¶ 1. Johnnie Carprue appeals from a judgment entered after a jury found him guilty of one count of second-degree sexual assault, use of force, contrary to Wis. Stat. § 940.225(2)(a) (2001-02).1 He also appeals from an order denying his postconviction motion. Carprue claims: (1) the trial court erred when it conducted its own investigation, which included questioning a witness, questioning the defendant, and using a file from another pending criminal case; and (2) trial counsel provided ineffective assistance by failing to object to the trial court's actions. *172Because the trial court's actions reflected an adversarial process and did not constitute harmless error, we reverse and remand for a new trial.

BACKGROUND

¶ 2. In May 2001, Carprue was charged with second-degree sexual assault. The case was tried to a jury. During the trial, the State presented evidence that when the police came to arrest Carprue, he fled and tried to hide. Carprue testified that he fled from police because he was violating rules of In-House Correctional Services related to a different matter, and that he feared the police would arrest him for those violations if they caught him. He indicated he did not flee from the police for fear of being apprehended for a sexual assault. It was Carprue's theory that the sexual intercourse with the victim was consensual.

¶ 3. After his testimony, but before the close of evidence, the trial court excused the jury and conducted its own investigation. The trial court called Kenneth Morrow, who was the operations manager of In-House Correctional Services. Morrow was present in the courtroom as he had been named by the defense as a "custodian of records." He brought Carprue's "in-house" file with him. The court questioned Morrow about "in-house" procedures relative to rules violations. Morrow testified that if an individual violates the telephonic monitoring system, "in-house" notifies the court, a bench warrant is issued, and the warrant is sent to the sheriffs department. Morrow stated that this procedure is explained to inmates and, once released, the individual would have no reason to believe that a rules violation would result in direct notification to the police.

*173¶ 4. After this questioning, the court then questioned Carprue about a letter he wrote to a different trial court in a separate matter, wherein he pleaded for leniency and referred to his fiancée. Following this exchange, the jury was brought back and the trial court asked the State whether they would be calling any rebuttal witnesses.

¶ 5. The State called two rebuttal witnesses. The first was Kenneth Morrow. The second witness called has no relevance to this appeal. Morrow testified as to the procedures taken when an individual violates "in-house" monitoring. Morrow stated that he would notify the court, the court would issue a bench warrant and send it to the sheriffs department. He indicated that all inmates are informed, both verbally and in writing, that this is the procedure. Morrow advised that an inmate would have no reason to think that a rules violation would result in direct contact with the police.

¶ 6. After instructions and closing arguments, the case was sent to the jury. The jury found Carprue guilty. The trial court sentenced him to twenty-five years; fifteen years of initial confinement, followed by ten years of extended supervision. Judgment was entered. Subsequently, Carprue filed a postconviction motion alleging that the trial court's conduct of calling its own witness, and questioning Carprue with respect to a letter in a separate matter, improperly prejudiced his case. The postconviction court noted that the procedure engaged in by the trial court was "out of the ordinary," but found that because it was done outside the presence of the jury, it was not prejudicial. Carprue now appeals.

*174DISCUSSION

¶ 7. Carprue asserts two claims of error: (1) the trial court's sua sponte investigation and questioning of witnesses resulted in improper adversarial conduct and exceeded judicial authority; and (2) defense counsel's failure to object to the trial court's actions constituted ineffective assistance of counsel. Carprue claims that the trial court's actions constituted a violation of due process and prevented him from receiving a fair trial. Our review is de novo. State v. Littrup, 164 Wis. 2d 120, 473 N.W.2d 164 (Ct. App. 1991). We conclude that because the trial court's actions resulted in improper partisanship and advocacy, we must reverse the judgment and order and remand for a new trial.

¶ 8. "Due process of law guarantees a criminal defendant a fair trial before an impartial judge and jury in a neutral atmosphere." State v. Brown, 741 A.2d 321, 323 (Conn. App. Ct. 1999); see U.S. Const. amend. Xiy § 1. There is no dispute that a trial court is permitted to question witnesses in order to clarify testimony. State v. Asfoor, 75 Wis. 2d 411, 437, 249 N.W.2d 529 (1977); Wis. Stat. § 906.14. "Clarification questions" do not interfere with a defendant's constitutional right to an impartial judge. Clarification is permitted because "[a] judge does have some obligation to see to it that justice is done." Asfoor, 75 Wis. 2d at 437. It is expected that a trial court will exercise its statutory authority to intervene in the trial by calling or questioning witnesses "only in the exceptional case." Wis. Stat. Ann. § 906.14, Judicial Council Committee Note 1974. The trial court should never "take an active role in trying the case." State v. Garner, 54 Wis. 2d 100, 104, 194 N.W.2d 649 (1972). *175"The judge who acts as an advocate skates on thin ice in any judicial hearing and runs the risk of turning the adversary system into an inquest wherein the trier of the fact calls and questions the witnesses." Id.; see also C.S. Wheatley, Jr., Annotation, Propriety of Trial Judge in Propounding Questions to Witnesses in Criminal Case, 84 A.L.R. 1172 (1933).

¶ 9. Accordingly, "potential prejudice lurks behind every intrusion a presiding judge makes into a trial." United States v. Slone, 833 F.2d 595, 597 (6th Cir. 1987) (citation omitted). In that vein, the Sixth Circuit identified three instances suggesting when a trial court has good reason to interject questions into a trial: (1) when a trial is lengthy and complex, calling for judicial intervention for clarification; (2) when counsel are unprepared or obstreperous and meddling with the facts, suggesting that judicial intervention is necessary for clarification; and (3) when a witness is difficult, confused and providing unbelievable information, that is not properly addressed. United States v. Hickman, 592 F.2d 931, 933 (6th Cir. 1979). Reviewing these factors can assist us in determining whether the trial court intervention was appropriate or prejudicial.

¶ 10. Neither the first nor second factor apply here. The dissent apparently argues that the third factor is present — that the trial court believed Carprue provided unbelievable information and that additional evidence was required to refute his claims. In addressing this factor, it is important to acknowledge that it is a far better policy "to allow counsel the opportunity to develop facts in the first instance" because counsel is in a better position to do so. See Auger v. Auger, 546 A.2d 1373, 1375 (Vt. 1988). The trial court here did not heed that advice; rather, the trial court interjected itself into *176the evidentiary proceedings before the State had an opportunity to present its rebuttal case.

¶ 11. The reason for the policy is premised on two grounds. First, the line is thin between judging and advocacy; a defendant's right to a fundamentally fair trial before a neutral bench is essential to the administration of justice. Second, as one judge noted:

The judge views the case from a peak of Olympian ignorance. His intrusions will in too many cases result from partial or skewed insights. He may expose the secrets one side chooses to keep while never becoming aware of the other's. He runs a good chance of pursuing inspirations that better informed counsel have considered, explored, and abandoned after fuller study. He risks at a minimum the supplying of more confusion than guidance by his sporadic intrusions.

Marvin E. Frankel, The Search for Truth: An Umpireal View, 123 U. Pa. L. Rev. 1031, 1042 (1975). The dissent suggests that the trial court here was fulfilling its obligation "to see to it that justice is done" and to "negate a portion of the defendant's testimony that did not ring true." Dissent op. at ¶ 10. This is an issue that repeatedly arises in cases across the nation. Although some courts and commentators agree with the philosophy espoused by the dissent, many others abide by the reasoning contained in the majority:

[JJudicial interrogation to search for the whole truth or to complete the evidence presupposes that the judge knows the truth better than the parties or their counsel.... There is something unseemly about a judge . . . who seeks to elicit testimony in aid of those answers.

Stephen A. Saltzburg, The Unnecessarily Expanding *177Role of the American Trial Judge, 64 Va. L. Rev. 1, 57-58 (1978). Here, the trial court's questioning demonstrated its belief that Carprue was not being forthright with respect to why he fled from police. We hold that when a trial court's questioning reveals its disbelief of the defendant's testimony, the court has crossed over the .line of impartiality. The trial court's conduct suggested favoritism toward the State and against the defendant and violated Carprue's due process rights.

¶ 12. The trial court is entrusted with the responsibility to conduct the trial in an "atmosphere of perfect impartiality," Glasser v. United States, 315 U.S. 60, 82 (1942), superseded on other grounds by Fed. R. Evid. 104(a), and must "never assume a position of advocacy, real or apparent," State v. Delarosa, 547 A.2d 47, 51 (Conn. App. Ct. 1988) (emphasis added).

¶ 13. Here, the trial court's conduct was not limited to "clarification," but resulted in the discovery of new information, challenged the credibility of the defendant, and threatened the fairness of the trial. In calling Morrow to testify, the trial court elicited information about "in-house" placement procedures as well as the information provided to inmates. The trial court questioned Morrow as to exactly what happens if the inmate violates the "in-house" rules associated with monitoring. As noted, Morrow testified that if a contact violation occurs, "in-house" notifies the court, not the police. This testimony was elicited from the court in direct response to Carprue's testimony that he was hiding from police not because he feared he would be arrested for sexual assault, but rather, because he knew he violated the "in-house" contact rules. The purpose of the trial court's independent questioning was to establish the procedures used by "in-house" when a violation *178occurs, and to demonstrate that Carprue should have known that the "in-house" violation would not result in sending the police to arrest him. The questioning by the trial court was adversarial because it was not simply done to clarify a matter that was confusing to the jury. Rather, the questioning by the court demonstrated the trial court's belief that Carprue was not being credible when he offered an explanation for fleeing from police.

¶ 14. By "target[ing Carprue's] credibility and challeng[ing] his story" the trial court acted "more in the manner of a prosecutor than an impartial judge." United States v. Filani, 74 F.3d 378, 385 (2d Cir. 1996). In Carprue's case, his credibility was a key issue. There were two versions of events presented in this case. One was Carprue's version, that no sexual assault occurred and he fled from the police only because he was violating "in-house" rules. The other was the State's and victim's version, that this was a forced sexual assault, evidenced by Carprue's fleeing from police because he did not want to be arrested for sexual assault.

¶ 15. The State responds that Carprue was not prejudiced as a result of the trial court's questioning because the interrogation was done outside the presence of the jury.2 We cannot agree. "Even without the presence of a jury to be influenced, the trial court has the duty to conduct proceedings in an impartial, deco*179rous manner, so as not to intimidate either party in presenting matters to the court or to demean the public image of judicial integrity." Bruce v. State, 375 N.E.2d 1042, 1056 (Ind. 1978) (citation omitted). Directly following the trial court's interrogation, the State called Morrow as a rebuttal witness. The State's questioning of Morrow mirrored that of the trial court. The State elicited testimony from Morrow as to the procedures utilized by "in-house" in the event an inmate violates the monitoring rules. Morrow provided the same testimony, this time in front of the jury, that "in-house" does not contact the police, but rather it notifies the court of a violation. Morrow also testified that each inmate is advised of these procedures, both verbally and in writing.

¶ 16. By sua sponte calling Morrow and questioning him on this subject, the trial court acted as a "hinting advocate" for the State. In other words, the trial court provided the prosecutor with information it believed the jury needed to hear in order to rebut Carprue's claims that he hid from police solely because he knew he had violated "in-house" rules, and not because he had just committed a sexual assault. Even the postconviction court determined that the "[trial court]'s active involvement in the trial was out of the ordinary. It was so active that it raised an unfortunate appearance that she was partial to the State, or at least partial against Mr. Ca[r]prue." The State suggests that the prosecutor would have called Morrow to offer the same rebuttal testimony even absent the involvement of the trial court. The record, however, reveals nothing to suggest that eventuality. Morrow was not listed as a *180witness by the State, and did not offer any testimony during the State's case-in-chief. Moreover, the State did not make any reference during the opening statement to suggest that it intended to call Morrow for such a purpose.

¶ 17. Because of the nature of this indiscretion, we cannot say with absolute certainty whether Morrow's testimony affected the jury's ultimate decision. It could be that Carprue's flight did not influence their decision at all. It could be that the victim's testimony was entirely more credible than Carprue's claim that the sexual act was consensual. Under the circumstances, however, we must apply the harmless error test to determine whether the trial court's conduct resulted in "a reasonable possibility that the error contributed to the conviction." See State v. Dyess, 124 Wis. 2d 525, 543, 370 N.W2d 222 (1985). The trial court's elicitation of Morrow's testimony inevitably suggested that Carprue's veracity was questionable. Although this was done outside the presence of the jury, the trial court's conduct communicated partisan information to the prosecutor, who immediately thereafter presented the substance of the trial court's actions to the jury during the rebuttal case. As a result, the jury was presented with evidence to doubt Carprue's credibility, which it otherwise may not have heard. When "such doubt is injected by the court in a case where credibility of a defendant-witness is a key issue, there has been a deprivation of a fair jury trial." United States v. Victoria, 837 F.2d 50, 55 (2d Cir. 1988). Due process and the right to an impartial tribunal are critically important to maintain the integrity of our justice system. The trial courts operate as a buffer between the adversarial conduct of the state, the protection of the *181rights of the accused, and the assurance to the community that justice will be served. A trial court must "exercise self-restraint and preserve an atmosphere of impartiality and detachment." Pariser v. City of New York, 146 F.2d 431, 433 (2d Cir. 1945). Our system requires that judges refrain from advocacy and advocates refrain from judging.

¶ 18. Thus, although we cannot say with absolute certainty that Carprue would have been acquitted absent the -trial court's intervention, we are compelled to the conclusion that a reasonable possibility of that outcome existed. It is evident that the trial court's conduct exceeded the parameters of acceptability, and rendered the trial fundamentally unfair. Under these facts and circumstances, we conclude that the trial court's actions steered the State toward evidence it otherwise might not have presented. This resulted in the appearance of advocacy and partisanship on the part of the trial court. The trial court's conduct exceeded the scope of authority intended by Wis. Stat. § 906.14 and crossed the "fine line" a trial court confronts when exercising its limited discretion to interrogate witnesses. See Garner, 54 Wis. 2d at 104. As a result, we have no choice but to conclude that Carprue's due process rights were violated. When the trial court engaged in partisanship and advocacy, it could no longer maintain impartiality and objectivity.

¶ 19. Although the dissent attempts to place blame with this court for reversing a conviction, which necessarily will result in implications for the victim, the issue in this case is not about assigning fault. It is often said that tough cases make bad law. It is our duty to avoid that cliché. Certainly, this court draws no pleasure from reversing a conviction in a sexual assault case, and we can empathize with what our decision *182means for the victim. However, our decision cannot be based on sympathy, but must be grounded in the constitutional principles upon which our system of justice was developed. Here, the requirements of proper process are not being proposed as a preventative to achieving justice. To the contrary, protestations by the dissent notwithstanding, proper process is quintessential to ensuring that ultimate justice is achieved. To the judiciary belongs this important and continuing obligation. Furthermore, because of the sensitive nature of sexual assault cases, it is incumbent upon the trial court to be vigilant in its duty and refrain from engaging in any conduct that suggests the court is partial to the State, or at least biased against the defendant. By failing to exercise such vigilance, the trial court runs the risk that the case may have to be retried. Based on the foregoing, the judgment and order in this case are reversed and the case is remanded for a new trial.3

By the Court. — Judgment and order reversed and cause remanded.

¶ 20 (dissenting). It is an oft-repeated cliché that a trial is a search for the truth. Consistent with that ideal paradigm, the rules of evidence permit trial judges to call and interrogate witnesses. See Wis. Stat. Rule 906.14.1 This is because, as Judge Learned *183Hand recognized almost sixty years ago, "[a] judge is more than a moderator; he is charged to see that the law is properly administered, and it is a duty which he cannot discharge by remaining inert." United States v. Marzano, 149 F.2d 923, 925 (2d Cir. 1945). This is also the law in Wisconsin:

[T]he trial judge is more than a mere referee. The judge does have a right to clarify questions and answers and make inquiries where obvious important evidentiary matters are ignored or inadequately covered on behalf of the defendant and the state. A judge does have some obligation to see to it that justice is done but must do so carefully and in an impartial manner.

State v. Asfoor, 75 Wis. 2d 411, 437, 249 N.W2d 529, 540-541 (1977) (emphasis added).

¶ 21. A trial judge in Wisconsin is given broad berth by the rules. As we have seen, Wis. Stat. Rule 906.14 is quite plain: "The judge may, on the judge's own motion. . ., call witnesses." Wis. Stat. Rule 906.14(1) (emphasis added). Further, "[t]he judge may interrogate witnesses, whether called by the judge or by a party." Wis. Stat. Rule 906.14(2) (emphasis added). Indeed, this power includes the discovery and development of new information. Thus, the trial judge has the authority to appoint expert witnesses — even on the "judge's own motion." Wis. Stat. Rule 907.06(1).

*184¶ 22. Of course, the trial" 'judge should not take an active role in trying the case for either the state or the defense.'" Asfoor, 75 Wis. 2d at 437, 249 N.W2d at 540 (quoting State v. Garner, 54 Wis. 2d 100, 104, 194 N.W.2d 649, 651 (1972)). Nevertheless, Asfoor affirmed the conviction in that case even though the judge "took a somewhat active role in questioning witnesses" in front of the jury. Asfoor, 75 Wis. 2d at 437, 249 N.W.2d at 540.

¶ 23. Garner, relied on by both Asfoor and the Majority here, is also instructive. There, the Wisconsin Supreme Court found no prejudice even though the trial judge in that case "served, in effect, as prosecuting attorney," during the probation-revocation hearing. Garner, 54 Wis. 2d at 104, 194 N.W.2d at 651. Indeed, the trial judge in that case "called and examined all the state witnesses himself." Ibid. Of course, Garner recognized:

A trial judge while exercising his limited discretion to call and examine a witness should not function as a partisan, as this could very well mislead the jury. But even where there is no jury, the judge should not take an active role in trying the case for either the state or the defense. The judge who acts as an advocate skates on thin ice in any judicial hearing and runs the risk of turning the adversary system into an inquest wherein the trier of the fact calls and questions the witnesses.

Ibid, (internal citation omitted). Nevertheless, even though the trial judge in Garner, as noted, "called and examined all the state witnesses himself," the supreme court found "no reversible error on these facts." Ibid.

¶ 24. An example where a trial judge falls through the "thin ice" of partisanship is Marzano. There, the trial judge not only called and interrogated witnesses who were awaiting sentence before him, but also, by his *185conduct of the questioning in front of the jury, made the jury believe that there was evidence beyond what they heard in open court that indicated that the defendant was guilty.

[I]t must have appeared to the jury extremely probable that, not only was he satisfied that [the witnesses he called and questioned] were accomplices [of the defendant], but that he was relying upon more than the testimony of the [other witnesses who were alleged to have been drug "decoys"]; that he had other grounds for his belief which had not come out upon the trial; for he would scarcely have gone the lengths which he did, merely upon such testimony.

Marzano, 149 F.2d at 926.

¶ 25. There are two matters that, in my view, are critical to this appeal. First, the jury neither saw nor knew that the trial court called and interrogated Kenneth Morrow. Thus, this case is different than all but one of the foreign cases the Majority marshals in support of reversal. Only Auger v. Auger, 546 A.2d 1373 (Vt. 1988), was a non-jury trial. Moreover, of all the jury-trial cases cited by the Majority, only United States v. Victoria, 837 F.2d 50 (2d Cir. 1988); United States v. Hickman, 592 F.2d 931 (6th Cir. 1979); and State v. Delarosa, 547 A.2d 47 (Conn. App. Ct. 1988), reversed convictions because of the trial judges' alleged lack of impartiality.

¶ 26. Second, Johnnie Carprue's trial lawyer never objected. See Wis. Stat. Rule 906.14(3) (specifying when a party may object "to the calling of witnesses by the judge"). Thus, this case must be evaluated in the context of whether Carprue was deprived of his right to effective assistance of counsel. Kimmelman v. Morrison, 477 U.S. 365, 374-375 (1986); State v. Damaske, 212 Wis. 2d 169, 200, 567 N.W.2d 905, 919 (Ct. App. *1861997). The Majority glissades over the first and it ignores the second, which, in my view, is dispositive.

¶ 27. To prevail here, Carprue must show both that his trial lawyer's performance was deficient, and that "the deficient performance prejudiced the defense." Strickland v. Washington, 466 U.S. 668, 687 (1984); see also State v. Sanchez, 201 Wis. 2d 219, 236, 548 N.W.2d 69, 76 (1996). Again, this is the test that we must apply, see Kimmelman, 477 U.S. at 374-375, not "harmless error." Assuming that Carprue's lawyer's failure to object to the trial court's calling Kenneth Morrow out of the presence of the jury was less-than "reasonably effective representation," State v. McMahon, 186 Wis. 2d 68, 80, 519 N.W.2d 621, 626 (Ct. App. 1994), Carprue has, in my view, utterly failed to show prejudice — that is, he has not even come close to demonstrating that " 'but for counsel's unprofessional errors, the result of the proceeding would have been different.'" Sanchez, 201 Wis. 2d at 236, 548 N.W2d at 76 (quoting Strickland, 466 U.S. at 694).

¶ 28. The evidence here was strong; the Majority does not opine that it was not. Further, the Majority recognizes that proof of "prejudice" in the Strickland sense has not been shown — the Majority concedes: "[W]e cannot say with absolute certainty whether Morrow's testimony affected the jury's ultimate decision. It could be that Carprue's flight did not influence their decision at all. It could be that the victim's testimony was entirely more credible than Carprue's claim that the sexual act was ■ consensual." Majority op. at ¶ 14. A defendant loses an ineffective-assistance-of-counsel claim when the prejudice/no-prejudice equation is so balanced.

¶ 29. Clearly, the Majority does not like what the trial court did in "hinting" to the State that it should call *187a witness, who was already in the courtroom, to negate a portion of the defendant's testimony that did not ring true. See Majority op. at ¶ 13. I disagree. In my view, silence by the trial judge in the face of something that she knew was wrong would háve been an abdication of her obligation "to see to it that justice is done" by making "inquiries where obvious important evidentiary matters are ignored or inadequately covered." Asfoor, 75 Wis. 2d at 437, 249 N.W.2d at 540-541. In my view, the "search for the truth" should not be a sporting event, where possible miscues can let the guilty go free:

Over 75 years ago, Roscoe Pound condemned American courts for ignoring "substantive law and justice," and treating trials as sporting contests in which the "inquiry is, Have the rules of the game been carried out strictly?" Pound, The Causes of Popular Dissatisfaction With the Administration of Justice, 29 ABA Ann. Rep. 395, 406 (1906). A criminal trial is not a "game."

Morris v. Slappy, 461 U.S. 1, 15 (1983). I agree. Wisconsin law, Asfoor and Garner, amply support what the trial court did here. It is that law, and not foreign law upon which the Majority relies, that governs this case. State v. Lossman, 118 Wis. 2d 526, 533, 348 N.W.2d 159, 163 (1984) (court of appeals bound by supreme court precedent).

¶ 30. The victim in this case claimed that she was raped. A jury believed her. The Majority now makes her endure either another trial or seeing the man who raped her get a plea-bargained deal. A victim's trauma on any retrial is an appropriate consideration in deciding whether marginal errors warrant a new trial. Morris, 461 U.S. at 14-15. In my view, however, there were no errors, and I would affirm.

*188¶ 31. I respectfully dissent.