The New York Republican State Committee will hold a primary on March 7, 1996 in which it will select its delegates to the 1996 Republican National Convention. The rules of the National Republican Party mandate that three convention delegates are to be elected in each congressional district. To be eligible for the ballot under New York law, delegate candidates need to collect signatures from the lesser of five percent or 1250 of the enrolled Republicans in the delegates' congressional district. The United States District Court for the Eastern District of New York (Korman, J.) issued an injunction on December 1, 1995, striking down the "five percent or 1250" rule as a violation of the Fourteenth Amendment's Equal Protection Clause and instituting in its place a "1.41 percent or 1250" rule. See Rockefeller v. Powers, 909 F. Supp. 863 (E.D.N.Y.1995).1 The question presented is whether the injunction should be upheld.

The Republican State Committee moved for an expedited appeal, so that delegate candidates who were in the process of collecting signatures for nominating petitions due on January 4, 1996, would know what the signature requirement was. On December 5, 1995, we granted the motion to expedite the appeal. On December 21, 1995, after oral argument, we vacated the injunction and announced that an opinion would follow.

* The plaintiffs are enrolled New York Republicans, residing in the Tenth, Eleventh, Twelfth, and Fourteenth congressional districts.2 On November 13, 1995, they sued the New York State Board of Elections, the New York City Board of Elections, four county boards of elections, the New York Republican State Committee, the commissioners of all these bodies, and the Republican National Committee. The plaintiffs alleged in their complaint that the ballot access rules for the 1996 New York Republican presidential primary election--requiring petitions signed by the lesser of five percent or 1250 registered Republicans in each congressional district--violate their equal protection and First Amendment rights.3 A clear understanding of those rules and their source is essential for a proper resolution of this case.

In presidential election years, the National Republican Party holds a convention at which it chooses its nominees for President and Vice-President, adopts a platform expressing Party policy positions, and takes care of internal Party affairs. Among other things, the delegates adopt the rules that govern delegate selection for the next convention. The rules adopted at the 1992 Convention provide that each state may send to the 1996 Convention six at-large delegates, several "victory bonus" delegates (assigned to states that voted Republican in the 1992 presidential election or that have prominent or numerous Republican officeholders), and three delegates "for each Representative" in the House of Representatives. Rules of the Republican Party (1992), Rule 31(a)(2). These last delegates "shall be elected by each [ ] Congressional district." Id. Rule 32(b)(5). All delegates "shall be elected ... [i]n accordance with any applicable laws of a state, insofar as the same are not inconsistent with these rules." Id. Rule 32(a)(1).

On August 8, 1995, the New York Legislature adopted Sessions Law 586 to govern the selection of delegates to national political party conventions in 1996. 1995 N.Y.Laws 586. Political parties must conduct their delegate selection processes under either section 2 or section 3 of the law. Id. Sec. 1.

Section 3, the path traditionally associated with the Democratic Party, we are told, gives three choices: one, the party can select its delegates by holding a statewide presidential primary "in which the names of candidates [for President] appear on the ballot," id. Sec. 3(1)(a); two, it can hold direct elections for delegates in each congressional district, id. Sec. 3(1)(b); three, it can decide on the party's nominee by an internal party meeting or convention. Id. Sec. 3(1)(c). If it chooses to elect delegates directly by congressional districts, delegate candidates must file petitions signed by the least of (i) 1,000 of the enrolled party voters in their district, (ii) 0.5 percent of such voters, or (iii) a number of such voters equal to 0.5 percent of all votes cast in the district in the 1992 New York Democratic presidential primary. Id. Sec. 3(6)(c). None of the provisions of section 3 is challenged here.

Section 2, we are told, is the provision traditionally associated with the Republican Party, and it was in fact the one chosen by the Republican State Committee for its 1996 primary. Section 2, once chosen by the political party, requires that delegates be selected in direct elections conducted in each congressional district. Id. Sec. 2(2)(a). When filing their candidacies under section 2, delegates either declare their alignment with a particular presidential candidate or declare themselves "uncommitted." Id. Sec. 2(4)(a). Their allegiance to a presidential candidate or their uncommitted status is shown on the ballot next to their names. Id. Sec. 2(4)(c). The business end of section 2 states: "Except as provided in this section and party rules and regulations, the provisions of the election law shall apply to elections conducted pursuant to this section." Id. Sec. 2(6).

Neither section 2 of Sessions Law 586 nor the Republican Party rules and regulations say anything about the signature requirements for placing delegate slates on the ballot. Therefore, "the provisions of the [New York State] election law" dealing with signature requirements apply to the 1996 Republican Party primary.4 Specifically, section 6-136 of New York Election Law provides that candidates "[f]or any office to be filled by all the voters of any congressional district" need to file petitions with signatures from five percent or 1250 of enrolled party voters from the district, whichever is less. N.Y.Elec.Law Sec. 6-136(2)(g) (McKinney 1995). This numerical requirement is applied to party convention delegates, on a permissive basis, by section 6-136(3), which states that "[t]he number of signatures on a petition ... to designate a candidate for the position of district delegate to a national party convention need not exceed the number required for a petition for representative in congress." Id. Sec. 6-136(3). Thus, for the 1996 Republican primary, the "five percent or 1250" signature rule applies.5

The gravamen of the plaintiffs' claim as it reaches us on appeal is that this signature requirement--the lesser of a fixed number or a fixed percentage in each congressional district--violates the Equal Protection Clause by discriminating against voters from congressional districts having relatively few enrolled Republicans. The plaintiffs explain this by statistics. New York has thirty-one congressional districts. Republicans are most numerous in the Third District in suburban Long Island: 158,097 Republicans are registered there. Since five percent of this number, 7905, exceeds 1250, the five percent threshold does not apply. Thus, Republican delegate candidates in the Third District need 1250 signatures to qualify for the ballot, a mere 0.79 percent of the district's enrolled Republicans. In contrast, Republicans are least numerous in the Eleventh District in Brooklyn, where only 11,814 Republicans are registered. Since five percent of this number, 591, is fewer than 1250, Republican delegate candidates in the Eleventh District need only 591 signatures to qualify for the ballot.

The plaintiffs thus demonstrate that a smaller percentage of Republicans in the Third (as compared to the Eleventh) congressional district can secure a spot on the ballot for a slate of candidates, and characterize this as a disparity of constitutional dimension. The indisputable arithmetic is that in districts like the Eleventh, having fewer than 25,000 enrolled Republicans, delegate candidates need the signatures of five percent of enrolled Republicans to get on the ballot, while in heavily Republican districts like the Third, delegate candidates need the signatures of (variously) only 0.81 percent (the Twenty-Second), 0.97 percent (the First) or 1.10 percent (the Nineteenth) of enrolled Republicans. Of the thirty-one districts, seven have an effective signature requirement for the 1996 primary of one percent or less, and sixteen (a majority) have a signature requirement of 1.41 percent or less.6

The result, say the plaintiffs, is less choice for Republican voters in districts like the Eleventh. As proof, they point to the last contested Republican presidential primary in New York, in 1988. The four principal contenders were George Bush, Robert Dole, Jack Kemp, and Pat Robertson. After analyzing the delegate slates in all congressional districts, the plaintiffs report that in the eight districts with the fewest registered Republicans, four had only Bush delegates on the ballot, three had Bush and Kemp delegates on the ballot, and one had Bush and Dole delegates. In contrast, in ten of the districts "with greater numbers" of enrolled Republicans, two districts had delegates representing all four candidates, and the other eight had delegates representing George Bush and two others.

According to the plaintiffs, the reason for this disparity in ballot choice is the disparity of burdens created by the split-level signature requirement. The plaintiffs claim that in the heavily-Republican rural or suburban districts, where gathering signatures is as easy as "stand[ing] in the mall or at the town grocer's and solicit[ing] signatures from everyone who walks by," the 1250 signature requirement softens an already easy job by reducing the required percentage of signatures well below five percent. On the other hand, the plaintiffs claim that in the largely urban districts with relatively few Republicans, where "a petition-carrier could wait for a week on Flatbush Avenue and not encounter a single Brooklyn Republican!", the onerous five percent requirement is unmitigated.

As one can see from the 1988 data (set forth in the margin7 ), however, there are additional observations that may be made from the same data upon which the plaintiffs rely. First, the ten districts "with greater numbers," cited by the plaintiffs for the proposition that heavily Republican districts have more ballot choices, were selectively chosen. Those ten districts range from the second most Republican district to the nineteenth: the more significant observation concerning these nineteen districts is that, while roughly half had three or more candidates (as the plaintiffs point out), almost half had either no delegates or only Bush delegates on their ballots. Second, if the districts are divided between the seventeen most Republican districts and the seventeen least Republican districts, exactly seven districts in each group had either no delegates or only Bush delegates on the ballots. Third, Kemp delegates do not seem to have been disadvantaged by the signature requirement: they appeared on ballots in eight of the top seventeen and seven of the bottom seventeen Republican districts. Finally, the only three districts in which no candidate qualified for the ballot were heavily Republican (the Twenty-Eighth, Twenty-Ninth, and Thirty-Fourth).8

Nevertheless, the district court agreed with the plaintiffs that the signature requirement created disparate levels of choice among districts, and that this disparity violated the Equal Protection Clause. The five percent or 1250 signature requirement "prevent[ed] 'independent' candidates from obtaining the requisite number of signatures," said the court, and harmed "particular voters whose right to vote in a particular election is rendered meaningless."9 Rockefeller v. Powers, 909 F. Supp. 863, 868 (E.D.N.Y.1995). Applying the strict scrutiny test from Illinois State Bd. of Elections v. Socialist Workers Party, 440 U.S. 173, 184, 99 S. Ct. 983, 59 L. Ed. 2d 230 (1979), the court held that no compelling interest existed for the five percent or 1250 signature requirement. In fact, wrote the court, "[t]he only justification appears to be a desire to increase the advantage already enjoyed by the presidential candidate favored by the Party organization," Rockefeller, 909 F. Supp. at 868, with the result that the Republican primary has now become "little more than a state-sponsored endorsement of the candidate of the party leadership." Id. at 869. The court found that the only "legitimate objective" advanced by the five percent or 1250 signature requirement is the assurance that delegate candidates have "a minimum modicum of support," but that this objective "could be achieved in a far more rational and less restrictive way."10 Id. at 869.

The district court thought that it had found such a way and embodied it in the injunction, explaining as follows. The "minimum modicum of support" rationale makes sense only if "the 1250 cap ... that liberalizes ballot access" was "an exception" to the five percent rule. Id.

at * 871. Under the framework established by New York's Legislature, however, twenty-five of the thirty-one districts "benefit" from the 1250 cap while only six districts have to meet the five percent requirement. The 1250 cap is therefore the "rule", reasoned the court, and the extra "burden" created by the five percent rule is the "exception." Since the district court viewed the 1250 cap as having a liberalizing or ameliorative influence on the burden imposed by the five percent rule, the signature requirement scheme could be deemed "least restrictive" only if the 1250 voter cap becomes the exception to the percentage rule. The proper fix, according to the district court, was to replace five percent with 1.41 percent,because in the majority of the districts, delegate candidates were required to collect signatures from 1.41 percent or more of the district's registered Republicans. See note 6, supra. In this way, the 1250 cap would be returned to its proper status as "an exception that liberalizes ballot access," id. at 871, since it would now apply in a minority of the districts, and the 1.41 percent requirement would now be the "rule," applying to sixteen of the thirty-one districts. See generally id. at * 864-66, 871. The district court issued an injunction effectively requiring that the defendants enforce the signature requirement as if it were a "1.41 percent or 1250 rule."11

The New York State Republican Committee and its chairman appealed. They argued that this case is nonjusticiable, that the plaintiffs lack standing, that the rational basis test applies to the disparate treatment created by the signature requirement, and that a rational basis for it exists. The plaintiffs cross-appealed, arguing that "the logic of the remedy granted by the district court requires that the modicum of support requirement be fixed at 0.79 percent, or at most 1.07 percent, rather than 1.41 percent."

On December 21, 1995, we reversed and restored the five percent or 1250 signature requirement. This opinion explains why.

The defendants argue that this case is nonjusticiable because the plaintiffs are inviting us to meddle in the internal affairs of a political party. However, the statutory signature requirement scheme at issue is not an internal party rule, and is easily classified as state action. The defendants' justiciability arguments therefore fail.

Under the National Republican Party rules, most of the delegates to the national convention are selected by registered Republicans in each of the nation's 435 congressional districts. The plaintiffs do not challenge this implicit determination by the Republican Party that a geographically diverse delegate group improves its choice of presidential and vice-presidential candidates. Rather, they challenge the State's signature requirement process, embodied in Sessions Law 586 and New York Election Law Sec. 6-136. The rules adopted by the National Republican Party and by the Republican State Committee have no five percent minimum, no 1250 cap, and in fact no signature requirement at all. The requirement, as applied to the 1996 primary, is purely a product of two state laws. We conclude that the signature requirement scheme challenged by the plaintiffs constitutes state action that courts may examine to ensure that it does not violate the Equal Protection Clause.

True, the State offered the Republican State Committee a choice of schemes, and the particular scheme at issue is being employed because it was chosen by the Committee. Although the Republican State Committee is therefore the but-for cause of the supposed disparity, the Committee's selection of one of two state-offered options does not abrogate the state action. Pursuant to Sessions Law 586, the State has determined that the five percent or 1250 signature requirement is an appropriate way to make sure that only delegate candidates with a modicum of support make it on the ballot. The fact that the State Republican Committee agrees with the State that this is an appropriate ballot access rule does not change our analysis. We do not consider the justiciability of a challenge to the State Republican Committee's choice of the option embodied in section 2. The challenge here is to state laws that create (although they do not necessarily impose) the five percent or 1250 signature requirement.

In Bullock v. Carter, 405 U.S. 134, 92 S. Ct. 849, 31 L. Ed. 2d 92 (1972)--the Supreme Court ballot access case most similar to ours in many respects--the Court struck down a Texas law requiring that candidates in primary elections pay a filing fee to get on the ballot. The law, however, set no fee amounts, leaving it completely up to the judgment of the state's political parties what fee to charge (so long as it was, "in their judgment [ ] just and equitable"). Id. at 138, 92 S.Ct. at 853. Each political party imposed its own fees (and different fees for different offices in different counties) with the result that ballot-access fees varied from $50 to $6,300. The Court found state action, even though the filing fee requirement applied only to primaries and the political parties were free to fix whatever fees they wished. Id. at 140, 92 S.Ct. at 854 ("the mechanism ... is the creature of state legislative choice and hence is 'state action' within the meaning of the Fourteenth Amendment"). If it is state action for a legislature to adopt a ballot access rule so permissive and indulgent as the one struck down in Bullock, it is state action a fortiori for the New York Legislature to adopt a ballot-access rule that (at least on its face) narrows to two the ballot-access choices available to each party.

In reaching our conclusion that this is state action, we are also guided by our prior decision in Montano v. Lefkowitz, 575 F.2d 378 (2d Cir.1978) (Friendly, J.). A Bronx congressman retired in mid-term. New York law allowed each political party to nominate a replacement candidate "in such manner as the party rules prescribed." Id. at 380. Although the congressional district covered less than all of the Bronx, the State Democratic Party delegated this task to the Executive Committee of the Bronx County Democratic Party. The Executive Committee included party members from throughout Bronx County, some of whom represented districts other than the newly vacated one. Democratic voters in the vacated district sued, claiming that their equal protection rights were violated by allowing party members from outside their district to participate in an intra-district nomination. Id.

Even though the State had not enacted or specified the nominating process, we held that "New York's delegation to the various parties of the right to nominate candidates for special elections renders the party selection process state action." Id. at 383 (citations omitted); see also id. at 383 n. 7 ("[W]e join with most commentators and many lower courts in holding that when the state grants political parties the right to nominate candidates and then gives those nominees special access to the ballot, the parties' procedures constitute state action." (internal quotations and citations omitted)).

The argument for state action is stronger here than in Montano. Here, the State itself created the scheme at issue. Faced with an allegedly unconstitutional act of the State, we inspect it. See Gray v. Sanders, 372 U.S. 368, 374, 83 S. Ct. 801, 805, 9 L. Ed. 2d 821 (1963) (political party's actions in conducting state-regulated primary constitute state action); Mrazek v. Suffolk County Bd. of Elections, 630 F.2d 890, 894 (2d Cir.1980).12

The defendants argue that the plaintiffs lack standing. To establish a case or controversy under Article III, " the plaintiff must have suffered an injury in fact ..., there must be a causal connection between the injury and the conduct complained of ..., [and 3] it must be likely, as opposed to merely speculative, that the injury will be redressed by a favorable decision." Lujan v. Defenders of Wildlife, 504 U.S. 555, 561, 112 S. Ct. 2130, 2136, 119 L. Ed. 2d 351 (1992) (internal quotations and citations omitted). The defendants point out that, although the plaintiffs claim standing by virtue of their alleged lack of choice in the Republican presidential primary, there is no guarantee that a change in the ballot access rule would widen their choices, particularly since the plaintiffs cannot identify candidates who have already been or definitely will be excluded by the signature requirement. Accordingly, say the defendants, there is no "injury in fact" that can be "redressed by a favorable decision."

This argument is insupportable in light of Northeastern Florida Chapter of Associated General Contractors v. City of Jacksonville, 508 U.S. 656, 113 S. Ct. 2297, 124 L. Ed. 2d 586 (1993), an equal protection case concerning a minority set-aside program for government contracts. There, the Court explained that contractors challenging the rule need not show that, but for the set-aside rule, they definitely would have received a contract. Rather, they merely had to establish that their ability to compete was impaired. Stated more generally, "[t]he 'injury in fact' in an equal protection case of this variety is the denial of equal treatment resulting from the imposition of the barrier, not the ultimate inability to obtain the benefit." Id. at ----, 113 S.Ct. at 2303. The plaintiffs have stated this kind of "injury in fact."

The plaintiffs claim that the five percent or 1250 formula operates to restrict their choice in Republican presidential primaries. If their statistical arguments are credited, voters living in districts with relatively few Republicans would likely have a more narrow choice of delegates pledged to the various presidential candidates. Under Northeastern Florida, they need not establish that, absent the current rule, they necessarily would see more candidates on their ballot. Their claim that the rule decreases the likelihood that they will have choices among delegates amounts to a sufficient pleading of "injury in fact" and a "causal connection."13

The redressability requirement is also satisfied, although not perhaps by the redress the plaintiffs would prefer. The district court imposed a 1.41 percent requirement for half the districts, and plaintiffs seek an across-the-board 0.79 percent requirement. We have some reservations about each of those remedies, since we suspect that the selection of a particular percentage requirement is legislative rather than judicial activity. However, a court redressing the plaintiffs' claim could certainly strike down the 1250 signature alternative, leaving a five percent requirement across-the-board (which no one seems to want); or it could strike down section 2 of Sessions Law 586 in its entirety, leaving the Republican Party with the other option in the statute--the one apparently associated with the Democratic Party. We conclude therefore that the plaintiffs' claim, if it exists, could be redressed.

Finally, we reject the defendants' argument that, because "the best party" to bring this claim is a delegate candidate, not non-candidate voters, the plaintiffs lack "prudential standing." Voters have standing to bring claims alleging that their district was unconstitutionally drawn because they suffer "special harms" from this constitutional violation. United States v. Hays, --- U.S. ----, ----, 115 S. Ct. 2431, 2436, 132 L. Ed. 2d 635 (1995). The plaintiffs allege here that the state action decreases their voting choices relative to voters in other districts. This harm is particular to them and would not be felt in those terms by the delegate candidates who abandoned the race or failed to collect enough signatures. As said in Bullock, "the rights of voters and the rights of candidates do not lend themselves to neat separation." 405 U.S. at 143, 92 S.Ct. at 856. We therefore conclude that the plaintiffs have standing to bring this suit.

To obtain a preliminary injunction in this Circuit, the moving party must show (1) the likelihood of irreparable injury, and (2) either (a) likelihood of success on the merits, or (b) sufficiently serious questions going to the merits and a balance of hardships tipping decidedly in the movant's favor.

Tough Traveler, Ltd. v. Outbound Prods., 60 F.3d 964, 967 (2d Cir.1995). Assuming irreparable injury, the district court premised the injunction on the basis of "likelihood of success on the merits." Rockefeller, 909 F. Supp. at 871-72. Our analysis leads us to the conclusion that the plaintiffs have not shown that they are likely to succeed on the merits.

The district court held that the State's five percent or 1250 rule does not withstand strict scrutiny. It chose the strict scrutiny test by applying Illinois State Bd. of Elections v. Socialist Workers Party, 440 U.S. 173, 99 S. Ct. 983, 59 L. Ed. 2d 230 (1979). There, the Court struck down Illinois's requirement that independent candidates and "new political parties" collect 37,000 signatures to secure a place on the ballot in Chicago general elections, but only 25,000 to get on the ballot in statewide general elections, despite the fact that the voting population of Chicago was one-sixth as large as the statewide voting population.14 Id. at 175-77, 183-84, 99 S. Ct. at 985-86, 990. The Court explained its decision to apply strict scrutiny as follows:

Restrictions on access to the ballot burden two distinct and fundamental rights, "the right of individuals to associate for the advancement of political beliefs, and the right of qualified voters, regardless of their political persuasion, to cast their votes effectively." Williams v. Rhodes, [393 U.S. 23, 30, [89 S. Ct. 5, 10, 21 L. Ed. 2d 24] (1968) ].... By limiting the choices available to voters, the State impairs the voters' ability to express their political preferences....

When such vital individual rights are at stake, a State must establish that its classification is necessary to serve a compelling interest.

Id. at 184, 99 S.Ct. at 990.15 We believe that the district court was incorrect to conclude, as it apparently did, that the broad language of Socialist Workers Party automatically requires strict scrutiny for ballot-access measures that result in any sort of disparate treatment.16 But we need not decide here the exact scope of Socialist Workers Party, because the plaintiffs have not shown that the disparity in signature percentages significantly burdens their fundamental right to vote or significantly impairs their ability to express political preferences. Thus, even considering Socialist Workers Party as the controlling legal framework, we still conclude that strict scrutiny is not required.17

The injury claimed by the plaintiffs is that on March 7, 1996, they and other voters in districts with few Republicans will likely have fewer delegate slates from which to choose, as compared with their fellow Republicans who live in heavily Republican districts. The plaintiffs characterize this alleged decrease in choice as a virtual denial of their fundamental right to vote: when "choice is foreclosed," they say, "voting itself becomes a sham." This is overstatement at best. The plaintiffs' argument is not really that choice is "foreclosed." Rather, they are making an empirical claim that the five percent or 1250 rule (as incorporated in Sessions Law 586) establishes a likelihood of significantly reduced choice in districts with fewer Republicans compared to more heavily Republican districts. In making this claim, the plaintiffs rely on evidence from the last contested Republican presidential primary in New York, the 1988 election in which George Bush, Robert Dole, Jack Kemp, and Pat Robertson were candidates. Plaintiffs report that in the 1988 primary, Bush delegates were the only delegates on the ballot in four of the eight districts with the fewest Republicans.18 In addition, as the chart at note 7 indicates, supra, in the fifteen districts with the fewest Republicans, no district had more than two candidates. In the fifteen districts with the most Republicans,19 on the other hand, six districts had three candidates and two had four.

The plaintiffs' statistical evidence does not establish, however, that the five percent signature requirement significantly limits their choices as compared with voters living in districts subject to the 1250 signature requirement. Specifically, the evidence does not indicate a causal connection between the challenged regulation and the alleged injury. True, a choice of one, or none, is arguably no choice at all. But a closer look at all of the districts substantially undermines the plaintiffs' argument. There appears to be no appreciable correlation between the number of registered Republicans and the incidence of "no choice" (i.e., one or no candidate) ballots. While the seventeen districts with the fewest Republicans (the bottom half of the districts) included seven single-candidate districts, the seventeen districts with the most Republicans (the top half) included four districts with one candidate and three with none.20 The top and bottom thirds each contain five such districts. Thus, one-candidate and no-candidate districts appear to be randomly distributed. Even the most heavily Republican district had only Bush delegates on the ballot.

There does appear to be a correlation between (a) the districts with more than two candidates on the ballot and (b) the number of registered Republicans in those districts; none of the fifteen least Republican districts had more than two candidates on the ballot. But we think that the difference between three slates and two slates is decidedly less significant than the difference between two slates and one slate. The contention that the fundamental right to vote is infringed in single-candidate districts (where there is arguably no choice) is a very different contention (and a much stronger one) than the contention that the fundamental right to vote is infringed in districts with two candidates on the ballot. Thus, the plaintiffs ultimately fail to prove this point by failing to show a correlation between single-candidate districts and the effective percentage signature requirement in those districts.

The plaintiffs' equal protection argument also fails because it focuses exclusively on the choice of presidential candidates, and focuses not at all on the delegate slates. Even if it is true that Republican voters in a district with few Republicans are less likely to have a choice on their primary ballot than voters in a heavily Republican district, the plaintiffs still have not established that there is a right to equal degrees of choice in voting for different people in different districts. The plaintiffs would quarrel with this characterization of the March 7 election as a vote for different people. They claim that the delegate selection process amounts to a unitary statewide election in which the primaries in all thirty-one congressional districts will be held on the same day, will designate delegates to the same national convention, and will be seen widely as a unitary state presidential primary. However, there is no getting around the fact that--notwithstanding the unitary impact that the overall vote will have--there will be thirty-one elections on March 7. Each district will send its own slate of three delegates to the Republican National Convention. No two districts will send the same person.

Not only will there be thirty-one different elections, but the elections will produce ninety-three different delegates rather than one presidential candidate. Theoretically, these ninety-three delegates could be pledged to as many as thirty-one different presidential candidates; or the delegates might be split, forty-eight for one candidate and forty-five for another. Some districts may elect delegates who are uncommitted. Delegates pledged to a particular candidate who thereafter drops out of the presidential race may arrive at the convention uncommitted, or committed (or brokered) to someone else.

Although popular attention may well focus on the number of delegates pledged to each candidate at the convention, the delegates themselves will also cast votes on platform issues and issues of party governance. No doubt, the chief purpose of many voters will be to send a message on presidential candidates. But that does not mean that we must treat these thirty-one elections as if they were a straw poll.21 In short, registered Republicans in each district will be electing a slate of three people who are pledged to vote for a particular candidate, who may be freed to vote for anyone, and who will vote at the convention on other issues as well.

The other flaw with the plaintiffs' argument is that the disparity in percentage requirements--which, they contend, unconstitutionally burdens their fundamental right to vote--provides them with a substantial benefit. The same statistics used by the plaintiffs to discount the value of a Republican vote in the Eleventh District can plausibly demonstrate the enhanced value and power of each such vote. If every registered Republican in the Eleventh District voted in a primary, each voter would have 1/11,814th of a say in deciding who the convention delegates would be. In the Third District, however, each voter has only 1/158,097th of a say in making the same decision. In a congressional election--for which state law allows use of the same formula--this advantage enjoyed by Republicans in the Eleventh District is offset by the fact that a Republican nominee in the heavily Democratic Eleventh is much less likely to arrive in Congress than a Republican nominee in the Third. In electing convention delegates, however, the Eleventh District's advantage remains undiluted. Thus the 11,814 registered Republicans in the Eleventh District (constituting 0.4 percent of the total number of registered Republicans in the State) send the same number of delegates to the national convention as the 158,097 registered Republicans in the Third (constituting 5.8 percent of statewide registered Republicans). Using this dynamic, the vote of the Eleventh District Republicans in Brooklyn is fourteen times more powerful than the vote of their fellow Republicans in Long Island's Third District.22 In percentage terms, it is perhaps more work for activists in the Eleventh to put a slate on the ballot, but the number of signatures required is fewer (591) and the influence of each vote is greater. This situation may or may not furnish enough incentive to produce multiple candidacies, but it cannot be fairly characterized as a foreclosure of choice, a sham, or a mockery of the election process. If anything, the plaintiffs' ability to have a say in who the Republican Party nominates for President is very well protected.

This case is therefore markedly different from Socialist Workers Party, relied on by the district court to support its determination that strict scrutiny is the appropriate test here. Socialist Workers Party concerned the right of voters to cast their votes in a general election for a wide variety of parties and candidates, other than those nominated by the two major parties. Without these choices, socialist workers could not "cast their votes effectively." 440 U.S. at 184, 99 S. Ct. at 990. See Norman v. Reed, 502 U.S. 279, 289, 112 S. Ct. 698, 116 L. Ed. 2d 711 (1992) (citing Socialist Workers Party for the proposition that "severe" burdens on the right to form new political parties merits strict scrutiny); Clements v. Fashing, 457 U.S. 957, 964-65, 102 S. Ct. 2836, 2844-45, 73 L. Ed. 2d 508 (1982) (plurality opinion) (Socialist Workers Party applies only to ballot access classifications involving small political parties or independent candidates); Unity Party v. Wallace, 707 F.2d 59, 63 (2d Cir.1983) (citing Socialist Workers Party for the proposition that classifications burdening independent candidates or small political parties are subjected to strict scrutiny). As discussed above, however, the plaintiffs are not deprived of their ability to "cast their votes effectively." Republican voters on March 7 will select delegates who may or may not vote at a subsequent party convention for the presidential candidate listed next to the delegate's name on the ballot, or who may be uncommitted from the start. Voters from districts with relatively few Republicans may perhaps enjoy less choice among delegates than their fellow party members in more heavily Republican districts, but they gain an amplified voice by the equal impact of their small numbers. Finally, New York's signature requirement requires fewer signatures in districts with fewer registered Republicans, in contradistinction to Socialist Workers Party, where more signatures were required in a place with fewer voters (a scheme that would apparently have failed even the rational basis test, under the Court's analysis).23 See note 15, supra.

At most, the plaintiffs can show that the signature requirement makes it less likely that they will have a broad choice in a primary among potentially uncommitted delegates to a national party convention. The case most analogous (in fact almost identical) to this one is Hewes v. Abrams, 884 F.2d 74 (2d Cir.1989), in which we examined a signature requirement similar to the one at issue here. That case reviewed a different provision of New York Election Law Sec. 6-136, which required candidates for Mayor of New York City to collect signatures from the lesser of five percent or 10,000 of the registered voters in the candidate's party in order to qualify for the ballot. The plaintiff was the candidate of the Right to Life Party. Because his party had only 6,000 registered voters, he needed signatures from five percent of them to get on the ballot. Candidates of the Democratic and Republican Parties, each party having more than 200,000 enrolled voters, could avail themselves of the 10,000-signature numerical cap. This was 0.5 percent of registered Democrats, and 2.4 percent of registered Republicans. Id. at 75; Hewes v. Abrams, 718 F. Supp. 163, 164 (S.D.N.Y.1989). Like the plaintiffs here, the plaintiff in Hewes claimed that this scheme violated his equal protection rights. After summarizing the facts and the parties' arguments, we said: "We affirm substantially for the reasons stated by [District Court] Judge Haight in his thorough opinion." 884 F.2d at 75. Judge Haight's reasoning is therefore the law of this Circuit. See United States v. Chila, 871 F.2d 1015, 1018 (11th Cir.), cert. denied, 493 U.S. 975, 110 S. Ct. 498, 107 L. Ed. 2d 501 (1989). Judge Haight held that rational basis review applied to the State's signature requirement: "Where, as here, the plaintiff has not shown invidious discrimination based upon wealth or race classifications or a preclusion of the right to vote, heightened scrutiny is not required." 718 F. Supp. at 166 (internal quotations omitted). The court upheld the requirement, finding that the State's "important" interest in requiring that candidates have "a modicum of support" before they get on the ballot supplied a rational basis for the rule. Id. at 166 (citing Jenness v. Fortson, 403 U.S. 431, 442, 91 S. Ct. 1970, 1976, 29 L. Ed. 2d 554 (1971)).

Here, the plaintiffs' right to vote is certainly not "preclu[ded]." Nor do we believe, after reviewing the plaintiffs' evidence and the nature of this election, that the disparity in the percentage signature requirements significantly burdens their right to vote. We therefore hold that strict scrutiny does not apply.

We apply a rational basis analysis. See McDonald v. Board of Election Comm'rs., 394 U.S. 802, 806-07, 89 S. Ct. 1404, 1407-08, 22 L. Ed. 2d 739 (1969); Schulz v. Williams, 44 F.3d 48, 58 (2d Cir.1994); Unity Party, 707 F.2d at 63. The plaintiffs point to a disparity between districts with many Republicans and districts with few Republicans. The State must advance a justification for the five percent or 1250 rule that is rationally related to a legitimate state interest. New York has determined that it is permissible to choose delegates to national conventions from congressional districts. It has also determined, consistent with the Supreme Court's holding in Jenness, 403 U.S. at 442, 91 S.Ct. at 1976, that requiring candidates to submit petitions signed by five percent of the district's enrolled Republicans is consistent with the goal of weeding out candidates who do not have "a modicum of support." The State has further determined that the five percent rule would impose an unnecessary burden in districts with many Republicans (like the Third), where an unmitigated five percent requirement would require 7905 signatures on a delegate's petition. New York has concluded that a petition signed by 1250 Republicans shows a "modicum of support." This effort to balance two goals--restricting ballot access to candidates who show some percentage of support, and providing a safety valve when the across-the-board percentage formula forces candidates in certain districts to collect more signatures than a number sufficient to show a modicum of support--is rational. See Socialist Workers Party, 440 U.S. at 187, 99 S.Ct. at 991 (striking down statute that imposed five percent signature requirement on all political subdivisions, and effectively converting statute to a five percent or 25,000 signature requirement, whichever is less).

Under the district court's logic, a numerical cap would, in virtually all cases, be an unconstitutional approach for a state to take in balancing the desire for liberal ballot access with the need to ensure that candidates achieve some modicum of support. As the plaintiffs point out in their cross-appeal, the district court's 1.41 percent solution creates its own disparity, the "deviation of two to one between the maximum percentage [1.41 percent] and the minimum percentage [0.79 percent]." The plaintiffs argue that this disparity is unjustifiable and that the only way to end it is to enforce a 0.79 percent or 1250 formula: that is, to adopt as the percentage requirement the lowest percentage of voters able to field a slate of delegates in any district under the statutory formula. In that district, the Third, the 1250 signatures required equals 0.79 percent of registered Republicans. Of course, this would eliminate the disparity by adopting what is in effect an across-the-board flat percentage, because the 0.79 percentage will require fewer than 1250 signatures in every district but the Third, where it will (by design) require exactly 1250. Under this logic, a state could never alleviate a perceived hardship created by the percentage-based signature requirement because any numerical cap--if it is efficacious--would create a disparity requiring the "correction" urged by the plaintiffs. But by remedying an unconstitutional ballot access rule through the imposition of its own numerical cap in Socialist Workers Party, the Supreme Court has taken precisely this approach and, in effect, held that it is a rational one. See id. at 187, 99 S. Ct. at 991.

There is another rational justification for the ballot access rule that is quite strong. As we have discussed, each vote in a district with relatively few Republicans is given much more weight that an individual vote in heavily Republican districts. Thus, a state may rationally require candidates in those districts to get support from a higher proportion of voters, and may rationally expect that the opportunity to elect the same number of delegates with many fewer votes will furnish an incentive for candidates to circulate petitions. In this way, a state counterbalances the disproportionate weight that each Republican's vote has in a less populous district. Still another rational justification might be based on a state's conclusion that 1250 signatures reflect a greater level of support in more rural districts, where people are widely dispersed and where the actual collection process may be more arduous. We do not need to decide whether each of these government interests by itself would support the signature requirement. Together, they certainly constitute sufficient, legitimate governmental interests to which New York's five percent or 1250 formula is rationally related.

We do not express any opinion as to the wisdom of the five percent requirement. That decision is for the state legislature and the Republican Party, not the federal courts.

The plaintiffs have failed to prove either a likelihood of success on the merits or sufficiently serious questions going to the merits. Therefore, the judgment of the district court is reversed, and the order is vacated.