This is our second opportunity to consider the allegation that Attala County's districting plans for the election of county supervisors and county constables dilute minority voting strength.1 Plaintiffs initially alleged violations of both § 2 of the 1965 Voting Rights Act, 42 U.S.C. 1973, and the one person-one vote principle of the Equal Protection Clause of the Fourteenth Amendment. Following a hearing, the district court found that the plaintiffs failed to prove either claim. On appeal, an earlier panel of this court affirmed the judgment on the constitutional claim but sent the § 2 issue back to the district court for a more thorough examination of plaintiffs' statistical evidence on racially polarized voting. On remand the district court again concluded that the districting plans in place in Attala County do not violate § 2 of the Voting Rights Act. We now find that the district court erred in drawing this conclusion in the face of plaintiffs' quantitative proof. We also find that the district court's crediting of the depressed level of black political participation in Attala County to black voter apathy untenable. We therefore reverse the decision of the district court and render judgment in favor of the plaintiffs.

The 1990 Census reported that blacks make up 7,299 or 39.5% of Attala County's total population of 18,481. Despite comprising such a significant proportion of the population, no black candidate has ever won a county-wide election or an election in a white majority district when pitted against a white candidate. That means, as the district court found, "no black candidates ever have won election to public office in Attala County as justice court judge, constable, sheriff, circuit clerk, coroner, county attorney, or state senator, during modern times." The district court provided a full list of the relevant black candidates and contests, as well as a racial breakdown of the districts in its first opinion, Teague v. Attala County, 807 F. Supp. 392, 398-99 (N.D.Miss.1992) ("Teague I ").

The county's current redistricting plan provides for the election of five county supervisors, one from each of five single member districts, and two justice court judges (constables), one from each of two single member districts. Blacks are the majority among voters in only one of the five supervisor districts. That district is District 4, in which 59.3% of the voting age population is black. Blacks are a minority in both of the county's constable districts.

Attala County adopted its current districting outline for the board of supervisors following its last apportionment in 1983. The United States Department of Justice precleared the plan in accordance with § 5 of the Voting Rights Act on December 23, 1983. Likewise, the Justice Department granted preclearance to the county's present justice court districting plan on June 10, 1983. Following the 1990 census, the Justice Department insisted that Attala County redistrict to create two majority-black districts. Attala County rejected this idea and chose instead to use the existing districts.

Plaintiffs' experts proffered evidence of what they claimed to be racially polarized voting. This evidence came by way of two standard methods for analyzing electoral data: ecological regression analysis and extreme case analysis. The experts examined a number of elections within Attala County in which black candidates opposed white candidates and generated an estimated level of support for black candidates by white voters and by black voters. Ecological regression analysis explains how voting and turnout correspond to the proportions of whites and blacks in each voting precinct. Extreme case analysis looks separately at the actual votes black candidates received in the most heavily black and white precincts to give a clear indication of the preference of voters for either black or white candidates in the two types of districts. The experts applied ecological regression analysis to show an average level of support for black candidates of 87% among black voters versus only 15% among white voters. According to plaintiffs' experts, these numbers jibe with the results of the extreme case analysis and demonstrate racial bloc voting.

At the first hearing, the district court was not convinced by the plaintiffs' evidence, concluding that "the current district plans neither dilute black voting strength nor deny blacks equal access to the political process in violation of § 2 of the Voting Rights Act.

Plaintiffs appealed the judgment. A panel of this court remanded on the issue of vote dilution. It found that the district court's findings on this claim were incomplete and required a fuller analysis of the statistical evidence.

On remand plaintiffs presented the testimony of another expert, Dr. Richard Engstrom. He examined three black-white post-remand elections in Attala County using ecological regression analysis. His report supported a finding of racial bloc voting. According to his analysis, support among black voters for black candidates ranged from 86.3% to 92.8% while support among white voters for black candidates ranged from 12% to 25.2%.

The United States offered its own expert, Dr. James W. Loewen, who looked at the same election results as did Dr. Engstrom. Dr. Loewen checked his regression results with complementary overlapping percentages analysis, a form of extreme case analysis.2 Dr. Loewen found that on average in selected contests in Attala County since 1991 black voters supported black candidates by 84.2% while the average of whites supporting black candidates was merely 16.8%.3 On the basis of these figures, Dr. Loewen concluded that "despite the strong political cohesion shown by the black community, the white community demonstrated bloc voting sufficient to usually defeat the blacks' candidates of choice, except in majority black districts."

Defense expert Dr. Ronald E. Weber also used ecological regression analysis and produced results similar to the statistical findings of the plaintiffs' experts. However, Dr. Weber did not use extreme case analysis to check his regression results because there is no supermajority black district comparable to areas where whites make up 80% or more of the voting age population.

Defendants also presented precinct election returns for separate analysis apart from extreme case or regression analysis. The district court identified election returns from two individual precincts, each from a different election, which it considered suggestive of particular instances of crossover voting.

The district court, after looking at the evidence a second time, again found that plaintiffs had failed to prove their § 2 claim. In particular, it found that plaintiffs established neither black political cohesion nor that voters in Attala County cast their votes on the basis of race to the exclusion of other nonracial factors, such as a candidate's experience, qualifications, education, and contact with the electorate.

Stressing language from the Act at 42 U.S.C. § 1973, the district court again concluded that "blacks have the same opportunity to participate in the political process and ... elect representatives of their choice," and that the "political processes leading to the nomination or election in Attala County are equally open to black and white participation." The district court found that voter apathy resulting in reduced registration and turnout is a better explanation for the apparent political impotence of blacks in Attala County than systemic vote dilution. The court pointed to two pieces of evidence to support this conclusion. The first was a showing that, statewide, black and white voter registration rates were comparable. The second was the testimony of defendants' expert, Arthur Whittemore, a retired political science professor, who described a sharp state and national decline in voter participation equally affecting blacks and whites in recent years, which he attributed to increased voter apathy.

The district court also found no evidence in the record to suggest a lack of responsiveness on the part of elected officials to the particularized needs of blacks in Attala County. To arrive at that conclusion, it relied on the testimony of several "political veterans" that black support is important to electability in Attala County.

In summary, the district court found that blacks do indeed have the ability to participate in the political process and elect candidates of their choice equal to the ability possessed by the rest of the electorate. The district court decided after its re-review of the evidence that no one factor is outcome determinative and that the current districting plan neither dilutes black voting strength nor denies blacks equal access to the political process in violation of § 2 of the Voting Rights Act.

The ultimate finding of vote dilution is a question of fact subject to the clearly erroneous standard of Federal Rule of Civil Procedure 52(a). Thornburg v. Gingles, 478 U.S. 30, 78-79, 106 S. Ct. 2752, 2780-81, 92 L. Ed. 2d 25 (1986). Under that standard, "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, 1511, 84 L. Ed. 2d 518 (1985) (quoting United States v. United States Gypsum Co., 333 U.S. 364, 395, 68 S. Ct. 525, 541-42, 92 L. Ed. 746 (1948)). The analysis the district court must make to evaluate a § 2 claim is fact-intensive, requiring "an intensely local appraisal of the design and impact of the contested electoral mechanisms." Gingles, 478 U.S. at 79, 106 S.Ct. at 2781. That the legal standard the district court must apply to the facts of the claim may come into play does not change the standard of review. Id.

The Supreme Court first interpreted § 2 of the Voting Rights Act after its 1983 amendment in Thornburg v. Gingles. According to the Court, "[t]he essence of a § 2 claim is that a certain electoral law, practice, or structure interacts with social and historical conditions to cause an inequality in the opportunities enjoyed by black and white voters to elect their preferred representatives." Gingles, 478 U.S. at 47, 106 S.Ct. at 2764. A violation has three elements: (1) the minority group must be sufficiently large and compact to constitute a majority in a single member district, (2) the minority group must be politically cohesive, and (3) it must be shown that the white majority votes sufficiently as a bloc to enable it usually to defeat the minority's preferred candidate. Id. at 50-51, 106 S.Ct. at 2766-67. The Supreme Court has held that the Gingles factors apply equally to challenges to single-member districts and challenges to multi-member districts. Growe v. Emison, 507 U.S. 25, 39-41, 113 S. Ct. 1075, 1084, 122 L. Ed. 2d 388, 404 (1993); Voinovich v. Quilter, 507 U.S. 146, 156-58, 113 S. Ct. 1149, 1157, 122 L. Ed. 2d 500, 514 (1993). Failure to establish all three elements defeats a § 2 claim. The first threshold factor of compactness/numerousness under Gingles is not disputed in this case.

Even if plaintiffs satisfy these prerequisites, the court must still look to the "totality of the circumstances" to determine whether the challenged electoral system is equally open to minority voters. Johnson v. De Grandy, 512 U.S. 997, ----, 114 S. Ct. 2647, 2657, 129 L. Ed. 2d 775 (1994).

Section 2(a) of the Voting Rights Act prohibits any "voting qualification or prerequisite to voting or standard, practice, or procedure ... which results in a denial or abridgement of the rights of any citizen of the United States to vote on account of race or color[.]" 42 U.S.C. 1973(a). Section 2(b) provides:

A violation of subsection (a) is established if, based on the totality of the circumstances, it is shown that the political processes leading to nomination or election in the State or political subdivision are not equally open to participation by members of a class of citizens protected by subsection (a) of this section in that its members have less opportunity than other members of the electorate to participate in the political process and to elect representatives of their choice.

42 U.S.C. 1973(b).

The existence of racial bloc voting pertains to a vote dilution claim in two ways. Bloc voting by blacks tends to prove that the black community is politically cohesive. In other words, it shows that blacks as a group tend to support the same candidate as they would elect if voting in a single member, black majority district. Naturally, however, politically cohesive voting behavior will not elect the minority's candidate of choice if the majority group is similarly politically cohesive. Bloc voting by a white majority tends to prove that blacks will generally be unable to elect representatives of their choice. Gingles, 478 U.S. at 68, 106 S.Ct. at 2775. The United States, as appellant, argues that the district court erred in ruling that voting in Attala County was not racially polarized. In particular it faults the district court for requiring plaintiffs to prove affirmatively that voting there was based on race to the exclusion of other nonracial factors.

In Teague I, the district court found that plaintiffs' and defendants' expert testimony on racial bloc voting patterns varied "considerably." Teague I, 807 F. Supp. at 402. However, the court contradicted itself on remand based on much of the same data and analysis when it said that the defendants' expert yielded results similar to plaintiffs' experts. It criticized the plaintiffs' experts' methodology with respect to extreme case analysis because their analysis did not include any precincts with a 90% black voting age population. Finding a district with a black voting age population to measure against a district with a comparably homogeneous white voting age population is ideal for such analysis. However, the most heavily black precinct in Attala County is the Northeast precinct in District 2, which has only a 71.81% black voting age population. Plaintiffs' expert Cheri McKinless used only precincts with at least 80% white majority voting age population as subjects for her extreme case analysis. As in Teague I, the court did not give much weight to extreme case analysis that did not include a homogeneously black precinct, despite the fact that such a district does not exist in Attala County. Still the court accepted the evidence of racial polarization from the regression analysis and allowed that given the "strong indication of white bloc voting using the extreme case analysis ... there can be no doubt that the statistical evidence in the record favors a finding of racial bloc voting in Attala County."

But the court chose to go beyond those findings and looked to two elections where there was evidence of black crossover voting. These were the same two elections the court discussed in Teague I. The first was the 1991 first primary election for sheriff in which there were five candidates--four white and one black. In the Sallis precinct of District 4, the existing black super majority district, the court found "substantial black crossover voting which resulted in a white candidate's victory." In that precinct the black voting age population numbered 564 and the white voting population was only 353. The court reasoned that if all white voters cast their ballots for the white candidates, then at least 232 black voters must have supported white candidates. The second election was the constable east district primary election of August 1987 where the black candidate, Michael Lee, received 96 votes from the McCool precinct. If all 96 of his votes had come from black voters, the court surmised, based on the black voting age population, black voter turnout would have had to have been around 88.1%. Given the historically low black voter turnout, the court found this rate to be incredible and more likely approximating the 12% level plaintiffs' own expert estimated. At that level, the black candidate would clearly have had to receive votes from whites in the precinct. According to the court, "[l]ack of black political cohesion ... accounted for his loss, not white animosity." The district court counted these two anecdotes against the statistical evidence of polarized voting in Attala County: "Although the court does not hold that the results in these two elections alone disprove plaintiffs' case, ... these two elections provide compelling evidence that voters in Attala County are not driven by race."

The Supreme Court in Gingles said that the results of a couple of elections do not discount the presence of racial bloc voting: "[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, "a showing that bloc voting is not absolute does not preclude a finding of racial polarization." Gingles stressed that, to be legally significant, racial polarization need only be the "usual[ ]" pattern for a significant proportion of voters. Id. Vote dilution is a determination that must be made over time and over the course of many elections. How many elections must be studied to make this determination depends on the particular circumstances of the locale. Id. at 57 n. 25, 106 S. Ct. at 2769 n. 25. However, we have no trouble concluding that the district court's assessment of vote dilution based on two precincts in two elections is insufficient to overcome the admittedly compelling statistical evidence of racial bloc voting in Attala County.

The language from Gingles quoted above directly undercuts the district court's reliance on the Sallis and McCool precinct returns. From those discrete examples of voting in two particular precincts in two different elections, the district court deduced that "voters in Attala County are not driven by race." This finding is clearly erroneous. The contrary statistical evidence is considerable, to say the least. In the case of the 1991 sheriff election, Michael Lee could not realistically be considered a serious candidate given the paucity of support (less than 10%) he received from any segment of the voting population. This election should not even have been examined because standard methodological procedure excludes races involving such marginal candidates. Campos v. City of Baytown, Tex., 840 F.2d 1240, 1245 n. 7; Citizens for a Better Gretna v. City of Gretna, La., 834 F.2d 496 (5th Cir.1987), cert. denied, 492 U.S. 905, 109 S. Ct. 3213, 106 L. Ed. 2d 564 (1989).

Plaintiffs' expert Cheri McKinless found sharp differences in the voting behavior of blacks and whites in Attala County and a distinct pattern of racial polarization between these groups. She analyzed eight elections ranging in office from district level supervisor to president of the United States and found little support for black candidates by white voters. In fact, in not one of the elections she analyzed was there any significant white voter support for the black candidate and in only two elections did that support exceed 25%. For the remaining six elections the support by whites is less than 15%. Overall, an average of 15% support is given to the black candidates by the white voters; whereas, 85% of the white voters preferred white candidates.

In all eight elections McKinless examined blacks supported the black candidate at a significantly higher rate than did whites. In addition, a majority of the blacks chose the black candidate in every election. In all but one of the elections, the support given to the black candidate was 75% and above. On average, the black support for the black candidate was 87% compared to 13% support given to white candidates. Taken together these percentages indicate strong divisions along racial lines in the selection of candidates. Black candidates have realized electoral success against white candidates only in black majority districts. This consistent lack of black electoral success suggests that majority white districts may not provide black voters in Attala County with an equal opportunity to elect their candidates of choice, according to Loewen.

Through ecological regression analysis, Dr. Loewen demonstrated "overwhelming" white bloc voting: An average of 83.2% of all whites who voted in six elections surveyed voted for white candidates. Conversely, blacks were politically cohesive as well: An average of 84.2% of their ballots went to black candidates. In these elections, all since 1991, the proportion of white voters who voted for black candidates was never above 28% and averaged less than 17%. Black voters gave black candidates an average of 84.2% of their votes; nonetheless, black candidates had little hope for success except in majority black districts, owing to the white racial bloc voting. Thus, despite the strong political cohesion shown by the black community, the white community demonstrated bloc voting usually sufficient to defeat the blacks' candidates of choice, except in majority black districts.

Dr. Loewen also calculated a high correlation between voting behavior and the racial composition of the districts. This relationship is expressed by a correlation coefficient, r, a statistical measure of the closeness of the fit between an independent variable (the racial composition of the election precinct) and a dependent variable (the level of support for a particular candidate). A high correlation coefficient justifies confidence in the underlying data. Correlation coefficients range in value from -1.00 to +1.00, from a perfectly inverse relationship to a perfectly positive one. The square of the correlation coefficient indicates the proportion of the variance that is associated with the independent variable.

Dr. Loewen figured the correlation coefficients and r2s for the six black-white elections in Attala County after 1991 for which they could be computed. All six rs are above 0.5, and all but one are 0.9 or above, allowing extreme confidence in the underlying data and the ultimate regression estimates. Similarly the r2s are high. All but one are above 0.8, and four of the six are above 0.9, indicating that race is the primary factor in the variation in the vote.

In the face of this compelling evidence the district court persisted in denying the racial basis of voting in Attala County. It repeated its statement in Teague I, 807 F. Supp. at 403, that the election analysis of the experts failed to prove racial polarization and that "[f]actors such as a candidate's experience, qualifications, education and contact with the electorate are of much greater significance than race to voters in Attala County." The district court opined that the record in this case is devoid of any proof that other factors such as these are as or more important than race in predicting voting behavior in Attala County. According to the court, "Statistical analyses do not encompass other factors and variables that provide further insight into voting behaviors and patterns. Statistics examine how, rather than why, people vote the way they do."

The district court's view disregards the established acceptance of regression analysis as a standard method for analyzing racially polarized voting. Gingles, 478 U.S. at 52-53 n. 20, 106 S. Ct. at 2767 n. 20. Furthermore, by this reasoning, the district court errs by placing the burden on plaintiffs to disprove that factors other than race affect voting patterns in Attala County. Plaintiffs are to present evidence of racial bias operating in the electoral system by proving up the Gingles factors. Defendants may then rebut the plaintiffs' evidence by showing that no such bias exists in the relevant voting community. Nipper v. Smith, 39 F.3d 1494, 1524 (11th Cir.1994). The district court betrays its misunderstanding of this burden-shifting when it concludes that "factors other than race influence voters in Attala County, and therefore, plaintiff has failed to prove racial polarization" on the basis of there being "no proof in the record that in any way compares the candidates on any basis other than race." Such a showing is for the defendants to make. The lack of evidence in the record on this point favors the plaintiffs, not the defendants. By the district court's own reasoning, then, the plaintiffs' have prevailed in demonstrating racial bloc voting.

The district court concluded its discussion on racial bloc voting by citing anecdotal evidence from the trial to support its opinion that factors other than race influence voters in Attala County. To that end, Troy Hodges, supervisor in District 3 in commenting on his experiences campaigning in the black area of Kosciusko testified that "[r]ace won't keep you from being elected," and that "blacks won't vote for a candidate just because [the candidate] is black or white." The district court also mentioned Alderman H.L. Myrick, who is black and who credited his success to significant white crossover voting as well as financial and campaign help from white citizens. He decried the assertion that Attala County voters mark their ballots on the basis of skin color instead of the candidate and his qualifications. Myrick did admit, however, on cross-examination, that whereas white and black volunteers together assisted a white candidate in a contest against another white candidate, no whites campaigned for black candidates.

Largely on the basis of this testimony, the district court found that black and white crossover voting has a significant impact on election outcomes in Attala County. The court found that "[a]lthough the statistical evidence may weigh in favor of a finding of racial polarization," this evidence is insufficient, failing "to establish that voters in Attala County cast their vote on the basis of race, to the exclusion of other nonracial factors, such as a candidate's experience, qualifications, education and contact with the electorate."Plaintiffs countered with the testimony of Jesse Fleming, a former supervisor from black majority District 4. Fleming testified that within Attala County, voting so severely splits along racial lines that black candidates cannot win elections unless they run in a majority black district. Fleming supplemented this testimony with accounts of racially hostile encounters he and other black candidates received from whites.

Plaintiffs presented additional lay testimony indicating the presence of voting along racial lines. Defense witness Charles England, a former County Chancery Clerk, admitted on cross-examination that it is easier for a candidate to become popular with a voter if the candidate and the voter are of the same race. England acknowledged that churches, private clubs, social organizations, and other aspects of social life in Attala County are racially segregated, and that as a result, personal familiarity is associated with race.

"The district court is not obliged to accept statistical evidence as conclusive on the question whether racially polarized voting exists." Teague, 17 F.3d at 798. Still, the district court does have an obligation to give particularized findings when it discredits the statistical evidence, especially where the evidence is weighty. And it is weighty here. Indeed, the district court did not even have to choose among competing opinions based on the statistics. The results of the regression analyses of both plaintiffs' and defendants' experts were consistent. And when the district court looked to lay testimony, even the anecdotal evidence was conflicting.

This is not to say that the district court rejected the statistical evidence outright, or that it questioned the accuracy of the numbers. Indeed, the court concluded that "there can be no doubt that the statistical evidence in the record favors a finding of racial bloc voting in Attala County." The court expressed its opinion, with which we agree, that the statistics do not capture all the reasons motivating a voter. The difficulty is that the district judge then refused to accord any presumption to the virtually unchallenged statistical case and placed the additional burden on the plaintiffs to prove that "racial polarization, experience, qualifications, education, and contact with the electorate" were not the true explanations for the voting patterns. As the district court put it, "Conspicuously missing from the record in this cause is any proof to the contrary. Of the black-white elections analyzed, there is no proof in the record that in any way compares the candidates on any basis other than race." In the face of a strong statistical case that defendants were unable to shake, general statements that race played no role at the polls carry little weight, and a district court errs by adjusting the plaintiffs' burden of proof on the basis of such testimony. In short, the district court placed upon the plaintiffs the insurmountable burden of coming forward with evidence disproving all nonracial reasons that can explain election results in spite of the fact that the defendant had itself produced no real evidence that factors other than race were at work.

This court remanded for a more thorough discussion by the district court of the statistics which made up the principal evidence plaintiffs offered. But the district court chose to skirt this charge by averring that "this evidence alone is not sufficient." Without substantially more in the way of particularized findings contradicting the numbers, the district court may not totally discount methodology that both the Supreme Court and this court have cited with approval. Gingles, 478 U.S. at 52-54, 106 S.Ct. at 2767-68; Campos, 840 F.2d at 1246 n. 9. Doing so was in disregard of the remand order.

After a careful and complete review of the trial court record, we find overwhelming evidence of racial polarization. The results of the statistical analyses in this case create a strong presumption in favor of a finding of black political cohesion and racial bloc voting. As in Teague I, the district court, on remand, did not accept this presumption. We hold that finding to constitute clear error. Even relying on the anecdotal evidence, the district court should have found it ambiguous at best and certainly not dispositive. Rather the court apparently chose to accept some and ignore others. And while the court does have discretion in the manner in which it weighs the facts, the statistical evidence here is so one-sided as to require a more convincing discussion of the lay testimony before refuting the objective results. The district court failed to give the serious consideration to the statistical evidence for which this case was remanded. Although there may be cases in which even proof of the three Gingles factors may not sustain a vote dilution claim, this is certainly not such a case. We find that the plaintiffs have successfully carried the burden on their vote dilution claim.

Again, a finding of the three Gingles preconditions does not end the inquiry. Reviewing courts are to look beyond the Gingles threshold factors when evaluating vote dilution claims. Nipper, 39 F.3d at 1513. As the Supreme Court said, "Lack of electoral success is evidence of vote dilution, but courts must also examine other evidence in the totality of circumstances, including the extent of the opportunities minority voters enjoy to participate in the political processes." De Grandy, 512 U.S. at ----, 114 S. Ct. at 2657. A defendant may try to rebut plaintiffs' claim of vote dilution via evidence of "objective, nonracial factors under the totality of the circumstances standard." Nipper, 39 F.3d at 1513.

The Voting Rights Act as amended invalidates any "standard, practice, or procedure ... 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 ..." It says that a denial or abridgment occurs where,

based on the totality of the 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) in that its members have less opportunity than other members of the electorate to participate in the political process and to elect representatives of their choice. The extent to which members of a protected class have been elected to office in the State or political subdivision is one circumstance which may be considered: Provided, That nothing in this section establishes a right to have members of a protected class elected in numbers equal to their proportion in the population.

42 U.S.C. § 1973(b).

Gingles says that the reviewing court is to be flexible in its totality inquiry and guided by factors drawn from the Senate Judiciary Committee report on the 1982 amendments to the Voting Rights Act and reference Zimmer v. McKeithen, 485 F.2d 1297 (5th Cir.1973). Gingles, 478 U.S. at 44, 106 S.Ct. at 2762-63. These 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.

Id. at 36-37, 106 S.Ct. at 2758-59 (quoting S.Rep. No. 417, 97th Cong., 2d Sess. 28-29 (1982), reprinted in 1982 U.S.C.C.A.N. 177, 205-06). Additional factors that may be probative in some cases as part of plaintiffs' evidence establishing a voting rights violation are8. 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.

Id. at 37, 106 S.Ct. at 2759. The district court on remand made the same totality determination as it did in Teague I. The court still found that Attala County's electoral system does not violate § 2 and that blacks have the same opportunity as whites to participate in Attala County's electoral system.

The district court held that evidence of past discrimination and present socioeconomic disparities between blacks and whites does not predominantly explain voting patterns in Attala County. These factors, it concluded, while they do exist and have a significant impact, still do not tip the scales in favor of a § 2 vote dilution violation. The court repeated its explanation that other factors such as voter apathy have resulted in low voter registration and turnout among blacks in Attala County. In its opinion, plaintiffs have failed to prove that past discrimination presently affects voting behavior in Attala County and, even if it does, that that impact is greater than many other influences.

The court next looked to the arguments that socioeconomic disparities between black and white voters have had an impact on political opportunity. Again the court credited the defendants' argument that voter apathy explains the depressed level of political participation by blacks in Attala County. Quoting Teague I, the court found:

Rather than unequal opportunity, the source of minority candidate defeat can be traced to the failure to energize an apathetic electorate. Voter apathy afflicts society in general, transgressing every socio-economic level and knowing no barriers. Political victories are the fruit of organizing, energizing and mobilizing an electorate to turn out and vote for a particular candidate.

The court cited the testimony of defendants' trial expert, Arthur Whittemore, who attributed sharp declines in voter participation and voter turnout by both blacks and whites from 1988 to 1990 to increased voter apathy statewide and nationally.

The district court found no evidence in the record that elected officials are unresponsive to the "particularized needs" of blacks relative to whites in Attala County. In support of this finding, the court looked to the lay testimony described earlier from county polls testifying to the necessity of black support for winning elections.

The court found no one factor determinative of blacks' ability to participate in the political process, and that in the totality of the circumstances blacks in Attala County have just as much opportunity to participate in the political process and elect candidates of their choosing as do whites in the county. This finding by the district court is clearly erroneous.

In Clark v. Calhoun County, 21 F.3d 92 (5th Cir.1994), this court held:

[I]t will be only the very unusual case in which the plaintiffs can establish the existence of the three Gingles factors but still have failed to establish a violation of § 2 under the totality of circumstances. In such cases, the district court must explain with particularity why it has concluded, under the particular facts of that case, that an electoral system that routinely results in white voters voting as a bloc to defeat the candidate of choice of a politically cohesive minority group is not violative of § 2 of the Voting Rights Act.

Id. at 97. Attala County is not that unusual case. The district court did not take adequate notice of the considerable statistical evidence of vote dilution. The conclusion it drew in its totality discussion is similarly dubious, relying entirely on the denials of a few lay witnesses and ignoring the contrary testimony of others. The district court did not place this testimony in the context of the area's history of voter exclusion nor give voice to any plausible nonracial explanation for Attala County's voting patterns.

That Mississippi has a long and dubious history of discriminating against blacks is indisputable. Furthermore, as the district opined on remand, there is no question that grave socioeconomic disparities exist between blacks and whites in Attala County. Still, what must be shown is that the effects of past discrimination impede the ability of blacks in Attala County to participate in the political process. League of United Latin Am. Citizens v. Clements, 999 F.2d 831, 866 (5th Cir.1993). Plaintiffs are not required to prove a causal connection between these factors and a depressed level of political participation. Quoting the Senate report, this court noted in LULAC that:

The courts have recognized that disproportionate educational, employment, income level and living conditions arising from past discrimination tend to depress minority political participation. Where these conditions are shown, and where the level of black participation in politics is depressed, plaintiffs need not prove any further causal nexus between their disparate socio-economic status and the depressed level of political participation.

Id. at 867 (quoting S.Rep. No. 417, 97th Cong., 2d Sess. 29 n. 114, reprinted in 1982 U.S.C.C.A.N. 177, 207 n. 114).

Dr. Loewen's analysis of elections since 1991 showed that black participation is lower than white participation in Attala County as did the report of defendants' expert Dr. Weber. This disparity in voting parallels racial disparities in registration. Dr. Loewen attributed these disparities to socioeconomic gaps between blacks and whites in Attala County. He calculated that 95.9% of the white voting age population was registered compared to 76.1% of the black voting age population.

Dr. Loewen discussed how these depressed levels of political participation correlate to blacks' socioeconomic status. He reported that black families are more than 3.5 times as likely to be poor than are white families. Almost 47% of black families live below the poverty level whereas that figure for whites is around 13%. Roughly three out of every four families with annual incomes below $5,000 are black. Sixty percent of the families making $5,000 to $10,000 are also black. The median family income for whites is almost twice the amount for black families.

Employment figures help explain why blacks fare so poorly economically in Attala County. The unemployment rate in the black community is 12.9%, nearly twice that in the white community. Among the employed, one-third of white workers are in white collar occupations. Admittedly, this is a low figure by national standards; but it nevertheless is three times the rate for black workers. Dr. Loewen built upon these socioeconomic indicators to surmise that blacks, being more heavily concentrated in wage-earning positions, do not have the same level of flexibility come election day to leave their jobs and vote as salaried employees typically do. These blue collar workers may have to wait until the end of the day when lines at the polls are longest and are therefore discouraged from making the commitment to wait it out to cast their ballot.

Finally, Dr. Loewen offered the differences in educational achievement between blacks and whites in Attala County as an explanation for socioeconomic disparities. He reported that whites are almost twice as likely as blacks to have graduated from high school. Whites are 2 1/2 times more likely to have graduated from college.

The district court determined that voter apathy influenced election results in Attala County largely on the basis of testimony from defendant's trial expert, Whittemore, who testified regarding a sharp decline in voter participation and voter turnout by both blacks and whites from 1988-1990. To conclude that black voter apathy is the reason for the failure of blacks to elect the candidates of their choice when apathy affects all voters is counterintuitive. The fact that blacks and whites in Attala County are going to the polls in decreasing proportions does not explain why blacks alone are essentially shut out of the political processes of the county.

The considerable evidence of the socioeconomic differences between black and white voters in Attala County argues against the district court's reiteration that black voter apathy is the reason for generally lower black political participation. The presence of voter apathy is not a matter for judicial notice. Kirksey v. Board of Supervisors, 554 F.2d 139, 145 (5th Cir.) (en banc), cert. denied, 434 U.S. 968, 98 S. Ct. 512, 54 L. Ed. 2d 454 (1977). The record before us does not contain evidence to support the district court's conclusion that voter apathy is the reason for the failure of blacks in Attala County to elect the candidates of their choice to political office. See United States v. Dallas County Comm'n, 739 F.2d 1529, 1536 (11th Cir.1984).

We are not denying that the ultimate inquiry of Section 2 is racial discrimination; nor are we holding that a defendant cannot offer evidence of the nonracial reasons for the voting patterns. In LULAC we did not hold that the plaintiff has the burden of negating all nonracial reasons possibly explaining plaintiffs' statistical case. That case concerned a different problem--the role of partisan politics.

For the foregoing reasons we find that the district court clearly erred in finding that the 1983 supervisory and justice court judge redistricting plans for Attala County do not diminish minority voting power in violation of § 2 of the Voting Rights Act, as amended. The case is remanded for the development and implementation of a remedial plan to cure the vote dilution caused by the existing redistricting plans and for a determination of what amount, if any, the plaintiffs are entitled to recover in court costs and attorneys' fees. REVERSED, RENDERED, AND REMANDED.