In 1986, the Supreme Court first construed amended § 2 of the Voting Rights Act of 1965 (VRA), Pub.L. No. 89-110, 79 Stat. 437, codified at 42 U.S.C. §§ 1973--1973%io, to conclude several North Carolina legislative districts impaired the opportunity of black voters "to participate in the political process and to elect representatives of their choice." Thornburg v. Gingles, 478 U.S. 30, 106 S. Ct. 2752, 92 L. Ed. 2d 25 (1986); § 2(b); 42 U.S.C. § 1973(b). Thus having launched a decade of challenges to multi-member election districts and at-large election procedures, Gingles now presides over a new round of VRA lawsuits involving the configuration of single-member districts, the 1990 census having necessitated redrawing numerous state and congressional districts. In this case, plaintiffs, Jennie Sanchez, Adeline Sanchez, and Debra Casanova, challenged the reconfiguration of their state legislative district, Colorado House District 60 (HD 60), drawn after the 1990 census, contending the resulting single-member district did not provide Hispanic voters in south central Colorado a fair opportunity to elect candidates of their choice. However, the district court denied them declaratory and injunctive relief concluding the Gingles' quantum of proof was unmet. Sanchez v. State of Colo., 861 F. Supp. 1516 (D.Colo.1994). With the guidance of two recently decided Supreme Court cases, Bush v. Vera, --- U.S. ----, 116 S. Ct. 1941, 135 L. Ed. 2d 248 (1996), and Shaw v. Hunt, --- U.S. ----, 116 S. Ct. 1894, 135 L. Ed. 2d 207 (1996) (Shaw II ),2 and Gingles' governing principles, we conclude plaintiffs have established under the totality of circumstances the presently configured HD 60 unlawfully dilutes the voting strength of Hispanic voters in south central Colorado. We therefore reverse.

Plaintiffs, Jennie Sanchez and Adeline Sanchez, are Hispanic residents and registered voters in Center, Colorado, located in Saguache County. Plaintiff, Debra Casanova, resides in the city of Alamosa, Colorado, in Alamosa County. Saguache and Alamosa along with Conejos, Costilla, Mineral, and Rio Grande Counties comprise the San Luis Valley (the Valley or SLV) in south central Colorado, a flatland whose western boundary is the San Juan Mountains. On its eastern boundary are the Sangre de Cristo Mountains, its peaks towering 14,000 feet, and the La Garita Mountains and Cochetopa Hills overlap to hem in its northern borders. Principally an agricultural area, the Valley encompasses a mosaic of range, pasture, and crop lands and boasts bumper potato harvests, ranking it fifth in potato production nationally. Water issues percolate through all of this economy, the Valley sitting upon one of the nation's largest aquifers, attracting populous urban areas in and out of Colorado which threaten to divert this vital resource to distant paper mills and suburban lawns.

The Valley embraces one of three of Colorado's largest native born Hispanic populations, its Hispanic residents tracing their history back to the original settlement of the Hispano Homeland which comprised extensive Mexican land grants in northern New Mexico and southern Colorado.3 Some Hispanic families in the SLV date back to the 1800's when their ancestors first settled there. Although many of these early landowners were later dispossessed of their land, substantial numbers remained in the Valley, lured away to northern cities like Pueblo and Denver when the local agrarian economy failed to provide the wages city jobs offered. Presently, Pueblo and Denver support similarly large, native-born Hispanic populations. Indeed, Hispanics comprise the largest ethnic minority in Colorado, representing 12.5% of its total population. Colorado Vital Statistics 1990, p.18.

In the wake of the decennial federal census, the Colorado Reapportionment Commission (the Commission) was reconvened to redraw the boundaries of state senate and house legislative districts to reflect the 14% increase in the state's population from 2,889,735 to 3,294,394. Operating under Colorado Constitutional mandate4 and federal law, the politically neutral Commission5 prioritized the criteria governing their redistricting: first, the equal population of districts with no more than a 5% deviation between the largest and smallest to maintain the principle of one person, one vote; and, second, compliance with the VRA. "Of secondary importance," the Commission's Final Report listed, "are preservation of county boundaries, the preservation of municipal lines, and the formation of compact districts," while "a third level" criterion was the preservation of communities of interest. Of final "unofficial" importance is the preservation of politically competitive districts. Final Report of the Colorado Reapportionment Commission, April 1992, p. 4.

In preparation for redistricting, the Commission retained Election Data Services (EDS), a Washington, D.C.-based research and consulting firm which analyses election behavior, to examine state legislative races to determine "if legally significant racial bloc voting exists and to ascertain the percentage minority population necessary in a district for minority voters to have the opportunity to elect candidates of choice." In its first report, EDS identified two state districts with 40 and 50% Hispanic populations which appeared insufficient to elect minority-preferred candidates, the Weld County area (House Districts 50 and 51); and the southeast, south central part of the state (House Districts 43, 60, and 63). EDS concluded, based on preliminary evidence, "[i]t may be necessary to create districts that have higher concentrations of Hispanics in them in these two areas of the state." Id. at 6. Subsequently, the Commission requested EDS perform additional racial bloc voting analyses of local elections in each of the counties comprising the Valley except for Mineral County and the surrounding counties, Archuleta, Huerfano, Las Animas, and Prowers.

In its second report to the Commission, EDS observed voting in the Valley "appear[ed] to be racially polarized ... in almost every election contest examined in which a Hispanic candidate competed in 1988 and 1990." Because EDS found in countywide elections studied approximately 20% Anglos cross over to vote for the Hispanic preferred candidate, it recommended the State did not have to create super-majority districts to compensate for what otherwise would be completely polarized voting. Consequently, given the slightly lower turnout of minority voters and its finding Hispanic voting cohesiveness, EDS concluded, "[i]t is necessary to create districts that are more heavily Hispanic in the San Luis Valley than elsewhere in the state because of the degree of racially polarized voting found in this area of the state."

Before drafting the final plan, the Commission held hearings throughout the state, in particular, Alamosa, Pueblo, and Trinidad, south central Colorado cities which would be affected by a possible reconfiguration of HD 60. Typical of such localized public fora, a range of opinions voicing hometown and area concerns was aired: whether the Valley should be split; whether the urban interests of Trinidad and Pueblo differed from the Valley's agricultural concerns; and how reconfiguration would impact incumbent representatives to the state legislature. With little disagreement, SLV citizens urged the Commission not to split the Valley after publication of a preliminary redistricting plan proposed splitting the Valley to increase the Hispanic vote. Support from the SLV chapter of the Hispanic League6 to keep the Valley in one district was, in fact, elicited by the SLV County Commissioner Association which, in return, publicly condemned racial bloc voting in the Valley and resolved to "establish policies that ensure that protected minorities have every opportunity to fully participate in the political process."

Nevertheless, the Valley encompasses only about 80% of the population necessary to comprise a district under Colorado law.7 The decision then whether to go west or east, whether to include the city of Trinidad with its large Hispanic population, or to carve into part of Pueblo, 40% of whose population is Hispanic, confronted the Commission when it convened on four occasions to weigh the alternative districts proposed for the Valley. Despite the EDS data on racial bloc voting and recommendation to create an Hispanic majority district; the awareness of the lack of success of Hispanic candidates from the SLV seeking election to the state legislature; and the concern about VRA liability, the Commission ultimately adopted the alternative which neither split the Valley nor touched Trinidad or Pueblo, preserving the incumbents' seats,8 and added part of Las Animas and all of Huerfano Counties. Except for Las Animas County, no counties are split. However, the Continental Divide slices through HD 60's western boundary, placing part of Mineral and Saguache Counties on the Divide's western slope, while Las Animas and Huerfano Counties sit on the eastern side of the Sangre de Cristo Mountains. Of the estimated 45.41% Hispanic population in the resulting HD 60, 42.38% comprise the Hispanic voting age population, a 5% increase in Hispanic voters from the prior HD 60. After this redistricting, voters in HD 60 reelected Lewis Entz, the incumbent Anglo representative, to the state legislature in 1992, maintaining the Anglo control of this seat that has existed since 1940.

Various objectors, including plaintiffs in this action, challenged the Final Plan in the Colorado Supreme Court. However, although it agreed with certain objections and remanded for corrections, the Colorado Supreme Court concluded the Final Plan satisfied constitutional criteria. In re Colorado General Assembly, 828 P.2d 185 (Colo.1992).9 Subsequently, plaintiffs filed this lawsuit against the state of Colorado, members of the Commission, and other state officials (the State, collectively) challenging the legality of HD 60 under § 2 of the VRA. They alleged the 1992 configuration of HD 60 disenfranchised Hispanic voters by permitting the white majority to vote as a bloc to defeat Hispanic-preferred candidates. Alleging Hispanic voters in the area constitute a sufficiently compact population and are politically cohesive, plaintiffs asked the district court to declare HD 60 violates § 2; permanently enjoin the State from conducting further elections for representative to the Colorado General Assembly from HD 60; and order the creation of a new HD 60 to protect Hispanic rights under § 2. After a five-day trial to the court, the district court concluded plaintiffs failed to establish Gingles' three preconditions and did not demonstrate based on a totality of circumstances HD 60 violates § 2(b). We address the rudiments of this conclusion in abecedarian order.

Gingles not only furnishes the substantive basis for review of cases under the VRA but also clarifies the proper standard of appellate review. Generally, the Court advises treating "the ultimate finding of vote dilution as a question of fact subject to the clearly-erroneous standard of Rule 52(a)." Gingles, 478 U.S. at 78, 106 S.Ct. at 2780. That is, we shall not disturb this finding unless there is no evidence in the record to support it; or, if there is some evidence substantiating it, we are, nevertheless, after reviewing all the evidence "left with a definite and firm conviction that a mistake has been made." Zimmerman v. Sloss Equip., Inc., 72 F.3d 822, 825 (10th Cir.1995) (citations omitted). However formidable a standard, Uno v. City of Holyoke, 72 F.3d 973, 978 (1st Cir.1995), it is not insurmountable, as Gingles counsels, for it "does not inhibit an appellate court's power to correct errors of law, including those that may infect a so-called mixed finding of law and fact, or a finding of fact that is predicated on a misunderstanding of the governing rule of law." Gingles, 478 U.S. at 79, 106 S.Ct. at 2781 (quoting Bose Corp. v. Consumers Union of U.S., Inc., 466 U.S. 485, 501, 104 S. Ct. 1949, 1960, 80 L. Ed. 2d 502 (1984)). Subjecting the trial court's legal analysis to plenary review "assures the appropriate legal standards are applied," Jenkins v. Red Clay Consol. School Dist. Bd. of Educ., 4 F.3d 1103, 1117 (3d Cir.1993), cert. denied, 512 U.S. 1252, 114 S. Ct. 2779, 129 L. Ed. 2d 891 (1994), balances the trial court's "intensely local appraisal" of these fact-intensive cases, and best "preserves the benefit of the trial court's particular familiarity with the indigenous political reality without endangering the rule of law." Gingles, 478 U.S. at 79, 106 S.Ct. at 2781. Thus, while we do not defer to an erroneous interpretation of the law, neither will we substitute our view of the evidence for the alternative choice the fact-finder made. Sanchez v. Bond, 875 F.2d 1488, 1495 (10th Cir.1989) (citing Anderson v. City of Bessemer City, N.C., 470 U.S. 564, 573-74, 105 S. Ct. 1504, 1511-12, 84 L. Ed. 2d 518 (1985)), cert. denied, 498 U.S. 937, 111 S. Ct. 340, 112 L. Ed. 2d 305 (1990).

As amended, § 2(a), 42 U.S.C. § 1973(a), declares:

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)10 of this title, as provided in subsection (b) of this section.

After the Supreme Court later interpreted § 2 to require plaintiffs prove the contested electoral practice was intentionally adopted and maintained by state officials for a discriminatory purpose, City of Mobile v. Bolden, 446 U.S. 55, 100 S. Ct. 1490, 64 L. Ed. 2d 47 (1980), Congress amended its language, codifying the "results tests" which the courts had applied before Bolden, and clarifying that plaintiffs need only prove the contested practice has a discriminatory effect. This amendment, § 2(b), reads:

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(b). As enacted, Section 2(b) promises protected minorities an even playing field, not a certain victory.

The Senate Report accompanying the 1982 amendments elaborates the proof for a § 2 violation, specifying a "variety of relevant factors, depending upon the kind of rule, practice, or procedure called into question." S.Rep. No. 417, 97th Cong., 2d Sess. 28-29 (1982), reprinted in 1982 U.S.C.C.A.N. 177, 206-07.11 The Senate Report added that "there is no requirement that any particular number of factors be proved, or that a majority of them point one way or the other." Id. at 207. Moreover, "[w]hile the enumerated factors will often be pertinent to certain types of § 2 violations, particularly to vote dilution claims, other factors may also be relevant and may be considered." Gingles, 478 U.S. at 45, 106 S.Ct. at 2763 (citation omitted) (footnote omitted). Most importantly, "the question whether the political processes are 'equally open' depends upon a searching practical evaluation of the 'past and present reality,' and on a 'functional' view of the political process." Id. (quoting S.Rep. No. 417, 1982 U.S.C.C.A.N. 177, 208). The lack of electoral opportunity is the key. "The 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." Id. at 47, 106 S.Ct. at 2764.

and its Progeny

Gingles shorthands the totality of circumstances proof provided in the Senate Report recognizing "the conjunction" of three circumstances which, as preconditions, must be present to establish a vote dilution claim. Indeed, their presence creates the inference the challenged practice is discriminatory. First, minority plaintiffs must prove their group is "sufficiently large and geographically compact to constitute a majority in a single-member district." Id. at 50, 106 S.Ct. at 2766. Second, plaintiffs must show the minority group is "politically cohesive." Id. at 51, 106 S.Ct. at 2766. Third, they must demonstrate "the white majority votes sufficiently as a bloc to enable it--in the absence of special circumstances, such as the minority candidate running unopposed--usually to defeat the minority's preferred candidate." Id. While these three conditions are necessary to establish a vote dilution claim, they are not sufficient. Johnson v. De Grandy, 512 U.S. 997, ----, 114 S. Ct. 2647, 2657, 129 L. Ed. 2d 775 (1994).

Under De Grandy, a court's examination of relevant circumstances is not complete "once the three factors were found to exist, or in the sense that the three in combination necessarily and in all circumstances demonstrated dilution." Id. This is so "because the ultimate conclusions about equality or inequality of opportunity were intended by Congress to be judgments resting on comprehensive, not limited, canvassing of relevant facts." Id. Failure to prove the totality of circumstances establishes the minority is not harmed by the challenged practice and rebuts the inference of discrimination arising from proof of the three preconditions. Uno, 72 F.3d at 980. In Growe v. Emison, 507 U.S. 25, 113 S. Ct. 1075, 122 L. Ed. 2d 388 (1993), the Court expressly applied the Gingles' paradigm to a vote dilution claim against a single member district.

A. "Sufficiently large and geographically compact"

The first question, whether the minority group is "sufficiently large and geographically compact to constitute a majority in a single-member district," simply asks whether any remedy is possible in the first instance. As Gingles noted:

The reason that a minority group making such a challenge must show, as a threshold matter, that it is sufficiently large and geographically compact to constitute a majority in a single-member district is this: Unless minority voters possess the potential to elect representatives in the absence of the challenged structure or practice, they cannot claim to have been injured by that structure or practice.

478 U.S. at 50 n. 17, 106 S. Ct. at 2766 n. 17 (emphasis in original). As a corollary, if the minority group is small and dispersed, no single member district could be created to remedy its grievance.

Hence, the first prerequisite asks about "the existence of a legally cognizable injury." NAACP, Inc. v. City of Niagara Falls, N.Y., 65 F.3d 1002, 1011 (2d Cir.1995). As such, this element of proof assists a court in finding "a reasonable alternative practice as a benchmark against which to measure the existing voting practice." Holder v. Hall, 512 U.S. 874, ----, 114 S. Ct. 2581, 2585, 129 L. Ed. 2d 687 (1994).12 "The inquiries into remedy and liability, therefore, cannot be separated: A district court must determine as part of the Gingles threshold inquiry whether it can fashion a permissible remedy in the particular context of the challenged system." Nipper v. Smith, 39 F.3d 1494, 1530-31 (11th Cir.1994), cert. denied, --- U.S. ----, 115 S. Ct. 1795, 131 L. Ed. 2d 723 (1995).

Because Gingles advances a functional evaluation of whether the minority population is large enough to form a district in the first instance, the Circuits have been flexible in assessing the showing made for this precondition.

The first Gingles precondition does not require some aesthetic ideal of compactness, but simply that the black population be sufficiently compact to constitute a majority in a single-member district. Moreover, plaintiffs' proposed district is not cast in stone. It was simply presented to demonstrate that a majority-black district is feasible.... If a § 2 violation is found, the county will be given the first opportunity to develop a remedial plan.

Clark v. Calhoun County, Miss., 21 F.3d 92, 95 (5th Cir.1994) (citations omitted); see also Houston v. Lafayette County, Miss., 56 F.3d 606, 611 (5th Cir.1995) ("Compactness is not as narrow a standard as the district court construed it to be.").

Geographical compactness, then, does not implicate constitutional principles. "The Constitution does not mandate regularity of district shape," the Court recently stated in Bush v. Vera, --- U.S. at ----, 116 S. Ct. at 1953 (citing Shaw v. Reno, 509 U.S. 630, 647, 113 S. Ct. 2816, 2826-27, 125 L. Ed. 2d 511 (1993)) (Shaw I ). Instead, Justice Kennedy clarified, "The first Gingles condition refers to the compactness of the minority population, not to the compactness of the contested district." Id. at ----, 116 S.Ct. at 1971, (Kennedy, J., concurring). Stating the concept in a different way, we must determine whether the affected minority is diffused and thus politically ineffective, not whether the area by which it is bound is geographically dense.

Nevertheless, as we shall see in this case, and as demonstrated recently in Shaw II and Bush, "compactness is the conceptual point at which the tension between the traditional American commitment to territorial districting and the VRA concern for fair representation of group interests must be resolved." Richard H. Pildes & Richard G. Niemi, Expressive Harms, "Bizarre Districts," and Voting Rights: Evaluating Election District Appearances after Shaw v. Reno, 92 Mich. L.Rev. 483, 535 (1993) [hereinafter Expressive Harms ]. And, as Shaw I warned, "appearances do matter," 509 U.S. at 647, 113 S.Ct. at 2827, triggering, as they do, Fourteenth Amendment equal protection scrutiny.

That the minority group demonstrates it is politically cohesive embodies a similarly functional focus. "If the minority group is not politically cohesive, it cannot be said that the selection of a multimember electoral structure thwarts distinctive minority group interests." Gingles, 478 U.S. at 51, 106 S. Ct. at 2766. Like the first Gingles' precondition, however, the Court does not expressly define political cohesiveness. Other courts have elaborated.

In Gomez v. City of Watsonville, 863 F.2d 1407, 1415 (9th Cir.1988), cert. denied, 489 U.S. 1080, 109 S. Ct. 1534, 103 L. Ed. 2d 839 (1989), the Ninth Circuit observed, "[t]he inquiry is essentially whether the minority group has expressed clear political preferences that are distinct from those of the majority." Thus, we judge political cohesiveness by looking at the "voting preferences expressed in actual elections." Id. Necessarily, when we examine the evidence of political cohesiveness as voting preferences, we look to the same statistical evidence plaintiffs must offer to establish vote polarization.13 Indeed, political cohesiveness is implicit in racially polarized voting.14

Gingles adopted a straightforward definition of racial bloc voting provided by the expert witness upon whom the district court had relied. Racial polarization or bloc voting "exists where there is a consistent relationship between the race of the voter and the way in which the voter votes ... or to put it differently, where black voters and white voters vote differently." 478 U.S. at 53 n. 21, 106 S. Ct. at 2768 n. 21 (internal quotation marks omitted). The Court's focus was twofold: to determine whether the minority group votes cohesively and "whether whites vote sufficiently as a bloc usually to defeat the minority's preferred candidates." Gingles, 478 U.S. at 56, 106 S.Ct. at 2769. The extent to which this bloc voting impairs the minority's ability to elect candidates of their choice, however, must be "legally significant," a sliding scale that varies with the district and a variety of factual circumstances and may emerge more distinctly over a period of time. Id. While the Court offered no "simple doctrinal test for the existence of legally significant racial bloc voting," id. at 58, 106 S.Ct. at 2770, it urged a flexible approach, noting that the isolated success of a minority candidate in a district that usually exhibits vote polarization will not alone negate plaintiffs' showing. Thus, while legally significant white bloc voting enables the majority "in the ordinary course, to trounce minority-preferred candidates most of the time," its presence may be more subtle requiring close inquiry over time. Uno, 72 F.3d at 980 (citing Voinovich v. Quilter, 507 U.S. 146, 156, 113 S. Ct. 1149, 1156-57, 122 L. Ed. 2d 500 (1993)).

To determine whether racial bias, in fact, motivated the targeted voting practice, the Court accepted the statistical method necessarily inhered to the definition of racial bloc voting. In Gingles, the district court had relied on expert testimony offered by Dr. Bernard Grofman, who used two methods of analysis of voting patterns, "bivariate ecological regression analysis" and "homogeneous precinct analysis," also called "extreme case analysis." Bivariate ecological regression analysis "determines the degree of relationship between two variables--here the relationship between the racial composition in each political unit (the independent variable) and the support provided a particular candidate within that political unit (the dependent variable)." Jenkins, 4 F.3d at 1119 n. 10. In an ecological regression analysis, the correlation coefficient shows which data points fall on the straight line. The linear relationship created by the two variables ideally then will pack closely together on a line. Extreme case analysis examines the actual voting percentages received by candidates in racially homogeneous precincts. The inferences that arise from the latter analysis are often graphically demonstrated by the former statistical method, the latter providing actual results to demonstrate the estimates. While homogeneous precinct analysis may be a useful check on ecological regression analysis, neither method is without disadvantage, and each, like all statistical evidence, is subject to interpretation.15

However, while the Supreme Court approved the statistical proof provided by bivariate ecological regression and homogeneous precinct analysis, it did not expressly preclude other methods of establishing the presence of racial bloc voting. Nevertheless, if the third Gingles' precondition asks whether whites vote sufficiently as a bloc to enable them usually to defeat the minority candidate, it is asking how voters vote, not why voters voted that way. Indeed, the searching evaluation done in the totality of circumstances perhaps reveals the answer to the latter question. However, at the threshold, we are simply looking for proof of the correlation between the race of the voter and the defeat of the minority's preferred candidate. We do not, therefore, reject multivariate regression analysis but prefer to reserve its use, if at all, to the more global picture plaintiffs must establish.

Before addressing the district court's order, we must reiterate our focus in this case. Plaintiffs contend the current district violates their federal rights guaranteed by the VRA to have a fair opportunity to elect candidates of their choice. While that question predominates our review, it is not exclusively answered by federal law because the Colorado Constitution and state law furnish those "traditional districting principles" of which the Court so frequently speaks. Bush, --- U.S. at ----, 116 S. Ct. at 1952. Even in its most recent opinions, the Court reminded, "we adhere to our longstanding recognition of the importance in our federal system of each State's sovereign interest in implementing its redistricting plan." Id. at ---- - ----, 116 S. Ct. at 1960-61 (citation omitted). "[T]he Constitution leaves with the States primary responsibility for apportionment of their federal congressional and state legislative districts." Growe, 507 U.S. at 34, 113 S.Ct. at 1081, citing U.S. Const. art. I, § 2. Given these principles, however, we must gingerly look to federal law to assure the equality of each vote because achieving "fair and effective representation for all citizens is concededly the basic aim of legislative apportionment." Reynolds v. Sims, 377 U.S. 533, 565-66, 84 S. Ct. 1362, 1383, 12 L. Ed. 2d 506 (1964). Just as the Court has affirmed its unfailing championship of the Fourteenth and Fifteenth Amendments in Shaw II and Bush, it has also declared the constitutionality of § 2 of the VRA, observing, "it would be irresponsible for a State to disregard the § 2 results test." Bush, --- U.S. at ----, 116 S. Ct. at 1969. (O'Connor, J. concurring, joined by Stevens, Ginsburg, Breyer, Souter, JJ.).

In posing the first Gingles' question, the district court misfocused its inquiry. It asked, "Is the Hispanic population sufficiently large and geographically compact to constitute a majority in the Plaintiffs' proposed H.D. 60?" 861 F. Supp. at 1524 (italics added). The district court proceeded to answer its question, looking to facts about the geographical uniqueness of the Valley, its agri-industries, and cities. Against these facts and the various mathematical tests used to determine compactness, the court concluded, based on "principles of contiguity, preservation of communities of interest, and minimizing the splitting of counties and municipalities, the proposed alternative H.D. 60 is not compact." Id. at 1525. It found plaintiffs' alternative split too many counties, was not contiguous, would "erode the number of political offices held by Hispanics in the Valley and in the adjacent districts," and burden representatives attempting to visit such a "large, geographically dispersed district." Id. We have no quarrel with any of these findings. However, they do not address whether plaintiffs have established the minority group is "sufficiently large and geographically compact to constitute a majority in a single-member district." 478 U.S. at 50, 106 S. Ct. at 2766.

As we have noted, satisfaction of the first precondition requires plaintiffs show a majority-Hispanic district is feasible; a remedy is possible. The evidence surely established the Hispanic population is "sufficiently large," the current district including 48.82% Hispanic population. Under Justice Kennedy's definition, the Hispanic population is also compact; for example, representing as much as 76.8% of Costilla County and at least 40% of both Alamosa and Rio Grande Counties. Nonetheless, as stated, the Valley alone does not contain sufficient population to satisfy the federal requirement of one person, one vote. In judging plaintiffs' proposed alternative, which necessarily had to exceed the Valley's bounds, the district court focused only on the shape of the geographical boundaries rather than the size and concentration of the minority population. Houston v. Lafayette County, Miss., 56 F.3d at 611.

When plaintiffs proposed an alternative, Gingles required no more. While that proposal splits more counties and two cities, much of its irregularity results from its maintaining the integrity of precinct lines throughout and observing natural boundaries created by the region's geography. Even under the Colorado Constitution, compactness is not a fixed principle, but a measure of combining like regions and natural boundaries in the shortest aggregate linear distance. Colo. Const. art. V, § 47(1).16 Indeed, the Commission crafted HD 7, which includes the city of Denver extended eastward to wrap around Denver International Airport with no "aesthetic ideal" in mind. Clark, 21 F.3d at 95. Nevertheless, plaintiffs, with minimal time on the State's computer used to draw districts, provided a feasible alternative.17 Under Gingles, plaintiffs satisfied their burden by showing that feasibility. Drawing the necessary district is not their onus because the State must be given the first opportunity to fashion a remedy. Id.18

We would also note by comparison to the districts the Court recently found "bizarre" in Shaw II and Bush, plaintiffs' proposed district is non-objectionable. Again, as Justice Kennedy distinguished, there is no benchmark here against which the proposed alternative district may be measured. Instead, the plaintiffs have demonstrated, as the evidence established, the Hispanic population is sufficiently large and compact to indicate the possibility of a remedy. Notwithstanding the conceptual burdens invested in the analysis, the district court's interpretation of this precondition unfortunately mischaracterized plaintiffs' burden and incorporated considerations and attributes to the requirement that are irrelevant to the inquiry. That is, while compact districting may "facilitate the representation of political communities," Expressive Harms at 501, and we recognize the role of communities of interest in the districting process, at this stage, the district court overburdened plaintiffs' showing with concerns better left for its later analysis. This misdirected view of the law led the district court to erroneously find plaintiffs failed to meet their burden of proving the first Gingles' precondition.

VII. Political Cohesiveness and Racial Bloc Voting

The district court essentially collapsed these two preconditions, intertwining observations about whether Hispanics vote cohesively on some issues with general observations about racial bloc voting. However, as we have stated, both inquiries are rooted in the same statistical evidence offered to show minorities "have expressed clear political preferences." Gomez, 863 F.2d at 1415. We therefore do not fault the conjunctive approach. However, the heart of each inquiry requires a searching look into the statistical evidence to discern the way voters voted. Missing this essential inquiry, the district court chose instead to examine plaintiffs' evidence and superimpose its view of "the reasons for, or causes of, voting behavior." Sanchez, 861 F. Supp. at 1527. Regrettably, those reasons are not relevant at this point.

Without expressly addressing any of plaintiffs' statistical evidence, the district court rejected bivariate ecological regression and homogeneous precinct analyses and embraced the State's proof based on multivariate regression analysis, "a statistical method that helps determine whether one variable--here race--makes an 'independent' contribution to voting decisions once other factors such as newspaper endorsements, incumbency, campaign spending, and the socioeconomic characteristics of the voters are taken into account." See Bernard Grofman, Lisa Handley, & Richard G. Niemi, Minority Representation and the Quest for Voting Equality 83 (1992). In this case, the variable of interest the State's expert selected was the percent of Democratic vote share judged against over forty other variables. In contrast, the bivariate ecological regression analysis presented to the district court used only two variables of interest: the percentage vote for the Hispanic candidate tracked along a vertical axis with the percentage of Hispanic voting age population appearing on the horizontal axis.

To us who are statistics neophytes, bivariate ecological regression analysis identifies the differences while multivariate regression analysis explains the differences. Id. at 100.19 However, "[t]he legal standard for the existence of racially polarized voting looks only to the difference between how majority votes and minority votes were cast; it does not ask why those votes were cast the way they were nor whether there were other factors present in contested elections, such as 'white backlash.' " Collins v. City of Norfolk, Va., 816 F.2d 932, 935 (4th Cir.1987).

Despite these differences, the district court cited no basis in the evidence to validate the use of either methodology or square that evidence with the proper legal focus. Although the court stated it would "not restrict its examination to Dr. Bardwell's bivariate analysis, but will also consider Dr. Zax's multivariate analysis," 861 F. Supp. at 1527, its conclusions made no reference to plaintiffs' statistical proof20 and did not address with any specificity the State's statistical evidence. Although the court referenced lay testimony figuring into its conclusion, it neither identified the lay witnesses nor examined how their testimony established, without more, the absence of racial bloc voting. Instead, in general, the court found "substantial evidence that many factors influence voting behavior and electoral success in H.D. 60." Id.21

At this juncture, it bears reminding Fed.R.Civ.P. 52(a) requires the district court, sitting without a jury, to "find the facts specially." That must be the premise of our clearly erroneous appellate review. Broad and general findings, not explicitly tethered to any particular testimony--especially in the VRA context which demands penetrating case by case, fact bound analysis--simply do not provide the foundation for proper appellate review. Thus, as the court stated in Teague v. Attala County, Miss., 17 F.3d 796, 798 (5th Cir.1994), "in making its intensely fact-specific inquiry here, the district court ought to have discussed appellants' statistical evidence more thoroughly because that was the principal evidence they offered and because their statistics had at least surface plausibility." In this case, the district court did not even address plaintiffs' statistical proof, leaving us little basis for review.

In that absence, we till the same ground. Dr. Bardwell testified he aimed his analysis to address the question whether there is legally significant racial bloc voting in HD 60. To that end, he explained he used a two equation method for ecological regression analysis and a one equation method for the scatter plots.22 This methodology mirrored EDS's although Dr. Bardwell stated he analyzed more elections in the targeted counties. Like EDS,23 he did not include elections in which an Anglo candidate opposed an Anglo candidate based on the presumption Hispanic voters prefer Hispanic candidates.24 In all, he analyzed 53 elections, 46 of which, he testified, evidenced statistically significant polarization. He included the elections for HD 60 representative, primary elections,25 and exogenous elections incorporating the proposed alternative district to examine Hispanic cohesiveness in these non-coincidental jurisdictions.26 His studies concluded in the proposed alternative district, Hispanics voted cohesively.

Of greatest interest are his results for the 1980, 1982, and 1992 elections for HD 60 representative to the Colorado General Assembly. In those "endogenous" races, that is, elections involving the legislative seats on which plaintiffs premise their vote dilution claim, the correlation coefficients, how the data points fall on a line, are 0.83%, 0.88%, and 0.82%, graphically demonstrating on the scatter plots the high correlation between the voter's ethnicity and vote for the Hispanic candidate of choice.27 This analysis revealed 86% Hispanic cohesiveness for the 1980 HD 60 election and 84% Anglo bloc vote; 89% Hispanic cohesiveness in 1982 and 90% Anglo bloc vote; and, in 1992, 89% Hispanic cohesiveness and 83% Anglo bloc vote. Homogeneous precinct analysis largely replicated this graphic picture. A similar analysis done for both the election of then Democrat Ben Nighthorse Campbell, a Native American who ran for the United States Senate in 1992, and for the 1988 Amendment 1 ballot initiative declaring English the official language of Colorado also mirrored this pattern of Hispanic political cohesiveness in HD 60 and its expanded borders.28 For all of the elections studied, Dr. Bardwell found a mean cohesiveness among Hispanic voters of 83% and among Anglos of 71%. Consequently, Dr. Bardwell concluded these figures established "a particularly high level of polarization."29

Dr. Bardwell testified his figures for the white crossover vote were slightly lower than those generated by EDS' studies: he calculated an Anglo crossover vote between 10-17% while EDS, which studied fewer elections, determined a 20% Anglo crossover vote. However, he also observed a predictable Hispanic crossover vote that ranged between 11-14% counterbalanced the Anglo crossover vote. Because of the virtual neutralization of the crossover vote demonstrating equal polarization, Dr. Bardwell concluded a majority district is necessary to give Hispanics the opportunity to elect candidates of their choice.

In contrast, the State's expert, Dr. Jeffrey Zax, testified his goal was to understand why in five counties with Hispanic voting age populations slightly under 40%, there is dramatic variation in the success of Hispanic candidates. Consequently, his focus was to account for the difference in the rate of minority success "explained by something other than the percent of Hispanics." To that end, he utilized the methodology of multivariate regression analysis, believing that voting behavior is a construct of complex factors. His analysis produced five conclusions: (1) heavily Hispanic precincts vote more heavily for Democratic candidates; (2) ethnicity plays a subsidiary role; (3) other factors, incumbency, gender, etc., affect who wins; (4) socioeconomic factors affect the way precincts vote; and (5) the differences in elections arise from differences in platforms, personalities, campaign financing, and often factors difficult to measure. Taken together, Dr. Zax concluded Hispanics vote for Democrats, and Democrats don't beat Republicans in HD 60.

To validate his finding, Dr. Zax relied on seven hypothetical contests, the most pertinent of which predicted that an Hispanic Democrat non-incumbent male will defeat an Anglo Republican non-incumbent male in a run for the current HD 60 seat. To unseat an incumbent Anglo Republican candidate, however, an Hispanic candidate needs a more heavily Hispanic district, he observed. Nevertheless, an Anglo Democrat running against an incumbent Anglo Republican also loses by an even greater percentage, he predicted.

Thus, although his statistical model produced a similar pattern of racial bloc voting, Dr. Zax infused the calculations with different meanings. Indeed, what seems most striking about his analysis is that he used approximately the same numbers to predict outcomes, rather than focusing on the actual elections to examine their outcomes. Based on these predictions, plumbed from the interplay of 49 explanatory variables, Dr. Zax concluded, "political affiliation is the single most important factor that distinguishes between the vote outcomes in precincts that are heavily Hispanics [sic] and heavily Anglo in House District 60. The role of ethnicity, on the other hand, is relatively minor." (italics added). That is, if Anglo Democrats compete against Anglo Republicans and partisanship is the prime determinant, the single variant of ethnicity virtually disappears. The inquiry sidesteps Gingles' primary analytic focus under § 2. Moreover, in this case, the theory does not appear to be supported by present or historical facts.

First, the naked figures demonstrate registered Democrats in HD 60 have a numerical edge in district-wide elections, if partisanship is the prime motivator of voting behavior. The district court found: (1) about 80% of the Hispanic voting age population is registered Democrat; (2) 20% is registered Republican; (3) 56% of the Anglo population is registered Republican; (4) 44% is registered Democrat; and (5) 16% is registered Independent. 861 F. Supp. at 1527-28.30 When the percentage of Hispanic and Anglo Democratic vote is combined, Democrat voters district-wide voting solely along party lines would appear to command a plurality. Clearly, some other factor, however, affects the presumed Anglo and Hispanic Democratic vote. EDS concluded it was ethnicity as did Dr. Bardwell. Dr. Zax disagreed, although he testified, "in a 100% Anglo voting age population precinct, the Anglo candidate will receive a majority of the votes regardless of whether that candidate is a democrat or a republican."31

In addition to relying on the State's multivariate analysis, the district court utilized EDS' figure that approximately 20% of Anglo voters in the Valley cross over to vote for the Hispanic candidate of choice and acknowledged Dr. Bardwell's estimate of 13-17% Anglo cross-over. Adding either of these estimates to 42.38% Hispanic voting age population, the court calculated approximately 60% of the voting age population could be projected to vote for the Hispanic candidate of choice. The court then concluded plaintiffs failed to prove the third Gingles' precondition, that Anglos vote sufficiently as a bloc usually to defeat the minority's preferred candidate. 861 F. Supp. at 1528.

Gingles, however, doesn't require an absolute monolith in the Anglo or Hispanic bloc vote and recognizes the existence and role of white crossover voting. It does ask, though, whether as a practical matter, whites usually vote as a bloc to defeat the minority preferred candidate. In the face of the array of statistical evidence and the discordant interpretations they produced, the historical fact that an Hispanic candidate has not won election to the state legislature from HD 60 since 1940 must also figure into the district court's evidentiary base. See Cane v. Worcester County, Md., 35 F.3d 921, 926 (4th Cir.1994). Thus, whatever the numbers may mean, as a practical matter an Hispanic candidate has never held the HD 60 seat since the district has been drawn to include the Valley. "[T]he lack of success of Hispanic candidates is a strong factor tending to show vote dilution." Sanchez v. Bond, 875 F.2d at 1496.

Moreover, the record does not validate Dr. Zax's predictions. He predicted, based on his analysis of 49 explanatory variables, an Hispanic Democrat non-incumbent male will defeat an Anglo Republican non-incumbent male for the HD 60 seat. In 1982, however, the first time the Valley was included in present HD 60, Republican Lewis Entz, the non-incumbent Anglo candidate, defeated Democrat Alex Marquez, the Hispanic non-incumbent. We believe, therefore, actual outcomes, not predictions of outcomes, provide a more appropriate test.

In concluding partisanship not ethnicity accounted for voting behavior in HD 60, the district court relied on League of United Latin American Citizens Council No. 4434 v. Clements, 999 F.2d 831, 853-54 (5th Cir.1993), cert. denied, 510 U.S. 1071, 114 S. Ct. 878, 127 L. Ed. 2d 74 (1994) (LULAC), where minority plaintiffs contended the Texas system of electing trial judges in countywide elections violated § 2. The case is not apposite. First, it operates in a special context in which the courts have weighed the linkage of the judicial and jurisdictional districts a state draws as an important factor in the totality of circumstances. Second, it was premised on the fact that plaintiffs made no effort to establish racial bloc voting in the first instance, relying instead on what they believed was the uncontrovertible evidence of minority failure at the polls. Given Shaw I 's caution, "it is not mere suffering at the polls but discrimination in the polity with which the Constitution is concerned," 509 U.S. at 661, 113 S.Ct. at 2835 (White, J., dissenting) (citing Whitcomb v. Chavis, 403 U.S. 124, 91 S. Ct. 1858, 29 L. Ed. 2d 363 (1971), and White v. Regester, 412 U.S. 755, 93 S. Ct. 2332, 37 L. Ed. 2d 314 (1973)), LULAC held "[e]lectoral losses that are attributable to partisan politics do not implicate the protections of § 2." 999 F.2d at 863. However, built-in bias at the polls differs from partisan defeat at the polls. And, LULAC actually left for another day the question whether "limiting the racial bloc voting inquiry to a determination whether or not divergent voting patterns are attributable to partisan differences or an underlying divergence in interests best captures the mandate of § 2," recognizing that "partisan affiliation may serve as a proxy for illegitimate racial considerations." Id. at 860. LULAC is, thus, distinguishable and, despite its particular factual context, does not refocus the probativeness of plaintiffs' substantial statistical evidence here.

We believe the district court relied too much on LULAC and not enough on Gingles in its ultimate analysis of plaintiffs' proof. We would also trace its mischaracterizing the evidence of racially polarized voting, in part, to its misperceiving Gingles' position on the minority's candidate of choice. Citing Justice Brennan's discussion in Part III-C in which only three Justices joined, the district court embraced the per se view that the race of the candidate is irrelevant to the racial bloc voting analysis, a view that five Justices rejected. 861 F. Supp. at 1526. In a separate concurrence, Justice White observed that if race was neutralized by removing the candidate's ethnicity from the racial bloc voting equation, and blacks and whites simply voted along party lines, we could substitute interest group politics "as a rule hedging against discrimination" to explain each outcome. 478 U.S. at 83, 106 S.Ct. at 2783 (White, J., concurring). "I doubt this is what Congress had in mind in amending § 2 as it did," Justice White wrote. Id. Justice O'Connor, joined by three other Justices, agreed that defendants cannot rebut a showing of racial bloc voting "by offering evidence that the divergent racial voting patterns may be explained in part by causes other than race, such as an underlying divergence in the interest of minority and white votes." 478 U.S. at 100, 106 S.Ct. at 2792 (O'Connor, J., concurring in the judgment).32 The Gingles' majority, then, concluded the candidate's race is never irrelevant but, generally, is "of less significance than the race of the voter--but only within the context of an election that offers voters the choice of supporting a viable minority candidate." Citizens for a Better Gretna v. City of Gretna, La., 834 F.2d 496, 503 (5th Cir.1987), cert. denied, 492 U.S. 905, 109 S. Ct. 3213, 106 L. Ed. 2d 564 (1989).

Second, while lay testimony is relevant to determine who is the candidate of choice, it is not alone dispositive. In Gingles, the Court acknowledged the district court "credited some testimony of lay witnesses, but relied principally on statistical evidence." 478 U.S. at 52, 106 S.Ct. at 2767. Sanchez v. Bond also approved the use of lay testimony. However, absent the finding the statistical evidence is unreliable, insufficient, or irrelevant, lay testimony should not eclipse the analysis.

Certainly, key to examining racial polarization in the challenged electoral mechanism is determining for whom voters vote. "Ascertaining whether legally significant white bloc voting exists begins with the identification of the minority members' 'preferred candidates' or 'representatives of their choice.' " Collins v. City of Norfolk, Va., 883 F.2d 1232, 1237 (4th Cir.1989), cert. denied, 498 U.S. 938, 111 S. Ct. 340, 112 L. Ed. 2d 305 (1990). Only then can the district court determine whether whites vote sufficiently as a bloc usually to defeat the minority's choice.

Although this court has accepted evidence of Anglo versus Anglo races on the ground such elections may be relevant to discern whether racially polarized voting is present, we cautioned the evidence is useful only "so long as one of the Anglo candidates can be considered a preferred candidate of the minority group." Sanchez v. Bond, 875 F.2d at 1495. Sanchez v. Bond offers no guidance for the district court to determine who is the minority's candidate of choice. However, without that evaluation and with a history of Anglo/Anglo contests, we fall prey to the myopic presumption there is a minority preferred candidate in any race in which the minority votes. While some courts shortcut the question looking only to the candidate who has received more than 50% of the minority vote, NAACP, 65 F.3d at 1019, we now believe some additional direction would be helpful.

Again we turn to Gingles which requires plaintiffs establish by a preponderance of the evidence who is the preferred minority candidate in each election. While "experience does demonstrate that minority candidates will tend to be candidates of choice among the minority," this inference alone is not sufficient to meet plaintiffs' burden. Jenkins, 4 F.3d at 1126. Hence, minority plaintiffs must introduce some additional evidence, anecdotal or otherwise, to satisfy this burden. "The additional evidence required to meet the threshold, however, is not very substantial, and the burden may be satisfied with a variety of evidence, including lay testimony or statistical analyses of voting patterns." Id.

We also embrace Jenkins' guidance to judge defendants' evidence that white candidates were, in fact, minority preferred. The Fourth Circuit counsels in measuring this evidence, "the court must engage in a detailed, practical evaluation of the extent to which any particular white candidate was, as a realistic matter, the minority voters' representative of choice." Id. at 1129. One factor it suggests is the "extent to which the minority community can be said to have sponsored the candidate," to examine minority involvement in originally sponsoring the candidate or helping to finance the candidate's campaign. Id. Another factor is the candidate's attention to the issues concerning minorities; the extent the candidate campaigned in the minority's neighborhoods or addressed predominantly minority crowds. Id. Evidence of minority turnout for the election would also be relevant as well as evidence of "disincentives" for minorities to run for office in the first instance. Id. Another factor is evidence of the Anglo candidate's ties to the minority community. For example, in this case, one of plaintiffs' witnesses, an Anglo elected Saguache County Commissioner, was the minority preferred candidate because, he testified, he is married to an Hispanic who helped him campaign in Hispanic precincts. This list, of course, is not exclusive, and the district court must be sensitive to the quality of the evidence plaintiffs and defendants offer to assist in this factfinding.

Thus, after such a searching evaluation, the district court may find the minority preferred candidate is, in fact, Anglo. However, the answer cannot be shorthanded by the candidate's partisan label or ethnic or racial status. As Justice O'Connor noted, the question arises within the context of determining whether racial bloc voting defeats the minority's opportunity to elect a candidate of its choice. Indeed, the VRA ensures members of a protected class equal opportunity "to elect representatives of their choice," not "necessarily members of their class." NAACP, 65 F.3d at 1015. However, § 2's "guarantee of equal opportunity is not met when '[c]andidates favored by blacks can win, but only if the candidates are white.' " Clarke v. City of Cincinnati, 40 F.3d 807, 812 (6th Cir.1994) (quoting Smith v. Clinton, 687 F. Supp. 1310, 1318 (E.D.Ark.1988) (three-judge panel)).

While any statistical analysis permits a body of data to tell a story, how the story is read, each reader bringing a different focus to the details, alters the theme. The glass, after all, is either half empty or half full. Gingles, however, instructs us to look for the theme of racial polarization and the extent to which that polarization robs the minority of meaningful access to the political process. While that may initially set a rather stark stage, "establishing vote dilution does not require the plaintiffs affirmatively to disprove every other possible explanation for racially polarized voting." Uno, 72 F.3d at 983. Requiring plaintiffs to rebut a showing that partisan politics and not racial bias operates to defeat their § 2 claim may arise in another setting. However, it is not the story here.

We therefore hold the district court committed reversible error in concluding plaintiffs failed to establish racial bloc voting. First, the district court rejected plaintiffs' evidence of racial bloc voting, even though they used the same statistical method approved in Gingles and most of the § 2 case law. Although criticizing various aspects of the methodology, without any searching analysis of the evidence in the first instance, the court's critique is without foundation. Second, it adopted the State's statistical theory on the mistaken view that why voters vote a certain way answers Gingles' question about the existence of racial bloc voting. Finally, the court overlooked the substantial evidence of Anglo bloc voting, incorrectly holding that the single factor of partisanship explained electoral outcomes in a § 2 challenge in HD 60.

Because the district court misread the governing law, we reverse its conclusion plaintiffs failed to establish geographical compactness, Hispanic cohesiveness, and Anglo bloc voting. We turn now to the ultimate issue of the totality of circumstances.

The Senate Report expands two of Gingles' preconditions which adumbrate our review of the totality of circumstances: the existence of racially polarized voting and the extent to which minorities are elected to public office. 478 U.S. at 48 n. 15, 106 S. Ct. at 2765 n. 15.33 Even if plaintiffs succeed in the preliminary Gingles' proof and create the inference that minority voters are harmed by the challenged electoral practice, the inference metamorphoses the fact of vote dilution when other explanations for voting outcomes are dispelled. Although § 2 violations may be pronounced only after a flexible, fact-intensive examination of this broadened evidentiary base, the Senate Report places three limits on such proof.

First, electoral devices ... may not be considered per se violative of § 2. Plaintiffs must demonstrate that, under the totality of the circumstances, the devices result in unequal access to the electoral process. Second, the conjunction of an allegedly dilutive electoral mechanism and the lack of proportional representation alone does not establish a violation. Third, the results test does not assume the existence of racial bloc voting; plaintiffs must prove it.

Gingles, 478 U.S. at 46, 106 S.Ct. at 2764 (citations omitted).

Plaintiffs are not required to rebut all the evidence of non-dilution to establish vote dilution. "Rather, the provision requires the court's overall judgment, based on the totality of circumstances and guided by those relevant factors in the particular case, of whether the voting strength of minority voters is, in the language of Fortson v. Dorsey, 379 U.S. 433, 85 S. Ct. 498, 13 L. Ed. 2d 401 (1965), 'minimized or canceled out.' " 1982 U.S.C.C.A.N. at 207 n.118. "[T]he ultimate conclusions about equality or inequality of opportunity were intended by Congress to be judgments resting on comprehensive, not limited, canvassing of relevant facts." De Grandy, 512 U.S. at ----, 114 S. Ct. at 2657.

In this case, even though the district court concluded plaintiffs failed to prove each of the Gingles' preconditions, it proceeded to the next phase of analysis and concluded based on the totality of circumstances plaintiffs had failed to establish vote dilution. While it 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, in this case, having concluded plaintiffs made no preliminary showing a remedy is even possible and then finding the totality of circumstances also defeat their claim is remarkable.

Nevertheless, the court's conclusion remains tied to its erroneous application of the contours of the Gingles' proof; therefore, it cannot stand. In particular, the court's determination of no racial polarization and a lack of Hispanic political cohesiveness predestined its review of all the evidence and submerged critical facts into less relevant factors. We canvass these findings to explain our conclusion.

First, the district court found no evidence of a history of official discrimination that has "touched the right of the members of the minority group to register, to vote, or otherwise to participate in the democratic process." 1982 U.S.C.C.A.N. at 206. Although plaintiff, Jennie Sanchez, testified about voter registration problems in Saguache County: voter registration branches placed in Anglo homes where Hispanics would feel uncomfortable entering, limited hours for field workers to register to vote, and the appointment of all Anglo election judges, the district court found plaintiff's official complaint to the State prompted remedial action.34 Plaintiff testified, in 1986 or 1988, the names of several Hispanic voters whom she had assisted in registering were absent from the voting rolls when they attempted to vote. Charles Grant, a Saguache County Commissioner from 1984-1990, testified he first instituted advertising vacancies on local boards to permit a more open selection process in what was otherwise a system in which Anglo commissioners tapped their friends for appointments. Plaintiff, Jennie Sanchez, also testified, in 1992, the Center Post-Dispatch, a Saguache County newspaper, published a list of absentee voters. The names were mostly Hispanic surnames, which, plaintiff thought masked a subtle racial appeal against Hispanics. Although we agree the list is "a matter of public record," as the district court found, 861 F. Supp. at 1529, the focus here must be the minority's perception of the action. To the Hispanic community, publishing a list of Hispanic absentee voters makes a statement about immigrant status, particularly offensive in a predominantly native-born Hispanic population. "In this enlightened day and age, bigots rarely advertise an intention to engage in race-conscious politics." Uno, 72 F.3d at 984. While there may be a host of other explanations, minority plaintiffs are entitled to present circumstantial evidence of racial discrimination to rebut the State's offering other explanations.

Finally, the court acknowledged plaintiffs "perceived" a discriminatory motive behind the ballot initiative making English the official language of Colorado, but stated, "the Amendment was soundly defeated by a solid coalition of Hispanic and Anglo voters." 861 F. Supp. at 1528. In fact, Colorado voters approved Amendment 1. See Colo. Const. art. II, § 30a. Hispanic voters in HD 60 overwhelmingly voted against it.

Gingles asks how the challenged electoral 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." 478 U.S. at 47, 106 S. Ct. at 2764 (italics added). Clearly, the Senate Report targets the vestiges of discrimination which may lurk like "a silent, shadowy thief of the minority's rights" continuing to disadvantage minority opportunity. Uno, 72 F.3d at 984.

The district court found most of this evidence too remote in time to establish a present impediment to voting in HD 60. We agree the record discloses no current metastases of past discriminatory practices. While it is difficult to draw a bright line demarcating the end of any history of official discrimination in voting, we are satisfied on the whole the district court did not err in concluding such past practices are sufficiently attenuated to mitigate their present effects. Nevertheless, while plaintiffs cannot rely on the lingering effects of alleged discrimination to bolster their case, neither can defendants explain the showing of vote dilution by eliminating this factor from the problem.

Another Senate factor asks the extent to which members of the minority "bear the effects of discrimination in such areas as education, employment and health, which hinder their ability to participate effectively in the political process." The Senate Report explains, "The courts have recognized that disproportionate educational, employment, income level and living conditions arising from past discrimination tend to depress minority political participation." 82 U.S.C.C.A.N. at 207 n.114.

Although the district court recognized the higher unemployment and poverty rates borne by the Hispanic community in HD 60, it credited certain lay testimony that today, "Hispanics have equal access to education, jobs, and loans," disconnecting any past socioeconomic discrimination from present participation in the political process. 861 F. Supp. at 1529. We believe this finding is, on the whole, contradicted by the overwhelming weight of plaintiffs' evidence, testimonial and documentary. One of plaintiff's experts, Dr. Rodolfo de la Garza, testified that poverty, unemployment, school drop-out, housing, and alcoholism disproportionately affect the Hispanic community in south central Colorado. Based on the 1990 census and demographic information from state agencies, he concluded education, age, income, and employment, accepted predictors of voter participation, create a "reinforcing milieu" which impedes Hispanic participation. This testimony was echoed by experts on both sides as well as lay witnesses and legislative representatives discussing the need to increase school funding in south central Colorado. While the record contains testimony about Hispanic students graduating from high school and attending area colleges and community colleges, the broader picture of lower educational achievement by Hispanic students predominates.35 Indeed, the district court noted broadly, "[a]ccording to the 1990 Census data, Anglos in the Valley have higher education and income levels than Hispanics, as well as lower unemployment and poverty rates...." 861 F. Supp. at 1529.

The district court found "Hispanics in H.D. 60 turn out to vote at rates only slightly less (1%) than non-Hispanics," while "Hispanic voter registration is virtually equal to Anglo voter registration." 861 F. Supp. at 1529. While it noted, "Plaintiffs themselves have been successful in establishing convenient voter registration hours and in substantially increasing Hispanic voter registration and turnout," id., the testimony reflects this effort redressed discriminatory barriers local officials erected to Hispanic voter registration. In most instances, their vigilance didn't supplement existing voter registration but permitted it to occur in the first instance.

We agree the record includes testimony offered by teachers, Hispanic business owners, elected and party officials to establish that these and other individuals have succeeded.36 However, the record is replete with demographic, historic, and socioeconomic data which comprehensively fills the canvas. Each facet must be considered under the totality, and the absence of one factor doesn't necessarily cancel out the presence of another. As we stated, the Senate Report doesn't require a final score. While there is some evidence to support the district court's general conclusion, the evidence on the whole points the other way underscoring the nexus upon which the Senate factor focusses.

The district court found no evidence in political campaigns of overt or subtle racial appeals, giving little weight to allegedly discriminatory comments that it found were either hearsay or too remote in time, and finding the publication of the list of absentee voters unnoteworthy. The court's finding is not clearly erroneous.

The second important carryover from the Gingles ' "gloss on the totality of the circumstances" is a thoughtful look at the extent to which Hispanics have been elected to office. Jenkins, 4 F.3d at 1115. The district court found besides controlling the Democratic party in the Valley, "Hispanics have run for and been elected to numerous political offices in the Valley [and] elected and appointed to local, county, and state-level boards, commissions, and offices." 861 F. Supp. at 1530. We believe the record does not justify the district court's credit of the extent to which minorities have been elected to public office in HD 60.

The Senate Report cautions, "the election of a few minority candidates does not necessarily foreclose the possibility of dilution of the [minority] vote, in violation of this section." 1982 U.S.C.C.A.N. at 207 n.15 (citation omitted); see also Gingles, 478 U.S. at 76, 106 S.Ct. at 2779-80. The Report explains, "Were we to hold that a minority candidate's success at the polls is conclusive proof of a minority group's access to the political process, we would merely be inviting attempts to circumvent the Constitution...." Id.

Thus, some success at the polls does not negate a showing of polarized voting or disprove the existence of vote dilution. Moreover, "exogenous elections--those not involving the particular office at issue--are less probative than elections involving the specific office that is the subject of the litigation." Clark v. Calhoun County, Miss., 88 F.3d 1393, 1397 (5th Cir.1996). Plaintiffs here complain they cannot elect their candidate of choice to the Colorado General Assembly because the configuration of HD 60 dilutes their vote. They offer the single fact an Hispanic has not been elected to this particular office since 1940. That fact is probative under the totality, notwithstanding the mayoral offices, rural electrical boards, and other seats Hispanics have achieved.

An additional factor the district court examined is whether elected officials are responsive to the particularized needs of the minority group. It found generally that HD 60 Representative Lewis Entz has addressed issues important to his Hispanic constituency, cooperates with Hispanic representatives in the legislature, and works for "economic development, education, and water." 861 F. Supp. at 1530. Although Mr. Entz was taken to task for referring to Hispanics as "wetbacks," he testified he didn't believe the term was derogatory but apologized for using it publicly.

While there is evidence to the contrary, we do not judge the district court's finding HD 60's elected official responsive to the minority community clearly erroneous. This factor, however, does not defeat plaintiffs' claim. "Unresponsiveness is not an essential part of plaintiff's case," the Senate Report states, and the cases consider it of only "limited relevance." 1982 U.S.C.C.A.N. at 207 n.116; Harvell v. Blytheville School Dist. No. 5, 33 F.3d 910, 917 (8th Cir.1994). In Clark, 88 F.3d at 1400, the Fifth Circuit concluded the district court attached too much weight to the finding the county was responsive to the minority community's needs. The Fifth Circuit counseled, "[r]esponsiveness, like many things, is a question of both kind and degree. While two cities may both be said to be responsive to minority needs, the two may vary greatly in approach and commitment. The totality-of-circumstances inquiry is not blind to those differences." Id. at 1401; see also Westwego Citizens for Better Government v. City of Westwego, 946 F.2d 1109, 1123 (5th Cir.1991) ( § 2 violated although plaintiffs did not prove lack of responsiveness). Thus, while the district court's finding is not clearly erroneous, its relevance and weight in the entire balance are limited.

Finally, the court found the policy underlying the state's drawing HD 60 was not tenuous, citing that evidence "show[ing] that the Colorado Reapportionment Commission followed the dictates of § 2 in drawing H.D. 60 from the beginning to the end of the reapportionment process." 861 F. Supp. at 1530. It noted the newly drawn district actually increased the Hispanic voting age population by 5%, was approved by the Colorado Supreme Court, and reflected the overwhelming community sentiment to keep the Valley whole.

The Senate Report explains this factor, indicating "[i]f the procedure markedly departs from past practices or from practices elsewhere in the jurisdiction, that bears on the fairness of its impact. But even a consistently applied practice premised on a racially neutral policy would not negate a plaintiff's showing through other factors that the challenged practice denies minorities fair access to the process." 1982 U.S.C.C.A.N. at 207 n.117.

The record casts doubt on the court's finding the Commission from beginning to end observed the tenets of § 2. Indeed, the final decision to draw HD 60 in its present form eschewed EDS's recommendation a majority Hispanic district was required based on the research the Commission specifically requested.37 Further, members of the Commission voiced concern that accommodations were made to create black majority districts elsewhere in the state38 when the case presented by the Hispanic community in south central Colorado was equally as compelling.39 These facts bear upon whether the resulting district is fair.

The question then becomes whether the decision to draw HD 60 in the complained of configuration "markedly departs from practices elsewhere in the jurisdiction, that bears on the fairness of its impact." As noted, the decision was a quintessentially political one, ramified by concerns about pitting incumbents against each other, splitting cities differently, and dividing the Valley to create the necessary population base. In the final vote, these "traditional districting principles," splitting counties and cities and the protection of incumbents, were elevated over compliance with § 2 despite the Commission's knowledge of a perceived vote dilution problem in HD 60. Although the resulting district increased the Hispanic voting age population by 5%, the record establishes the Commission was fully aware that effort was insufficient to create the Hispanic majority district EDS recommended based on the extent of racially polarized voting. We do not suggest traditional districting principles were used as a pretext for discrimination. Nevertheless, "it would be irresponsible for a State to disregard the § 2 results test." Bush, --- U.S. at ----, 116 S. Ct. at 1969. Again, plaintiffs have offered evidence of the process which the Senate Report envisioned placing within the quantum tending to show a violation under the totality of circumstances. Plaintiffs did not have to show the resulting district was a pretext for discrimination. The "fairness of the impact" was demonstrated by the 1992 election in which the incumbent was returned to the General Assembly for his sixth consecutive term.

Thus, "through other factors," racially polarized voting and electoral success for the HD 60 seat, we believe plaintiffs have presented evidence to rebut the State's contention the resulting district does not deny them "fair access to the process." Again, the district court misjudged the weight to be attributed this factor and allowed secondary evidence to subsume the primary evidence of racial polarization.

and Bush

In its conclusion, the district court characterized plaintiffs' § 2 claim, "[a]s a practical political matter," an effort "to segregate political districts by race [which] can only serve to deepen racial divisions by destroying any need for voters or candidates to build bridges between racial groups or to form voting coalitions." 861 F. Supp. at 1531 (citing Holder v. Hall, 512 U.S. at ----, 114 S. Ct. at 2599 (Thomas, J., concurring) (size of voting district does not violate § 2)). "The Plaintiffs would have the court assume 'that members of the racial group must think alike and that their interests are so distinct that the group must be provided a separate body of representatives in the legislature to voice its unique point of view.' " Id. (quoting Holder, 512 U.S. at ----, 114 S. Ct. at 2599). Surely, all voting rights cases are premised on minority status and, necessarily, seek some "special" treatment of minority voters shown to have been denied that guarantee. Eliminating poll taxes, literacy tests, and multimember districting schemes which submerged minority voters in majority populations perhaps were the easier remedies. In this new round, the challenge has turned to single member districts drawn to remedy minority vote dilution which ostensibly balkanize minority and majority populations in alleged violation of the Fourteenth Amendment.

At this point, we must digress from the immediate resolution of this case to the Supreme Court's most recent guidance to address two of the district court's concerns and preempt the argument, based on these recent cases, that any remedy here may unwittingly violate the Equal Protection Clause. Despite this apparent detour, we shall return to the substance of plaintiffs' claim: a fair opportunity to elect representatives of their choice.

On the one hand, adherence to Gingles to remedy violations of § 2 necessarily implicates race. On the other hand, how closely that remedy comes to triggering "the Court's equal protection angst" seems to arise when a constitutional line is drawn "at cartographic departures from 'traditional districting principles.' " S. Issacharoff, The Constitutional Contours of Race and Politics, 1995 S.Ct. Rev. 45, 47 (1996). The latest round of Supreme Court cases from Miller v. Johnson, --- U.S. ----, 115 S. Ct. 2475, 132 L. Ed. 2d 762 (1995), to Shaw I, Shaw II, and Bush scrutinizes these outer limits of race-based districts where § 5 of the VRA mandated a cure which the Court then found violated the Equal Protection Clause. Upon this new terrain, Gingles retains viability to determine the § 2 violation, but, once found, Shaw II and Bush teach us that race cannot override all other traditional districting principles any more than reasonably necessary to remedy the violation.

In each of the recent cases, the proposed district's bizarre shape triggered strict scrutiny, unmistakably suggesting that race was the legislature's predominant consideration in redistricting. Miller, --- U.S. at ----, 115 S. Ct. at 2488. In Miller, the newly created Eleventh District was geographically "a monstrosity, stretching from Atlanta to Savannah." --- U.S. at ----, 115 S. Ct. at 2484. However, "[s]hape is relevant not because bizarreness is a necessary element of the constitutional wrong or threshold requirement of proof, but because it may be persuasive circumstantial evidence that race for its own sake, and not other districting principles, was the legislature's dominant and controlling rationale in drawing its district lines." Id. at ----, 115 S.Ct. at 2486. Consequently, to satisfy the strict scrutiny the Eleventh District triggered, the Court held, "the State must demonstrate that its districting legislation is narrowly tailored to achieve a compelling interest." Id. at ----, 115 S.Ct. at 2490. Miller left open the question whether compliance with the VRA, "standing alone, can provide a compelling interest independent of any interest in remedying past discrimination." Id. at ---- - ----, 115 S. Ct. at 2490-91. However, "under a correct reading of the statute," id. at ----, 115 S.Ct. at 2491, the proposed district was not required.

Bush answered the question left open in Miller. In her separate concurrence, in which four Justices joined, Justice O'Connor announced, "compliance with the results test of § 2 of the Voting Rights Act (VRA) is a compelling state interest." --- U.S. at ----, 116 S. Ct. at 1968. As we noted, in Colorado, compliance with the VRA falls second only to the one person, one vote requirement and precedes all traditional districting concerns. Consequently, if a § 2 violation is found, Colorado has a compelling state interest in seeking a remedy tailored by its traditional districting principles.

It is now a given that shape triggers strict scrutiny, and in this case the State argued, and the district court agreed, plaintiffs' proposed alternative district was not geographically compact. In a vote dilution case, once plaintiffs have established the second and third Gingles' preconditions, the Gingles' compactness requirement is the entry into the narrow tailoring question. If the state has a compelling interest in avoiding § 2 liability and if the minority population is concentrated enough to form a majority-minority district, the state must "tailor its districts narrowly to serve that interest." Id. at ----, 116 S.Ct. at 1971 (Kennedy, J., concurring). Although Justice O'Connor concluded § 2 does not require a non-compact majority-minority district, Justice Kennedy observed, "neither does [ § 2] forbid it, provided that the rationale for creating it is proper in the first instance. Districts not drawn for impermissible reasons or according to impermissible criteria may take any shape, even a bizarre one." Id. at ----, 116 S.Ct. at 1972.

Justice O'Connor clarified the contours of proof necessary for a state to have a compelling interest in remedying discrimination under § 2. Two conditions, Justice O'Connor wrote, must be satisfied: "First, the discrimination that the State seeks to remedy must be specific, identified discrimination; second, the State must have had a strong basis in evidence to conclude that remedial action was necessary, before it embarks on an affirmative action program." Id. at ---- - ----, 116 S. Ct. at 1962-63 (citation omitted) (internal quotations omitted). For purposes of proof, Justice O'Connor accepted as valid a claim of vote dilution caused by racial bloc voting for the state's § 2 compliance defense. Id.

The factual predicates for these recent cases cannot be ignored. In each challenged district, voters were grouped solely on the basis of race, even when the district didn't address the precise violation found, Shaw II, when geographical and political boundaries were ignored, Shaw II (district followed interstate highway); and Bush (district fragmented precincts and neighborhoods). In each instance, the individuals were selected solely on the basis of their race, raising the specter of a new genre of political apartheid.40 In our case, plaintiffs' proposed alternative district attempted to bring together Hispanic voters who also live in geographically connected areas that share the same agricultural and rural communities of interest, along with various socioeconomic concerns. Thus, although race figures in the configuration of the proposed alternative Hispanic majority district, it is not "in substantial disregard of customary and traditional districting practices." Miller, --- U.S. at ----, 115 S. Ct. at 2497 (O'Connor, J., concurring). Moreover, Justice O'Connor acknowledged "the nature of the expressive harms with which we are dealing, and the complexity of the districting process, are such that bright-line rules are not available." Bush, --- U.S. at ----, 116 S. Ct. at 1964.

Miller and its progeny, thus, offer a solution for remedying a found violation. Four principles emerge from these cases which sharpen the focus on the existing alleged violation. First, redistricting in which racial concerns predominate, done even for remedial purposes, is subject to strict scrutiny. Second, compliance with § 2 of the VRA constitutes a compelling governmental interest. Third, the discrimination the state seeks to remedy must be specific. Fourth, the state must have a strong basis in evidence to conclude the Gingles ' preconditions exist to justify the redistricting as reasonably necessary to comply with § 2. Fifth, states may intentionally create majority-minority districts and otherwise take race into consideration without coming under strict scrutiny so long as traditional districting criteria are not subordinated. Id. at ----, 116 S.Ct. at 1969.

In response to our order to the parties to address these latest cases, plaintiffs rely upon Justice Kennedy's statement the compactness requirement refers "to the compactness of the minority population, not to the compactness of the contested district". Id. at ----, 116 S.Ct. at 1971. Plaintiffs urge the distinction was lost on the district court which superimposed the shape of the proposed alternative district exclusively and ignored the size and compactness of the Hispanic population in south central Colorado. Plaintiffs insist the Bush plurality would permit even a bizarrely shaped district that is narrowly tailored to fit the found VRA violation. The State reads these recent cases to underline the correctness of the district court's order.

As we earlier discussed, the district court's definition of geographically compact was too limited and did not incorporate the notion of compactness of the minority population itself, not simply the area enclosing them. Second, in deciding whether any remedy was available, the district court diffused the compactness inquiry with concerns over impacting adjacent districts. While it cannot be gainsaid that increasing the minority population in one area will necessarily extract minority population from another, "without this phenomenon, no majority-[minority] districts would ever be created." Clark, 21 F.3d at 95; Houston v. Lafayette County, Miss., 56 F.3d at 610-11. In fact, no county in south central Colorado including the Valley, Pueblo, and Trinidad, emerged after the 1992 redistricting with an Hispanic voting age majority population.41 In any event, the concern is better addressed to the remedial phase of this litigation in which the State must have the first opportunity to draw HD 60.

Again, because plaintiffs sought only to establish the possibility of an alternative district, we do not linger on its specifics, the counties and cities it splits, or the mathematical scores it generates. "Districting plans are integrated bundles of compromises, deals, and principles ... representing an array of values, some relatively neutral, some intensely partisan." Expressive Harms, 585. At this stage, it remains academic to compare the proposed alternative to other oddly shaped or unusually large districts in Colorado. Suffice we recognize the proposed alternative evidences none of the extreme bizarreness of any of the districts that offended the Court and leave the drawing of boundaries to the area of politics in which it belongs.

The "expressive harms" plaintiffs have alleged cannot be recast as political balkanization, as the district court decried, especially in light of the Supreme Court's directive that compliance with § 2 is a compelling state interest. Plaintiffs, instead, availed themselves of the protections of § 2, seeking to establish they do not presently have the same fair opportunity to elect representatives of their choice to the General Assembly. Their proof echoed what the State's redistricting Commission considered before it drew the currently complained of district. They further established under the totality of circumstances racial polarization drives the voting community in HD 60 despite limited local success in being elected or appointed to political office. Consequently, plaintiffs have satisfied the two conditions necessary for the State to have a compelling interest to remedy their § 2 claim. Although we conclude the district court incorrectly applied the governing law, nonetheless, we recognize Congress has handed to the courts the task of interpreting and applying a law which appears deceptively simple. Yet, that law is exasperatingly complex, requiring application of principles and concepts drawn from disciplines foreign to most judges. Therefore, criticism is not to be leveled at anyone who conscientiously attempts to come to grips with the monumental and salutary task given to us. We must, however, REVERSE the judgment and REMAND the case to the district court with directions to order the State to implement a remedial plan of redistricting consistent with this opinion and the dictates of § 2.