Opinion

JUSTICE O'CONNOR delivered the opinion of the Court.

In this case, we decide whether Georgia's State Senate redistricting plan should have been precleared under § 5 of the Voting Rights Act of 1965, 79 Stat. 439, as renumbered and amended, 42 U. S. C. § 1973c. Section 5 requires that before a covered jurisdiction's new voting "standard, prac tice, or procedure" goes into effect, it must be precleared by either the Attorney General of the United States or a federal court to ensure that the change "does not have the purpose and will not have the effect of denying or abridging the right to vote on account of race or color." 42 U. S. C. § 1973c. Whether a voting procedure change should be precleared depends on whether the change "would lead to a retrogression in the position of racial minorities with respect to their effective exercise of the electoral franchise." Beer v. United States, 425 U. S. 130, 141 (1976). We therefore must decide whether Georgia's State Senate redistricting plan is retrogressive as compared to its previous, benchmark districting plan.

I

A

Over the past decade, the propriety of Georgia's state and congressional districts has been the subject of repeated litigation. In 1991, the Georgia General Assembly began the process of redistricting after the 1990 census. Because Georgia is a covered jurisdiction under § 5 of the Voting Rights Act, see Miller v. Johnson, 515 U. S. 900, 905 (1995), Georgia submitted its revised State Senate plan to the United States Department of Justice for preclearance. The plan as enacted into law increased the number of majorityminority districts from the previous Senate plan. The Department of Justice nevertheless refused preclearance because of Georgia's failure to maximize the number of majority-minority districts. See Johnson v. Miller, 929 F. Supp. 1529, 1537, and n. 23 (SD Ga. 1996). After Georgia made changes to the Senate plan in an attempt to satisfy the United States' objections, the State again submitted it to the Department of Justice for preclearance. Again, the Department of Justice refused preclearance because the plan did not contain a sufficient number of majority-minority districts. See id., at 1537, 1539. Finally, the United States precleared Georgia's third redistricting plan, approving it in the spring of 1992. See id., at 1537.

Georgia's 1992 Senate plan was not challenged in court.

See id., at 1533-1534. Its congressional districting plan, however, was challenged as unconstitutional under the Equal Protection Clause of the Fourteenth Amendment. See Shaw v. Reno, 509 U. S. 630 (1993). In 1995, we held in Miller v. Johnson that Georgia's congressional districting plan was unconstitutional because it engaged in "the very racial stereotyping the Fourteenth Amendment forbids" by making race the "predominant, overriding factor explaining" Georgia's congressional districting decisions. 515 U. S., at 928, 920. And even though it was "safe to say that the congressional plan enacted in the end was required in order to obtain preclearance," this justification did not permit Georgia to engage in racial gerrymandering. See id., at 921. Georgia's State Senate districts served as "building blocks" to create the congressional districting plan found unconstitutional in Miller v. Johnson. Johnson v. Miller, 929 F. Supp., at 1533, n. 8 (internal quotation marks omitted); see also id., at 1536.

Georgia recognized that after Miller v. Johnson, its legislative districts were unconstitutional under the Equal Protection Clause. See 929 F. Supp., at 1533, 1540. Accordingly, Georgia attempted to cure the perceived constitutional problems with the 1992 State Senate districting plan by passing another plan in 1995. The Department of Justice refused to preclear the 1995 plan, maintaining that it retrogressed from the 1992 plan and that Miller v. Johnson concerned only Georgia's congressional districts, not Georgia's State Senate districts. See 929 F. Supp., at 1540-1541.

Private litigants subsequently brought an action challenging the constitutionality of the 1995 Senate plan. See id., at 1533. The three-judge panel of the District Court reviewing the 1995 Senate plan found that "[i]t is clear that a black maximization policy had become an integral part of the sec tion 5 preclearance process... when the Georgia redistricting plans were under review. The net effect of the DOJ's preclearance objection[s]... was to require the State of Georgia to increase the number of majority black districts in its redistricting plans, which were already ameliorative plans, beyond any reasonable concept of non-retrogression." Id., at 1539-1540. The court noted that in Miller v. Johnson, we specifically disapproved of the Department of Justice's policy that the maximization of black districts was a part of the § 5 retrogression analysis. See 929 F. Supp., at 1539. Indeed, in Miller, we found that the Department of Justice's objections to Georgia's redistricting plans were "driven by its policy of maximizing majority-black districts." 515 U. S., at 924. And "[i]n utilizing § 5 to require States to create majority-minority districts wherever possible, the Department of Justice expanded its authority under the statute beyond what Congress intended and we have upheld." Id., at 925.

The District Court stated that the maximization of majority-minority districts in Georgia "artificially push[ed] the percentage of black voters within some majority black districts as high as possible." 929 F. Supp., at 1536. The plan that eventually received the Department of Justice's preclearance in 1992 "represented the General Assembly's surrender to the black maximization policy of the DOJ." Id., at 1540. The court then found that the 1995 plan was an unconstitutional racial gerrymander. See id., at 1543.

Under court direction, Georgia and the Department of Justice reached a mediated agreement on the constitutionality of the 1995 Senate plan. Georgia passed a new plan in 1997, and the Department of Justice quickly precleared it. The redrawn map resembled to a large degree the 1992 plan that eventually received preclearance from the Department of Justice, with some changes to accommodate the decision of this Court in Miller v. Johnson, and of the District Court in Johnson v. Miller. All parties here concede that the 1997 plan is the benchmark plan for this litigation because it was in effect at the time of the 2001 redistricting effort. The 1997 plan drew 56 districts, 11 of them with a total black population of over 50%, and 10 of them with a black voting age population of over 50%. See Record, Doc. No. 148, Pl. Exh. 1C (hereinafter Pl. Exh.). The 2000 census revealed that these numbers had increased so that 13 districts had a black population of at least 50%, with the black voting age population exceeding 50% in 12 of those districts. See 195 F. Supp. 2d 25, 39 (DC 2002).

After the 2000 census, the Georgia General Assembly began the process of redistricting the Senate once again. No party contests that a substantial majority of black voters in Georgia vote Democratic, or that all elected black representatives in the General Assembly are Democrats. The goal of the Democratic leadership-black and white-was to maintain the number of majority-minority districts and also increase the number of Democratic Senate seats. See id., at 41-42. For example, the Director of Georgia's Legislative Redistricting Office, Linda Meggers, testified that the Senate Black Caucus" 'wanted to maintain'" the existing majorityminority districts and at the same time" 'not waste'" votes. Id., at 41.

The Vice Chairman of the Senate Reapportionment Committee, Senator Robert Brown, also testified about the goals of the redistricting effort. Senator Brown, who is black, chaired the subcommittee that developed the Senate plan at issue here. See id., at 42. Senator Brown believed when he designed the Senate plan that as the black voting age population in a district increased beyond what was necessary, it would "pus[h] the whole thing more towards [the] Republican[s]." Pl. Exh. 20, at 24. And "correspondingly," Senator Brown stated, "the more you diminish the power of African-Americans overall." Ibid. Senator Charles Walker was the majority leader of the Senate. Senator Walker testified that it was important to attempt to maintain a Democratic majority in the Senate because "we [AfricanAmericans] have a better chance to participate in the political process under the Democratic majority than we would have under a Republican majority." PI. Exh. 24, at 19. At least 7 of the 11 black members of the Senate could chair committees. See 195 F. Supp. 2d, at 41.

The plan as designed by Senator Brown's committee kept true to the dual goals of maintaining at least as many majority-minority districts while also attempting to increase Democratic strength in the Senate. Part of the Democrats' strategy was not only to maintain the number of majorityminority districts, but to increase the number of so-called "influence" districts, where black voters would be able to exert a significant-if not decisive-force in the election process. As the majority leader testified, "in the past, you know, what we would end up doing was packing. You put all blacks in one district and all whites in one district, so what you end up with is [a] black Democratic district and [a] white Republican district. That's not a good strategy. That does not bring the people together, it divides the population. But if you put people together on voting precincts it brings people together." PI. Exh. 24, at 19.

The plan as designed by the Senate "unpacked" the most heavily concentrated majority-minority districts in the benchmark plan, and created a number of new influence districts. The new plan drew 13 districts with a majority-black voting age population, 13 additional districts with a black voting age population of between 30% and 50%, and 4 other districts with a black voting age population of between 25% and 30%. See PI. Exh. 2C. According to the 2000 census, as compared to the benchmark plan, the new plan reduced by five the number of districts with a black voting age population in excess of 60%. Compare PI. Exh. 1D with PI. Exh. 2C. Yet it increased the number of majority-black voting age population districts by one, and it increased the number of districts with a black voting age population of between 25% and 50% by four. As compared to the benchmark plan enacted in 1997, the difference is even larger. Under the old census figures, Georgia had 10 Senate districts with a majority-black voting age population, and 8 Senate districts with a black voting age population of between 30% and 50%. See PI. Exh. 1 C. The new plan thus increased the number of districts with a majority black voting age population by three, and increased the number of districts with a black voting age population of between 30% and 50% by another five. Compare PI. Exh. 1C with PI. Exh. 2C.

The Senate adopted its new districting plan on August 10, 2001, by a vote of 29 to 26. Ten of the eleven black Senators voted for the plan. 195 F. Supp. 2d, at 55. The Georgia House of Representatives passed the Senate plan by a vote of 101 to 71. Thirty-three of the thirty-four black Representatives voted for the plan. Ibid. No Republican in either the House or the Senate voted for the plan, making the votes of the black legislators necessary for passage. See id., at 41. The Governor signed the Senate plan into law on August 24,2001, and Georgia subsequently sought to obtain preclearance.

B

Pursuant to § 5 of the Voting Rights Act, a covered jurisdiction like Georgia has the option of either seeking administrative preclearance through the Attorney General of the United States or seeking judicial preclearance by instituting an action in the United States District Court for the District of Columbia for a declaratory judgment that the voting change comports with § 5. 42 U. S. C. § 1973c; Georgia v. United States, 411 U. S. 526 (1973). Georgia chose the latter method, filing suit seeking a declaratory judgment that the State Senate plan does not violate § 5.

Georgia, which bears the burden of proof in this action, see Pleasant Grove v. United States, 479 U. S. 462 (1987), attempted to prove that its Senate plan was not retrogres sive either in intent or in effect. It submitted detailed evidence documenting in each district the total population, the total black population, the black voting age population, the percentage of black registered voters, and the overall percentage of Democratic votes (i. e., the overall likelihood that voters in a particular district will vote Democratic), among other things. See 195 F. Supp. 2d, at 36; see also PI. Exhs. 2C, 2D. The State also submitted evidence about how each of these statistics compared to the benchmark districts. See 195 F. Supp. 2d, at 36; see also PI. Exhs. lC, lD, lE (revised).

Georgia also submitted testimony from numerous people who had participated in enacting the Senate plan into law, and from United States Congressman John Lewis, who represents the Atlanta area. These witnesses testified that the new Senate plan was designed to increase black voting strength throughout the State as well as to help ensure a continued Democratic majority in the Senate. The State also submitted expert testimony that African-American and non-African-American voters have equal chances of electing their preferred candidate when the black voting age population of a district is at 44.3%. Finally, in response to objections raised by the United States, Georgia submitted more detailed statistical evidence with respect to three proposed Senate districts that the United States found objectionableDistricts 2, 12, and 26-and two districts that the intervenors challenged-Districts 15 and 22.

The United States, through the Attorney General, argued in District Court that Georgia's 2001 Senate redistricting plan should not be precleared. It argued that the plan's changes to the boundaries of Districts 2, 12, and 26 unlawfully reduced the ability of black voters to elect candidates of their choice. See Brief for Federal Appellees 8; 195 F. Supp. 2d, at 72. The United States noted that in District 2, the black voting age population dropped from 60.58% to 50.31 %; in District 12, the black voting age population dropped from 55.43% to 50.66%; and in District 26, the black voting age population dropped from 62.45% to 50.80%.1 Moreover, in all three of these districts, the percentage of black registered voters dropped to just under 50%. The United States also submitted expert evidence that voting is racially polarized in Senate Districts 2, 12, and 26. See id., at 69-71. The United States acknowledged that some limited percentage of whites would vote for a black candidate, but maintained that the percentage was not sufficient for black voters to elect their candidate of choice. See id., at 70-71. The United States also offered testimony from various witnesses, including lay witnesses living in the three districts, who asserted that the new contours of Districts 2, 12, and 26 would reduce the opportunity for blacks to elect a candidate of their choice in those districts; Senator Regina Thomas of District 2, the only black Senator who voted against the plan; Senator Eric Johnson, the Republican leader of the Senate; and some black legislators who voted for the plan but questioned how the plan would affect black voters. See Vols. 25-27 Record, Doc. No. 177, United States Exhs. 707-736 (Depositions). As the District Court stated, "the United States' evidence was extremely limited in scope-focusing only on three contested districts in the State Senate plan. That evidence was not designed to permit the court to assess the overall impact of [the Senate plan]." 195

Pursuant to Federal Rule of Civil Procedure 24, the District Court also permitted four African-American citizens of Georgia to intervene. The intervenors identified two other districts-Districts 15 and 22-where they alleged retrogression had occurred. The intervenors "present[ed] little evidence other than proposed alternative plans and an expert report critiquing the State's expert report." 195 F. Supp. 2d, at 37.

A three-judge panel of the District Court held that Georgia's State Senate apportionment violated § 5, and was therefore not entitled to preclearance. See id., at 97. Judge Sullivan, joined by Judge Edwards, concluded that Georgia had "not demonstrated by a preponderance of the evidence that the State Senate redistricting plan would not have a retrogressive effect on African American voters" effective exercise of the electoral franchise. Ibid. The court found that Senate Districts 2, 12, and 26 were retrogressive because in each district, a lesser opportunity existed for the black candidate of choice to win election under the new plan than under the benchmark plan. See id., at 93-94. The court found that the reductions in black voting age population in Districts 2, 12, and 26 would "diminish African American voting strength in these districts," and that Georgia had "failed to present any... evidence" that the retrogression in those districts "will be offset by gains in other districts." Id., at 88. Judge Edwards, joined by Judge Sullivan, concurred.

Judge Edwards emphasized that §§ 5 and 2 are "procedurally and substantively distinct provisions." Id., at 97. He therefore rejected Georgia's argument that a plan preserving an equal opportunity for minorities to elect candidates of their choice satisfies § 5. Judge Edwards also rejected the testimony of the black Georgia politicians who supported the Senate plan. In his view, the testimony did not address whether racial polarization was occurring in Senate Districts 2, 12, and 26. See id., at 101-102.

Judge Oberdorfer dissented. He would have given "greater credence to the political expertise and motivation of Georgia's African-American political leaders and reasonable inferences drawn from their testimony and the voting data and statistics." Id., at 102. He noted that this Court has not answered "whether a redistricting plan that preserves or increases the number of districts statewide in which minorities have a fair or reasonable opportunity to elect candidates of choice is entitled to preclearance, or whether every district must remain at or improve on the benchmark probability of victory, even if doing so maintains a minority supermajority far in excess of the level needed for effective exercise of [the] electoral franchise." Id., at 117.

After the District Court refused to preclear the plan, Georgia enacted another plan, largely similar to the one at issue here, except that it added black voters to Districts 2, 12, and 26. The District Court precleared this plan. See 204 F. Supp. 2d 4 (2002). No party has contested the propriety of the District Court's preclearance of the Senate plan as amended. Georgia asserts that it will use the plan as originally enacted if it receives preclearance.

We noted probable jurisdiction to consider whether the District Court should have precleared the plan as originally enacted by Georgia in 2001, 537 U. S. 1151 (2003), and now vacate the judgment below. II

Before addressing the merits of Georgia's preclearance claim, we address the State's argument that the District Court was incorrect in allowing the private litigants to intervene in this lawsuit. Georgia maintains that private parties should not be allowed to intervene in § 5 actions because States should not be subjected to the political stratagems of intervenors. While the United States disagrees with Georgia on the propriety of intervention here, the United States argues that this question is moot because the participation of the intervenors did not affect the District Court's ruling on the merits and the intervenors did not appeal the court's ruling.

We do not think Georgia's argument is moot. The intervenors did not have to appeal because they were prevailing parties below. Moreover, the District Court addressed the evidence that the intervenors submitted, which is now in front of this Court. The issue whether intervenors are proper parties still has relevance in this Court because they argue here that the District Court correctly found that the Senate plan was retrogressive.

The District Court properly found that Federal Rule of Civil Procedure 24 governs intervention in this case. Section 5 permits a State to bring "an action in the United States District Court for the District of Columbia for a declaratory judgment." 42 U. S. C. § 1973c. Section 5 does not limit in any way the application of the Federal Rules of Civil Procedure to this type of lawsuit, and the statute by its terms does not bar private parties from intervening. In NAACP v. New York, 413 U. S. 345, 365 (1973), we held that in an action under § 5, "[i]ntervention in a federal court suit is governed by Fed. Rule Civ. Proc. 24."

To support its argument, Georgia relies on Morris v. Gressette, 432 U. S. 491 (1977). In Morris, we held that in an administrative preclearance action, the decision to object belongs only to the Attorney General and is not judicially reviewable. See id., at 504-505. But Morris concerned the administrative preclearance process, not the judicial preclearance process. Morris itself recognized the difference between administrative preclearance and judicial preclearance. See id., at 503-507.

Here, the District Court granted the motion to intervene because it found that the intervenors' "analysis of the... Senate redistricting pla[n] identifies interests that are not adequately represented by the existing parties." App. to Juris. Statement 218a. Private parties may intervene in § 5 actions assuming they meet the requirements of Rule 24, and the District Court did not abuse its discretion in granting the motion to intervene in this case. See NAACP v. New York, supra, at 367.

III

A

Section 5 of the Voting Rights Act "has a limited substantive goal: "'to insure that no voting-procedure changes would be made that would lead to a retrogression in the position of racial minorities with respect to their effective exercise of the electoral franchise.''' Miller, 515 U. S., at 926 (quoting Beer v. United States, 425 U. S., [at 141])." Bush v. Vera, 517 U. S. 952, 982-983 (1996). Thus, a plan that merely preserves "current minority voting strength" is entitled to § 5 preclearance. City of Lockhart v. United States, 460 U. S. 125, 134, n. 10 (1983); Bush v. Vera, supra, at 983. Indeed, a voting change with a discriminatory but nonretrogressive purpose or effect does not violate § 5. See Reno v. Bossier Parish School Bd., 528 U. S. 320, 341 (2000). And "no matter how unconstitutional it may be," a plan that is not retrogressive should be precleared under § 5. Id., at 336. "[P]reclearance under § 5 affirms nothing but the absence of backsliding." Id., at 335.

Georgia argues that a plan should be precleared under § 5 if the plan would satisfy § 2 of the Voting Rights Act of 1965, 42 U. S. C. § 1973. We have, however, "consistently understood" § 2 to "combat different evils and, accordingly, to impose very different duties upon the States." Reno v. Bossier Parish School Bd., 520 U. S. 471, 477 (1997) (Bossier Parish 1). For example, while § 5 is limited to particular covered jurisdictions, § 2 applies to all States. And the § 2 inquiry differs in significant respects from a § 5 inquiry. In contrast to § 5's retrogression standard, the "essence" of a § 2 vote dilution claim is that "a certain electoral law, practice, or structure... cause[s] an inequality in the opportunities enjoyed by black and white voters to elect their preferred representatives." Thornburg v. Gingles, 478 U. S. 30, 47 (1986); see also id., at 48-50 (enunciating a three-part test to establish vote dilution); id., at 85-100 (O'CONNOR, J., concurring in judgment); 42 U. S. C. § 1973(b). Unlike an inquiry under § 2, a retrogression inquiry under § 5, "by definition, requires a comparison of a jurisdiction's new voting plan with its existing plan." Bossier Parish I, supra, at 478. While some parts of the § 2 analysis may overlap with the § 5 inquiry, the two sections "differ in structure, purpose, and application." Holder v. Hall, 512 U. S. 874, 883 (1994) (plurality opinion).

In Bossier Parish I, we specifically held that a violation of § 2 is not an independent reason to deny preclearance under § 5. See 520 U. S., at 477. The reason for this holding was straightforward: "[R]ecognizing § 2 violations as a basis for denying § 5 preclearance would inevitably make compliance with § 5 contingent upon compliance with § 2. Doing so would, for all intents and purposes, replace the standards for § 5 with those for § 2." Ibid.

Georgia here makes the flip side of the argument that failed in Bossier Parish I-compliance with § 2 suffices for preclearance under § 5. Yet the argument fails here for the same reasons the argument failed in Bossier Parish 1. We refuse to equate a § 2 vote dilution inquiry with the § 5 retrogression standard. Georgia's argument, like the argument in Bossier Parish I, would "shift the focus of § 5 from nonretrogression to vote dilution, and [would] change the § 5 benchmark from a jurisdiction's existing plan to a hypothetical, undiluted plan." Id., at 480. Instead of showing that the Senate plan is nondilutive under § 2, Georgia must prove that its plan is nonretrogressive under § 5.

B

Georgia argues that even if compliance with § 2 does not automatically result in preclearance under § 5, its State Senate plan should be precleared because it does not lead to "a retrogression in the position of racial minorities with respect to their effective exercise of the electoral franchise." Beer v. United States, supra, at 141. See, e. g., Brief for Appellant 32, 36.

While we have never determined the meaning of "effective exercise of the electoral franchise," this case requires us to do so in some detail. First, the United States and the District Court correctly acknowledge that in examining whether the new plan is retrogressive, the inquiry must encompass the entire statewide plan as a whole. See 195 F. Supp. 2d, at 73; Tr. of Oral Arg. 28-29. Thus, while the diminution of a minority group's effective exercise of the electoral franchise in one or two districts may be sufficient to show a violation of § 5, it is only sufficient if the covered jurisdiction cannot show that the gains in the plan as a whole offset the loss in a particular district.

Second, any assessment of the retrogression of a minority group's effective exercise of the electoral franchise depends on an examination of all the relevant circumstances, such as the ability of minority voters to elect their candidate of choice, the extent of the minority group's opportunity to participate in the political process, and the feasibility of creating a nonretrogressive plan. See, e. g., Johnson v. De Grandy, 512 U. S. 997, 1011-1012, 1020-1021 (1994); Richmond v. United States, 422 U. S. 358, 371-372 (1975); Thornburg v. Gingles, supra, at 97-100 (O'CONNOR, J., concurring in judgment). "No single statistic provides courts with a shortcut to determine whether" a voting change retrogresses from the benchmark. Johnson v. De Grandy, supra, at 1020-1021.

In assessing the totality of the circumstances, a court should not focus solely on the comparative ability of a minority group to elect a candidate of its choice. While this factor is an important one in the § 5 retrogression inquiry, it cannot be dispositive or exclusive. The standard in § 5 is simplewhether the new plan "would lead to a retrogression in the position of racial minorities with respect to their effective exercise of the electoral franchise." Beer v. United States, 425 U. S., at 141.

The ability of minority voters to elect a candidate of their choice is important but often complex in practice to determine. In order to maximize the electoral success of a minority group, a State may choose to create a certain number of "safe" districts, in which it is highly likely that minority voters will be able to elect the candidate of their choice. See Thornburg v. Gingles, 478 U. S., at 48-49; id., at 87-89 (O'CONNOR, J., concurring in judgment). Alternatively, a State may choose to create a greater number of districts in which it is likely-although perhaps not quite as likely as under the benchmark plan-that minority voters will be able to elect candidates of their choice. See id., at 88-89 (O'CONNOR, J., concurring in judgment); cf. Pildes, Is Voting-Rights Law Now at War With Itself? Social Science and Voting Rights in the 2000s, 80 N. C. L. Rev. 1517 (2002).

Section 5 does not dictate that a State must pick one of these methods of redistricting over another. Either option "will present the minority group with its own array of electoral risks and benefits," and presents "hard choices about what would truly 'maximize' minority electoral success." Thornburg v. Gingles, supra, at 89 (O'CONNOR, J., concurring in judgment). On one hand, a smaller number of safe majority-minority districts may virtually guarantee the election of a minority group's preferred candidate in those districts. Yet even if this concentration of minority voters in a few districts does not constitute the unlawful packing of minority voters, see Voinovich v. Quilter, 507 U. S. 146, 153154 (1993), such a plan risks isolating minority voters from the rest of the State, and risks narrowing political influence to only a fraction of political districts. Cf. Shaw v. Reno, 509 U. S., at 648-650. And while such districts may result in more "descriptive representation" because the representatives of choice are more likely to mirror the race of the majority of voters in that district, the representation may be limited to fewer areas. See H. Pitkin, The Concept of Representation 60-91 (1967).

On the other hand, spreading out minority voters over a greater number of districts creates more districts in which minority voters may have the opportunity to elect a candidate of their choice. Such a strategy has the potential to increase "substantive representation" in more districts, by creating coalitions of voters who together will help to achieve the electoral aspirations of the minority group. See id., at 114. It also, however, creates the risk that the minority group's preferred candidate may lose. Yet as we stated in Johnson v. De Grandy, supra, at 1020:"[T]here are communities in which minority citizens are able to form coalitions with voters from other racial and ethnic groups, having no need to be a majority within a single district in order to elect candidates of their choice. Those candidates may not represent perfection to every minority voter, but minority voters are not immune from the obligation to pull, haul, and trade to find common political ground, the virtue of which is not to be slighted in applying a statute meant to hasten the waning of racism in American politics." Section 5 gives States the flexibility to choose one theory of effective representation over the other.

In addition to the comparative ability of a minority group to elect a candidate of its choice, the other highly relevant factor in a retrogression inquiry is the extent to which a new plan changes the minority group's opportunity to participate in the political process. "'[T]he power to influence the political process is not limited to winning elections.'" Thornburg v. Gingles, supra, at 99 (O'CONNOR, J., concurring in judgment) (quoting Davis v. Bandemer, 478 U. S. 109, 132 (1986)); see also White v. Regester, 412 U. S. 755, 766-767 (1973); Whitcomb v. Chavis, 403 U. S. 124, 149-160 (1971); Johnson v. De Grandy, 512 U. S., at 1011-1012.

Thus, a court must examine whether a new plan adds or subtracts "influence districts"-where minority voters may not be able to elect a candidate of choice but can play a substantial, if not decisive, role in the electoral process. Cf. Shaw v. Hunt, 517 U. S. 899, 947, n. 21 (1996) (STEVENS, J., dissenting); Hays v. Louisiana, 936 F. Supp. 360, 364, n. 17 (WD La. 1996); Johnson v. De Grandy, supra, at 1011 1012; Thornburg v. Gingles, 478 U. S., at 98-100 (O'CONNOR, J., concurring in judgment). In assessing the comparative weight of these influence districts, it is important to consider "the likelihood that candidates elected without decisive minority support would be willing to take the minority's interests into account." Id., at 100 (O'CONNOR, J., concurring in judgment). In fact, various studies have suggested that the most effective way to maximize minority voting strength may be to create more influence or coalitional districts. See, e. g., Lublin, Racial Redistricting and African-American Representation: A Critique of "Do Majority-Minority Districts Maximize Substantive Black Representation in Congress?" 93 Am. Pol. Sci. Rev. 183, 185 (1999) (noting that racial redistricting in the early 1990's, which created more majorityminority districts, made Congress "less likely to adopt initiatives supported by blacks"); Cameron, Epstein, & O'Halloran, Do Majority-Minority Districts Maximize Substantive Black Representation in Congress? 90 Am. Pol. Sci. Rev. 794, 808 (1996) (concluding that the "[d]istricting schemes that maximize the number of minority representatives do not necessarily maximize substantive minority representation"); C. Swain, Black Faces, Black Interests 193234 (1995); Pildes, 80 N. C. L. Rev., at 1517; Grofman, Handley, & Lublin, Drawing Effective Minority Districts: A Conceptual Framework and Some Empirical Evidence, 79 N. C. L. Rev. 1383 (2001).

Section 5 leaves room for States to use these types of influence and coalitional districts. Indeed, the State's choice ultimately may rest on a political choice of whether substantive or descriptive representation is preferable. See Pitkin, supra, at 142; Swain, supra, at 5. The State may choose, consistent with § 5, that it is better to risk having fewer minority representatives in order to achieve greater overall representation of a minority group by increasing the number of representatives sympathetic to the interests of minority voters. See Thornburg v. Gingles, supra, at 87-89, 99 (O'CONNOR, J., concurring in judgment); cf. Johnson v. De Grandy, 512 U. S., at 1020.

In addition to influence districts, one other method of assessing the minority group's opportunity to participate in the political process is to examine the comparative position of legislative leadership, influence, and power for representatives of the benchmark majority-minority districts. A legislator, no less than a voter, is "not immune from the obligation to pull, haul, and trade to find common political ground." Ibid. Indeed, in a representative democracy, the very purpose of voting is to delegate to chosen representatives the power to make and pass laws. The ability to exert more control over that process is at the core of exercising political power. A lawmaker with more legislative influence has more potential to set the agenda, to participate in closeddoor meetings, to negotiate from a stronger position, and to shake hands on a deal. Maintaining or increasing legislative positions of power for minority voters' representatives of choice, while not dispositive by itself, can show the lack of retrogressive effect under § 5.

And it is also significant, though not dispositive, whether the representatives elected from the very districts created and protected by the Voting Rights Act support the new districting plan. The District Court held that the support of legislators from benchmark majority-minority districts may show retrogressive purpose, but it is not relevant in assessing retrogressive effect. See 195 F. Supp. 2d, at 89; see also post, at 503 (SOUTER, J., dissenting). But we think this evidence is also relevant for retrogressive effect. As the dissent recognizes, the retrogression inquiry asks how "voters will probably act in the circumstances in which they live." Post, at 509. The representatives of districts created to ensure continued minority participation in the political process have some knowledge about how "voters will probably act" and whether the proposed change will decrease minority voters' effective exercise of the electoral franchise.

The dissent maintains that standards for determining nonretrogression under § 5 that we announce today create a situation where "[i]t is very hard to see anything left of" § 5. Post, at 495. But the dissent ignores that the ability of a minority group to elect a candidate of choice remains an integral feature in any § 5 analysis. Cf. Thornburg v. Gingles, supra, at 98 (O'CONNOR, J., concurring in judgment). And the dissent agrees that the addition or subtraction of coalitional districts is relevant to the § 5 inquiry. See post, at 492, 504. Yet assessing whether a plan with coalitional districts is retrogressive is just as fact-intensive as whether a plan with both influence and coalitional districts is retrogressive. As JUSTICE SOUTER recognized for the Court in the § 2 context, a court or the Department of Justice should assess the totality of circumstances in determining retrogression under § 5. See Johnson v. De Grandy, supra, at 1020-1021. And it is of course true that evidence of racial polarization is one of many factors relevant in assessing whether a minority group is able to elect a candidate of choice or to exert a significant influence in a particular district. See Thornburg v. Gingles, 478 U. S., at 37; id., at 100104 (O'CONNOR, J., concurring in judgment); see also White v. Regester, 412 U. S. 755 (1973); Zimmer v. McKeithen, 485

The dissent nevertheless asserts that it "cannot be right" that the § 5 inquiry goes beyond assessing whether a minority group can elect a candidate of its choice. Post, at 494. But except for the general statement of retrogression in Beer, the dissent cites no law to support its contention that retrogression should focus solely on the ability of a minority group to elect a candidate of choice. As JUSTICE SOUTER himself, writing for the Court in Johnson v. De Grandy, supra, at 1011-1012, has recognized, the "extent of the opportunities minority voters enjoy to participate in the political processes" is an important factor to consider in assessing a § 2 vote-dilution inquiry. See also Thornburg v. Gingles, supra, at 98-100 (O'CONNOR, J., concurring in judgment). In determining how the new districting plan differs from the benchmark plan, the same standard should apply to § 5.

C

The District Court failed to consider all the relevant factors when it examined whether Georgia's Senate plan resulted in a retrogression of black voters' effective exercise of the electoral franchise. First, while the District Court acknowledged the importance of assessing the statewide plan as a whole, the court focused too narrowly on proposed Senate Districts 2, 12, and 26. It did not examine the increases in the black voting age population that occurred in many of the other districts. Second, the District Court did not explore in any meaningful depth any other factor beyond the comparative ability of black voters in the majority minority districts to elect a candidate of their choice. In doing so, it paid inadequate attention to the support of legislators representing the benchmark majority-minority districts and the maintenance of the legislative influence of those representatives.

The District Court correctly recognized that the increase in districts with a substantial minority of black voters is an important factor in the retrogression inquiry. See 195 F. Supp. 2d, at 75-78. Nevertheless, it did not adequately apply this consideration to the facts of this case. The District Court ignored the evidence of numerous other districts showing an increase in black voting age population, as well as the other evidence that Georgia decided that a way to increase black voting strength was to adopt a plan that "unpacked" the high concentration of minority voters in the majority-minority districts. Its statement that Georgia did not "presen[t] evidence regarding potential gains in minority voting strength in Senate Districts other than Districts 2, 12 and 26" is therefore clearly erroneous. Id., at 94. Like the dissent, we accept the District Court's findings that the reductions in black voting age population in proposed Districts 2, 12, and 26 to just over 50% make it marginally less likely that minority voters can elect a candidate of their choice in those districts, although we note that Georgia introduced evidence showing that approximately one-third of white voters would support a black candidate in those districts, see id., at 66, and that the United States' own expert admitted that the results of statewide elections in Georgia show that "there would be a 'very good chance' that... African American candidates would win election in the reconstituted districts." Id., at 71; see also id., at 84-85. Nevertheless, regardless of any racially polarized voting or diminished opportunity for black voters to elect a candidate of their choice in proposed Districts 2, 12, and 26, the District Court's inquiry was too narrow. In the face of Georgia's evidence that the Senate plan as a whole is not retrogressive, the United States introduced nothing apart from the evidence that it would be more difficult for minority voters to elect their candidate of choice in Districts 2, 12, and 26. As the District Court stated, the United States did not introduce any evidence to rebut Georgia's evidence that the increase in black voting age population in the other districts offsets any decrease in black voting age population in the three contested districts: "[T]he United States' evidence was extremely limited in scope-focusing only on three contested districts in the State Senate plan." Id., at 37. Indeed, the District Court noted that the United States' evidence "was not designed to permit the court to assess the overall impact" of the Senate plan. Ibid.

Given the evidence submitted in this case, we find that Georgia likely met its burden of showing nonretrogression. The increase in black voting age population in the other districts likely offsets any marginal decrease in the black voting age population in the three districts that the District Court found retrogressive. U sing the overlay of the 2000 census numbers, Georgia's strategy of "unpacking" minority voters in some districts to create more influence and coalitional districts is apparent. Under the 2000 census numbers, the number of majority black voting age population districts in the new plan increases by one, the number of districts with a black voting age population of between 30% and 50% increases by two, and the number of districts with a black voting age population of between 25% and 30% increases by another 2. See PI. Exhs. lD, 2C; see also supra, at 470-471.

Using the census numbers in effect at the time the benchmark plan was enacted to assess the benchmark plan, the difference is even more striking. Under those figures, the new plan increases from 10 to 13 the number of districts with a majority-black voting age population and increases from 8 to 13 the number of districts with a black voting age population of between 30% and 50%. See PI. Exhs. lC, 2C. Thus, the new plan creates 8 new districts-out of 56-where black voters as a group can playa substantial or decisive role in the electoral process. Indeed, under the census figures in use at the time Georgia enacted its benchmark plan, the black voting age population in Districts 2, 12, and 26 does not decrease to the extent indicated by the District Court. District 2 drops from 59.27% black voting age population to 50.31 %. District 26 drops from 53.45% black voting age population to 50.80%. And District 12 actually increases, from 46.50% black voting age population to 50.66%. See PI. Exhs. lC, 2C.2 And regardless of any potential retrogression in some districts, § 5 permits Georgia to offset the decline in those districts with an increase in the black voting age population in other districts. The testimony from those who designed the Senate plan confirms what the statistics suggest-that Georgia's goal was to "unpack" the minority voters from a few districts to increase blacks' effective exer cise of the electoral franchise in more districts. See supra, at 469-471.

Other evidence supports the implausibility of finding retrogression here. An examination of black voters' opportunities to participate in the political process shows, if anything, an increase in the effective exercise of the electoral franchise. It certainly does not indicate retrogression. The 34 districts in the proposed plan with a black voting age population of above 20% consist almost entirely of districts that have an overall percentage of Democratic votes of above 50%. See PI. Exh. 2D. The one exception is proposed District 4, with a black voting age population of 30.51% and an overall Democratic percentage of 48.86%. See ibid. These statistics make it more likely as a matter of fact that black voters will constitute an effective voting bloc, even if they cannot always elect the candidate of their choice. See Thornburg v. Gingles, 478 U. S., at 100 (O'CONNOR, J., concurring in judgment). These statistics also buttress the testimony of the designers of the plan such as Senator Brown, who stated that the goal of the plan was to maintain or increase black voting strength and relatedly to increase the prospects of Democratic victory. See supra, at 469-470.

The testimony of Congressman John Lewis is not so easily dismissed. Congressman Lewis is not a member of the State Senate and thus has less at stake personally in the outcome of this litigation. Congressman Lewis testified that "giving real power to black voters comes from the kind of redistricting efforts the State of Georgia has made," and that the Senate plan "will give real meaning to voting for African Americans" because "you have a greater chance of putting in office people that are going to be responsive." PI. Exh. 21, at 21-23. Section 5 gives States the flexibility to implement the type of plan that Georgia has submitted for preclearance-a plan that increases the number of districts with a majority-black voting age population, even if it means that in some of those districts, minority voters will face a somewhat reduced opportunity to elect a candidate of their choice. Cf. Thornburg v. Gingles, supra, at 89 (O'CONNOR, J., concurring in judgment).

The dissent's analysis presumes that we are deciding that Georgia's Senate plan is not retrogressive. See post, at 501508. To the contrary, we hold only that the District Court did not engage in the correct retrogression analysis because it focused too heavily on the ability of the minority group to elect a candidate of its choice in the majority-minority districts. While the District Court engaged in a thorough analysis of the issue, we must remand the case for the District Court to examine the facts using the standard that we announce today. We leave it for the District Court to determine whether Georgia has indeed met its burden of proof. The dissent justifies its conclusion here on the ground that the District Court did not clearly err in its factual determination. But the dissent does not appear to dispute that if the District Court's legal standard was incorrect, the decision below should be vacated.

The purpose of the Voting Rights Act is to prevent discrimination in the exercise of the electoral franchise and to foster our transformation to a society that is no longer fixated on race. Cf. Johnson v. De Grandy, 512 U. S., at 1020; Shaw v. Reno, 509 U. S., at 657. As Congressman Lewis stated: "I think that's what the [civil rights] struggle was all about, to create what I like to call a truly interracial democracy in the South. In the movement, we would call it creating the beloved community, an all-inclusive community, where we would be able to forget about race and color and see people as people, as human beings, just as citizens." PI. Exh. 21, at 14. While courts and the Department of Justice should be vigilant in ensuring that States neither reduce the effective exercise of the electoral franchise nor discriminate against minority voters, the Voting Rights Act, as properly interpreted, should encourage the transition to a society where race no longer matters: a society where integration and color-blindness are not just qualities to be proud of, but are simple facts of life. See Shaw v. Reno, supra, at 657.

IV

The District Court is in a better position to reweigh all the facts in the record in the first instance in light of our explication of retrogression. The judgment of the District Court for the District of Columbia, accordingly, is vacated, and the case is remanded for further proceedings consistent with this opinion.

It is so ordered.

Notes

* A brief of amicus curiae urging affirmance was filed for the Georgia Coalition for the Peoples' Agenda by Laughlin McDonald, Neil Bradley, Barbara R. Arnwine, Thomas J. Henderson, Anita Hodgkiss, Elaine R. Jones, Norman J. Chachkin, and Todd A. Cox.

1 Georgia and the United States have submitted slightly different figures regarding the black voting age population of each district. The differing figures depend upon whether the total number of blacks includes those people who self-identify as both black and a member of another minority group, such as Hispanic. Georgia counts this group of people, while the United States does not do so. Like the District Court, we consider all the record information, "including total black population, black registration numbers and both [black voting age population] numbers." 195 F. Supp. 2d 25, 79 (DC 2002). We focus in particular on Georgia's black voting age population numbers in this case because all parties rely on them to some extent and because Georgia used its own black voting age population numbers when it enacted the Senate plan. Moreover, the United States does not count all persons who identify themselves as black. It counts those who say they are black and those who say that they are both black and white, but it does not count those who say they are both black and a member of another minority group. U sing the United States' numbers may have more relevance if the case involves a comparison of different minority groups. Cf. Johnson v. De Grandy, 512 U. S. 997 (1994); Bush v. Vera, 517 U. S. 952 (1996). Here, however, the case involves an examination of only one minority group's effective exercise of the electoral franchise. In such circumstances, we believe it is proper to look at all individuals who identify themselves as black.

2 The dissent summarily rejects any inquiry into the benchmark plan using the census numbers in effect at the time the redistricting plan was passed. See post, at 506. Yet we think it is relevant to examine how the new plan differs from the benchmark plan as originally enacted by the legislature. The § 5 inquiry, after all, revolves around the change from the previous plan. The 1990 census numbers are far from "irrelevant." Ibid. Rather, examining the benchmark plan with the census numbers in effect at the time the State enacted its plan comports with the one-person, one-vote principle of Reynolds v. Sims, 377 U. S. 533 (1964), and its progeny. When the decennial census numbers are released, States must redistrict to account for any changes or shifts in population. But before the new census, States operate under the legal fiction that even 10 years later, the plans are constitutionally apportioned. Mter the new enumeration, no districting plan is likely to be legally enforceable if challenged, given the shifts and changes in a population over 10 years. And if the State has not redistricted in response to the new census figures, a federal court will ensure that the districts comply with the one-person, one-vote mandate before the next election. See, e. g., Branch v. Smith, 538 U. S. 254 (2003); Lawyer v. Department of Justice, 521 U. S. 567 (1997); Growe v. Emison, 507 U. S. 25 (1993).