BATCHELDER J., delivered the opinion of the court, in which CLAY, J., joined. NATHANIEL R. JONES, J. (pp. 844-47), issued a separate concurring opinion.
OPINION
BATCHELDER, Circuit Judge.Don Sundquist, Governor of the State of Tennessee, and other state officials appeal from a district court order finding that a districting plan for the Tennessee House of Representatives unlawfully dilutes African-American voting strength in violation of § 2 of the Voting Rights Act, 42 U.S.C. § 1973, and enjoining further use of that plan. The State further requests a stay of the district court’s order for elections scheduled to begin in August of 2000. For the reasons set forth below, we affirm the order of the district court, and deny the motion for a stay as moot.
I
In April of 1992, the Tennessee General Assembly enacted legislation reapportioning the State’s single-member House of Representatives and Senate districts. Tenn.Code Ann. §§ 3-1-102 & 103 (1992) (repealed 1994). Prior to the 1992 primaries, the Rural West Tennessee African-American Affairs Council (“RWTAAC”) and certain registered voters in Tennessee filed suit charging that both the Senate Plan and the House Plan violated § 2 of the Voting Rights Act and the Thirteenth, Fourteenth, and Fifteenth Amendments to the United States Constitution. Phillip R. Langsdon and other registered voters in Tennessee filed suit challenging the validity of the House Plan on the grounds that it violated § 2 of Voting Rights Act and the “one person, one vote” doctrine of the Fourteenth Amendment; the plaintiffs also challenged the House Plan on other grounds subsequently dismissed.
A three-judge panel of the district court convened and ordered the two cases consolidated. On September 15, 1993, the panel held that the House Plan was unconstitutional because it violated the one person, one vote doctrine of the Equal Protection Clause. RWTAAC v. McWherter, 836 F.Supp. 447, 452 (W.D.Tenn.1993). The court ordered the defendants to prepare and submit a constitutional apportionment plan by January 25, 1994. Id. The State appealed to the United States Supreme Court.
On November 4, 1993, the district court ruled that the Senate Plan violated § 2 of the Voting Rights Act by affording African-American voters in west Tennessee less opportunity than other members of the electorate to participate in the political process and to elect representatives of their choice. RWTAAC v. McWherter, 836 F.Supp. 453, 466 (W.D.Tenn.1993) (“RWTAAC /”). The State appealed, and the Supreme Court vacated the panel’s order and remanded for further consideration in light of Johnson v. DeGrandy, 512 U.S. 997, 114 S.Ct. 2647, 129 L.Ed.2d 775 (1994). RWTAAC v. McWherter, 512 U.S. 1249, 114 S.Ct. 2776, 129 L.Ed.2d 888 (1994). On remand, the district court reversed its decision in RWTAAC I and held that the Senate Plan conformed to the Voting Rights Act. RWTAAC v. McWhert*838er, 877 F.Supp. 1096, 1098 (W.D.Tenn.1995) (“RWTAAC II”). The plaintiffs appealed.
While the appeals in the Senate and House Plan cases were pending, the Tennessee General Assembly adopted a new House Plan and submitted it to the district court. Tenn.Code Ann. § 3-1-103 (1994). The court found that the new Plan complied with the Equal Protection Clause’s one person, one vote requirement. It delayed consideration of the other challenges to the House Plan until the Supreme Court ruled on the appeals pending in the Senate case.
The Supreme Court affirmed the dismissal of the plaintiffs’ claims as to the Senate Plan on October 4, 1995. RWTAAC v. Sundquist, 516 U.S. 801, 116 S.Ct. 42, 133 L.Ed.2d 9 (1995). RWTAAC then amended its complaint to challenge the House Plan on the sole ground that it violated § 2 of the Voting Rights Act. Because the amended complaint contained no constitutional claims, the three-judge court disbanded itself. After a trial on the merits of the plaintiffs’ consolidated claim of vote dilution, the Honorable Jerome Turner, on November 6, 1998, declared the 1994 House Plan to be violative of the Voting Rights Act and enjoined the defendants from using it in future elections. RWTAAC v. Sundquist, 29 F.Supp.2d 448, 450 (W.D.Tenn.1998). The state defendants bring this timely appeal.
II
“A district court’s factual findings regarding Section 2 violations and the determination of whether vote dilution has occurred are ordinarily reviewed for clear error.” Cousin v. McWherter, 46 F.3d 568, 574 (6th Cir.1995) (citing Fed.R.Civ.P. 52(a) and Thornburg v. Cingles, 478 U.S. 30, 79, 106 S.Ct. 2752, 92 L.Ed.2d 25 (1986)). However, “Rule 52(a) does not inhibit an appellate court’s power to correct errors of law, including those that may infect a so-called mixed finding of law and fact, or a finding of fact that is predicated on a misunderstanding of the governing rule of law.” Cingles, 478 U.S. at 79, 106 S.Ct. 2752 (internal quotation marks omitted).
A
Section 2 of the Voting Rights Act provides:
(a) No voting qualification or prerequisite to voting or standard, practice, or procedure shall be imposed or applied by any State or political subdivision in a manner which results in a denial or abridgement of the right of any citizen of the United States to vote on account of race or color, or in contravention of the guarantees set forth in section 1973b(f)(2) of this title, as provided in subsection (b) of this section.
(b) A violation of subsection (a) of this section is established if, based on the totality of circumstances, it is shown that the political processes leading to nomination or election in the State or political subdivision are not equally open to participation by members of a class of citizens protected by subsection (a) of this section in that its members have less opportunity than other members of the electorate to participate in the political process and to elect representatives of their choice. The extent to which members of a protected class have been elected to office in the State or political subdivision is one circumstance which may be considered: Provided, That nothing in this section establishes a right to have members of a protected class elected in numbers equal to their proportion in the population.
42 U.S.C. § 1973.
Before considering whether the House Plan dilutes minority voting strength in rural west Tennessee and thus denies members of the minority group a fair opportunity to elect representatives of their choice, we must determine whether the plaintiffs have met the three preconditions announced by the Supreme Court in Thornburg v. Gingles, 478 U.S. at 50-51, 106 S.Ct. 2752. The plaintiffs must dem*839onstrate that 1) the minority group is sufficiently large and geographically compact to constitute a majority in a single-member district; 2) the minority group is politically cohesive; and 3) the white majority votes sufficiently as a bloc to enable it usually to defeat the minority’s preferred candidate. Id. The parties agree that the first two Gingles preconditions have been satisfied. They clash on the question whether whites vote as a bloc usually to defeat the candidates of choice of African-American voters, contesting most hotly the weight to be given to particular sets of election data.
B
Chapter 536 of the Public Acts of 1994 provides a three-part reapportionment plan for Tennessee’s ninety-nine house districts. Plan A, at issue in this case, creates twelve majority-African-American districts. None of these districts, however, lies in the area that plaintiffs describe as rural west Tennessee, which includes Madison, Haywood, Hardeman, Tipton, Fayette, and Lauderdale counties. Chapter 536 provides that should a court find that Plan A unlawfully dilutes minority voting strength, Plan B, which creates thirteen majority-African-American house districts, including one in rural west Tennessee, will take effect. Plan C, which would have reinstated the 1992 house redistricting plan had the State prevailed on its claim that the 1992 plan complied with the Equal Protection Clause, is moot. See Millsaps v. Langsdon, 510 U.S. 1160, 114 S.Ct. 1183, 127 L.Ed.2d 534 (1994) (affirming the district court’s ruling that the 1992 house redistricting plan violated the “one person, one vote” doctrine).
At trial, both parties presented expert testimony regarding voter behavior in rural west Tennessee. Plaintiffs’ expert, Dr. Steven Cole, analyzed eleven legislative elections which had both black and white candidates from 1974 to 1996 using bivariate ecological regression analysis, and found that the black preferred candidates lost nine times (82%). Average black cohesion was 67%, and average white cohesion was 89%. An analysis of homogenous white precincts for legislative contests from 1994 to 1996 showed that on average 80% of whites voted for the white candidates over the black candidates. There were no homogenous black precincts. Based on this evidence, Dr. Cole concluded that white voters tend to vote as a bloc so as usually to defeat the candidate of choice of African-American voters.
The State’s experts, Dr. William Lyons and Dr. Michael Gant, analyzed ten black-white legislative contests in rural west Tennessee from 1986 to 1996, and found that the minority — preferred candidates had been defeated nine times (90%). Average black cohesion was 64%, and average white cohesion was 86%. Because black voter turnout was roughly equal to white voter turnout in black-white elections and in elections that had only white candidates, Drs. Lyons and Gant also analyzed eleven white-white legislative contests from 1986 to 1996. Adding the results to those from the black-white contests, Drs. Lyons and Gant found that the preferred candidate of black voters was defeated eleven out of twenty-one times (52.38%).
While Drs. Lyons and Gant acknowledged that the minority-preferred candidate usually lost, and hence that the results of state legislative contests in rural west Tennessee were indicative of minority vote dilution, they noted that a difference in one election would change this outcome. They therefore examined twenty-four black-white and twenty-six white-white countywide elections from 1986 to 1996. In the black-white contests, they found that the black-preferred candidate was defeated thirteen times (54%). For white-white countywide elections, the black-preferred candidate lost six times (23%). Combining all the countywide contest from 1986 to 1996, the black-preferred candidate lost eighteen out of fifty times (36%). Noting that if all of the countywide elections were combined with all of the legislative elections, the results showed that the black-preferred candidate lost only twenty-*840nine of seventy-one elections (41%), the defense experts concluded that election results in rural west Tennessee were not, in general, indicative of minority vote dilution.
After hearing this testimony, the district court held that the plaintiffs had made a showing sufficient to satisfy the third Crin-gles precondition. The district court placed primary emphasis on the results of the black-white contests and, due to the significant number of legislative elections analyzed, determined that the results from the countywide contests were not particularly useful or necessary. In light of its limited probativeness, the district court concluded, the white-white and countywide evidence was insufficient to overcome the fact that a white voting bloc defeats the’ minority-preferred candidate at least 82% of the time in interracial legislative elections in rural west Tennessee.
On appeal, the State argues that the district court erred. The State contends, in essence, that the district court was obliged to give equal and controlling weight to white-white and non-legislative contests in' its analysis under the’ third Gingles factor.
C
This court has made clear that white-white elections are relevant in the analysis of a voting dilution claim. In Cousin v. Sundquist, we considered a § 2 challenge to the at-large method of electing judges utilized by Hamilton County, Tennessee. Cousin, 145 F.3d 818, 820 (6th. Cir.1998). The plaintiffs’ expert used average cohesion figures from black-white elections in connection with voter turnout information to determine that, in order to succeed in a Hamilton County election, a hypothetical black candidate would need a number of white crossover votes exceeding the average crossover suggested by the cohesion figures. Id. at 824. The defendants’ expert, by contrast, used both black-white and white-white election results, identified the minority’s preferred candidate in each, and determined that the white majority did not regularly vote in such a way as to deprive black voters in Hamilton .County of the opportunity to elect their candidate of choice. Id. In preferring the methodology of the defense expert, we noted: “The proper inquiry is not whether white candidates do or do not usually defeat black candidates, but whether minority-preferred candidates, whatever their race, usually lose.” Id. at 825.
While the plain import of Cousin is that courts are not foreclosed from considering electoral races involving only white candidates, that case does not suggest (as the State seems to argue) that white-white contests are necessarily entitled to the same weight as those involving a minority candidate. As Judge Richard Arnold has pithily stated, the Voting Rights Act’s guarantee of equal opportunity is not met when “[cjandidates favored by blacks can win, but only if the candidates are white.” Smith v. Clinton, 687 F.Supp. 1310, 1318 (E.D.Ark.1988) (three judge panel). This court, along with others, has accordingly held that a candidate’s race can be relevant to a § 2 inquiry under certain circumstances. See Clarke v. City of Cincinnati, 40 F.3d 807, 812 (6th Cir.1994); see also, e.g., Nipper v. Smith, 39 F.3d 1494, 1540 (11th Cir.1994) (en banc) (holding that white-white elections may be considered, but are less probative than those involving black candidates). One such circumstance occurs when white bloc voting is “targeted” against black candidates. Clarke, 40 F.3d at 812.
In this case, there is marked evidence of targeting; the experts for both the plaintiffs and the defendants agreed that white voter cohesion in rural west Tennessee increases from 59% in white-white elections to 86% in black-white elections. Perhaps not unrelatedly, no African-American candidate has ever won an interracial legislative contest in the six-county area, despite many candidacies. In view of such evidence that a white voting bloc coalesces to frustrate African-American candidacies, the district court properly considered the race of candidates in its § 2 *841analysis, and accorded greater weight to the results of black-white elections.
Similarly, the district court was correct to discount the results of countywide contests. The parties stipulated that legislative elections were the most legally relevant, and the rulings of our sister circuits support the parties’ appraisal. See, e.g., Citizens for a Better Gretna v. City of Gretna, 834 F.2d 496, 502 (5th Cir.1987) (suggesting that exogenous elections alone cannot prove racially polarized voting, but can be considered as “additional evidence”). At trial, data were presented from twenty-one legislative elections over a period of ten years — a substantial body of evidence. See Jenkins v. Red Clay Consol. Sch. Dist. Bd. of Educ., 4 F.3d 1103, 1130 (3d Cir.1993) (holding that seven elections over a period of ten years was a suitable sample such that the court could discern the presence of a pattern of white bloc voting usually defeating the minority voters’ candidate of choice). The State nevertheless advocates the inclusion of the countywide elections on the ground that a different outcome in one of the eleven legislative elections in which the preferred candidate of African-Americans was defeated would alter the minority vote dilution determination. But the record contains no information that would allow us to evaluate the claimed marginal character of the statistics derived from the legislative contests. More importantly, the State’s claim rests on the assumption that black-white and white-white legislative elections should be given equal weight. As we have pointed out, however, that assumption is not valid in this case.
In sum, we find no clear error in the district court’s determination that the plaintiffs have satisfied the third Gingles precondition. The data from the legislative elections were sufficiently robust for the district court to discern whether there existed a pattern of white bloc voting. When the black-white legislative elections are afforded greater weight, the data show that the white majority votes sufficiently as a bloc to enable it usually to defeat the minority’s preferred candidate in rural west Tennessee.
III
Having determined that the plaintiffs have met the three Gingles preconditions, we turn to the question whether, given the “totality of the circumstances,” the House Plan has in fact “diluted” African-American electoral strength and thus denied African-Americans in rural west Tennessee a fair opportunity to elect representatives of their choice. See Clarke, 40 F.3d at 811. The Senate Report which accompanied the 1982 amendments to the Voting Rights Act specifies factors which typically may be relevant to a § 2 claim: the history of voting-related discrimination in the State or political subdivision; the extent to which voting in the elections of the State or political subdivision is racially polarized; the extent to which the State or political subdivision has used voting practices or procedures that tend to enhance the opportunity for discrimination against the minority group, such as unusually large election districts, majority vote requirements, and prohibitions against bullet voting; the exclusion of members of the minority group from candidate slating processes; the extent to which minority group members bear the effects of past discrimination in areas such as education, employment, and health, which hinder their ability to participate effectively in the political process; the use of overt or subtle racial appeals in political campaigns; and the extent to which members of the minority group have been elected to public office in the jurisdiction. Gingles, 478 U.S. at 44-45, 106 S.Ct. 2752 (citing S.Rep. No. 97-417, at 28-29 (1982), U.S. Code Cong. & Admin. News at 177, 205-207). The Report notes also that there may be probative value to evidence demonstrating that elected officials are unresponsive to the particularized needs of the members of the minority group and that the policy underlying the State’s or the political subdivision’s use of the contested *842practice or structure is tenuous. Id. at 45, 106 S.Ct. 2752.
The Supreme Court has recently highlighted a new element — proportionality'—■ to be weighed in the totality of the circumstances. In Johnson v. De Grandy, a case involving a challenge to Florida’s legislative reapportionment plan, the Court concluded that § 2 relief should not be granted because, notwithstanding the presence of continued discrimination and racial bloc voting, minority voters were able to form effective voting majorities in a number of legislative districts that were roughly proportional to their respective shares in the voting age population. De Grandy, 512 U.S. 997, 1024, 114 S.Ct. 2647, 129 L.Ed.2d 775 (1994). The Court emphasized, however, that proportionality is not a “safe harbor:” “the degree of probative value assigned to proportionality may vary with other facts. No single statistic provides courts with a shortcut to determine whether a set of single-member districts unlawfully dilute minority voting strength.” Id. at 1020-21, 114 S.Ct. 2647.
A
The district court found that the totality of the circumstances indicated that the House Plan unlawfully dilutes minority voting strength in rural west Tennessee. Beginning with the two factors that the Supreme Court has declared are the most important in balancing the totality of the circumstances, see Gingles, 478 U.S. at 48 n. 15, 106 S.Ct. 2752, the court found that no African-American had been elected to the state legislature from the six-county area, and that the experts at the trial agreed that voting in the area is racially polarized. The court next took judicial notice of findings from the Senate Plan cases, RWTAAC I & II, regarding the history and effect of discrimination in voting in western Tennessee. The court in the Senate Plan cases had recounted the entire history of official discrimination from the pre-Civil War era, a time-frame whose use we disapproved of in the Cousin case, 145 F.3d at 882, but went on to cite two cases from the 1980s which indicate that voting rights violations by public officials in rural west Tennessee are ongoing. See Taylor v. Haywood County, 544 F.Supp. 1122, 1131 (W.D.Tenn.1982) (holding that change to an at-large election scheme was a result of purposeful intention to dilute black voting strength); Bills v. Alexander, No. 83-12220 (W.D.Tenn.1983) (approving settlement setting up a new system to ensure the opportunity of black citizens to participate meaningfully in the political process). The court concluded that as a result of those ongoing violations, African-Americans had suffered disadvantages in such areas as education, employment, and health.
The district court further found that the Tennessee House of Representatives was responsive to the needs of black voters in the rural western part of the state, a factor upon which this court has laid heavy emphasis in the past. See Cousin, 145 F.3d at 833. The district court questioned the state policy of maintaining municipal boundaries wherever possible in configuring the legislative districts, since the plan itself fractured certain cities. Under the heading of “miscellaneous” factors, the court concluded that there was no evidence of suspect electoral practices, racial appeals in political campaigns, or a slating process or other mechanism used to prevent minority candidacies.
Turning to the question of proportionality, the district court noted that blacks make up 31.01% of the voting age population of the six counties comprising rural western Tennessee, but that none of the five house districts covering that area contains a majority of black voters. The court acknowledged that in four of the five districts minority members make up 25-55% of the population and hence could meaningfully affect election outcomes in those four districts, but concluded that in the absence of a single majority black district, this fact had little probative significance. Balancing the lack of proportionality with the other factors from the Senate Report, the court concluded that black vot*843ers in rural west Tennessee do not have equal opportunity in the political process.
The State’s primary quarrel with this determination concerns the geographical area on which the court focused in conducting the proportionality analysis.1 The State argues that the court should have considered proportionality in the entire state or, in the alternative, in an area that includes Shelby County along with the six counties of rural west Tennessee. On a statewide basis, there are twelve majority black house districts out of a total of ninety-nine, which, the State contends, is substantially proportional to Tennessee’s black voting age population of 14.4%. In the seven-county area that includes Shelby County, the black voting age population is 37.9%; since 42.85% of the house districts are majority black in that seven-county area, the State says, blacks are in fact overrepresented.
B
In Johnson v. De Grandy, the Supreme Court explicitly left open the question of the proper frame of reference for analyzing § 2 claims. De Grandy, 512 U.S. at 1022, 114 S.Ct. 2647 (“[W]e have no occasion to decide which frame of reference should have been used if the parties had not apparently agreed in the District Court on the appropriate geographical scope for analyzing the alleged § 2 violation and devising its remedy.”). Nevertheless, the reasoning in that case, and in Shaw v. Hunt, 517 U.S. 899, 917, 116 S.Ct. 1894, 135 L.Ed.2d 207 (1996), persuades us that neither over-proportionality in one area of the State nor substantial proportionality in the State as a whole should ordinarily be used to offset a problem of vote dilution in one discrete area of the State.
In De Grandy, the State of Florida argued that, as a matter of law, no dilution occurs whenever the percentage of single-member districts in which minority voters form an effective majority mirrors the minority voters’ percentage of the relevant population. De Grandy, 512 U.S. at 1017, 114 S.Ct. 2647. Critiquing this “safe harbor” argument, the Supreme Court remarked on the State’s “unexplored premise of highly suspect validity:”
that in any given voting jurisdiction (or portion of that jurisdiction under consideration), the rights of some minority voters under § 2 may be traded off against the rights of other members of the same minority class. Under the State’s view, the most blatant racial gerrymandering in half of a county’s single-member districts would be irrelevant under § 2 if offset by political gerrymandering in the other half, so long as proportionality was the bottom line.
Id. at 1019, 114 S.Ct. 2647. Similarly, in Shaw, the Court considered North Carolina’s argument that a bizarrely shaped majority-black congressional district was a narrowly tailored remedy for a § 2 violation elsewhere in the State. Shaw, 517 U.S. at 916-17, 116 S.Ct. 1894. Finding this position “singularly unpersuasive,” the Court stated:
If a § 2 violation is proved for a particular area, it flows from the fact that individuals in this area “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 vote-dilution injuries suffered by these persons are not remedied by creating a safe majority-black district somewhere else in the State....
... To accept that the district may be placed anywhere implies that the claim, and hence the coordinate right to an undiluted vote (to cast a ballot equal among voters), belongs to the minority as a group and not to its individual members. It does not.
*844Id. at 917, 116 S.Ct. 1894. Taken together, the admonitions of De Grandy and Shaw dissuade us from accepting the Tennessee’s invitation to append Shelby County, or the State as whole, to the geographical frame of reference that the plaintiffs have selected because to do so would require us to trade the § 2 rights of individual African-Americans in rural west Tennessee against those of African-American groups elsewhere in the State.
The State complains that by allowing the plaintiffs to define the frame of reference for their § 2 claim, we will enable future litigants to carve up successively smaller areas of the State until they are able to maximize the number of majority-minority legislative districts — a result not countenanced by the Voting Rights Act. See De Grandy, 512 U.S. at 1017, 114 S.Ct. 2647 (“Failure to maximize cannot be the measure of § 2.”). As the district court pointed out, however, the Gingles preconditions operate to prevent just the sort of limit-lessly small “reverse gerrymander” whose specter the State raises here. See, e.g., Campos v. City of Houston, 113 F.3d 544, 547-48 (5th Cir.1997) (holding that minority group was not sufficiently large and geographically compact to sustain a § 2 claim). In this regard, we note that the region selected by the plaintiffs in this case is a sensible one — indeed, more sensible than the seven-county area including Shelby County urged by the State. While Shelby County is the southernmost and westernmost county in the State, and, like the six neighboring counties of rural west Tennessee, has a large African-American population, the seven counties do not form a coherent demographic unit. The African-American population in Shelby County is concentrated in inner-city Memphis, and is largely set off from rural west Tennessee by a “buffer zone” of white suburbs. As a result, the African-American populations in these two areas are not likely to be particularly cohesive. For this reason, the district court properly restricted the geographic scope of relevant statistical data to the six counties of rural west Tennessee.
Having correctly defined its frame of reference, the district court made no clear error in weighing the totality of the circumstances. Rural west Tennessee has an African-American voting age population of 31%, but none of its five house districts is majority-African-American. Legislative elections in the six county area are racially polarized, and no African-American has ever won one. These circumstances simply overwhelm those factors that might favor the State of Tennessee (such as a lack of suspect electoral practices, responsive (albeit mainly white) state legislators, and the possibility that minority voters in minority-African-American districts will be able to influence electoral outcomes). The district court therefore properly held that the plaintiffs had proved a § 2 violation. The remedy that the plaintiffs seek for this violation, creation of one majority-African-American house district in rural west Tennessee through implementation of Plan B of Chapter 536, will take place as a matter of state law.
IV
The late Judge Turner ably considered a complex body of statistical and anecdotal evidence to determine that Plan A of Chapter 536 unlawfully dilutes African-American voting strength in rural west Tennessee. His order enjoining use of the House Plan in future elections is hereby affirmed. The defendants’ motion for a stay pending appeal is denied as moot.
. The State also challenges the district court’s finding regarding the history of official discrimination in voting in western Tennessee. We have carefully considered this question, and find no error in the district court’s determination.