Opinion by Judge KOZINSKI; Dissent by Judge TALLMAN.
OPINION
KOZINSKI, Circuit Judge:We consider the constitutionality of Tucson’s unusual system for electing members of its city council.
FACTS
Tucson’s elections are ordinary in many ways. The city is divided into six wards of approximately equal population, and each ward is allotted one seat on the city council. A candidate for city council must run for the seat in the ward where he resides. See Tucson City Charter ch. Ill, § 1; ch. XVI, §§ 5, 8, 9. From there, things take an odd turn.
In some American cities, council seats are filled at large, with the entire city voting for each seat in the primary and general elections. In other cities, council members are nominated and elected by the residents of particular districts. Tucson splits the difference: Since 1930, the city has used a “hybrid system” that combines ward-based primaries with at-large general elections.
The first step in the hybrid system is a partisan primary. Each ward holds its own primary limited to residents of that ward. The winners of the ward primaries advance to the general election, where they compete against the other candidates nominated from that ward. In the general election, all Tucson residents can vote for one council member from each ward that held a primary during the same election cycle. See Charter ch. XVI, § 9. Thus, a resident of Ward 1 can’t vote in the Ward 2 primary, but can vote for one of the Ward 2 candidates in the general election. The parties agree that, once elected, council members represent the entire city, not just the ward from which they were nominated. See City of Tucson v. State, 229 Ariz. 172, 273 P.3d 624, 631 (2012) (“Tucson council members, although nominated by ward, represent the entire city, just as do council members elected at large in other cities.”); see also Dallas Cty. v. Reese, 421 U.S. 477, 480, 95 S.Ct. 1706, 44 L.Ed.2d 312 (1975) (“[E]lected officials represent all of those who elect them. . . .”); Fortson v. Dorsey, 379 U.S. 433, 438, 85 S.Ct. 498, 13 L.Ed.2d 401 (1965) (similar).
Council seats are filled in staggered elections, with three council members elected every other year. Once elected, a council member serves a four-year term. See Charter ch. XVI, §§ 3-4. The council members from Wards 1, 2 and 4 will be elected in 2015, and the council members from Wards 3, 5 and 6 will be elected in 2017. Because only half of the council seats are up for election in any given year, only half of Tucsonans can vote in a primary in each election cycle. And approxi*879mately 83 percent of the electorate that votes for any given council seat in the general election has no say in selecting the nominees competing for that seat.
Plaintiffs are five Tucson voters and a non-profit corporation called the Public Integrity Alliance (collectively “PIA”). PIA concedes that the city could use ward-based primaries and ward-based general elections without offending the Constitution. Similarly, the city could use at-large primaries and at-large general elections. But PIA argues that combining these two options into a hybrid system violates the federal and Arizona Constitutions1 by depriving Tucson voters of their right to vote in primary elections for individuals who will ultimately serve as their at-large representatives. PIA sued the city seeking to enjoin the hybrid system and secure a declaration that the scheme is unconstitutional. The district court ruled in favor of the city. We have jurisdiction under 28 U.S.C. § 1291.
DISCUSSION
We start by resolving a dispute between the parties that has a substantial bearing on our analysis and, ultimately, on the result we reach: Are Tucson’s primary and general elections two separate contests, each governed by rules that must be judged independently of one another — as the city contends? Or are they two parts of a single election cycle, which must be considered in tandem when determining their constitutionality — as PIA claims? The difference matters a great deal., If the two elections were separate, PIA’s constitutional objections would largely evaporate and this would become a simple case. This is so because there would be no mismatch between the voting constituency and the represented constituency in the two elections. It’s only if we view the two elections as one that serious constitutional doubts arise.
Unfortunately, the easy solution is not available because it is perfectly clear that the two contests are not independent. Instead, they are complementary components of a single election. Although the two contests are separated in time by ten weeks, they are entirely co-dependent. Without the primary, there could be no candidate to compete in the general election; without the general election, the primary winners would sit on their hands. Because a candidate must win a primary in order to compete in the general election, the “right to choose a representative is in fact controlled by the primary.” United States v. Classic, 313 U.S. 299, 319, 61 S.Ct. 1031, 85 L.Ed. 1368 (1941). Thus, the Supreme Court has held that the primary and general elections are a “single instrumentality for choice of officers.” Smith v. Allwright, 321 U.S. 649, 660, 64 S.Ct. 757, 88 L.Ed. 987 (1944); see Newberry v. United States, 256 U.S. 232, 284-86, 41 S.Ct. 469, 65 L.Ed. 913 (1921) (Pitney, J., concurring in part) (noting that the. primary and general elections are “essentially but parts of a single process”).
Because the primary and general elections are two parts of a “unitary” process, Allwright, 321 U.S. at 660-61, 64 S.Ct. 757, a citizen’s right to vote in the general election may be meaningless unless he is also permitted to vote in the primary. If a *880voter’s preferred candidate is defeated in a primary from which the voter is excluded, the voter would never have the chance to cast a ballot for his candidate of choice. Cf. Morse v. Republican Party of Va., 517 U.S. 186, 205, 116 S.Ct. 1186, 134 L.Ed.2d 347 (1996) (invalidating registration fee for Virginia senatorial nominating convention because the fée limited voters’ “influence on the field of candidates whose names [would] appear on the ballot” and thus “weakened] the ‘effectiveness’ of their votes cast in the general election itself’); Bullock v. Carter, 405 U.S. 134, 146, 92 S.Ct. 849, 31 L.Ed.2d 92 (1972) (“[T]he primary election may be more crucial than the general election.... ”); Classic, 313 U.S. at 319, 61 S.Ct. 1031 (observing that “the practical influence of the choice of candidates at the primary may be so great as to affect profoundly the choice at the general election”); Ayers-Schaffner v. DiStefano, 37 F.3d 726, 728 n. 5 (1st Cir.1994) (noting that “the ability to vote in the general election [is not] a satisfactory alternative for those voters not allowed to vote in the primary, as the candidate of their choice may have been excluded in the preliminary election from which they were barred”).
This case illustrates the point. . Although Arizona as a whole generally votes Republican, Tucson generally votes Democratic. This means that the Democratic nominee from each ward will likely win the general election regardless of whether the ward from which he was nominated is principally Republican or Democratic.- Indeed, the city’s current mayor and all six council members are Democrats. See Tucson City Council Democratic Incumbents Re-Elected, Arizona Public Media (Nov. 6, 2013), available at https://goo.gl/oMkOxi. In most cases, then, the Democratic ward primary is the only election that matters; the general election is a mere formality. Even if electing the Democratic nominee is not automatic, there is no dispute that the Democratic nominee enters the general election with an enormous advantage. Thus the vote in the primary — and particularly the Democratic primary — has a commanding influence on the outcome of the general election. Yet five-sixths of Tucson’s voters have not even a theoretical possibility of participating in the primary that will, for all practical purposes, determine who will represent them in the city council.
The Supreme Court has indicated that, “[o]nce the geographical unit for which a representative is to be chosen is designated, all who participate in the election are to have an equal vote” no matter where “their home may be in that geographical unit.” Gray v. Sanders, 372 U.S. 368, 379, 83 S.Ct. 801, 9 L.Ed.2d 821 (1963). Gray defines the “geographical unit” by reference to the constituency of “the representative to be chosen.” Id. at 379, 83 S.Ct. 801; see id. at 382, 83 S.Ct. 801 (Stewart, J., concurring) (“Within a given constituency, there can be room for but a single constitutional rule — one voter, one vote.” (emphasis added)). All parties before us agree that the constituency of each Tucson council member is the entire city. Thus, the relevant geographical unit is the city at large. Because the constituency of the representative to be elected remains static throughout the election process, the geographical unit must also remain static throughout that process.2
*881If the city were permitted to change the geographical unit between the primary and general elections, it could decouple the representative to be elected from his constituency. For example, Tucson could decree that only voters living on Main Street are eligible to vote in primaries, thereby forcing the entire city to choose among nominees selected by a tiny minority of residents. Or the State of New York, in an effort to cap its number of city-slicker senators, could limit the primary for its junior senator to Manhattanites and the primary for its senior senator to the rest of the state. We do not believe that such mismatches between voters at different stages of a single election cycle are constitutionally permissible.
Given the city’s concession that each council member represents all of Tucson, it’s clear that the representational nexus runs between the city and the council member, not between the ward and the council member. But the hybrid system makes the tenure of each at-large council member largely dependent on the preferences of voters of his home ward; without their support, a council member could not be nominated (or re-nominated) in the first place. Given that reality, each council member will be disproportionately responsive to voters from his home ward, especially those of his own party. The city claims that this is a redeeming benefit of its hybrid system. The exact opposite is true. The practical effect of the Tucson system is to give some of a representative’s constituents — those in his home ward — a vote of disproportionate weight. That is the very result the Supreme Court’s one person, one vote jurisprudence is meant to foreclose. See Reynolds v. Sims, 377 U.S. 533, 560-64, 84 S.Ct. 1362, 12 L.Ed.2d 506 (1964). We cannot endorse an election system that encourages at-large representatives to prioritize kissing babies and currying favor in their home wards over the interests of their constituents who happen to live in other parts of the city. As the Supreme Court itself has noted, an at-large representative “must be vigilant to serve the interests of all the people in the [city]; and not merely those of people in his home [ward].” Fortson, 379 U.S. at 438, 85 S.Ct. 498.
We hold that every otherwise eligible voter who will be a constituent of the winner of the general election must have an equal opportunity to participate in each election cycle through which that candidate is selected. Just as the city could not exclude a resident of Ward 1 from voting in the general election for his council member from Ward 2, so the city may not exclude that resident from a primary election for the same official. See Alhmight, 321 U.S. at 664, 64 S.Ct. 757 (“[T]he same tests to determine the character of discrimination or abridgement should be applied to the primary as are applied to the *882general election.”); Classic, 313 U.S. at 318, 61 S.Ct. 1031 (“[The] right of participation [in the nominating process] is protected just as is the right to vote at the election.... ”).
The city’s final argument is that the hybrid system is a reasonable “residency restriction” on the right to vote. But when two groups of citizens share identical interests in an election, the city may not use a residency requirement to exclude one group while including the other. See Town of Lockport v. Citizens for Comm. Action at the Local Level, Inc., 430 U.S. 259, 268, 97 S.Ct. 1047, 51 L.Ed.2d 313 (1977) (residency requirements must be premised on a “genuine difference in the relevant interests of the groups that the state electoral classification has created”); id. (excluded group must be permitted to vote if it has “substantially identical interests” as included group); Evans v. Comman, 398 U.S. 419, 422-26, 90 S.Ct. 1752, 26 L.Ed.2d 370 (1970) (residents of federal enclave within Maryland couldn’t be excluded from the franchise because they had “a stake equal to that of other Maryland residents”); see also Holt Civic Club v. City of Tuscaloosa, 439 U.S. 60, 72 n. 8, 99 S.Ct. 383, 58 L.Ed.2d 292 (1978) (suggesting that a city might be required to enfranchise non-residents if it were “exercising precisely the same governmental powers over [them] as it does over those residing within its corporate limits”).3 In this case, the out-of-ward Tucsonans who are excluded from the ward primaries have precisely the same interests in those primaries as do the ward residents who are permitted to participate. The nominees selected in the ward primaries will advance to the general election; if elected there, they will represent the entire city. Because all Tucsonans have an equal interest in determining who the nominees will be, the city may not exclude out-of-ward voters from the primaries.
Little Thunder v. South Dakota, 518 F.2d 1253 (8th Cir.1975), is instructive. That case involved a challenge to South Dakota’s scheme for governing its unorganized counties. The residents of the unorganized counties were governed by elected officials in the nearest organized county, but only the residents of the organized county were allowed to vote for those officials. The state defended this scheme as a reasonable residency requirement. Id. at 1255. In its view, the residents of the unorganized counties (who were mainly Native Americans) did not share the same interests in the elections as did the residents of the organized counties. The Eighth Circuit rejected the argument. Citing Comman, that court held that a state may not use a residency requirement to prevent citizens from voting for “those who will function as their elected officials.” Id. at 1258. The court applied strict scrutiny and invalidated the scheme. Id.
The fact that two groups live on opposite sides of a political boundary does not necessarily mean they can be treated differ*883ently for voting purposes. This is the teaching of Little Thunder. 518 F.2d at 1256; see English v. Bd. of Educ. of Town of Boonton, 301 F.3d 69, 77, 79 (3d Cir.2002); United States v. South Dakota, 636 F.2d 241, 245 (8th Cir.1980); see also Holt Civic Club, 439 U.S. at 81, 86, 99 S.Ct. 383 (Brennan, J., dissenting) (cautioning against “eed[ing] to geography a talismanic significance”). If two groups are represented by the same politician, they are necessarily part of a “single unit of local government.” Little Thunder, 518 F.2d at 1256. Any boundary that purports to subdivide that single unit is hopelessly arbitrary, and any “residency restriction” that disenfranchises citizens based on where they live in relation to that arbitrary boundary cannot stand. Excluding out-of-ward voters from the primary election discriminates among residents of the same governmental unit in violation of the Equal Protection Clause of the Fourteenth Amendment.
REVERSED.
. PIA alleges that the hybrid system violates the Free and Equal Elections Clause of the Arizona Constitution. Ariz. Const, art. II, § 21. We have been cited no authority indicating that the rights guaranteed by that document differ from those guaranteed by the federal Constitution. Because PIA did not develop any state-law arguments in its appellate briefing, we consider the state-law claims abandoned. See Greenwood v. F.A.A., 28 F.3d 971, 977 (9th Cir.1994).
. We are not persuaded by the city's reliance on two decades-old district court opinions that dealt with hybrid systems for judicial elections. Holshouser v. Scott and Stokes v. Fortson involved challenges to state laws providing that judges would be nominated from their districts but elected statewide in the general election. In both cases, three-judge district courts ruled that the principle of one person, one vote is not applicable to judicial *881elections. Both courts went on to observe that, even if that were not the case, the hybrid schemes would not violate one person, one vote because they didn’t involve dilution or an unequal counting of votes. See Holshouser, 335 F.Supp. 928, 930, 933 (M.D.N.C.1971); Stokes, 234 F.Supp. 575, 577 (N.D.Ga.1964). The city argues that the Supreme Court’s summary affirmance in Holshouser, 409 U.S. 807, 93 S.Ct. 43, 34 L.Ed.2d 68 (1972), is a ruling on the merits that requires us to uphold Tucson’s hybrid system. But a “summary affirmance without opinion in a case within the Supreme Court’s obligatory appellate jurisdiction has very little precedential significance.” Dillenburg v. Kramer, 469 F.2d 1222, 1225 (9th Cir.1972). It does not enshrine as Supreme Court precedent every stroke of the pen in the district court's opinion. The summary disposition in Holshouser was likely intended to affirm the proposition that one person, one vote does not apply to judicial elections, as the Court eventually held in Chisom v. Roemer, 501 U.S. 380, 402-03, 111 S.Ct. 2354, 115 L.Ed.2d 348 (1991).
. Nothing we say has any bearing on the city’s existing candidate-residency requirement, which requires each council member to run for the seat from the ward in which he resides. See Tucson City Charter ch. XVI, §§ 5, 9. The Supreme Court has twice upheld similar schemes. Dusch v. Davis, 387 U.S. 112, 114-16, 87 S.Ct. 1554, 18 L.Ed.2d 656 (1967); Dallas Cty. v. Reese, 421 U.S. 477, 480-81, 95 S.Ct. 1706, 44 L.Ed.2d 312 (1975). In light of Dusch and Reese, the city argues that we are bound to approve the voter-residency requirements imposed by the hybrid system. But, despite the similarity in names, candidate-residency requirements are quite different than voter-residency requirements. Neither Dusch nor Reese requires that the same constitutional principles governing candidate-residency requirements also apply to voter-residency requirements. See Dusch, 387 U.S. at 115-16, 87 S.Ct. 1554.