concurring in part and dissenting in part.
In Part III of its opinion, the Court points out that the law ordinarily treats municipalities as creatures of the State. See Reynolds v. Sims, 377 U. S. 533, 575 (1964). Hence the fact that a state statute, rather than a municipal ordinance, limits the use of the municipality’s payroll deduction system is beside the point. I agree that this is so, and I agree with Justice Souter’s discussion about the relationship between the State and the municipality.
I do not agree, however, with the Court’s further analysis of the pertinent legal question — whether the state statute violates the First Amendment. Nor do I agree with its ultimate conclusion. Rather, in my view, we should remand this case for further consideration.
The Court’s First Amendment analysis emphasizes its characterization of the statute as not “abridging” a union’s or a worker’s “freedom of speech,” but rather “declin[ing] to promote” that speech. Ante, at 355 (internal quotation marks omitted). I agree that the First Amendment does not prohibit government from “declining to promote” speech. It says that government shall not “abridg[e] the freedom of *366speech.” (Emphasis added.) But I do not think the distinction particularly useful in this case.
That is because here the distinction is neither easy to draw nor likely to prove determinative. Sometimes, I concede, the distinction may help. Were there no payroll deduction system at all and were the unions arguing for the creation of such a system from scratch, one might characterize their claim as seeking the promotion of speech. But that is not the situation here. A deduction system already exists. The unions attack a separate statutory provision that removes politically related deductions from that system. And linguistically speaking, one need not characterize such an attack as (1) seeking speech promotion rather than (2) seeking to prevent an abridgment of political-speech-related activity that otherwise (i e., in the absence of the exception) would occur. In such an instance, the debate over characterization is more metaphysical than practical.
More importantly, the characterization quite possibly does not matter. Suppose, for example, a somewhat similar statutory exception picks and chooses among political causes, prohibiting deductions that help one political party while permitting deductions that help another. The First Amendment result could not turn upon whether one described the exception as an “abridgment” or a “promotion” failure. And, as I shall explain, infra, at 369-370, such may be the case here.
I disagree with the Court’s characterizations in another respect. The Court says that because the exception “has not infringed the unions’ First Amendment rights,” “strict scrutiny” does not apply and, thus, the State “need only demonstrate a rational basis” — the standard of review applicable to any ordinary legislation that does not infringe fundamental rights — “to justify the ban on political payroll deductions.” Ante, at 359 (emphasis added). I agree that the exception does not call for “strict scrutiny” — a categorization that almost always proves fatal to the law in question. *367After all, the exception does not restrict the content of the unions’ speech, impose a prior restraint on that speech, or ban union speech on political issues altogether.
But I disagree with the Court in that I believe there is a First Amendment interest at stake. The exception affects speech, albeit indirectly, by restricting a channel through which speech-supporting finance might flow. As a result, the alternative to “strict scrutiny” is not necessarily a form of “rational basis” review — a test that almost every restriction will pass. And instead of applying either “strict scrutiny” or “rational basis” review to the statutory exception, I would ask the question that this Court has asked in other speech-related contexts, namely, whether the statute imposes a burden upon speech that is disproportionate in light of the other interests the government seeks to achieve. See Burdick v. Takushi, 504 U. S. 428, 433-434 (1992) (election regulation); Nixon v. Shrink Missouri Government PAC, 528 U. S. 377, 403 (2000) (Breyer, J., concurring) (collecting cases); see also, e. g., Thompson v. Western States Medical Center, 535 U. S. 357, 388 (2002) (Breyer, J., dissenting) (discussing the Court’s application of this approach in the commercial speech context); Denver Area Ed. Telecommunications Consortium, Inc. v. FCC, 518 U. S. 727, 740-747 (1996) (plurality opinion) (cable programming regulation); Pickering v. Board of Ed. of Township High School Dist. 205, Will Cty., 391 U. S. 563, 568 (1968) (government employee speech). Constitutional courts in other nations also have used similar approaches when facing somewhat similar problems. See, e. g., Libman v. Quebec (Attorney General), [1997] 3 S. C. R. 569 (Canada) (applying proportionality in the campaign finance context); Bowman v. United Kingdom, 26 Eur. Ct. H. R. 1 (1998) (same); Midi Television (Pty) Ltd v. Director of Public Prosecutions 2007 SCA 56 (S. Afr.) (applying proportionality in the freedom of press context); Bakri v. Israel Film Council, HCJ 316/303 (Isr. Sup. Ct. 2003) (applying proportionality in the freedom of expression context).
*368In these cases the Court has sought to determine whether the harm to speech-related interests is disproportionate in light of the degree of harm, justifications, and potential alternatives. In doing so, it has considered the seriousness of the speech-related harm the provision will likely cause, the importance of the provision’s countervailing objectives, the extent to which the statute will tend to achieve those objectives, and whether there are other less restrictive ways of doing so. In light of these considerations, it has determined whether ultimately the statute works speech-related harm that is out of proportion to its justifications. See Board of Trustees of State Univ. of N. Y. v. Fox, 492 U. S. 469, 480 (1989) (describing need for a “fit” between legislative ends and means “whose scope is ‘in proportion to the interest served’”); United States v. American Library Assn., Inc., 539 U. S. 194, 217-218 (2003) (Breyer, J., concurring in judgment).
Where context calls for “strict scrutiny,” one would not necessarily ask these proportionality questions; but I would ask them in other contexts calling for less than ordinary legislative leeway in light of the fact that constitutionally protected expression is at issue. See id., at 218. To do so, in my view, helps structure what the Court sometimes calls an “intermediate scrutiny” inquiry. See Turner Broadcasting System, Inc. v. FCC, 512 U. S. 622 (1994).
Applying this analysis here, I would find the statutory exception constitutional, but only if I were convinced that the exception applied evenhandedly among similar politically related contributions. If so, the provision would still negatively affect speech-related interests, for it would close off one channel through which individuals might provide speech-enabling funds to political institutions. But, as the majority points out, many other channels for those funds exist, and the State has a strong interest in “avoiding the reality or appearance of government favoritism or entanglement with partisan politics,” ante, at 359. I would conse*369quently find the restriction justified as proportionately serving a legitimate, important governmental need. Cf. Fox, supra, at 480.
It is not clear, however, whether the particular exception before us does, in fact, operate evenhandedly. To read the statute without more, I concede, suggests evenhandedness. The provision says that “[deductions for political activities as defined in chapter 26, title JM,, Idaho Code, shall not be deducted from the wages, earnings or compensation of an employee.” Idaho Code §44-2004(2) (Michie 2003) (emphasis added). And chapter 26, title 44, Idaho Code, defines “political activities” without special reference to labor organizations. See § 44-2602(l)(e).
Nonetheless, certain features of the provision suggest it may affect some politically related deductions, namely, labor-related deductions, but not others. Title 44 of the Idaho Code — entitled “Labor” — is about labor activities. And the ban on payroll deductions for political activities was enacted as part of a statute in which every other provision is concerned solely with union activities. See Voluntary Contributions Act, 2003 Idaho Sess. Laws chs. 97 and 340 (codified at Idaho Code §§44-2601 through 44-2605 and §44-2004). At the same time, the provision containing the payroll deduction ban is immediately followed by another related provision that expressly mentions labor unions. See §44-2004(3) (“Nothing in this chapter shall prohibit an employee from personally paying contributions for political activities ... to a labor organization unless such payment is prohibited by law” (emphasis added)).
It is important to know whether the exception concerns only labor-related political deductions (while allowing other similar deductions) or treats all alike. A restriction that applies to the political activities of unions alone would seem unlikely to further the government’s justifying objective, namely, providing the appearance of political neutrality. And in that case, the provision could well bring about *370speech-related harm that is disproportionate to the statute’s tendency to further the government’s “neutrality” objective.
Because the Court of Appeals analyzed the issue as if the State “regulated” its municipalities (as government might regulate a private firm), it did not resolve the questions I have just described. I would remand the case so that it can decide whether the parties appropriately raised those matters and, if so, consider them. Accordingly, I would vacate the Court of Appeals’ judgment and remand the case.