The State of Oregon appeals a district court order declaring Or.Rev.Stat. § 280.070(4)(a) unconstitutional and permanently limiting its enforcement. Section 280.070(4)(a) requires that ballots for initiatives proposing local option taxes include a statement: “This measure may cause property taxes to increase more than three percent.” The district court deemed this requirement constitutionally infirm, concluding that inclusion of the “three-percent warning” violated appellee Michael Caruso’s First Amendment rights as a petition circulator and his due process rights as a voter. We conclude that the requirement does not violate the U.S. Constitution, reverse the decision of the district court, and vacate the injunction limiting enforcement of section 280.070(4)(a).

I. BACKGROUND

The Oregon Constitution reserves to the people “the initiative power, which is to propose laws and amendments to the Constitution and enact or reject them at an election independently of the Legislative Assembly.” Or. Const. art. IV, § 1. Under Oregon law, the ballot titles for initiatives that propose the imposition of a local option tax must include an additional statement commonly referred to as the “three-percent warning”: “This measure may cause property taxes to increase more than three percent.” Or.Rev.Stat. § 280.070(4)(a).

Appellee Michael Caruso was among the chief petitioners for an initiative measure which appeared on the ballot for the March 9, 2004, Yamhill County Special Election. Consistent with section 280.070(4)(a), the Yamhill County Commission adopted, on November 26, 2003, the following ballot title for the measure:

See also Or.Rev.Stat. § 250.035(1) (providing that the ballot title of any non-state initiative shall consist of a caption, question, and summary).

Caruso challenged the constitutionality of section 280.070(4)(a) before the district court, asserting that the required inclusion of the three-percent warning violated his First Amendment rights as a petition cir-culator and his due process rights as a voter. Caruso requested that the district court declare the provision unconstitutional, both as applied to Measure 36-55 and on its face, and enter preliminary and permanent injunctions against its enforcement.

With the parties’ consent, the district court consolidated the hearing on Caruso’s motion for a preliminary injunction with the trial on the merits. In an opinion issued the day after the hearing, the district court held section 280.070(4)(a) unconstitutional as applied and as enacted. Adopting the reasoning set forth in a companion decision, Horton v. Multnomah County, No. Civ. 03-1257-HA, 2004 WL 1745789 (D.Or. Aug.4, 2004), the district court deemed the three-percent warning “false and misleading” because it implied that the initiative measure “by itself’ “may cause property taxes to increase more than three percent” when the increase proposed by Measure 36-55 was in fact much lower: only “$0,003 per $1,000 of assessed valuation.”

The district court reasoned that section 280.070(4)(a) was therefore constitutionally infirm. First, it impeded Caruso’s ability to communicate the actual tax consequences of Measure 36-55 and forced him to be associated with the State’s misleading message. In light of these burdens on “core political speech,” the district court determined that section 280.070(4)(a) was subject to, and did not survive, strict scrutiny under the First Amendment. Second, section 280.070(4)(a) substantially chilled protected speech. Specifically, because the three-percent warning applied to all initiatives proposing local option taxes— including those which posed no threat of themselves increasing property taxes more than three percent — it discouraged others from circulating such initiatives by erecting the additional “hurdle of convincing voters of the false nature of the state mandated ‘warning.’ ” Finally, the three-percent warning obscured the actual subject of Measure 36-55, upsetting the even-handedness of the election and working a fundamental unfairness on the voters.

To remedy these constitutional infirmities, the district court enjoined the government defendants from enforcing section 280.070(4)(a) in relation to Measure 36-55, and, more broadly, in relation to “all ballot measures that by themselves cannot cause an increase in property taxes of more than three percent.” The State of Oregon, which had intervened to oppose Caruso’s claims, timely appealed.

II. DISCUSSION

A. Mootness

Pursuant to the district court injunction, Measure 36-55 appeared on the ballot for the March 9, 2004, Yamhill County Special Election without the three-percent warning. The measure failed, with 3,250 voters favoring the levy and 9,153 opposing it. The State argues that the election rendered moot Caruso’s claim that section 280.070(4)(a) is unconstitutional as applied to the defeated measure.

As a general rule, a case is moot “ ‘when the issues presented are no longer ‘live’ or the parties lack a legally cognizable interest in the outcome.’ ” Schaefer v. Townsend, 215 F.3d 1031, 1033 (9th Cir.2000) (quoting Powell v. McCormack, 395 U.S. 486, 496 , 89 S.Ct. 1944 , 23 L.Ed.2d 491 (1969)). There is an exception, however, for challenged practices that are “ ‘capable of repetition, yet evading review.’ ” Id. (quoting Dunn v. Blumstein, 405 U.S. 330 , 333 n. 2, 92 S.Ct. 995 , 31 L.Ed.2d 274 (1972)). Under this exception, a court may consider the merits of a case that would otherwise be deemed moot when “(1) the challenged action is in its duration too short to be fully litigated prior to cessation or expiration, and (2) there is a reasonable expectation that the same complaining party will be subject to the same action again.” Spencer v. Kemna, 523 U.S. 1, 17 , 118 S.Ct. 978 , 140 L.Ed.2d 43 (1998) (alterations and citation omitted).

Cases challenging election laws often fall within the “capable of repetition, yet evading review” exception “because the inherently brief duration of an election is almost invariably too short to enable full litigation on the merits.” Porter v. Jones, 319 F.3d 483, 490 (9th Cir.2003). For this reason, “[i]f such cases were rendered moot by the occurrence of an election, many constitutionally suspect election laws ... could never reach appellate review.” Joyner v. Mofford, 706 F.2d 1523, 1527 (9th Cir.1983). Here, the State cannot seriously contest that the period between adoption of the ballot title and election on the measure — in this case, four months — is too brief to permit “full litigation on the merits.” See Meyer v. Grant, 486 U.S. 414 , 417 n. 2, 108 S.Ct. 1886 , 100 L.Ed.2d 425 (1988) (finding the “evading review” prong satisfied where the state “grant[ed] the proponents of an initiative only six months in which to obtain the necessary signatures”).

The State more forcefully contests the expectation that Caruso “will be subject to the same action again.” It observes that there is no evidence in the record that Caruso has circulated or intends to circulate another initiative proposing the imposition of a local option tax. Although this is true, we have rejected the analogous argument that a candidate’s challenge to an election law is not moot “only when [the] candidate plans to seek reelection.” Schaefer, 215 F.3d at 1033 . In Schaefer , we observed that the Supreme Court had proceeded to the merits of a voter’s challenge to the residency requirements for voting even though by the time of the Court’s decision, the voter had resided in the state long enough to vote in the next election. Id. at 1033 (citing Blumstein, 405 U.S. at 331 -33 & n. 2, 92 S.Ct. 995 ). We accordingly examined the merits of a prospective candidate’s challenge to the residency requirements for nomination even though by the time of our decision, the contested seat had been filled and the candidate “refuse[d] to disclose his intentions” for running for state office in the future. Id.; accord Merle v. United States, 351 F.3d 92, 95 (3d Cir.2003) (disagreeing with the suggestion that a candidate must allege his intent to run in a future election to satisfy the “capable of repetition” requirement).

Consistent with Schaefer , we conclude that Caruso’s claims are not moot. Although Caruso has not expressed an intention to circulate similar initiatives in the future, enforcement of section 280.070(4)(a) is nevertheless “capable of repetition, yet evading review.” Accordingly, we proceed to the merits.

B. Constitutional and Statutory Context

We begin by setting forth the context in which we must examine section 280.070(4)(a). Before 1997, property in Oregon was assessed and taxed at its real market value. See Or.Rev.Stat. § 308.232 (1995). Then, in 1997, Oregon voters adopted Measure 50, which amended the Oregon Constitution to limit the property taxes imposed by Oregon counties. See Flavorland Foods v. Wash. County Assessor, 334 Or. 562 , 54 P.3d 582, 583-84 (2002) (en banc). First, Measure 50 “rolled back” a property’s maximum assessed value for the tax year beginning July 1, 1997, to its “real market value for the tax year beginning July 1, 1995, reduced by 10 percent.” Or. Const, art. XI, § ll(l)(a). Second, “[f]or tax years beginning after July 1, 1997,” Measure 50 provided that a “property’s maximum assessed value shall not increase by more than three percent from the previous tax year.” Id. § ll(l)(b). Third, Measure 50 established a “permanent limit on the rate of ad valorem property taxes imposed” by each local taxing district. Id. § ll(3)(b). This limit is commonly referred to as the “permanent rate.” See Shilo Inn v. Multnomah County, 333 Or. 101 , 36 P.3d 954, 958 (2001), modified by 334 Or. 11 , 45 P.3d 107 (2002).

The limits imposed by Measure 50, however, were subject to an exception for taxes submitted to and approved by the voters:

Or. Const, art. XI, § ll(4)(a)(A). The conditions for submitting local option taxes to the voters are provided for by Oregon statutes. See Or.Rev.Stat. §§ 280.070-.075. These statutes require that the ballot titles for such measures include, in addition to the three-percent warning, “the period during which the proposed local option tax will be imposed,” “[t]he first fiscal year in which the proposed local option tax will be imposed,” and “the total amount of money to be raised by the proposed local option tax, in dollars and cents.” Id. §§ 280.070(4)(a), 280.070(5)(a)-(b), 280.075(1). Against this constitutional and statutory backdrop, we consider Caruso’s First and Fourteenth Amendment challenges.

C. First Amendment

Caruso contends that section 280.070(4)(a) infringed upon his complementary First Amendment rights “to speak freely and ... to refrain from speaking at all.” Wooley v. Maynard, 430 U.S. 705, 714 , 97 S.Ct. 1428 , 51 L.Ed.2d 752 (1977). Specifically, he contends that the three-percent warning restricted his ability to communicate accurately the tax consequences of Measure 36-55 and forced him instead to be associated with the State’s misleading communication. The district court agreed. Where, as here, a district court sustains a constitutional challenge to a restriction on speech, its findings of fact are reviewed for clear error. Lovell ex. rel. Lovell v. Poway Unified Sch. Dist., 90 F.3d 367 , 370 (9th Cir.1996). Its application of law to those facts and its ultimate determination that the restriction is unconstitutional, however, are reviewed de novo. Mont. Right to Life Ass’n v. Eddleman, 343 F.3d 1085, 1090 (9th Cir.2003).

1. Level of Scrutiny

We turn at the outset to the appropriate level of First Amendment scrutiny, pausing first to reject the State’s suggestion that no scrutiny is warranted because the speech regulated by section 280.070(4)(a) is the government’s. We have elsewhere identified “several recognized instances of constitutional limitations on government speech.” R.J. Reynolds Tobacco Co. v. Shewry, 384 F.3d 1126, 1140 (9th Cir.2004). For example, the First Amendment may limit government speech that “make[s] private speech difficult or impossible,” id., or that attributes a government message to a private speaker, see Johanns v. Livestock Mktg. Ass’n, — U.S. -, - & nn. 7-8, 125 S.Ct. 2055 , 2064-65 & nn. 7-8, 161 L.Ed.2d 896 (2005). Caruso argues that section 280.070(4)(a) inflicted both types of harm.

a. Election Laws

An election law that burdens First Amendment rights is not necessarily subject to strict scrutiny, however. See Clingman v. Beaver, — U.S. -, -, 125 S.Ct. 2029, 2038 , 161 L.Ed.2d 920 (2005). Rather, such laws are generally subject to a balancing standard, under which a reviewing court weighs the “character and magnitude” of the burden imposed against the interests advanced to justify that burden. See e.g., Timmons v. Twin Cities Area New Party, 520 U.S. 351, 358 , 117 S.Ct. 1364 , 137 L.Ed.2d 589 (1997); Burdick v. Takushi, 504 U.S. 428, 434 , 112 S.Ct. 2059 , 119 L.Ed.2d 245 (1992); Anderson v. Celebrezze, 460 U.S. 780, 789 , 103 S.Ct. 1564 , 75 L.Ed.2d 547 (1983). As the Supreme Court explained in Burdick , it has typically applied this “more flexible” standard to election laws because “[cjommon sense, as well as constitutional law, compels the conclusion that government must play an active role in structuring elections.” 504 U.S. at 433 , 112 S.Ct. 2059 ; accord Clingman, 125 S.Ct. at 2039 (explaining that subjecting every electoral regulation to strict scrutiny would “hamper the ability of States to run efficient and equitable elections”).

To be sure, the Supreme Court has in several other cases subjected election laws to strict scrutiny rather than flexible balancing. See, e.g., McIntyre v. Ohio Elections Comm’n, 514 U.S. 334, 345-46 , 115 S.Ct. 1511 , 131 L.Ed.2d 426 (1995); Meyer v. Grant, 486 U.S. 414, 420 , 108 S.Ct. 1886 , 100 L.Ed.2d 425 (1988). Unlike the cases described above, these cases did not involve regulations of the “voting process”; they instead involved regulations of — or, more precisely, limitations on — “pure speech.” See McIntyre, 514 U.S. at 345 , 115 S.Ct. 1511 ; accord Campbell v. Buckley, 203 F.3d 738, 745 (10th Cir.2000) (“[Sjtrict scrutiny is applied where the government restricts the overall quantum of speech available to the election or voting process.”). Although these cases “do not lend themselves to a simple synthesis,” Campbell, 203 F.3d at 745 , we regard them as distinguishable in two respects.

First, unlike the provisions challenged in McIntyre and Meyer (which respectively prohibited the distribution of anonymous campaign literature, McIntyre, 514 U.S. at 336 , 115 S.Ct. 1511 , and the payment of petition circulators, Meyer, 486 U.S. at 416 , 108 S.Ct. 1886 ), section 280.070(4)(a) governs the political process more than it does political speech. See McIntyre, 514 U.S. at 344 , 115 S.Ct. 1511 . Significantly, the provision regulates only what is said through the ballot itself. See Timmons, 520 U.S. at 362-63 , 117 S.Ct. 1364 (applying flexible balancing where the challenged law burdened a political party’s ability to communicate through the ballot only); Rubin v. City of Santa Monica, 308 F.3d 1008, 1015 (9th Cir.2002) (similar). All other means of communication, including the “interactive” discussions for which First Amendment protection is “at its zenith,” Meyer, 486 U.S. at 422, 425 , 108 S.Ct. 1886 , fall outside the provision’s regulatory scope.

Caruso responds that the ballot is not merely a means of communication; it is the most direct and credible means of communication. This argument has some force. Indeed, the Supreme Court has characterized “ ‘the instant before the vote is cast’ ” as “ ‘the most crucial stage in the election process.’ ” Cook v. Gralike, 531 U.S. 510, 525 , 121 S.Ct. 1029 , 149 L.Ed.2d 44 (2001) (quoting Anderson v. Martin, 375 U.S. 399, 402 , 84 S.Ct. 454 , 11 L.Ed.2d 430 (1964)). But the fact that the ballot is “crucial” to an election does not imply that Caruso therefore has a First Amendment right to communicate a specific message through it. To the contrary, the Supreme Court in Timmons rejected the “contention that[a political party] has a right to use the ballot itself to send a particularized message”: “Ballots serve primarily to elect candidates, not as forums for political expression.” 520 U.S. at 362-63 , 117 S.Ct. 1364 ; accord Rubin, 308 F.3d at 1016 (“A ballot is a ballot, not a bumper sticker.”).

Caruso argues that cases such as Tim-mons are inapposite because they involved candidates rather than initiatives. We disagree. In Buckley v. American Constitutional Law Foundation, Inc., 525 U.S. 182 , 119 S.Ct. 636 , 142 L.Ed.2d 599 (1999), a case involving ballot initiatives, the Supreme Court expressly affirmed the Tenth Circuit’s reliance on Timmons. See id. at 190, 119 S.Ct. 636 . The Supreme Court further affirmed that “States allowing ballot initiatives have considerable leeway to protect the integrity and reliability of the initiative process, as they have with respect to election processes generally.” Id. at 191 , 119 S.Ct. 636 . It is true that the Supreme Court has elsewhere highlighted the differences between initiatives and candidates, but it appears to have done so not because the First Amendment is more protective of initiatives than of candidates, but because elections on initiatives pose lesser risks of quid pro quo corruption. See, e.g., First Nat’l Bank v. Bellotti, 435 U.S. 765, 790 , 98 S.Ct. 1407 , 55 L.Ed.2d 707 (1978) (“The risk of corruption perceived in cases involving candidate elections simply is not present in a popular vote on a public issue.” (citations omitted)).

Caruso also observes that notwithstanding Timmons , Chief Justice Rehnquist in Cook applied strict scrutiny to an amendment governing the composition of Missouri ballots. See Cook, 531 U.S. at 530-32 , 121 S.Ct. 1029 (Rehnquist, C.J., concurring in the judgment). Specifically, the amendment required that ballots label candidates for congressional office who “DISREGARDED VOTERS’ INSTRUCTIONS ON TERMS LIMITS” or “DECLINED TO PLEDGE TO SUPPORT TERM LIMITS.” Id. at 514-15 , 121 S.Ct. 1029 . Chief Justice Rehnquist described the requirement as constitutionally suspect in part because it was “not neutral as to issues or candidates.” Id. at 532 , 121 S.Ct. 1029 . To the contrary, only one issue was selected for comment and “only those candidates who fail[ed] to conform to the State’s position reeeive[d] derogatory labels.” Id.

Caruso maintains that here, just as Chief Rehnquist described in Cook , the State selected a single consideration for comment on the ballot. As an initial matter, we note that Chief Justice Rehnquist’s opinion in Cook was a concurrence, joined only by Justice O’Connor. Id. at 530 , 121 S.Ct. 1029 . In contrast to Chief Justice Rehnquist’s concurring opinion, the majority opinion reviewed Missouri’s labeling requirement under the Tenth Amendment and the Elections Clause rather than under the First Amendment. Id. at 518-27 , 121 S.Ct. 1029 . In fact, the majority specifically rejected the respondents’ reliance on “First Amendment cases upholding ‘time, place, and manner’ regulations of speech,” explaining that “[although the Elections Clause uses the same phrase as that branch of our First Amendment jurisprudence, it by no means follows that such cases have any relevance to our disposition of this case.” Id. at 527 n. 20, 121 S.Ct. 1029 .

But even apart from its precedential value, we regard Chief Justice Rehnquist’s concurring opinion as distinguishable. Unlike Missouri’s labeling requirement, which applied to “only those candidates who fail[ed] to conform to the State’s position,” id. at 532 , 121 S.Ct. 1029 , Oregon’s three-percent warning applies to all “measure[s] authorizing the imposition of local option taxes,” Or.Rev.Stat. § 280.070(4)(a). Accordingly, no measure or group of measures is “singled out.” Compare Cook, 531 U.S. at 532 , 121 S.Ct. 1029 . And unlike congressional candidates, who “may debate tax reform, Social Security, national security, and a host of other issues,” id., initiative measures, by law, may “embrace one subject only,” Or. Const. art. IV, § (2)(d). For Measure 36-55, that subject was a one-time increase in property taxes. The measure’s possible effect on property taxes, then, was not simply one facet of many. Finally, unlike the State of Missouri, which “ha[d] chosen one and only one issue to comment on,” id., the State of Oregon has chosen to comment not only on a measure’s possible effect on property taxes, but also on the rate of the special levy, the voter turnout required to pass it, the proposed uses for the funds, the total amount of funds anticipated, and the estimated additional cost to property owners.

Second, and also unlike the provisions challenged in McIntyre and Meyer , section 280.070(4)(a) does not have “the inevitable effect of reducing the total quantum of speech on a public issue.” Meyer, 486 U.S. at 423 , 108 S.Ct. 1886 ; accord McIntyre, 514 U.S. at 346 , 115 S.Ct. 1511 . In Meyer , the Supreme Court determined that Colorado’s ban on the use of the paid petition circulators “limitfed] the number of voices who will convey appellees’ message and the hours they can speak,” thus restricting their ability to reach a larger audience, “garner the number of signatures necessary to place the matter on the ballot,” and “make the matter the focus of statewide discussion.” 486 U.S. at 422-23 , 108 S.Ct. 1886 . The Court quoted the observation made by the Colorado Supreme Court in another context: “ ‘[T]he solicitation of signatures on petitions is work. It is time-consuming and it is tiresome — so much so that it seems that few but the young have the strength, the ardor and the stamina to engage in it, unless, of course, there is some remuneration.’ ” Id. at 423-24 , 108 S.Ct. 1886 (quoting Urevich v. Woodard, 667 P.2d 760, 763 (Colo.1983)).

In contrast, the effect of the three-percent warning on the “total quantum of speech” is uncertain. In a declaration submitted to the district court, Caruso pre- dieted that continued enforcement of the three-percent-warning requirement would discourage petition circulators “as they w[ould] realize that all of their work can and will be nullified by the false statement that will appear on the ballot.” But Caruso’s prediction — unsupported by any evidence in the record, compare Am. Constitutional Law Found., 525 U.S. at 198, 119 S.Ct. 636 (describing the testimony of several petition circulators who recounted the harassment and fear that they and others had experienced as a result of the requirement that they wear identification badges) — is not the only plausible one. In fact, the requirement “might encourage more speech,” Campbell, 203 F.3d at 745 , with petition circulators working not only to collect signatures, but also to inform voters that their measures would not, by themselves, increase property taxes more than three percent. Along these lines, we note that the three-percent warning required by section 280.070(4)(a) is “separated from the moment the circulator speaks,” leaving circulators free to contest its accuracy in their conversations with electors. Compare Am. Constitutional Law Found., 525 U.S. at 198-200, 119 S.Ct. 636 (explaining that Colorado’s affidavit requirement, unlike its identification badge requirement, “exemplifies the type of regulation for which McIntyre left room” because it “must be met only after circulators have completed their conversations with electors”).

b. Compelled Speech

Caruso contends that section 280.070(4)(a) should be subject to strict scrutiny for the additional reason that it forced him to be associated with the State’s message. See Pac. Gas & Elec. Co. v. Pub. Utils. Comm’n, 475 U.S. 1, 32-33 , 106 S.Ct. 903 , 89 L.Ed.2d 1 (1986); Wooley v. Maynard, 430 U.S. 705, 714 , 97 S.Ct. 1428 , 51 L.Ed.2d 752 (1977). Yet unlike the requirements challenged in Pacific Gas & Electric and Wooley , section 280.070(4)(a) does not require that owners use their private property to transmit the State’s message. See Pac. Gas & Elec. Co., 475 U.S. at 17-18 , 106 S.Ct. 903 (invalidating an order requiring a utility company “to use its property — the billing envelopes — to distribute the message of another”); Wooley, 430 U.S. at 714 , 97 S.Ct. 1428 (invalidating a New Hampshire law requiring drivers to use their cars as a “ 'mobile billboard’ for the State’s ideological message”). The provision instead provides for the State’s message to be transmitted through ballots, documents prepared, printed, and distributed by— and therefore attributed to — State and local governments. See Timmons, 520 U.S. at 356 , 117 S.Ct. 1364 ; see also Johanns, 125 S.Ct. at 2065 , 125 S.Ct. 2055 (“[T]his second [compelled speech] theory might ... form the basis for an as-applied challenge — if it were established, that is, that individual beef advertisements were attributed to respondents.”).

Nor does section 280.070(4)(a) require the “affirmation of a belief.” Bd. of Educ. v. Barnette, 319 U.S. 624, 633 , 63 S.Ct. 1178 , 87 L.Ed. 1628 (1943). Oregon law requires that the ballot title for a county initiative be prepared by the district attorney, not by the petitioner. See Or.Rev. Stat. § 250.175(3). Indeed, in his declaration, Caruso opined that voters view the three-percent warning as “an official statement of government.” No part of section 280.070(4)(a) required Caruso to affirm the government’s statement; rather, he remained free to publicly dissociate himself from it. See PruneYard Shopping Ctr. v. Robins, 447 U.S. 74, 88 , 100 S.Ct. 2035 , 64 L.Ed.2d 741 (1980).

We thus find the present appeal distinguishable from both the election law cases and the compelled speech cases in which the Supreme Court has applied strict scrutiny. We accordingly apply the Supreme Court’s more flexible balancing standard.

2. Balancing Standard

Under this standard, “the rigorousness of [a reviewing court’s] inquiry into the propriety of a state election law depends upon the extent to which a challenged regulation burdens First and Fourteenth Amendment rights.” Burdick, 504 U.S. at 434 , 112 S.Ct. 2059 . “[W]hen those rights are subjected to ‘severe’ restrictions, the regulation must be ‘narrowly drawn to advance a state interest of compelling importance.’ ” Id. (quoting Norman v. Reed, 502 U.S. 279, 289 , 112 S.Ct. 698 , 116 L.Ed.2d 711 (1992)). But when those rights are subjected only to “ ‘reasonable, nondiscriminatory’ ” restrictions, “ ‘the State’s important regulatory interests are generally sufficient.’ ” Id. (quoting Anderson, 460 U.S. at 788 , 103 S.Ct. 1564 ).

a. Burden

Caruso argues that section 280.070(4)(a) imposed a “severe” burden on his communication with voters by requiring that the ballot title for Measure 36-55 include the “false and misleading” warning that “[t]his measure may cause property taxes to increase more than three percent.” This warning, Caruso maintains, implied that Measure 36-55 by itself might cause property taxes to increase more than three percent when the measure in fact proposed a much lower increase of 0.0003 percent. Because the material facts are not in dispute, “the character and extent of the statute’s burden involves a question of law,” which we review de novo. Krislov v. Rednour, 226 F.3d 851 , 859 (7th Cir.2000) (citing (WIN) Wash. Initiatives Now v. Rippie, 213 F.3d 1132 , 1137 (9th Cir.2000)).

Mindful of the constitutional backdrop against which section 280.070(4)(a) was enacted, we reject Caruso’s assumption that his is the only reasonable interpretation of the three-percent warning. As discussed above, the Oregon Constitution provides that a “property’s maximum assessed value shall not increase by more than three percent from the previous tax year.” Or. Const, art. XI, § ll(l)(b). But this limitation may be exceeded by local option taxes submitted to and approved by voters in the taxing district. Id. § ll(4)(a)(A). When construed in context, then, the three-percent warning could be read as an accurate reminder that the local option tax proposed by Measure 36-55 would not be subject to the limitation imposed by the Oregon Constitution. In other words, absent approval of Measure 36-55, maximum assessed values and, in turn, property taxes could increase by no more than three percent. Id. § ll(l)(b). If the measure were approved, however, property taxes could increase by more than that amount. Id. § ll(4)(a)(A).

Caruso offers several reasons to reject this reading: (a) voters are unlikely to know of the constitutional three-percent limit; (b) the constitutional three-percent limit applies to maximum assessed values, not property taxes; (c) there is only an “infinitesimal possibility” that other property taxes will increase three percent; and (d) even if other property taxes increased three percent, it would not be Measure 36-55 that would “cause” total taxes to increase more than that amount. We address these reasons in turn.

As to voters’ knowledge, we note that Measure 50, which imposed the three-percent limit, was adopted by referendum. Flavorland Foods, 54 P.3d at 583-84 . Before the election on Measure 50, voters would have been presented with summaries of its impact on the ballot and in the voters’ pamphlet. Or.Rev.Stat. §§ 250.035(2)(d), 251.185(1). It is also likely that the ballot issue was the subject of discussion and press coverage leading up to election day. We further note that Measure 50 replaced Measure 47, a “short-lived constitutional amendment” which had been enacted by the voters by ballot initiative the year before the approval of Measure 50, and which included substantially the same property tax limitations. See Flavorland Foods, 54 P.3d at 583 (explaining that “ ‘[c]ertain practical and technical difficulties in the application of Measure 47 led the legislature to propose, and the people to adopt, Measure 50 as its effective replacement’ ” (quoting Shilo Inn, 36 P.3d at 958 n. 6)). As such, voters would have been presented with substantially identical summaries before the election on Measure 47 as well. Accordingly, although voters might not have had perfect knowledge of the Oregon property tax system, they were likely aware of the constitutional three-percent limit, which not only was the subject of two separate ballot measures, both of which the voters approved, but which also affected the calculation of property taxes for every year since its adoption.

As to the difference between maximum assessed values and property taxes and the likelihood that the latter will increase by three percent, Caruso’s assertions are undermined by other limitations imposed by Measure 50. Taken together, these limitations create more than an “infinitesimal possibility” that property taxes will increase three percent. As described above, in addition to imposing the three-percent limit, Measure 50 “rolled back” properties’ maximum assessed values for the tax year beginning July 1, 1997, and established a “permanent limit on the rate of ad valorem property taxes imposed” by each local taxing district. Or. Const, art. XI, § ll(l)(a)-(3)(b). The State informs us that real market values, meanwhile, have been “generally rising.” The result is that for tax years beginning after July 1, 1997, a property’s assessed value — defined as “the lesser of: (a) The property’s maximum assessed value; or (b) The property’s real market value,” Or.Rev.Stat. § 308.146 — is likely to equal its maximum assessed value. And as maximum assessed values increase by three percent per year, property taxes increase by the same amount because they are imposed at the permanent rate established by Measure 50.

Finally, as to the “cause” of a greater-than-three-percent increase in property taxes, predictable three-percent increases in property taxes suggest that local option taxes may in fact “produce[ ]” an increase of more than that amount. See Black’s Law Dictionary 234 (8th ed.2004) (defining “cause” as “[sjomething that produces an effect or result”). That is, absent approval of a local option tax, maximum assessed values and, frequently, assessed values and property taxes could increase by no more than three percent. Or. Const, art. XI, § ll(l)(b). If a local option tax is approved, however, it could produce an increase of more than that amount. Id. § ll(4)(a)(A).

Certainly, the three-percent warning could also be read to suggest that Measure 36-55 by itself might cause property taxes to increase more than three percent, but we regard its interpretation as an accurate reminder that the measure would not be subject to the constitutional three-percent limit to be at least as plausible. For this reason, we do not characterize the burden imposed by section 280.070(4)(a) as “severe.” Rather, like the Fourth Circuit, we believe that we may not declare a State’s ballot language unconstitutionally burdensome “merely because it could conceivably mislead some individuals and could have been crafted more adroitly.” Cf. McLaughlin v. N.C. Bd. of Elections, 65 F.3d 1215 , 1227 (4th Cir.1995) (rejecting constitutional challenges to the allegedly misleading heading printed on voter petitions because the court could not say that the language “permits of only one interpretation” (internal quotation marks omitted)).

b. Interest

Where, as here, a state election law imposes restrictions on speech that are not severe, “ ‘the State’s important regulatory interests are generally sufficient’ ” to justify it. Burdick, 504 U.S. at 434 , 112 S.Ct. 2059 (quoting Anderson, 460 U.S. at 788 , 103 S.Ct. 1564 ). The State maintains that section 280.070(4)(a) advances its interest in reminding voters that approving a local option tax effectively waives the constitutional three-percent limit. The sufficiency of this interest presents a question of law, which we review de novo. See McClure v. Galvin, 386 F.3d 36, 41-45 (1st Cir.2004).

The Supreme Court has consistently acknowledged states’ legitimate interest in “fostering informed and educated expressions of the popular will.” Anderson, 460 U.S. at 796 , 103 S.Ct. 1564 ; see, e.g., Eu v. S.F. County Democratic Cent. Comm., 489 U.S. 214, 228 , 109 S.Ct. 1013 , 103 L.Ed.2d 271 (1989); Jenness v. Fortson, 403 U.S. 431, 442 , 91 S.Ct. 1970 , 29 L.Ed.2d 554 (1971). By requiring inclusion of the three-percent warning as a “shorthand designation” of an initiative’s potential tax consequences, section 280.070(4)(a) “plays a role in the process by which voters inform themselves for the exercise of the franchise.” Cf. Tashjian v. Republican Party, 479 U.S. 208, 220 , 107 S.Ct. 544 , 93 L.Ed.2d 514 (1986) (observing that “party labels provide a shorthand designation of the views of party candidates on matters of public concern”); compare McIntyre, 514 U.S. at 348-49 , 115 S.Ct. 1511 (noting that “in the case of a handbill written by a private citizen who is not known to the recipient, the name and address of the author add little, if anything, to the reader’s ability to evaluate the document’s message”).

We are mindful that a state’s asserted interest in informing voters will not necessarily justify the burdens its regulations impose on First Amendment rights. For example, “[a] state’s claim that it is enhancing the ability of its citizenry to make wise decisions by restricting the flow of information to them must be viewed with some skepticism.” Anderson, 460 U.S. at 798 , 103 S.Ct. 1564 . Skepticism is also warranted if a state goes beyond “merely” providing information about candidates to “handicap[ping]” them. Cook, 531 U.S. at 525 , 121 S.Ct. 1029 (internal quotation marks omitted). Section 280.070(4)(a), however, neither restricts information nor handicaps initiatives. It instead provides additional information, “ ‘opening] the channels of communication rather than ... closfing] them,’ ” and applies equally to all initiatives proposing local option taxes, leaving no measure at a “political disadvantage.” Compare Anderson, 460 U.S. at 798, 103 S.Ct. 1564 (quoting Va. Pharmacy Bd. v. Va. Consumer Council, 425 U.S. 748, 770 , 96 S.Ct. 1817 , 48 L.Ed.2d 346 (1976)); Cook, 531 U.S. at 525 , 121 S.Ct. 1029 . We note, too, that the State enacted section 280.070(4)(a) as part of a statutory scheme requiring that ballot titles for local option taxes include not only the three-percent warning, but also the rate, period, amount, and purpose of the proposed option tax. See Or.Rev.Stat. §§ 280.070, 280.075. The State’s directing attention to this variety of considerations is in our view more consistent with an interest in informing voters than with an interest in influencing them. Compare Anderson, 375 U.S. at 402 , 84 S.Ct. 454 (“[B]y directing the citizen’s attention to the single consideration of race or color, the State ... may [decisively influence the citizen to cast his ballot along racial lines.”).

As stated above, the three-percent warning might have been more clearly worded. For example, Article XI-A, section 4 of the Oregon Constitution, which authorizes a separate statewide property tax for the purpose of repaying the principal and interest of bonds, provides that “said tax levy hereby authorized shall be in addition to all other taxes which may be levied according to law.” Or. Const, art. XI-A, § 4. But because the three-percent warning, even as presently worded, does not impose a severe burden on Caruso’s First Amendment rights, “the State need not narrowly tailor the means it chooses to promote [its interests].” See Timmons, 520 U.S. at 365 , 117 S.Ct. 1364 .

The district court deemed section 280.070(4)(a) insufficiently tailored in a different respect: specifically, in its application to all measures proposing the imposition of a local option tax, even those which by themselves cannot cause property taxes to increase more than three percent. The district court concluded that the general applicability of section 280.070(4)(a) rendered it substantially overbroad and therefore facially invalid. See Broadrick v. Oklahoma, 413 U.S. 601, 615-16 , 93 S.Ct. 2908 , 37 L.Ed.2d 830 (1973).

We disagree. If, as the district court ordered, application of section 280.070(4)(a) were limited to measures which themselves could cause property taxes to increase more than three percent, the statute would be rendered largely ineffective in increasing the information available to voters. Under a separate provision of Oregon law, the ballot titles for measures proposing local option taxes must declare the “dollar rate per thousand dollars assessed value.” Or.Rev.Stat. § 280.060(l)(b). Voters can thus readily identify ballot measures that themselves may cause property taxes to increase more than three percent. What voters are less likely to readily identify are those ballot measures that propose lower dollar rates, but which, if approved, will be levied outside the three-percent limit imposed by the Oregon Constitution. See Or. Const, art. XI, § ll(4)(a)(A). In our view, then, the general applicability of section 280.070(4)(a) does not burden substantially more speech than is justified by the State’s important interests.

In sum, we conclude that the First Amendment burden imposed by section 280.070(4)(a) is not severe and further that that burden is justified by the State’s important interest in encouraging informed and educated voting. We accordingly conclude that section 280.070(4)(a) does not violate the First Amendment, either as applied to Measure 36-55 or on its face.

D. Fourteenth Amendment

In addition to his First Amendment claims, Caruso asserts a Fourteenth Amendment claim, arguing that required inclusion of the “false and misleading” three-percent warning violates his rights to substantive due process. We review the claim de novo. See Mont. Chamber of Commerce v. Argenbright, 226 F.3d 1049 , 1054 (9th Cir.2000).

“Several appellate courts, including our own, have held that an election is a denial of substantive due process if it is conducted in a manner that is fundamentally unfair.” Bennett v. Yoshina, 140 F.3d 1218, 1226 (9th Cir.1998). To prevail on his substantive due process claim, then, Caruso must demonstrate that “ ‘the state’s choice of ballot language so upset[s] the evenhandedness of the [election] that it work[s] a ‘patent and fundamental unfairness’ on the voters.’ ” Nat’l Audubon Soc’y, Inc. v. Davis, 307 F.3d 835, 858 (9th Cir.2002) (quoting Burton v. Georgia, 953 F.2d 1266, 1269 (11th Cir.1992)). The parties agree that such an exceptional case would arise if, for example, the ballot language were so misleading as to deceive voters about the subject of the measure at issue. See id.

Applying this standard in National Audubon Society, we rejected a challenge to ballot material which allegedly misled voters by understating the scope of proposed changes to California law. Id. at 859. The district court there concluded that an argument favoring passage of the proposition was not “materially misleading” in part because it “was not completely inaccurate.” Id. at 858. Rather, although the argument allegedly understated the scope of the broader of two proposed bans, it accurately described the narrower one. Id.

In addition, “other materials accompanying the ballot” included arguments against the proposed changes. Id. at 858-59. We agreed that under those circumstances, “the ballot material did not rise to the level of a substantive due process violation.” Id. at 859.

Like the material challenged in National Audubon Society, the three-percent warning is “not completely inaccurate.” To be sure, the three-percent warning might have been read as a misleading suggestion that Measure 36-55 by itself might cause property taxes to increase more than three percent. But, as described above, the warning might also have been read, in context, as an accurate reminder that the proposed local option tax would not be subject to the three-percent limit imposed by the Oregon Constitution. Moreover, although an average voter might have read the three-percent warning as Caruso does “ ‘[i]f [he] had to decide what he was voting on from the [warning] alone, .... he did not have to decide from this summary.’ ” Burton, 953 F.2d at 1271 (quoting Kohler v. Tugwell, 292 F.Supp. 978, 981 (E.D.La.1968)). Instead, he could look to “other materials” — including the text of Measure 36-55 and the remainder of the ballot title as it appeared in the voters’ pamphlet and, indeed, on the ballot itself, see Or.Rev.Stat. §§ 250.235, 251.315(l)(g), 280.060(l)(b) — which indicated the actual increase proposed by Measure 36-55, disclosing both the dollar rate of the proposed tax and the estimated levy for a house with an assessed value of $150,000. See Nat’l Audubon Soc’y, 307 F.3d at 858-59 . We are thus unpersuaded that the State’s choice of ballot language rose to the level of a due process violation under National Audubon Society.

The fact that the three-percent warning would have appeared in the ballot title for Measure 36-55 rather than in an “avowedly partisan portion of the materials” does not change our conclusion. Compare Nat’l Audubon Soc’y, 307 F.3d at 858 (internal quotation marks omitted). Because the three-percent warning could have been interpreted accurately, and because “other materials” would have enforced this inter pretation, we cannot say that including the three-percent warning in the ballot title would have “infected” the entire election with “patent and fundamental unfairness.” Burton, 953 F.2d at 1271 (internal quotation marks omitted).

III. CONCLUSION

For the above reasons, we conclude that Or.Rev.Stat. § 280.070(4)(a) does not violate Caruso’s First Amendment or Fourteenth Amendment rights. Accordingly, we reverse the contrary determinations of the district court and vacate the injunction permanently limiting the enforcement of section 280.070(4)(a).

JUDGMENT REVERSED; INJUNCTION VACATED.

. Formation of the Yamhill County People's Utility District (“PUD”) was proposed by Measure 36-54, which appeared on the same ballot. . The named defendant, Yamhill County, also participated in the proceedings below. It did not appeal the judgment of the district court. . We do not dispute that the real market values for some properties may be decreasing. For these properties, the likelihood that property taxes would increase by three percent in a given year would be much lower, because assessed values would likely equal decreasing real market values rather than incrementally increasing maximum assessed values. But particularly in the absence of evidence that a substantial portion of Oregon properties fit this description, we decline to condition application of section 280.070(4)(a) on the real market values of individual properties. Doing so would effectively “tie the hands” of the State. See Burdick, 504 U.S. at 434 , 112 S.Ct. 2059 .