We are called upon to decide whether Oregon Ballot Measure 26’s prohibition of payment to electoral petition signature gatherers on a piece-work or per signature basis unconstitutionally burdens core political speech. Because the district court did not clearly err in determining that the plaintiffs failed to establish that the challenged measure significantly burdens speech, we cannot hold the Measure imposes a severe burden under the First Amendment. Therefore, because the defendant has established an important regulatory interest in support of the Measure, the plaintiffs have failed to prove that the prohibition violates the First Amendment.

In November 2002, Oregon voters approved Ballot Measure 26 (“Measure 26”), a voter initiative, by a margin of 75 per cent to 25 percent. Measure 26 reads: 1863

Or. Const., art. IV, § lb.

Barbara and Eugene Prete and Jason Williams (collectively “plaintiffs”), as chief petitioners, later coordinated signature gathering to place various initiative measures on the February and November 2004 general election ballots. Oregon’s Elections Division office sent inquiry letters to plaintiffs in November 2003, advising plaintiffs that the Elections Division had received complaints alleging plaintiffs had paid signature gatherers on the basis of the number of signatures collected, in violation of Measure 26. The inquiry letters requested additional information from plaintiffs.

Plaintiffs responded by bringing an action in federal district court against defendant, alleging Measure 26 violated the First Amendment. Plaintiffs sought declaratory and injunctive relief. Six days later, Tim Nesbitt and the Oregon AFL-CIO (collectively “intervenor-defendants”) brought a motion to intervene as of right under Fed.R.Civ.P. 24(a)(2), and alternatively, for permissive intervention under Fed.R.Civ.P. 24(b). Nesbitt, president of the Oregon AFL-CIO, was chief petitioner for Measure 26, and the Oregon AFL-CIO was a major supporter of Measure 26. Plaintiffs opposed the motion; Bill Bradbury, in his official capacity as the Secretary of State of Oregon (hereinafter “de fendant”), did not. The district court granted the motion to intervene as of right.

Plaintiffs then brought a motion for a preliminary injunction to enjoin defendant from enforcing Measure 26. After oral argument on the motion, the parties stipulated no further discovery was needed and the court could issue a final ruling on the merits pursuant to Fed.R.Civ.P. 65(a)(2).

In its amended opinion and order, the district court found Measure 26 was targeted at electoral processes rather than at the communicative aspect of petition circulation. The court reasoned Measure 26 prohibited only one method of payment for petition circulators, “a matter entirely between the circulator, his or her employer, and the chief petitioner.” Next, the court found Measure 26 imposed no severe or substantial burdens on the circulation of initiative or referendum petitions, and defendant’s interest in protecting the integrity of the initiative process justified the lesser burdens imposed by the measure. The court, therefore, denied plaintiffs’ motion for a preliminary injunction and entered judgment in favor of defendant and intervenor-defendants. Plaintiffs timely appealed.

On appeal, plaintiffs assert (1) the district court erred in granting intervenor-defendants’ motion to intervene as of right, and (2) Measure 26 violates the First Amendment of the United States Constitution. We have jurisdiction under 28 U.S.C. § 1291 and we hold: (1) the district court erred in granting intervenor-defen-dants’ motion to intervene but that error was harmless; and (2) the district court did not err in determining plaintiffs failed to establish Measure 26 violates the First Amendment. Accordingly, we AFFIRM the judgment of the district court.

II.

This court reviews de novo a district court’s ruling on a motion to intervene as of right pursuant to Fed.R.Civ.P. 24(a)(2). United States v. Alisal Water Corp., 370 F.3d 915, 918 (9th Cir.2004).

Under Fed.R.Civ.P. 24(a)(2), an applicant for intervention as of right must demonstrate that: (1) the intervention application is timely; (2) the applicant has a “significant protectable interest relating to the property or transaction that is the subject of the action”; (3) “the disposition of the action may, as a practical matter, impair or impede the applicant’s ability to protect its interest”; and (4) “the existing parties may not adequately represent the applicant’s interest.” Alisal Water Corp., 370 F.3d at 919 (internal quotation marks and citations omitted). Although the party seeking to intervene bears the burden of showing those four elements are met, “the requirements for intervention are broadly interpreted in favor of intervention.” Id.

A. Timeliness, “Significant Protectable Interest,” and Impairment

Here, plaintiffs wisely concede the intervenor-defendants’ application was timely and the intervenor-defendants have a “significant protectable interest” relating to the subject of this action. First, intervenor-defendants brought the motion to intervene only six days after plaintiffs brought the action. Second, for purposes of intervention as of right, a public interest group that has supported a measure (such as an initiative) has a “significant protectable interest” in defending the legality of the measure. Sagebrush Rebellion, Inc. v. Watt, 713 F.2d 525, 528 (9th Cir.1983). Third, an adverse court decision on such a measure may, as a practical matter, impair the interest held by the public interest group. Id.

In Sagebrush Rebellion, this court held that a public interest group may have a protectable interest in defending the legality of a measure it had supported. Id. at 527 . There, a public interest group which had supported the creation of a conservation area in Idaho sought to intervene on behalf of the government in an action challenging the federal statute that created that conservation area. Id. at 526 . The district court denied the motion to intervene. This court reversed, holding the group had a protectable interest in defending the creation of the conservation area. We stated in broad language that “a public interest group [is] entitled as a matter of right to intervene in an action challenging the legality of a méasure which it had supported.” Id. at 527 . Further, an adverse decision against the conservation area “would impair the society’s interest in the preservation of birds and their habitats,” an interest the conservation area was designed to protect. Id. at 528 . This court also held the government’s representation of the group’s interest “may be inadequate” (for reasons discussed infra); thus, this court reversed and remanded to the district court for it to grant the motion to intervene. Id. at 529 .

Here, Nesbitt was chief petitioner for the measure, and the Oregon AFL-CIO was a main supporter of the measure. Under the rule from Sagebrush Rebellion, intervenor-defendants thus have a “significant protectable interest” related to this action, and an adverse judgment might impede or impair that interest.

Plaintiffs contend, however, that Arizonans for Official English v. Arizona, 520 U.S. 43 , 117 S.Ct. 1055 , 137 L.Ed.2d 170 (1997) (hereinafter “AOE”), controls here and bars initiative sponsors from intervening in judicial challenges to the initiative. Plaintiffs misread AOE. There, the plaintiff (Yniguez), a state employee, brought an action against the State of Arizona alleging the adoption of an initiative which declared English “the official language of [Arizona]” violated the First Amendment. Id. at 49 , 117 S.Ct. 1055 . Yniguez complained she often spoke Spanish with Spanish-speaking persons as part of her state job, and the initiative’s mandate for state employees to “act in English” could expose her to sanctions. Id. at 50 , 117 S.Ct. 1055 . After a bench trial, the district ruled the initiative was unconstitutional as overbroad. Id. at 54 , 117 S.Ct. 1055 . The Arizonans for Official English Committee (“AOE”) — which was the principal sponsor of the initiative' — then brought a motion to intervene, seeking to defend the constitutionality of the initiative on appeal. Id. at 56, 117 S.Ct. 1055 . The district court denied the motion. Yniguez then resigned from her employment with the state. AOE appealed nonetheless, and this court determined AOE had Article III standing to pursue the appeal in defense of the initiative, and the action was not moot because of Yniguez’s resignation. Id. at 58-60 , 117 S.Ct. 1055 . The U.S. Supreme Court reversed. The Court observed that AOE was not an elected representative, nor did any Arizona state law appoint initiative sponsors as agents “to defend, in lieu of public officials, the constitutionality of initiatives made law of the State.” Id. at 65 , 117 S.Ct. 1055 . On that basis, the Court stated: “We thus have grave doubts whether AOE ... ha[s] standing under Article III to pursue appellate review. Nevertheless, we need not definitively resolve the issue. Rather, we will follow a path we have taken before and inquire, as a primary matter, whether originating plaintiff Yniguez still has a case to pursue.” Id. at 66 , 117 S.Ct. 1055 . The Court concluded Yniguez’s resignation after the district court’s judgment but before appeal mooted the case, and the Court then vacated the decision of the district court and the court of appeals. Id. at 72, 75 , 117 S.Ct. 1055 .

AOE did not hold that initiative sponsors do not have an interest in defending the initiative sufficient to support intervention. The main issue presented in AOE was whether the intervenor-applicant there had Article III standing to pursue an appeal when a step taken by the original plaintiff (resignation of her job) rendered the entire case or controversy moot. Such a scenario is not at issue here. Therefore, we hold that under Sagebrush Rebellion, intervenor-defendants have a “significant protectable interest” related to this action and an adverse judgment may impair or impede that interest.

B. Adequacy of Representation

A closer issue is presented whether in-tervenor-defendants established that “the existing parties may not adequately represent the applicant’s interest.” See Alisal Water Corp., 370 F.3d at 919 . Plaintiffs contend that because defendant is defending the constitutionality of Measure 26, intervenor-defendants’ interest in defending the constitutionality of Measure 26 is adequately represented. The district court disagreed, concluding defendant might not adequately represent in-tervenor-defendants’ interests because in-tervenor-defendants “claim an interest in preventing the gathering and eventual counting of invalid signatures for initiatives opposing union interests,” and thus defendant possibly could make different arguments than intervenor-defendants.

In assessing whether a present party will adequately represent an intervenor-applicant’s interests, we “consider several factors, including whether [a present party] will undoubtedly make all of the intervenor’s arguments, whether [a present party] is capable of and willing to make such arguments, and whether the intervenor offers a necessary element to the proceedings that would be neglected.” Sagebrush Rebellion, 713 F.2d at 528 . The burden of showing inadequacy of representation is minimal and “is satisfied if the applicant shows that representation of its interests ‘may be’ inadequate....” Id. (internal citations omitted).

Although the burden of establishing inadequacy of representation may be minimal, the requirement is not without teeth:

Arakaki v. Cayetano, 324 F.3d 1078, 1086 (9th Cir.2003) (internal citations omitted). Additionally, “[t]here is also an assumption of adequacy when the government is acting on behalf of a constituency that it represents. In the absence of a very compelling showing to the contrary, it will be presumed that a state adequately represents its citizens when the applicant shares the same interest.” Id. (internal citations and quotation marks omitted).

In Sagebrush Rebellion, discussed supra, we held the public interest group seeking to intervene as of right established that the defendant (the Secretary of the Interior) might not adequately represent the group’s interest. 713 F.2d at 528 . We reasoned that the Secretary of the Interi- or, James Watt, had previously been head of the foundation which was representing the plaintiff in the present action. Thus, the public interest group — intervening on the defendant’s side — might bring a perspective materially different from that of the present parties and was entitled to intervene. Id.

In League of United Latin Am. Citizens v. Wilson, however, this court recognized that when an intended intervenor and a party in the action seek the same ultimate objective, a presumption arises that the intervenor’s interests are adequately presented. 131 F.3d 1297 . In League of United Latin Am. Citizens, the plaintiff brought a lawsuit challenging California’s Proposition 187, which had been enacted into law. Id. at 1300. A public interest group brought a motion to intervene as of right, claiming it participated in the drafting and sponsorship of the proposition and desired to intervene in support of its defense. Id. at 1301. The district court denied the motion, and we affirmed. This court recognized the defendant (the State of California) and the public interest group sought the same ultimate objective — i.e., to defend the constitutionality of Proposition 187 — and thus a presumption of adequacy of representation arose. Id. at 1305. Hence, we held the intervenor-applicant’s interests were adequately represented by the state defendant and affirmed the denial of the motion to intervene.

Here, the ultimate objective for both defendant and intervenor-defendants is upholding the validity of Measure 26. Thus, a presumption arises that defendant is adequately representing intervenor-de-fendants’ interests. See id. at 1305. Second, defendant is the Oregon government, and intervenor-defendants (the Oregon AFL-CIO and its president) share the same interest with defendant, i.e., defending Measure 26. Therefore, it is assumed that defendant is adequately representing intervenor-defendants’ interests. Arakaki, 324 F.3d at 1086 . While it is unclear whether this “assumption” rises to the level of a second presumption, or rather is a circumstance that strengthens the first presumption, it is clear that “[i]n the absence of a ‘very compelling showing to the contrary,’ it will be presumed that” the Oregon government adequately represents the interests of the intervenor-defendants. See id.

Intervenor-defendants fail to present that compelling showing of inadequate representation. In their motion to intervene, intervenor-defendants stated first that defendant may not be able to provide a complete defense of Measure 26 due to “budget constraints.” Virtually all governments face budget constraints generally, and if such a basis were sufficient to establish inadequate representation, it would eliminate the presumption of adequate representation when the government and the intervenor-applicant share the same interest. Most importantly, there is no evidence in the record that defendant is unable to mount an effective defense of Measure 26 due to alleged “budget constraints.” See League of United Latin Am. Citizens, 131 F.3d at 1307 (citing Moosehead San. Dist. v. S.G. Phillips Corp., 610 F.2d 49, 54 (1st Cir.1979) (hold ing “a petitioner must produce something more than speculation to the purported inadequacy in order to justify intervention as of right”)).

Second, intervenor-defendants assert defendant “may be inclined [to] give an unnecessarily narrow construction of Measure 26 in the face of legal attacks on the measure.” Yet neither plaintiffs nor defendant have argued for a narrowing construction of Measure 26, and Measure 26 does not seem susceptible to any narrowing construction. Thus, intervenor-de-fendants have failed to present evidence sufficient to meet their burden of a “compelling showing” on this score as well.

Third, intervenor-defendants contend defendant “does not have the breadth of knowledge regarding the signature gathering process to fully develop the record and respond to plaintiffs factual allegations.” Intervenor-defendants assert they have “particular expertise in the subject of the dispute.” For example, they “have direct knowledge and experience in how well a signature gathering campaign staffed by hourly circulators can run.”

Yet defendant, as Oregon’s Secretary of State, is undoubtedly familiar with the initiative process and the requisite signature-gathering; indeed, defendant is the government party responsible for counting the signatures. Defendant also administers Oregon’s election processes and promulgates regulations to give effect to the state’s election statutes. See Or. Admin. R. 165 -014-0260 (interpreting Measure 26). Although intervenor-defendants may have some specialized knowledge into the signature gathering process, they provided no evidence to support their speculation that the Secretary of State lacks comparable expertise. To the contrary, defendant presumably is sufficiently acquainted with the signature gathering process and could also acquire additional specialized knowledge through discovery (e.g., by calling upon intervenor-defendants to supply evidence) or through the use of experts. Thus, such a reason is insufficient to provide the “compelling showing” necessary to overcome the presumption of adequate representation discussed supra.

Accordingly, while we emphasize that the burden of showing inadequacy of representation is generally minimal, here in-tervenor-defendants failed to present evidence sufficient to support a finding that their interests are not adequately represented by the defendant in this action. We hold, therefore, that the district court erred in granting the motion to intervene as of right.

C. Remedy

The remedy for an improper grant of intervention has not been clearly established. It is more common for appellate courts to consider the denial of a motion to intervene, and the few cases reversing the grant of a motion to intervene are distinguishable because here the district court did not enter separate judgments for the defendant and the intervenor-defen-dants as in the cited cases, but entered a single judgment in favor of both defendant and intervenor-defendants: that Measure 26 does not violate the First Amendment.

Here, the district court erred in granting intervention as of right and thereby allowing intervenor-defendants to present evidence and argument. Under 28 U.S.C. § 2111 and Federal Rule of Civil Procedure 61, however, we may not reverse the district court’s judgment unless this error affected the “substantial rights of the parties.” Cf. Texas Co. v. Hogarth Shipping Corp., 256 U.S. 619, 629 , 41 S.Ct. 612 , 65 L.Ed. 1123 (1921) (applying harmless error review to the erroneous grant of intervention as amicus curiae); Alaska v. Suburban Propane Gas Corp., 123 F.3d 1317, 1321-22 (9th Cir.1997) (applying harmless error review under Rule 61 to the erroneous denial of a motion to intervene); California ex rel. State Lands Com’n v. United States, 805 F.2d 857 , 866 n. 6 (9th Cir.1986) (declining to “consider whether or under what circumstances an erroneous grant of intervention could constitute reversible error under Fed.R.Civ.P. 61”); Hackin v. Lockwood, 361 F.2d 499 (9th Cir.1966) (the improper joinder of a civil defendant does not prevent this court from addressing the merits of the action as to the proper parties).

Here, the district court’s error in granting the motion to intervene did not affect the substantial rights of the parties. In its amended opinion and order, the district court discussed only one piece of evidence submitted by intervenor-defendant: an affidavit submitted by Ted Blaszak of Democracy Resources of Oregon, Inc., a signature-gathering firm. Blaszak averred that the requirement to pay petition circu-lators by the hour rather than by the signature did not significantly increase his costs or decrease productivity. Although helpful to defendant’s case, the evidence was not crucial. As noted infra in footnote 21, consideration of that affidavit does not make it more probable than not that the district court’s error tainted the judgment.

Accordingly, the district court erred in granting intervenor-defendants’ motion to intervene as of right, but the error was harmless and, therefore, does not require vacating the judgment of the district court.

III.

In reviewing a district court’s final judgment after consolidation of its preliminary injunction ruling with its decision on the merits pursuant to Fed.R.Civ.P. 65(a)(2), we review the district court’s factual findings for clear error and its conclusions of law de novo. Associated Builders & Contractors of S. California v. Nunn, 356 F.3d 979 , 984 (9th Cir.2004). When the issue presented involves the First Amendment, however, the standard of review is modified slightly. Historical questions of fact (such as credibility determinations or ordinary weighing of conflicting evidence) are reviewed for clear error, while constitutional questions of fact (such as whether certain restrictions create a “severe burden” on an individual’s First Amendment rights) are reviewed de novo. Planned Parenthood of the Columbia/Willamette, Inc. v. Am. Coalition of Life Activists, 290 F.3d 1058 , 1070 (9th Cir.2002).

The First Amendment, incorporated and made applicable to the states by the Fourteenth Amendment, prohibits state governments from enacting a “law ... abridging the freedom of speech.” McIntyre v. Ohio Elections Comm’n, 514 U.S. 334 , 336 & n. 1, 115 S.Ct. 1511 , 131 L.Ed.2d 426 (1995). As discussed infra, the circulation of initiative and referendum petitions involves “core political speech,” and is, therefore, protected by the First Amendment. See Meyer v. Grant, 486 U.S. 414, 421-22 , 108 S.Ct. 1886 , 100 L.Ed.2d 425 (1988).

The First Amendment does not, however, prohibit all restrictions upon election processes: “States may, and inevitably must, enact reasonable regulations of parties, elections, and ballots to reduce election— and campaign-related disorder.” Timmons v. Twin Cities Area New Party, 520 U.S. 351, 358 , 117 S.Ct. 1364 , 137 L.Ed.2d 589 (1997). Indeed, the U.S. Supreme Court has recognized “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.” Buckley, 525 U.S. at 191, 119 S.Ct. 636 (1999).

For purposes of determining whether a state election law violates an individual’s First Amendment rights, we

Arizona Right to Life Political Action Comm., 320 F.3d at 1007-08 (quoting Timmons, 520 U.S. at 358 , 117 S.Ct. 1364 ) (emphases added and internal quotation marks omitted). The U.S. Supreme Court has counseled against establishing any bright-line rule in this field: “no litmus-paper test will separate valid ballot-access provisions from invalid interactive speech restrictions; we have come upon no substitute for the hard judgments that must be made.” Buckley, 525 U.S. at 192 , 119 S.Ct. 636 (internal quotation marks omitted).

In Meyer v. Grant, the Supreme Court recognized the expressive nature of petition circulation and held the wholesale prohibition of paid petition circulators imposed an impermissible burden on free speech under the First Amendment. 486 U.S. 414 , 108 S.Ct. 1886 , 100 L.Ed.2d 425 . In Meyer , the plaintiffs challenged an amendment to the Colorado constitution which made it a felony to pay money or anything of value to petition circulators who circulated initiative or referendum petitions. Id. at 415 , 108 S.Ct. 1886 . After a bench trial, the district court upheld the statute, but the court of appeals reversed. Id. at 418-420 , 108 S.Ct. 1886 . The U.S. Supreme Court affirmed, explaining:

Id. at 421-22 , 108 S.Ct. 1886 .

The Court recognized that a wholesale prohibition of paid petition circulators limited such “core political speech” in two ways: (1) “it limits the number of voices who will convey [plaintiffs’] message and the hours they can speak and, therefore, limits the size of the audience they can reach”; and (2) “it makes it less likely that [plaintiffs] will garner the number of signatures necessary to place the matter on the ballot, thus limiting their ability to make the matter the focus of statewide discussion.” Id. at 423-424 , 108 S.Ct. 1886 . The Court rejected Colorado’s argument that the prohibition was justified by the state’s interest in protecting the integrity of the initiative process, reasoning that Colorado presented no evidence that paid petition circulators are more likely to accept fraudulent signatures over those of a volunteer, and that other Colorado statutes prohibited accepting forged or fraudulent signatures. Id. at 426-27 , 108 S.Ct. 1886 . The Court thus concluded the prohibition “imposes a burden on political expression that the State has failed to justify,” and hence the prohibition violated the First Amendment. Id. at 428 , 108 S.Ct. 1886 .

Similarly, in Buckley the Supreme Court struck down a Colorado statute which required: (1) petition circulators be registered voters in Colorado; (2) petition circulators wear an identification badge bearing the circulator’s name; and (3) initiative proponents publicly disclose the names and amounts paid to all paid circu-lators. 525 U.S. at 186 , 119 S.Ct. 636 . First, the Court observed the registered voter requirement “decreases the pool of potential circulators as certainly as that pool is decreased by the prohibition of payment to circulators. Both provisions limit the number of voices who will convey the initiative proponents’ message and, consequently, cut down the size of the audience proponents can reach.” Id. at 194-95, 119 S.Ct. 636 (internal quotation marks and alterations omitted). The Court rejected Colorado’s assertion that the registered voter requirement was not a severe burden because it was not difficult to register to vote; although failure to register sometimes results from ignorance or apathy, the decision not to register can also implicate “political thought and expression.” Id. at 195-96, 119 S.Ct. 636 . The Court also struck down the name badge requirement and the disclosure provisions, explaining that both provisions forced circulators to surrender the anonymity enjoyed by their volunteer counterparts and had only a tenuous relationship to Colorado’s interest in ensuring the integrity of the initiative process. Id. at 198-204, 119 S.Ct. 636 .

Unlike Meyer , Measure 26 does not completely prohibit the payment of initiative-petition circulators. Instead it prohibits one method of payment. Plaintiffs claim that Measure 26 in practice limits the available pool of people willing to circulate petitions. To the extent Meyer may be read to indicate that any resulting decrease in the pool of available circulators is sufficient to constitute a “severe burden” under the First Amendment, in Buckley the Court refined its analysis and made clear that the degree of the decrease resulting from the measure is properly considered in determining the severity of the burden. See id. at 192, 119 S.Ct. 636 (analyzing the degree of the effect of the challenged provisions on the pool of available circulators and explaining: “We therefore detail why we are satisfied that, as in Meyer , the restrictions in question significantly inhibit communication with voters about proposed political change, and are not warranted by the state interests ... alleged to justify those restrictions.”) (emphasis added).

Unlike Buckley , where the pool was limited to state residents registered to vote, here, anyone may serve as a petition circu-lator, regardless of residence or registration. Therefore, we find the Eight Circuit’s analysis of a North Dakota state law more analogous to Measure 26, and thus more persuasive. In Initiative & Referendum Inst. v. Jaeger, 241 F.3d 614 (8th Cir.2001), the Eighth Circuit distinguished North Dakota’s prohibition on paying initiative-petition circulators “on a basis related to the number of signatures obtained” (i.e., the same type of restriction at issue here) from the complete prohibition on paid petition circulators in Meyer . In Jaeger, the court noted that the state had an “important interest in preventing signature fraud” in the initiative process, and that the state had supported that interest with evidence that paying petition circula-tors per signature encouraged such fraud. Id. at 618. Further, the plaintiffs had “produced no evidence that payment by the hour, rather than on commission, would in any way burden their ability to collect signatures. The [plaintiffs] have only offered bare assertions on this point.” Id. Thus, because the state asserted an important interest in preventing signature fraud, supported that interest with evidence that signature fraud was actually a problem in North Dakota, and the plaintiffs failed to present evidence the restriction would otherwise burden their ability to collect signatures, the court upheld the ban on paying petition circulators on the basis of the number .of signatures collected. Id. For reasons discussed further below, our case is more properly analyzed under the framework applied in Jaeger than under Meyer or Buckley .

A. “Severe” or “Lesser” Burden

Plaintiffs contend Measure 26 imposes a severe burden on the circulation of initiative petitions because the measure makes paid signature gathering prohibitively expensive, inefficient, and results in a higher rate of invalid signatures. Plaintiffs thus contend that strict scrutiny should apply and Measure 26 is not narrowly tailored to serve a compelling governmental interest. The district court rejected this argument, finding plaintiffs did not prove that Measure 26 imposed severe burdens on the circulation of initiative petitions, and any lesser burdens imposed by Measure 26 were reasonably related to and justified by the state’s interest in preventing fraud in the initiative process.

In reaching that conclusion, the district court assessed plaintiffs’ claims that: (1) Measure 26 eliminates an avenue of signature-gathering and decreases the available pool of petition circulators; (2) Measure 26 increases the costs of gathering signatures, making it more difficult to circulate petitions and qualify initiative or referendum measures for the ballot; and (3) Measure 26 resulted in a significant decrease in the number of valid signatures collected by signature gatherers. Because these are claims of historical fact, we review the district court’s findings regarding these claims for clear error. See Planned Parenthood, 290 F.3d at 1070. Because the district court did not clearly err in rejecting each of these claims, we affirm the district court’s' holding that Measure 26 imposes only a lesser burden on the circulation of initiative petitions. .

1. The Effect Of Measure 26 On The Pool of Petition Circulators in Oregon

Plaintiffs presented affidavits from William Arno of Arno Political Consultants (“APC”), a California petition circulation firm, and Tracy Taylor of Taylor Petition Management, LLC, a Washington state petition circulation firm (under contract with APC). Arno averred Measure 26 “make[s] it less likely that companies such as APC will continue to do business in Oregon” and stated he had “personal knowledge that at least three of [his] chief competitors will not do business in Oregon” because of Measure 26. Yet Arno later testified that his “personal knowledge” of those companies came only from Taylor.

Taylor averred that several professional petition circulators he knew were not interested in working in Oregon on an hourly wage basis when they could work in other states on a per-signature basis. Yet he did not identify those circulators, nor state whether they would work in Oregon absent Measure 26. Indeed, both Arno and Taylor conceded that other factors might cause petition circulators to leave or not to work in Oregon. For example, Arno noted that access to private property for petition circulators was particularly strict in Oregon. Both Arno and Taylor also noted that “harassment” of paid petition circulators by the Voter Education Project discouraged some companies from working in Oregon. Last, Arno also suggested that Measure 26 makes it more difficult for petition circulation companies to operate because those companies have to treat petition circulators as employees rather than independent contractors. Yet Oregon employment law requires that petition circulators must be treated as employees rather than independent contractors; Measure 26 does not itself mandate such treatment. See Canvasser Services, Inc. v. Employment Dep’t, 163 Or.App. 270 , 987 P.2d 562, 568 (1999) (holding, under Oregon law, petition circulators are employees rather than independent contractors).

The district court rejected Arno and Taylor’s averments of circulators leaving or refusing to work in Oregon as “unsupported speculation” because Arno received his information from Taylor, Taylor only repeated the basic claim that paid circula-tors would not work in Oregon because of Measure 26, and several factors other than Measure 26 could explain the alleged reluctance of petition circulators to work in Oregon. The district court’s factual conclusion is supported by the record and is not clearly erroneous.

Plaintiffs also presented affidavits from David Rubin of Universal Petitions, a southern California petition circulating firm; Lura Lucille Cordes, who employs initiative-petition circulators to gather signatures in California; and Angelo Paparel-la of Progressive Campaigns, Inc., a national signature gathering firm. Rubin averred that because — in his opinion — payment by signature is more efficient than payment by the hour, and Measure 26 would thus make signature gathering more difficult in Oregon, he is “sure I would never be asked to go to Oregon to coordinate a petition drive with Measure 26 restrictions in effect.” Similarly, Cordes stated because of the burdens imposed by Measure 26, she would “not come to Oregon to circulate petitions/gather signatures.” Paparella also stated that because of Measure 26, his company “will not circulate petitions in Oregon because the cost of hiring and maintaining a workforce of hourly wage workers is very, very high when compared to using petition circula-tors who are independent contractors.” Yet none of these affiants stated they had ever circulated petitions in Oregon or would do so in the absence of Measure 26. Further, Paparella’s averment suggests he would not come to Oregon because he would have to treat petition circulators as employees rather than as independent contractors, which is the law in Oregon notwithstanding Measure 26. Therefore, the district court’s conclusion that plaintiffs did not prove Measure 26 “caused a reduction in the number of available circulators or otherwise limit[ed] the size of plaintiffs audience” is supported by the record and is not clearly erroneous.

2. The Effect Of Measure 26 On The Cost of Signature Gathering in Oregon

•Arno averred that Measure 26 would increase the cost of signature collection by 35-45 percent. Taylor similarly averred Measure 26 would increase the cost of gathering signatures in Oregon. Yet both Arno and Taylor based their predictions on the misapprehended fact that Measure 26 converted circulators from independent contractors into employees, resulting in increased payroll costs. As noted above, Oregon law recognizes petition circulators as employees, rather than independent contractors, notwithstanding Measure 26.

Further, Arno and Taylor had little, if any, experience in initiative-petition circulation in Oregon before Measure 26 was passed. Arno testified he had worked on one initiative campaign in Oregon “around 1992,” but that campaign “ended up folding prior to turning in signatures.” Arno had not worked on any other initiative campaigns in Oregon. Similarly, Taylor testified that apart from an unrelated petition (Referendum Petition 401 placed on the November 2004 ballot), he had never worked on any initiative campaigns in Oregon before or after Measure 26 was passed. Thus, as noted by the district court, neither Arno nor Taylor could “offer a reliable comparison on the added costs, if any, imposed by Measure 26.”

Plaintiffs submitted several other affidavits which they contend support their claim that Measure 26 poses a severe burden by increasing costs. Jason Williams (one of the plaintiffs) averred he did not circulate an initiative petition “due in large part to the fact that the cost of circulating the petition, using paid signature gatherers, has increased significantly.” Yet Williams does not aver that Measure 26 is responsible for any such price increase. R. Russell Walker, chief petitioner for an unrelated initiative (initiative petition 59), averred that he did not circulate that petition “due in large part to the fact that the cost of circulating the petition, using paid signature gatherers, has increased significantly.” Similarly, he makes no averment that Measure 26 is to blame.

The district court ultimately concluded “Measure 26 imposes no appreciable burden in terms of costs for an initiative or referendum campaign.” Implicit in this finding is the conclusion that Measure 26 does not substantially increase the cost of initiative-petition circulation. This finding is supported by the record and is not clearly erroneous.

3. The Effect of Measure 26 On The Invalidity Rate of Signatures Gathered for Initiatives in Oregon

Arno averred he “noticed a significant decrease in the number of valid signatures collected by signature gatherers since Measure 26 became law.” Taylor similarly averred he “discovered a disproportionate number of signatures to be invalid [in Oregon], as opposed to the validity rates I am encountering in Washington and Ohio [which states do not prohibit payment by signature].” Yet Taylor does not attribute the higher invalidity rate to Measure 26 or suggest any reason for the higher invalidity rate. He stated, however, that signature gatherers paid by the hour “have more of an incentive to defraud me [compared to signature gatherers paid per signature] because they know that regardless of whether I think the signatures are valid, the signature gatherer must still be paid an hourly wage.” As the district court noted, however, both Arno and Taylor testified they had limited to no experience in initiative and referendum processes in Oregon. Thus, their assertions that paying petition circulators by the hour, instead of per signature, results in higher signature invalidity rates carry little weight.

Both Williams and Walker also averred Oregon had a higher signature invalidity rate post-Measure 26. Yet neither affiant attributed the higher invalidity rate directly to Measure 26.

In contrast, defendant submitted an affidavit from Richard J. Ellis, Ph.D., a political science professor at Willamette University in Oregon. Ellis averred that “the available evidence — though limited — suggests that circulators paid by the hour also have a higher validity rate than those paid by the signature.” For example, in Oregon’s 2002 election year, Measure 26 (which used only circulators paid by the hour) had a signature validity rate of 73.43 percent, higher than the ten other initiative petitions submitted for that election. Ellis also states the overall signature validity rates have dropped in Oregon not because of Measure 26, but because a March 2000 directive by the Oregon Elections Division instructed county clerks (who confirm the validity of signatures on petitions) not to count initiative signatures by “inactive voters” (ie., voters who have registered but have not voted in a certain number of past elections).

Further, Referendum Petition 401, which was qualified for the February 2004 Oregon ballot, after the passage of Measure 26, had a signature validity rate of 84.55 percent. Arno, Taylor, and Williams were involved in the circulation and gathering of signatures for Referendum Petition 401. The high validity rate of the collection of signatures for Referendum Petition 401, conducted after the adoption of Measure 26, weighs against plaintiffs’ claim. The record, therefore, supports the district court’s conclusion that Measure 26 results in higher validity rates for signature collection, rather than lower validity rates. Therefore, the district court’s finding is not clearly erroneous.

In sum, plaintiffs’ presentation of proof falls short here. The district court did not clearly err in finding that Measure 26 did not decrease the pool of petition circula-tors in Oregon; did not increase the costs of signature gathering; and did not result in a higher invalidity rate of signatures gathered for initiatives.

We next review the district court’s determination that Measure 26 creates only a “lesser burden” on plaintiffs’ First Amendment rights. Because this question relates to a constitutional fact (i.e., what constitutes a “severe burden” or a “lesser burden”), we review the district court’s determination de novo. See Planned Parenthood, 290 F.3d at 1070.

As noted supra, the district court did not clearly err in finding plaintiffs failed to prove Measure 26 resulted in any burden on their First Amendment rights. Unlike Meyer , plaintiffs did not prove that Measure 26 limited “the number of voices who will convey [plaintiffs’] message and the hours they can speak”; that Measure 26 “limits the size of the audience [plaintiffs’] can reach”; or that Measure 26 makes it “less likely that [plaintiffs] will garner the number of signatures necessary to place the matter on the ballot, thus limiting their ability to make the matter the focus of statewide discussion.” See Meyer, 486 U.S. at 423-24 , 108 S.Ct. 1886 . Moreover, unlike Buckley , plaintiffs did not prove that Measure 26 significantly limits the available pool of people willing to circulate petitions or constrains petition circulators’ “political thought and expression.” See Buckley, 525 U.S. at 194-96 , 119 S.Ct. 636 .

Of course, from an economic perspective, eliminating one method of payment (but not every method, a la Meyer) for petition circulators could result in some barriers to entry in the signature procurement market. For a task like signature gathering, it is possible that paying per signature (i.e., a commission basis) can be more productive of signatures than paying an hourly wage. Whether Measure 26 creates such barriers to entry, however, is a question of historical fact reviewed for clear error. Here, the district court did not clearly err in finding plaintiffs failed to prove the existence of such barriers to entry or that, if present, they diminished petition circulators’ ability to garner the requisite number of signatures to qualify initiatives for the ballot. Absent proof that such barriers to entry existed and had the claimed result, we are not left with a “definite and firm conviction that a mistake has been made” by the district court. See Sawyer v. Whitley, 505 U.S. 333 , 346 n. 14, 112 S.Ct. 2514 , 120 L.Ed.2d 269 (1992); SEC v. Rubera, 350 F.3d 1084, 1093 (9th Cir.2003) (citing Easley v. Cromartie, 532 U.S. 234 , 242, 121 S.Ct. 1452 , 149 L.Ed.2d 430 (2001)) (“Under the clearly erroneous standard, we defer to the lower court’s determination unless, based on the entire evidence, we are possessed of a ‘definite and firm conviction that a mistake has been committed.’ ”).

Moreover, even if such barriers to entry did arise, they would result in only a “lesser burden” under the First Amendment. Measure 26 is quite limited in its proscription, barring only payment of petition cir-culators on the basis of the number of signatures gathered. It does not prohibit adjusting salaries or paying bonuses according to validity rates or productivity, see Or. Admin. R. 165 -014-0260, which could likely counter any barriers to entry.

In the absence of proof that Measure 26 creates such barriers to entry or otherwise burdens their First Amendment rights, plaintiffs have established only that Measure 26 imposes “lesser burdens” upon the initiative process. Generally, the finding of a “lesser burden” triggers a “less exacting review” under which an “important regulatory interest[]” will support a finding that the measure is a “reasonable, nondiscriminatory restriction! ].” See Bayless, 320 F.3d at 1007.

B. Oregon’s “Important Regulatory Interest”

Defendant has an important regulatory interest in preventing fraud and its appearances in its electoral processes. See Bayless, 320 F.3d at 1013; see also Timmons, 520 U.S. at 364 , 117 S.Ct. 1364 (“States certainly have an interest in protecting the integrity, fairness, and efficiency of their ballots and election processes as means for electing public officials.”). Further, the record supports the conclusion that Measure 26 is aimed at combating actual instances of fraud and forgery committed by petition circulators paid on the basis of the number of signatures gathered.

First, the voter pamphlet circulated to the voters in consideration of Measure 26 supports the conclusion that Measure 26 is aimed at combating fraud in the signature gathering process. See Ecumenical Ministries v. Oregon State Lottery Comm’n, 318 Or. 551 , 871 P.2d 106 , 111 n. 8 (1994) (“In considering the history of a constitutional provision adopted through the initiative process, [Oregon courts] examine[ ], as legislative facts, other sources of information that were available to the voters at the time the measure was adopted and that disclose the public’s understanding of the measure ... [such as] the ballot title and arguments for and against the measure included in the voters’ pamphlet .... ”). The voter pamphlet states in support of Measure 26 that “[t]his most recent election cycle saw convictions [of paid petition circulators] on a variety of forgery, fraud, and identity theft counts, charges pending against others and allegations of dozens more.” Measure 26 would combat such fraud, the pamphlet states, by removing the “incentive for fraud out of the system” by mandating hourly pay rather than per signature.

As evidence of the actual existence of fraud and forgery in the initiative process, defendant presented an affidavit from Bill Carroll, a criminal investigator in the Oregon Department of Justice. He averred that paying petition circulators per signature leads to two types of fraud. First, the signature gatherers often forge signatures, thus receiving payment for a collected signature even though the signature is invalid. Second, the signature gatherers falsely certify the petition signature sheets, either for petitions submitted by themselves or for other petition circula-tors.

As attachments to his affidavit, Carroll supplied reports of interviews of various signature gatherers (paid per signature) who had forged signatures on their petitions; purchased signature sheets filled with signatures, then submitted them with their petitions as if they had collected the signatures themselves; or participated in “signature parties” in which multiple petition circulators would gather and sign each others’ petitions.

Defendant also submitted an affidavit by-John Lindback, Director of Oregon Secretary of State’s Elections Division. He averred “the practice of paying signature gatherers by the signature is a substantial case of ... fraud” and forgery in the initiative process.

Plaintiffs point to the Arno and Taylor affidavits, however, which aver that signature gatherers would not engage in fraud or forgery

Although such a general proposition may be sound, it does not controvert defendant’s evidence discussed above that some signature gatherers paid per signature have engaged in fraud and forgery, nor does it diminish defendant’s important regulatory interest in preventing such fraud.

Like Jaeger, defendant asserted an important regulatory interest in preventing-fraud and forgery in the initiative process. Defendant supported that interest with evidence that signature gatherers paid per signature actually engage in such fraud and forgery. This court’s duty is not to determine whether the state’s chosen method for prevention of fraud is the best imaginable. Once the burden is found to be of the “lesser” variety, our inquiry is limited to whether the chosen method is reasonably related to the important regulatory interest. Last, as the district court correctly determined, plaintiffs did not prove Measure 26 would otherwise burden their ability to collect signatures. See Jaeger, 241 F.3d at 618.

In sum, because plaintiffs failed to prove the district court erred in determining that Measure 26 does not severely burden their First Amendment rights in circulating initiative petitions, and defendant has established that Measure 26 serves the important regulatory interest in preventing fraud and forgery in the initiative process, we hold that Measure 26 does not violate the First Amendment, as applied, and AFFIRM the judgment of the district court.

AFFIRMED.

.Oregon's Secretary of State issued an administrative rule interpreting Measure 26. The rule states in part Measure 26: bans the practice of paying circulators or others involved in an initiative or referendum effort if the basis for payment is the number of signatures obtained. This means that payment cannot be made on a per signature basis. Employment relationships that do not base payment on the number of signatures collected are allowed. Allowable practices include: paying an hourly wage or salary, establishing either express or implied minimum signature requirements for circulators, terminating circula-tors who do not meet the productivity requirements, adjusting salaries prospectively relative to a circulator’s productivity, and paying discretionary bonuses based on reliability, longevity and productivity, provided no payments are made on a per signature basis. Or. Admin. R. 165 -014-0260. A violation of Measure 26 will result in civil penalties of a minimum of $100 for each individual signature sheet containing signatures collected in violation of Measure 26. Id. . Under Oregon law, a petition for a ballot measure must designate one to three "chief petitioners,” who are the main sponsors of the measure. See Or. Rev. St. § 250.045(3). . As the district court noted, the inquiry letters did not threaten prosecution, but "simply notified] plaintiffs that complaints were filed and requested] additional information. Nonetheless, the inquiry letters sent to plaintiffs are sufficient to establish standing.” For First Amendment purposes, a plaintiff demonstrates an "injury-in-fact” where "the plaintiff intends to engage in a course of conduct arguably affected with a constitutional interest and that there is a credible threat that the challenged provision will be invoked against the plaintiff.” Arizona Right to Life PAC v. Bayless, 320 F.3d 1002, 1006 (9th Cir.2003) (internal quotation marks omitted). Here, plaintiffs intended to engage in signature gathering where payment is made per signature, and receipt of the inquiry letters is sufficient to establish a "credible threat” that Measure 26 will be invoked against plaintiffs. . The district court initially issued an opinion and order construing the parties' submissions as motions for summary judgment and granting summary judgment for defendant and in-tervenor-defendants. The court later vacated that opinion and order because the parties intended the court to consolidate trial on the merits with plaintiffs' motion for a preliminary injunction, pursuant to Fed.R.Civ.P. 65(a)(2). The court then issued an amended order and opinion, granting final judgment on the merits for defendant and intervenor-de-fendants. Plaintiffs do not contest the steps taken by the court to convert cross-motions for summary judgment to a trial on the merits, and thus we consider the amended order and opinion the final judgment of the trial court in this matter. . To be clear, we do not hold that Measure 26 is facially constitutional. Rather, as discussed infra, we hold that because the district court did not clearly err in determining plaintiffs failed to establish that Measure 26 significantly diminishes the pool of potential petition circulators, increases the cost of signature gathering, or increases the invalidity rate of signatures gathered, we cannot conclude that Measure 26 imposes a "severe burden” under the First Amendment. Because plaintiffs have established only a "lesser burden,” and defendant has offered "an important regulatory interest” in preventing fraud, we conclude the district court did not err in upholding the constitutionality of Measure 26 as applied. We express no opinion, however, regarding whether Measure 26 could withstand strict scrutiny had plaintiffs proven the measure imposed a "severe burden” under the First Amendment. See Buckley v. Am. Constitutional Law Found., 525 U.S. 182, 192 , 119 S.Ct. 636 , 142 L.Ed.2d 599 (1999) (requiring plaintiffs to establish the challenged restrictions resulted in a significant decrease in the available pool of petition cir-culators to support a finding of a "severe burden”). . This court reviews for abuse of discretion a district court's ruling on a motion for permissive intervention pursuant to Fed.R.Civ.P. 24(b)(2). Kootenai Tribe of Idaho v. Veneman, 313 F.3d 1094, 1110 (9th Cir.2002). Although intervenor-defendants brought a motion in the alternative for permissive intervention, the district court did not rule on that motion because it granted the motion to intervene as of right. . Fed.R.Civ.P. 24(a)(2) provides in part: Upon timely application anyone shall be permitted to intervene in an action ... when the applicant claims an interest relating to the property or transaction which is the subject of the action and the applicant is so situated that the disposition of the action may as a practical matter impair or impede the applicant’s ability to protect that interest, unless the applicant's interest is adequately represented by existing parties. . There is some question., however, whether an intervenor-applicant must independently establish Article III standing to intervene as of right. For example, in the case at hand, Article III standing is satisfied between plaintiffs and defendant. But a circuit split exists whether an intervenor-applicant must also independently satisfy Article III standing to intervene as of right. Compare Planned Parenthood of Mid-Missouri & Eastern Kansas, Inc. v. Ehlmann, 137 F.3d 573, 576-77 (8th Cir.1998) (requiring independent inlervenor standing) and Building & Const. Trades Dep’t v. Reich, 40 F.3d 1275, 1282 (D.C.Cir.1994) (same), with Associated Builders & Contractors v. Perry, 16 F.3d 688, 690 (6th Cir.1994) (no independent intervenor standing required), and U.S. Postal Serv. v. Brennan, 579 F.2d 188 , 190 (2d Cir.1978) (same). The U.S. Supreme Court has not yet settled the issue. See 7C Charles Alan Wright et al., Federal Practice and Procedure § 1908 (2d ed.2005). This court also has not definitively ruled on the issue. Although some sources (such as Federal Practice and Procedure) cite Yniguez v. Arizona, 939 F.2d 727 (9th Cir.1991), for the proposition that this court does not require independent Article III standing for in-tervenors, id. at 731 , that' opinion was vacated by the U.S. Supreme Court. See AOE, 520 U.S. at 80, 117 S.Ct. 1055 ; League of United Latin Am. Citizens v. Wilson, 131 F.3d 1297 , 1305 n. 5 (9th Cir.1997) (noting Yniguez was vacated by the U.S. Supreme Court and "is thus wholly without precedential authority"). Regardless, we need not reach this issue because, as discussed infra, the district court erred in granting intervenor-defendants’ motion to intervene on grounds other than whether intervenor-defendants had independent standing. . Further, the court distinguished Sagebrush Rebellion, noting that in that case the defendant, Secretary of the Interior Watt, had previously served as the head of the foundation representing the plaintiff. Id. at 1305. In League of United Latin Am. Citizens, however, this court noted that the defendant had vigorously defended Proposition 187, and there was no evidence the defendant would cease to do so in the future. Id. . We do not hold that budgetary constraints that impact the ability of the government to adequately litigate its position can never support a motion to intervene as of right by a citizen of that government. Rather, we hold only that absent any evidence of Oregon’s alleged budgetary constraints and the impact of said constraints on this litigation, interve-nor-defendants failed to meet their burden to present a compelling showing of inadequate representation. . Intervenor-defendants also argue that Measure 26 protects their interest in "preventing fraudulent signatures from being gathered for initiatives that are contrary to their union interests,” and defendant may not defend that interest. We will read this claim as evincing an interest in preventing fraudulent signatures on all petitions, not just those petitions which are "contrary to their union interests.” We are left ignorant of what constitutes intervenors' union interests, if any, apart from preventing fraudulent signatures on petitions. Absent any basis for determining there are "union interests” separate and distinct from the prevention of fraudulent signatures, we see no basis for intervention on this score. . To be precise, the Oregon Secretary of State does not count every signature on the petitions. Instead, it employs a statistical sampling technique to determine whether the petitions contain the requisite number of signatures to support certification of the initiative for the ballot. Or.Rev.Stat. § 250.105(4). . While we recognize intervenor-defendants may have greater first-hand knowledge than the Secretary of State regarding the impact of Measure 26 on petition circulation, it will often be the case that a private party has greater first hand knowledge of the impact of legislation oh private individuals than the government. Such knowledge may support a trial judge's discretionary grant of permissive intervention, but it is not sufficient by itself to support intervention as of right in this case. See Garza v. County of Los Angeles, 918 F.2d 763, 777 (9th Cir.1990) (Unlike intervention as of right, "[t]he decision to grant or denyfpermissive] intervention is discretionary, subject to considerations of equity and judicial economy.” Therefore, while an interest may not be "sufficiently weighty to warrant intervention as of right, the court may nevertheless consider eligibility for permissive intervention under Fed.R.Civ.Pro. 24(b)(2).”). In this case intervenor-defendants failed to present any evidence that the Oregon government could not have obtained any knowledge it lacked through the use of discovery and expert testimony. Absent such evidence, in-tervenor-defendants failed to make a compelling showing of inadequate representation sufficient to support intervention as of right. . The reason for this disparity is straightforward: the denial of a motion to intervene is a final order and is thus immediately appeal-able. See Stringfellow v. Concerned Neighbors in Action, 480 U.S. 370, 375-76 , 107 S.Ct. 1177 , 94 L.Ed.2d 389 (1987). Yet the grant of a motion to intervene is not a final order and is not appealable until after final judgment. Id. at 379-80 , 107 S.Ct. 1177 . . See Mothersill D.I.S.C. Corp. v. Petroleos Mexicanos, S.A., 831 F.2d 59, 60-63 (5th Cir.1987) (the plaintiff brought a contracts action against the defendant, and the parties agreed to settle; a corporate employee moved to intervene as of right, claiming he was due payment for services rendered under the contracts; the district court granted the motion to intervene and later entered judgment for the intervenor, awarding him payment for past services under the contracts; the Fifth Circuit reversed, holding intervention was improper and vacating the judgment in favor of the intervenor.); Stockton v. United States, 493 F.2d 1021, 1022-24 (9th Cir.1974) (the plaintiff sought a tax refund from the defendant, and the district court entered judgment for the plaintiff; the plaintiff's attorney then moved to intervene as of right, claiming an interest in his attorneys fees (to be paid from the refund recovery); the district court granted the motion to intervene and granted judgment for the plaintiff's attorney; this court reversed, holding intervention was improper and vacating judgment for the intervenor). . “On the hearing of any appeal or writ of certiorari in any case, the court shall give judgment after an examination of the record without regard to errors or defects which do not affect the substantial rights of the parties.” 28 U.S.C. § 2111 . . No error in either the admission or exclusion of evidence and no error or defect in any ruling or order or in anything done or omitted by the court or by any of the parties is ground for granting a new trial or for setting aside a verdict or for vacating, modifying, or otherwise disturbing a judgment or order, unless refusal to take such action appears to the court inconsistent with substantial justice. The court at every stage of the proceeding must disregard any error or defect in the proceeding which does not affect the substantial rights of the parties. Fed.R.Civ.P. 61. . In addition, had the district court denied Nesbit and the Oregon AFL-CIO’s motion to intervene, defendant could have offered the same evidence in cooperation with Nesbitt and the Oregon AFL-CIO. Moreover, Nesbitt and the Oregon AFL-CIO could have presented argument as amicus rather than as a full-fledged party. There is nothing in the record to suggest that the status of Nesbitt and the Oregon AFL-CIO as intervenors, rather than as amici, materially affected plaintiffs' pretrial preparation, discovery, trial tactics (such as its motion for summary judgment) or the case as a whole. For instance, nothing indicates that intervenor-defendants paid defendant’s litigation expenses conditioned on in-tervenors procuring intervenor status. . Fed.R.Civ.P. 65(a)(2) provides in part: "Before or after the commencement of the hearing of an application for a preliminary injunction, the court may order the trial of the action on the merits to be advanced and consolidated with the hearing of the application.” . The Voter Education Project ("VEP”) bills itself as an "educational watchdog organization” organized to protect the integrity of the initiative system. It researches potential fraud and forgery committed by paid signature gatherers. It makes direct contact with voters through its "Think Before You Ink” program, which informs voters that petition circulators paid by the signature use devious tactics and commit fraud and forgery during the collection of signatures. The record also includes a news article that VEP also engages in harassment of paid petition circulators, "yelling at them, threatening them, even following them in cars.” The article also states VEP has links to the AFL-CIO. Other than the news article, plaintiffs presented no evidence that the VEP (or the Oregon AFL-CIO) supported Measure 26 to weaken in some manner the initiative process in Oregon. . In making that finding, the district court also looked to an affidavit by Ted Blaszak of Democracy Resources of Oregon, Inc., a signature gathering firm. The Blaszak affidavit was submitted by intervenor-defendants. Blaszak had run the signature gathering campaign for Measure 26, during which the petition circulators were paid by the hour, not per-signature. Blaszak averred that he had worked on about ten other initiative campaigns in Oregon, and "the requirement to pay employees by the hour rather than by the signature has not significantly increased my costs or decreased productivity.” As noted in Section I, although the district court erred in granting intervenor-defendants' motion to intervene, that error was harmless. The Blaszak affidavit supports defendant's position that Measure 26 does not increase the costs of initiative petition circulation, yet the district court largely found that plaintiffs, through their own offer of proof, did not prove that Measure 26 would impose such a burden. According to its amended opinion and order, the district court did not rely upon any other evidence submitted by intervenor-defendants. Thus, the district court’s consideration of the Blaszak affidavit does not make it more probable than not that the district court’s error in granting intervention tainted the verdict. Furthermore, there is no evidence that defendant would not have obtained and submitted the Blaszak affidavit if Nesbitt and the Oregon AFL-CIO had not intervened. . Whether the proven barriers, if any, constitute a severe or lesser burden is a question of constitutional fact reviewed de novo. See Planned Parenthood, 290 F.3d at 1070. . "To be clearly erroneous, a decision must strike us as more than just maybe or probably wrong; it must ... strike us as wrong with the force of a five-week-old, unrefrigerated dead fish." Hayes v. Woodford, 301 F.3d 1054 , 1067 n. 8 (9th Cir.2002) (citing Fisher v. Roe, 263 F.3d 906, 912 (9th Cir.2001) (quoting Parts & Elec. Motors, Inc. v. Sterling Elec., Inc., 866 F.2d 228, 233 (7th Cir.1988))). . "The 'principal inquiry' in determining whether a regulation is content-neutral or content-based 'is whether the government has adopted the regulation because of agreement or disagreement with the message it conveys.' " Crawford v. Lungren, 96 F.3d 380, 384 (9th Cir.1996) (quoting Turner Broadcasting Sys., Inc. v. FCC, 512 U.S. 622, 642 , 114 S.Ct. 2445 , 129 L.Ed.2d 497 (1994)) (internal alterations omitted). "[Ljaws that by their terms distinguish favored speech from disfavored speech on the basis of the ideas or views expressed are content-based.” Id. Here, Measure 26 does not regulate what can be said in an initiative or referendum petition, nor does it adopt or reject any particular subject that can be raised in a petition. It may be argued that a restriction on the initiative process itself, which is a means to wrest power from the legislature, is inherently content-based. This argument must fail because in Oregon the initiative power may be used to amend the constitution to grant additional power to the legislature. See Or. Const. Art. IV § 1. See also Stranahan v. Fred Meyer Inc., 331 Or. 38 , 11 P.3d 228, 242 (2000) ("In sum, the case law demonstrates that Article IV, section 1, confers an unfettered right to propose laws and constitutional amendments by initiative petition, and to approve or reject such proposed laws or amendments through the voting process.”); Bernstein Bros. Inc. v. Dept. of Revenue, 294 Or. 614 , 661 P.2d 537, 539 (1983) ("The power to invoke a referendum is a constitutional power reserved by the people. The creation of the referendum power (along with the initiative power) changes the allocation of legislative power within a state, because after this creation the legislative power is shared between the people and their representatives.”); Zilesch et al. v. Polk County et al., 107 Or. 659 , 215 P. 578, 582 (1923) ("[T]he legislature and the people, through the initiative or referendum, [are] coordinate legislative bodies, and[ ] either [may] independently repeal an act passed by the other.... ”). .Plaintiffs alternatively contend that Measure 26 is subject to strict scrutiny because it is content-based, in that it applies only to initiative and referendum petitions, not to recall or candidate sponsorship petitions. See Bayless, 320 F.3d at 1009 (noting strict scrutiny automatically applies to content-based restrictions). The district court rejected this argument, noting "the content of an initiative petition itself is not restricted, regulated or otherwise affected by Measure 26.” We agree with the district court. "The 'principal inquiry’ in determining whether a regulation is content-neutral or content-based 'is whether the government has adopted the regulation because of agreement or disagreement with the message it conveys.’ ” Crawford, 96 F.3d at 384 (quoting Turner Broadcasting Sys., Inc. v. FCC, 512 U.S. 622, 642 , 114 S.Ct. 2445 , 129 L.Ed.2d 497 (1994)) (internal alterations omitted). "[Ljaws that by their terms distinguish favored speech from disfavored speech on the basis of the ideas or views expressed are content-based.” Id. Measure 26 does not regulate what can be said in an initiative or referendum petition, nor does it adopt or reject any particular subject that can be raised in a petition. Therefore, Measure 26 is not a content-based restriction and strict scrutiny does not apply. . Petition circulators must certify that the signatures on the petitions were obtained in the presence of the circulator and that upon belief, each signature is that of a registered Oregon voter. Or. Rev. St. § 250.045(7). It is unlawful to make a false certification. Or. Rev. St. § 260.715(1). . Or. Rev. St. § 260.558(2) makes it unlawful "to sell, offer to sell, purchase or offer to purchase, for money or other valuable consideration, any signature sheet of an initiative, referendum or recall petition or any other portion of the petition used to gather signatures.” . Or. Rev. St. § 260.555(3)-(4) makes it unlawful to obtain a signature on an initiative "knowing that the person signing the petition is not qualified to sign it”; or that the person has already signed the petition once. . Plaintiffs also rely on four district court cases. Those cases are distinguishable, however, because in each case the state defending the prohibition on per-signature payment for petition circulators failed to present any evidence that per-signature payments increased fraud. Hence, in those cases, the states presented no evidence to support their assertions that a per-signature ban was necessary to promote the state interest in preventing fraud and forgery in the initiative process. See Idaho Coalition United for Bears v. Cenarrusa, 234 F.Supp.2d 1159, 1165-66 (D.Idaho 2001) (the plaintiffs challenged an Idaho prohibition making it a felony to "offer ... or attempt to sell ... any petition or any part thereof or of any signatures” for initiative petitions; the district court granted summary judgment for the plaintiffs; construing the prohibition to prohibit payment of petition circulators per signature, the court found Idaho presented no evidence of fraud in the signature gathering process and thus struck down the prohibition as violating the First Amendment); On Our Terms '97 PAC v. Sec'y of State of Maine, 101 F.Supp.2d 19, 25-26 (D.Me.1999) (the plaintiffs challenged Maine's prohibition on paying petition circulators per signature; following a bench trial, the district court ruled for the plaintiffs, finding the prohibition burdened the signature gathering process but noting that Maine provided "no evidence whatsoever that fraud is more pervasive among circula-tors paid per signature, or even that fraud in general has been a noteworthy problem in the lengthy history of the Maine initiative and referendum process.”); Terms Limits Leadership Council, Inc. v. Clark, 984 F.Supp. 470, 471 (S.D.Miss.1997) (the plaintiffs challenged a Mississippi prohibition on paying petition circulators per signature; the district court granted summary judgment for plaintiffs, finding "plaintiffs have shown that the[] statute[ ] burden[s] their right to political expression, [and] the State has failed to present evidence of fraud or actual threat to its citizens' confidence in government on account of the per-signature payment of petition circula-tors.”); LIMIT v. Maleng, 874 F.Supp. 1138, 1140-41 (W.D.Wash.1994) (the plaintiffs challenged Washington's prohibition on paying petition circulators on a per signature basis; the district court granted summary judgment for the plaintiffs, finding Washington presented "no actual proof of fraud stemming specifically from the payment per signature method of collection,” and thus Washington's unsupported interest in maintaining the integrity of the initiative process did not outweigh the burdens imposed by the prohibition). Here, we have the affidavits of Carroll and Lind-back, infra p. 969 - 970.