dissenting.
I would hold that the rejection by the Michigan Court of Appeals of Ferensic’s claims of error as to the testimony of witnesses Shulman and St. John was neither contrary to nor an unreasonable application of federal law. I would therefore reverse the district court’s grant of the writ of habeas corpus.
I. RIGHT TO PRESENT A DEFENSE
As the majority notes, in Taylor v. Illinois, the Supreme Court recognized that the Sixth Amendment affords a defendant the right “to compel a witness’[s] presence in the courtroom” and “to offer testimony.” 484 U.S. at 409, 108 S.Ct. 646. However, the right is not absolute; “[t]he accused does not have an unfettered right to offer testimony that is incompetent, privileged, or otherwise inadmissible under standard rules of evidence. The Compulsory Process Clause provides him with an effective weapon, but it is a weapon that cannot be used irresponsibly.” Id. at 410, 108 S.Ct. 646. As the Court further explained in United States v. Scheffer,
state and federal rulemakers have broad latitude under the Constitution to establish rules excluding evidence from criminal trials. Such rules do not abridge an accused’s right to present a defense so long as they are not “arbitrary” or “disproportionate to the purposes they are designed to serve.” Moreover, we have found the exclusion of evidence to be unconstitutionally arbitrary or disproportionate only where it has infringed upon a weighty interest of the accused.
523 U.S. at 308, 118 S.Ct. 1261 (quoting Rock, 483 U.S. at 56, 107 S.Ct. 2704).
A. Dr. Shulman
In considering Ferensic’s claim that the exclusion of Shulman’s testimony violated his right to present a defense, the Michigan Court of Appeals applied the following standard:
When fashioning a remedy for noncompliance with a discovery order, a trial court must first determine whether the objecting party’s interest in preparing its own case or its opportunity to test the authenticity of its opponent’s evidence has been prejudiced by the *485noncompliance, and then, if that be the case, must consider what remedy may be appropriate, giving due regard to the competing interests of the opposing party, the court and the public. The trial court must also inquire into all the relevant circumstances, including the reasons behind noncompliance and whether the objecting party was actually prejudiced. Finally, the remedy for noncompliance should not put the objecting party in a better position than the party would have enjoyed if the discovery order had been complied with. Thus, the exclusion of otherwise admissible evidence should be limited to the “most egregious cases,” in which other less severe remedies would fail to protect the parties’ competing interests.
2001 WL 865089, at *1 (citations omitted) (citing People v. Davie (After Remand), 225 Mich.App. 592, 571 N.W.2d 229, 231-32 (1997); People v. Taylor, 159 Mich.App. 468, 406 N.W.2d 859, 868-69 (1987)). As stated above, the Supreme Court analyzes the exclusion of defense evidence pursuant to evidentiary rules to determine whether the rules are “‘arbitrary’ or ‘disproportionate to the purposes they are designed to serve,’ ” and to determine whether “the exclusion of evidence ... has infringed upon a weighty interest of the accused.” Scheffer, 523 U.S. at 308, 118 S.Ct. 1261.
The majority concludes principally that the Michigan Court of Appeals unreasonably applied established law, but also asserts, as to the exclusion of St. John’s testimony, that prejudice is not “the proper test for analyzing a right-to-present-a-defense claim.” 1 Opinion at 476. However, the Michigan Court of Appeals did not perform a prejudice analysis in evaluating Ferensic’s challenge to the exclusion of Shulman’s testimony: Rather, the court held that “the ruling did not put the prosecution in a better position than it would have enjoyed had the order been complied with because defense counsel was otherwise able to effectively challenge inconsistencies in the victims’ identification testimony.” 2001 WL 865089, at *2. This is merely an application of the Michigan rule, quoted above, that “the remedy for noncompliance should not put the objecting party in a better position than the party would have enjoyed if the discovery order had been complied with.” This rule seems to be a fair rendition of the first Scheffer principle that evidentiary rules must not be “disproportionate to the purposes they are designed to serve.” The Michigan rule that a decision to exclude evidence “must consider what remedy may be appropriate, giving due regard to the competing interests of the opposing party, the court and the public” would seem to encompass the other portion of the federal rule, namely, that exclusion not be “arbitrary.”
The majority also concludes that the trial court’s exclusion of Shulman’s testimony was arbitrary and disproportionate to the discovery violation, and that the Michigan Court of Appeals unreasonably *486applied federal law in holding to the contrary. The trial court excluded Shulman’s testimony in response to an oral motion by the prosecution on the first day of trial, June 28, 1999. The prosecution had originally moved to exclude this evidence in a pretrial motion which the trial court denied on December 18, 1998; the court ordered instead that Shulman would be permitted to testify if the defense provided the prosecution with Shulman’s report two months before trial. In the brief hearing on the prosecution’s later motion to exclude Shulman’s evidence, the trial court ruled,
It doesn’t make a difference [that defense counsel sent the report to the prosecution the day he received it]. It was your responsibility. If you’re going to call somebody, and you’re under a court order to get the report to the prosecution two months before trial, especially when I ordered it back in December, you were supposed to make that person get you the report. If not, you should have come into court and said, “Judge, we need more time so that I can get the report to the prosecution.” You will not be able to use it.
Tr. of June 28,1999, at 193.
In view of the significant time in which the defense could have complied with the court order, the exclusion of Shulman does not appear to be arbitrary, nor disproportionate to the purpose for which the original order was made. As the trial court explained when the defense sought reconsideration of the decision at the close of testimony on the first day of trial, “I ordered you to give it to him two months ahead of time so that he could get his own expert. I wanted to have enough time for the prosecution to be prepared to try this case without having to have an adjournment.” Id. at 307-08, 118 S.Ct. 1261. This is precisely the reason the Supreme Court recognized in Taylor v. Illinois as an appropriate reason to exclude defense evidence:
The adversary process could not function effectively without adherence to rules of procedure that govern the orderly presentation of facts and arguments to provide each party with a fair oppoHunity to assemble and submit evidence to contradict or explain the opponent’s case. The trial process would be a shambles if either party had an absolute right to control the time and content of his witnesses’ testimony.
484 U.S. at 410-11, 108 S.Ct. 646 (emphasis added).
As the majority points out, unlike Taylor, this case does not involve a defendant who announces the existence of witnesses only after the trial has begun. See id. at 403, 108 S.Ct. 646. In this case, the trial court gave its order regarding Shulman’s testimony more than six months before trial, and Shulman was not a witness belatedly discovered by the defense. According to Shulman’s testimony before the district court, defense counsel had actually retained him as an expert witness in 1998, and Shulman had initially been prepared for a trial date of September 22, 1998; the trial was subsequently postponed. Shul-man’s testimony that in his discussions with Ferensic’s counsel after that date counsel made a “request for an updated report or a revised report” indicates that Shulman had already provided a report by September of 1998, which apparently Fer-ensic desired Shulman to modify in some fashion. Ferensic thus should have had no difficulty in complying with the discovery order.
Moreover, as the trial court pointed out, the December order gave Ferensic four months to provide the prosecution with the required report, and he could have asked the court for an extension if this for some reason proved impossible to do. The trial *487court could, of course, have permitted a mid-trial adjournment in which the prosecution could obtain its own expert witness. However, the avoidance of such an adjournment was precisely the reason the trial court ordered that defendant provide Shulman’s report two months before trial. As the Supreme Court noted in Taylor, “a less drastic sanction is always available.” 484 U.S. at 413, 108 S.Ct. 646. However, that does not mean that a less drastic sanction must always be used. Given the situation, the exclusion of Shulman’s testimony was not arbitrary, and not disproportionate to the purpose the order was designed to accomplish, which was that it would not be necessary to adjourn the trial in order for the prosecution to respond to Shulman’s evidence. The Michigan Court of Appeals’s conclusion that the exclusion of Shulman was “nothing more than the court had warned of six months before trial,” 2001 WL 865089, at *2, was reasonable, and a reasonable application of the standard governing the exclusion of defense evidence.
The majority also argues that the apparent purpose of the trial court’s order, namely, ensuring that the prosecution have an opportunity to secure its own witness, was not a genuine one; that is, that the trial court was “indifferen[t] to whether the prosecution was actually prejudiced,” and excluded Shulman’s testimony without regard to whether the prosecution actually desired to secure its own eyewitness identification expert. Opinion at 477. However, this also does not present the full picture of the proceedings before the trial court. The trial court explained,
[0]ne of the things I’ve got to do is I’ve got to make sure that whatever you get ready to put in evidence, the other side has had the opportunity to prepare to do something about it. If not, I have to stop in the middle of the trial and give them time. Now, you know I’m not going to do that in this case, don’t you? You know I’m not going to let you put in your expert, and then adjourn this case for two months to let the prosecutor go running around trying to find another expert that will testify differently. You know that’s not going to happen, don’t you?
Tr. of June 28, 1999, at 308-09. Defense counsel responded, “Well, I’ve never had any indication from [the prosecutor] that he wants to hire an expert to contradict our expert.” Id. at 309. The trial court replied, “I don’t care what he — what I’m trying to do is I’m making sure — when I gave the order, the reason for the two months was so I would not run into this problem, and you didn’t follow the order. I’m sorry.” Id.
While the trial court certainly indicated its indifference to defense counsel’s statements regarding whether the prosecutor had ever indicated an interest in securing an opposing expert, this does not entail indifference to the effect of its ruling on the prosecution. The court clearly was concerned that it might “have to stop in the middle of trial and give [the prosecution] time,” and was enforcing its order “so I would not run into this problem.” Id. at 308, 309. The fact that the prosecutor had not then represented to defense counsel that he intended to secure an opposing expert does not mean that the need to do so might not later appear “in the middle of trial,” necessitating the adjournment which the court’s order had carefully avoided.
In addition, the Supreme Court in Schef-fer held that “we have found the exclusion of evidence to be unconstitutionally arbitrary or disproportionate only where it has infringed upon a weighty interest of the accused.” 523 U.S. at 308, 118 S.Ct. 1261. Supreme Court precedent, including the three cases Scheffer specifically cites in support of this proposition, indicate that a “weighty interest of the accused” does not *488include any interest in offering evidence, but is limited to an interest in offering the accused’s own testimony, Rock, 483 U.S. at 58. n. 15, 107 S.Ct. 2704, or an interest in using the only available means of addressing a relevant issue, see, e.g., Chambers v. Mississippi, 410 U.S. 284, 300-02, 93 S.Ct. 1038, 35 L.Ed.2d 297 (1973) (defendant forbidden to offer evidence that another person had confessed on several different occasions to the crime of which he stood accused); Washington v. Texas, 388 U.S. 14, 22, 87 S.Ct. 1920, 18 L.Ed.2d 1019 (1967) (defendant forbidden to offer exculpatory testimony from a witness because the witness was an accomplice in the crime and had not been acquitted). The majority responds that the list of “weighty interests” is not so limited. Even assuming these Supreme Court decisions do not constitute an exhaustive list of “weighty interests” and different interests could meet the weighty-interest standard in the future, it cannot be said that any other interests are clearly established federal law as required under AEDPA.
The Michigan Court of Appeals correctly noted that in this case, the exclusion of Shulman’s testimony did not prevent Fer-ensic from addressing the issue of flaws in eyewitness identifications; “defense counsel was otherwise able to effectively challenge inconsistencies in the victims’ identification testimony” through cross-examination of the victims. 2001 WL 865089, at *2. Habeas relief stemming from the exclusion of Shulman’s testimony is thus also inappropriate because the exclusion did not infringe on a “weighty interest of the accused.”
B. Danny St. John
The Michigan Court of Appeals analyzed. Ferensic’s claim regarding the trial court’s failure to grant an adjournment to permit the appearance of St. John under the following standard:
In determining whether a trial court abused its discretion in denying a motion for an adjournment, this Court should consider “whether defendant (1) asserted a constitutional right, (2) had a legitimate reason for asserting the right, (3) had been negligent, and (4) had requested previous adjournments.” Defendant must also show that he was prejudiced by the denial of the adjournment. The witness in question would have testified to some differences in identification of the offenders and the vehicle they were driving. The purported testimony was not especially strong, however, and inconsistency inherent in the victims’ identification was otherwise shown. Accordingly, defendant has failed to satisfy the requirement that he show prejudice.
2001 WL 865089, at *2 (citations omitted) (quoting People v. Lawton, 196 Mich.App. 341, 492 N.W.2d 810, 814 (1992)).
The majority primarily criticizes the Michigan Court of Appeals’s decision to affirm on the basis that it unreasonably applied clearly established federal law. However, the majority also asserts that the Michigan Court of Appeals applied harmless error analysis in disposing of the argument, and implies that since Scheffer requires inquiry into whether evidentiary rules are arbitrary or disproportionate to the purposes they are designed to serve, the Michigan Court of Appeals analysis is contrary to clearly established federal law. See Opinion at 481-83. This is not true. As the Supreme Court has explained, avoiding contrariness to federal law “does not require citation of our cases — indeed, it does not even require awareness of our cases, so long as neither the reasoning nor the result of the state-court decision contradicts them.” Early v. Packer, 537 U.S. 3, 8, 123 S.Ct. 362, 154 L.Ed.2d 263 (2002). The majority points to no Supreme Court *489case forbidding the application of harmless error analysis.
The majority also concludes that the Michigan Court of Appeals unreasonably applied federal law to Ferensic’s claim regarding the exclusion of St. John’s testimony. The Court of Appeals relied on defense counsel’s proffer of evidence at trial to determine that St. John’s testimony “was not especially strong,” and therefore that “defendant has failed to satisfy the requirement that he show prejudice.” 2001 WL 865089, at *2. This evaluation of the proffered evidence is confirmed by St. John’s testimony at the evidentiary hearing before the district court below. St. John testified that he was “laying carpet” in one of the homes on the victims’ street, and his attention was drawn to a truck because “the truck that was parked in front of the [victims’] house, my neighbor’s got one exactly like it.” Tr. of June 19, 2006, at 6. He saw two men standing outside the truck; both men were white, and “[o]ne looked like he was probably in his early forties.” Id. at 7-8, 18. He confirmed that he was “not able to provide a full description of either one of these individuals.” Id. at 8. On cross-examination by the government, he clarified that he “wasn’t actually parked in the driveway outside” when he saw the men and the truck; he was driving and “passed it on the street,” and he got “[j]ust a glance” at the two men. Id. at 12-13. St. John also answered “No, sir” to the district court’s question, “Never saw him before?” referring to the defendant sitting in court at the evidentiary hearing. Id. at 13.
The import of St. John’s testimony was thus that he got very little chance to look at either of the two men and could not describe them. In view of this evidence, it is not clear that St. John’s testimony at the trial would have had any probative value at all, let alone sufficient probative value to outweigh the considerations of delay attendant upon granting an adjournment. The district court below concluded that “Mr. St. John also would have helped the defense, because his description of a suspect with curly black hair matched the description that the Kostoffs gave the police. At trial, the Kostoffs denied, telling the police that Petitioner had curly black hair.” Ferensic v. Birkett, 451 F.Supp.2d 874, 883 (E.D.Mich.2006). It is true that at trial, the Kostoffs denied describing the robber as having curly black hair. However, St. John’s testimony at the evidentiary hearing does not include such a description of either man. Perhaps more tellingly, Ferensic’s counsel never even asked whether St. John would describe either man in this way, or what color hair either had. The only suggestion that St. John would have so testified was counsel’s proffer at trial.
Even supposing that St. John would have testified that one of the men had black curly hair, and that counsel below simply forgot to ask him about this description, that testimony would at most have contradicted Mr. Kostoffs testimony that he never described one of the robbers as having black, curly hair. However, the fact that the Kostoffs initially described one of the robbers as having black, curly hair appears in the police report, which was used for impeachment at trial. This would seem to be a considerably stronger refutation of their testimony that they did not describe the robber that way than would St. John’s testimony that one of the robbers had black, curly hair. Moreover, Ferensic does not contend that St. John’s testimony would actually have indicated that neither robber resembled him, or otherwise tended to exonerate him. The conclusion of the Michigan Court of Appeals that St. John’s testimony “was not especially strong” is supported by the record.
The majority’s assertion that, on the other hand, Ferensic “was asking for the *490briefest of delays” in order to present St. John’s testimony does not tell the whole story. It is true that Ferensic finished his evidence at 10:25AM and trial counsel stated that St. John was due to arrive at the courthouse at 11:00AM. However, the trial court also noted for the record that “we were supposed to start at 9:00 this morning. Your witness was not here. I told you to put on your next witness. He’s not here.” Tr. of June 29, 1999, at 73-74. Two of the scheduled defense witnesses were thus not present when proceedings began for the day, even though defense counsel knew that there was an outstanding question as to permissible impeachment testimony which would determine whether- or not Ferensic himself elected to testify. In fairness, Ferensic did have one witness available to testify that morning, but that witness was Ferensic’s father, who likely was in attendance throughout the trial.
In its full context, the trial court’s refusal to grant an adjournment until 11:00 when St. John was expected to arrive does not appear the “ ‘myopic insistence upon expeditiousness in the face of a justifiable request for delay’ ” that the majority terms it. On the contrary, in view of the absence of multiple witnesses from the beginning of proceedings that morning and the very limited helpfulness of St. John’s testimony, even assuming it would have been consistent with the proffer, the district court’s denial of a continuance was not arbitrary or disproportionate to the interest it was designed to serve, and the Michigan Court of Appeals did not unreasonably apply established law in affirming Ferensic’s conviction on that basis.
Because the Michigan Court of Appeals decision did not unreasonably apply clearly established federal law, I believe that the district court was not authorized to grant a writ of habeas corpus. See Taylor, 288 F.3d at 850. A harmless error review, including the cumulative prejudice analysis of the majority, is therefore unnecessary. See Penry v. Johnson, 532 U.S. 782, 795, 121 S.Ct. 1910, 150 L.Ed.2d 9 (2001) (“Even if our precedent were to establish squarely that the prosecution’s use of the Peebles report violated Penry’s Fifth Amendment privilege against self-incrimination, that error would justify overturning Penry’s sentence only if Penry could establish that the error ‘had substantial and injurious effect or influence in determining the jury’s verdict.’ ”) (emphasis added) (quoting Brecht, 507 U.S. at 637, 113 S.Ct. 1710); see also Fulcher v. Motley, 444 F.3d 791, 808-09 (6th Cir.2006) (performing harmless error review only after finding that habeas relief was otherwise warranted); Biros v. Bagley, 422 F.3d 379, 388 (6th Cir.2005) (same); Stapleton v. Wolfe, 288 F.3d 863, 867 (6th Cir.2002) (same).2
II. INEFFECTIVE ASSISTANCE OF COUNSEL
Although the majority does not reach the issue, because I would hold that Fer-*491ensic is not entitled to habeas relief on the basis of his right to present a defense argument, I will briefly address his second issue, ineffective assistance of counsel. See McCalvin v. Yukins, 444 F.3d 713, 724 (6th Cir.2006) (Cole, J., dissenting) (addressing in dissent a secondary argument for habeas relief not reached by the majority). The standard for determining whether a defendant has received ineffective assistance was set forth by the Supreme Court in Strickland v. Washington. Under that standard, a defendant claiming ineffective assistance must show both “that counsel’s representation fell below an objective standard of reasonableness” and that “there is a reasonable probability that, but for counsel’s unprofessional errors, the result of the proceeding would have been different.” 466 U.S. at 688, 694, 104 S.Ct. 2052. Though the Michigan Court of Appeals did not cite Strickland by name, it analyzed whether “counsel’s failure to secure the testimony of these two witnesses was objectively unreasonable” and whether “absent this deficient performance the outcome of his trial would have been different.” 2001 WL 865089, at *2. This clearly is the correct standard.
The Michigan Court of Appeals determined that counsel’s actions were objectively unreasonable, but that Ferensic did not meet the prejudice prong. Id. With regard to St. John’s testimony, this seems difficult to dispute. According to his testimony at the evidentiary hearing, he could have testified at trial only that the two men were white, one was in his mid-forties, and he could further describe neither of them. The helpfulness of this testimony to Ferensic is at best negligible; it could fairly be argued that by confirming the presence of two white men near a black truck in front of the victims’ home, the victims’ testimony was corroborated. The fact that Ferensic was considerably younger than forty does not make St. John’s testimony exculpatory, since he had no idea of the age of the second man. Even if St. John had been prepared to testify that one of the men had black, curly hair, it is hard to see that there is a reasonable probability that with this testimony, the outcome of the trial would have been different. St. John would have described one of the robbers as having black, curly hair; the victims would have denied that they ever described either of the robbers in this way; and no witness would have described the second man in a manner inconsistent with Ferensic’s appearance. This is not sufficient to show prejudice under Strickland.
With regard to Shulman’s testimony, the prejudice question is perhaps closer. Certainly, as the majority notes, expert testimony on eyewitness identifications is now widely admissible in court. However, Michigan courts have held that an eyewitness identification expert is not constitutionally required. People v. Cooper, 236 Mich.App. 643, 601 N.W.2d 409, 418 (1999) (holding that the decision not to call an identification expert is not objectively unreasonable). The Sixth Circuit in unpublished decisions has also held that counsel’s failure to offer an identification expert is not ineffective assistance. Dorch, 105 Fed.Appx. at 656-657 (approving state appellate court decision that habeas petitioner “could not satisfy these two [Strickland ] elements” given that trial counsel “ ‘presented several witnesses who testified as to [the petitioner’s] whereabouts on the weekend of the incident’ ” and cross-examined the identifying witness); Tipton, 1996 WL 549802, at *1, 2 (holding that a federally convicted habeas petitioner “was not prejudiced by any allegedly ineffective assistance provided by his counsel” although counsel did “not hir[e] an expert in eyewitness identification”).
Moreover, the government showed during the evidentiary hearing that Shulman’s *492evidence undermining the victims’ testimony would be equivocal at best. Shulman conceded that if one witness identified a suspect from a lineup and the other failed to do so, that would be “a sign of independence” of their memories, indicating that their visual memory was not contaminated by “collaboration” regarding what the robber looked like. Tr. of June 19, 2006, at 47-48. He conceded that if the robber did not brandish a weapon when Mr. Kostoff first saw him, “there’s no weapons focus for that period of time, and that gives that witness a better opportunity to focus on the face of the person.” Id. at 48. He also conceded that if the interaction between the witness and the robber took place over a forty-five minute period and involved several encounters, not just an initial meeting at the door, the witness’s memory would be more accurate. He conceded that a witness’s initial verbal description of a suspect to a police officer would not be influenced by his later memories of seeing the suspect in a lineup or a photo array, and would therefore be relatively reliable. Thus, Shulman conceded that many if not most of the factors that he described as limiting a witness’s ability to identify a stranger accurately did not apply under the circumstances in this case.
Finally, trial counsel did thoroughly cross-examine the victims regarding their identification of Ferensic, as well as questioning the police witnesses about the composite sketch, photo array, and lineup, and offered the testimony of Ferensic’s father regarding scars on Ferensic’s face, which the victims did not remember. Given that Shulman’s testimony would have suggested at most slight potential weaknesses in the victims’ identification of Ferensic, and that defense counsel thoroughly argued the issue of identification, it does not appear reasonably probable that Shulman’s testimony would have altered the outcome of the trial.
I do not agree with the majority that the note the jury sent out during deliberations asking to see the police report indicates that the jury “had doubts about the strength of the case against Ferensic.” Opinion at 483. Although, as the majority observes, the police report contained the artist’s sketch, it contained other information as well, and the record contains no indication that the jury was interested in the sketch particularly. In any case, the trial court refused to provide the jury with the report, as it had not been entered into evidence, and after further deliberations, the jury found Ferensic guilty.
The finding of the district court below that “the record indicates that the jurors were unable to agree on a verdict at one point during their deliberations” is not helpful to a prejudice analysis. First of all, findings of fact by the district court are entitled to no deference if they are based entirely on the record from the state court proceedings. Taylor, 288 F.3d at 850. Second, the “one point” during deliberations at which the jurors requested the police report occurred after they had been deliberating for only ninety minutes, as the trial court noted on the record. Finally, and most importantly, the jury’s question does not indicate that they entertained doubts about their ultimate decision. This case is readily distinguishable from this court’s decision in Powell v. Collins, 332 F.3d 376, 401 (6th Cir.2003), in which the jury had “informed the court that it was ‘at a stalemate’ and could not agree whether to impose a death sentence.” Here, the jurors merely asked for information, which is not an indication that the verdict is unreliable, but a part of the appropriate role of the jury. See Weeks v. Angelone, 528 U.S. 225, 235-26, 120 S.Ct. 727, 145 L.Ed.2d 727 (2000) (“This particular jury demonstrated that it was not too shy to ask questions....”).
*493The determination of the Michigan Court of Appeals that the outcome would not have been different is thus not an unreasonable application of Strickland, especially in view of this court’s deference to state court application of federal law in habeas cases. As the Supreme Court held in Wiggins v. Smith, “In order for a federal court to find a state court’s application of our precedent ‘unreasonable,’ the state court’s decision must have been more than incorrect or erroneous. The state court’s application must have been ‘objectively unreasonable.’ ” 539 U.S. at 520-21, 123 S.Ct. 2527 (citations omitted) (quoting Williams, 529 U.S. at 409, 120 S.Ct. 1495). The conclusions of the Michigan Court of Appeals were not objectively unreasonable.
III. CONCLUSION
For these reasons, I would reverse the district court’s grant of the writ of habeas corpus.
. Presumably, this assertion, like the contention that the Michigan Court of Appeals applied the wrong standard in examining the exclusion of St. John’s testimony, is directed toward the requirement that habeas relief not be granted unless the state court decision is “contrary to ... clearly established federal law.” 28 U.S.C. § 2254(d)(1). However, the Supreme Court has explained that a state court decision is “contrary to” federal law only if “the state court arrives at a conclusion opposite to that reached by this Court on a question of law or if the state court decides a case differently than this Court has on a set of materially indistinguishable facts.” Williams v. Taylor, 529 U.S. at 413, 120 S.Ct. 1495 (O’Connor, J., delivering the opinion of the Court as to this part). The majority does not assert that the Supreme Court has concluded that a prejudice analysis is not appropriate in a right-to-present-a-defense case, nor that the Supreme Court has ever been presented with facts materially indistinguishable from those here.
. In Eddleman v. McKee, 471 F.3d 576, 583 n. 3 (6th Cir.2006), this court recently held that "the Brecht standard continues to apply when federal courts must make a harmless-error determination in the first instance, as when a state court found no error and therefore did not address whether an error would be harmless.” However, Eddleman cites for this proposition only cases, including Pemy, Ful-cher, Biros, and Stapleton, that either have performed a harmless error analysis after finding a constitutional error, or that have held any error to be harmless and not reached the question of whether constitutional error occurred. Thus, the rule enunciated in Eddleman apparently means that, regardless of whether the state court performed a harmless error analysis, Brecht applies where the federal court on habeas review has found constitutional error, but not where it has found none.