, 42 U.S.C. § 1973 (1994). Finding clear error, in part because of the court’s reliance on the electoral success of a black candidate during the pendency of this litigation, we reverse and remand for the implementation of a remedy.
I. PROCEDURAL HISTORY
This lawsuit began in 1985 in the United States District Court for the Northern District of Florida. Appellants sued Liberty County’s Commission, School Board, and commissioners and members in their official capacities (collectively appellees), alleging that at-large elections unlawfully diluted minority voting strength, in violation of section 2 of the Voting Rights Act, 42 U.S.C. § 1973.1 In March 1986, the district court conducted a five-day bench trial and, a year later, entered final judgment in favor of ap-pellees. See Solomon v. Liberty County, Fla., Nos. 85-7009-MMP, 85-7010-MMP (N.D.Fla. May 4,1987) (unpublished).
On appeal, a panel of this court vacated and remanded for further findings of fact. Solomon v. Liberty County, Fla., 865 F.2d 1566 (11th Cir.1988) (Solomon I), vacated, 873 F.2d 248 (11th Cir.1989). In April 1990, the court reheard the case en banc. Solomon v. Liberty County, Fla., 899 F.2d 1012 (11th Cir.1990) (en banc) (Solomon II), cert. denied, 498 U.S. 1023, 111 S.Ct. 670, 112 L.Ed.2d 663 (1991). Unanimously vacating and remanding for further proceedings, the court held that contrary to the district court’s judgment, appellants satisfied the three factors set forth in Thornburg v. Gingles, 478 U.S. 30, 106 S.Ct. 2752, 92 L.Ed.2d 25 (1986), “as a matter of law[.]” Solomon II, 899 F.2d at 1013 (per curiam).
The court, however, divided five-five concerning the significance of that holding. See Solomon II, 899 F.2d at 1013-21 (Kravitch, J., specially concurring), 1021-37 (Tjoflat, C.J., specially concurring). According to five judges, appellants established section 2 violations and deserved a remedy. See Solomon II, 899 F.2d at 1021 (Kravitch, J., specially concurring). The other five judges, however, reaffirmed the prior panel’s opinion that further findings of fact were warranted. See Solomon II, 899 F.2d at 1021, 1037 (Tjoflat, C.J., specially concurring). Unable to create a majority opinion, the en banc court simply instructed the district court “to proceed in accordance with Gingles, giving due consideration to the views expressed in Chief Judge Tjoflat’s and Judge Kravitch’s specially concurring opinions.” Solomon II, 899 F.2d at 1013 (per curiam).
On remand, the district court granted ap-pellees’ motion to supplement the record and conducted a half-day trial. Based on the evidence it admitted in the first and second trials and two judicially noticed facts, the district court again found that Liberty County’s at-large elections do not violate section 2 and entered final judgment in favor of appel-lees. See Solomon v. Liberty County, Fla., 957 F.Supp. 1522 (N.D.Fla.1997) (Solomon III).
II. BACKGROUND
Liberty County is located in northwest Florida. It is especially rural; the Apalachicola National Forest encompasses much of *1139its 836 square miles. As Florida’s least densely-populated county, it averages seven persons per square mile. Liberty County’s population totals only 5,569.2 Of Liberty County’s 5,569 residents, 982- — or 17.63 percent — are black.3 The county’s total voting-age population (that is, persons over the age of 18) is 3,320, of whom 831' — or 25.03 percent — are black.
Residents vote at one of eight precincts.4 Candidate seat-wise, however, five equally populous residential districts divide Liberty County. “To a great extent, the county is ... racially segregated.” Solomon III, 957 F.Supp. at 1558. Approximately 90 percent of its black population lives in “residential district 1,” which corresponds somewhat to voting precincts 1 and 2.5 Of residential district l’s 640 registered voters in 1991, blacks comprise 295 — or 46.09 percent. Although “significant socio-economic disparities” exist between Liberty County’s white.and black residents, substantially all of its voters are Democrats. Solomon III, 957 F.Supp. at 1559.
Liberty County’s Commission and School Board each consists of five members who serve staggered four-year terms. Candidates run for numbered seats that correspond to their residential district. Voters elect candidates at-large, that is, the entire county electorate votes for one candidate from each residential district. To become their party’s nominee, candidates must receive a majority (that is, greater than 50 percent) of the county-wide vote. If no candidate receives a majority in the first primary election, the county conducts a second (or run-off) primary election. To win the general election, candidates must obtain a plurality (that is, more than their opponents) of the county-wide vote. Because most candidates in Liberty County are Democrats, voters usually decide races during the primary elections.
Prior to the commencement of this case, four black candidates unsuccessfully ran for county-wide office. “Black candidates for seats on the school board included Charles Berrium in 1968, and Earl Jennings in 1980 and 1984. In 1984, Gregory Solomon ran for a seat on the county commission.” Solomon III, 957 F.Supp. at 1557. Six months after 1990’s en banc mandate, however, Jennings won the residential district 1 seat on the county commission, defeating a white opponent in the primary and a white Republican incumbent in the general election. In 1992, voters reelected Jennings in the run-off primary over three white opponents, two of whom failed to make it past the first primary. In 1996, two black candidates and one white candidate challenged Jennings, but he again prevailed, in the ruri-off primary election. According to the record, Jennings is the only black candidate that Liberty County voters have elected to public office.
III. DISTRICT COURT’S FINDINGS OF FACT
In its memorandum opinion, the district court rendered findings of fact that favored both appellants and appellees before it ultimately found no vote dilution. As to findings of fact that favored appellants, the district court first recognized the unanimous en banc holding that appellants had proved the three Cingles factors, that is, Liberty County’s (1) black voters are “sufficiently large and geographically compact to constitute a majority in a single-member district[,]” namely residential district 1; (2) black voters are “politically cohesive”; and (3) the “white majority votes sufficiently as a bloc to enable it” usually to defeat black voters’ “preferred candidate.” Gingles, 478 U.S. at 50-51, 106 S.Ct. 2752; Solomon III, 957 F.Supp. at 1555. *1140Additionally, the court found that “a high degree of racially polarized voting” exists. Solomon III, 957 F.Supp. at 1560. Fifth, it stated that “the continued use of the majority vote requirement can work to the detriment of black candidates!!,]” as can “the size of Liberty County[.]” 957 F.Supp. at 1560-61. Finally, the court pointed out that “[bjlacks have not achieved proportional representation on the Liberty County School Board.” 957 F.Supp. at 1570.
As to findings of fact that favored appel-lees, the district court first found “no evidence” that the historical and “remaining vestiges of official discrimination in Liberty County” hinder “the ability of blacks to participate in the political process[.]” Solomon III, 957 F.Supp. at 1559. Next, it held that because of Jennings’s 1990 electoral success, “blacks have not been excluded from Liberty County’s informal slating process” that consists of “large white families” with which candidates “must align themselves” “[i]n order ... to be successful!!.]” 957 F.Supp. at 1561-63 & nn. 96-97. Additionally, according to the court, black candidates’ “lack of knowledge of the dynamics of running effective campaigns!!,]” as opposed to “the present socio-economic effects of past discrimination!.,]” “significantly] imped[e] ... black political participation in Liberty County.” 957 F.Supp. at 1564-65. Fourth, the court found “no record evidence of any recent overt or subtle racial appeals in campaigns in Liberty County.” 957 F.Supp. at 1565. Further, it stated that “the more recent history of elections in Liberty County shows consistent electoral success by the black candidate of choice for public office,” Jennings. 957 F.Supp. at 1566. Sixth, the court found that “with few exceptions, the Liberty County School Board and County Commission have been responsive to the needs of the black community.” 957 F.Supp. at 1567. Moreover, “as a matter of law[,]” the court concluded, appellants “failed to establish that the policy underlying the use of at-large elections in Liberty County is tenuous.” 957 F.Supp. at 1568. Eighth, it stated that “racial animus” does not drive Liberty County’s electoral process. 957 F.Supp. at 1570. Finally, citing Jennings’s election and re-elections, the court found that blacks “clearly” enjoy “proportional representation on the county commission.” 957 F.Supp. at 1570.
IV. ISSUE
The encompassing issue before us is whether the district court erred in finding that Liberty County’s at-large method of electing its county commissioners and school board members does not violate section 2 of the Voting Rights Act, 42 U.S.C. § 1973. We review the district court’s historical, subsidiary and ultimate findings of fact for clear error. See Gingles, 478 U.S. at 79, 106 S.Ct. 2752 (“[T]he clearly-erroneous test of [Federal Rule of Civil Procedure] 52(a) is the appropriate standard for appellate review of a finding of no vote dilution.”). “[A] finding is ‘clearly erroneous’ when although there is evidence to support it, the reviewing court on the entire evidence is left with the definite and firm conviction that a mistake has been committed.” Anderson v. City of Bessemer City, N.C., 470 U.S. 564, 573, 105 S.Ct. 1504, 84 L.Ed.2d 518 (1985) (internal quotation marks and citation omitted). Additionally, “we will correct a district court’s errors of law and its findings of fact based upon misconceptions of law.” Negron v. City of Miami Beach, Fla., 113 F.3d 1563, 1566 (11th Cir.1997).
V. DISCUSSION
To prevail on a claim of vote dilution under section 2 of the Voting Rights Act, plaintiffs must establish that, “based on the totality of circumstances,” “the political processes leading to nomination or election in the State or political subdivisions are not equally open to [racial minority] participation ... 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. At minimum, section 2 plaintiffs must prove “the three now-familiar Gin-gles factors (compactness/numerousness, minority cohesion or bloc voting, and majority bloc voting)[.]” Johnson v. De Grandy, 512 U.S. 997, 1011, 114 S.Ct. 2647, 129 L.Ed.2d 775 (1994). However, *1141other factors ... may, in “the totality of circumstances,” support a claim of racial vote dilution. Derived from the Senate Report accompanying the 1982 amendment to section 2, those 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; [and]
7. the extent to which members of the minority group have been elected to public office in the jurisdiction.
Gingles, 478 U.S. at 36-37, 106 S.Ct. at 2759 (quoting S.Rep. No. 417, at 28-29, reprinted in 1982 U.S.C.C.A.N. at 177, 206-07). Additional factors that may be probative of vote dilution in some cases are:
[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.
Gingles, 478 U.S. at 37, 106 S.Ct. at 2759.
Nipper v. Smith, 39 F.3d 1494, 1511 (11th Cir.1994) (en banc) (plurality opinion of Tjoflat, C.J.), cert. denied, 514 U.S. 1083, 115 S.Ct. 1795, 131 L.Ed.2d 723 (1995). Also, in a section 2 case involving state legislative single-member districts, the Supreme Court announced another factor, proportionality, that it defined as the relationship between “the number of majority-minority voting districts [and] minority members’ share of the relevant population.” De Grandy, 512 U.S. at 1013-14 & n. 11, 114 S.Ct. 2647. Finally, two judges of this court consider racial bias in the voting community to be a relevant factor. Nipper, 39 F.3d at 1524 (“The defendant may rebut the plaintiffs evidence by demonstrating the absence of racial bias in the voting eommunity[.]”) (plurality opinion of Tjoflat, C.J., with Anderson, J.).
In this case, the district court outlined what it believed to be “a reasoned methodology for examining claims under section 2”:
In order to prove unlawful vote dilution, in violation of section 2, a plaintiff must meet the following test. First, the plaintiff must demonstrate the three Gingles factors of geographic compactness, political cohesiveness, and racial bloc voting. If one or more of the Gingles factors is not shown, then the defendant prevails. If all three factors are proven, then the court must review all relevant evidence under the totality of the circumstances. The defendant may present evidence of the lack of racial bias in the community, proportionate representation, past and present electoral success, as well as proof of the other factors which are indicative of the existence or non-existence of vote dilution. The plaintiff may respond in kind. Without treating any single factor as dispositive, the reviewing court will finally determine through a searching inquiry whether the members of the minority group are denied equal political opportunity with respect to their race *1142or color. If they are, then a claim under section 2 has been established.
Solomon III, 957 F.Supp. at 1553-54. The parties do not dispute the legal accuracy of this framework. Rather, they dispute the district court’s application of it as well as the dilution factors. As we outlined in section III of this opinion, the district court found that the three Gingles factors, Senate Factors 2 and 3, and a lack of proportionality (only as to the school board) cut in favor of appellants. On the other hand, the court weighed Senate Factors 1, 4-9, proportionality (only as to the county commission), and the absence of racial bias in favor of appel-lees. Appellants argue that the district court clearly erred in rendering findings adverse to them, including its ultimate determination of no vote dilution. Appellees defend the district court’s judgment, contending that it properly resolved conflicts in the evidence and considered the totality of the circumstances.
A.
We begin with appellants’ principal assignment of error, the district court’s reliance on the electoral success of Earl Jennings, a black county commissioner. One of the two “most important Senate Report factors” is the “extent to which minority group members have been elected to public office in the jurisdiction.” Southern Christian Leadership Conference of Alabama v. Sessions, 56 F.3d 1281, 1301 (11th Cir.1995) (en banc) (quoting Gingles, 478 U.S. at 48 n. 15, 106 S.Ct. 2752, and Senate Report at 28-29) (Hatchett, J., dissenting), cert. denied, 516 U.S. 1045, 116 S.Ct. 704, 133 L.Ed.2d 660 (1996). It is undisputed that prior to 1990, no black candidate — including Jennings — had been elected to a county-wide office. At the second trial, appellees proffered the results of the September 1990 Democratic primary, in which Jennings became the party’s nominee:
[[Image here]]
In 1990 s general election, Jennings narrowly beat the Republican incumbent:
[[Image here]]
After the second trial, at appellees’ request, the district court took judicial notice of Jennings’s 1992 reelection. The September 1992 first primary results were as follows:
[[Image here]]
*1143In October 1992’s run-off primary election, facing no Republican opposition, Jennings retained his seat on the county commission:
[[Image here]]
Similarly, the district court judicially no-tieed Jennings’s 1996 reelection. In the first primary, Jennings faced two black and one white challengers:
[[Image here]]
Jennings defeated Willis Brown — the same white opponent he beat in 1992 — in the 1996 run-off primary:
[[Image here]]
The district court cited this data almost exclusively in finding that: (1) Jennings achieved consistent electoral success in 1990, 1992 and 1996 as the black candidate of choice (Senate Factor 7); (2) the informal but controlling white family slaters endorsed Jennings in 1990 and, therefore, did not exclude black candidates in general (Senate Factor 4); and (3) blacks enjoyed proportional representation on the county commission because Jennings held one of five seats, a proportion in line with Liberty County’s black population (De Grandy). See Solomon III, 957 F.Supp. at 1561-63 & nn. 96-97, 1565, 1570.
We agree with appellants that the district court clearly erred in rendering the first two findings (Senate Factors 7 and 4), but we hold that the district court did not clearly err with the third finding (DeGrandy /proportionality). It is well-recognized that “[ejections of minority candidates during the pen-dency of Section Two litigation ... have little probative value.” Davis v. Chiles, 139 F.3d 1414, 1417 n. 2 (11th Cir.1998) (assigning “little probative value” to the post-trial appointment and reelection of a black judge in Florida’s Second Judicial Circuit, which includes Liberty County, and the “recent” election of a black judge who faced a white opponent in Leon County, Florida). In crafting this rule, Congress, the Supreme Court and the former Fifth Circuit reasoned that “the possibility exists that the majority citizens might evade [§ 2] by manipulating the election of a ‘safe’ minority candidate” and that pending litigation “might” work “a onetime advantage for black candidates in the *1144form of unusual organized political support by white leaders concerned to forestall single-member districting.” Gingles, 478 U.S. at 75-76, 106 S.Ct. 2752 (quoting in part Senate Report at 29, and Zimmer v. McKeithen, 485 F.2d 1297, 1307 (5th Cir.1973) (era banc), aff'd sub nom., East Carroll Parish School Bd. v. Marshall, 424 U.S. 636, 96 S.Ct. 1083, 47 L.Ed.2d 296 (1976) (per curiam)). Indeed, because it is couched in “possibilities” and “mights,” this rule may very well be unconditional, that is, not requiring proof beyond the timing of the election. See also NAACP v. Gadsden County Sch. Bd., 691 F.2d 978, 983 (11th Cir.1982) (“[Politicians might find it politically expedient to have a ‘token’ black school board member.”) (emphasis added).
In any event, to be sure, the circumstances surrounding Jennings’s 1990 election fit squarely within the rule’s rationale. Following two unsuccessful school board elections, Jennings became Liberty County’s first elected county official after not only the filing of this lawsuit, but also the era banc court’s mandate that appellants proved the three Gingles factors as a matter of law. Indeed, it occurred in the wake of five circuit judges concluding that appellants established liability and deserved a remedy. See Solomon II, 899 F.2d at 1021 (Rravitch, J., specially concurring). Furthermore, as the district court found, Jennings would not have won the Democratic primary but for the support of “large white families” that ran Liberty County’s informal slating process. See Solomon III, 957 F.Supp. at 1561-62. The “possibility” of vote “manipulation” further intensifies when we consider the anomalous polarization data associated with Jennings’s pre-litigation races. Gingles, 478 U.S. at 75-76, 106 S.Ct. 2752. According to all ten judges of the era banc court who reviewed that data, the white majority likely perceived Jennings as a “safe” minority candidate. See Solomon II, 899 F.2d at 1019-20 nn. 8 & 9 (“[T]he white establishment in Liberty County backed Earl Jennings’ 1980 school board candidacy.”) (Kravitch, J., specially concurring); Solomon I, 865 F.2d at 1579-81 & n. 26 (“[Vjoting patterns were not polarized in Earl Jennings’s 1980 candidacy. The reason for this lack of polarization may well be explained by the district court’s observation that Jennings had run on a ‘white slate.’ ”). Jennings’s beating a Republican in a predominately-Democrat county and twice retaining his seat as an incumbent are other “special circumstances” that likely belie his electoral success. Meek v. Metropolitan Dade County, Fla., 985 F.2d 1471, 1483-84 (11th Cir.1993) (“[T]he presence of special circumstances such as incumbency or lack of opposition may explain the success of the minorities’ preferred candidate in some elections.”) (per curiam) (emphasis added). But cf. Solomon III, 957 F.Supp. at 1556 n. 79 (“[A]ny suggestion that racial polarization in the county was somehow caused by partisan politics ... would have no meaning here.”) (distinguishing League of United Latin Am. Citizens, Council No. 4434 v. Clements, 999 F.2d 831 (5th Cir.1993) (era banc), cert. denied, 510 U.S. 1071, 114 S.Ct. 878, 127 L.Ed.2d 74 (1994)) (emphasis added). So is 1992’s first primary, where Jennings faced three white challengers. See Rollins v. Fort Bend Indep. Sch. Dist., 89 F.3d 1205, 1212 (5th Cir.1996) (“The evidence demonstrates that minority candidates have been elected to the ... school board without special circumstances, such as a candidate running unopposed or opposed by two strong Anglo candidates.’’) (emphasis added).
Contrary to the district court’s view, appellees, not appellants, bore the burden of proving that Jennings’s success had probative value. See Solomon III, 957 F.Supp. at 1566 (“There has been no suggestion nor proof offered by Plaintiffs that white officials or white voters in Liberty County have manipulated the electoral success of Mr. Jennings in order to defeat Plaintiffs vote dilution challenge.”) (emphasis added). Although section 2 plaintiffs bear the burden of proving the Gingles factors and other factors in the totality of circumstances that support a finding of vote dilution, defendants bear the burden of proving any factor that they believe weighs in their favor. See Sanchez v. Colorado, 97 F.3d 1303, 1322 (10th Cir.1996) (“Plaintiffs are not required to rebut all the evidence of non-dilution to establish vote dilution.”), cert. denied, — U.S.-, 117 S.Ct. 1820, 137 L.Ed.2d 1028 *1145(1997); Teague v. Attala County, Miss., 92 F.3d 283, 290 (5th Cir.1996) (“[T]he district court errs by placing the burden on plaintiffs to disprove that factors other than race affect voting patterns in Attala County.”), cert. denied, - U.S. -, 118 S.Ct. 45, 139 L.Ed.2d 12 (1997); accord Nipper, 39 F.3d at 1513 (“A defendant in a vote dilution case may always attempt to rebut the plaintiffs claim by introducing evidence of objective, non-racial factors under the totality of the circumstances standard.”) (plurality opinion of Tjoflat, C.J.). Consistent with this standard and the district court’s undisputed legal framework, as proponents of the post-ew banc evidence (and to rebut the presumption that the election of a minority candidate during section 2 litigation carries no more than nominal probative value), appellees bore the burden of proving that Jennings was, in fact, the black candidate of choice. Cf. United States v. Marengo County Comm’n, 731 F.2d 1546, 1574-75 (11th Cir.) (“In view of the evidence already in the record, the defendants bear the burden of establishing that circumstances have changed sufficiently to make our finding of discriminatory results in 1978 inapplicable in 1984.”), cert. denied, 469 U.S. 976, 105 S.Ct. 375, 83 L.Ed.2d 311 (1984). In proffering only the election results, however, they failed to meet that burden as a matter of law. Unquestionably, mere proof of Jennings’s race could not support a finding that he was the choice of black voters. See Sanchez, 97 F.3d at 1320. Importantly, appellees did not proffer any statistical analysis at the second trial. See Solomon III, 957 F.Supp. at 1560 n. 89 (appellees did not “bother[ ] to analysis of [Jennings’s] elections in 1990, 1992, and 1996”). Nor could appellees rely on pre-en banc statistical evidence concerning Jennings to show that he was the black candidate of choice. Even the original appellate panel considered that evidence to be of little value. See Solomon I, 865 F.2d at 1581 (“Earl Jennings, although black, may not have been perceived as a black candidate”).
The district court’s sua sponte mathematical analysis of the 1990, 1992 and 1996 elections results further evinces clear error. Although neither party argued percentages at trial or in their proposed findings of fact and conclusions of law, the court became its own math advocate. In finding Jennings to be the “overall ...' black candidate of choice[,]” the court stated that
[i]n the 1990 election, Mr. Jennings received 75.52 percent of the votes east in district 1 in the first primary, and 76.81 percent of the votes cast in district 1 in the general election.... In the 1992 election, Mr. Jennings received 55.84 percent of the votes cast in district 1 in the first primary (a figure less conclusive of black support), and 72.04 percent of the votes cast in district 1 in the second primary.... In the 1996 election, black support in the first primary was apparently divided between the three black candidates (Mr. Jennings received 24.06 percent of the vote in residential district 1, compared to 52.83 percent for Stafford Stanley Dawson and 12.26 percent for Helen Hall), but was squarely behind Mr. Jennings in the second primary when he garnered 73.02 percent of the votes cast in district 1.
Solomon III, 957 F.Supp. at 1562-63 n. 97 (emphasis added). As is readily. apparent, although ■ it divided the numbers correctly, the court used data from “voting precinct 1” to reach conclusions about “residential district 1.”6 This was clear error. “Voting precinct 1” simply refers to one of eight physical *1146polling locations. “Residential district 1,” in contrast, represents one of five seats for which county commission and school board candidates run, as well as a would-be single-member district where blacks, together with the white cross-over vote, are “sufficiently large in number and geographically compact to constitute a majority[.]” Gingles, 478 U.S. at 50, 106 S.Ct. 2752; see Solomon I, 865 F.2d at 1574 (“[I]f white crossover votes are included, an effective black majority would be achieved in the district.”). Plainly, residential district l’s borders do not equate to those of voting precinct 1; according to the district court .itself, “residential district 1 ... includes the first and second precincts.” Solomon III, 957 F.Supp. at 1556 (emphasis added). Nor does residential district 1 represent the simple combination of voting precincts 1 and 2. Compare Solomon III, 957 F.Supp. at 1556 (“[BJlacks comprise 46.09 percent of all the registered voters in residential district one.”) with 957 F.Supp. at 1562 n. 96 (voting precinct 1 is 58.05 percent black, and voting precinct 2 is 56.92 percent black).
Our finding clear error in the district court’s reliance on Jennings’s electoral success, although “inherently fact-intensive,” comports with precedent. Nipper, 39 F.3d at 1498 (plurality opinion of Tjoflat, C.J.). In NAACP v. Gadsden County Sch. Bd., this court held that the district court attached “inordinate significance” to a single black candidate’s successful, at-large election to the school board that occurred after plaintiffs filed the lawsuit. 691 F.2d at 983 (“The election of a single black ... is not conclusive evidence that the votes of the minority are not being diluted.”). Similarly, in United States v. Marengo County Comm’n, this court discounted the probative value of a black candidate’s election and reelection to county coroner in an at-large system. See 731 F.2d at 1572 (“The district court’s conclusion that this nearly complete lack of success did not indicate a lack of effective access to the system ... is clearly erroneous.”). And, the court in McMillan v. Escambia County, Fla. did not even consider — and thus relegated to a footnote — the at-large election of one black county commissioner during the pen-dency of litigation. See 748 F.2d 1037, 1045 n. 20 (Former 5th Cir.1984). See also generally White v. Regester, 412 U.S. 755, 766-67, 93 S.Ct. 2332, 37 L.Ed.2d 314 (1973) (finding significance in the fact that the “white-dominated organization that is in effective control of Democratic Party candidate slating” had endorsed only two black candidates) (cited in Marengo, 731 F.2d at 1569); City of Carrollton Branch of the NAACP v. Stallings, 829 F.2d 1547, 1560 (11th Cir.1987) (finding “no sustained minority electoral success” within the county government despite one minority candidate’s election to a municipal position after the filing of the lawsuit), cert. denied, 485 U.S. 936, 108 S.Ct. 1111, 99 L.Ed.2d 272 (1988). Unlike the district court, these appellate panels did not require plaintiffs to advance proof of voter manipulation. See Campbell, 691 F.2d at 983; Marengo, 731 F.2d at 1572; McMillan, 748 F.2d at 1045 & n. 20; accord Davis, 139 F.3d at 1417 n. 2.7 Although we find that the district court did not clearly err in finding that blacks had achieved proportional representation (only as to the county commission) pursuant to De-Grandy, we accord this factor less weight because of the timing of Jennings’s election. See De Grandy, 512 U.S. at 1020, 114 S.Ct. 2647 (“[T]he degree of probative value assigned to proportionality may vary with the facts.”)
*1147B.
We turn next to appellants’ contention that the district court erred in finding that black candidates’ “lack of knowledge of the dynamics of running effective campaigns[,]” as opposed to “the present socio-economic effects of past discrimination[,]” “significantly] imped[e] ... black political participation in Liberty County.” Solomon III, 957 F.Supp. at 1564-65 (Senate Factor 5). Appellants argue that the court assumed without foundation that black candidates and would-be candidates are more “apathetic”' than their white counterparts. In response, appellees contend that the record supports the district court’s finding.
We do not find that the district court clearly erred in rendering its finding on Senate Factor 5. In this circuit, it is well-established that “when there is clear evidence of present socioeconomic or political disadvantage resulting from past discrimination, ... the burden is not on the plaintiffs to prove that this disadvantage is causing reduced political participation, but rather is on those who deny the causal nexus to show that the cause is something else.” Marengo, 731 F.2d at 1569 (collecting eases). Thus, Senate Factor 5 should have weighed in appellants’ favor unless appellees offered a preponderance of evidence that something other than the lingering effects of past discrimination caused the low number of black candidates.
Appellees point to four portions of the record that they contend support the district court’s finding: (1) testimony that “blacks have frequently not known nor asked about alternatives to paying qualifying fees”; (2) Solomon’s testimony that he “did not know a voter registration list was available to assist in targeting people who are more likely to vote”; (3) Solomon’s testimony that “getting to know the public is also a big part of’ a successful campaign; and (4) testimony from Dr. Billings, appellee’s expert witness, that “Jennings’s electoral success in 1990 resulted in large part from becoming more knowledgeable about the political process in Liberty County, and running a very personal kind of campaign.” Solomon III, 957 F.Supp. at 1564-65 n. 101 (internal quotation marks and record citations omitted). In addition, the district court provided additional examples supporting its finding that socioeconomic disparities did not hinder black political participation in Liberty County. We believe that appellee’s contentions, along with other subsidiary findings that the district court made, support the conclusion that the district court’s finding that “the greatest obstacle to effective black participation in the political process has been a lack of knowledge of the dynamics of running effective campaigns” was not clearly erroneous. Solomon III, 957 F.Supp. at 1564.
C.-
Finally, we discuss appellants’ assertion that the district court clearly erred in its overall finding of no vote dilution. Our review of this issue, of course, must take into account the error that we have already found. In section V, part A, we found clear error in the district court’s reliance on Jennings’s post-en banc electoral success. This error tainted the court’s weighing Senate Factors 4 and 7 (but not proportionality as to the county commission) in favor of appellees.8
That leaves us with the three Gingles factors, Senate Factors 2 and 3, and lack of proportionality (only as to the school board) weighing in favor of appellants, proportionality (only as to the county commission, and accorded less weight because of the timing of Jennings’s election) and Senate Factor 5 weighing in favor of appellees, and Senate Factors 1, 6, 8 .and 9 and the absence of racial bias also weighing in favor of appel-lees. A review of this record leaves us with “a definite and firm conviction that a mistake has been committed.” Anderson, 470 U.S. at *1148573, 105 S.Ct. 1504. First and foremost, appellants’ unquestionable establishment of the three Gingles factors, that is, compactness/numerousness, minority cohesion and majority bloc voting, cuts very strongly in their favor. As the Third, Fifth and Tenth Circuits have observed, “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.” Jenkins v. Red Clay Consol. Sch. Dist. Bd. of Educ., 4 F.3d 1103, 1135 (3d Cir.), cert. denied, 512 U.S. 1252, 114 S.Ct. 2779, 129 L.Ed.2d 891 (1994); see also Clark v. Calhoun County, Miss., 21 F.3d 92, 97 (5th Cir.1994); Sanchez, 97 F.3d 1303, 1322 (10th Cir.1996) (“[I]t is the rare case for minority plaintiffs to satisfy the Gingles ’ preconditions and fail to overcome defendants’ evidence that other factors rebut that initial proof[.]”). Although this case may be “hard[J” we do not find it particularly “unusual” or “rare.” Solomon III, 957 F.Supp. at 1525; Jenkins, 4 F.3d at 1135; Sanchez, 97 F.3d at 1322. All ten judges of the en banc court agreed that appellants proved the Gingles factors “as a matter of law,” and at least five of those judges found that appellants “adduced strong evidence establishing the other supportive [that is, Senate] factors.” Solomon II, 899 F.2d at 1013 (per curiam), & 1016-17 n. 3 (Kraviteh, J., specially concurring) (emphasis added). Although eight years have elapsed between the en banc session and this appeal, the only new analytical evidence that appellees presented — Jennings’s electoral success and the results of a 1990 referendum where approximately 59.1 percent of Liberty County’s black voters supported the retention of at-large elections for the county commission, and approximately 60 percent supported at-large elections for the school board — have not changed the record in their favor. See Solomon I, 865 F.2d at 1584 (“[C]lass opposition to the remedy that may result from the successful litigation of a section 2 claim is irrelevant in weighing the totality of circumstances.”); Solomon III, 957 F.Supp. at 1555 n. 75 (even if the district court considered appellees’ new evidence concerning the threshold factors, it “would still hold that the three Gingles factors have been established as a matter of law”); 1987 Memorandum Opinion at 24 (“[T]he propriety of Liberty County’s election system does not depend on whether, why, or what proportion of blacks favor a ehange[.]”).
Greatly underscoring the strength of appellants’ evidence is Liberty County’s high degree of racially polarized voting (Senate Factor 2), a factor that courts consider the keystone of a section 2 claim. Gingles, 478 U.S. at 55, 106 S.Ct. 2752; McMillan, 748 F.2d at 1043; Marengo, 731 F.2d at 1566; cf. Nipper, 39 F.3d at 1511-12 (“racially polarized voting[] is the linchpin of a § 2 vote dilution claim”) (internal quotation marks and citation omitted) (plurality opinion of Tjoflat, C.J.). Indeed, it serves as one of the two “most important Senate Report factors[.]” Sessions, 56 F.3d at 1301 (quoting Gingles, 478 U.S. at 48 n. 15, 106 S.Ct. 2752, and Senate Report at 28-29) (Hatchett, J., dissenting). Moreover, as the district court found, “the continued use of the majority vote requirement can work to the detriment of black candidates[,]” as can “the size of Liberty County[.]” 957 F.Supp. at 1560-61 (Senate Factor 3); see generally Nipper, 39 F.3d at 1510 (“Of particular concern are electoral structures, such as at-large elections in areas with white majorities, that produce racial vote dilution because ‘[t]he right to vote can be affected by a dilution of voting power as well as by an absolute prohibition on casting a ballot.’ ”) (quoting Allen v. State Bd. of Elections; 393 U.S. 544, 569, 89 S.Ct. 817, 22 L.Ed.2d 1 (1969)) (plurality opinion of Tjoflat, C.J.). In its original memorandum opinion, the district court pointedly explained the significance of Liberty County’s size: “Liberty County does present a vast area to cover in an at-large campaign, and this factor could work to the detriment of black candidates because of their tower average income.” 1987 Memorandum Opinion at 13; cf. McMillan, 748 F.2d at 1044 & n. 17 (county’s size of 657 square miles “enhance[d] the problems faced by blacks seeking access to the political process”); Marengo, 731 F.2d at 1570 (county’s “small population” and “large, rural area” rendered it expensive to campaign “countywide ... for an at-large position”). The *1149disturbing fact that Liberty County’s blacks earn, on average, less than a third of the amount that whites earn intensifies this effect. See Solomon III, 957 F.Supp. at 1563 (“While the average per capita income for whites in the county is $ 13,127, the average per capita income for blacks is only $ 3,935.”); cf. Marengo, 731 F.2d at 1570 (“Since blacks earn, on average, less than half of the amount that whites earn, the district court correctly found that the size of the county contributes to dilution.”) (emphasis added).9 Finally, to date, “[b]lacks have not achieved proportional representation on the Liberty County School Board.” Solomon III, 957 F.Supp. at 1570 (De Grandy factor).
In contrast to the evidence favoring appellants, the evidence relative to the remaining Senate and other factors that purportedly cut in appellees’ favor&emdash;Senate Factors 1, 6, and 8 and the absence of racial bias&emdash;was anything but strong, and the district court’s findings regarding Senate Factor 9 were clearly erroneous. Regarding Senate Factor 1, that appellants presented, “no evidence” that Liberty County’s undisputed and extensive history of official discrimination hinders blacks’ ability to participate in the democratic process is, at most, a wash. Solomon III, 957 F.Supp. at 1559. As we explained in section V, part A of this opinion, appellees, not appellants,-bore the burden of proving any Senate factor that they believed weighed in their favor. Thus, appellants’ failure to produce sufficient evidence on this factor should not hurt their claim. Equally as weak is the lack of “recent overt or subtle racial appeals in campaigns in Liberty County.” Solomon III, 957 F.Supp. at 1566 (Senate Factor 6); see Marengo, 731 F.2d at 1571 (lack of evidence of racial appeals “should not weigh heavily against a plaintiff proceeding under the results test of section 2” because “overtly bigoted behavior has become more unfashionable”) (internal quotation marks and citation omitted).
Nor is the district court’s finding of responsiveness (Senate Factor 8) significant. As the court itself acknowledged, this factor is “less important than other factors” in the totality of circumstances. Solomon III, 957 F.Supp. at 1566 (citing Senate Report at 29; and Marengo, 731 F.2d at 1572); see also Gadsden, 691 F.2d at 983 (“Responsiveness .... has nothing to do with impact.”). Strangely, the court’s finding contradicts its original view of the evidence. Soon after the first trial, the court found that appellants had “proved a significant lack of responsiveness to the interests of blacks by Liberty County’s elected officials.” 1987 Memorandum Opinion at 19 (emphasis added). Yet, after remand, without any new relevant evidence, the court found just the opposite, that is, that both the county commission and the school board respond to the needs of the black community. See Solomon III, 957 F.Supp. at 1567. These findings are irreconcilable. At best, then, responsiveness is a close call.
With regard to Senate Factor 9, however, we find substantial evidence of clear error in the district court’s finding that the policy underlying at-large election of school board members, adopted in 1953, was not tenuous but instead was the result of a “citizen’s reform movement in the county to abolish the existing ward-type political system.” Solomon III, 957 F.Supp. at 1568. The district court acknowledged that “[t]here is little question that the Florida legislature’s 1947 legislation allowing counties to adopt at-large districts for the election of school boards ... was designed to dilute the voting power of the black community.” Solomon III, 957 F.Supp. at 1567. The district court characterized this statute as “allowing” the adoption of at-large elections for school boards; however, the Florida Legislature chose mandatory language in constructing this statute and this court has characterized the statute as a “requirement.” See, e.g., Solomon I, 865 F.2d at 1578 (Tjoflat, C.J., concurring); McMillan v. Escambia County, Fla., 638 F.2d 1239, 1245-49 (5th Cir.Feb.), cert. dismissed, 453 U.S. 946, 102 S.Ct. 17, 69 L.Ed.2d 1033 (1981).
The district court also relied upon the testimony of Dr. Billings to support its *1150conclusion on Senate Factor 9. On cross-examination, however, Dr. Billings stated specifically that he was testifying about the use of at-large elections in Liberty County only with respect to the county commission— not the school board. The district court further relied upon the results of a 1990 county referendum in which voters chose not to abolish at-large elections for the school board, with a resulting vote of 72.67% to 27.43%. Statistical analysis of this result showed that black as well as white voters preferred to retain the at-large system, a fact that the district court ruled “undercut any suggestion that the continued policy of maintaining at-large elections ... is somehow discriminatory or otherwise tenuous.” Solomon III, 957 F.Supp. at 1568. Whether the protected class supports the challenged system, however, is irrelevant in determining whether the system is based upon a discriminatory policy. See Solomon I, 865 F.2d at 1584 (“[C]lass opposition to the remedy that may result from the successful litigation of a section 2 claim is irrelevant in weighing the totality of circumstances.”). Thus, we conclude that the district court clearly erred in finding that Senate Factor 9 did not support appellants’ position in this ease.
With regard to racial animus, even if we assume, without deciding, that the district court’s findings are sound, they certainly do not tip the scales in appellees’ favor. In fact, the district court essentially neutralized that finding: “the lack of racial animus [in the voting community] does not cancel out the proof provided under the other factors.” Solomon III, 957 F.Supp. at 1570. Moreover, under the approach that only two judges of this court who advocate the relevance of this factor advance, appellees bore the burden of proof. See Nipper, 39 F.3d at 1515 (“[I]f the evidence [that defendant sponsors] shows, under a totality of the circumstances, that the community is not motivated by racial bias in its voting patterns, then a case of vote dilution has not been made.”) (plurality opinion of Tjoflat, C.J., and Anderson, J.). But see Nipper, 39 F.3d at 1547-60 (neither the presence nor absence of racial bias in the voting community is relevant to section 2’s results test) (Hatchett, J., dissenting). It is safe to say that appellees barely exceeded, if not failed to meet altogether, that burden through cross-examination of Solomon, appellants’ witness. See Solomon III, 957 F.Supp. at 1569 (citing Solomon’s testimony at 59, 61, 66 and 82-84).
In short, the neutral findings and evidence favoring appellees did not begin to outweigh appellants’ strong proof of the three Gingles factors and other vote dilution factors. The district court clearly erred in finding to the contrary. We are mindful that no mechanistic formula exists under section 2 of the Voting Rights Act. See De Grandly, 512 U.S. at 1011, 114 S.Ct. 2647 (courts should decide section 2 claims based on a “comprehensive, not limited, canvassing of relevant facts”); Marengo, 731 F.2d at 1574 (“No formula for aggregating the factors applies in every case.”). Our careful, searching and practical review of the record, however, compels us to hold that both the Liberty County Commission and the Liberty County School Board violated section 2 through their at-large election system. See Gingles, 478 U.S. at 79-80, 106 S.Ct. 2752 (courts should “carefully” consider the totality of circumstances “based upon a searching practical evaluation of the past and present reality”) (internal quotation marks and citation omitted).
VI. CONCLUSION
For the foregoing reasons, we reverse the judgment of the district court and remand the case for the implementation of a remedy and other proceedings consistent with this opinion.
REVERSED and REMANDED.
. Appellants' lawsuit against the School Board encompassed not only section 2 of the Voting Rights Act but also the Fourteenth and Fifteenth Amendments to the United States Constitution. Because of our holding, we need not reach the constitutional claim. Accord Solomon v. Liberty County, Fla., 865 F.2d 1566, 1569 n. 2 (11th Cir.1988) {Solomon I) (appellants’ “constitutional claims are not before us on appeal”), vacated, 873 F.2d 248 (11th Cir.1989).
. Current figures are based on the 1990 census and other evidence presented at the second trial. At the time of the first trial, Liberty County’s population was 4,260.
. At the time of the first trial, blacks totaled 471 — or 11.06 percent — of Liberty County’s 4,260 population.
. Apparently, at the time of the first trial, seven voting precincts existed. See Solomon v. Liberty County, Fla., Nos. 85-7009-MMP, 85-7010-MMP, at 11 (N.D.Fla. May 4, 1987) (unpublished).
. According to the evidence at the first trial, blacks comprise 58.05 percent of precinct 1 and 56.92 percent of precinct 2.
. Although no party points out these errors to us, they are patently obvious. According to the table that we have reproduced in the text of the opinion, in the 1990 primary, Jennings received 75.52 percent (108 / (108 + 35)) of the votes in precinct 1&emdash;a figure that the district court erroneously attributed to “district 1.” Likewise, (1) in the 1990 general election, Jennings received 76.81 percent (106 / (106 + 32)) of the votes in precinct 1, not "district 1”; (2) in the 1992 first primary election, Jennings received 55.84 percent (110 / (110 + 11 + 30 + 46)) of the votes in precinct 1, not "district 1"; (3) in the 1992 runoff primary election, Jennings received 72.04 percent (134 / (134 + 52)) of the votes in precinct 1, not "district 1"; (4) in the 1996 first primary election, Jennings received 24.06 percent (51 /(51 + 23 + 112 + 26)) of the votes in precinct 1, not "residential district 1”; and (5) in the 1996 run-off primary election, Jennings received 73.02 percent (138 / (138 + 51)) of the votes in precinct 1, not "district 1." Solomon III, 957 F.Supp. at 1562-63 n. 97.
. At first blush, this case may appear to contradict Askew v. City of Rome, where this court adopted the district court’s judgment that no vote dilution resulted from Rome’s at-large method of electing its commissioners and board of education members, in part because plaintiffs "produced no evidence” that the white electorate engaged in "some sort of active conspiracy ... to allow the black community a token representative, but no more.” 127 F.3d 1355, 1385 (11th Cir.1997) (per curiam). A close read of Askew, however, reveals that it is materially distinguishable from this case. First, unlike the voters in Liberty County, the voters in Rome elected the black candidate prior to litigation. Next, in contrast to appellants, the Askew plaintiffs failed to establish one of the Gingles factors, racial polarization. Finally, the plaintiffs in Askew bore the burden of discounting minority success because they were attempting to prove an "essential element” of their section 2 claim, whereas appellees in this case bore the burden of proving those Senate factors (e.g., Senate Factor 7, minority electoral success) that they believed weighed in their favor. 127 F.3d at 1385.
. Because the district court’s finding of proportionality regarding Jennings’s election to the county commission is accorded less weight, and because the lack of proportionality on the school board is not dispositive to our reversal, we assume without deciding that De Grandy’s proportionality factor properly applies where, as here, an at-large election system — as opposed to a single-member voting district — is challenged. See De Grandy, 512 U.S. at 1014 n. 11, 114 S.Ct. 2647 (defining “proportionality” as the relationship between "the number of majority-minority [single-member] voting districts [and] minority members’ share of the relevant population! ]”).
. See also 1987 Memorandum Opinion at 15 ("[BJlacks in Liberty County currently bear the effects of past discrimination in areas such as education, employment, and health.”).