The jury in this case, after receiving instructions from the trial court that erroneously defined one element of robbery, returned verdicts convicting defendant of robbery and certain robbery-related offenses. On defendant’s appeal, the Court of Appeal reversed these convictions. This court granted review to decide whether, as the Court of Appeal concluded, the erroneous jury instruction so tainted the verdicts on the robbery and robbery-related offenses as to require that the convictions be reversed and the matter retried.
The determination of prejudice presents unusual difficulties. The charge of robbery was presented to the jury under alternative factual theories involving the taking of different items of property at different times from the same victim, the erroneous instruction was clearly misleading as to some theories but not others, and the verdict does not disclose which theory or theories the jury accepted. Because the error implicates the right to jury trial under the federal Constitution, the determination of prejudice must be made according to federal standards set by the United States Supreme Court, but the high court’s recent decisions leave room for doubt as to the correct analytical approach in cases such as this one.
The majority’s analysis I find difficult to follow. As I understand it, the majority concludes that the instructional error was not prejudicial because we as a reviewing court may infer that the jury actually rested its robbery verdict on a factual theory or theories as to which the erroneous instruction was not implicated. We may so infer, the majority reasons, because the evidence in support of guilt on those theories was so overwhelming as to establish guilt as a matter of law. I am not persuaded, however, that the relevant decisions of the United States Supreme Court permit us to draw such an inference about the actual basis of a jury’s verdict.
I agree that defendant was not prejudiced by the instructional error, but I arrive at this conclusion by a different approach that, in my view, is more *453easily reconciled with federal precedent. What makes the error harmless, under this approach, is that defendant, through his testimony at trial and his attorney’s argument to the jury, admitted every fact needed to prove his guilt of the crime of robbery. Because the elements of robbery were established by undisputed facts, there were no factual issues for the jury to decide and therefore it is of no constitutional significance that, for all we know, the jury may have relied to some extent on the erroneous instruction in reaching its guilty verdict on the robbery charge.
Were this not a case of guilt established by undisputed facts, reversal of the robbery conviction would appear to be required by controlling decisions of the United States Supreme Court, as explained in the dissenting opinion of Justice Mosk. Because this is a case of guilt established by undisputed facts, and not simply a case of overwhelming evidence of guilt, I do not join Justice Mosk in dissent.
I
The relevant facts are these:
Three young men were attempting to steal an automobile when its owner appeared. The young men subdued the vehicle’s owner, whom they took, along with his automobile, to a house where defendant was staying. One of the young men entered the house and told defendant what had happened, while the other two remained outside with their captive. Defendant, whom the others regarded as their leader, gave instructions to bind and blindfold the victim. When his instructions had been carried out (using handcuffs defendant had supplied), defendant went out to the automobile in which the victim was confined. One of the young men handed defendant the victim’s wallet. Defendant removed and kept for himself the money he found in the wallet.
Thereafter, defendant and the three members of his “team” proceeded to loot the victim’s home and business and to use the victim’s automatic teller machine access cards to obtain money from his accounts. Defendant was a full participant in these activities, and it was defendant who rented a motel room in which the victim was confined against his will. In defendant’s absence, the others took the victim to a remote location where they shot him to death. When informed of the murder, defendant assisted in concealing the body, withdrew more money from the victim’s accounts, and attempted to sell the victim’s automobile.
During his testimony at trial, defendant admitted each of these facts, although he maintained that his initial motive in assisting the kidnapping was merely to control the behavior of his subordinates so that the situation would not get out of hand.
*454In argument to the jury, defendant’s trial attorney conceded that defendant was guilty of most of the crimes with which he was charged, including kidnapping, three counts of burglary, two counts of auto theft, grand theft of a firearm, and fraudulent use of bank access cards. Defense counsel challenged only the charges of robbery, kidnapping for robbery, and murder.
Regarding the robbery charge, defense counsel conceded that defendant’s three companions had committed robbery by forcibly taking the victim’s automobile, his keys, and his wallet. But counsel argued that, as to the taking of these items, the crime of robbery had terminated before defendant joined the enterprise, and therefore defendant was guilty not of robbery but only of the lesser offense of receiving stolen property. Counsel conceded that defendant had participated in the later thefts from the victim’s home, business, and bank accounts, but counsel argued that none of these thefts would support a robbery conviction because none of the items had been taken from the victim’s person or immediate presence.
The prosecutor, in argument to the jury, disputed defense counsel’s assertion that the robbery involving the taking of the victim’s automobile, keys, and wallet had ceased before defendant joined the enterprise. He argued that the robbery was ongoing so long as the robbers were “encumbered” with the robbery victim and that defendant, by assisting in the restraint and transportation of the victim, had aided and abetted the continuing robbery. But the prosecutor did not limit the robbery charge to the taking of the victim’s automobile, keys, and wallet. He also argued that defendant could be found guilty of the robbery charge based on the later thefts from the victim’s house and business. Relying on an erroneous jury instruction that an act of robbery occurs in the immediate presence of the victim so long as the victim perceives any overt act connected with the offense, the prosecutor argued that the “immediate presence” element was satisfied because the victim had overheard conversations in which the “team” discussed plans to loot the victim’s residence and business and because those who participated in these takings had questioned the victim about what they would find inside.1
*455II
In this state, robbery is defined as “the felonious taking of personal property in the possession of another, from his person or immediate presence, and against his will, accomplished by means of force or fear.” (Pen. Code, §211.) Thus, one of the elements of robbery is that the personal property of the victim be taken either from the victim’s person or from the victim’s immediate presence. In People v. Hayes (1990) 52 Cal.3d 577, 626-628 [276 Cal.Rptr. 874, 802 P.2d 376], this court held that the term “immediate presence,” as used in the definition of robbery, means that the property is “ ‘ “so within [the victim’s] reach, inspection, observation or control” ’ ” that the victim could have retained possession had the victim not been prevented or deterred from doing so by the perpetrator’s use of force or fear.
This case was tried before this court’s decision in People v. Hayes, supra, 52 Cal.3d 577. Instructing the jury on the meaning of “immediate presence” as used in the definition of robbery, the trial court told the jury: “The act of robbery is deemed to have occurred within the immediate presence of a victim so long as the victim perceived any overt act connected with the commission of the offense. Any and all of the sensory perceptions of the victim are to be considered in determining presence.” As this court explained in People v. Hayes, this instruction is erroneous because it deprives the term “immediate presence” of any independent meaning, making it duplicative of the “force or fear” element of robbery. (Id. at pp. 627-628.)
When a trial court, in instructions to the jury, incorrectly defines one of the elements of the crime with which the defendant is charged, the defendant’s conviction of that crime can be affirmed on appeal only if the reviewing court is persuaded beyond a reasonable doubt that the error did not “contribute to” the verdict. (Chapman v. California (1967) 386 U.S. 18, 24 [17 L.Ed.2d 705, 710-711, 87 S.Ct. 824,24 A.L.R.3d 1065]; see also Pope v. Illinois (1987) 481 U.S. 497, 502 [95 L.Ed.2d 439, 446, 107 S.Ct. 1918]; People v. Hayes, supra, 52 Cal.3d 577, 628.) To determine whether an error “contributed to” a verdict, a reviewing court does not ask whether a hypothetical jury in a hypothetical trial in which the error did not occur would *456surely have reached the same verdict. (Sullivan v. Louisiana (1993) 508 U.S. _,_ [124 L.Ed.2d at pp. 182, 188-189, 113 S.Ct. 2078, 2081].) Rather, the reviewing court must ask whether the guilty verdict actually rendered in this trial was “surely unattributable to the error.” (Ibid.) This is because the Sixth Amendment right to jury trial means that the jury, and not a reviewing court, must find beyond a reasonable doubt every fact needed to convict.
To determine whether a verdict was “surely unattributable” to a trial error, the reviewing court ordinarily should first determine what evidence the jury actually considered in reaching its verdict by analyzing the trial evidence in light of the instructions, presuming that the jury considered relevant evidence when told it might do so. Next, the reviewing court must weigh the evidence properly considered against any tainted evidence or other improper jury influence resulting from the error. The conviction may be upheld only if this weighing process shows that the improper influence was “unimportant in relation to everything else the jury considered on the issue in question, as revealed in the record.” (Yates v. Evatt (1991) 500 U.S. 391, 403 [114 L.Ed.2d 432, 449, 111 S.Ct. 1884].)
In this case, the task of assessing harmless error by means of the described analysis is complicated because (1) the prosecutor urged the jury to base its robbery verdict on any of several different takings, (2) the jury’s verdict does not disclose which taking or takings the jury based its verdict upon, and (3) the significance of the instructional error varies greatly depending upon which taking or takings the jury considered in reaching its verdict.
If the jury based its robbery verdict on the taking of the victim’s wallet, for example, the instructional error would be inconsequential. The wallet was taken from the victim’s person, and because robbery may be committed by a taking from either the person or the immediate presence of the victim, the jury would never have occasion to apply the definition of “immediate presence” if it based its verdict entirely upon a taking from the victim’s person. On the other hand, if the jury based its robbery verdict on the thefts from the victim’s residence, the instructional error would most likely be prejudicial. These thefts did not consist in the taking of items from the victim’s person, and whether the items were taken from the victim’s immediate presence presented a factual issue for the jury. (See People v. Hayes, supra, 52 Cal.3d 577, 628-629.) Because the jury presumably would have used the erroneous “immediate presence” instruction to decide this issue, it would have proceeded by asking the wrong questions and would never have made the required factual determination.
The majority evidently believes that we, as a reviewing court, may infer that the jury rested its robbery verdict on the taking of the wallet or similar *457items from the victim’s person because the prosecution’s evidence in support of this robbery theory was so strong as to establish defendant’s guilt as a matter of law. But, when viewed in the distorting light of the erroneous jury instruction on “immediate presence,” the evidence that defendant was guilty of robbery for the thefts from the victim’s home and residence may well have seemed equally compelling. To base the robbery verdict on the taking of the wallet and other items from the victim’s person, the jury would have had to decide whether the robbery as to these items had terminated before defendant’s involvement. To base the robbery verdict on the takings from the victim’s home or residence, the jury would have had to decide whether the taking of these items satisfied the element of immediate presence. Either theory would require the jury to resolve an issue that was vigorously contested by counsel in closing argument. Given the trial court’s instructions (including the erroneous “immediate presence” instruction) and counsel’s arguments, these alternative theories might well have appeared equally problematic, or equally sound, to a reasonable juror. How then can a reviewing court be confident, beyond a reasonable doubt, that the jury actually rested its verdict on the taking of the wallet or some other item from the victim’s person?
Justice Mosk’s concurrence and dissent takes the view that because the verdict does not reveal which robbery theory or theories the jury relied upon, and because it is certainly conceivable, if not likely, that the jury based the robbery verdict in whole or in part upon a theory as to which the “immediate presence” instruction would have been prejudicially misleading, a reviewing court may not say beyond a reasonable doubt that the robbery verdict was “surely unattributable” to the instructional error. Although I agree with Justice Mosk’s concurrence and dissent that we, as a reviewing court, may not infer that the jury actually rested its verdict on a taking from the victim’s person, nonetheless there appears to be a sound basis upon which to conclude that defendant was not prejudiced by the erroneous jury instruction.
Under the United States Supreme Court’s decisions in Yates v. Evatt, supra, 500 U.S. 391, and Sullivan v. Louisiana, supra, 508 U.S._ [124 L.Ed.2d 182, 113 S.Ct. 2078], the focus of the harmless error analysis is upon the verdict actually rendered in this case, not the verdict that would have been rendered in a hypothetical trial free of the instructional error. Thus, if an instructional error prevents the jury from making a necessary factual finding, it is generally irrelevant that the evidence supporting that finding was overwhelming. In the words of Justice Scalia, “misdescription of an element of the offense . . . deprives the jury of its factfinding role” and thus is “not curable by overwhelming record evidence of guilt.” (Carella v. California (1989) 491 U.S. 263, 270 [105 L.Ed.2d 218, 225, 109 S.Ct. 2419] (conc. opn. of Scalia, J.).)
*458Here, however, the record shows not merely overwhelming evidence, but undisputed facts. The distinction is critical, as I shall explain.
The United States Supreme Court commented on the role of undisputed facts in harmless error analysis in Connecticut v. Johnson (1983) 460 U.S. 73 [74 L.Ed.2d 823,103 S.Ct. 969], There, the trial court had instructed the jury that “ ‘the law presumes that a person intends the ordinary consequences of his voluntary acts.’ ” (Id. at pp. 74-75 [74 L.Ed.2d at p. 826].) The high court had previously held, in Sandstrom v. Montana (1979) 442 U.S. 510 [61 L.Ed.2d 39, 99 S.Ct. 2450], that such an instruction violates the due process clause of the Fourteenth Amendment. In Connecticut v. Johnson, four justices of the United States Supreme Court joined in a plurality opinion stating that when such an erroneous instruction has been given, a reviewing court may not find the error harmless on the basis that the evidence of criminal intent was overwhelming. (460 U.S. at p. 85 [74 L.Ed.2d at p. 833].) But the plurality also stated that the error might be found harmless if the fact of criminal intent was undisputed: “[A] Sandstrom error may be harmless if the defendant conceded the issue of intent. [Citations.] In presenting a defense such as alibi, insanity, or self-defense, a defendant may in some cases admit that the act alleged by the prosecution was intentional, thereby sufficiently reducing the likelihood that the jury applied the erroneous instruction as to permit the appellate court to consider the error harmless.” (Id. at p. 87 [74 L.Ed.2d at p. 835].)2 Justice Stevens concurred in the judgment but declined to reach the issue of harmless error. (Id. at p. 88 [74 L.Ed.2d at pp. 835-836].) The four dissenting justices would have adopted a less restrictive harmless error analysis. (Id. at pp. 98-99 [74 L.Ed.2d at pp. 835-836].) Thus, all but one member of the high court gave assent to the proposition that an instructional error effectively removing an issue from the jury’s consideration may be harmless if the defendant conceded the issue.
In a concurring opinion joined by three other justices, Justice Scalia has likewise acknowledged that instructional error effectively taking an issue from the jury may be deemed harmless when it relates to “an element of the crime that the defendant in any case admitted.” (Carella v. California, supra, 491 U.S. 263, 270 [105 L.Ed.2d 218, 225] (cone. opn. of Scalia, J.).) Although the United States Supreme Court has not yet directly held that undisputed facts may render harmless a trial court’s error in misdefining an element of a charged offense, I am persuaded, based on this concurring *459opinion by Justice Scalia and on the plurality opinion in Connecticut v. Johnson, supra, 460 U.S. 73, that it would so conclude. (See also U.S. v. Warren (9th Cir. 1993) 984 F.2d 325, 327 [stating that failure to instruct on an element of an offense is harmless if the omitted element was undisputed].)
Why is it that instructional error may be “cured” by undisputed facts but not by overwhelming evidence? When a fact is undisputed, it is reasonable to assume that the jury accepted it as true, because ordinarily the jury would have no reason to do otherwise. But it does not follow that the jury actually rested its verdict on the undisputed fact. An erroneous jury instruction may cause a jury to dismiss as irrelevant an undisputed fact that is essential to the guilt determination. Thus, an undisputed fact does not “cure” an instructional error by providing assurance that the jury actually rested its verdict on the undisputed fact.
The reason undisputed facts “cure” instructional error may be found in the nature of the constitutional guarantee of jury trial. In criminal cases, the right to jury trial is, primarily, a right to have the jury rather than a court decide every “issue of fact” arising in the trial of the criminal charge. But when a fact is undisputed, or the parties have stipulated to its existence, there is no “issue of fact” for the jury to resolve, and this aspect of the Sixth Amendment right to jury trial is not implicated.3 Otherwise stated, the federal Constitution gives an accused no right to have the jury decide the truth of a fact that the accused has elected not to contest.
Here, every fact needed to convict defendant of robbery based on the taking of the victim’s wallet was undisputed at defendant’s trial. It was undisputed (1) that defendant or a member of his team took the wallet from the victim’s person; (2) that the wallet was taken against the victim’s will; (3) that the taking was accomplished by force; and (4) that defendant and his companions intended to permanently deprive the victim of at least the cash contained in the wallet, if not the wallet itself. Defendant disputed none of these facts, admitting them through his testimony and through his lawyer’s argument to the jury. In that jury argument, defense counsel relied upon a theory that although the victim had been robbed of his wallet, the robbery *460had terminated before defendant received the wallet, and therefore defendant was guilty only of receiving stolen property. This theory was legally defective; the purported defense was no defense at all.
The flaw in counsel’s theory of defense is apparent upon review of the law of aiding and abetting. For purposes of determining liability as an aider and abettor, “the commission of a robbery continues until all acts constituting the offense have ceased.” (People v. Cooper (1991) 53 Cal.3d 1158, 1164 [282 Cal.Rptr. 450, 811 P.2d 742], italics & fn. omitted.) And this court has recently emphasized that “in determining the duration of an offense, for the purpose of aider and abettor liability, the court must take into account the nature of the interests that the penal provision is intended to protect.” (People v. Montoya (1994) 7 Cal.4th 1027, 1040 [31 Cal.Rptr.2d 128, 874 P.2d 903].) This court noted also that “both the victim of a crime and a potential aider and abettor frequently will not perceive an offense as ‘completed’ simply because all elements necessary to establish guilt already have been satisfied.” (Ibid.)
Here, the force used to effect the taking of the wallet had not ceased when defendant joined the enterprise. The members of defendant’s team had overpowered the victim and brought him directly to defendant. The victim was still being forcibly restrained and held captive when defendant obtained the wallet. Thus, the acts constituting the robbery of the wallet (that is, the application of force or fear to separate the victim from his property) had not ceased. (See People v. Cooper, supra, 53 Cal.3d 1158, 1164.) Moreover, robbery is punished to protect the victim’s interests in avoiding both loss of property and physical harm. (People v. Guerin (1972) 22 Cal.App.3d 775, 782 [99 Cal.Rptr. 573].) Consideration of these interests dictates that a robbery not be considered terminated while the robbers continue to hold the victim’s property and the victim remains their captive and at risk of immediate physical harm.
Here, by appropriating the wallet that had been forcibly taken from the victim, with the intent to permanently deprive the victim of the cash in the wallet, and by knowingly and intentionally assisting in the victim’s continued restraint, defendant joined the continuing robbery either as an aider and abettor or as an additional direct perpetrator.
III
The diversity of views reflected in the several opinions authored in this court suggests that the controlling federal prejudice standard is not as clear *461as it might be, and that further elucidation by the United States Supreme Court would be useful.4
The difficulty of determining and applying the federal harmless error standard in a case like this becomes apparent upon examination of the high court’s decision in Pope v. Illinois, supra, 481 U.S. 497, which involved a prosecution for selling pornographic literature. In its instructions to the jury, the trial court had erroneously instructed the jury to use “contemporary community standards,” rather than an objective “reasonable person” standard, to determine whether the allegedly obscene material lacked serious literary, artistic, political, or scientific value. (See Smith v. United States (1977) 431 U.S. 291, 301 [52 L.Ed.2d 324, 334-335, 97 S.Ct. 1756]; Miller v. California (1973) 413 U.S. 15, 24 [37 L.Ed.2d 419, 430-431, 93 S.Ct. 2607].) The United States Supreme Court held that the instructional error, because it did not entirely preclude the jury from considering the question of value, would be harmless if “no rational juror, if properly instructed, could find value in the magazines” (Pope v. Illinois, supra, 481 U.S. at p. 503 [95 L.Ed.2d at p. 447]), and it remanded the case to a state court for the resolution of this question.
Justice Stevens dissented, taking the position that misinstruction on an element of an offense could not be harmless: “It is fundamental that an appellate court (and for that matter, a trial court) is not free to decide in a criminal case that, if asked, a jury would have found something that it did not find. We have consistently rejected the possibility of harmless error in these circumstances.” (Pope v. Illinois, supra, 481 U.S. 497, 509-510 [95 L.Ed.2d 439, 451] (dis. opn. of Stevens, J.).)
Although this was the most recent case in which the United States Supreme Court has explained the harmless error analysis that a reviewing court should use when a trial court’s jury instructions incorrectly defined an element of a charged offense, and although the holding in that case has not been expressly overruled, there are grounds to doubt that the harmless error analysis set forth in the majority opinion in Pope v. Illinois, supra, 481 U.S. 497, has continuing vitality as precedent.
*462As authority for its harmless error methodology, the court had relied on one of its earlier decisions: “The situation here is comparable to that in Rose v. Clark 478 U.S. 570 [92 L.Ed.2d 460, 106 S.Ct. 3101] (1986). In Rose, the jury in a murder trial was incorrectly instructed on the element of malice, yet the Court held that a harmless-error inquiry was appropriate. The Court explained that in the absence of error that renders a trial fundamentally unfair, such as denial of the right to counsel or trial before a financially interested judge, a conviction should be affirmed ‘[w]here a reviewing court can find that the record developed at trial established guilt beyond a reasonable doubt . . . .’ Id., at 579.” (Pope v. Illinois, supra, 481 U.S. 497, 502-503 [95 L.Ed.2d 439, 446], fn. omitted.)
Just four years later, the court stated that this very language (that misinstruction on an element of an offense is harmless “ ‘[w]here a reviewing court can find that the record developed at trial establishes guilt beyond a reasonable doubt’ ”) was, at least when viewed in isolation, “not correct.” (Yates v. Evatt, supra, 500 U.S. 391, 402-403, fn. 8 [114 L.Ed.2d 432, 448].)
In a still later case, the high court approvingly cited Justice Stevens’s dissent in Pope v. Illinois, supra, 481 U.S. 497, for the proposition that “to hypothesize a guilty verdict that was never in fact rendered—no matter how inescapable the findings to support that verdict might be—would violate the jury-trial guarantee.” (Sullivan v. Louisiana, supra, 508 U.S._,_ [124 L.Ed.2d 182, 189, 113 S.Ct. 2078, 2082].)
Thus, it would seem that the harmless error analysis prescribed in Pope v. Illinois, supra, 481 U.S. 497, has been disapproved and replaced by the form of analysis outlined in Yates v. Evatt, supra, 500 U.S. 391, and Sullivan v. Louisiana, supra, 508 U.S._[124 L.Ed.2d 182, 113 S.Ct. 2078]. Yet the high court has not itself applied the newer mode of analysis to a jury instruction misdefining an element of a charged offense, but only to jury instructions containing improper presumptions or misdefining the “reasonable doubt” standard of proof. And the high court has never attempted to apply the Yates!Sullivan analysis to a situation, such as presented in this case, in which the jury is given alternative factual theories of guilt, the verdict does not disclose which theory or theories the jury accepted, and the erroneous instruction would be misleading as to some theories of guilt but not others. Lacking firm guidance from the high court in situations directly analogous to the one we face here, this court’s resolution of the harmless error problem must of necessity involve a certain degree of inference and *463informed speculation based on our readings of the pertinent decisions of the high court.5
IV
Although a defendant is entitled to have the jury decide every issue of fact material to guilt, facts that are undisputed do not raise “issues of fact.” Here, the verdict convicting defendant of robbery may be deemed to incorporate findings of each material undisputed fact, and those findings establish every element of robbery. Applying the federal harmless-error standard as best I am able to determine it, I conclude beyond a reasonable doubt that the robbery verdict against defendant was “surely unattributable to the error” that the trial court committed when it instructed the jury with an incorrect definition of the “immediate presence” element of robbery.
“Now, the judge told you yesterday what that meant. An act of robbery occurs in the immediate presence of the person who is robbed as long as that person perceives any overt, any overt act connected with the offense.
“Now, when you look at that issue, you are to look at all of that person’s perceptual abilities: namely, his ability to see, his ability to hear, and the other sensory perceptions that a person might have.
“What about the items that were taken from the business? They weren’t taken from his person. They weren’t taken from his immediate presence. He wasn’t in actual possession, but he had constructive possession of those items of property. Taken from his immediate presence because he was able to perceive several overt acts associated with the taking of that property. When he was sitting out in front of Shon Maple’s house, clearly he must have heard them *455talking about going back to the business. On the way to the business, Valdez F[.] was asking him questions. Once at the business, there was a point in time when Shon Maple—and again, this is before any property is taken—specifically asked [the victim] about the building that he had come out of.
“I suggest to you, ladies and gentlemen, that these overt acts associated with the commission of the taking of that property made the taking of that property within [the victim]’s immediate presence as defined by the law, even though [the victim] was out in the car at the time the items were taken.”
In a footnote, the plurality noted that “a defendant in a criminal trial is justified, of course, in defending solely in reliance on the presumption of his innocence and the State’s burden of proof.” (Connecticut v. Johnson, supra, 460 U.S. 73, 87, fn. 16 [74 L.Ed.2d 823, 835].) In other words, a fact becomes undisputed in a criminal trial only if the defense affirmatively concedes it.
I do not mean to suggest that if all facts material to guilt were undisputed, the trial court could direct a verdict for the government or dispense with a jury verdict entirely and enter a judgment of conviction. I understand the Sixth Amendment to the federal Constitution as requiring a guilt determination by the jury, not by the court, even where the material facts are undisputed. What I am suggesting is that once the jury has made that determination, misinstruction on any single element of the charged offense, or indeed on every element of the offense, will be harmless if the element or elements in question were established by undisputed facts.
Quoting from my dissenting opinion in People v. Cahill (1993) 5 Cal.4th 478, 554 [20 Cal.Rptr.2d 582, 853 P.2d 1037], Justice Mosk argues that we stand in no need of “guidance” from the United States Supreme Court because we are “no less capable . . . than any other court” of interpreting the federal Constitution. (Conc, and dis. opn. of Mosk, J., ante, p. 449, fn. 1.) But the quotation from Cahill is taken out of context. In that case, this court was interpreting our state Constitution and therefore was under no obligation to follow holdings of the United States Supreme Court construing similar provisions in the federal Constitution. On matters of federal constitutional law, such as application of the federal harmless error standard under Chapman v. California, supra, 386 U.S. 18, 24 [17 L.Ed.2d 705, 710-711], the situation is otherwise. The United States Supreme Court’s rulings are binding on this court. (See, e.g., People v. Clair (1992) 2 Cal.4th 629, 662 [7 Cal.Rptr.2d 564, 828 P.2d 705].)
At least one federal court agrees that there is ambiguity in the current state of law. In a recent opinion, the First Circuit remarked: “One might, or might not, read recent Supreme Court decisions to mean that where an incorrect instruction [on an element of an offense] is given, it may not be adequate for the government to show that the record evidence assured that a reasonable jury under proper instructions would have found the disputed element in favor of the government; rather, it may be the law that the jury must in fact have made this finding despite the erroneous instruction.” (U.S. v. Whiting (1st Cir. 1994) 28 F.3d 1296, 1309, fn. omitted.)