delivered the opinion of the Court.
The question presented in this case is whether the First Amendment permits the Minnesota Supreme Court to prohibit candidates for judicial election in that State from announcing their views on disputed legal and political issues.
h — I
Since Minnesota’s admission to the Union in 1858, the State’s Constitution has provided for the selection of all state judges by popular election. Minn. Const., Art. VI, §7. Since 1912, those elections have been nonpartisan. Act of June 19, ch. 2,1912 Minn. Laws Special Sess., pp. 4-6. Since 1974, they have been subject to a legal restriction which states that a “candidate for a judicial office, including an incumbent judge,” shall not “announce his or her views on disputed legal or political issues.” Minn. Code of Judicial Conduct, Canon 5(A)(3)(d)(i) (2000). This prohibition, promulgated by the Minnesota Supreme Court and based on Canon 7(B) of the 1972 American Bar Association (ABA) Model Code of Judicial Conduct, is known as the “announce clause.” Incumbent judges who violate it are subject to discipline, including removal, censure, civil penalties, and suspension without pay. Minn. Rules of Board on Judicial Standards 4(a)(6), 11(d) (2002). Lawyers who run for judicial office also must comply with the announce clause. Minn. Rule of Professional Conduct 8.2(b) (2002) (“A lawyer who is a candidate for judicial office shall comply with the applicable provisions of the Code of Judicial Conduct”). Those who violate it are subject to, inter alia, disbarment, suspension, and probation. Rule 8.4(a); Minn. Rules on Lawyers Professional Responsibility 8-14, 15(a) (2002).
In 1996, one of the petitioners, Gregory Wersal, ran for associate justice of the Minnesota Supreme Court. In the course of the campaign, he distributed literature criticizing several Minnesota Supreme Court decisions on issues such as crime, welfare, and abortion. A complaint against Wersal *769challenging, among other things, the propriety of this literature was filed with the Office of Lawyers Professional Responsibility, the agency which, under the direction of the Minnesota Lawyers Professional Responsibility Board,1 investigates and prosecutes ethical violations of lawyer candidates for judicial office. The Lawyers Board dismissed the complaint; with regard to the charges that his campaign materials violated the announce clause, it expressed doubt whether the clause could constitutionally be enforced. Nonetheless, fearing that further ethical complaints would jeopardize his ability to practice law, Wersal withdrew from the election. In 1998, Wersal ran again for the same office. Early in that race, he sought an advisory opinion from the Lawyers Board with regard to whether it planned to enforce the announce clause. The Lawyers Board responded equivocally, stating that", although it had significant doubts about the constitutionality of the provision, it was unable to answer his question because he had not submitted a list of the announcements he wished to make.2
Shortly thereafter, Wersal filed this lawsuit in Federal District Court against respondents,3 seeking, inter alia, a *770declaration that the announce clause violates the First Amendment and an injunction against its enforcement. Wersal alleged that he was forced to refrain from announcing his views on disputed issues during the 1998 campaign, to the point where he declined response to questions put to him by the press and public, out of concern that he might run afoul of the announce clause. Other plaintiffs in the suit, including the Minnesota Republican Party, alleged that, because the clause kept Wersal from announcing his views, they were unable to learn those views and support or oppose his candidacy accordingly. The parties filed cross-motions for summary judgment, and the District Court found in favor of respondents, holding that the announce clause did not violate the First Amendment. 63 F. Supp. 2d 967 (Minn. 1999). Over a dissent by Judge Beam, the United States Court of Appeals for the Eighth Circuit affirmed. Republican Party of Minn. v. Kelly, 247 F. 3d 854 (2001). We granted certio-rari. 534 U. S. 1054 (2001).
II
Before considering the constitutionality of the announce clause, we must be clear about its meaning. Its text says that a candidate for judicial office shall not “announce his or her views on disputed legal or political issues.” Minn. Code of Judicial Conduct, Canon 5(A)(3)(d)(i) (2002).
We know that “announcing]... views” on an issue covers much more than promising to decide an issue a particular way. The prohibition extends to the candidate’s mere statement of his current position, even if he does not bind himself to maintain that position after election. All the parties agree this is the case, because the Minnesota Code contains a so-called “pledges or promises” clause, which separately prohibits judicial candidates from making “pledges or promises of conduct in office other than the faithful and impartial performance of the duties of the office,” ibid. — a prohibition that is not challenged here and on which we express no view.
*771There are, however, some limitations that the Minnesota Supreme Court has placed upon the scope of the announce clause that are not (to put it politely) immediately apparent from its text. The statements that formed the basis of the complaint against Wersal in 1996 included criticism of past decisions of the Minnesota Supreme Court. One piece of campaign literature stated that “[t]he Minnesota Supreme Court has issued decisions which are marked by their disregard for the Legislature and a lack of common sense.” App. 37. It went on to criticize a decision excluding from evidence confessions by criminal defendants that were not tape-recorded, asking “[s]hould we conclude that because the Supreme Court does not trust police, it allows confessed criminals to go free?” Ibid. It criticized a decision striking down a state law restricting welfare benefits, asserting that “[fit’s the Legislature which should set our spending policies.” Ibid. And it criticized a decision requiring public financing of abortions for poor women as “unprecedented” and a “pro-abortion stance.” Id., at 38. Although one would think that all of these statements touched on disputed legal or political issues, they did not (or at least do not now) fall within the scope of the announce clause. The Judicial Board issued an opinion stating that judicial candidates may criticize past decisions, and the Lawyers Board refused to discipline Wersal for the foregoing statements because, in part, it thought they did not violate the announce clause. The Eighth Circuit relied on the Judicial Board’s opinion in upholding the announce clause, 247 F. 3d, at 882, and the Minnesota Supreme Court recently embraced the Eighth Circuit’s interpretation, In re Code of Judicial Conduct, 639 N. W. 2d 55 (2002).
There are yet further limitations upon the apparent plain meaning of the announce clause: In light of the constitutional concerns, the District Court construed the clause to reach only disputed issues that are likely to come before the candidate if he is elected judge. 63 F. Supp. 2d, at 986. The *772Eighth Circuit accepted this limiting interpretation by the District Court, and in addition construed the clause to allow general discussions of case law and judicial philosophy. 247 F. 3d, at 881-882. The Supreme Court of Minnesota adopted these interpretations as well when it ordered enforcement of the announce clause in accordance with the Eighth Circuit’s opinion. In re Code of Judicial Conduct, supra.
It seems to us, however, that — like the text of the announce clause itself — these limitations upon the text of the announce clause are not all that they appear to be. First, respondents acknowledged at oral argument that statements critical of past judicial decisions are not permissible if the candidate also states that he is against stare decisis. Tr. of Oral Arg. 33-34.4 Thus, candidates must choose between stating their views critical of past decisions and stating their views in opposition to stare decisis. Or, to look at it more concretely, they may state their view that prior decisions were erroneous only if they do not assert that they, if elected, have any power to eliminate erroneous decisions. Second, limiting the scope of the clause to issues likely to come before a court is not much of a limitation at all. One would hardly expect the “disputed legal or political issues” raised in the course of a state judicial election to include such matters as whether the Federal Government should end the embargo of Cuba. Quite obviously, they will be those legal or political disputes that are the proper (or by past decisions have been made the improper) business of the state courts. And within that relevant category, “[t]here is almost no legal or political issue that is unlikely to come before a judge of an American court, state or federal, of general jurisdiction.” *773Buckley v. Illinois Judicial Inquiry Bd., 997 F. 2d 224, 229 (CA7 1993). Third, construing the clause to allow “general” discussions of case law and judicial philosophy turns out to be of little help in an election campaign. At oral argument, respondents gave, as an example of this exception, that a candidate is free to assert that he is a “ ‘strict constructionist.’” Tr. of Oral Arg. 29. But that, like most other philosophical generalities, has little meaningful content for the electorate unless it is exemplified by application to a particular issue of construction likely to come before a court — for example, whether a particular statute runs afoul of any provision of the Constitution. Respondents conceded that the announce clause would prohibit the candidate from exemplifying his philosophy in this fashion. Id., at 43. Without such application to real-life issues, all candidates can claim to be “strict constructionists” with equal (and unhelpful) plausibility.
In any event, it is clear that the announce clause prohibits a judicial candidate from stating his views on any specific nonfanciful legal question within the province of the court for which he is running, except in the context of discussing past decisions — and in the latter context as well, if he expresses the view that he is not bound by stare decisis.5
*774Respondents contend that this still leaves plenty of topics for discussion on the campaign trail. These include a candidate’s “character,” “education,” “work habits,” and “how [he] would handle administrative duties if elected.” Brief for Respondents 35-36. Indeed, the Judicial Board has printed a list of preapproved questions which judicial candidates are allowed to answer. These include how the candidate feels about cameras in the courtroom, how he would go about reducing the caseload, how the costs of judicial administration can be reduced, and how he proposes to ensure that minorities and women are treated more fairly by the court system. Minnesota State Bar Association Judicial Elections Task Force Report & Recommendations, App. C (June 19, 1997), reprinted at App. 97-103. Whether this list of preapproved subjects, and other topics not prohibited by the announce clause, adequately fulfill the First Amendment’s guarantee of freedom of speech is the question to which we now turn.
III
As the Court of Appeals recognized, the announce clause both prohibits speech on the basis of its content and burdens a category of speech that is “at the core of our First Amendment freedoms” — speech about the qualifications of candidates for public office. 247 F. 3d, at 861, 863. The Court of Appeals concluded that the proper test to be applied to determine the constitutionality of such a restriction is what our cases have called strict scrutiny, id., at 864; the parties do not dispute that this is correct. Under the strict-scrutiny test, respondents have the burden to prove that the an*775nounce clause is (1) narrowly tailored, to serve (2) a compelling state interest. E. g., Eu v. San Francisco County Democratic Central Comm., 489 U. S. 214, 222 (1989). In order for respondents to show that the announce clause is narrowly tailored, they must demonstrate that it does not “unnecessarily circumscrib[e] protected expression.” Brown v. Hartlage, 456 U. S. 45, 54 (1982).
The Court of Appeals concluded that respondents had established two interests as sufficiently compelling to justify the announce clause: preserving the impartiality of the state judiciary and preserving the appearance of the impartiality of the state judiciary. 247 F. 3d, at 867. Respondents reassert these two interests before us, arguing that the first is compelling because it protects the due process rights of litigants, and that the second is compelling because it preserves public confidence in the judiciary.6 Respondents are rather vague, however, about what they mean by “impartiality.” Indeed, although the term is used throughout the Eighth Circuit’s opinion, the briefs, the Minnesota Code of Judicial Conduct, and the ABA Codes of Judicial Conduct, none of these sources bothers to define it. Clarity on this point is essential before we can decide whether impartiality is indeed a compelling state interest, and, if so, whether the announce clause is narrowly tailored to achieve it.
A
One meaning of “impartiality” in the judicial context — and of course its root meaning — is the lack of bias for or against either party to the proceeding. Impartiality in this sense *776assures equal application of the law. That is, it guarantees a party that the judge who hears his case will apply the law to him in the same way he applies it to any other party. This is the traditional sense in which the term is used. See Webster’s New International Dictionary 1247 (2d ed. 1950) (defining “impartial” as “[n]ot partial; esp., not favoring one more than another; treating all alike; unbiased; equitable; fair; just”). It is also the sense in which it is used in the cases cited by respondents and amici for the proposition that an impartial judge is essential to due process. Tumey v. Ohio, 273 U. S. 510, 523, 531-534 (1927) (judge violated due process by sitting in a case in which it would be in his financial interest to find against one of the parties); Aetna Life Ins. Co. v. Lavoie, 475 U. S. 813, 822-825 (1986) (same); Ward v. Monroeville, 409 U. S. 57, 58-62 (1972) (same); Johnson v. Mississippi, 403 U. S. 212, 215-216 (1971) (per curiam) (judge violated due process by sitting in a case in which one of the parties was a previously successful litigant against him); Bracy v. Gramley, 520 U. S. 899, 905 (1997) (would violate due process if a judge was disposed to rule against defendants who did not bribe him in order to cover up the fact that he regularly ruled in favor of defendants who did bribe him); In re Murchison, 349 U. S. 133, 137-139 (1955) (judge violated due process by sitting in the criminal trial of defendant whom he had indicted).
We think it plain that the announce clause is not narrowly tailored to serve impartiality (or the appearance of impartiality) in this sense. Indeed, the clause is barely tailored to serve that interest at all, inasmuch as it does not restrict speech for or against particular parties, but rather speech for or against particular issues. To be sure, when a case arises that turns on a legal issue on which the judge (as a candidate) had taken a particular stand, the party taking the opposite stand is likely to lose. But not because of any bias against that party, or favoritism toward the other party. *777Any party taking that position is just as likely to lose. The judge is applying the law (as he sees it) evenhandedly.7
B
It is perhaps possible to use the term “impartiality” in the judicial context (though this is certainly not a common usage) to mean lack of preconception in favor of or against a particular legal view. This sort of impartiality would be concerned, not with guaranteeing litigants equal application of the law, but rather with guaranteeing them an equal chance to persuade the court on the legal points in their case. Impartiality in this sense may well be an interest served by the announce clause, but it is not a compelling state interest, as strict scrutiny requires. A judge’s lack of predisposition regarding the relevant legal issues in a case has never been thought a necessary component of equal justice, and with good reason. For one thing, it is virtually impossible to find a judge who does not have preconceptions about the1 law. As then-JuSTiCE Rehnquist observed of our own Court: “Since most Justices come to this bench no earlier than their middle years, it would be unusual if they had not by that time formulated at least some tentative notions that would influence them in their interpretation of the sweeping clauses of the Constitution and their interaction with one another. It would be not merely unusual, but extraordinary, if they had *778not at least given opinions as to constitutional issues in their previous legal careers.” Laird v. Tatum, 409 U. S. 824, 885 (1972) (memorandum opinion). Indeed, even if it were possible to select judges who did not have preconceived views on legal issues, it would hardly be desirable to do so. “Proof that a Justice’s mind at the time he joined the Court was a complete tabula rasa in the area of constitutional adjudication would be evidence of lack of qualification, not lack of bias.” Ibid. The Minnesota Constitution positively forbids the selection to courts of general jurisdiction of judges who are impartial in the sense of having no views on the law. Minn. Const., Art. VI, §5 (“Judges of the supreme court, the court of appeals and the district court shall be learned in the law”). Arid since avoiding judicial preconceptions on legal issues is neither possible nor desirable, pretending otherwise by attempting to preserve the “appearance” of that type of impartiality can hardly be a compelling state interest either.
C
A third possible meaning of “impartiality” (again not a common one) might be described as openmindedness. This quality in a judge demands, not that he have no preconceptions on legal issues, but that he be willing to consider views that oppose his preconceptions, and remain open to persuasion, when the issues arise in a pending case. This sort of impartiality seeks to guarantee each litigant, not an equal chance to win the legal points in the case, but at least some chance of doing so. It may well be that impartiality in this sense, and the appearance of it, are desirable in the judiciary, but we need not pursue that inquiry, since we do not believe the Minnesota Supreme Court adopted the announce clause for that purpose.
Respondents argue that the announce clause serves the interest in openmindedness, or at least in the appearance of openmindedness, because it relieves a judge from pressure to rule a certain way in order to maintain consistency with *779statements the judge has previously made. The problem is, however, that statements in election campaigns are such an infinitesimal portion of the public commitments to legal positions that judges (or judges-to-be) undertake, that this object of the prohibition is implausible. Before they arrive on the bench (whether by election or otherwise) judges have often committed themselves on legal issues that they must later rule upon. See, e. g., Laird, supra, at 831-833 (describing Justice Black’s participation in several cases construing and deciding the constitutionality of the Fair Labor Standards Act, even though as a Senator he had been one of its principal authors; and Chief Justice Hughes’s authorship of the opinion overruling Adkins v. Children's Hospital of D. C., 261 U. S. 525 (1923), a case he had criticized in a book written before his appointment to the Court). More common still is a judge’s confronting a legal issue on which he has expressed an opinion while on the bench. Most frequently, of course, that prior expression will have occurred in ruling on an earlier case. But judges often state their views on disputed legal issues outside the context of adjudication — in classes that they conduct, and in books and speeches. Like the ABA Codes of Judicial Conduct, the Minnesota Code not only permits but encourages this. See Minn. Code of Judicial Conduct, Canon 4(B) (2002) (“A judge may write, lecture, teach, speak and participate in other extra-judicial activities concerning the law . . .”); Minn. Code of Judicial Conduct, Canon 4(B), Comment. (2002) (“To the extent that time permits, a judge is encouraged to do so . . .”). That is quite incompatible with the notion that the need for openmindedness (or for the appearance of openmindedness) lies behind the prohibition at issue here.
The short of the matter is this: In Minnesota, a candidate for judicial office may not say “I think it is constitutional for the legislature to prohibit same-sex marriages.” He may say the very same thing, however, up until the very day before he declares himself a candidate, and may say it repeat*780edly (until litigation is pending) after he is elected. As a means of pursuing the objective of openmindedness that respondents now articulate, the announce clause is so woefully underinclusive as to render belief in that purpose a challenge to the credulous. See City of Ladue v. Gilleo, 512 U. S. 43, 52-53 (1994) (noting that underinclusiveness “diminish[es] the credibility of the government’s rationale for restricting speech”); Florida Star v. B. J. F, 491 U. S. 524, 541-542 (1989) (Scalia, J., concurring in judgment) (“[A] law cannot be regarded as protecting an interest of the highest order, and thus as justifying a restriction upon truthful speech, when it leaves appreciable damage to that supposedly vital interest unprohibited” (internal quotation marks and citation omitted)).
Justice Stevens asserts that statements made in an election campaign pose a special threat to openmindedness because the candidate, when elected judge, will have a particular reluctance to contradict them. Post, at 801. That might be plausible, perhaps, with regard to campaign promises. A candidate who says “If elected, I will vote to uphold the legislature’s power to prohibit same-sex marriages” will positively be breaking his word if he does not do so (although one would be naive not to recognize that campaign promises are — by long democratic tradition — the least binding form of human commitment). But, as noted earlier, the Minnesota Supreme Court has adopted a separate prohibition on campaign “pledges or promises,” which is not challenged here. The proposition that judges feel significantly greater compulsion, or appear to feel significantly greater compulsion, to maintain consistency with nonpromissory statements made during a judicial campaign than with such statements made before or after the campaign is not self-evidently true. It seems to us quite likely, in fact, that in many cases the opposite is true. We doubt, for example, that a mere statement of position enunciated during the pendency of an election will be regarded by a judge as more binding — or as more likely *781to subject him to popular disfavor if reconsidered — than a carefully considered holding that the judge set forth in an earlier opinion denying some individual’s claim to justice. In any event, it suffices to say that respondents have not carried the burden imposed by our strict-scrutiny test to establish this proposition (that campaign statements are uniquely destructive of openmindedness) on which the validity of the announce clause rests. See, e. g., Landmark Communications, Inc. v. Virginia, 435 U. S. 829, 841 (1978) (rejecting speech restriction subject to strict scrutiny where the State “offered little more than assertion and conjecture to support its claim that without criminal sanctions the objectives of the statutory scheme would be seriously undermined”); United States v. Playboy Entertainment Group, Inc., 529 U. S. 803, 816-825 (2000) (same).8
Moreover, the notion that the special context of electioneering justifies an abridgment of the right to speak out on disputed issues sets our First Amendment jurisprudence on its head. “[Djebate on the qualifications of candidates” is “at the core of our electoral process and of the First Amendment freedoms,” not at the edges. Eu, 489 U. S., at 222-223 (internal quotation marks omitted). “The role that elected officials play in our society makes it all the more imperative that they be allowed freely to express themselves on matters *782of current public importance.” Wood v. Georgia, 370 U. S. 375, 395 (1962). “It is simply not the function of government to select which issues are worth discussing or debating in the course of a political campaign.” Brown, 456 U. S., at 60 (internal quotation marks and citation omitted). We have never allowed the government to prohibit candidates from communicating relevant information to voters during an election.
Justice Ginsburg would do so — and much of her dissent confirms rather than refutes our conclusion that the purpose behind the announce clause is not openmindedness in the judiciary, but the undermining of judicial elections. She contends that the announce clause must be constitutional because due. process would be denied if an elected judge sat in a case involving an issue on which he had previously announced his view. Post, at 816, 819. She reaches this conclusion because, she says, such a judge would have a “direct, personal, substantial, and pecuniary interest” in ruling consistently with his previously announced view, in order to reduce the risk that he will be “voted off the bench and thereby lose [his] salary and emoluments,” post, at 816 (internal quotation marks and alterations omitted). But elected judges — regardless of whether they have announced any views beforehand — always face the pressure of an electorate who might disagree with their rulings and therefore vote them off the bench. Surely the judge who frees Timothy McVeigh places his job much more at risk than the judge who (horror of horrors!) reconsiders his previously announced view on a disputed legal issue. So if, as Justice Ginsburg claims, it violates due process for a judge to sit in a case in which ruling one way rather than another increases his prospects for reelection, then — quite simply — the practice of electing judges is itself a violation of due process. It is not difficult to understand how one with these views would approve the election-nullifying effect of the announce *783clause.9 They are not, however, the views reflected in the Due Process Clause of the Fourteenth Amendment, which has coexisted with the election of judges ever since it was adopted, see infra, at 785-786.
Justice Ginsburg devotes the rest of her dissent to attacking arguments we do not make. For example, despite the number of pages she dedicates to disproving this proposition, post, at 805-809, we neither assert nor imply that the First Amendment requires campaigns for judicial office to sound the same as those for legislative office.10 What we do assert, and what Justice Ginsburg ignores, is that, even if the First Amendment allows greater regulation of judicial election campaigns than legislative election campaigns, the announce clause still fails strict scrutiny because it is woefully underinclusive, prohibiting announcements by judges (and would-be judges) only at certain times and in certain forms. We rely on the cases involving speech during elections, supra, at 781-782, only to make the obvious point that this underinclusiveness cannot be explained by resort to the notion that the First Amendment provides less protection during an election campaign than at other times.11
*784But in any case, Justice Ginsburg greatly exaggerates the difference between judicial and legislative elections. She asserts that “the rationale underlying unconstrained speech in elections for political office — that representative government depends on the public’s ability to choose agents who will act at its behest — does not carry over to campaigns for the bench.” Post, at 806. This complete separation of the judiciary from the enterprise of “representative government” might have some truth in those countries where judges neither make law themselves nor set aside the laws enacted by the legislature. It is not a true picture of the American system. Not only do state-court judges possess the power to “make” common law, but they have the immense power to shape the States’ constitutions as well. See, e. g., Baker v. State, 170 Vt. 194, 744 A. 2d 864 (1999). Which is precisely why the election of state judges became popular.12
*785IV
To sustain the announce clause, the Eighth Circuit relied heavily on the fact that a pervasive practice of prohibiting judicial candidates from discussing disputed legal and political issues developed during the last half of the 20th century. 247 F. 3d, at 879-880. It is true that a “universal and long-established” tradition of prohibiting certain conduct creates “a strong presumption” that the prohibition is constitutional: “Principles of liberty fundamental enough to have been embodied within constitutional guarantees are not readily erased from the Nation’s consciousness.” McIntyre v. Ohio Elections Comrrin, 514 U. S. 334, 375-377 (1995) (Scalia, J., dissenting). The practice of prohibiting speech by judicial candidates on disputed issues, however, is neither long nor universal.
At the time of the founding, only Vermont (before it became a State) selected any of its judges by election. Starting with Georgia in 1812, States began to provide .for judicial election, a development rapidly accelerated by Jacksonian democracy. By the. time of the Civil War, the great majority of States elected their judges. E. Haynes, Selection and Tenure of Judges 99-135 (1944); Berkson, Judicial Selection in the United States: A Special Report, 64 Judicature 176 (1980). We know of no restrictions upon statements that could be made by, judicial candidates (including judges) throughout the 19th and the first quarter of the 20th century. Indeed, judicial elections were generally partisan during this period, the movement toward nonpartisan judicial elections not even beginning until the 1870’s. Id., at 176-177; *786M. Comisky & P. Patterson, The Judiciary — Selection, Compensation, Ethics, and Discipline 4, 7 (1987). Thus, not only were judicial candidates (including judges) discussing disputed legal and political issues on the campaign trail, but they were touting party affiliations and angling for party nominations all the while.
The first code regulating judicial conduct was adopted by the ABA in 1924. 48 ABA Reports 74 (1923) (report of Chief Justice Taft); P. McFadden, Electing Justice: The Law and Ethics of Judicial Election Campaigns 86 (1990). It contained a provision akin to the announce clause: “A candidate for judicial position . . . should not announce in advance his conclusions of law on disputed issues to secure class support . . . .” ABA Canon of Judicial Ethics 30 (1924). The States were slow to adopt the canons, however. “By the end of World War II, the canons ... were binding by the bar associations or supreme courts of only eleven states.” J. MacKenzie, The Appearance of Justice 191 (1974). Even today, although a majority of States have adopted either the announce clause or its 1990 ABA successor, adoption is not unanimous. Of the 31 States that select some or all of their appellate and general-jurisdiction judges by election, see American Judicature Society, Judicial Selection in the States: Appellate and General Jurisdiction Courts (Apr. 2002), 4 have adopted no candidate-speech restriction comparable to the announce clause,13 and 1 prohibits only the discussion of “pending litigation.”14 This practice, relatively new to judicial elections and still not universally adopted, does not compare well with the traditions deemed worthy of our attention in prior cases. E. g., Burson v. Freeman, 504 U. S. 191, 205-206 (1992) (crediting tradition of prohibiting speech around *787polling places that began with the very adoption of the secret ballot in the late 19th century, and in which every State participated); id., at 214-216 (Scalia, J., concurring in judgment) (same); McIntyre, supra, at 375-377 (SCALIA, J., dissenting) (crediting tradition of prohibiting anonymous election literature, which again began in 1890 and was universally adopted).
* * *
There is an obvious tension between the article of Minnesota’s popularly approved Constitution which provides that judges shall be elected, and the Minnesota Supreme Court’s announce clause which places most subjects of interest to the voters off limits. (The candidate-speech restrictions of all the other States that have them are also the product of judicial fiat.15) The disparity is perhaps unsurprising, since the ABA, which originated the announce clause, has long been an opponent of judicial elections. See ABA Model Code of Judicial Conduct, Canon 5(C)(2), Comment (2000) (“[M]erit selection of judges is a preferable manner in which to select the judiciary”); An Independent Judiciary: Report of the ABA Commission on Separation of Powers and Judicial Independence 96 (1997) (“The American Bar Association strongly endorses the merit selection of judges, as opposed to their election .... Five times between August 1972 and August 1984 the House of Delegates, has approved recommendations stating the preference for merit selection and encouraging bar associations in jurisdictions where judges are elected ... to work for the adoption of merit selection and retention”). That opposition may be well taken (it certainly had the sup*788port of the Founders of the Federal Government), but the First Amendment does not permit it to achieve its goal by-leaving the principle of elections in place while preventing candidates from discussing what the elections are about. “[T]he greater power to dispense with elections altogether does not include the lesser power to conduct elections under conditions of state-imposed voter ignorance. If the State chooses to tap the energy and the legitimizing power of the democratic process, it must accord the participants in that process ... the First Amendment rights that attach to their roles.” Renne v. Geary, 501 U. S. 312, 349 (1991) (Marshall, J., dissenting); accord, Meyer v. Grant, 486 U. S. 414, 424-425 (1988) (rejecting argument that the greater power to end voter initiatives includes the lesser power to prohibit paid petition-circulators).
The Minnesota Supreme Court’s canon of judicial conduct prohibiting candidates for judicial election from announcing their views on disputed legal and political issues violates the First Amendment. Accordingly, we reverse the grant of summary judgment to respondents and remand the case for proceedings consistent with this opinion.
It is so ordered.
The Eighth Circuit did not parse out the separate functions of these two entities in the case at hand, referring to the two of them collectively as the “Lawyers Board.” We take the same approach.
Nor did Wersal have any success receiving answers from the Lawyers Board when he included “concrete examples,” post, at 799, n. 2 (Stevens, J., dissenting), in his request for an advisory opinion on other subjects a month later:
“As you are well aware, there is pending litigation over the constitutionality of certain portions of Canon 5. You are a plaintiff in this action and you have sued, among others, me as Director of the Office of Lawyers Professional Responsibility and Charles Lundberg as the Chair of the Board of Lawyers Professional Responsibility. Due to this pending litigation, I will not be answering your request for an advisory opinion at this time.” App. 153.
Respondents are officers of the Lawyers Board and of the Minnesota Board on Judicial Standards (Judicial Board), which enforces the ethical rules applicable to judges.
Justice Ginsburg argues that we should ignore this concession at oral argument because it is inconsistent with the Eighth Circuit’s interpretation of the announce clause. Post, at 810 (dissenting opinion). As she appears to acknowledge, however, the Eighth Circuit was merely silent on this particular question. Ibid. Silence is hardly inconsistent with what respondents conceded at oral argument.
In 1990, in response to concerns that its 1972 Model Canon — which was the basis for Minnesota’s announce clause — violated the First Amendment, see L. Milord, The Development of the ABA Judicial Code 50 (1992), the ABA replaced that canon with a provision that prohibits a judicial candidate from making “statements that commit or appear to commit the candidate with respect to cases, controversies or issues that are likely to come before the court.” ABA Model Code of Judicial Conduct, Canon 5(A)(3)(d)(ii) (2000). At oral argument, respondents argued that the limiting constructions placed upon Minnesota’s announce clause by the Eighth Circuit, and adopted by the Minnesota Supreme Court, render the scope of the clause no broader than the ABA’s 1990 canon. Tr. of Oral Arg. 38. This argument is somewhat curious because, based on the same constitutional concerns that had motivated the ABA, the Minnesota Supreme Court was urged to replace the announce clause with the new ABA language, but, unlike other jurisdictions, declined. Final Report of the Advi*774sory Committee to Review the ABA Model Code of Judicial Conduct and the Rules of the Minnesota Board on Judicial Standards 5-6 (June 29, 1994), reprinted at App. 367-368. The ABA, however, agrees with respondents’ position, Brief for ABA as Amicus Curiae 5. We do not know whether the announce clause (as interpreted by state authorities) and the 1990 ABA canon are one and the same. No aspect of our constitutional analysis turns on this question.
Although the Eighth Circuit also referred to the compelling interest in an “independent” judiciary, 247 F. 3d, at 864-868, both it and respondents appear to use that term, as applied to the issues involved in this case, as interchangeable with “impartial.” See id., at 864 (describing a judge’s independence as his “ability to apply the law neutrally”); Brief for Respondents 20, n. 4 (“[J]udicial impartiality is linked to judicial independence”).
Justice Stevens asserts that the announce clause “serves the State’s interest in maintaining both the appearance of this form of impartiality and its actuality.” Post, at 801. We do not disagree. Some of the speech prohibited by the announce clause may well exhibit a bias against parties — including Justice Stevens’s example of an election speech stressing the candidate’s unbroken record of affirming convictions for rape, ante, at 800-801. That is why we are careful to say that the announce clause is “barely tailored to serve that interest,” supra, at 776 (emphasis added). The question under our strict scrutiny test, however, is not whether the announce clause serves this interest at all, but whether it is narrowly tailored to serve this interest. It is not.
We do not agree with Justice Stevens’s broad assertion that “to the extent that [statements on legal issues] seek to enhance the popularity of the candidate by indicating how he would rule in specific cases if elected, they evidence a lack of fitness for office.” Post, at 798 (emphasis added). Of course all statements on real-world legal issues “indicate” how the speaker would rule “in specific cases.” And if making such statements (of honestly held views) with the hope of enhancing one’s chances with the electorate displayed a lack of fitness for office, so would similarly motivated honest statements of judicial candidates made with the hope of enhancing their chances of confirmation by the Senate, or indeed of appointment by the President. Since such statements are made, we think, in every confirmation hearing, Justice Stevens must contemplate a federal bench filled with the unfit.
Justice Ginsburg argues that the announce clause is not election nullifying because Wersal criticized past decisions of the Minnesota Supreme Court in his campaign literature and the Lawyers Board decided not to discipline him for doing so. Post, at 811-812. As we have explained, however, had Wersal additionally stated during his campaign that he did not feel bound to follow those erroneous decisions, he would not have been so lucky. Supra, at 772-773. This predicament hardly reflects “the robust communication of ideas and views from judicial candidate to voter.” Post, at 812.
Justice Stevens devotes most of his dissent to this same argument that we do not make.
Nor do we assert that candidates for judicial office should be compelled to announce their views on disputed legal issues. Thus, Justice Ginsburg’s repeated invocation of instances in which nominees to this Court declined to announce such views during Senate confirmation hearings is pointless. Post, at 807-808, n. 1, 818-819, n. 4. That the practice *784of voluntarily demurring does not establish the legitimacy of legal compulsion to demur is amply demonstrated by the unredacted text of the sentence she quotes in part, post, at 819, from Laird v. Tatum, 409 U. S. 824, 836, n. 5 (1972): “In terms of propriety, rather than disqualification, I would distinguish quite sharply between a public statement made prior to nomination for the bench, on the one hand, and a public statement made by a nominee to the bench.” (Emphasis added.)
Although Justice Stevens at times appears to agree with Justice Ginsburg's premise that the judiciary is completely separated from the enterprise of representative government, post, at 798 (“[Ejvery good judge is fully aware of the distinction between the law and a personal point of view”), he eventually appears to concede that the separation does not hold true for many judges who sit on courts of last resort, ante, at 799 (“If he is not a judge on the highest court in the State, he has an obligation to follow the precedent of that court, not his personal views or public opinion polls”); post, at 799, n. 2. Even if the policymaking capacity of judges were limited to courts of last resort, that would only prove that the announce clause fails strict scrutiny. “[I]f announcing one’s views in the context of a campaign for the State Supreme Court might be” protected speech, ibid., then — even if announcing one’s views in the context of a campaign for a lower court were not protected speech, ibid. — the announce clause would not be narrowly tailored, since it applies to high- *785and low-court candidates alike. In fact, however, the judges of inferior' courts often “make law,” since the precedent of the highest court does not cover every situation, and not every case is reviewed. Justice Stevens has repeatedly expressed the view that a settled course of lower court opinions binds the highest court. See, e. g., Reves v. Ernst & Young, 494 U. S. 56, 74 (1990) (concurring opinion); McNally v. United States, 483 U. S. 350, 376-377 (1987) (dissenting opinion).
Idaho Code of Judicial Conduct, Canon 7 (2001); Mich. Code of Judicial Conduct, Canon 7 (2002); N. C. Code of Judicial Conduct, Canon 7 (2001); Ore. Code of Judicial Conduct, Rule 4-102 (2002). All of these States save Idaho have adopted the pledges or promises clause.
Ala. Canon of Judicial Ethics 7(B)(1)(c) (2002).
These restrictions are all contained in these States’ codes of judicial conduct, App. to Brief for ABA as Amicus Curiae. “In every state, the highest court promulgates the Code of Judicial Conduct, either by express constitutional provision, statutory authorization, broad constitutional grant, or inherent power.” In the Supreme Court of Texas: Per Curiam Opinion Concerning Amendments to Canons 5 and 6 of the Code of Judicial Conduct, 61 Tex. B. J. 64, 66 (1998) (collecting provisions).
2 J. Campbell, Lives of the Chief Justices of England 208 (1873) (quoting Hale’s Rules For His Judicial Guidance, Things Necessary to be Continually Had in Remembrance).