(concurring in part and dissenting in part). I concur with my colleagues that the rebuttal testimony was erroneously admitted. However, I believe that the harmless error standard for preserved nonconstitutional error should remain harmless beyond a reasonable doubt. I further believe that the error in this case was not harmless under any “lesser” *223standard. Nevertheless, in order to provide the bar with one standard adopted by a majority of this Court, I would concur with the standard proposed by Justice Levin in his separate opinion.
i
Today, by distinguishing nonconstitutional error from constitutional error, this Court has added yet another category to the labyrinth of harmless error in Michigan. It has done so without explaining why, or how, the uniform standard that this Court adopted in People v Robinson, 386 Mich 551; 194 NW2d 709 (1972), is unsatisfactory.1 Robinson restated the longstanding definition of harmless error:
The appropriate considerations are described in People v Wichman, 15 Mich App 110, 116 [166 NW2d 298] (1968):
*224“Where it is claimed that error is harmless, two inquiries are pertinent. First, is the error so offensive to the maintenance of a sound judicial process that it never can be regarded as harmless? See People v Bigge, 288 Mich 417, 421 [285 NW 5] (1939); People v Berry, 10 Mich App 469, 474 [157 NW2d 310] (1968); People v Mosley, 338 Mich 559, 566 [61 NW2d 785] (1953). See also Chapman v California, 386 US 18, 23, 24 (87 S Ct 824; 17 L Ed 2d 705 [1967]), rehearing denied 386 US 987 (87 S Ct 1283; 18 L Ed 2d 241) [(1967)]. Second, if not so basic, can we declare a belief that the error was harmless beyond a reasonable doubt? See People v Liggett, 378 Mich 706, 716, 717 [148 NW2d 784] (1967); Chapman v California, supra.” [Robinson, 386 Mich 562-563.]
This Court later unanimously applied the Robinson standard in a case with strong similarities to the instant case. People v Belenor, 408 Mich 244, 247; 289 NW2d 719 (1980). The Belenor Court held that error, like that which occurred here, was to be judged under the Robinson harmless beyond a reasonable doubt standard. Id. at 247. The Court did not engage in any discussion of whether this improper impeachment evidence was nonconstitutional error or constitutional error (perhaps as prosecutorial misconduct or as a violation of the right to a fair trial). Again, the Belenor Court must be presumed to have been aware of the distinction between nonconstitutional error and constitutional error and have concluded that the Robinson harmless error standard applied to all error that was amenable to harmless error analysis. See also People v Howe, 392 Mich 670, 678; 221 NW2d 350 (1974) (jury request to reread testimony).
Even if we were writing on a clean slate today, I would adopt a uniform harmless error standard for all *225errors that are subject to harmless error analysis.2 The line between evidentiary error and constitutional error is rarely clear. One prominent example is the case law involving the statement against penal interest exception to the hearsay rule. MRE 804(b)(3). That particular “evidentiary” rule’s outer parameters are directly determined by the constitutional ramifications. For instance, if the statement is used to inculpate the accused, the defendant’s constitutional right of confrontation is implicated. People v Poole, 444 Mich 151; 506 NW2d 505 (1993). In contrast, if the statement is used to exculpate the accused, the defendant’s constitutional right to present a defense is implicated. Chambers v Mississippi, 410 US 284; 93 S Ct 1038; 35 L Ed 2d 297 (1973). When the issue is whether a hearsay statement should be admitted under MRE 804(b)(3), the constitutional concerns are inseparable from the determination whether the trial court erred in its evidentiary ruling. People v Barrera, 451 Mich 261, 279; 547 NW2d 280 (1996).
Additionally, by holding today that there is a difference between nonconstitutional error and constitutional error, this Court is opening the doors to increasing litigation, by adding yet another issue to an appellate court’s analysis. This Court today telegraphs that there must be a significant difference between the harmless error standards for constitutional error and for nonconstitutional error and that the difference between the two standards will be outcome-determinative — for if the difference is not outcome-*226determinative, then this Court has no need today to distinguish a separate standard. After today, if the Court of Appeals applies the wrong standard, and my colleagues have yet to agree on the right standard, a remand, if not reversal, by definition, will be required because the difference between the standards is outcome-determinative. Likewise, if the defense counsel erroneously argues that the error falls within the wrong category, there will be an additional, and meritorious, claim of ineffective assistance of counsel. My colleagues have failed to explain persuasively why this nebulous maze is necessary.
Moreover, I believe that the adoption of a “lesser” standard with respect to nonconstitutional error implicitly threatens constitutional rights. I believe that for those errors that are amenable to harmless error analysis, the harmless error standard should remain: the nontainted evidence was so overwhelming and the error was so relatively insignificant, that the appellate court is led to the conclusion that the actual jury would have convicted regardless of the error because there is no reasonable possibility that the error could have contributed to the verdict. In other words, the error was harmless beyond a reasonable doubt.
The most basic justification commanding this uniform standard is that the constitution requires that every factual element of guilt be proven beyond a reasonable doubt. In re Winship, 397 US 358, 364; 90 S Ct 1068; 25 L Ed 2d 368 (1970). In sharp contrast, under my colleagues’ views, something short of proven beyond a reasonable doubt will be sufficient. This inherently translates into an appellate court *227impermissibly directing a verdict for the prosecution, violating the defendant’s right to a jury trial.
Starting with that axiom, I note that I am not the first to believe that a uniform standard must be applied, and I quote below the persuasive reasoning of two courts that have adopted such a standard, expressly rejecting the federal approach. The Pennsylvania Supreme Court stated in Commonwealth v Story, 476 Pa 391, 406-409; 383 A2d 155 (1978):
Several considerations persuade us that the “beyond a reasonable doubt” standard is the proper standard to apply in determining the harmlessness of any criminal trial error. First, this standard is commensurate with the standard of proof in criminal trials — that an accused cannot be convicted unless the trier of fact is convinced beyond a reasonable doubt that the accused is guilty as charged. In re Winship, 397 US 358; 90 S Ct 1068; 25 L Ed 2d 368 (1970). In order to maintain the integrity of this standard, appellate courts should utilize a comparable standard in determining whether an error was harmless. Professor Saltzburg has observed:
“It would make little sense to adopt the Winship standard, which is designed to prevent criminal convictions if there is even a reasonable doubt in the minds of jurors as to the guilt of the person charged, and then on appeal to emasculate that evidentiary standard when the trial court has violated evidentiary rules which might have influenced the jury by creating the requisite doubt. . . . [Cjonvictions must be reversed where the appellate court cannot arrive at a conclusion about the impact of an error on the jury verdict with the same degree of certainty demanded at trial.”
Saltzburg, The harm of harmless error, 59 Va L Rev 988, 992 (1973).12
Second, there are sound reasons for applying the same standard for determining harmless error whether the error violates state or federal law. State rules often implicate constitutional values, and the violation of a state rule may rise *228to the level of a federal constitutional violation. The protection of constitutional rights, as well as the development of a coherent doctrine of harmless error, militate in favor of the application of the same standard for constitutional and non-constitutional errors. Because it may be unclear whether a well established state rule is also constitutionally mandated, separate harmless error standards might prove to be unworkable. Moreover, a more relaxed harmless error standard for errors perceived as violations of state rules, but which might also be violations of the federal Constitution, would leave constitutional values inadequately protected.
Third, it is irrelevant whether an error is constitutional or non-constitutional in determining whether the error is prejudicial to the accused. Constitutional errors are not inherently more injurious to an accused than errors under state law. There is no reason why a state court should apply a stricter harmless error standard to federal constitutional rules than to state rules, especially since the purpose of most state rules is to assure a fair trial.
Finally, there is the danger that a lenient harmless error rule may denigrate the interests and policies which both constitutional and non-constitutional rules promote. We are convinced that the “beyond a reasonable doubt” standard reaches the proper balance of competing considerations implicated in the harmless error rule. . . . We believe that the “beyond a reasonable doubt” standard reaches the most reasonable balance between the consideration of judicial economy and the important policies which underlie constitutional and non-constitutional rules.
Having articulated the proper standard of proof in determining whether an error is harmless, we must now address the proper definition of harmlessness, for any theory of harmless error must include both the standard of the degree to which an appellate court must be convinced that an error is harmless and the definition of harmlessness. We adopt the standard that an error cannot be held harmless unless the appellate court determines that the error could not have contributed to the verdict. Whenever there is a “ ‘reasonable possibility’ ” that an error “ ‘might have contributed to the conviction,’ ” the error is not harmless.
*22912 Accord, Commonwealth v Davis, 452 Pa 171, 177; 305 A2d 715, 719 (1973) (citations omitted):
“This reasonable doubt standard reflects a fundamental belief that once . . . error has been established, it is far worse to conclude incorrectly that the error was harmless than it is to conclude incorrectly that the error was reversible.”_
[Citations omitted.]
Similarly, the Maryland Court of Appeals explained in Dorsey v State, 276 Md 638, 657-658; 350 A2d 665 (1976):
As we see it, there is no sound reason for drawing a distinction between the treatment of those errors which are of constitutional dimension and those other evidentiary, or procedural, errors which may have been committed during a trial. Although the Amendments to the United States Constitution are commonly considered a source of fair judicial procedure, other nonconstitutional evidentiary and procedural rules, signifying state policy with respect to judicial fairness, are often a defendant’s primary source of protection. An evidentiary or procedural error in a trial is bound, in some fashion, to affect the delicately balanced, decisional process. The abnegation of a particular rule upon which the defense intended to rely may often inflict more damage than initially apparent; a meritorious line of defense may be abandoned as a result; an important witness may not be called; strategies are often forsaken. The future course of the trial inevitably must be changed to accommodate the rulings made. It is the impact of the erroneous ruling upon the defendant’s trial and the effect it has upon the decisional process which is of primary concern, not whether the error is labelled as constitutional or nonconstitutional. Invariably, a number of constitutional rights, be they of federal or state origin, are inexorably intertwined with state rules of evidence and procedure. Regardless of the generic nature of the error, we believe that upon appellate review, a uniform test should be applied in all criminal cases to determine the effect the error may have had on the verdict. [Citation omitted.]
*230Later, the court noted:
We think it worthwhile to repeat here the caveat set forth in Younie v State [272 Md 233, 248; 322 A2d 211 (1974)], that the harmless error rule “has been and should be carefully circumscribed for the reasons given in People v Jablonski, 38 Mich App 33, 39; 195 NW2d 777, 780 (1972), where it is said that:
‘Continued expansion of the harmless error rule will merely encourage prosecutors to attempt to get such testimony in, since they know that, if they have a strong case, such testimony will not be considered to be reversible error, yet if they have a weak case, they will use such testimony to buttress the case to gain a conviction and then hope that the issue is not raised on appeal.’ ” [Dorsey, 276 Md 661.]
For all these reasons, I would maintain the Robinson course in Michigan of using a uniform harmless error standard for preserved errors that are subject to harmless error analysis.
n
Although Justice Boyle declines to adopt a standard, and consequently the remainder of her opinion is dicta, she does hint at her view of what that standard should and should not be.
I do not disagree with Justice Boyle that where a reviewing court affirmatively3 finds that a preserved, nontechnical error did not prejudice the defendant it should not reverse. Nor do I disagree that where a reviewing court affirmatively finds that such error did prejudice the defendant it must reverse. However, *231Justice Boyle and I disagree with respect to the cases that fall in between those two extremes.
The issue then becomes, on which side should the reviewing court land. In those cases where the reviewing court is not sure whether the error was harmful or harmless, I believe that the constitutional guarantees to the defendant leave the reviewing court no choice but to err on the side of caution and reverse. In contrast, Justice Boyle seems to suggest that the reviewing court may affirm a conviction even if it cannot affirmatively say that the error was harmless. This is not correct. The question always must be whether an error was harmless.
The reviewing court should start its analysis from the premise that a preserved error requires reversal unless it finds under the requisite level of confidence that the error was harmless. It should not start from the premise that reversal is not required unless it finds that such error was harmful. Justice Boyle’s statements suggest that, when in doubt, the reviewing court should side with the prosecutor. This is a misreading of the harmless error standard advocated by Justice Traynor to which Justice Boyle alludes.4
I believe that a test for harmful error violates the constitutional guarantees to the defendant of having guilt proven beyond a reasonable doubt and of having a fair trial. Accordingly, I would reject any such approach.
*232m
Turning now to the instant case, I agree with the separate opinion that under the Robinson harmless beyond a reasonable doubt standard, the error here was not harmless. Accordingly, I would reverse the conviction. However, even under a “lesser” standard, I believe that the error was not harmless.
The separate opinion states that if “the whole record” consisted of the testimony of the victim Jose Cantu, the defense witness Crystal Blair, and the improper rebuttal testimony of Jennifer Brecht, it would hold that the error required reversal. Post at 264. However, by adding in the hearsay statement of the victim’s sister Elva Lulgjurflj, the separate opinion finds that the error becomes harmless. I cannot agree with this latter conclusion.
Cantu testified that he had been staying temporarily with Lulgjurflj. Defendant Mateo was also living there, as well as Lulgjurflj’s and Mateo’s three children. At around 12:15 A.M. on the night of the assault, Cantu brought home a friend, Darrell Wilder, to play cards. Cantu alleged that after playing cards for up to three hours, he heard Mateo arrive home. Cantu then asked his friend to leave.
Cantu testified that (while he was brushing his teeth in a bathroom) Mateo instigated an exchange with him about when he would be moving out. Mateo then retrieved a small silver gun from a nearby bedroom and threatened Cantu by placing the gun next to his chin. Cantu then slapped the gun out of Mateo’s hand, and it landed under the sink or on the floor. (Police officers recovered the gun from the sink.) Cantu testified that he pushed Mateo out of the bathroom and closed the door. Mateo quickly returned *233with a steak knife in each hand and proceeded to knock the door down. Cantu sustained several cuts in the ensuing attack until he was able to escape out the front door of the house and run down the street to a gas station.
Mateo’s defense was alibi, presenting the testimony of Blair who stated that Mateo was with her at the time of the assault.
My colleagues properly find sufficient evidence that Cantu was assaulted by someone. However, I cannot agree with the separate opinion that the straw that tips the scales in favor of the prosecution can be the hearsay statement of Lulgjurflj. Even if a reviewing court can properly consider unobjected-to hearsay statements — and I emphasize that this issue of first impression was neither briefed nor argued by the parties in this case — a hearsay statement that fails to fit within an exclusion or exception to the hearsay rule remains hearsay. It is inherently unreliable evidence.5 *234Moreover, on the facts in this case, the statement is especially unreliable.
Foremost, we should note that the officers did not attempt to retrieve fingerprints from the gun. They recovered it from the sink, although Cantu testified that it fell under the sink or on the floor. More importantly, the officers did not conduct a thorough search of the scene and never recovered any bloody knives- — -and, in fact, probably never entered the kitchen, where the last part of the assault allegedly occurred, even though it was only about ten feet from the bathroom. In sum, there was no physical evidence to identify Mateo as the assailant.
Additionally, the officers did not write down Lulgjurflj’s statement, even though it was customary to do so. Their only notes with respect to her statement were that she confirmed Cantu’s story. Their express testimony is informative.
Officer Wendy Keelty testified:
Q. And who did [Cantu] say did this to him?
A. He said- — he stated his sister’s boyfriend, Raul.
Q. And did the sister confirm that also?
A. Yes, she did.
Officer Eulalio Reyes testified:
Q. And did you also see his sister, by the name of Alva [sic], at the mother’s home?
A. Yes, I did.
* * *
Q. And did she tell you who did this?
A. Yes, she did.
Q. And who did she say?
A. Raul Mateo.
*235On redirect Officer Reyes further testified:
Q. And what did you put in your police report in regards to his sister?
A. That she stated the same as the victim and that she was the girlfriend of the responsible.
Q. So you put two things, that she stated the same thing as the victim did about the incident?
A. Correct.
Q. And that she was the girlfriend of Raul Mateo?
A. Correct.
Q. And those are the only two things you put in the report, is that correct?
A. Yes, it is.
Further, the detective in charge of the case, Santiago Serna, testified that Lulgjurflj refused to cooperate in the investigation.
Q. Did you also attempt to contact his sister, Alva?
A. Yes.
Q. And did you want her to make a statement, a written statement?
A. I wanted to meet with her. And the only time I met with her is when I served the subpoena. She did not want to cooperate with me at all.
I could not get her to talk to me about anything.
Q. Now at that time were you aware that she had children by the defendant?
A. Yes, I was.
Q. Now how many times did you serve a subpoena on Alva?
A. Yesterday . . . was the fourth time that I served the subpoena. . . .
Q. Did she ever appear in court pursuant to any of the subpoenas that you’ve served upon her?
A. No, she has not.
*236Moreover, the separate opinion fails to note that Cantu himself testified that Lulgjurflj could not have seen the stabbing during the assault because she was shut in another room. Most importantly, the trial judge, who was listening firsthand to the parties involved, did not believe that Lulgjurflj’s testimony would assist the prosecutor:
The Court: Certainly. And you think she would testify for the People . . .
[The Prosecutor]: If she was here . . .
The Court: It doesn’t sound like it from what the detective said. And [after] all of your investigation, she refuses to come in.
Lulgjurflj was the mother of three of Mateo’s children. The trial testimony revealed that he had other girlfriends during their time together. Lulgjurflj’s mother’s testimony clearly indicated that this was no secret. If Cantu had a motive to lie, as implicitly recognized by the separate opinion, I believe that Lulgjurflj may have had an even stronger motive to lie. Perhaps Lulgjurflj was angry with Mateo on the night in question and had a retaliatory motive in identifying him as the assailant. Or perhaps, Lulgjurflj did not see the assault at all and out of loyalty to her brother merely repeated what her brother had related to her.
Nevertheless, the reliability of Lulgjurflj’s statement was not tested through cross-examination. Further, in light of the fact that it was not reduced to written form, there is a likely possibility that at the time of the trial the officers did not completely remember her statement, which had been spoken to them six months earlier.
*237The bottom line is that this particularly unreliable hearsay statement cannot and should not be the outcome-determinative fact in this case. I believe that the separate opinion is focusing on an incorrect premise. There was not overwhelming evidence of the defendant’s guilt. The only evidence identifying the defendant as the assailant was Cantu’s testimony and the police officers’ testimonies restating Cantu’s and his sister’s statements. In contrast, the defendant presented his alibi witness. If the sister’s statement was cumulative or merely derivative of Cantu’s statement, a jury confronted with two opposing versions of the events at issue would obviously be swayed by improper and highly prejudicial rebuttal testimony that both painted the alibi witness as a liar and painted Mateo’s character as one prone to assaultive violence.6 Absent eyewitness testimony from a third party or corroborating physical evidence of identity, this Court cannot, under any standard, say that the erroneously admitted testimony did not affect the jury’s decision and influence the result.
In sum, I would reverse.
Levin, J. (separate opinion). I am in substantial agreement with the majority concerning the harmless error standard for preserved nonconstitutional error.
I would, however, decide the question of the level of assurance that an appellate court must have that *238the error was not prejudicial and thus harmless, a question that the majority puts off until another day.
Further, I would not reformulate “miscarriage of justice” to preserve some legitimacy for the statutory standard, but, rather, would declare a harmless error standard for preserved error applicable on direct review of judgments in criminal and civil cases alike. I would do so in the exercise of the Court’s authority under the constitution, which confers the “judicial power of the state” on “one court of justice” headed by this Court,1 and which confers on this Court “appellate jurisdiction as provided by rules of the supreme court,”2 and which provides that “[t]he supreme court shall by general rules establish, modify, amend and simplify the practice and procedure in all courts of this state.”3
i
I agree with the majority that the standard on review for nonconstitutional error is not whether the error is harmless “beyond a reasonable doubt.” I further agree that the harmlessness of preserved error should be “reviewed in terms of its effect on the factfinder,”4 and that a reviewing court should engage “in a comparative analysis of the likely effect of the error in light of the other evidence,”5 and that the judgment entered by the trial court should be *239affirmed if the “error had only slight or negligible influence on the verdict . . . .”6
Although the majority speaks of error that is harmful7 as well as error that is harmless, it is clear, I think, that the majority and I are in agreement that if the error affects a substantial right of a defendant in a criminal case, and was prejudicial to the defendant and affected the verdict, that the error was harmful and hence not harmless.
As I read the majority opinion,
• “affirmatively appears” means what the reviewing court has found from an examination of the whole record;
• there has been “miscarriage of justice” if the error affects a substantial right of the defendant, and was prejudicial or harmful to the defendant;
• The reviewing court should find that the error is prejudicial or harmful to the defendant unless it can state that it affirmatively appears from an examination of the whole record that the error did not affect the factfinder’s decision.8
*240I would, reaching the level of confidence or assurance question, hold that
• While the harmless error inquiry for preserved constitutional error is whether the error is harmless beyond a reasonable doubt, that is not the inquiry when the issue is whether preserved nonconstitutional error is harmless.
• The core concept is whether the defendant was prejudiced. If he was prejudiced, the error was harmful to him — and thus not harmless — and is error requiring reversal.
• The inquiry is whether the error was harmless, not whether it was harmful.
• Whether the accused was prejudiced is to be determined on the basis of the effect or influence of the error on the trier of fact, rather than on whether it appears that an actually innocent defendant has been convicted, or whether, on a retrial, following reversal of the defendant’s conviction because the error was not harmless, another jury would assuredly convict him.9
*241• Whether error had an effect or influence on the jury is to be determined upon examination of the totality of the record.10
• The inquiry on appellate review should not focus alone on the sufficiency of the evidence that supports the verdict finding the defendant guilty, but rather on all the evidence in the whole record, both the untainted evidence tending to show guilt and the evidence tending to create a doubt concerning guilt.
• On direct review, a preserved error that affects substantial rights is prejudicial to the defendant, and hence harmful to the defendant — and thus not harmless — unless the reviewing court finds that it affirmatively appears from a review of the whole record *242that it is highly probable that the error did not affect the factfinder’s decision.
n
We granted leave to appeal to consider and decide what is the appropriate standard for determining when preserved nonconstitutional error in admitting evidence requires reversal.11 We should adopt wholeheartedly the approach of the United States Supreme Court in Kotteakos v United States, 328 US 750, 759; 66 S Ct 1239; 90 L Ed 1557 (1946), applicable in determining whether nonconstitutional error in a trial in a federal court is harmless.
We should conclude, in agreement with the prosecutor and amicus curiae Prosecuting Attorneys Association of Michigan, that the appropriate standard is not the Chapman12 standard, whether the error was harmless “beyond a reasonable doubt,” but, rather, the standard set forth in Kotteakos.
As stated in Kotteakos, “a conviction cannot stand” unless the reviewing court can say
• “with fair assurance, after pondering all that happened without stripping the erroneous action from the whole, that the [jurors’] judg*243ment was not substantially swayed by the error,” or
• “the conviction is sure that the error did not influence the jury, or had but very slight effect . . .
If the reviewing court is “left in grave doubt” whether the jurors’ judgment was “substantially swayed by the error” or whether the error did substantially “influence the jury or had but very slight effect” the conviction cannot stand.13 (Emphasis added.)
Most recently, in O’Neal v McAninch, 513 US 432, 436; 115 S Ct 992; 130 L Ed 2d 947 (1995), the United States Supreme Court held, in explication of the “grave doubt” component of the Kotteakos standard, that a defendant seeking relief from conviction in a criminal case does not have the burden of establishing whether error was prejudicial. Declining to phrase the issue in terms of burden of proof, the Court declared that if the reviewing court “is in grave doubt” about whether the error had “substantial and injurious effect or influence in determining the jury’s verdict,’’ the error is not harmless.
The three dissenting justices in O’Neal had no “quarrel with the majority’s conclusion that once an error has been shown on direct appeal, the government must demonstrate that it was harmless if the conviction is to stand.”14 (Emphasis added.)
*244in
Mateo was convicted of assault with intent to murder,15 felonious assault,16 and of being an habitual offender, second offense.17
We agree with the Court of Appeals18 that the trial court erred in permitting Jennifer Brecht to testify over objection as a rebuttal witness to the testimony of Crystal Blair, a witness for Mateo. The Court of Appeals found that the error was “harmless beyond a reasonable doubt.”
A
Mateo was living with Elva Lulgjurflj and their three children. Lulgjurflj’s brother, Jose Cantu, was living with them temporarily. Cantu testified that he and Mateo became involved in an argument early in the morning concerning whether Cantu would be moving out. Cantu said that Mateo threatened to blow Cantu’s brains out with a pistol and slashed him repeatedly with a knife. Hospital records verified the slashes.
Cantu’s testimony identifying Mateo as his assailant was supported by hearsay testimony, for which no *245objection was raised, by two police officers who indicated that Lulgjurflj told them shortly after the assault that Mateo was the person who assaulted Cantu.
B
Mateo did not testify. He presented an alibi through Crystal Blair, who said that Mateo was with her when the assault was alleged to have taken place.
The prosecutor cross-examined Blair concerning conversations with Jennifer Brecht, who also had been a girlfriend of Mateo. Blair denied having conversations with Brecht about Mateo. Blair said she only spoke with Brecht when she telephoned to talk with Mateo when Mateo was living with Brecht. The prosecutor then called Brecht who, over objection, testified that Blair had telephoned Brecht and warned that Mateo was “very violent.”
c
The Court of Appeals held that Brecht’s testimony was improper because it impeached Blair on a collateral matter rather than a material issue.19 The Court added that the prosecutor may not elicit a denial from a witness on cross-examination “only to inject a new issue on rebuttal.”20 The Court continued that the
*246prosecutor apparently asked Blair about the assault during cross-examination to interject evidence of prior assaultive acts by Mateo and to “set-up” Blair for the purpose of impeaching her credibility with prior inconsistent statements.
I agree with the Court of Appeals that it was error to admit the rebuttal testimony of Brecht, both because it involved impeaching a witness on a collateral matter, and also because the inconsistency was created by eliciting a denial on cross-examination.21
D
The Court of Appeals found that the error was “harmless.”
The victim testified that defendant assaulted him with a knife and a gun. The victim’s sister who witnessed the assaults confirmed his version of the events to the investigating officers. The victim was treated for multiple lacerations and a handgun was recovered from the bathroom where the assault took place. Further, the officers noted that the bathroom door had been broken.[22] In light of the *247overwhelming evidence against defendant, the improper rebuttal testimony was harmless. [Unpublished opinion per curiam, issued February 17, 1993 (Docket No. 134528), slip op, p 1.]
IV
Mateo contends that a reviewing court should assess the harmlessness of the error under the harmless-beyond-a-reasonable-doubt standard announced by the United States Supreme Court in Chapman v California, 386 US 18, 24; 87 S Ct 824; 17 L Ed 2d 705 (1967). The Court there held that “before a federal constitutional error can be held harmless, the court must be able to declare a belief that it was harmless beyond a reasonable doubt” (emphasis added), and that the repeated references to the defendants’ failure to testify did not constitute harmless error.23
*248In the instant case, the error in admitting Brecht’s statements was not federal constitutional error.24 While, as Mateo contends, this Court has employed the Chapman harmless-beyond-a-reasonable-doubt formulation in cases in which the error did not constitute federal constitutional error,25 in none of those cases did this Court consider that the Chapman standard enunciated by the United States Supreme Court is not applied by that Court when the error is not of constitutional dimension.
A
Over twenty years before Chapman was decided, the United States Supreme Court, in Kotteakos, construed the federal harmless-error statute26 applicable in reviewing error occurring during the course of a federal civil or criminal trial.
*249The Michigan harmless-error statute,27 first enacted in 1915, and the federal harmless-error statute, first enacted in 1919, have a common history.28 The United States Supreme Court, in Kotteakos, said that it was “not necessary to review in detail the history of the abuses which led to the agitation or of the progress of the legislation” resulting in the enactment of the federal harmless-error statute, and that “anyone familiar with it knows that” that legislation “and similar *250state legislation grew out of widespread and deep conviction over the general course of appellate review in American criminal causes.”29 (Emphasis added.)
B
The Michigan harmless-error statute30 and court rule31 express, as stated in People v Robinson, 386 Mich 551, 562; 194 NW2d 709 (1972), essentially the *251same concept. This Court observed in Robinson that the “strictures” of the statute and the court rale are but “different articulations of the same idea,” and stated that “same idea” as follows:
[A]ppellate courts should not reverse a conviction unless the error was prejudicial. As stated in [People v] Nichols [341 Mich 311; 67 NW2d 230 (1954)] “. . . the rule always in effect in Michigan, . . . has been and is that the question of reversal is controlled by determination of whether the error was prejudicial.” [Id., p 562 (emphasis added).]
In Nichols, the case cited in Robinson, the prosecutor had urged that “the entire record is persuasive of defendant’s guilt.” The Court responded, citing a civil case, Soltar v Anderson, 340 Mich 242; 65 NW2d 777 (1954), where, as stated in Nichols, this Court had held that “the rale always in effect in Michigan, both before and after the enactment of the mentioned statutes and unaffected thereby, has been and is that the question of reversal is controlled by determination of whether the error was prejudicial . ...” Id., p 332 (emphasis added).32
*252In People v Bigge, 288 Mich 417, 421; 285 NW 5 (1939), this Court said respecting the provision of the 1927 statute concerning harmless error in a criminal case (which has not been changed by subsequent legislation, see n 27):
The responsibility of maintaining the right of fair trial and due process of law is placed with the judicial branch and cannot be otherwise by legislative permission. We are not concerned with the guilt or innocence of the accused, for we are not triers of the facts and must apply the law to the case as tried.
The foregoing statements of this Court over forty years ago in Nichols, Soltar, and Bigge find support in the following statement by the United States Supreme Court in Kotteakos-.
[I\t is not the appellate court’s function to determine guilt or innocence. Nor is it to speculate upon probable reconviction . . . . [Id., p 763 (citations omitted; emphasis added).]
V
Having in mind the Michigan pre-Chapman harmless-error standard — whether the error was prejudicial — and the common history of enactment of the federal and Michigan harmless-error statutes, I am persuaded that the harmless-error standard announced in Kotteakos by the United States Supreme Court in the construction of the federal harmless-error statute should be adopted by this Court as the harmless-error standard to be applied when appellate courts of this state review preserved nonconstitutional error.
*253As stated at the outset of this opinion, I would adopt the harmless-error standard stated in Kotteakos, that a finding or verdict of guilty is required to be set aside and a new trial granted unless the reviewing court can say
• “with fair assurance, after pondering all that happened without stripping the erroneous action from the whole, that the [jurors’] judgment was not substantially swayed by the error,” or
• “the conviction is sure that the error did not influence the jury, or had but very slight effect.”
If the reviewing court is “left in grave doubt” whether the jurors’ judgment was “substantially swayed by the error” or whether the error did substantially “influence the jury or had but very slight effect” the conviction cannot stand.33 (Emphasis added.)
*254Recently, in Brecht v Abrahamson, 507 US 619; 113 S Ct 1710; 123 L Ed 2d 353 (1993), the United States Supreme Court held that the use of the accused’s post-Miranda warning silence for impeachment purposes, while violative of the Due Process Clause, would, because the question arose in a habeas coipus proceeding and not on direct appeal, be reviewed under Kotteakos formulation to determine whether the error “had substantial and injurious effect or influence in determining the jury’s verdict.”34
Justice Stevens, concurring in Brecht v Abraham-son, said that Kotteakos requires de novo review “in the context of the entire trial record,”35 and “requires a reviewing court to decide that ‘the error did not influence the jury,’ ” and that “the judgment was not substantially swayed by the error . . . ,”36
*255As previously noted, last year, in O’Neal v McAninch, supra, the United States Supreme Court elaborated on the Kotteakos formulation,37 and addressed the “special circumstance in which record review leaves the conscientious judge in grave doubt about the likely effect of an error on the jury’s verdict.” The Court elaborated on what constitutes “grave doubt,” stating:
(By “grave doubt” we mean that, in the judge’s mind, the matter is so evenly balanced that he feels himself in virtual equipoise as to the harmlessness of the error.) We conclude that the uncertain judge should treat the error, not as if it were harmless, but as if it affected the verdict (i.e., as if it had a “substantial and injurious effect or influence in determining the jury’s verdict”).[38]
The Court expressly declined to phrase the issue in terms of burden of proof.
*256[W]e deliberately phrase the issue in this case in terms of a judge’s grave doubt, instead of in terms of “burden of proof.”[39]
The dissenting justices in O’Neal had no “quarrel with the majority’s conclusion that once an error has been shown on direct appeal, the government must demonstrate that it was harmless if the conviction is to stand.”40 (Emphasis added.) The dissenters would have required a federal habeas corpus petitioner to bear the burden of persuasion that the challenged actions caused harm — that the “error was harmful”— and that the error was prejudicial.
Mateo is seeking relief on direct appeal, and the question is whether the error was harmless, not whether it was harmful. The reviewing court should *257ask whether the error is harmless, not whether it is harmful.
VI
The majority puts off to another day deciding the level of assurance or confidence that a reviewing court must have that a preserved, nonconstitutional error was not prejudicial and, thus, harmless.41 It states that “[t]he Kotteakos test, as articulated by the prosecutor, does not definitively settle on this question,”. and that “federal courts have variously adopted tests described as a highly probable, more probable than not, and a reasonable likelihood that the error affected substantial rights.”42 The majority continues that Chief Justice Traynor, in his book on harmless error, discussed the “highly probable” and “more probable than not” tests, and “suggests that the highly probable test strikes the appropriate balance between protecting both the public and defendants’ interests in fair trials.”
A
Most circuits that have considered the level of confidence or assurance issue employ the “highly probable” test:
United States v Cudlitz, 72 F3d 992, 999 (CA 1, 1996) (“we are instructed to ask whether it is ‘highly probable’ that the error did not ‘contribute to the verdict’ ”).
United States v Tussa, 816 F2d 58, 67 (CA 2, 1987) (“in order to find harmless error, we must find it *258'“highly probable” that the error did not contribute to the verdict’ ”).
United States v Quintero, 38 F3d 1317, 1331 (CA 3, 1994) (“we must evaluate whether it is ‘highly probable that the evidence did not contribute to the jury’s judgment of conviction.’ ” Government of Virgin Islands v Toto, 529 F2d 278, 284 [CA 3, 1976]).
United States v Ince, 21 F3d 576, 583 (CA 4, 1994) (“whether we believe it ‘highly probable that the error did not affect the judgment’ ”).
United States v Mackin, 163 US App DC 427, 435; 502 F2d 429, 437 (1974) (“In reviewing the proceedings this court must decide whether it was ‘highly probable that the error had substantial and injurious effect or influence in determining the jury’s verdict’ ”).
The United States Court of Appeals for the Tenth Circuit in United States v Wacker, 72 F3d 1453, 1473 (CA 10, 1995), declared that error is “harmless ‘unless it has a “substantial influence” on the outcome or leaves one in “grave doubt” as to whether it had such effect.’ ”
The United States Courts of Appeals for the Fifth Circuit and Eleventh Circuit (a split-off of the Fifth) appear to reiterate the Kotteakos formulation and then announce a decision.
In United States v Neuroth, 809 F2d 339, 342 (CA 6, 1987), the United States Court of Appeals for the Sixth Circuit declared: “An error, not of constitutional dimension, is harmless unless it is more probable than not that the error materially affected the verdict.” Neuroth, did not refer to Kotteakos. Subsequent cases do, and do not include the “more probable than not” formulation. See, e.g., United States v Dean, 969 F2d 187, 197 (CA 6, 1992), and United States v *259Wiedyk, 71 F3d 602, 607 (CA 6, 1995) (using “fair assurance” language from Kotteakos).
The United States Court of Appeals for the Seventh Circuit does not appear to have moved beyond the language of Kotteakos, often relying on the “substantial influence” language. United States v Grier, 866 F2d 908, 920 (CA 7, 1989). There was a time, however, when the circuit employed a “significant possibility” standard, but that seems to have been abandoned in favor of reliance on Kotteakos itself. United States v Marrero, 486 F2d 622, 627 (CA 7, 1973).
The United States Court of Appeals for the Eighth Circuit in United States v Nevils, 897 F2d 300, 307 (CA 8, 1990) said: “We must be able to say with ‘fair assurance’ that the error did not sway the jury’s verdict.”
The United States Court of Appeals for the Ninth Circuit, in United States v Hitt, 981 F2d 422, 425, n 2 (CA 9, 1992), observed that at times the circuit had applied a “fair assurance” test and at others a “more probable than not” test, and said:
A 55% likelihood that the error was harmless qualifies as “more probable than not,” but it’s hardly a “fair assurance” of harmlessness.
Kotteakos contains its own level of confidence and assurance tests, namely, the question is whether the error had “substantial influence” on the trier of fact, and if the reviewing court “is left in grave doubt, the conviction cannot stand”:
If, when all is said and done, the conviction is sure that the error did not influence the jury, or had but very slight effect, the verdict and the judgment should stand, except perhaps where the departure is from a constitutional norm *260or a specific command of Congress. But if one cannot say, with fair assurance, after pondering all that happened without stripping the erroneous action from the whole, that the judgment was not substantially swayed by the error, it is impossible to conclude that substantial rights were not affected. The inquiry cannot be merely whether there was enough to support the result, apart from the phase affected by the error. It is rather, even so, whether the error itself had substantial influence. If so, or if one is left in grave doubt, the conviction cannot stand. [Id., pp 764-765 (citation omitted).]
It appears that the alternative expressions of the level of confidence or assurance an appellate court on direct review must have before declaring that a preserved nonconstitutional error was not prejudicial and, thus, harmless, are the “substantial influenceTgrave doubt” standard set forth in Kotteakos itself, and the alternatives suggested by Justice Traynor of “highly probable,” adopted in most United States Courts of Appeals, or “more probable than not,” not regularly employed in any circuit, the vague “fair assurance” test and the “reasonable likelihood” test, adverted to in footnote 19 of the majority opinion, set forth in United States v Hawkins, 905 F2d 1489, 1493 (CA 11, 1990).
This Court is adequately informed concerning the alternatives, and the briefing in this case is more than adequate.
B
Further, the bench and bar have been awaiting a definitive statement from this Court since People v Anderson (After Remand), 446 Mich 392, 407, n 39; 521 NW2d 538 (1994), was decided with the following observation:
*261Accordingly, we save for another day, after full briefing and argument, the distinction between, and the enunciation of, the constitutional harmless error test and Michigan’s nonconstitutional harmless error test.
On the same day that Anderson was decided, this Court decided People v Dunn, 446 Mich 409; 521 NW2d 255 (1994), where this Court concluded that the admission of tainted evidence was not harmless, but did so without reference to this Court’s harmless error jurisprudence in deference to the decision reflected in Anderson to save until another day, after full briefing and argument, the elucidation of the standard for reviewing preserved nonconstitutional error. Also on the same day that Anderson and Dunn were decided, this Court decided People v Stanaway, 446 Mich 643; 521 NW2d 557 (1994), where, in a separate opinion, a colleague observed:
Despite guidance from both our court rules and statute, this Court has yet to fully examine the relevant considerations for nonconstitutional harmless error and certainly has failed to set forth a clear and concise nonconstitutional harmless-error test. Instead, this Court has generally chosen to rely on the harmless-error statute and court rules for the limited guidance provided therein in making this determination. People v Travis, 443 Mich 668, 686; 505 NW2d 563 (1993). From a plain reading of these rules, however, the vague concept of injustice certainly does not provide any meaningful help to appellate courts in reviewing nonconstitutional error. [Id., pp 697-698 (Riley, J., concurring).]
Lastly, Mateo’s application for leave to appeal was held in abeyance pending this Court’s decision in People v Dunn, which, for reasons already stated, did not decide the appropriate standard for determining when preserved nonconstitutional error requires reversal. In *262granting leave to appeal in this case, we asked the parties to brief “what is the appropriate standard for determining when nonconstitutional error in admitting evidence is reversible.” See n 11.
c
I would, therefore, complete the task, and decide the level of assurance or confidence question under the circumstances that the alternative possible tests have been considered and discussed in countless cases, the Court is fully informed, both from our own experiences in reviewing literally hundreds, if not thousands, of cases where the harmless error issue permeates the disposition at the Court of Appeals level and at this level, and the more than adequate briefing and independent research as reflected in the majority discussion of Justice Traynor’s work on harmless error.
vn
As stated in the concurring opinion in Stanaway, “the vague concept of injustice certainly does not provide any meaningful help to appellate courts in reviewing nonconstitutional error.”43 Nothing is gained, and much lost, by preserving the reformulated “miscarriage of justice” formula.
There can be little doubt that it is ultimately this Court’s responsibility, and not that of the Legislature, to determine whether it is consistent with the concept of appellate judicial review, and the proper exercise of the judicial power, and the concept of a fair trial embodied in the Due Process Clause, to *263ignore events at a trial that taint a conviction or encourage practices that debilitate or undermine the high standards one expects to be observed and enforced in a court of justice.44
Continued verbalization of miscarriage of justice as a test is inconsistent with this Court’s reaffirmation today that a reviewing court may not affirm a conviction “simply because it concludes the jury reached the right result.”45
The impropriety of continuing the miscarriage of justice formula then becomes clearer upon consideration of this Court’s recent decision in People v Grant, 445 Mich 535; 520 NW2d 123 (1994), where, following the lead of the United States Supreme Court in United States v Olano, 507 US 725; 113 S Ct 1770; 123 L Ed 2d 508 (1993), this Court recognized that even where the defendant fails to preserve nonconstitutional error, an appellate court may reverse for “plain forfeited error” even though, as stated in Olano, there has not been a miscarriage of justice in the sense that the defendant is actually guilty, if the error “seriously affects” the fairness, integrity or public reputation of judicial proceedings.
vm
Applying46 the Kotteakos standard, applicable to nonconstitutional error, as elaborated in Brecht and *264O’Neal, I turn to a consideration, on de novo review of the whole record, whether the error in admitting Brecht’s testimony was harmless.
I agree with Mateo, contrary to a statement of the Court of Appeals, that the evidence against him was not “overwhelming.” While there was possibly overwhelming evidence that Cantu had been assaulted by someone,47 no one other than Cantu testified that Mateo was the assailant. Cantu clearly was an interested witness, and therefore his credibility could be questioned. Mateo did not testify, but, as a matter of law, that cannot be weighed in evaluating the harmlessness of the error in admitting Jennifer Brecht’s testimony that Blair had warned Brecht that Mateo was “very violent,” thereby impeaching Blair — who had testified that Mateo was with her when the assault was claimed to have taken place — and who had denied having conversations with Brecht about Mateo.
If that were the whole record, I would be at least in grave doubt whether the error had substantial and *265injurious effect or influence in determining the jury’s verdict. There would only be Cantu’s testimony identifying Mateo as the assailant, and Blair’s alibi testimony that Mateo was with her at the time of the assault. Brecht’s testimony impeaching Blair’s testimony and indicating that Mateo was at times a violent womanizer, might very well then have swayed the jury.
Additionally, however, there was the testimony of two police officers that Cantu’s sister, Elva Lulgjurflj, had confirmed, at the scene of, and shortly after, the alleged assault, that Mateo was the assailant.
The officers’ testimony was hearsay, but Mateo’s lawyer failed to object. Cantu’s mother’s testimony that he identified Mateo as his assailant shortly after the incident, might have been admissible, although hearsay, as an excited utterance or some other basis. But, Mateo’s lawyer’s failure to object leaves the record silent in that regard. Courts disagree concerning the weight to accord hearsay admitted without objection. The majority view is that hearsay admitted without objection has probative value.48
On the same premise that the harmless-error analysis for preserved error differs from that where the error has not been preserved,49 an appellate court undertaking review for harmlessness should consider, as part of the whole record, unobjected to hearsay— *266in this case, the hearsay statements identifying Mateo as the assailant.
To be sure, the inquiry in assessing the harmlessness of error is whether the error may have swayed the jury, not the quantity or quality of the other evidence. Nevertheless, in deciding whether the error may have swayed the jury, the reviewing court must consider the whole record, not for the purpose of itself making a judgment concerning the defendant’s guilt or innocence, but rather for the purpose of determining whether the jury might have been swayed by the erroneously admitted testimony.50
Cantu’s testimony identifying Mateo as the assailant, together with the hearsay corroboration, was opposed by Blair’s testimony that Mateo was with her at the time of the assault. It is questionable whether any juror’s assessment of the credibility of the opposing testimony was swayed by Brecht’s assertion that Blair had, contrary to her denial, conversed with Brecht concerning Mateo disparagingly.51
There is a further consideration, however. The concept of harmless error has become part of the equation for the purpose of determining whether there is error at all, as in the assessment of whether a defendant was deprived of the effective assistance of counsel. In People v Pickens, 446 Mich 298; 521 NW2d 797 (1994), this Court, again following the lead of the United States Supreme Court, adopted the approach set forth in Strickland v Washington, 466 US 668; 104 S Ct 2052; 80 L Ed 2d 674 (1984), which imposes on *267the defendant the burden of showing that, absent counsel’s errors, the factfinder would have had a reasonable doubt respecting guilt. This Court said that “under Michigan law, counsel’s ineffective assistance must be found to have been prejudicial in order to reverse an otherwise valid conviction.”52
The Court’s analysis in Grant and Pickens considered whether the unpreserved error in Grant and the error of counsel in Pickens, might have been outcome-determinative.53
Clearly, the error would be outcome-determinative if the reviewing court would be in “grave doubt” or be unable to declare that it is “highly probable” that the error did not affect the verdict, but for error of counsel that permitted inadmissible evidence to become part of the record that may have influenced or swayed the jury.
No possible trial strategy could have justified Mateo’s lawyer’s failure to have objected to the inadmissible hearsay. His failure to do so fell below an objective standard of reasonableness in the representation of persons accused of crime.
For the foregoing reasons, I conclude that the error was not harmless, and would remand for a new trial.
The majority asserts that the evidence of Mateo’s guilt is “overwhelming.” The mantra “overwhelming” ignores the imperative of considering not only evidence tending to establish guilt, but also the evidence tending to refute the evidence of guilt. The Court of Appeals and this Court, on review of the whole record, were and are required to assess not only *268Cantu’s testimony, but also Crystal Blair’s alibi testimony tending to show that Mateo could not have been the assailant.
I would reverse and remand for a new trial.
My colleagues imply that the Robinson Court did not understand the significance of citing Chapman v California, 386 US 18; 87 S Ct 824; 17 L Ed 2d 705 (1967), when it defined its harmless error standard. I disagree. In Robinson, there were two primary categories of alleged errors: (1) the admission of an involuntary confession, and (2) evidentiary errors, which were the equivalent of improper bad-acts evidence. Id. at 556. The Court reversed the defendant’s conviction on the basis of the admission of the involuntary confession, without engaging in any harmless error analysis. Id. at 559. It was within the discussion of the alleged evidentiary errors, or what would now be termed nonconstitutional errors, that Robinson cited Chapman for support of the harmless error test.
Given that Robinson was decided in 1972, five years after Chapman was decided in 1967, and twenty-six years after Kotteakos v United States, 328 US 750, 764; 66 S Ct 1239; 90 L Ed 1557 (1946), was decided, we must presume that this Court understood the significance of citing Chapman rather than Kotteakos. It knew then that the federal harmless error standards were different for constitutional and nonconstitutional error. If this Court had wanted to adopt such a dichotomy, it would have done so. It strains belief that any of my colleagues would suggest that this Court did not recognize what it was doing before now. In reality, my colleagues simply disagree with Robinson, which they should openly acknowledge, rather than undermining the work of our predecessor.
Some errors have been held to be “so offensive to the maintenance of a sound judicial process that [they] cannot be regarded as harmless error.” People v Bentley, 402 Mich 121, 124; 261 NW2d 716 (1978) (unconstitutional denial of right to counsel).
For the sake of argument, I am assuming that “affirmatively finds,” which I interpret as a confident and positive assertion, equates with “finding beyond a reasonable doubt.”
Justice Boyle quotes from Justice Traynor:
“Unless the appellate court believes it highly probable that the error did not affect the judgment, it should reverse." [Ante at 219 (citation omitted and emphasis added).]
As this Court has held:
A witness’ perception of persons and events, the clarity and accuracy of the witness’ memory, and the lucidity of the witness’ description of persons and events are critical in evaluating the credibility of testimony. The law requires that witnesses be present at trial, take an oath of truthfulness and be subject to cross-examination so that their credibility may be properly evaluated. The admission of hearsay evidence is disfavored because it is difficult, if not impossible, for the trier of fact to assess the reliability of hearsay statements or of the hearsay declarant. The trier of fact is unable to view and evaluate the demeanor and manner of the declarant while making the hearsay statement. The hearsay statement has not been vouched and thus the declarant may not have felt the “special obligation to tell the truth” that results from the taking of an oath. Most importantly, there is no opportunity for cross-examination of the declarant regarding the content of the hearsay statement. [Poole, 444 Mich 160.]
Justice Boyle refers to Cantu’s statement to his mother. Ante at 208. However, this statement would still be nothing more than a cumulative restatement of Cantu’s version of the events. It was not based on any independent knowledge on his mother’s part.
Const 1963, art 6, § 1.
Const 1963, art 6, § 4.
Const 1963, art 6, § 5.
Ante, p 221.
Id., p 206.
Id., p 221.
[R]eversal is not required unless an error is harmful. [Id., p 212.]
Justice Scalia, writing for a unanimous court in a constitutional error case, said that the harmless error inquiry concerns the effect of the error on the jury that was actually impaneled, not the effect on a hypothetical jury:
The most an appellate court can conclude is that a jury would surely have found petitioner guilty beyond a reasonable doubt — not that the jury’s actual finding of guilty beyond a reasonable doubt would surely not have been different absent the constitutional error. That is not enough. See Yates [v Evatt, 500 US 391, 403-404; *240111 S Ct 1884; 114 L Ed 2d 432 (1991)] (Scalia, J., concurring in part and concurring in judgment). The Sixth Amendment requires more than appellate speculation about a hypothetical jury’s action, or else directed verdicts for the State would be sustainable on appeal; it requires an actual jury finding of guilty. See Bollenbach v United States, 326 US 607, 614 [66 S Ct 402; 90 L Ed 350] (1946). [.Sullivan v Louisiana, 508 US 275, 280; 113 S Ct 2078; 124 L Ed 2d 182 (1993).]
See n 4; Kotteakos v United States, 328 US 750, 759, 763; 66 S Ct 1239; 90 L Ed 1557 (1946).
In Kotteakos, the Court said:
[T]his does not mean that the appellate court can escape altogether taking account of the outcome. To weigh the error’s effect against the entire setting of the record without relation to the verdict or judgment would be almost to work in a vacuum. ... In criminal causes that outcome is conviction. This is different, or may be, from guilt in fact. It is guilt in law, established by the judgment of laymen. And the question is, not were they right in their judgment, regardless of the error or its effect upon the verdict. It is rather what effect the error had or reasonably may be taken to have had upon the jury’s decision. The crucial thing is the impact of the thing done wrong on the minds of other men, not on one’s own, in the total setting. . . .
This must take account of what the error meant to them, not singled out and standing alone, but in relation to all else that happened. And one must judge others’ reactions not by his own, but with allowance for how others might react and not be regarded generally as acting without reason. This is the important difference, but one easy to ignore when the sense of guilt comes strongly from the record. [Id., n 9 supra, pp 764-765 (emphasis added).]
448 Mich 868 (1995). Mateo’s application for leave to appeal was held in abeyance pending this Court’s decision in People v Dunn, 446 Mich 409; 521 NW2d 255 (1994). Dunn did not decide the appropriate standard for determining when nonconstitutional error requires reversal. This Court resolved the pending abeyance in this case by entering an order granting leave to appeal, limited to the issues “(1) whether the trial court erred in permitting witness Brecht to testify as a rebuttal witness, (2) what is the appropriate standard for determining when nonconstitutional error in admitting evidence is reversible, and (3) whether any error in the admission of the testimony of witness Brecht was error requiring reversal.”
Chapman v California, 386 US 18; 87 S Ct 824; 17 L Ed 2d 705 (1967).
Kotteakos, supra, pp 764-765.
Id., 513 US 448.
Amicus curiae Prosecuting Attorneys Association of Michigan agrees, and further agrees that the power to determine the harmless error standard for preserved nonconstitutional error, except where the error alleged is a failure to observe a mandate created by statute, is “committed to the *244judicial branch of government,” and thus the literal language of § 26, ch EX of the Code of Criminal Procedure (see n 30) is not controlling.
O’Neal, in contrast with the instant case, was not a direct appeal, but rather a federal habeas corpus proceeding, requiring that the petitioner establish a violation of the federal constitution or a federal statute. The dissenters would have required a federal habeas coipus petitioner to bear the burden of persuasion that the error caused harm and that the error was prejudicial.
MCL 750.83; MSA 28.278.
MCL 750.82; MSA 28.277.
MCL 769.10; MSA 28.1082.
Unpublished opinion per curiam issued February 17, 1993 (Docket No. 134528), slip op, p 1.
The Court of Appeals cited People v Holland, 179 Mich App 184, 193; 445 NW2d 206 (1989). See n 18.
Id., slip op, p 1. The Court again cited Holland, n 19 supra, p 194. The Court in Holland said:
Rebuttal testimony is that used to contradict, repel, explain, or disprove evidence produced by the other party, tending directly to weaken or impeach the same. People v Kelly, 423 Mich 261, 281; 378 NW2d 365 (1985). There are, however, limitations. First, if the evidence should have been introduced in the case in chief, rebuttal is improper. People v Losey, 413 Mich 346, 351; 320 NW2d 49 *246(1982). Second, rebuttal evidence must be on a material matter, not a collateral issue. People v Teague, 411 Mich 562, 566; 309 NW2d 530 (1981). Third, if the evidence was not raised in the prosecutor’s case in chief, the defense must have raised a new issue. The prosecutor is not allowed to elicit a denial from the defendant on cross-examination only to iryect a new issue on rebuttal or revive the right to introduce evidence that should have been introduced in the case in chief. People v Bennett, 393 Mich 445, 449-450; 224 NW2d 840 (1975). [Id., pp 193-194.]
See People v McGillen No 1, 392 Mich 251; 220 NW2d 677 (1974); People v Bennett, n 20 supra.
The Court of Appeals added:
Intent to murder may be inferred from the facts and circumstances. Id. The victim testified that defendant held a handgun under the victim’s chin and told him that he was going to “blow his brains out.” After the victim knocked the gun out of defendant’s *247hand and locked the bathroom door, defendant smashed in the upper door panel and started swinging two steak knives at the victim. When the victim attempted to flee, defendant cut him off and continued swinging the knives, cutting the victim twice in the head and once on the chin and hand, all requiring sutures. [Id., slip op, p 2.]
The Court said that although the government had “presented a reasonably strong ‘circumstantial web of evidence’ against petitioners, 63 Cal 2d [178, 197; 404 P2d 209, 220 (1965)],” “absent the constitutionally forbidden comments, honest, fair-minded jurors might very well have brought in not-guilty verdicts.” Chapman v California, supra, pp 25-26.
Nearly twenty-five years after Chapman was decided, the United States Supreme Court, in Arizona v Fulminante, 499 US 279, 307; 111 S Ct 1246; 113 L Ed 2d 302 (1991), characterized some constitutional errors as “trial error,” subject to harmless-error analysis, and other constitutional errors as “structural defects” that “defy analysis by ‘harmless-error standards.’ ”
Two years later, in Sullivan v Louisiana, 508 US 275; 113 S Ct 2078; 124 L Ed 2d 182 (1993), the United States Supreme Court ruled, on direct appeal from the Supreme Court of Louisiana, that a constitutionally deficient criminal jury instruction concerning the definition of reasonable doubt, was structural error not amenable to harmless-error analysis.
Mateo and amicus curiae Criminal Defense Attorneys of Michigan argue that the distinction between constitutional and nonconstitutional violation is not easily drawn, and the overlap supports the adoption of a single, beyond-a-reasonable-doubt standard. We are not persuaded that because the task may at times be daunting that it would be responsible to ignore the cost to the administration of justice of so simply cutting the Gordian knot.
People v Liggett, 378 Mich 706, 717; 148 NW2d 784 (1967); People v Robinson, 386 Mich 551, 563; 194 NW2d 709 (1972); People v Gash, 419 Mich 230, 249; 351 NW2d 822 (1984).
But see the opinion in People v Lee, 434 Mich 59, 86, 123; 450 NW2d 883 (1990) (opinions of Brickley and Boyle, JJ.); People v Straight, 430 Mich 418, 427; 424 NW2d 257 (1988); People v Johnson, 427 Mich 98, 115; 398 NW2d 219 (1986) (Boyle, J.).
On the hearing of any appeal, certiorari, writ of error, or motion for a new trial, in any case, civil or criminal, the court shall give judgment after an examination of the entire record before the court, without regard to technical errors, defects, or exceptions which do not affect the substantial rights of the parties. [Section 269 of the Judicial Code, 28 USC 391, enacted in 1919 (40 Stat 1181) (emphasis added).]
Four years before the enactment of the federal harmless-error statute, the Legislature enacted the following law:
No judgment or verdict shall be set aside or reversed, or a new trial be granted by any court of this State in any case, civil or criminal, on the ground of misdirection of the jury, or the improper admission or rejection of evidence, or for error as to any matter of pleading or procedure, unless, in the opinion of the court, after an examination of the entire cause, it shall affirmatively appear that the error complained of has resulted in a miscarriage of justice. [1915 PA 89 (emphasis added).]
This provision was codified in the Code of Criminal Procedure, 1927 PA 175, ch IX, § 26, with the substitution of “in any criminal case” for “in any case, civil or criminal.” The language so enacted, which has continued to remain in force without amendment to the present, provides as follows:
No judgment or verdict shall be set aside or reversed or a new trial be granted by any court of this state in any criminal case, on the ground of misdirection of the jury, or the improper admission or rejection of evidence, or for error as to any matter of pleading or procedure, unless in the opinion of the court, after an examination of the entire cause, it shall affirmatively appear that the error complained of has resulted in a miscarriage of justice. [MCL 769.26; MSA 28.1096 (emphasis added).]
The language of the original Michigan harmless-error statute (1915 PA 89) was enacted as a section of the Judicature Act of 1915, 1915 PA 314, ch L, § 28, with the substitution of “in any civil case” for “any case, civil or criminal.” This language was carried forward through 1948 CL 650.28, but was not continued in the Revised Judicature Act of 1961. Language respecting civil harmless error was continued forward as GCR 1963, 529.1, which is set forth in n 31.
The Michigan harmless error statute, as first enacted in 1915, applied in both civil and criminal trials. See n 27 for the text of the 1915 statute.
Kotteakos, supra, p 759.
See n 27 for the text of the original Michigan harmless-error statute, enacted as 1915 PA 89, and for the text of the harmless-error statute enacted as part of the Code of Criminal Procedure, as 1927 PA 175, ch IX, § 26, which remains in force without amendment to the present as MCL 769.26; MSA 28.1096.
GCR 1963, 529.1 was modeled on FR Civ P 61, and read as follows:
No error in either the admission or the exclusion of evidence and no error or defect in any ruling or order or in anything done or omitted by the court or by any of the parties is ground for granting a new trial or for setting aside a verdict or for vacating, modifying, or otherwise disturbing a judgment or order, unless refhsal to take such action appears to the court inconsistent with substantial justice. The court at every stage of the proceeding shall construe these rules to secure the just, speedy, and inexpensive determination of every action so as to avoid the consequences of any error or defect in the proceeding which does not affect the substantial rights of the parties. [Emphasis added.]
The current civil rule reads as follows:
An error in the admission or the exclusion of evidence, an error in a ruling or order, or an error or defect in anything done or omitted by the court or by the parties is not ground for granting a new trial, for setting aside a verdict, or for vacating, modifying, or otherwise disturbing a judgment or order, unless refusal to take this action appears to the court inconsistent with substantial justice. [MCR 2.613(A) (emphasis added).]
The Michigan Rules of Evidence, modeled on the Federal Rules of Evidence, state essentially the same concept:
(a) Effect of erroneous ruling. Error may not be predicated upon a ruling which admits or excludes evidence unless a substantial right of the party is affected, and
* * *
*251(d) Plain error. Nothing in this rule precludes taking notice of plain errors affecting substantial rights although they were not brought to the attention of the court. [MRE 103 (emphasis added).]
In Soltar, supra, p 244, it was urged that instructional error in a civil case should not result in reversal because it did not “ ‘affirmatively appear that the error complained of has resulted in a miscarriage of justice’ . . . .” The defendant contended “that an examination of the entire cause discloses the verdict to be in accord with justice.” This Court responded:
The statute is ineffective to change the rule always in effect in Michigan, both before and after the enactment, that the question of reversal is controlled by determination of whether the error was prejudicial. [Emphasis added.]
The full text of the paragraph in Kotteakos from which the standard I would adopt is derived, reads as follows:
If, when all is said and done, the conviction is sure that the error did not influence the jury, or had but very slight effect, the verdict and the judgment should stand, except perhaps where the departure is from a constitutional norm or a specific command of Congress. But if one cannot say, with fair assurance, after pondering all that happened without stripping the erroneous action from the whole, that the judgment was not substantially swayed by the error, it is impossible to conclude that substantial rights were not affected. The inquiry cannot be merely whether there was enough to support the result, apart from the phase affected by the error. It is rather, even so, whether the error itself had substantial influence. If so, or if one is left in grave doubt, the conviction cannot stand. [Emphasis added; citations omitted.]
See also United States v Lane, 474 US 438, 450; 106 S Ct 725; 88 L Ed 2d 814 (1986), in which the United States Supreme Court found that “[i]n *254the face of overwhelming evidence of guilt shown here” that any error “failed to have any ‘substantive influence’ on the verdict,” citing Kotteakos, supra, p 765. Justices Brennan’s and Stevens’ separate opinions are instructive.
See 507 US 631.
The Court said that the “Kotteakos [harmless-error] standard is thus better tailored to the nature and purpose of collateral review, and more likely to promote the considerations underlying our recent habeas cases.” [Id., pp 637-638.]
Id., p 641.
The majority and the dissenters in O’Neal v McAninch, supra, were in agreement that the reviewing court should study the entire record. The majority said that a record review ordinarily will enable the reviewing court to make up its mind about the harmlessness of the error. And, indeed, the reviewing court has an obligation to review the record in order to make such a judgment.
Justice Stevens continued his exposition of Kotteakos, stating:
In a passage that should be kept in mind by all courts that review trial transcripts, Justice Rutledge wrote that the question is not
“were they [the jurors] right in their judgment, regardless of the error or its effect upon the verdict. It is rather what effect the error *255had or reasonably may be taken to have had upon the jury’s decision. The crucial thing is the impact of the thing done wrong on the minds of other men, not on one’s own, in the total setting.
“This must take account of what the error meant to them, not singled out and standing alone, but in relation to all else that happened. And one must judge others’ reactions not by his own, but with allowance for how others might react and not be regarded generally as acting without reason. This is the important difference, but one easy to ignore when the sense of guilt comes strongly from the record.” [Id., pp 642-643 (citations omitted).]
O’Neal v McAninch was a federal habeas corpus proceeding. Although a federal habeas corpus court first must find federal constitutional or federal statutory error, the harmless-error analysis proceeds, by reason of the Court’s decision in Brecht v Abrahamson, under the approach set forth in Kotteakos and not the Chapman beyond-a-reasonable-doubt standard.
O’Neal, 513 US 435.
Id., 513 US 436.
The Court continued:
The case before us does not involve a judge who shifts a “burden” to help control the presentation of evidence at a trial, but rather involves judges who apply a legal standard (harmlessness) to a record that the presentation of evidence is no longer likely to affect. In such a case, we think it conceptually clearer for the judge to ask directly, “Do I, the judge, think that the error substantially influenced the jury’s decision?” than for the judge to try to put the same question in terms of proof burdens (e.g., “Do I believe the party has borne its burden of showing . . . ?”) As Chief Justice Traynor said:
“Whether or not counsel are helpful, it is still the responsibility of the . . . court, once it concludes there was error, to determine whether the error affected the judgment. It must do so without benefit of such aids as presumptions or allocated burdens of proof that expedite fact-finding at the trial.” R. Traynor, The Riddle of Harmless Error 26 (1970) .... [Id.\
Id., 513 US 448.
O’Neal, in contrast with the instant case, was not a direct appeal, but rather a federal habeas corpus proceeding, requiring that the petitioner establish a violation of the federal constitution or a federal statute.
Ante, pp 207 and 218.
Id., pp 218-219.
Stanaway, supra, p 698 (Riley, J., concurring).
See ns 1-3, 14, and discussing Nichols, Soltar, and Bigge, n 32 supra, and accompanying text.
Ante, p 206.
The United States Supreme Court in O’Neal did not itself undertake a review of the record to determine whether the error was harmless. It remanded for further proceedings to the United States Court of Appeals *264for the Sixth Circuit after having stated its conclusion that the Sixth Circuit had erred in stating that the habeas petitioner must bear the burden of establishing whether the error was prejudicial.
In the instant case, in contrast with O’Neal at the federal Court of Appeals level, the Michigan Court of Appeals applied at least as high a standard in reviewing for harmlessness as we' conclude is applicable. We note, however, Justice Brennan’s observation in United States v Lane, n 33 supra, p 465, that the United States Supreme Court, and, I would add immodestly, this Court, is ordinarily manifestly ill-equipped to undertake harmless-error analysis without the benefit of prior harmless-error analysis by the Court of Appeals.
Cantu testified that he was assaulted with a gun and a knife. As the Court of Appeals observed, Cantu was treated at a hospital for multiple lacerations, and a hand gun was recovered from the bathroom where he claimed the assault took place. Police officers testified that the bathroom door had been broken, thereby confirming that aspect of Cantu’s testimony.
29A Am Jur 2d, Evidence, § 1441, pp 822-823. See People v Maciejewski, 68 Mich App 1, 3; 241 NW2d 736 (1976).
See People v Grant, supra.
We decline the invitation of the amicus curiae Prosecuting Attorneys Association of Michigan, to expand on what was said in Grant respecting review for plain error when the issue has not been preserved for appellate review.
United States v Lane, n 33 supra, pp 464-465 (Brennan, J.).
None of these witnesses, Cantu, who testified that Mateo assaulted him, Cantu’s mother, and Brecht and Blair, former girlfriends of Mateo, were disinterested witnesses.
Pickens, supra, p 314.
Grant, supra, p 553; Pickens, supra, p 312.