dissenting.
Assuredly, much progress has been made since the enactment of the Voting Rights Act in 1965. But in spite of the contributions the majority quite properly credits with eradicating many of the most glaring forms of discrimination in voting, the law has yet to ensure that members of minority groups will have an equal opportunity to participate in the political process. We cannot pretend, and I do not read the majority opinion to suggest, that the discrimination prohibited by the Voting Rights Act has been relegated to an unfortunate but closed chapter of American history. Discrimination and its effects remain a part of our present reality. If we deny the continued existence of this problem, we not only lose our ability to recognize and remedy present instances of unlawful inequality; we also guarantee that discrimination and the damage it does to the integrity and effectiveness of democratic government will be a more prevalent and intractable feature of our country’s future.
The majority and I share these broader concerns, but I have a different understanding of the way in which they are implicated in this case. In my view, the district court erred in concluding that Pennsylvania’s nonvoting purge statute does not violate § 2 of the Voting Rights Act. Moreover, the purge statute serves as a clear reminder that the law has yet to eliminate discrimination and its enduring effects in the area of voting. For these reasons, I dissent.
Because of my deep concerns about the impact that I fear today’s decision will have, both on the residents of Philadelphia who will not be able to exercise their right to vote in upcoming elections, and on future efforts to properly apply the Voting Rights Act, I have set forth my views at considerable length.
I.
Early every January, in keeping with Pennsylvania’s non-voting purge law, the Voter Registration Division of Philadelphia’s City Commissioners Office determines whether registered voters have voted during the previous two calendar years. Those who have not done so are slated for purging. After identifying these voters, the City sends them “intent to purge notices.” The notices, which are printed only in English, state:
NOTICE OF FAILURE TO VOTE WITHIN TWO YEARS
Our Records indicate you have failed to vote for the last two years. As required by law, we will cancel your Registration, unless you vote in the next Primary or Election or File with this Commission a written request for Reinstatement (10) ten days prior to the next Primary or Election, signed by you, giving your present residence. This is the only notice you will receive.
Appendix (“A.”) at 362. The address of the Voter Registration Division appears on the *320notices beneath this statement.1 The notices do not, however, provide any instructions as to how an individual might “File with this Commission a written request for Reinstatement,” nor do they contain any information regarding what such a “request for Reinstatement” must contain (other than a signature and address).
Voters who are sent an intent to purge notice and who fail either to vote in the upcoming spring primary or to file a written request for reinstatement are purged from the registration rolls. Once purged, an individual must re-register in order to vote in the future.
The non-voting purge has a substantially disparate impact on black and Latino Phila-delphians. The uncontroverted statistical evidence presented at trial, which the district court credited, showed that each year, greater percentages of black and Latino voters were slated for purging than were white voters. The evidence further showed that white voters were reinstated at higher rates than blacks and Latinos, thus increasing the adverse disparate impact on these minority groups as a result of the non-voting purge. A. at 32-35.
The differences between the way this challenged voting practice affected Latinos and blacks, as compared to the way it affected whites, were substantial and consistent. According to the plaintiffs’ expert, Dr. Alan Lichtman, the statistical data showed that Pennsylvania’s non-voting purge created a “clear and consistent pattern” of systematically purging black and Latino voters at significantly greater rates than whites. See, e.g., A. at 38-39 (testimony of Dr. Lichtman, describing the differences as far exceeding the standards of statistical significance, and as reflecting “systematic” results of the purge process); A. at 385, 394-95 (declaration of Dr. Lichtman, stating: “The disparate impact on minorities of Philadelphia’s nonvoting purge is both substantial and systematic, extending through a full four-year electoral cycle.”). Dr. Lichtman’s expert analysis of the registration and purging statistics was uncontroverted and scientifically sound. The district court, relying on his trial testimony and declaration, found that the expert evidence “clearly reveals that African-American and Latino voters are slated for purging at higher rates than their white counterparts, and further, that minorities are purged at higher rates than white registrants.” Ortiz v. City of Philadelphia, 824 F.Supp. 514, 530 (E.D.Pa.1993). “This finding,” the court stated, “is not limited by the fact that there were fluctuations in the purge rate from year to year.” Id.2
The district court made a number of other factual findings, which the majority accurately recounts. Particularly significant among those, for reasons I discuss below, the court found that Latino and black Philadelphians have suffered significant disadvantages in ed-*321ueation, employment, housing, health care, and income.3 These “substantial socioeconomic disparities among African-American, Latino, and white residents of the City of Philadelphia,” the court concluded, “affect the ability of these minority groups to participate in the political process and to elect candidates of their choice.” Ortiz, 824 F.Supp. at 535. In the court’s view, that conclusion was “further supported by statistical evidence demonstrating that minority voters in Philadelphia do not exercise their right to vote to the same extent as white voters, which in part may be attributable to discrimination and the overall socioeconomic status of minorities in Philadelphia.” Id.
The record thus shows that Pennsylvania’s non-voting purge law operates to remove blacks and Latinos from Philadelphia registration rolls at substantially higher rates than whites. In addition, it establishes that black and Latino Philadelphians suffer disadvantages and discrimination in various socioeconomic categories. Specifically, the findings indicate that members of these groups are less educated. They are in poorer health, and they experience greater difficulty acquiring adequate medical care. They own fewer homes. They are less frequently employed. They have lower incomes. Finally, they do not vote as often. The district court even demonstrated its understanding of the relationship between these familiar facts: that socioeconomic disparities, which are linked to past and present discrimination, make members of minority groups less able and less likely to participate in the political process. Prior to today’s decision, I would have thought that given this proven reality, courts could easily appreciate the discriminatory effect of Pennsylvania’s non-voting purge law, as well as the legal significance of that effect under § 2 of the Voting Rights Act.
II.
In 1982, in response to the Supreme Court’s decision in Mobile v. Bolden, 446 U.S. 55,100 S.Ct. 1490, 64 L.Ed.2d 47 (1980), Congress amended § 2 to make clear that in determining whether this provision of the Voting Rights Act has been violated, courts should apply the “results test” that the Supreme Court articulated in White v. Regester, 412 U.S. 755, 93 S.Ct. 2332, 37 L.Ed.2d 314 (1973), rather than require plaintiffs to prove that a voting practice or procedure was motivated by discriminatory intent. E.g., Thornburg v. Gingles, 478 U.S. 30, 35, 106 S.Ct. 2752, 2758, 92 L.Ed.2d 25 (1986). Accordingly, § 2(a) prohibits all states and political subdivisions from applying any qualification or prerequisite to voting, or any standard, practice, or procedure “in a manner which results in a denial or abridgement of the right of any citizen of the United States to vote” who is a member of a protected class. 42 U.S.C. § 1973(a) (emphasis added). Section 2(b) sets forth the legal standard for meeting the results test adopted in § 2(a); drawn directly from the Court’s language in White, § 2(b) states:
A violation of subsection (a) 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) in that its members have less opportunity than other members of the electorate to participate in the political process and to elect representatives of their choice.
42 U.S.C. § 1973(b). Distilling the central meaning of this statutory language in Gin-gles, the Supreme Court stated: “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.” Gingles, 478 U.S. at 47, 106 S.Ct. at 2764 (emphasis added).4
*322When Congress amended § 2, it instructed that in determining whether political processes remain equally open to members of protected groups, in the way the Act requires, courts must conduct “a searching practical evaluation of the ‘past and present reality.’ ” S.Rep. No. 97-417, 97th Cong., 2d Sess. 30 (1982), reprinted in 1982 U.S.C.C.A.N. 177 (“Sen.Rep.”).5 The Supreme Court had taken such an approach in White, 412 U.S. at 769-70, 93 S.Ct. at 2341, and, following the 1982 amendments, it has emphasized the importance of the congressional mandate to maintain a searching practical perspective when evaluating the effects, and thus the lawfulness, of a challenged voting practice or procedure. See Gingles, 478 U.S. at 45, 62-63, 66, 106 S.Ct. at 2772-78, 2774 (repeatedly quoting and following Congress’s instruction to conduct a searching practical evaluation of reality in § 2 inquiries). Additionally, in expanding on this required approach, Congress instructed that in applying § 2, courts should take a functional view of political processes as opposed to a formalistic one. Sen.Rep. at 30 n. 120; see also Jenkins v. Red Clay Consolidated School Dist. Board of Education, 4 F.3d 1103, 1122 (3d Cir.1993) (“The Senate Report repeatedly emphasizes that the court must evaluate plaintiffs’ claims under ‘a searching practical evaluation of the past and present reality and on a functional view of the political process.’ ” (quoting Gingles, 478 U.S. at 45, 106 S.Ct. at 2764; omitting internal quotation and citation)), cert. denied, — U.S. —, 114 S.Ct. 2779, 129 L.Ed.2d 891 (1993).
In keeping with this clearly expressed congressional intent, as well as with the post-1982 Supreme Court and Third Circuit decisions which faithfully do so, we must take a functional rather than a formalistic view of political processes and the way in which they are affected by challenged voting measures. Our understanding of what the Voting Rights Act means when it requires that such processes remain “equally open to participation” by members of a protected class must be informed by a searching practical evaluation of past and present reality. With this functional, practical, and realistic perspective— which, in my view, the majority fails to achieve — we can then turn to the question that the Supreme Court has identified as the essence of a § 2 claim: whether “a certain electoral law, practice, or structure interacts with social and historical conditions to cause an inequality in the opportunities” enjoyed by members of racial and language minorities. Gingles, 478 U.S. at 47, 106 S.Ct. at 2764.
A.
The majority devotes considerable effort to supporting the proposition that in order to prove a violation of § 2, plaintiffs must demonstrate “some causal connection between the challenged electoral practice and the alleged discrimination.” Maj. op. at 310. In fact, in its view, that is “the primary legal issue before us.” Id. I agree that § 2 plaintiffs must show a causal connection between the voting practice they challenge and the deprivation of equal political opportunity they allege. This point — which, I believe, the plaintiffs also accept6 — is not open to question. The language of the Voting Rights Act, the legislative history of the 1982 amendments, and controlling precedent all foreclose any argument that a voting practice can violate § 2 without resulting in, or play*323ing some part in causing, the abridgement of citizens’ voting rights. The majority,'however, actually requires something more than and different from the causal connection it initially describes. From its uncontroversial answer to what it presents as the primary legal question, it departs from and distorts the meaning of § 2.
On two occasions, the district court stated that the plaintiffs had not proven a violation of § 2 because they had failed to demonstrate that “the purge law is the dispositive force in depriving minority voters of equal access to the political process.... ” Ortiz, 824 F.Supp. at 539 (emphasis added); see also id. at 524. That is not the proper legal standard. Section 2 does not require plaintiffs to prove that a challenged voting practice or procedure is “the dispositive force,” or the only cause, or even the principal cause, of unequal political opportunity. Neither the statute, nor its legislative history, nor the relevant case law supports such a reading. To the contrary, that authority requires us to determine whether a challenged law interacts with other, external conditions to limit the political opportunities available to members of protected classes. Gingles, 478 U.S. at 47, 106 S.Ct. at 2764. As White and § 2(b) state, we must consider the totality of the circumstances. Those circumstances — the “social and historical conditions” that make up the “past and present reality” surrounding political processes — create the factual environment in which a challenged voting practice operates. As such, they necessarily contribute to the effect that a practice has on individuals’ political opportunities. In order to properly apply § 2 in this case, we cannot require the plaintiffs to prove that Pennsylvania’s non-voting purge law operates, by itself and independently of other circumstances, to produce discriminatory results. The challenged law must certainly be a contributing cause of unequal opportunity, but it need not be, as the district court twice suggested, “the dispositive force.”
At oral argument, the City conceded this point. It did not defend the “dispositive force” language that appears in the district court’s opinion, but instead argued that in spite of these erroneous comments, the court understood and applied - the proper legal standard. To an extent, I am inclined to agree with that position. A fair (though slightly charitable) reading of the district court’s analysis reveals that, at least on occasion, it seems to have understood that while § 2 plaintiffs must demonstrate that a voting practice interacts with other, external circumstances to cause an actionable effect, such plaintiffs need not establish that the practice is the dispositive or even primary cause of the results giving rise to their claim.
The majority, however, in contrast to the City, does not in any way qualify its support for the district court’s statements.7 Instead, it adopts the view that in order to have established a violation of § 2, the plaintiffs would have needed to prove that the nonvoting purge operated independently of social and historical conditions to cause the inequality of which they complain. Thus, the majority concludes that while black arid Latino voters have turned out in lower numbers, the purge law did not cause that statistical disparity. Maj. op. at 314. And because the plaintiffs have not shown that the purge law itself has prevented members of minority groups from voting, they have failed to prove the kind of causation that the majority reads § 2 to require. Under this construction, a challenged voting law is permissible unless it is entirely responsible for the abridgment of plaintiffs’ political opportunities; if other factors interact with the challenged law to bring about such a discriminatory result — absent proof that those factors are themselves prod*324ucts of the law — no violation of § 2 has occurred.8
Accordingly, the majority states that the discrimination and disadvantages that the plaintiffs experienced in education, employment, health care, housing, and income, which the district court found and documented, “just are not relevant.” Maj. op. at 315; see also id. at 317 (stating that societal disadvantages are “extraneous” to the plaintiffs’ legal claim). Such social and historical conditions do not matter, it explains, because “the record reveals no link”, by which the majority means no causal link, between those factors and the purge law. Id. at 315-16. After all, nothing in the record suggests that the non-voting purge makes black and Latino residents of Philadelphia so much worse off. It is not the purge law that causes members of minority groups to vote less frequently, and therefore to lose the ability to vote more often by removing them at disproportionately high rates from the registration rolls. That is the kind of “legally dispositive” causal connection that the majority would require the plaintiffs to establish in order to prevail on their § 2 claim.9
*325Finally, the majority reasons that this type of causation could not, under any circumstances, exist here, because a non-voting purge law, by its very nature, only affects people who have managed to register at least once. Maj. op. at 315. Such people, it points out, must have already demonstrated their ability to overcome whatever barriers might prevent them from participating in the political process, and any law that merely requires them to do so again could not violate § 2. Id. The premise underlying this reasoning can only be that unless a law has prevented members of a protected class from ever having registered or east a vote, it is permissible under the Voting Rights Act.
According to the Supreme Court, the essence of a § 2 claim is that a certain electoral law interacts with other circumstances to cause unequal political opportunity. Gingles, 478 U.S. at 47, 106 S.Ct. at 2764. The majority takes a different position. In its view, the essence of a § 2 claim is that a certain electoral law actually causes whatever circumstances might contribute to the deprivation of political equality of which a plaintiff complains. Respectfully, that is an unsupported and insupportable distortion of the Voting Rights Act with which I could not disagree more strongly.
The majority offers a simple explanation for the fact that a disproportionately high number of Latinos and blacks have been removed from Philadelphia’s registration rolls: minority citizens are purged more because they vote less. The majority’s reading of § 2 does not require it to know anything more about the underlying causes of the law’s effect. It need not look any deeper, or any more closely. It need not concern itself with why blacks and Latinos vote less; so long as the purge law itself is not responsible, no violation of § 2 has occurred.
The majority’s explanation for the purge law’s effect — the fact that minority voters simply do not vote as often — does not explain much. Rather, that fact lies near the surface of the past and present reality confronting members of minority groups who might try to participate in the political process. As the district court demonstrated, it is not difficult to look beneath the surface. And contrary to the majority’s view, that inquiry is not only important, but necessary. The majority has erred in concluding that § 2 does not require a more searching evaluation of the reasons why Latinos and blacks do not vote as often.
The district court recognized that the lower minority turnout statistics “may in part be attributable to discrimination and the overall socioeconomic status of minorities in Philadelphia.” Ortiz, 824 F.Supp. at 535. And regardless of any uncertainty the court had about the relationship between discrimination and socioeconomic disadvantages, on the one hand, and lower participation rates, on the other, Congress and the Supreme Court have provided answers. The Senate Report states that, as the Court observed in White, disadvantages in education, employment, income levels, and living conditions arising from past discrimination tend to depress minority political participation. Sen.Rep. at 29. “Where these conditions are shown,” the Report continues, “and where the level of black participation in politics is depressed, plaintiffs need not prove any further causal nexus between their socioeconomic status and the depressed level of political participation.” Id.10
*326As early as 1899, W.E.B. Du Bois described the barriers to meaningful political participation that black Philadelphians experienced as a result of their poverty, lack of education, and susceptibility to manipulation by local party organizations. See W.E.B. Du Bois, The Philadelphia Negro: A Social Study 373-75 (1899). Gunnar Myrdal, in a work on which the Supreme Court has relied in defining equal justice under law, has also noted “the striking relationship between nonvoting and poverty.” Gunnar Myrdal, An American Dilemma: The Negro Problem and Modem Democracy, 493 (Í944) (cited in Brown v. Board of Education, 347 U.S. 483, 494 n. 11, 74 S.Ct. 686, 692 n. 11, 98 L.Ed. 873 (1954)). But we should not have to depend on landmark works of social science to understand this point. We need only follow Congress’s instruction that in assessing the way a voting law affects individuals’ opportunities to participate in political processes, we conduct an evaluation of past and present reality.11 Councilman Ortiz described that reality at trial, where he testified that in Philadelphia’s Latino community:
[Y]ou have a population that is surviving from day to day[,] and to try to bring the concept of voting into their world, into their vista, is difficult. [T]hey have a survival mode. They have to be able to put some food on the table for the next day. And we’ve had' difficulties trying to convince folks that the' political system is going to make a difference in their lives, that by participating in the process, they’re going to effect a change.
A. at 127 (repetitive speech omitted). Black and Latino residents of Philadelphia, who, as the district court found, have suffered discrimination and disadvantage in every socioeconomic category, have fewer resources to devote to the tasks of registration and voting. Moreover, their experience has often taught them that such efforts will not improve their fives. Frustration and alienation can lead to apathy, which decreases participation, which further erodes the political power of minority groups. Judge Wisdom described this cycle of discrimination, disadvantage, and disenfranchisement in United States v. Marengo County Comm’n, 731 F.2d 1546, 1567 (11th Cir.1984), where he noted that “[p]ast discrimination may cause blacks to register or vote in lower numbers than whites [and] may also lead to present socioeconomic disadvantages, which in turn can reduce participation and influence in political affairs.” A law that purges registrants for failing to vote can only contribute to the symbiotic causes and effects of socioeconomic and political inequality. Thus, for Latinos and blacks, having their registration nullified only confirms what they are more likely to already believe: that given the higher costs and lower benefits of participating in the political process, there is simply no point in making the effort.
Even if we could not achieve our own understanding of this reality, its truth is plain enough that Congress and the Supreme Court have told us to take it on faith. Again, *327citing White, the Senate Report states that where plaintiffs can show socioeconomic disadvantage and discrimination along with depressed rates of political participation, they need not prove any further causal nexus between these facts. Sen.Rep. at 29 n. 114. Of course, if the majority is right, the existence of such a causal nexus would defeat any challenge to a law that had a disparate impact on those who voted less often. The majority, once again, would require the plaintiffs here to have proven that the purge law itself caused minority voters to be removed from the registration rolls at disproportionately high rates. Factors like poverty, lack of education, and inability to find work or housing as a result of discrimination, it states, are simply not relevant. Maj. op. at 315. However, we know that while the nonvoting purge law may play some part in further discouraging black and Latino political participation, it is largely societal discrimination and socioeconomic disadvantage, of the sort the district court found, that operate to depress minority turnout rates. If discrimination and disadvantage cause blacks and Latinos to vote less often, which we discover by consulting either the record or reality, then the purge law itself cannot be responsible for removing so many more members of these groups from the registration rolls. That is undeniable. The plaintiffs do not and cannot argue otherwise. So, if the majority is right, Congress and the Supreme Court — by relieving plaintiffs of the need to prove a causal connection between discrimination and socioeconomic disadvantage, on the one hand, and lower participation rates, on the other — have effectively immunized non-voting purge laws from § 2 challenges. Such laws cannot be responsible for the disparate impact they have when, as a matter of law, discrimination and disadvantage are at least principally to blame. Clearly, though, Congress and the Court have done no such thing, and the majority has erred in its construction of § 2.
The district and appellate court decisions in Mississippi State Chapter, Operation PUSH v. Allain 674 F.Supp. 1245, 1263-68 (N.D.Miss.1987) (“Operation PUSH I ”), aff'd sub nom. Mississippi State Chapter, Operation PUSH, Inc. v. Mabus (“Operation PUSH III”), 932 F.2d 400 (5th Cir.1991), a ease which closely parallels this one, properly apply § 2 and reject the majority’s position. In Operation PUSH, the plaintiffs challenged two Mississippi registration laws: one, the State’s “dual registration process,” required citizens to register first with a county registrar in order to vote in federal, state, and county elections, and again with a municipal clerk in order to vote in municipal elections; the other, the State’s prohibition of “satellite registration,” severely limited the extent to which citizens could register at places other than the registrar’s office. Operation PUSH III, 932 F.2d at 402. The district court found that these procedures had a substantial disparate impact on black registration rates. That disparity, it concluded, resulted in minority members of the electorate having less of an opportunity to participate in the political process and thus established a violation of § 2. Operation PUSH I, 674 F.Supp. at 1268. Of course, Mississippi’s registration laws had not themselves caused the failure of blacks to register as often as whites. The inequality occurred, as the court of appeals put it, “[bjecause a significant percentage of Mississippi’s black citizens do not have (1) access to the transportation necessary to travel to the county courthouse to register or (2) the type of jobs that would allow them to leave work during business hours to register to vote.... ” Operation PUSH III, 932 F.2d at 403; see also Operation PUSH I, 674 F.Supp. at 1255-56.12 *328Those circumstances interacted with the challenged laws to create a disparity in registration rates, which gave rise to a § 2 violation. Operation PUSH I, 674 F.Supp. at 1255-56, 1268. In affirming that ruling, the Eleventh Circuit did not express any doubt as to the soundness of the district eourt’s reasoning. The only question, in its view, was whether the court had erred in finding that blacks had registered in significantly lower numbers. Operation PUSH III, 932 F.2d at 409-13. Given the evidentiary support for the trial judge’s determination that such a disparity did exist, the Eleventh Circuit upheld the decision that the challenged laws, and the effect they had, violated § 2.
Unlike the majority, the Operation PUSH courts did not merely note that blacks registered less often, and that the registration laws themselves were not responsible for the disparity. Instead, the courts looked to the way in which the challenged laws interacted with social and historical conditions, such as the class members’ occupational status and limited access to transportation and information, to result in a lack of equal opportunity for the plaintiffs to participate in the political process. The obvious fact that the causes of those conditions had little or nothing to do with the challenged registration laws was legally irrelevant to the Operation PUSH courts, which clearly did not share the majority’s understanding of § 2.13
United States v. Marengo County Comm’n, 731 F.2d 1546 (11th Cir.1984), is similar, in significant respects, to Operation PUSH, and also rejects the majority’s view of the kind of causation a plaintiff must show in order to prevail on a § 2 claim. In Mar-engo County, black plaintiffs challenged laws that provided for the at-large election of members of the County Commission and the school board. The district court, in dismissing the case, had “attributed the absence of elected black officials to ‘voter apathy’ and ‘a failure of blacks to turn out their votes.’ ” Marengo County, 731 F.2d at 1568 (quoting the district court opinion). The Eleventh Circuit rejected the district court’s findings as clearly erroneous and reversed.
In reaching its decision, the Marengo County court relied, in part, on the plaintiffs’ showings that (1) the County Board of Registrars was open infrequently and never visited outlying areas to register rural voters, and (2) the County had not permitted a black person to serve as a deputy registrar. Id. at 1570. The district court had stated that it did not see how such policies discriminated against blacks. The court of appeals responded:
These policies, however, unquestionably discriminated against blacks because fewer blacks were registered.... By holding short hours the Board made it harder for unregistered voters, more of whom are black than white, to register. By meeting only in Linden the Board was less accessible to eligible rural voters, who were more black than white. By having few black poll officials and spurning the voluntary offer of a black citizen to serve as a registrar, county officials impaired black access to the political system and the confidence of blacks in the system’s openness.
*329Marengo County, 731 F.2d at 1570 (footnote and citation omitted). Like the Operation PUSH courts, the Eleventh Circuit did not require the plaintiffs to prove that the measures they challenged were responsible for their failure to register or vote. In fact, in rejecting the district court’s finding that blacks had simply failed to overcome their apathy, the court explained that a variety of social and historical circumstances accounted for the fact that blacks participated in significantly lower numbers than whites. Id. at 1568, 1574. Those proven circumstances, in conjunction with the challenged voting practices, had a discriminatory effect on blacks. For that reason, the court concluded, the record compelled a finding that the County’s at-large election system violated § 2. Id. at 1574.14
The majority relies on Salas v. Southwest Texas Junior College Dist., 964 F.2d 1542 (5th Cir.1992), for the proposition that “a protected class is not entitled to § 2 relief merely because it turns out in a lower percentage than whites to vote.” Maj. op. at 314. The context of that statement, however, reveals that Salas provides no support for the majority’s holding. In Salas, Hispanic voters, who constituted a majority of the registered voters in an at-large district, challenged the use of the at-large system (as opposed to single-member districts) under § 2. The court held that the plaintiffs had failed to satisfy the third element of the threshold test for vote dilution claims that the Supreme Court adopted in Thornburg v. Gingles, because they had not demonstrated that white bloc voting usually prevented Hispanics from electing their preferred representatives. Salas, 964 F.2d at 1555. The Fifth Circuit reasoned that even though the plaintiffs were members of a class that included a majority of the registered voters in the challenged district, they might nonetheless, in some instances, have been able to show that minority bloc voting frustrated their chances of electoral success. Id. For example, if such plaintiffs could establish that they faced “practical impediments to voting” or that “low turnout at elections was the result of prior official discrimination,” they could prove that their registered majority did not allow the class to overcome opposition from a cohesive minority of whites. Id. at 1555-56. In such circumstances, which the plaintiffs in Salas had failed to prove, an at-large district would be subject to a § 2 challenge brought by members of a majority class. But the Salas plaintiffs had not accounted, in any way, for their inability to take advantage of their majority status. In that context, we can understand the court’s statement that “a protected class is not entitled to § 2 relief merely because it turns out in a lower percentage than whites to vote.” Salas, 964 F.2d at 1556.
Here, the plaintiffs are not members of a majority class confronting the difficult task of satisfying the Gingles preconditions. The questions this case presents, and the factors that we must take into account in providing answers, are somewhat different. Nonetheless, to the extent that Salas does provide guidance, the decision actually rejects the majority’s reading of the Voting Rights Act. According to the Fifth Circuit, the plaintiffs in Salas could have prevailed if they had shown the existence of “practical impediments to voting” or low turnout rates resulting from prior discrimination (as may have occurred here, according to the district court). Those circumstances, the court reasoned, could interact with the state’s district-ing plan in such a way as to deprive class members of an equal opportunity to participate in the political process and to elect their chosen representatives. Under the majority’s construction of § 2, however, the Salas plaintiffs would have needed to prove that the voting measure they challenged, an at-*330large district, actually caused the practical impediments to voting or the low turnout rates that compromised their chances for success. Of course, that is an impossible requirement. District lines, like purge laws, cannot by themselves cause the “totality of the circumstances” that courts are supposed to consider in applying § 2. The Fifth Circuit would not have required the Salas plaintiffs to make such a showing. The majority here, in contrast, holds that because Pennsylvania’s non-voting purge law is not itself the cause of blacks’ and Latinos’ failure to vote more often, the plaintiffs have not established a violation of § 2. Salas does not take that view, and neither do I.15
In amending § 2, Congress devoted most of its attention to vote dilution cases. White v. Regester, for example, the case from which Congress drew the legal standard for defining a violation of § 2, involved a challenge to state reapportionment plans. And Zimmer v. McKeithen, 485 F.2d 1297 (5th Cir.1973), which articulated and enumerated the factors that the Court had taken into consideration in White, addressed the legality of multi-member districts. The Senate Report’s now-familiar list of “typical” (but not exclusive or required) factors that plaintiffs can show in establishing a violation of § 2 is drawn directly from Zimmer. See Sen.Rep. at 29. Congress made clear, however, that while it had focused and relied mainly on vote dilution eases in amending § 2 and explaining the new provision’s meaning, § 2 would continue to prohibit “all voting rights discrimination,” including “episodic” practices which did not involve the “permanent structural barriers” to political equality seen in the vote dilution context. Sen.Rep. at 30, 1982 U.S.C.C.A.N. p. 207. The Report explains:
If the challenged practice relates to such a series of events or episodes, the proof sufficient to establish a violation would not necessarily involve the same factors as the courts have utilized when dealing with permanent structural barriers. Of course, the ultimate test would be the White standard codified by this amendment of Section 2: whether, in the particular situation, the practice operated to deny the minority plaintiff an equal opportunity to participate and to elect candidates of their choice.
Id. Thus, events or episodes occurring during the administration of elections that resulted in unequal political opportunity would remain unlawful, although the factors relevant to proving such discriminatory effects might well be different from those involved in vote dilution cases.
In a footnote to the statement quoted above, Congress mentions purging as one variety of the “episodic discrimination” that § 2 would continue to prohibit.16 The note *331states: “[Pjurging of voters could produce a discriminatory result if fair procedures were not followed, Toney v. White, 488 F.2d 310 (5th Cir.1973), or if the need for a purge were not shown or if opportunities for re-registration were unduly limited.” Sen.Rep. at 30 n. 119,1982 U.S.C.C.A.N. p. 208 (emphasis added).17 For two reasons, this statement does not support an affirmance here. First, as I discuss at greater length below, the City did not prove “the need” for Pennsylvania’s nonvoting purge law. The district court found that the law served the legitimate purpose of preventing fraud, but laws are not necessary merely because they serve a valid purpose. Second, Congress mentioned purging as an example of the “episodic discrimination” that would remain within the scope of § 2, and the types of purges it listed in the one sentence it devoted to the subject are merely more specific examples of the way in which purging can have a “discriminatory result.” Congress provided no indication that purging could not produce discriminatory results, and thus violate § 2, in other ways. Indeed, such a suggestion would run directly counter to the flexible approach to § 2 cases that Congress intended courts to follow in situations involving different factual and legal variables.18
In the majority’s view, however, the Senate Report’s brief and illustrative reference to purging contains an “implicit” recognition “that a purge statute which was administered fairly, and in an even-handed manner, would not run afoul of the law.” Maj. op. at 311 n. 6. I have no idea where the majority finds such an implication. The text of the footnote plainly does not support it. See Ortiz, 824 F.Supp. at 522 (recognizing that the list in the footnote merely “provides several examples of how purging voters could violate § 2,” and that it does not “establish any prerequisites that plaintiffs ... must establish in order to state a cognizable claim under the Act”); see also, Steve Barber et al., Comment, The Purging of Empowerment: Voter Purge Laws and the Voting Rights Act, 23 Harv.C.R-C.L.L.Rev. 483, 520 (1988) (stating that as the footnote makes clear, the types of purges it lists “are merely examples of situations that might produce a discriminatory result, rather than an exhaustive list of the only ways in which purges could violate Section 2”). Moreover, Congress expressly disavowed the kind of purpose the majority believes it implied, stating: “[E]ven a consistently applied practice premised on a racially neutral policy would not negate a plaintiffs showing through other factors that the challenged practice denies minorities fair access to the process.” Sen.Rep. at 29 n. 117, 1982 U.S.C.C.A.N. p. 207. Finally, under the majority’s perception of what Congress implicitly recognized, not even all the examples included in the Report’s brief reference to purging would constitute violations of § 2. An unnecessary purge law that “unduly limited” opportunities for re-registration, for instance, would not, in the majority’s view, run afoul of the Voting Rights Act so long as the law was administered fairly. Congress’s statements, which express a different view, do not dissuade the majority from relying upon an unsupported implication that would severely, and I believe improperly, limit the scope of § 2 in challenges to the purging of registered voters.
*332The Senate Report acknowledges that especially outside the context of vote dilution claims, the factors that courts should consider in determining whether a challenged voting practice violates § 2 may well be different than those appearing in the list of “typical” factors drawn from White and Zimmer. The Report does not, however, provide any guidance with respect to what factors courts should consider in the type of case we confront here. The majority makes no effort to identify or explore the considerations that figure into an evaluation of the effect and legality of a non-voting purge law, probably because, in light of its theory of causation, it sees no need to do so. The district court, in contrast, addressed this important question, see Ortiz, 824 F.Supp. at 523-24, and I agree with some of its conclusions. As the district court recognized, the factors making up the threshold test for vote dilution claims that the Supreme Court adopted in Thornburg v. Gingles, 478 U.S. at 49-51, 106 S.Ct. at 2766-67, have little place in a § 2 challenge to a purge law. Ortiz, 824 F.Supp. at 523 (describing the Gingles preconditions as “peripheral issues bearing limited relevance to the plaintiffs’ claim”); see also Operation PUSH I, 674 F.Supp. at 1264 (concluding that the occurrence of polarized voting is not germane to a § 2 challenge to registration procedures); cf. Major v. Treen, 574 F.Supp. 325, 350 (E.D.La.1983) (three-judge court) (“To the extent that the enumerated factors are not factually relevant, they may be replaced or substituted by other, more meaningful factors”). The size, geographic insularity, and political cohesion of minority or majority groups are all irrelevant to the question of whether a non-voting purge statute has a discriminatory effect. Such a law operates to nullify the registration of individuals who fail to vote or reinstate themselves. Therefore, the important factors are those that influence peoples’ willingness and ability to participate or seek reinstatement. As discussed, the existence of discrimination and socioeconomic disadvantage is of primary importance among those factors.19 Other significant considerations could include: a history of official discrimination in voting, efforts made with the design or effect of limiting the political power of minority groups, the difficulty minority constituents have experienced in trying to elect their preferred candidates, and a lack of responsiveness on the part of elected officials to the particular needs and concerns of minority communities. Each of those factors could further discourage participation and therefore, each could interact with a non-voting purge statute to cause an unlawful abridgement of a group’s opportunity to participate in the political process and to elect its chosen representatives.20
Given the majority’s erroneous construction of § 2, however, this task of identifying and weighing relevant factors is of secondary importance. The critical point, it bears repeating, is that contrary to the majority’s view, whatever the relevant factors are, they need not be products of, or causally linked to, a challenged voting law in order to merit consideration in a court’s determination of the law’s effect and legality. It is a law’s interaction with the totality of the circumstances, and not its creation of those circumstances, that § 2 requires us to evaluate.
B.
The plaintiffs have argued that in order to prove that Pennsylvania’s non-voting purge law violates § 2, they must, in addition to showing the law’s disparate impact on blacks and Latinos, demonstrate that the State could achieve its legitimate goal of preventing election fraud in a less discriminatory *333way. As noted above, the Senate Report provides some support for that position by stating that if the “need for a purge” is not shown, the practice could constitute an instance of prohibited “episodic discrimination.” Sen.Rep. at 30 n. 119. Here, the plaintiffs want to take on the burden of proving that Pennsylvania does not need a purge law that results in the disenfranchisement of disproportionately high numbers of Latinos and blacks. Moreover, they have asked us to find that they carried that burden at trial.
I agree with much, though not all, of the plaintiffs’ argument. In enacting and amending § 2, Congress did not intend to mandate equal political opportunity at the expense of state and local governments’ abilities to pursue legitimate goals such as the prevention of voting fraud. Where measures are in fact necessary to achieve such objectives, Congress would have surely meant for them to remain in effect. There is no point in guaranteeing equal access to corrupt political processes that have no democratic value. To the contrary, the Voting Rights Act is premised on the existence of free and fair elections that reflect popular will with accuracy and integrity, and it should not be interpreted to prevent such elections from occurring.21 However, a voting practice that compromises equality, as I believe Pennsylvania’s purge statute does, should only be permitted if it is in fact necessary to achieve or preserve this type of public interest.22
Title VII disparate-impact law, in my view, provides a particularly appropriate analogy supporting the adoption of such a standard. In disparate-impact cases, employers can justify their use of challenged practices that have a greater negative effect on members of a protected class by proving that such practices constitute a “business necessity.” See, e.g., Griggs v. Duke Power, 401 U.S. 424, 431-32, 91 S.Ct. 849, 853-54, 28 L.Ed.2d 158 *334(1971) (first articulating the concept of business necessity); Dothard v. Rawlinson, 433 U.S. 321, 329-32, 97 S.Ct. 2720, 2726-28, 53 L.Ed.2d 786 (1977) (holding that defendant-employer failed to show business necessity, and that therefore, plaintiffs properly prevailed in disparate-impact Title VII case); see also 42 U.S.C.A § 2000e-2(k)(l)(A), (C) (codifying the disparate-impact standards developed in and following Griggs, including the business necessity defense, and nullifying Wards Cove Packing Co. v. Atonio, 490 U.S. 642, 109 S.Ct. 2115, 104 L.Ed.2d 733 (1989)). I would take the same approach here. Title VII disparate-impact law, like § 2 of the Voting Rights Act, adopts a results test for ensuring that protected individuals will not suffer a deprivation of the type of equal opportunity that the law mandates. Given the similarities between these two statutes, it is not surprising that in Griggs, the Supreme Court relied on a Voting Rights Act case, Gaston County, North Carolina v. United States, 395 U.S. 285, 89 S.Ct. 1720, 23 L.Ed.2d 309 (1969) (discussed infra p. 338 n. 30), in first explaining how and why Title VII prohibited the sort of discrimination challenged in a disparate-impact suit.23 Griggs’s reliance on Gaston County supports the analogy I am suggesting.
If an employer can show that an employment practice that operates to the disadvantage of minority workers is necessary to run its business, that practice is permissible. Similarly, in my view, if a state or local government can show that a voting practice that operates to the disadvantage of minority registrants is necessary to run a valid election, that practice would not constitute a violation of § 2.24 While the plaintiffs here have volunteered to prove that Pennsylvania does not need its non-voting purge law to guard against fraud, I would not require them to do so. Consistent with other results-based anti-discrimination law, defendants should bear the burden of proving necessity once plaintiffs establish the existence of a disparate impact.
The plaintiffs argue that the uncontrovert-ed evidence they introduced at trial “indisputably establishes that there are reasonable, less discriminatory alternatives to Philadelphia’s non-voting purge.” Appellants’ Brief at 32. I disagree. Evidence does not become indisputable merely because it is never disputed. Some evidence fails on its own, without any help from opposing evidence. Trial judges are not obligated to credit un-controverted testimony; they can simply remain unpersuaded. In this case, the district court never considered whether the City needed a non-voting purge law to prevent electoral fraud because the legal standard it adopted did not include this element. The court found that the policy reasons for the law were “not tenuous,” and that the law in fact served its intended purpose. “Not tenuous,” however, is a long way from necessary. The plaintiffs’ evidence would have easily supported a finding that practical, effective, less discriminatory alternatives to the purge do exist, but it did not “indisputably establish” anything. As a court of appeals, we are in no position to resolve this question. Therefore, under my view of § 2 and the evidence presented, a remand would be appropriate.
The majority, however, without the assistance of the district court, and without evi-*335dentiary support, now finds that the nonvoting purge law actually is “needed” to prevent electoral fraud. Maj. op. at 316 (relying on “a review of the record and present reality”). That is an unfounded conclusion. Again, I do not dispute that the purge law serves a valid purpose; I accept the district court’s finding that the law is useful in preventing fraud. But neither the anecdotal incidents the majority mentions nor the record show anything more. Contrary to the majority’s apparent view, usefulness and necessity are two entirely different things. In order to find that the purge law is needed, as the majority now does, a court must first consider whether equally effective but less discriminatory alternatives exist, which neither the district court nor the majority has ever done.
The testimony of Emmett Fremaux, Jr., whom the district court qualified as an expert on election and voter registration procedures, provides the only evidence in the record on the subject of alternatives to the non-voting purge law. Fremaux testified that the law’s important purposes could be achieved more effectively through other mechanisms which, unlike purge laws, would not have a pronounced discriminatory effect. A. at 311, 316. He based his opinion, to a significant extent, on his experience in administering registration and election procedures in Washington, D.C., which had successfully replaced its non-voting purge law with a “mail canvass” system. In addition, Fremaux stated, other large cities run clean elections without relying on non-voting purges. A. at 331.
Through this testimony, the plaintiffs introduced uncontroverted evidence that more effective, less discriminatory alternatives to a non-voting purge law do exist. Moreover, the City has not suggested that come January, 1995, when the National Voter Registration Act becomes effective and puts an end to the non-voting purge (at least in elections for federal office),25 alarming numbers of voters will emerge from prior residences or the grave to defile the integrity of Philadelphia elections. In a matter of months, purging will cease. Elections, presumably, will not. At oral argument, the City stated that it had every intention of complying with the new law. Apparently, Philadelphia believes that starting next year, it will somehow manage without the purge statute that the majority finds it needs this year.26
The district court may not have been persuaded by the evidence that viable, less discriminatory alternatives to purging do exist. That, again, is its prerogative. Moreover, before considering the question on remand, the court might properly exercise its discretion to allow the parties to present additional evidence on this issue. The City did not do so at trial, no doubt, because it did not (and still does not) believe the question was legally relevant.
While I do not accept the plaintiffs’ argument that they have “indisputably established” the existence of effective, less discriminatory alternatives to the statute they challenge, they can at least point to substantial, even uncontroverted evidence in support of that position. The majority cannot make nearly as strong a case for its unassisted finding that Philadelphia does need the nonvoting purge law to prevent electoral fraud. Unlike the majority, I do not believe that we are in any position to decide this question. Thus, while the purge statute may or may not be necessary, a remand certainly is.
*336C.
To summarize, largely as a result of social and historical conditions reflected in the district court’s findings, black and Latino Phila-delphians vote less often. Because they vote less often, Pennsylvania’s non-voting purge law operates to remove members of these protected classes from the registration rolls at disproportionately higher rates. The purge law, as the majority emphasizes, is not the sole or primary cause of the circumstances with which it interacts to limit the opportunity of blacks and Latinos to vote in Philadelphia. In my view, however, neither § 2 nor the decisions construing it require the plaintiffs to make such an implausible showing, and the majority has erred in concluding otherwise. Finally, given the demonstrated disparate impact of the non-voting purge law, and the way in which it interacts with social and historical conditions to deprive members of protected classes of an equal opportunity to participate in the political process, I would require the City to establish that the law is necessary to prevent voting fraud. That is a question that should be left, in the first instance, to the district court.
III.
Both the district court and the majority place some importance on the ability of black and Latino voters to reinstate themselves after being slated for purging, or to re-register once purged. Neither opinion clearly spells out the legal significance of the possibility of reinstatement or re-registration. The district court stated its ultimate conclusion as follows: “Plaintiffs have failed to demonstrate that the purge law interacts with social and historical conditions to deny minority voters equal access to the political process and to elect their preferred representatives, particularly since it is undisputed that the purge procedure is administered fairly and that there is ample opportunity for purged voters to re-register to vote.” Ortiz, 824 F.Supp. at 539.27 The court further stated that “[vjoters in jeopardy of being purged who are interested in preserving their right to participate in the political process need only vote or request reinstatement to preserve that right.” Id The majority takes a similar position; in its view, disproportionate numbers of Latinos and blacks are purged “because they do not vote, and they do not take the opportunity of voting in the next election or requesting reinstatement.” Maj. op. at 314.
As I explain above, the district court’s conclusion that the non-voting purge law does not interact with social and historical conditions to deny blacks and Latinos equal access to the political process is irreconcilable with the factual findings on which it bases its decision. Thus, in my view, that conclusion is erroneous.28 Moreover, the *337plaintiffs’ ability to reinstate or re-register themselves after being purged does not cure the deprivation of equality that unlawfully results from the law’s interaction with the totality of the circumstances. A functional view of the political process, informed by a searching evaluation of past and present reality, reveals that for members of racial and language minority groups, neither reinstatement nor re-registration is an insignificant barrier to participation. The majority’s suggestion to the contrary follows from its formalistic understanding of political opportunity and its failure to consider the reality that Congress instructs us to consult in evaluating the legality of challenged voting laws under § 2.
Interestingly, the majority contrasts Pennsylvania’s non-voting purge law with some better-known measures that states and localities have employed with the purpose and effect of disenfranchising minority voters. It points out, for example, that unlike “‘race-neutral’ literacy tests,” the purge law is not the type of device “which discriminates against minorities, which has no rational basis, and which is beyond the control of minority voters.” Maj. op. at 313. But really, what would distinguish a § 2 challenge brought against a facially neutral and evenly applied literacy test, which had been adopted to further a state’s legitimate interest in ensuring participation by a minimally informed electorate, from a § 2 challenge such as the one the plaintiffs have brought here?29 Surely, members of protected classes can take the opportunity to learn to read (English). Surely, the literacy test itself would not be at fault if they failed to do so. The state would not have prevented such plaintiffs from voting. Rather, under the majority’s reasoning, the legally permissible explanation for the literacy test’s disparate impact would be that minority citizens learn to read at a statistically lower rate than white voters. Literacy, after all, like filing a written Request for Reinstatement or a voter registration form, is not “beyond the control of minority voters.” In fact, reinstatement and re-registration require literacy, along with other skills and resources; without the ability to read, an individual will not even know that he or she has been slated for purging, let alone be able to respond to such a notice by reinstatement or re-registration. So what makes a well-intentioned, evenly applied, and race-neutral literacy test objectionable under § 2, as the majority seems to agree it is, while a device like Pennsylvania’s non-voting purge law has no trouble passing scrutiny? Why does one measure unlawfully discriminate against minorities, while the other does not? The majority does not suggest an an*338swer to these questions, and I cannot imagine one.30
Of the more than 190,000 people who were slated for purging in 1991, all of whom, of course, had already taken the trouble to register in the first place, fully 81% failed to reinstate themselves. Reinstatement rates were consistently higher for white voters than for blacks and Latinos.- The district court and majority opinions leave the possible impression that the 156,000 or so voters who did not object to getting purged had just lost interest in political participation. After all, nothing prevented them from properly filing a written Request for Reinstatement. They had the opportunity. They simply did not take it. If these citizens had really wanted to preserve their ability to vote, surely they could and would have done so. As the district court put it, registrants who were “interested in preserving their right to participate in the political process need[ed] only vote or request reinstatement to preserve that right.” Ortiz, 824 F.Supp. at 539. Mass apathy, particularly among blacks and Latinos, is the only explanation, and disinterest is not actionable under the Voting Rights Act. See, e.g., Maj. op. at 314 (relying on Salas for the proposition that members of a protected class are not entitled to relief under § 2 merely because they vote less often). As the City states in its brief, the “extra effort” required to reinstate or re-register oneself “is entirely the result of the apathy of individual voters.” Appellee's Brief at 42.
Viewed formalistically, the task of filing a written Request for Reinstatement with the Voter Registration Division of the City Commissioners Office might seem like an easy thing to do. From a more practical perspective, however, based on the reinstatement statistics and a sense for past and present reality that is not diminished by troubling assumptions about the behavior and motivations of black and Latino citizens, we should know better. See Marengo County Comm’n, 731 F.2d at 1568 (“Both Congress and the courts have rejected efforts to blame reduced black participation on ‘apathy’.”).
The record does not contain statistics showing how many voters re-register after being purged, or how many such voters are black or Latino. We know that as a general matter, members of minority groups register in lower percentages than whites. At trial, Sandy Newman, the Executive Director of the plaintiff Project Vote, testified that in his organization’s efforts to register minority voters, individuals attempting to fill out registration forms make errors or omissions in “roughly half the cases.” A. at 93. Councilman Ortiz and his aide, Israel Colon, described the special difficulties they experienced in efforts to register Latinos, especially as a result of the language barrier. A. at 50-51, 129. According to testimony presented by Robert Lee, the Director of the City’s Voter Registration Division, his office rejected about 8,800 of the registration forms submitted between April and October of 1992; about one out of every twenty forms failed to make the grade. A. at 173. Apparently, realistically, the task of re-registration does, in fact, create a significant obstacle to participation, especially for members of minority groups.31
*339Any doubt that the record leaves regarding the burden of re-registration is answered by the recent passage of the National Voter Registration Act of 1993, P.L. 103-31, 42 U.S.C.S. §§ 1973gg et seq. (popularly known as the motor-voter law). In introducing the National Voter Registration Act, Congress stated its finding that “discriminatory and unfair registration laws and procedures can have a direct and damaging effect on voter participation ... and disproportionately harm voter participation by various groups, including racial minorities.” 42 U.S.C.S. § 1973gg(a)(3). In light of that finding, and with the stated objectives of promoting the exercise of the fundamental right to vote, protecting the integrity of the electoral process, and ensuring that accurate and current voter registration rolls are maintained, Congress enacted a law designed to increase registration rates by eliminating, to a significant extent, the need for citizens to make independent efforts to register to vote. See 42 U.S.C.S. § 1973gg-3 (providing for simultaneous driver’s license and registration application). By drastically reducing the need for people to complete and submit separate registration applications, Congress hoped to “expand[] the rolls of the eligible citizens who are registered” and thus “give the greatest number of people the opportunity to participate” in federal elections. H.Rep. No. 103-9, 103d Cong., 1st. Sess. 3 (1993), reprinted in 1993 U.S.C.C.A.N. 105,107. During the debates, supporters of the bill repeatedly expressed the view that practical and procedural obstacles, and not a lack of desire to participate, were primarily responsible for the unfortunate fact that roughly 70 million eligible voters in this country (about 40 percent of the voting-age population) had failed to register. E.g., 139 Cong.Rec. H488 (daily ed. Feb. 4, 1993) (remarks of Rep. Meek) (“[TJhere are 70 million eligible voters who are not registered because of the burdensome registration policies and procedures which we have in this country.”); id. at S2470 (daily ed. Mar. 5, 1993) (remarks of Sen. Bradley) (stating that 40 percent of the voting-age population cannot vote “because obstacles are placed in the path of them registering to vote.”) No legislator ever questioned the fact that the bill would substantially increase registration rates. 139 Cong.Rec. H517 (daily ed. Feb. 4, 1993) (remarks of Rep.' Condit) (opposing the bill, but acknowledging that it would increase registration). Common sense and the experience of states that had already adopted motor-voter mechanisms would have defeated any such suggestion. E.g., President’s Remarks on Signing the National Voter Registration Act of 1993, 29 Weekly Comp.Pres.Doc. 914, 915 (May 20, 1993) (“The State of Washington instituted a similar measure during the *3401992 election, and their motor voter program registered in that state alone an additional 186,000 people.”); 139 Cong.Rec. H495 (daily ed. Feb. 4, 1993) (remarks of Rep. Tucker) (noting experience of states); id. at S2390 (daily ed. Mar. 4, 1993) (remarks of Sen. Ford) (same).
Once again, in assessing the effect and legality of a challenged voting law, we are supposed to maintain a practical and realistic perspective. We are supposed to reject formalistic views of what it means and what it takes to participate in political processes. The district court and majority opinions suggest that registering and re-registering to vote is an easy thing to do. The way they see it, anyone who wants to register certainly can. If that were true, if the task of filling out and submitting an application to register did not represent a substantial obstacle to political participation, the National Voter Registration Act would not exist. As Representative Lewis put it: “[F]or many Americans, it is not easy to register to vote. It is difficult.” 139 Cong.Rec. H488 (daily ed. Feb. 4, 1993). For that reason, Congress made “an effort to make democracy a little more real for all citizens of the United States.” Id. at H490 (remarks of Rep. McKinney).
Courts applying § 2 of the Voting Rights Act must take a similar approach. Unless we ignore Congress’s instructions, divorce ourselves from past and present reality, and adhere to a narrow and formalistic view of political opportunity, we will understand that the discriminatory effects of Pennsylvania’s non-voting purge law are not cured or at all alleviated by the fact that Latinos and blacks are free to reinstate themselves or re-register after being purged from the registration rolls.
IV.
In Part III.C of its opinion, the majority devotes special attention to the district court’s findings regarding the success that black and Latino candidates have had in Philadelphia elections. It reads the district court’s opinion to make “an explicit finding of fact that Philadelphia’s minority population has not had difficulty electing minority representatives.” Maj. op. at 315. That finding, the majority seems to argue, precludes the plaintiffs from prevailing on their § 2 claim. The majority has committed two significant errors here. First, the district court did not make the finding the majority describes. Second, even if such a finding had been made, the majority has misinterpreted its legal significance.
A.
1. What the District Court Did and Did Not (and Could Not) Find
Again, according to the majority, the district court “made an explicit finding that Philadelphia’s minority population has not had difficulty electing minority representatives.” Id. (citing Ortiz, 824 F.Supp. at 539). I disagree. The section of the district court opinion to which the majority refers is headed: “The Extent to Which African-Americans and Latinos Have Been Elected to Public Office in Philadelphia.” Ortiz, 824 F.Supp. at 537. The court’s headings faithfully track the typically relevant factors listed in the Senate Report, which include “the' extent to which members of the minority group have been elected to office in the jurisdiction.” Sen.Rep. at 29. 1982 U.S.C.C.A.N. p. 207. Accordingly, the district court considered, and ultimately rejected, the plaintiffs’ contention that “minority candidates have experienced considerable difficulty being elected to public office in Philadelphia.” Ortiz, 824 F.Supp. at 537. In supporting this conclusion, the court specifically mentioned minority representation in the state legislature and on the City Council. Id. at 538. Thus, the district court’s discussion, as its heading indicated, described the extent to which blacks and Latinos had been elected to office in Philadelphia.
Somewhere in this section of the district court’s opinion, the majority finds “an explicit finding of fact that the City’s minority population has not had difficulty electing minority representatives.” The district court, however, never made any finding regarding — and in fact, never mentioned — the difficulty minority voters might have experienced in their efforts to elect candidates; instead, in keep*341ing with the Senate Report’s instructions, it considered the extent to which black and Latino candidates had won election to public office. There is a substantial legal and factual difference between the difficulty experienced by minority candidates, which the district court considered, and the difficulty experienced by minority voters, which the district court neither considered nor addressed in a factual finding.32
If this were a vote dilution case, the plaintiffs would have had to satisfy the Gingles threshold factors, the third of which requires proof “that the white majority votes sufficiently as a bloc to enable it — in the absence of special circumstances ... — usually to defeat the minority’s preferred candidate.” Gingles, 478 U.S. at 51, 106 S.Ct. at 2766-67. Of course, this is not a vote dilution case, and I have already explained why the factors that are most relevant in that context are of little if any relevance here. Supra p. 332. Assuming, however, that the plaintiffs did need to make a showing regarding the difficulty they have experienced in electing their preferred representatives — and as I discuss below, I believe the majority has erred in apparently concluding that they did need to do so — they could not rely, and a district court could not base a finding, entirely on the win-loss records of minority candidates. We have not lost sight of the encouraging fact that Latino and black constituents can and sometimes do prefer white candidates, even in elections where a minority candidate is also running.33 So, as we explained in Jenkins, “[wjhile it may be tempting to assume that a minority candidate is always the candidate of choice among minority voters, this is not always true.” Jenkins, 4 F.3d at 1126; see generally id. at 1124-1131 (setting forth and applying the proper standards for determining which candidates are preferred by § 2 plaintiffs). The success of black and Latino candidates, therefore, does not necessarily prove the success of black and Latino voters.34
*342Contrary to the majority’s assertion, the district court did not find that minority voters had not experienced difficulty electing minority candidates; it merely concluded that, at least as far as the plaintiffs could prove, minority candidates had not experienced difficulty getting elected.35 We do not know which of those candidates were actually preferred by minority voters. And even if the district court had found, as the majority claims, that minority voters experienced no difficulty electing candidates, that finding could only have been based on the unsupported and legally erroneous assumption that blacks and Latinos always vote for blacks and Latinos. Thus, such a finding could not survive review.
2. The Unanswered Need to Consider Special Circumstances
Assuming that the plaintiffs’ case did depend on an independent showing that they have not been able to elect their preferred representatives (which it does not), and further assuming that minority voters always prefer minority candidates (which they do not), the district court’s findings still would not defeat the plaintiffs’ § 2 claim, because the court never considered the kinds of “special circumstances” that might undermine the legal significance of the minority representation it cited. Gingles mandates that in determining whether § 2 plaintiffs have usually been prevented from electing their preferred candidates, courts take into account “special circumstances, such as the absence of an opponent, incumbency, or the utilization of bullet voting,” which -could explain the success of minority candidates even in the face of a districting scheme that unlawfully diluted the plaintiffs’ vote. Gingles, 478 U.S. at 57, 106 S.Ct. at 2769; see also Jenkins, 4 F.3d at 1119 n. 9 (“consideration of special circumstances is specifically mandated within the definition of the third Gingles factor”). As the Gingles Court took care to make clear, its “list of special circumstances is illustrative, not exclusive.” Gingles, 478 U.S. at 57 n. 26, 106 S.Ct. at 2770 n. 26.
The uncontroverted evidence presented at trial showed that of the seven black or Latino representatives serving on the Philadelphia City Council, five were elected from “minority districts.” Ortiz, 824 F.Supp. at 537; id. at 538 (acknowledging that “some of the [minority] councilmembers represent wards that are highly populated by minority constituents.”) The record does not reflect whether minority state legislators have also been elected largely from districts in which black or Latino citizens constituted a voting majority. Nor does it disclose whether other special circumstances, such as incumbency or the nature (or lack) of a candidate’s opposition, might account for the success of minority candidates in some instances.
It seems obvious that the existence and effect of minority districts is the type of special circumstance that courts must take into account in determining whether minority voters have experienced difficulty electing their candidates of choice. Of course, in the vote dilution context, this consideration will never arise. Vote dilution claims invariably challenge districting schemes that do not provide minority voters with an effective voting majority. A court could never learn anything about the occurrence of dilution by looking to the results of elections in districts where majority-minority constituencies controlled the outcome. Outside the vote dilution context, however, the demographic composition of a district must be taken into account in assessing the effect of a challenged voting practice. If a discriminatory law results in the disenfranchisement of a disproportionate number of minority voters, but the remaining voting members of the group still constitute an effective majority within a district, then plainly, the success of a minority candidate elected from that district would have no legal significance. That result would not provide any indication whatsoever that the challenged law had not impaired the *343ability of minority voters to elect their chosen representatives. Thus, the existence of majority-minority districts is the type of special circumstance that can explain, and thereby lessen or eliminate the significance of, the political success enjoyed by minority-preferred candidates.
Here, the district court never considered the existence or effect of special circumstances such as majority-minority districts. The evidence presented at trial, which the court described and at least partially credited in its opinion, would have supported a finding that these circumstances did exist. In all likelihood, the district court never had any reason to conduct this sort of inquiry, because it was only considering the extent to which minority candidates had been elected to office; it was not making a finding regarding the difficulty minority voters had experienced in trying to elect representatives, which is entirely consistent with its conclusion that the Gingles preconditions were “peripheral issues bearing little relevance” in this case. Ortiz, 824 F.Supp. at 523. However, if the district court had wanted to move from its findings of minority representation on local and state governing bodies, to a conclusion that legally significant minority political success disproved the plaintiffs’ claim that their ability to influence the outcome of elections had been impaired, it would have needed to consider the presence of special circumstances. A failure to do so would constitute reversible legal error. Harvell v. Ladd, 958 F.2d 226, 230 (8th Cir.1992) (reversing for failure to consider special circumstances). While the district court did not commit such error, because of the limited nature of its findings, the majority does. It briefly describes the electoral successes of black and Latino candidates, and concludes that those facts refute the suggestion that the plaintiffs either have been denied fair access to the political process or have suffered an impairment of their ability to influence the outcome of elections. The majority fails to acknowledge the possibility that special circumstances could have accounted for the results on which it relies, despite the uncontroverted evidence that such circumstances did, in fact, exist. Of course, appellate courts are not competent to resolve this type of question on their own, but that is precisely what the majority would have to do in order to avoid its apparent error.36
B.
Even if I accepted the majority’s account of what the district court found, I would not agree with its apparent view of that finding’s significance.
I confess that I am not entirely sure I understand the legal argument that the majority advances in Part III.C of its opinion. It at least strongly intimates, however, that the plaintiffs’ failure to prove that they have experienced difficulty electing representatives, in and of itself, would preclude them from prevailing on a § 2 claim, and thus would provide an alternative ground for affirming the district court’s decision. If that is the majority’s position, and if the majority is right, then any voting practice that prevented members of protected classes from participating in the political process — in whatever manner, to whatever extent, and for whatever reason — would be permissible under § 2, provided those groups could still eke out a reasonably fair share of election victories. Thankfully, the Voting Rights Act allows no such thing.
The majority’s legal argument, one of the two sources of Part III.C’s troubles, itself contains two errors. First, and most impor*344tantly, the majority has misread the Supreme Court’s decision in Chisom v. Roemer, 501 U.S. 380, 111 S.Ct. 2354, 115 L.Ed.2d 348 (1991). Second, it has again misconstrued § 2.
According to the majority, Chisom holds that “Section 2 plaintiffs must demonstrate that they had less opportunity both (1) to participate in the political process, and to elect representatives of their choice.” Maj. op. at 314 (emphasis added); see also Maj. op. at 315 (referring to “both elements of a Section 2 claim”). Its discussion in Part III.C seems to focus on the second of those requirements; by pointing to the electoral successes enjoyed by Latino and black candidates, the majority suggests that the plaintiffs failed to prove that their ability to influence the outcome of elections — or, in other words, their ability to elect representatives of their choice — has been impaired. Maj. op. at 314.37 Because of this purported failure, the majority concludes, the plaintiffs have “failed to satisfy both elements of a § 2 cause of action and, accordingly, [have] failed to establish a basis upon which [their] requested relief could be granted.” Maj. op. at 318.
In the part of Chisom on which the majority relies, the Supreme Court addressed the position “that § 2 provides two distinct types of protection for minority voters — it protects their opportunity ‘to participate in the political process’ and their opportunity to elect representatives of their choice.’ ” Chisom, 501 U.S. 380, 111 S.Ct. 2354, 115 L.Ed.2d at 364. The majority of the Fifth Circuit Court of Appeals, sitting en banc, had adopted that view in League of United Latin American Citizens Council No. 4434 v. Clements, 914 F.2d 620 (5th Cir.1990) (en banc) (“LULAC’’), rev’d sub nom. Houston Lawyers’ Ass’n v. Attorney General of Texas, 501 U.S. 419, 111 S.Ct. 2376, 115 L.Ed.2d 379 (1991) (companion ease to Chisom). The LULAC majority determined that the word “representatives,” as used in the phrase “to elect representatives of their choice,” did not include judges. Consequently, it held that § 2 plaintiffs could not challenge a districting scheme that allegedly prevented them from electing the judicial candidates they preferred. However, because the word “representatives” did not limit the first of the two distinct protections that, according to the LULAC majority, § 2 provided — namely, the opportunity to participate in the political process — a voting practice or procedure that compromised minority voters’ opportunities to participate in judicial elections remained subject to challenge. Chisom describes the Fifth Circuit’s rationale as follows:
[A] standard, practice, or procedure in a judicial election, such as a limit on the times that polls are open, which has a disparate impact on black voters’ opportunity to cast their ballots under § 2, may be challenged even if a different practice that merely affects their opportunity to elect representatives of their choice to judicial office may not.
Chisom, 501 U.S. 380, 111 S.Ct. 2354, 115 L.Ed.2d at 364. The construction of § 2 the majority appears to adopt in Part III.C is similar but more demanding. Like the LU-LAC court, it divides § 2; the opportunity to participate, in its view, is distinct from the opportunity to elect. Unlike the LULAC court, however, the majority reads § 2 to contain two “elements,” each of which a plaintiff must satisfy in order to prevail, rather than two alternative “protections”; thus, while the Fifth Circuit required plaintiffs to prove only one of two different abridgements of political opportunity, the majority here requires proof of both.
The Supreme Court rejected both majority positions — the one advanced in LULAC and the one advanced here. According to Chi-som, the Fifth Circuit had erroneously at*345tempted “to divide a unitary claim created by Congress.” Chisom, 501 U.S. 880, 111 S.Ct. 2354, 115 L.Ed.2d at 365. The majority of this panel has done precisely the same thing. However, as the Chisom Court put it, § 2 cannot be compelled to undergo the “radical surgery [that] would be required to separate the opportunity to participate from the opportunity to elect.” Id., 501 U.S. 380, 111 S.Ct. 2354,115 L.Ed.2d at 364. “The statute does not,” the Court explained, “create two separate and distinct rights.” Id. Rather: “Any abridgement of the opportunity of members of a protected class to participate in the political process inevitably impairs their ability to influence the outcome of an election.” Id. (emphasis added). As the Court had stated in White v. Regester and Whitcomb v. Chavis, Chisom, continues, “the opportunity to participate and the opportunity to elect [are] inextricably linked.” Id., 501 U.S. 380, 111 S.Ct. 2354, 115 L.Ed.2d at 365 (emphasis added) (citing White, 412 U.S. at 766, 93 S.Ct. at 2339, and Whitcomb v. Chavis, 403 U.S. 124, 149, 91 S.Ct. 1858, 1872, 29 L.Ed.2d 363 (1971)).
Thus, in Chisom, the Court clearly and repeatedly rejected the majority’s apparent view that § 2 plaintiffs must satisfy each of two distinct elements (lesser opportunity both to participate and to elect) in order to prevail. Instead, if such plaintiffs can prove that a voting law has abridged their opportunity to participate in the political process— by, for example, removing them from the registration rolls and thus effectively nullifying their franchise — that law has also necessarily and “inevitably” impaired the plaintiffs’ ability to influence the outcome of an election. Such is the nature of the “unitary claim” Congress created. Under this sensible and well-supported construction of § 2, a voting practice is impermissible if it operates to deny the members of a protected class of an equal opportunity to participate in the political process — even if the group has managed to overcome that disadvantage and to elect representatives of its choice.38
The majority does not agree. Again, looking exclusively to electoral results, it concludes that the plaintiffs cannot prevail, because there is no evidence that the purge law has denied them fair access to the political process or impaired their ability to influence the outcome of elections. Maj. op. at 314-15. This position could be based on one of several rationales, none of which are availing. Perhaps the majority disagrees with, or has overlooked, Chisom’s use of words like “inevitably” and “inextricably linked”, which describe the relationship between the abridgement of a group’s opportunity to participate in the political process, on the one hand, and the impairment of that group’s ability to elect representatives, on the other. Chisom, 501 U.S. 380, 111 S.Ct. 2354, 115 L.Ed.2d at 364. The majority would apparently require the plaintiffs to produce evidence of a connection that, at least according to Chisom, necessarily exists as a matter of law and logic. Alternatively, the majority might believe that by purging people from the registration rolls, the City has not in any way lessened their opportunity to participate in the political process. But I would hope we could agree about this much: when individuals are purged, they cannot vote; if they cannot vote, their opportunity to participate is diminished; and, once again, according to the Supreme Court, if their opportunity to participate is diminished, their ability to elect their preferred representatives is necessarily, “inevitably” impaired. A final possibility remains. The majority might simply be restating, in a somewhat oblique way, its view that in order to establish a violation of § 2, the plaintiffs would have had to establish that the purge law itself operates to prevent Latinos and blacks from voting and that, therefore, the *346law is entirely responsible for its discriminatory effect. If that is the majority’s purpose, however, then the argument advanced in Part III.C would not provide any additional support for its conclusion, and thus would not require a response.
In arguing that the plaintiffs’ failure to prove difficulty electing representatives defeats their claim, the majority not only misreads Chisom; it also, once again, misreads § 2. The district court, in contrast, recognized that the success black and Latino candidates have had in winning public office is merely one factor among others that are relevant in a § 2 inquiry. As § 2(b) quite plainly states: “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.” 42 U.S.C. § 1973(b) (emphasis added).39 According to the majority’s apparent view, however, that circumstance is somehow disposi-tive. Maj. op. at 314-15. Congress disagrees. Sen.Rep. at 29 n. 118 (rejecting the notion that a failure “to establish any particular factor” defeats a § 2 claim). The Supreme Court disagrees. See Gingles, 478 U.S. at 45,106 S.Ct. at 2763 (recognizing that no particular number of factors need be proved, but that § 2 analysis instead depends on a searching practical evaluation of past and present reality and on a functional view of the political process). I disagree as well.
V.
Congress, in its continued efforts to ensure that the democratic processes occurring in this country will be equally open to members of racial and language minorities, has had the good sense to prohibit non-voting purges in federal elections. The legislation it recently enacted to eliminate “discriminatory and unfair registration laws” does not overlook laws such as the one the plaintiffs have challenged in this ease. Thus, when the National Voter Registration Act becomes effective, Pennsylvania will no longer be able to remove individuals from the registration rolls because of their failure to vote. See 42 U.S.C.S. § 1973gg-6(b)(2).40 As I stated at the outset, however, I am concerned that today’s decision will reach beyond its context and misdirect future efforts to properly apply the Voting Rights Act.
The majority makes a valid and important observation: the law, in various ways, has indeed “sought to raze any enduring bastions of state-administered voting discrimination.” Maj. op. at 313. We cannot discount the achievement that separates Philadelphia, Mississippi in 1964 from Philadelphia, Pennsylvania in 1994. However, the goal the majority describes, and in which it undoubtedly believes, has not been attained. Even after Congress’s most recent contribution to the cause of equal political opportunity, discrimination in voting remains. A proper application of the Voting Rights Act in this case would have resulted in precisely the kind of additional progress the Act was intended to bring about.
According to the majority, the right to vote, “the very essence of democratic soei-*347ety,” has now been “extended to every American citizen, without regard to race....” Maj. op. at 313. That accomplishment loses meaning, however, when discriminatory voting laws make the right to vote harder for some to exercise than others. When such discrimination occurs, I fear that today’s decision will make it more difficult for us to respond in the way the law requires. To borrow a phrase from a particularly meaningful source, there is “unfinished work” to be done before we achieve the equality mandated by § 2 of the Voting Rights Act.41 Because this ease demonstrates that fact, I respectfully dissent.
Present: SLOVITER, Chief Judge, BECKER, STAPLETON, MANSMANN, GREENBERG, HUTCHINSON, SCIRICA, COWEN, NYGAARD, ALITO, ROTH, LEWIS, McKEE and GARTH *, Circuit Judges.. Additionally, in smaller type and without any explanation, "MU6-1500” appears in the upper left corner of the notices, and "MU6-1501” appears in the upper right corner. At trial, Robert Lee, the Director of the Voter Registration Division, testified that those markings represent the Division’s phone numbers. According to Lee, that was "obvious.” A. at 283. Lee further testified that his office did receive calls from registered voters who asked for explanations of intent to purge notices that they had received. A. at 286.
. In his declaration, based on the statistics that appear in the district court and majority opinions, Dr. Lichtman explained: "To put this [disparate impact of the non-voting purge] into perspective, if the black [purge] rate in 1991 had been as low as the white rate, about 28,000 fewer black registrants would have been [slated for purging] in 1991. Conversely, if the white purge rate in 1991 had been as high as the black rate, about 41,000 more white registrants would have been [slated for purging] in 1991.” A. at 386.
The 1991 purge was the largest of the four years studied, as it affected all registrants who had not voted since the 1988 presidential election. About 21% of all registrants were slated for purging, and over 80% of those slated were ultimately purged. (Again, whites reinstated themselves at higher rates than blacks and Latinos.) According to Dr. Lichtman, the large numbers involved in the 1991 purge made that the "key year for assessing racially differential impact” of Pennsylvania’s law. A. at 385. While the numbers of registrants purged, as well as the magnitude of the disparate impact, decreased in other years, that fact does not undermine the significance of the 1991 figures. To the contrary, as Dr. Licht-man explained, "Given that a relatively smaller proportion of black than white registrants were available for purging after 1991, it is striking that in 1992 even a small difference between whites and blacks persists.” A. at 391.
. In stating its findings with respect to housing and employment, the district court observed that the City has been held liable for practicing discrimination in these areas. Ortiz, 824 F.Supp. at 533-35.
. The Gingles Court only referred to blacks because of the facts of that case. Its discussion of § 2 applies equally to Latinos, who are also members of a protected class under the Voting Rights Act. 42 U.S.C. §§ 1973b(f)(2), 19731(c)(3).
. The Supreme Court has relied upon the Report of the Senate Judiciary Committee as the "authoritative source for legislative intent” behind the 1982 amendments. Gingles, 478 U.S. at 43 n. 7, 106 S.Ct. at 2763 n. 7.
. According to the majority, the plaintiffs have argued that they need not show a causal connection between a challenged voting practice and the discriminatory effect of which they complain. Maj. op. at 312. That is not an accurate characterization of the plaintiffs’ position. In their brief, the plaintiffs insist that they have "never asserted that the discriminatory result need not be in part attributable to the purge.” Appellants’ Brief at 41 n. 8. The district court misunderstood them, they explain, when it formed the impression that they denied that causation was relevant to their claim. Id. Neither the majority nor the City has advanced a waiver argument. Thus, regardless of whether the plaintiffs ever took the untenable position for which they received credit in the district court, the important point, for present purposes, is that contrary to the majority’s description of their argument, they have not denied the relevance of causation here.
. The majority understands the phrase "the dis-positive force,” as the district court used it, to mean "a cause which, in that context, would he legally dispositive.” Maj. op. at 313 n. 11. I do not grasp the purpose or effect of this translation. The majority does not disagree with the substance of the court's statements; instead, it reads § 2 to require proof of a “legally disposi-tive” cause, by which it means a single cause of all relevant circumstances, that is entirely responsible for the discriminatory effect of which a plaintiff complains. There is no place, in the majority’s construction of § 2, for contributing causes of inequality which are not themselves products of the challenged voting law. As I explain below, I do not share this view.
. Interestingly, while the majority first asserts that the plaintiffs have argued that causation is unimportant (see supra p. 322 n. 6), it later claims that Ortiz's entire complaint "is drawn to allege that Pennsylvania’s purge statute 'caused' the disparate purge rates between Philadelphia's white and minority communities.” Maj. op. at 313 (quotation unattributed); see also Maj. op. at 310 n. 4 and accompanying text (perhaps suggesting that the plaintiffs’ complaint improvidently endorsed the majority’s construction of § 2). That would certainly be a curious way for a party who rejected the relevance of causation to plead a § 2 case. In fact, the plaintiffs have never pleaded, accepted, or claimed to have met the legal standard of causation that the majority now adopts. See Appellants’ Brief at 37 (arguing that "[t]he Act does not require a demonstration that the practice in and of itself results in a discriminatory result,” and claiming to have "demonstrated that historic, political, and socioeconomic discrimination interacts with the nonvoting purge to disproportionately disenfranchise Latinos and African-American voters’’); id. at 14, 31, 41 n. 8 (arguing that the purge law, operating in conjunction with social and historical conditions, was a contributing cause of the disproportionate purging of Latinos and blacks); see also A. at 18-20 (plaintiffs’ complaint, not pleading any view of § 2 causation). Nor have the plaintiffs pleaded or argued the opposite extreme — that causation is irrelevant. As they state in their brief, they accept that in order to prevail, they must show that the discriminatory result of which they complain must be attributable, in part, to the law they challenge. Appellants’ Brief at 41 & n. 8. Of course, even if the plaintiffs had taken either of the directly opposing erroneous positions the majority attributes to them, that would not require us to adopt an erroneous construction of § 2 in deciding this case.
. The majority’s reading of the Senate Report casts additional light on the substance of its rationale. In its view, the Report expresses the expectation that courts would "focus on the challenged procedure and its causal consequences....” Maj. op. at 311-12 n. 6. The "focus” the majority has in mind, however, is quite restrictive; it encompasses nothing more than the challenged law, and it excludes all other factors that the law has not created, but which nonetheless might contribute to the law’s effect. Thus, under the majority's reading of the Senate Report, Congress has instructed that § 2 might prohibit a purge statute which “itself” produces discriminatory results. Id.
The Senate Report does support the well-settled point that plaintiffs must prove a causal connection between the law they challenge and the discriminatory result they identify. Congress in no way, however, expressed any agreement with the majority’s view that the challenged law must itself account for its unlawful effects. According to the Senate Report, a court's “focus” must remain broad and varying enough to encompass the totality of the circumstances with which a voting practice invariably interacts to influence the nature of political processes.
In concluding its opinion, the majority appears to abandon its causation argument. The dissent might have meaning, it states, in a § 2 challenge to registration procedures in which plaintiffs alleged that as a result of socioeconomic disadvantages, they experienced greater difficulty traveling to registration centers or in some other way availing themselves of registration opportunities. Maj. op. at 317-18. That surprising comment cannot be reconciled with the majority’s legal argument. If the conditions of inequality the district court found here are extraneous to the plaintiffs' legal claim, those circumstances could not be relevant in the hypothetical suit the majority describes. In neither case could members of a protected class show that the laws they challenged caused the practical difficulties they experienced. In , neither case would the laws themselves be responsible for their disparately harmful effects on minority citizens. Rather, under the majority's theory of § 2 causation, the unfortunate results of whatever registration laws it has in mind would merely represent another of "society’s voting ills” that would not be subject to challenge under the Voting Rights Act.
I do not know and cannot guess why the majority believes that I am actually concerned with *325Philadelphia registration procedures that the plaintiffs have not challenged, rather than with what I view as the discriminatory effect of the non-voting purge law. I also cannot perceive a difference between the majority's hypothetical case — where, it states, socioeconomic disadvantages might have meaning — and the one before us now. In any event, I do not read the majority’s conclusion as expressing any purposeful modification or qualification of its earlier analysis of § 2 causation.
. Significantly, in White — the case from which Congress borrowed the words that now make up § 2, and on which Congress relied most heavily in explaining what it intended that language to mean — the Court did not look for or require any "link” between the challenged reapportionment plan and the factors that contributed to the plan’s discriminatory effect. White recounts, among other circumstances, the following findings: Texas had a "history of official racial discrimination ..., which at times touched upon the right of Negroes to register and vote and participate in the democratic processes”, 412 U.S. at 767, 93 S.Ct. at 2340; "the Bexar community, along with other Mexican-Americans in Texas, had long 'suffered from, and continues to suffer from, the results and effects of invidious discrimination and treatment in the fields of education, employment, economics, health, politics, *326and others’ ”, 412 U.S. at 768, 93 S.Ct. at 2340 (footnote omitted; quoting district court opinion); and "[t]he typical Mexican-American suffers a cultural and language barrier that makes his participation in community processes extremely difficult...." Id. (footnote omitted). "The residual impact of this history,” the Court observed, "reflected itself in the fact that Mexican-American voting registration remained very poor in the County...." Id. Plainly, these social and historical conditions had not been brought about by a reapportionment plan that Texas promulgated in 1971. Rather, they interacted with the challenged multimember districts to result in invidious discrimination. As the Supreme Court stated, "[biased on the totality of the circumstances,” the district court had properly assessed the lawfulness of the multimember district, "overlaid, as it was, on the cultural and economic realities” of the plaintiff class. White, 412 U.S. at 769, 93 S.Ct. at 2341 (emphasis added). Those realities, of course, predated and existed independently of the recent reapportionment. For the White Court, that did not matter. For the majority, nothing else would. Thus, in codifying and relying on White when it amended the Voting Rights Act, Congress could not have shared the majority’s understanding of the kind of causation § 2 requires.
. The evaluation need not be particularly searching. As the district court put it, "In fact, in evaluating a challenge to an election procedure pursuant to § 2, it would be myopic for a court to overlook the impact that [socio-economic] factors may have on minority voter turnout." Ortiz, 824 F.Supp. at 525. Under the majority’s construction of § 2, however, courts have no need for even limited vision of the underlying causes which contribute to a challenged law’s effect.
. The district court found that the registration laws had "a disparate impact on the opportunities of black citizens in Mississippi to register to vote because of their socio-economic and occupational status.” Operation PUSH I, 674 F.Supp. at 1255. In particular, the court focused on three significant forms of socioeconomic disadvantage: (1) “that black workers in Mississippi predominate in blue-collar and service worker positions in which they tire likely to be working for an hourly wage and are less likely to be able to take off from work to register to vote during the regular office hours (2) that "black citizens are also much less able than whites to travel to the county courthouse, or other centralized voter registration locations” because black households were much less likely to have a vehicle available; and (3) that "a disproportionately high percentage of black households do not have a telephone available for household use,” which *328made it more difficult for blacks to obtain information about procedures for registering. Id. at 1253, 1256. Because of these disadvantages, the district court concluded, blacks experienced greater difficulty in overcoming the administrative barriers to registering that Mississippi had created through its challenged laws. Id. at 1256.
. The majority’s attempt to distinguish the Operation PUSH cases is unpersuasive. It states that in Operation PUSH, unlike this case, the challenged voting laws were "the causes of black voters' registering at lower rates than white voters.” Maj. op. at 315 n. 17 (emphasis added). The Operation PUSH courts, however, clearly did not find, and did not require the plaintiffs to show, that Mississippi’s registration procedures were "the causes” of the statistical disparity in registration rates. There was no established causal "link” — as the majority puts it elsewhere — between the challenged laws, on the one hand, and either the types of jobs black residents held, or the facts that they possessed fewer cars and telephones, on the other. Rather, those conditions interacted with the registration procedures to produce a discriminatory result, and that constituted a violation of § 2.
The majority’s treatment of Operation PUSH also suggests that it sees a meaningful difference between laws that make registration more difficult to begin with, on the one hand, and laws that make it more difficult for someone to stay registered, on the other. In my view, that is not a reasonable distinction.
. Other cases that do not share the majority's view of § 2 include: Harris v. Graddick, 593 F.Supp. 128 (M.D.Ala.1984) (awarding an injunction to black citizens who claimed that a state's failure to appoint greater numbers of black poll officials violated § 2), and sub nom., Harris v. Siegelman, 695 F.Supp. 517 (M.D.Ala.1988) (holding that plaintiffs established violations of § 2, based on the conclusion that social and historical circumstances interacted with the challenged law, as well as with the disproportionately low number of black polling officials, to discourage black political participation); and Brown v. Dean, 555 F.Supp. 502 (D.R.I.1982) (enjoining a city's use of a polling facility based on a finding that the facility's location would be a substantial deterrent to voting by a class of black residents).
. In.addition to Salas, the majority relies on Irby v. Virginia State Board of Elections, 889 F.2d 1352 (4th Cir.1989). It is difficult to overstate the differences between Irby and the case before us. In Irby, a group of individual black plaintiffs and two civil rights organizations challenged a law that provided for the appointment, rather than the election, of school board members. First, assuming appointive systems are subject to challenge under § 2 — a proposition about which the Fourth Circuit expressed "considerable doubt”, Irby, 889 F.2d at 1357 — the legal and factual variables involved in such a case, where no voting and no electing has ever occurred, could not be comparable to those involved here (or elsewhere). Second, in Irby, the plaintiffs had failed to show any causal connection between the selection system they challenged and the underrepresentation of blacks on school boards; for example, the court pointed out, in one of the two counties where a statistically significant disparity did exist, every black resident who had requested appointment to the school board since 1971 had been selected as a member. Id. at 1358. In light of those facts, the Fourth Circuit concluded that the appointive system did not produce a disparate effect. Id. at 1359. The Irby court had no occasion even to consider a causation requirement such as the one the majority adopts here, because the challenged practice played no part in bringing about the lower number of blacks on the two school boards. Such a situation is not at all similar to one in which a purge law directly removes disproportionately high numbers of minority citizens from the registration rolls.
At the very most, Irby provides incremental support for the position that § 2 plaintiffs must show some causal connection between a challenged voting practice and the disparity at issue. As I have stated, that is not a controversial point, and neither the plaintiffs nor I suggest otherwise.
. While purging could be conducted "episodically” — that is, on discrete occasions or in a related series of events — I would not describe the annual operation of Pennsylvania's non-voting purge law as "episodic.” It is not "structural” in the same way as a redistricting plan; nor is it temporary, however, and the law produces its *331discriminatory effects in a reliable and consistent manner from year to year. Therefore, I will not refer to the kind of purging that occurred in this case as an “episodic” voting practice.
. By making clear that purging can violate § 2, the Senate Report creates a potential problem for the majority’s argument that these laws, because they only affect individuals who have already registered, must be permissible. The majority deals with this problem by emphasizing that in each example of unlawful purging the Senate Report mentions, it is the statute "itself” that produces a discriminatory result. Putting aside, for the moment, my more important disagreements with the majority’s reading of § 2's legislative history, and with its theory of causation, the fact remains that even a purge law that itself produces unlawful results can only affect voters who have already registered. This troublesome point, however, is only a minor symptom of the infirmity afflicting the majority’s argument.
. As the district court accurately stated, the legislative history expresses "the importance of applying a flexible approach when evaluating § 2 challenges, rather than a mechanical’ point-counting analysis.” Ortiz, 824 F.Supp. at 524; see also Jenkins, 4 F.3d at 1125, 1129 (recognizing the need to maintain a flexible approach in order to promote the goals and underlying principles of the Voting Rights Act).
. Contrary to the majority's assertion, I have not stated, nor do I believe, that societal conditions such as discrimination and socioeconomic disadvantage "constitute” the totality of the circumstances that a court should consider in evaluating the legality of a non-voting purge statute. Rather, those factors are particularly significant circumstances, among others, which contribute to the totality. Additionally, as should be clear, I do not mean for the list of relevant factors I mention here to be exhaustive.
. The existence of racial appeals, in contrast, could conceivably increase minorily turnout and thus work against any discriminatory effect that a non-voting purge might have. I imagine that this factor could point either way, depending on the nature of the evidence presented in a particular case.
. I share the majority’s concern about Philadelphia's need to guard against fraudulent voting practices in administering its elections. The dispute arising out of the State Senate race between Bruce Marks and William Stinson provides a recent reminder of the seriousness of this problem. See Marks v. Stinson, 19 F.3d 873 (3d Cir.1994). Of course, as we all recognize, the City is no Newcomer to dishonest election tactics. Nearly 100 years ago, Dr. Du Bois described the widespread practice of buying votes in Philadelphia, and stated: "To-day the government of both city and State is unparalleled in the history of republican government for brazen dishonesty and bare-faced defiance of public opinion." W.E.B. Du Bois, The Philadelphia Negro: A Social Study 372, 376-77 (1899).
. According to the majority, "it is well established that purge statutes are a legitimate means by which the State can attempt to prevent voter fraud.” Maj. op. at 314 & n. 13. By "legitimate," the majority cannot mean permissible under § 2 of the Voting Rights Act. The published reports do not contain a single case, other than this one, that decides a § 2 challenge to a non-voting purge law. The only scholarship directly addressing the question argues that such laws are likely to violate § 2. Barber, supra p. 26, at 517-527.
The cases the majority credits with recognizing the well-established legitimacy of non-voting purge statutes all resolve constitutional, not statutory challenges to voting laws. And in fact, of the decisions on which the majority relies, only Hoffman v. Maryland, 928 F.2d 646 (4th Cir.1991), involves a non-voting purge law. In Hoffman, the court held that Maryland's five-year non-voting purge did not violate the First Amendment because the law was content-neutral, was designed to serve the important state interest of curbing vote fraud, and did not unreasonably restrict alternative avenues of communication; the court also rejected the plaintiffs’ Fourteenth Amendment claim because they were not, as non-voters, members of a suspect class. This case, of course, unlike those the majority mentions, has been brought under § 2 of the Voting Rights Act, which no court has read to permit any and all voting laws that serve a valid purpose.
The majority surmises that § 2 challenges to purge laws have not been brought because they would necessarily fail. Maj. op. at 315. The assumption on which that theory rests, of course, is that if an untried claim had any merit, it would not be untried. I do not believe that is so. In the first of the series of lectures that would be published as The Nature of the Judicial Process, then-Judge Cardozo described the judge's search for applicable precedent as an attempt "to match the colors of the case at hand against the many sample cases spread out upon their desk.” Cardozo likened this process to a simple task of flipping through a card index of decisions. However, he continued, "It is when the colors do not match, when the references in the index fail, when there is no decisive precedent, that the serious business of the judge begins." Benjamin N. Cardozo, The Nature of the Judicial Process 20-21 (1921). The absence of cases deciding § 2 challenges to purge laws only means that the legitimacy of such measures is not well-established, and that we have serious business to do here.
. Gaston County was brought under § 4(a) of the Voting Rights Act, 42 U.S.C. § 1973b(a) (currently 42 U.S.C. § 1973b(a)(l)). Section 4(a) was — and following its amendment, still is — similar to § 2, aside from its placement of the burden of proof.. It provides that a state or subdivision, in order to reinstate a suspended voting practice, must show that the practice has not been used "for the purpose or with the effect of denying or abridging the right to vote on account of race or color.”
. The majority relies on Wesley v. Collins, 791 F.2d 1255 (6th Cir.1986), which involved a § 2 challenge to a law disenfranchising convicted felons. As the majority notes, in concluding that the law did not violate the Voting Rights Act, the Wesley court emphasized “the state's legitimate and compelling rationale for enacting the statute here in issue.” Id. at 1261. Wesley is based, to a significant extent, on legal principles that justifiably limit the civil rights of convicted felons. Id. at 1261-62. In that context, a court could readily conclude that a statute disenfranchising felons was necessary to serve the state’s compelling penalogical interests. It is doubtful that the objective of restricting offenders' rights to participate in civil society could be met in a different way. In any event, because the plaintiffs here have not lost their rights by committing crimes, Wesley is distinguishable.
. See 42 U.S.C.S. § 1973gg-6(b). I discuss this legislation at greater length below, infra pp. 339-40, 346 & n. 40.
. I do not mean to suggest that the impending effectiveness of the National Voter Registration Act constitutes conclusive proof that the purge statute is not necessary. The City might be able to demonstrate that compliance will require financial, administrative, or informational resources that it does not presently possess. The district court could properly consider such an argument on remand. The impending unlawfulness of purging also does not constitute a reason to take this case any less seriously. First, elections will be held later this year, and as the majority points out, the City has already begun to purge. Citizens whose registration is nullified will not be able to vote, and they will not be automatically reinstated when the practice becomes prohibited. Second, this decision, like the discriminatory effects of the purge statute, is likely to outlive the statute itself. The cessation of purging will not cure the majority's erroneous construction of § 2.
. This statement reveals that the district court based its conclusion, at least in part, on its view of the plaintiffs’ opportunity to re-register. However, the court did not make any factual findings on this subject; we do not know how many purged voters re-registered, or how many of those who did so are black or Latino. Consequently, while the opportunity to re-register undeniably exists in a purely formal sense, we do not have even the roughest measure of its practical effect. The district court also did not explain why the possibility of re-registration was legally significant. If for no other reason — and there are other reasons — this lack of factual and analytical precision, in what is otherwise an admirably thorough and conscientious opinion, should necessitate a remand. In Jenkins, we stated that "[b]ecause the complexity of the proof and the importance of the issues involved, the district courts must be particularly thorough in explaining their findings on both the ultimate issue of vote dilution and on their subsidiary findings as well.” Jenkins, 4 F.3d at 1135 n. 35. Section 2 cases that do not involve vote dilution claims raise similarly complex and important issues and thus merit the same thoroughness. The district court, which decided Ortiz before we issued Jenkins, should be given a chance to comply with this instruction. The remainder of its opinion demonstrates that even without our guidance, the court recognized the propriety of devoting special efforts to this type of decision.
. Because the majority is in complete agreement with the district court’s legal reasoning, it treats the court's ultimate conclusion that no § 2 violation occurred here as a purely factual one that we review only for clear error. Maj. op. at 312 & n. 8. It is worth noting, however, that there is a substantial legal component to § 2 decisions, and that therefore, our standard of review, in this context is not as lenient as the majority opinion might suggest. Judge Becker, writing for the court in Jenkins v. Red Clay Consolidated School Dist. Bd. of Education, 4 F.3d 1103 (3d Cir.1993), recently provided an instruc*337tive (and controlling) explanation of the applicable standard of review in § 2 cases. Discussing the same language in Thornburg v. Gingles on which the majority relies for the standard it applies, Jenkins points out that while the district court's factual findings are entitled to deference under Rule 52(a)’s clear error standard:
"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.' " This result is clearly necessary since both the ultimate determination of vote dilution and the accompanying subsidiary determinations often depend on important legal questions. Accordingly, we must subject the district court's underlying legal analysis to plenary review to ensure that the appropriate legal standards are applied. As the [Gingles ] Court concluded, it is this combination of factual deference and legal review that best "preserves the benefit of the trial court's particular familiarity with the indigenous political reality without endangering the rule of law.”
Jenkins, 4 F.3d at 1116-17 (quoting Gingles, 478 U.S. at 79, 106 S.Ct. at 2781, in turn quoting Bose Corp. v. Consumers Union, 466 U.S. 485, 501, 104 S.Ct. 1949, 1959, 80 L.Ed.2d 502 (1984)). While the majority may not disagree with the more complex standard of review we described in Jenkins, its opinion does not fully convey the significance of the legal component of § 2 decisions.
I do not take issue with any of the subsidiary factual findings the district court made in this case. In my view, however, as I have stated, those findings cannot be reconciled with its ultimate conclusion that no § 2 violation occurred. My disagreement with the district court, therefore, strikes me as more legal than factual. In any event, my position does not depend on the applicable standard of review.
. I am aware that § 201 of the Voting Rights Act Amendments of 1970, as amended, 42 U.S.C. § 1973aa, specifically prohibits the use of literacy tests. Hypothetically, however, such prerequisites to voting could also be challenged under § 2.
. In Gaston County, North Carolina v. United States, 395 U.S. 285, 89 S.Ct. 1720, 23 L.Ed.2d 309 (1969), a county brought suit under § 4(a) of the Voting Rights Act seeking to reinstate a literacy test that it had formerly used as a prerequisite to registration. The Supreme Court held that even if the test was administered in a fair and impartial manner, it would not be permissible. Gaston County, 395 U.S. at 296, 89 S.Ct. at 1725. In the Court's view, the County had failed to disprove that "use of its literacy test, coupled with its racially segregated and unequal school system, discriminatorily deprived Negroes of the franchise.” Id. at 293, 89 S.Ct. at 1724. Of course, as the brief history of segregation and educational inequality set forth in the Court's opinion made clear, 395 U.S. at 294-96, 89 S.Ct. at 1724-26, the literacy test was not responsible for the fact that a disproportionate percentage of the black residents of Gaston County could not pass such a test. There was no evidence of a "link,” as the majority might put it, between the challenged voting law and the plaintiffs' lack of reading skills. Nonetheless, according to the Court, that social and historical condition, "coupled with” the use of even an impartially applied literacy test, resulted in a discriminatory disenfranchisement of black citizens and therefore was not permissible under the Voting Rights Act.
. The majority believes that I have overlooked the fact that individuals cannot be purged without having already demonstrated their ability to register. Maj. op. at 315. Registering, it points out, can be no harder the second time around *339ñían it was the first. Id. at 314 n. 14. I have not overlooked this point. Rather, I believe it is partly unimportant and partly untrue.
As a general matter, doing something twice is more difficult than doing it once. Registering to vote is no exception. Just because some black and Latino citizens have overcome obstacles to political participation and registered at some point in the past, there is no reason to believe that they will be able to do so again. And even if they continue to have the ability to register (with whatever assistance they may have received previously), the task of filing a registration form, in a proper and timely manner, remains. Quite simply, registering twice is harder, and will occur less frequently, than registering once.
Moreover, re-registering after being purged is different, and will often be more difficult, than registering to begin with, because individuals who register and are later purged might not even be aware that they have lost their ability to vote. Representative Collins made exactly this point during the motor-voter law debates, stating: "[I]f you are busy, as most people are, with the day to day tasks of raising a family and working or trying to find a job or whatever else, you might not realize that you have missed the last election and have been removed from the voter registration list until it is too late.” 139 Cong.Rec. H517 (daily ed. Feb. 4, 1993) (remarks of Rep. Collins). Once registered, would-be voters might also reasonably believe that the City would not nullify their efforts. Indeed, as the authors of the only commentary on purge laws point out, “Purging qualified voters who have fulfilled registration requirements will likely increase their frustration and humiliation over continued obstacles to the franchise. Furthermore, it is likely to deter voting by increasing alienation and apathy toward the political process and by intimidating qualified voters and creating a fear of encounters with election officials in the reinstatement process or at the polling place.” Barber, supra p. 26, at 523.
For the majority, the fact that purged voters have registered before means that the statute could not have violated § 2. In my view, that fact only further demonstrates and contributes to the statute’s discriminatory effect.
. The majority’s misreading of the district court’s opinion is probably a result of the following sentence: "The Court concludes that plaintiffs have failed to demonstrate that minority candidates experience difficulty electing representatives to office." Ortiz, 824 F.Supp. at 538 (emphasis added). Clearly, that statement contains a mistake. The district court was not referring to a situation, such as that following a plurality vote in the electoral college, in which (successful) candidates for office are called upon to elect other representatives. There are two possible ways to make sense of the district court’s misstatement: either the court meant to say "voters” instead of "candidates", or it meant to say “getting elected to office” instead of "electing representatives to office." The first possibility, while perhaps the one the majority would prefer, is not realistic. The district court’s heading, its description and rejection of the plaintiffs' argument, its reliance on the Senate Report, and the content of its discussion — in which it referred only to candidates who had been elected, and never to voter preferences — all point to the unmistakable conclusion that the court made a finding regarding the difficulty minority candidates had experienced in their efforts to get elected, and not the difficulty minority voters had experienced in their efforts to elect the candidates they supported.
. Conversely, and of at least equal promise, whites can and do vote for minority candidates. The record even reveals that such laudable events may occur, as one might hope, in the City of Brotherly Love; as the district court pointed out, two minority members of the Philadelphia City Council were elected in at-large contests.
.Section 2 plaintiffs bear the burden of establishing which candidates they have preferred. Jenkins, 4 F.3d at 1126. Minority voters are not entitled to an unsupported assumption that the defeat of minority candidates, in and of itself, demonstrates the frustration of minority voting interests. In this case, because the district court never considered whether black and Latino citizens had succeeded in electing representatives of their choice, we do not have any of the factual information a court would need to determine which candidates were preferred by minority constituents. For example, we do not know anything about the losing candidates in elections where a black or Latino prevailed, nor do we have any idea if statistical or even anecdotal evidence supports the permissible inference that minority voters in fact preferred any given minority representative. Again, in cases (unlike this one) where such a showing is necessary, the plaintiffs must make it. Here, however, it is the majority that equates the success of minority candidates with the success of minority voters. That assumption is not legally or factually sound.
I am aware of the district court's finding that, as a general matter, voting in Philadelphia is racially polarized. Ortiz, 824 F.Supp. at 532-33. That finding, however, is insufficient to support a conclusion that minority voters preferred particular minority candidates. As we pointed out in Jenkins, this inquiry must be conducted "on an election-by-election basis.” Jenkins, 4 F.3d at 1126.
. Putting aside the majority's misunderstanding of whose experience (candidates’ or voters') the district court was considering, it is worth noting that the court only concluded that the plaintiffs had failed to prove that someone had experienced some sort of political hardship. It did not, as the majority states, make an affirmative finding that no one had done so. This second inaccuracy in the majority's version of the district court's finding may he more subtle, but it is not insignificant.
. As a final point, by relying on Wilson Goode’s mayoral victories in support of its position that minority voters have not had difficulty electing minority candidates, the majority has adjusted or reinterpreted the district court’s findings. Goode testified that he was the only black candidate, of the twelve who ran, to be elected mayor during the past thirty years. Thus, when the district court found substantial minority representation in Philadelphia, it was in spite of Goode’s success, not because of it. Ortiz, 824 F.Supp. at 538 ("Although there has been one minority mayor over the past several years, the city council....” (emphasis added)). The majority’s use of these findings is quite different. Instead of viewing Goode’s mayoral victories, in their historical context, as facts that detract from but do not defeat an argument that minority candidates have achieved substantial success in Philadelphia elections, the majority cites Goode’s victories as only supporting that position.
. Part III.C does contain a brief reference to a district court "finding of minority participation in the political process.” Maj. op. at 314 (citing Ortiz, 824 F.Supp. at 539). However, the "finding” the majority has in mind could only be the district court's statement of its ultimate conclusion, where it simply echoes the language of § 2 in holding that no violation occurred. That "finding” (which appears under the heading "CONCLUSIONS OF LAW”) cannot support itself. In any event, once again, the majority’s argument only relies on the findings regarding electoral results. It is the success of minority candidates, it reasons, that demonstrates that minority voters have not been denied either fair access to the political process or an equal opportunity to influence the outcome of elections.
. This is not the way Justice Scalia read the Court’s opinion. In fact, Justice Scalia criticized the Chisom majority for construing § 2 in such a way that members of a minority group that was not large enough to control the outcome of elections, but who nonetheless had suffered an abridgement of their opportunity to participate in the political process, would not be protected. Chisom, 501 U.S. 380, 111 S.Ct. 2354, 115 L.Ed.2d at 372 (Scalia, J., dissenting). The Chi-som majority, however, expressly rejected that interpretation of its opinion. Id., 501 U.S. 380, 111 S.Ct. 2354, 115 L.Ed.2d at 365 n. 24. Thus, regardless of whether Justice Scalia’s criticism had merit, not one member of the Court was willing to adopt a construction of § 2 under which a measure that operated to abridge a group’s opportunity to participate, but that did not itself prevent that group from electing representatives, would be permissible.
. As I state above, supra p. 332, the extent to which minority voters have experienced difficulty electing candidates of their choice — and thus re-Iatedly, the extent to which minority candidates have won election to office — is one relevant factor, among others, that a court can take into account in assessing the effect and legality of a non-voting purge law. Political success could well encourage participation, while repeated failure could lead to disillusionment, alienation, and the belief that voting is pointless. Therefore, because this factor might influence minority participation rates, it is a relevant consideration in a challenge to a law that disenfranchises those who fail to vote. The extent of a group’s political success, however, while relevant, is not of central importance among the circumstances that affect its members' inclination or ability to participate, and is certainly not anything approaching a threshold or essential factor that § 2 plaintiffs must establish in order to successfully challenge a non-voting purge law.
. The statute provides:
Any State program or activity to protect the integrity of the electoral process by ensuring the maintenance of an accurate and current voter registration roll for elections for Federal office ... shall not result in the removal of any person from the official list of voters registered to vote in an election for Federal office by reason of the person’s failure to vote.
42 U.S.C.S. § 1973gg-6(b). Theoretically, states and localities could creatively struggle to avoid this mandate by maintaining separate registration rolls for federal elections. At oral argument, the City stated that it had no intention of making such a determined and impractical effort to preserve, as best it can, the non-voting purge law.
. See Abraham Lincoln, Address at Gettysburg, Pennsylvania (Nov. 19, 1863), in 2 Abraham Lincoln: Speeches and 'Writings 536 (Don E. Fehren-bacher, ed., 1989). Appropriately, the "unfinished work” to which Lincoln referred was the preservation of democratic government. The debates preceding the enactment of the most recent voting rights legislation show that Lincoln continues to inform our ideals about political participation. E.g., 139 Cong.Rec. S2471 (daily ed. Mar. 5, 1993) (remarks of Sen. Bradley); id. at H506 (daily ed. Feb. 4, 1993) (remarks of Rep. Thomas).