OPINION
We are asked to vacate a final opinion of the Minnesota Court of Appeals on the grounds that the author of the opinion, Judge Roland Amundson, was disqualified because the law firm that represented certain respondents was the same firm that represented a trust of which Judge Amundson was the trustee. Under the peculiar circumstances of this case, we determine that Judge Amundson was disqualified and that vacatur is necessary. Accordingly, we reverse the decision of the court of appeals denying the motion to vacate, vacate the opinion, and remand the underlying appeal to the court of appeals for redetermination by a new panel.
A. The Parties
In the early 1950s, the late Walter G. Anderson founded respondent Walter G. Anderson, Inc. (Anderson Inc.), a paper box manufacturing and printing company. In the 1970s, Walter G. Anderson gave a significant number of shares in Anderson Inc. to his two children, appellant Jacque-lin Powell and respondent Richard Anderson.
In 1980, respondent A & P Partnership was formed to own real estate, some of which was to be leased to Anderson Inc. Powell and Richard Anderson were made 50% general partners in the partnership.
In 1991, Richard Anderson and another executive of Anderson Inc., respondent Walter Gervais, founded respondent Walter G. Anderson Corp. (Anderson Corp.), also a paper box manufacturing and printing company.
Walter G. Anderson died on February 21, 1997. After his death, the sole shareholders of Anderson Inc. were Powell, Richard Anderson, and Walter G. Anderson’s estate.
B. The District Court Proceedings
Powell filed suit against her brother, Richard Anderson; Anderson Inc.; Anderson Corp.; A & P Partnership; Walter Gervais; and the estate of her late father. Powell sought dissolution of Anderson Inc. and Anderson Corp., payment for the fair value of her shares in Anderson Inc., damages for past misconduct, and an accounting for the affairs of A & P Partnership. She accused her father, her brother, and Gervais of fraud, misrepresentation and waste in connection with their management of Anderson Inc.1 *111She claimed that the formation of Anderson Corp. represented a usurpation of a corporate opportunity in breach of the fiduciary duty to Anderson Inc. owed by Gervais and Richard Anderson. Finally, she alleged that her brother violated their partnership agreement and Minnesota partnership law by excluding her from the management of A & P Partnership.
The estate, A & P Partnership, and Anderson Inc. were all represented by the law firm of Rider, Bennett, Egan & Arun-del, LLP (Rider Bennett). Richard Anderson, Gervais, and Anderson Corp. were represented by the law firm of Anthony, Ostlund & Baer, P.A. In their answers, the defendants relied in part on the terms of a Share Retirement Agreement, signed by Powell in 1976, which provided that she would sell her shares of Anderson Inc. to Anderson Corp. at book value upon the later death of Walter G. Anderson or his wife Lillian Anderson.
After court-ordered mediation failed, the six defendants moved for summary judgment. Powell opposed the motion, arguing that crucial factual issues, including the value of Anderson Inc. and Anderson Corp. and the proportion of Anderson Inc.’s stock held by Powell, remained in dispute. She requested discovery regarding these and other issues under Minnesota Rule of Civil Procedure 56.06.2
The district court granted the respondents’ summary judgment motion in part and denied it in part. The court, sua sponte, found the Share Retirement Agreement was enforceable but “exercise[d] the equitable powers granted to it” under Minn.Stat. § 302A.751, subd. 1, to award Powell fair value instead of book value for her shares. The court then decided the fair-value issue. The court used financial analyses of Anderson Inc. performed by the parties’ respective experts for the confidential use of the court in a settlement conference, rejected the financial analysis provided by Powell’s expert, and accepted the financial analysis provided by respondents’ expert to set the fair value of Anderson Inc. at $13,985,000. The court then determined that Powell’s share of Anderson Inc. was 24.73% and valued her shares at $3,464,449.
Although the district court’s first summary judgment order denied the respondents’ motions for summary judgment with regard to Powell’s claims for fraud and breach of fiduciary duty, the court entered a second order for “clarification” that stated the only claim remaining for trial was usurpation of corporate opportunity arising out of the creation of Anderson Corp. The order denied Powell’s request to present evidence on any other claims, including the claims for fraud and breach of fiduciary duty.
When Powell moved for partial summary judgment on liability on the usurpation claim, Gervais and Richard Anderson stipulated to liability, and the district court granted the partial summary judgment. A bench trial was then held on the value of Powell’s interest in Anderson Corp. The court limited the trial testimony to the separate value of Anderson Corp., exclud*112ed Powell’s proffered evidence of the combined value of Anderson Inc. and Anderson Corp., and determined that the value of Powell’s share of Anderson Corp. was $605,390.40.
By its last order, the district court directed Powell to tender her shares to Anderson Inc. and ordered that, upon her failure to do so, Anderson Inc. should pay the $3,464,449 into court and could cancel Powell’s stock certificates as being null and void. The court also adopted the appraisal of the assets of A & P Partnership, for a total value of $7,150,000, which the court essentially divided equally between the partners.
C. The Initial Appeal
Powell and respondents each filed a notice of review from the several orders and judgments. Powell made several claims of error. First, regarding the valuation of Anderson Inc., Powell argued that the issue was not ripe for summary judgment because: (1) she had never been given the opportunity to conduct discovery regarding Anderson Inc.’s financial data (a stay of discovery had been in effect before the summary judgment motion); (2) there were factual disputes regarding the number of shares she owned; (3) there were factual disputes regarding the value of Anderson Inc.; (4) the district court improperly relied upon appraisal reports that were not in the record but were only preliminary drafts for use at the settlement conference;3 (5) the court rejected the appraisal report submitted by her expert even though that report presented genuine issues of material fact; and (6) the court relied upon the expert report submitted by respondents even though it had been provided to the court in confidence and Powell had not been allowed to see it or test its conclusions.
Next, Powell argued that she had not been allowed to present evidence on the combined value of Anderson Inc. and Anderson Corp. She contended that the combined value increased her interest by over three million dollars. She also argued that the district court erred in applying minority and marketability discounts to her interest in Anderson Corp. Finally, she complained that the court had refused to order an accounting of A & P Partnership, dismissed her fraud and breach of fiduciary duty claims and denied attorney fees, all without explanation or rationale.
On July 11, 2000, in a unanimous opinion authored by Judge Amundson, the court of appeals rejected Powell’s claims in all respects except as to the minority and marketability discounts the district court had applied to the value of Powell’s shares of Anderson Corp. Powell v. Anderson, No. C5-99-1755, 2000 WL 943842, at *5, *6 (Minn.App. July 11, 2000). We denied Powell’s petition for further review on September 27, 2000.
D. Events after the Court of Appeals’ Opinion
Approximately one and one-half years after he authored the Powell opinion, Judge Amundson resigned his seat on the court of appeals. Soon afterward, Judge Amundson pleaded guilty to theft of over $300,000 from a trust fund for which he was acting as a trustee.
*113On March 10, 2002, in one of its series of articles regarding Judge Amundson, the Star Tribune newspaper mentioned that the trust in question was represented by “trust attorney R. Thomas Greene Jr.” Greene is a partner at the Rider Bennett law firm. According to Greene’s affidavit submitted in this case, he “began representing [Amundson] in his capacity as trustee of the trust * * * on October 25, 1995;” he joined Rider Bennett in March 1996; and he “withdrew from representing Amundson in this capacity on September 27, 2001, for reasons unrelated to this proceeding.” Greene was therefore at Rider Bennett, representing the trust for which Judge Amundson served as a trustee, for the entire duration of the court of appeals’ proceedings in Powell. 4
Powell’s counsel saw the Star Tribune article, recognized Greene’s name, and filed a motion in the court of appeals to vacate the court’s earlier decision based on the attorney-client relationship between Judge Amundson and Rider Bennett. A new panel of the court of appeals unanimously denied the motion. The court wrote:
Appellant has not established that the relief sought is authorized by Minnesota law or is warranted. The unanimous decision in this case was made by a panel, and that decision is the decision of the entire court. Minn.Stat. § 480A.08, subd. 1 (2000). After a careful review of the internal handling of this matter, the submissions of all parties, and the appellate record, the panel has concluded that there has not been a sufficient showing for the extraordinary remedy sought.
Powell again filed a petition for further review; this time we granted the petition.
The parties apparently agree that jurisdiction to vacate a final decision of the court of appeals arises under Minnesota Rule of Civil Appellate Procedure 102, which provides that, for good cause shown, the supreme court or the court of appeals may suspend the requirements of the appellate rules. Such jurisdiction also exists in this court under our supervisory power to ensure the fair administration of justice. See, e.g., State v. Scales, 518 N.W.2d 587, 592 (Minn.1994). We further note that in Liljeberg v. Health Services Acquisition Corp., 486 U.S. 847, 863, 108 S.Ct. 2194, 100 L.Ed.2d 855 (1988), a case with comparable facts to this one, the United States Supreme Court held that it had the power to vacate a final district court judgment under Federal Rule of Civil Procedure 60(b)(6), which is substantially identical to Minnesota Rule of Civil Procedure 60.02(f).
We view this appeal as presenting two separate questions:
1. Did the circumstances of Judge Amundson’s relationship to the Rider Bennett firm present a basis to reasonably question his impartiality, sufficient to require disqualification?
2. If so, was Judge Amundson’s failure to disqualify himself sufficiently prejudicial to require vacatur of the court of appeals’ opinion?
I.
We have had only rare occasion to address the question of the disqualification of an appellate judge. In one of those cases, State ex rel. Wild v. Otis, 257 N.W.2d 361 (Minn.1977), we looked for guidance on such a question to the ABA Standards of Judicial Administration, Standards Relat*114ing to Appellate Courts. We do so again now.
A Grounds for Disqualification
Section 3.42 of the 1994 edition of the Standards Relating to Appellate Courts provides:
Disqualification of Judges. A judge of an appellate court should be subject to disqualification on the grounds set forth in the American Bar Association, Model Code of Judicial Conduct (1990) and in any case in which the judgment under review is one in which the judge participated as a judge in a lower court.
3 American Bar Ass’n, Judicial Admin. Div., Standards of Judicial Administration: Standards Relating to Appellate Courts § 3.42, at 80 (1994) (hereinafter Standards Relating to Appellate Courts).
On November 1, 1995, we ordered “that the Code of Judicial Conduct on the Rules of the Board on Judicial Standards, as amended, ‘shall govern all matters which come before the Board on Judicial Standards on or after the 1st day of January 1996.’ ”
Canon 3D(1) of the Minnesota Code of Judicial Conduct provides that
[a] judge shall disqualify himself or herself in a proceeding in which the judge’s impartiality might reasonably be questioned, including but not limited to instances where:
(a) the judge has a personal bias or prejudice concerning a party or a party’s lawyer * * *.
The Advisory Committee’s Commentary to Canon 3D(1) explains:
Under this rule, a judge is disqualified whenever the judge’s impartiality might reasonably be questioned, regardless whether any of the specific rules in Section 3D(1) apply. * * *
A judge should disclose on the record information that the judge believes the parties or their lawyers might consider relevant to the question of disqualification, even if the judge believes there is no real basis for disqualification.
Commentary to the Minnesota Code of Judicial Conduct, Comment — Canon 3 (Nov. 22,1995).
Respondents argue that the Code of Judicial Conduct may lead to a remedy in disciplinary proceedings but is not a controlling basis for disqualification of a judge. They rely on the court of appeals’ decision in Nachtsheim v. Wartnick, 411 N.W.2d 882, 890-91 (Minn.App.1987), which held that Canon 3C(1) did not provide a basis to disqualify a district court judge under Minnesota Rule of Civil Procedure 63.03 because the rale requires an affirmative showing of prejudice to remove a successor judge after a party has already disqualified the predecessor as a matter of right. See Minn. R. Civ. P. 63.03; see also Miller v. Michel, 409 N.W.2d 11, 14 (Minn.App.1987) (stating that district court judge’s violation of Canon 3C may be grounds for discipline but not for a new trial). These cases, however, do not deal with the present Canon 3D(1) and, in any event, focus not on the question of disqualification but on the remedy available to a party who seeks relief from a decision already made by an allegedly disqualified district court judge.
To the extent that Nachtsheim and Miller might be interpreted as standing for the proposition that disqualification of an appellate judge could not be based upon the Code of Judicial Conduct, we specifically overrule them and accept the recommendations of the ABA Standards Relating to Appellate Courts that an appellate judge should be subject to disqualification on the grounds set forth in the Code of Judicial Conduct. We note that Canon *1153D(1) is not purely aspirational in its terms, because it uses the operative verb “shall disqualify.”
Having so concluded, we also note that the grounds for disqualification in Canon 3D(1) are stated broadly, leaving considerable room for interpretation in their application to any given set of circumstances. Canon 3D(1) does not provide a precise formula that can automatically be applied.
B. Procedure for Disqualification
The Commentary to section 3.42 of the ABA Standards Relating to Appellate Courts suggests that the procedure for determining disqualification of an appellate judge differs from that applicable to a district court judge.
An appellate judge should be subject to challenge for cause on the same grounds as a trial judge, and also when an appeal involves review of that judge’s decision. The most difficult problem concerns the procedure to be employed. As in the challenge of a trial judge, if the challenge is sufficient on its face and any reasonable doubt of the judge’s disinterestedness is suggested, the judge may be expected to act favorably on the recu-sal. If the judge does not do so, at the trial level, factual issues relating to disqualification should properly be determined by another judge. See Section 2.32. In the case of an appellate judge, however, that procedure would subject the judge to decision of disinterestedness by peers with whom that judge continues to serve in a collegial capacity in deciding the case. Moreover, because an appellate court decides questions of law rather than fact, the question of an appellate judge’s ‘bias’ is often practically indistinguishable from the question of the judge’s views on the law, which are not properly subject to challenge through the recusal procedure. Given these complications, it is better that the question of recusal be decided by the challenged judge. There remains the remedy of appeal at the supreme court level from the decision in which the judge participates; it the challenged judge is a justice of the supreme court, reliance must be placed on the justice’s recognition that a court should not only be disinterested but that it should appear to be so.
Standards Relating to Appellate Courts, supra, at 80-81.
This suggestion is consistent with the approach we have taken in the few cases involving disqualification of appellate judges. Thus, we have held that “[i]t has long been the practice of [the Minnesota Supreme Court] to honor decisions of its individual members as to whether to participate in a pending proceeding.” In re Modification of Canon SA(7) of the Minn. Code of Judicial Conduct, 438 N.W.2d 95, 95 (Minn.1989); see also Wild, 257 N.W.2d at 363-64. The Minnesota Court of Appeals Special Rule of Practice 10 reflects the same approach by providing that “[a]p-plication of the principles governing recu-sal is ultimately the responsibility of the individual judge.”5
*116This procedure and the broad language of Canon 3D(1) imply that appellate judges must be allowed some discretion in deciding their own disqualification under any particular set of facts. A judge has a duty to minimize the burdens imposed on the judicial system by disqualification. But any discretion must be tempered by the unavoidable conflict of interest that is presented when a judge considers his or her own disqualification. These competing considerations appear to have been recognized in the development of the standard of review that has been applied when a judge’s decision not to disqualify is challenged, a standard which requires an objective examination of whether the judge’s impartiality could reasonably be questioned. According to the court of appeals in State v. Laughlin, 508 N.W.2d 545, 548 (Minn.App.1993), because Minnesota Code of Judicial Conduct Canon 3C(1) “require[s] disqualification where there is an appearance of partiality, a [reviewing] court’s inquiry must go beyond the challenged judge’s statements and include an objective examination into the circumstances surrounding the removal request.” This objective examination displaces any deference that might otherwise be paid to the challenged judge’s decision to not re-cuse.6
Respondents argue that our standard of review should be considerably narrower in this case because a separate panel of the court of appeals already reviewed the question of Judge Amundson’s disqualification and the decision of that panel should be reversed by this court only for a clear abuse of discretion. But the decision of the separate court of appeals’ panel did not address the underlying issue of disqualification under Canon 3D(1). It focused instead on the issue of vacatur. For this reason, we will consider and apply de novo the “objective examination” standard of review for the purpose of determining the initial issue of disqualification.
C. Objective Examination
The parties differ about whether an objective examination of the facts of this case would cause a reasonable examiner to question Judge Amundson’s impartiality. Powell refers to: (1) “Judge Amundson’s continuing, concurrent attorney-client relationship with the Rider Bennett Firm;” (2) the fact that Judge Amundson “had to submit all of his bills and accounts to [Rider Bennett] so Greene could prepare the annual trust petitions;” (3) the letters between Judge Amundson and Rider Bennett “immediately before and after the Powell oral argument;” and (4) the fact that “Judge Amundson was stealing from that trust.”
Respondents emphasize that the trust attorney, Greene, had no contact with the attorneys who argued the Powell case, and they argue that Judge Amundson’s resignation and misconduct are “wholly unrelated to this case.” Respondents cite Desnick v. Mast, 311 Minn. 356, 249 N.W.2d 878 (1976), as controlling precedent. Des-nick involved a plaintiffs contention that he should be granted a new trial because the attorney for one of the defendants was concurrently representing the trial judge as a nominal party in a legal malpractice suit. Id. at 362-63, 249 N.W.2d at 882-83. This court held that “[t]he circumstances of the relationship * * * clearly indicate it was a technical, rather than a direct or a personal, relationship,” and that the plain*117tiff was therefore “not denied a fair trial.”7 Id.
We do not consider Desnick to be controlling precedent on the issue of disqualification. First, Desnick did not address the question of disqualification under the Code of Judicial Conduct but rather went directly to the question of whether the plaintiff was denied a fair trial, which is relevant to the issue of vacatur. Second, as we will discuss below, the issue of disqualification is fact-dependent, and Des-nick is factually distinguishable from the present case.
The caselaw from other jurisdictions regarding the effect on disqualification of the existence of an attorney-client relationship between the judge and an attorney or firm appearing before the judge is mixed and also largely fact-dependent.8 Several courts have held that, under their peculiar facts, the appearance of impropriety arising from an attorney-client relationship mandated disqualification of the conflicted judge.9 A smaller number of courts have refused to require disqualification in cases involving attorney-client relationships between judges and attorneys appearing be*118fore them.10
Our review of the applicable precedent leads us to the conclusion that a reviewing court, pursuing an objective examination of a case where a judge has had an attorney-client relationship with an attorney appearing before the judge, should generally weigh four factors in deciding on disqualification.
First, the reviewing court should consider the extent of the attorney-client relationship. If the relationship consisted of a single, short episode, or even a series of sporadic contacts, disqualification is less likely than if it consisted of a long-term, continuous course of representation. Similarly, representation that had been concluded prior to the instant case is less likely to lead to disqualification than representation that is concurrent with the case.11
Second, the reviewing court should examine the nature of the representation. A direct relationship, where the judge is represented personally, is more indicative of a reasonable question regarding the judge’s impartiality than a relationship that only involves the judge in some institutional or technical role.12 Further, the more serious the matter for the judge, the greater the impact of the representation on the judge’s impartiality.
Third, the reviewing court should consider the frequency, volume and quality of contacts between the judge and the attorney or law firm. The more frequent and substantial these contacts, the more likely the relationship is to create a reasonable question as to impartiality. Likewise, the closer the contacts come to the subject of the case before the judge, the greater the impact on impartiality.
Fourth, the reviewing court should take into account any special circumstances that might either enhance or limit (1) the importance of the attorney or firm to the judge and/or (2) the appearance of impropriety to the public.13
On the facts before us, the first three of these factors reveal a close ease *119that would perhaps have fallen within the discretion of Judge Amundson. With regard to the extent of the representation, this weighs toward disqualification because Greene represented the trust for nearly six years, and this time period overlapped the entire course of the Powell litigation through our denial of Powell’s petition for further review in September 2000. As to the nature of the representation, this weighs against disqualification because Rider Bennett’s representation was somewhat less personal than a prototypical attorney-client relationship, even though Judge Amundson was not an entirely nominal party. As to the frequency and quality of contact, this weighs against disqualification because Judge Amundson did not communicate extensively with Rider Bennett attorneys, and when he did, it concerned technical aspects of the management and representation of the trust. As respondents emphasize, the connection between Rider Bennett and Judge Amund-son’s role in Powell is further attenuated by the fact that the Rider Bennett attorneys who argued this case were unaware of the concurrent representation.
However, the “special circumstances” have more impact in this case than in any other we have seen. Specifically, Judge Amundson’s theft from the trust that was the subject of Rider Bennett’s representation significantly enhanced the importance of that law firm to him. Judge Amundson was undoubtedly aware that the Rider Bennett attorneys were in a position to discover his wrongdoing — a fact that provided him with a strong motive to favor that firm. Judge Amundson’s theft also made the risk of an appearance of impropriety to the public extremely high.
Moreover, the details of Judge Amund-son’s misdeeds undercut respondents’ argument about the technical nature of the representation. While it is true that the trust, not Judge Amundson, was technically the client, it is also true that Judge Amundson treated the trust funds as his own.
We conclude that an objective examination into the Powell appeal reveals that Judge Amundson’s impartiality is subject to reasonable question. We do not conclude that Judge Amundson was actually partial to the Rider Bennett respondents, but instead rely on the appearance of partiality that inevitably arises from this unique combination of facts. Therefore, Judge Amundson should have disqualified himself from hearing this case.
II.
We turn now to the issue of whether Judge Amundson’s failure to disqualify himself requires that the court of appeals’ opinion be vacated and the case reversed and remanded. We have found no Minnesota case addressing the issue of vacatur on the basis of the participation in an appellate proceeding by a disqualified judge. Therefore, this is a case of first impression for this court.
Respondents argue that we should adopt the standard followed by Minnesota appellate courts when reviewing an appeal from the denial of a motion to disqualify a district court judge. In that context, our courts have engaged in a two-part inquiry — asking, first, whether there was judicial misconduct and, second, whether the misconduct denied the litigants a fair trial. See, e.g., Uselman v. Uselman, 464 N.W.2d 130, 139 (Minn.1990) superseded by statute on other grounds as stated in Radloff v. First Am. Nat’l Bank of St. Cloud, N.A., 470 N.W.2d 154 (Minn.App.1991); Desnick, 311 Minn. at 362-63, 249 N.W.2d at 882-83; Nachtsheim, 411 N.W.2d at 891. As to the first part — judicial misconduct— respondents argue that Powell should be *120required to show that Judge Amundson was in fact biased against her.
Powell responds that requiring her to prove actual bias and injury would eviscerate the objective appearance-of-impropriety standard of Canon 3D(1) and create a standard that is “nigh-impossible to meet.” She also forecasts a “procedural morass” if attorneys are forced to engage in invasive discovery to discern what contacts presiding judges have with the parties and law firms that appear before them and what actual impact those contacts may have had.
We share Powell’s concerns regarding an actual bias standard. While the existence of actual bias may be relevant to one of the objectives of disqualification — to provide a fair trial to litigants— it is not necessarily relevant to the other objective — to promote confidence in the judiciary.14 But we agree with respondents that not every case involving judicial disqualification deserves vacatur.
The United States Supreme Court offers a third option. In Liljeberg v. Health Services Acquisition Corp., the Court adopted the “appearance of partiality” test urged by Powell but tempered it with a harmless error analysis. 486 U.S. 847, 864, 108 S.Ct. 2194, 100 L.Ed.2d 855 (1988). The Lüjeberg Court addressed a case regarding the ownership of St. Jude Hospital of Kenner, Louisiana. Id. at 849, 108 S.Ct. 2194. Loyola University had an interest in the hospital, and the district court judge was a Loyola trustee. Id. at 850-51, 108 S.Ct. 2194. The judge claimed to have forgotten that Loyola held an interest in St. Jude. Id. at 851, 108 S.Ct. 2194. Petitioner could not prove that the judge was actually aware of the conflict, but the Court nonetheless found that his actions violated the federal judicial disqualification statute because “an objective observer would have questioned Judge Gollins’ impartiality.” Id. at 861, 108 S.Ct. 2194.
The Lüjeberg Court rejected any requirement of proof of actual bias and upheld the Fifth Circuit’s reading of the federal disqualification statute:
“The goal of [28 U.S.C.] section 455(a) is to avoid even the appearance of partiality. If it would appear to a reasonable person that a judge has knowledge of facts that would give him an interest in the litigation then an appearance of partiality is created even though no actual partiality exists because the judge does not recall the facts, because the judge actually has no interest in the case or because the judge is pure in heart and incorruptible.”
Id. at 860, 108 S.Ct. 2194 (quoting Health Servs. Acquisition Corp. v. Liljeberg, 796 F.2d 796, 802 (5th Cir.1986)).
However, the Court recognized that not every case presenting an appearance of impropriety merits vacatur:
A conclusion that a statutory violation occurred does not, however, end our inquiry. As in other areas of the law, there is surely room for harmless error committed by busy judges who inadvertently overlook a disqualifying circumstance.
*121 Id. at 862, 108 S.Ct. 2194. The Court then developed the following balancing test to be used for purposes of vacatur:
We conclude that in determining whether a judgment should be vacated for a violation of § 455(a), it is appropriate to consider the risk of injustice to the parties in the particular case, the risk that the denial of relief will produce injustice in other cases, and the risk of undermining the public’s confidence in the judicial process. We must continuously bear in mind that to perform its high function in the best way justice must satisfy the appearance of justice.
Id. at 864, 108 S.Ct. 2194 (internal quotation marks and citations omitted). Applying that test, the Court concluded that vacatur was the appropriate remedy on its facts. Id. at 867-70, 108 S.Ct. 2194.
We find the Supreme Court’s test for vacatur appropriate. It properly considers each of the two objectives of disqualification: to provide the parties a fair trial (and appeal) and to promote confidence in the judicial system. And by using a risk analysis, it allows the balancing of those objectives against the potential burdens placed on the judicial system and the parties by reopening a final judgment.
A. Risk of Injustice to the Parties
We begin the analysis of this factor by expressing our grave concern with the correctness of the court of appeals’ opinion authored by Judge Amundson. While the issues discussed in that opinion are not directly before us and we do not propose to decide them, we have examined the opinion in the light of Powell’s arguments and have concluded that the apparent failure of the opinion to recognize and address those arguments is relevant to the first factor of the Liljeberg test.
First, the court of appeals did not apply the proper standard of review for issues presented by summary judgment. Summary judgment should be granted only where there are no genuine issues as to any material fact. Minn. R. Civ. P. 56.03. Additionally, the district court is authorized to continue a motion for summary judgment if further discovery is needed before responding. Minn. R. Civ. P. 56.06.
As to Powell’s argument that there were genuine issues of material fact about her ownership share of Anderson Inc., the opinion noted that Powell “presented admissible evidence to support a much higher percentage,”- yet it affirmed the district court’s “determination” of a 24.73% share because there was evidence supporting that determination. Powell, 2000 WL 943842, at *4. This standard of review would be appropriate after trial, but it was not appropriate for summary judgment.
As to Powell’s argument that there were genuine issues of material fact regarding the valuation of Anderson Inc., the opinion does not address: (1) the propriety of the district court’s rebanee on an expert appraisal that was not in the record or (2) the fact dispute created by the competing valuation presented by Powell’s expert. It only discussed the narrow, and largely irrelevant, question of whether the appraisals were admissible under Minnesota Rule of Evidence 408 even though they "were presented in settlement discussions. Id. at *3.
As to Poweb’s request for discovery on the valuation issue, the court correctly observed that a continuance of a motion for summary judgment is generally discretionary, but it overlooked the effect of the stay of discovery that had been imposed to facilitate settlement negotiations. Id. at *4.
As to Powell’s argument that she should have been able to present evidence of the combined value of Anderson Inc. and *122Anderson Corp., the court of appeals treated the issue as an evidentiary one, decided by the district court’s evaluation of expert-witness credibility. Id. Since the valuation of Anderson Inc. was decided by summary judgment, the district court was not authorized to make credibility determinations.
Finally, the court of appeals did not discuss the fraud and fiduciary duty claims, as to which the district court first denied summary judgment and later adopted respondents’ proposed order that dismissed the claims without explanation.
Respondents argue that any error resulting from Judge Amundson’s participation in this case was harmless because the initial court of appeals’ opinion was unanimous — it was joined by two judges who were not disqualified from hearing the case. The parties cite conflicting caselaw regarding whether an appellate panel’s decision should be vacated when a member whose vote is not decisionally necessary is later shown to be disqualified from participating. See, e.g., Johnson v. Sturdivant, 295 Ark. 663B, 758 S.W.2d 415, 415-16 (1988) (ordering vacatur and recusing the entire Arkansas Supreme Court for the rehearing because one justice with a conflict of interest had participated in a 6-1 decision); Goodheart v. Casey, 523 Pa. 188, 565 A.2d 757, 761-62 (1989) (holding that, even if two Pennsylvania Supreme Court justices hearing an earlier case had been improperly “interested,” the judgment would not be vacated because the judges’ participation was “mere surplusage” in the 6-1 decision).
In Aetna Life Insurance Co. v. Lavoie, 475 U.S. 813, 106 S.Ct. 1580, 89 L.Ed.2d 823 (1986), the United States Supreme Court addressed a case in which a conflicted justice of the Alabama Supreme Court had authored the opinion and cast the “deciding” vote in a 5A decision. Id. at 827-28, 106 S.Ct. 1580. “Because of [the conflicted justice’s] leading role in the decision under review,” the Court wrote, “we conclude that the ‘appearance of justice’ will best be served by vacating the decision and remanding for further proceedings.” Id. at 828, 106 S.Ct. 1580. But the concurring opinions argued that the fact that the disqualified Alabama justice’s vote was “deciding” was irrelevant. Id. at 830-31, 106 S.Ct. 1580 (Brennan, J., concurring), 831-32, 106 S.Ct. 1580 (Blackmun, J., concurring in the judgment). According to Justice Blackmun, the justice’s “mere participation in the shared enterprise of appellate decisionmaking — whether or not he ultimately wrote, or even joined, the Alabama Supreme Court’s opinion — posed an unacceptable danger of subtly distorting the decisionmaking process.” Id. at 831, 106 S.Ct. 1580. Justice Brennan added:
The description of an opinion as being “for the court” connotes more than merely that the opinion has been joined by a majority of the participating judges. It reflects the fact that these judges have exchanged ideas and arguments in deciding the case. It reflects the collective process of deliberation which shapes the court’s perceptions of which issues must be addressed and, more importantly, how they must be addressed. And, while the influence of any single participant in this process can never be measured with precision, experience teaches us that each member’s involvement plays a part in shaping the court’s ultimate disposition. The participation of a judge who has a substantial interest in the outcome of a case of which he knows at the time he participates necessarily imports a bias into the deliberative process. This deprives litigants of the assurance of impartiality that is the fundamental requirement of due process.
Id.
Although we can envision a case in which a disqualified judge’s participation in *123an appellate panel was inconsequential enough that it would not infect the decision, this is not one of those cases. Because of our substantive concerns regarding the correctness of the initial court of appeals’ opinion in Powell and because Judge Amundson authored the opinion, we conclude that Judge Amundson’s participation was not cured by the fact that two qualified judges joined his opinion.
Second, respondents argue that any error should be regarded as harmless because the subsequent review by the second court of appeals’ panel prevented any injustice to Powell. But it is unclear from the second panel’s order whether the court conducted a complete review of the underlying merits of the case. As written, the order is too summary to consider it an independent analysis of the merits, the necessity of disqualification, or the effects of Judge Amundson’s failure to disqualify. The second panel’s review therefore does not render the error regarding Judge Amundson’s participation harmless.
Finally, respondents argue that our September 2000 denial of Powell’s initial petition for further review demonstrates the reliability of Judge Amundson’s opinion. Our consideration of petitions for further review is guided by Minnesota Rule of Civil Appellate Procedure 117, subd. 2. Because of the criteria in subdivision 2 (e.g., whether “the question presented is an important one upon which the Supreme Court should rule,” whether “the lower courts have so far departed from the accepted and usual course of justice as to call for an exercise of the Supreme Court’s supervisory powers,” and whether “a decision by the Supreme Court will help develop, clarify, or harmonize the law”), our denial of a petition for further review does not provide any confirmation of the correctness of the court of appeals’ decision. Minn. R. Civ.App. Proc. 117, subd. 2(a), (c), (d). As we have held,
denial of a petition for further review means no more than that the supreme court has declined, at that time and for whatever undisclosed reasons, to consider the matter.
⅜ ⅜ * *
The denial does not give the court of appeals decision any more or less prece-dential weight than a court of appeals decision from which no review was sought. The denial means only that, for one reason or another which is seldom disclosed, and not infrequently for conflicting reasons which may have nothing to do with the merits and certainly may have nothing to do with any view of the merits taken by a majority of the Court, there were not three members of the Court who thought the case should be heard.
Murphy v. Milbank Mut. Ins. Co., 388 N.W.2d 732, 739 (Minn.1986) (internal quotation marks and brackets omitted).
Respondents also argue that a remand in this case would be “an extreme remedy” and would be “to the unfair and extreme hardship of Respondents.” While we do not doubt that the burden of a remand on all of the litigants is serious, we do not believe that that burden is severe enough to allow the decision to stand. Therefore, we conclude that the first Liljeberg factor weighs strongly in favor of vacatur.
B. Risk that Denial of Relief will Produce Injustice in Other Cases
Because the initial opinion in Powell was unpublished and therefore had limited value as precedent, we doubt that denying Powell’s motion here would prejudice parties in other cases. Accordingly, this factor provides no support for vacatur.
*124 C. Risk of Undermining the Public’s Confidence in the Judicial Process
The circumstances of Judge Amundson’s theft from the trust have been widely publicized. As noted above, Judge Amund-son’s actions present a reasonable question about the impartiality of his decision in Powell. Thus, there is much reason to believe that the risks of weakening the public’s trust in the judicial branch from allowing the decision to stand are significant.
Indeed, this factor weighs more heavily in the instant case than it does in most other disqualification decisions involving judicial conflicts. Whereas the judges in Desnick, Liljeberg, Sturdivant, Goodheart, Aetna and other similar cases had connections that could have implicated their civil litigation or other property interests, Judge Amundson is unique in having criminal liability potentially turn on maintaining a beneficial relationship with the Rider Bennett firm. This distinction makes the appearance of impropriety in this case more troublesome, and it substantially increases the risk that Judge Amundson’s participation in the appeal would undermine the public’s confidence in the judicial process. Therefore, we conclude that third factor weighs in favor of vacatur.
Applying the Liljeberg factors to Judge Amundson’s participation in the Powell appeal, we conclude that the risk of prejudice to Powell is substantial and that the risk of undermining the public’s confidence in the judicial process is also significant. Accordingly, we hold that the court of appeals erred, we reverse its June 17, 2002, order, and we vacate its July 11, 2000, Powell opinion.
III.
Having decided that the court of appeals’ opinion should be vacated, we are left with the question of the proper forum for remand.
Powell argues that the district court’s errors were so “egregious” that the case should be remanded to the district court, essentially to start over. In the alternative, she seeks a remand to the court of appeals for another full argument on the issues raised by her appeal.
While we agree with Powell that we have the power, under Minnesota Rule of Civil Appellate Procedure 102 and our supervisory role, to remand this case to the district court, we conclude that the court of appeals is the proper place for the remand because it was the appellate process that was tainted by the failure of Judge Amundson to disqualify himself from participation.
Order denying motion to vacate reversed, opinion of July 11, 2000, vacated, and appeal remanded to the court of appeals for redetermination.
GILBERT, J., took no part in the consideration or decision of this case.