State v. Flores Ramos

— 292 — Argued and submitted August 18; decision of Court of Appeals affirmed in part and reversed in part, judgment of circuit court affirmed in part and reversed in part, and case remanded to circuit court for further proceedings December 24, 2020 STATE OF OREGON, Respondent on Review, v. ISIDRO FLORES RAMOS, aka Santiago Flores Martinez, Petitioner on Review. (CC 17CR30088) (CA A167187) (SC S067105) 478 P3d 515 At defendant’s trial, and over his objection, the jury was instructed that it could return nonunanimous guilty verdicts. The jury returned five guilty verdicts, four of which were unanimous and one of which was nonunanimous. The Court of Appeals affirmed defendant’s convictions. Held: (1) The jury instruction permitting the jury to return nonunanimous verdicts violated defendant’s Sixth Amendment rights under Ramos v. Louisiana, 590 US ___ , 140 S Ct 1390 , 206 L Ed 2d 583 (2020); (2) the instructional error was not a structural error; (3) the erroneous instruction was harmless beyond a reasonable doubt as to the convictions based on unanimous verdicts; (4) under State v. Ulery, 366 Or 500 , 464 P3d 1123 (2020), defendant was entitled to reversal of the single conviction based on a nonunanimous verdict. The decision of the Court of Appeals is affirmed in part and reversed in part. The judgment of the circuit court is affirmed in part and reversed in part, and the case is remanded to the circuit court for further proceedings. En Banc On review from the Court of Appeals.* Erik Blumenthal, Deputy Public Defender, Office of Public Defense Services, Salem, argued the cause and filed the briefs for petitioner on review. Also on the brief were Ernest G. Lannet, Chief Defender, and Joshua B. Crowther, Deputy Public Defender. Christopher A. Perdue, Assistant Attorney General, Salem, argued the cause and filed the brief for respondent on ______________ * On appeal from Clackamas County Circuit Court, Douglas V. Van Dyk, Judge. 298 Or App 841 , 449 P3d 572 (2019). Cite as 367 Or 292 (2020) 293 review. Also on the brief were Ellen F. Rosenblum, Attorney General, Benjamin Gutman, Solicitor General, and Doug M. Petrina, Assistant Attorney General. Scott Sell, Thomas, Coon, Newton & Frost, Portland, filed the brief on behalf of amicus curiae Street Roots. Jonathan Zunkel-deCoursey, Schwabe, Williamson & Wyatt, P.C., Portland, filed the brief on behalf of amicus curiae Immigrant and Refugee Community Organization. Also on the brief was Jeanice Chieng, Immigrant and Refugee Community Organization, Portland. Cody Hoesly, Larkins Vacura Kayser LLP, Portland, filed the brief on behalf of amici curiae NAACP Corvallis Branch #1118, NAACP Eugene-Springfield Branch, #1119, NAACP Portland Chapter 1120B, and NAACP Salem-Keizer Branch #1166. Timothy Wright, Tonkon Torp LLP, Portland, filed the brief for amicus curiae Don’t Shoot Portland. Also on the brief was J. Ashlee Albies, Albies & Stark, Portland. Nathan R. Morales, Perkins Coie LLP, Portland, filed the brief on behalf of amici curiae The Coalition of Communities of Color and Latino Network. Also on the brief was Misha Isaak. Aliza B. Kaplan filed the brief on behalf of amicus curiae Criminal Justice Reform Clinic at Lewis & Clark Law School. Also on the brief was Sarah Laidlaw. GARRETT, J. The decision of the Court of Appeals is affirmed in part and reversed in part. The judgment of the circuit court is affirmed in part and reversed in part, and the case is remanded to the circuit court for further proceedings. 294 State v. Flores Ramos GARRETT, J. In this case, we again consider the effect of the United States Supreme Court’s decision in Ramos v. Louisiana, 590 US ___ , 140 S Ct 1390 , 206 L Ed 2d 583 (2020), which held that the Sixth Amendment1 requires a jury to be unanimous in order to convict a defendant of a serious offense. We have held that Ramos requires reversal of Oregon convictions based on nonunanimous jury verdicts. State v. Ulery, 366 Or 500 , 464 P3d 1123 (2020). This case presents a different issue: After being instructed that it could convict defendant by a vote of 10 to two, the jury found defendant guilty of five crimes, four by unanimous verdicts and one by a nonunanimous verdict. Under Ramos and Ulery, the one conviction based on a nonunanimous verdict must be reversed. The additional question that we must answer in this case is whether the convictions based on unanimous verdicts must also be reversed, because the jury that returned them was instructed that it could convict defendant without reaching unanimity. Although we agree with defendant that instructing the jury that it could convict him by a nonunanimous vote violated the Sixth Amendment, we conclude that the error does not require any of defendant’s unanimous convictions to be reversed. I. BACKGROUND A. Legal Context We first clarify what we already have decided and the limited scope of the issues to be decided in this case. In Ramos, the Supreme Court held that the Sixth Amendment requires that the jury be unanimous to convict a criminal defendant of a serious offense and that that requirement is binding on the states through the Due Process Clause of the Fourteenth Amendment. 590 US at ___, 140 S Ct at 1397 . The rule announced in Ramos applies to all cases now on appeal—regardless of whether the trial occurred before or after Ramos. Griffith v. Kentucky, 479 US 314 , 107 S Ct 708 , 93 L Ed 2d 649 (1987) (holding that new rules of — 1 — The Sixth Amendment to the United States Constitution provides that, “[i]n all criminal prosecutions, the accused shall enjoy the right to a speedy and public trial, by an impartial jury of the State and district wherein the crime shall have been committed[.]” Cite as 367 Or 292 (2020) 295 constitutional law apply to all cases still on direct appeal). Before Ramos, in every felony case tried to a jury in Oregon, a nonunanimous verdict of 10 votes out of 12 was sufficient for a conviction of any offense other than murder, and juries were so instructed. See Or Const, Art I, § 11 (“[I]n the circuit court ten members of the jury may render a verdict of guilty or not guilty, save and except a verdict of guilty of first degree murder, which shall be found only by a unanimous verdict, and not otherwise[.]”). In many of those cases, the jury was polled, and the jury poll revealed that only 10 or 11 jurors agreed with the verdict on one or more counts of conviction. Ramos makes clear that all convictions for serious offenses that were based on nonunanimous verdicts involved constitutional error—a violation of the defendant’s Sixth Amendment right to jury unanimity. Not every constitutional error requires reversal of a conviction, but, in Ulery, we held that the receipt of a nonunanimous guilty verdict always does. 366 Or at 504 . That is, we held that acceptance of a nonunanimous guilty verdict represents a sufficiently grave error to require reversal of the conviction, when the error is properly presented to an appellate court on appeal. We further held in Ulery that reversal of nonunanimous convictions was appropriate even if the error had not been preserved in the trial court. As a general rule, Oregon appellate courts will consider assignments of error only where the error was properly objected to at trial. ORAP 5.45(1). In many cases, jurors were instructed that they could return nonunanimous guilty verdicts, and nonunanimous guilty verdicts were received, without any objection from the defendant—a circumstance that ordinarily would preclude appellate review. However, the state has conceded, and we have agreed, that receipt of nonunanimous verdicts qualifies as plain error, which is subject to reversal even when the assignment of error was not preserved. Ulery, 366 Or at 503 . And the receipt of a nonunanimous verdict is an error sufficiently grave that appellate courts should exercise their discretion to correct the error on appeal, despite the state’s interest “in avoiding the expense and difficulty associated with a retrial.” Id. at 504 . Further, in State v. Williams, 366 Or 495 , 466 P3d 55 (2020), we held that it was 296 State v. Flores Ramos appropriate to waive the rules of appellate procedure to permit consideration of the nonunanimous jury issue in cases where the issue might not otherwise be considered properly presented on direct appeal. Thus, under Ramos, Ulery, and Williams, the substantial majority of nonunanimous convictions on appeal at the time that Ramos was decided must be reversed, and many such convictions already have been reversed, typically by order rather than by published opinion. That much has already been decided. A significant question not yet resolved is whether Ramos requires convictions to be reversed when the jury was erroneously instructed that it could convict without being unanimous, but it nonetheless voted unanimously to convict—which is what happened with four of the counts in this case. The state presents a straightforward argument that a unanimous conviction renders the instructional error harmless because defendant ultimately received that to which he was entitled: unanimity. Thus, although the state agrees that defendant’s lone nonunanimous conviction must be reversed, it contends that the unanimous convictions should be upheld. Defendant advances several contrary arguments, which we address in this opinion. Before taking up those questions, we recite the facts of this case. B. Factual and Procedural Background Defendant broke into a home and sexually assaulted a nine-year-old girl. Defendant was charged with firstdegree unlawful sexual penetration, first-degree sexual abuse, attempted first-degree rape, first-degree burglary, and coercion. Before trial, he filed a motion requesting that the jury be instructed that it needed to be unanimous to convict. The trial court denied that motion. The jury was instructed that, “[a]s to each count, ten or more jurors must agree on your verdict. So whether your verdict is not guilty or guilty, at least ten of you must agree on that verdict. If you are divided nine to three, for example, you do not have a verdict.” After deliberations that lasted approximately an hour and a half, the jury returned guilty verdicts on each of the five counts. The trial court polled the jury by asking the jurors who voted “guilty” on each count to raise their hands. Cite as 367 Or 292 (2020) 297 The poll indicated that the jury had reached unanimous guilty verdicts on all counts except for the attempted firstdegree rape count. On that count, only 10 jurors had voted to convict. Defendant did not object to the manner in which the trial court polled the jury, and defense counsel indicated that he was satisfied by the poll. The trial court received the verdicts and entered a judgment based on them.2 Defendant appealed. As relevant here, he assigned error to both the use of the nonunanimous jury instruction and the receipt of the nonunanimous verdict—assignments of error that he had preserved in the trial court. He argued that those errors required reversal of all his convictions. In a decision issued before Ramos, the Court of Appeals affirmed defendant’s convictions without opinion. State v. Flores Ramos, 298 Or App 841 , 449 P3d 572 (2019). Defendant filed a petition for review in this court, which we held in abeyance until the Supreme Court issued its decision in Ramos. After Ramos was decided, we allowed review. As noted, the jury returned five guilty verdicts, four of which were unanimous. On the charge of attempted first-degree rape, the jury was not unanimous. The state concedes that, under Ramos, defendant’s conviction on that count cannot stand. We agree and reverse that part of the trial court’s judgment. What we address in this opinion are the other four counts, where, despite being instructed incorrectly, the jury nonetheless voted unanimously to convict. II. CONSTITUTIONAL ERROR The central dispute in this case is whether the trial court’s instructional error permitting nonunanimous guilty verdicts requires defendant’s unanimous convictions to be reversed, either because it amounted to a “structural” error that always requires reversal or, in the alternative, because the error was not harmless. Before turning to those arguments, we briefly address an additional argument made by the state, which suggests that no constitutional error occurred at all. — 2 — Although the jury returned five guilty verdicts, the first-degree unlawful sexual penetration and first-degree sexual abuse counts merged for purposes of conviction, so the judgment reflects four convictions. 298 State v. Flores Ramos Defendant argues that the Sixth Amendment, as incorporated through the Due Process Clause of the Fourteenth Amendment, was violated when the trial court instructed the jury that it could return a nonunanimous conviction. Although the state agrees that the instruction was erroneous, it disagrees that the Sixth Amendment was violated simply by the giving of the instruction. The state argues that an erroneous jury instruction amounts to a federal constitutional violation only if there is “ ‘a reasonable likelihood that the jury has applied the challenged instruction in a way’ that violates the Constitution.” Estelle v. McGuire, 502 US 62, 72 , 112 S Ct 475 , 116 L Ed 2d 385 (1991) (quoting Boyde v. California, 494 US 370, 380 , 110 S Ct 1190 , 108 L Ed 2d 316 (1990)). The state further argues that, because the jury was unanimous on the four counts in question, the jury necessarily did not apply the instruction in a way that violated the constitution. Boyde and McGuire are inapplicable. Those cases articulate a standard that applies to “claims that allegedly ambiguous instructions caused jury confusion.” Jones v. United States, 527 US 373, 390 , 119 S Ct 2090 , 144 L Ed 2d 370 (1999). “In such cases, constitutional error exists only if ‘there is a reasonable likelihood’ that the jury so interpreted the instruction.” Calderon v. Coleman, 525 US 141, 146 , 119 S Ct 500 , 142 L Ed 2d 521 (1998). McGuire suggests nothing different, as a fuller quotation of the passage excerpted by the state makes clear: “In addition, in reviewing an ambiguous instruction such as the one at issue here, we inquire ‘whether there is a reasonable likelihood that the jury has applied the challenged instruction in a way’ that violates the Constitution.” 502 US at 72 (quoting Boyde, 494 US at 380 ) (emphasis added). The state appears to read Boyde and McGuire to hold that whether a jury instruction violates the constitution depends on whether the instruction affected the jury’s verdict, even where the jury would certainly have understood the instruction in a manner that violated the constitution. But “[t]he Boyde analysis does not inquire into the actual effect of the error on the jury’s verdict[.]” Coleman, 525 US Cite as 367 Or 292 (2020) 299 at 147. That is, when the claim is that “the jury was given an ambiguous instruction that it might have interpreted” in an impermissible manner, the question is whether “there is a reasonable likelihood that the jury so interpreted the instruction.” Id. at 146 (emphasis added; internal quotation marks omitted). See United States v. Doyle, 130 F3d 523, 536 (2d Cir 1997) (“In other words, then, we do not engage in an inquiry of harmless error review such as was enunciated in Chapman v. California, 386 US 18, 23 , 87 S Ct 824 , 17 L Ed 2d 705 (1967), which looked at the case in its entirety to analyze the effect of the error on the jury’s verdict. Rather, we assess only the charge, taken as a whole, in order to determine whether there is a reasonable likelihood that the jury misinterpreted the reasonable doubt instruction.” (Emphases in original.)). Boyde and McGuire thus articulate an inquiry applicable only to ambiguous instructions. But the problem with the instruction challenged in this case is not that it was ambiguous. Rather, in light of Ramos, it was unambiguously wrong; it expressly told the jury that it could do what the Sixth Amendment forbids. Boyde and McGuire thus do not apply. And even if they did, they would have little to add; because the instruction was unambiguously incorrect, there is more than a reasonable likelihood “that the jury so interpreted the instruction.” Coleman, 525 US at 146 . We conclude that the Sixth Amendment is violated when a trial court tells the jury that it can convict a defendant of a serious offense without being unanimous. A unanimous verdict may render that constitutional violation harmless, as we explain in detail below, but it does not operate retroactively to prevent the violation from having occurred. III. STRUCTURAL ERROR Having concluded that the Sixth Amendment was violated when the jury was instructed that it could return a nonunanimous guilty verdict, we turn to the central question presented—whether that error requires reversal of defendant’s unanimous convictions. Most federal constitutional errors require reversal unless the error can be found “harmless beyond a reasonable doubt.” Chapman, 386 US at 24 . That is, the reviewing court must be satisfied that the 300 State v. Flores Ramos error did not affect the outcome. We address harmless error in section IV below. However, some federal constitutional violations qualify as “structural” errors, which is to say that the error is a “structural defect affecting the framework within which the trial proceeds, rather than simply an error in the trial process itself.” Arizona v. Fulminante, 499 US 279, 310 , 111 S Ct 1246 , 113 L Ed 2d 302 (1991). Structural error is not susceptible to a harmlessness analysis; if a structural error occurred, the conviction must be reversed. Defendant argues that the instructional error that occurred in this case was structural, requiring reversal of all his convictions. A. Structural Error Defined In arguing that the instructional error was structural, defendant must satisfy a high standard. The Supreme Court has “found an error to be ‘structural,’ and thus subject to automatic reversal, only in a ‘very limited class of cases.’ ” Neder v. United States, 527 US 1, 8 , 119 S Ct 1827 , 144 L Ed 2d 35 (1999) (quoting Johnson v. United States, 520 US 461, 468 , 117 S Ct 1544 , 137 L Ed 2d 718 (1997)). And, “[i]f the defendant had counsel and was tried by an impartial adjudicator, there is a strong presumption that any other errors that may have occurred are subject to harmless-error analysis.” Rose v. Clark, 478 US 570, 579 , 106 S Ct 3101 , 92 L Ed 2d 460 (1986). Consistent with that presumption, the Supreme Court has held that a wide variety of trial errors are subject to harmlessness analysis.3 The concept of structural error, — 3 — In Fulminante, 499 US at 306-07 , the Supreme Court offered the following partial list of errors subject to harmlessness analysis: “unconstitutionally overbroad jury instructions at the sentencing stage of a capital case”; “admission of evidence at the sentencing stage of a capital case in violation of the Sixth Amendment Counsel Clause”; “jury instruction containing an erroneous conclusive presumption”; “jury instruction misstating an element of the offense”; “jury instruction containing an erroneous rebuttable presumption”; “erroneous exclusion of defendant’s testimony regarding the circumstances of his confession”; “restriction on a defendant’s right to cross-examine a witness for bias in violation of the Sixth Amendment Confrontation Clause”; “denial of a defendant’s right to be present at trial”; “improper comment on defendant’s silence at trial, in violation of the Fifth Amendment Self-Incrimination Clause”; “statute improperly forbidding trial court’s giving a jury instruction on a lesser included offense in a capital case in violation of the Due Process Clause”; “failure to instruct the jury Cite as 367 Or 292 (2020) 301 by contrast, has been reserved for “basic protections” without which “a criminal trial cannot reliably serve its function as a vehicle for determination of guilt or innocence, and no criminal punishment may be regarded as fundamentally fair.” Clark, 478 US at 577-78 (citation omitted). The error at issue here is instructional, and the Supreme Court has held an instructional error to be structural only once. In Sullivan v. Louisiana, 508 US 275 , 113 S Ct 2078 , 124 L Ed 2d 182 (1993), the Court held that failure to properly instruct the jury on the “beyond a reasonable doubt” standard was structural error. As the Court explained in Neder, Sullivan’s holding rested on the fact that an improper reasonable-doubt instruction “ ‘vitiates all the jury’s findings’ and produces ‘consequences that are necessarily unquantifiable and indeterminate.’ ” Neder, 527 US at 11 (quoting Sullivan, 508 US at 281-82 ) (emphasis in Sullivan; internal citations omitted). By contrast, the Supreme Court has several times held significant instructional errors to be subject to a harmlessness analysis. In Neder the Court held that failure to instruct the jury as to an element of an offense is not structural error. The Court emphasized that “[the defendant] was tried before an impartial judge, under the correct standard of proof and with the assistance of counsel; a fairly selected, impartial jury was instructed to consider all of the evidence and argument in respect to [his] defense against the tax charges.” 527 US at 9. The Court has similarly held that unconstitutional mandatory presumptions and misinstruction on a single element of an offense are errors subject to harmlessness analysis. See Carella v. California, 491 US 263 , 109 S Ct 2419 , 105 L Ed 2d 218 (1989); Pope v. Illinois, 481 US 497 , 107 S Ct 1918 , 95 L Ed 2d 439 (1987); Clark, 478 US 570 . on the presumption of innocence”; “admission of identification evidence in violation of the Sixth Amendment Confrontation Clause”; “admission of the out-of-court statement of a nontestifying codefendant in violation of the Sixth Amendment Confrontation Clause”; “confession obtained in violation of Massiah v. United States, 377 US 201 , 84 S Ct 1199 , 12 L Ed 2d 246 (1964)”; “admission of evidence obtained in violation of the Fourth Amendment”; [and] “denial of counsel at a preliminary hearing in violation of the Sixth Amendment Confrontation Clause.” 302 State v. Flores Ramos In those cases, the Court has explained that, depending on the strength of the evidence presented at trial, the “erroneous instruction” may be “simply superfluous.” Clark, 478 US at 581 . Although removing an element from the jury’s consideration entirely, or incorrectly permitting the element to be decided based on a mandatory presumption, are undoubtedly serious Sixth Amendment violations, the Court has nonetheless been clear that such errors are not structural. The Supreme Court also has applied harmless error analysis even where the error was necessarily one that would have made an impression on the jury. In Fulminante, the Court held that admission of a defendant’s coerced confession, in violation of the Fifth Amendment, was subject to review for harmlessness. The Court recognized that “an involuntary confession may have a more dramatic effect on the course of a trial than do other trial errors—in particular cases it may be devastating to a defendant—but this simply means that a reviewing court will conclude in such a case that its admission was not harmless error; it is not a reason for eschewing the harmless-error test entirely.” Fulminante, 499 US at 312 . Similarly, in Bruton v. United States, 391 US 123 , 88 S Ct 1620 , 20 L Ed 2d 476 (1968), the Supreme Court held that a defendant’s Confrontation Clause rights were violated by the admission against a nontestifying codefendant at a joint trial of a confession by the codefendant that implicated the defendant as well. Even though the jury was instructed that it could not consider the confession as evidence against the defendant, the Court explained that the jury could not be presumed to have followed those instructions where “the powerfully incriminating extrajudicial statements of a codefendant, who stands accused side-by-side with the defendant, are deliberately spread before the jury in a joint trial.” 391 US at 135-36 . Nonetheless, even though the premise of the Bruton line of cases is that certain evidence put before the jury may be so powerful that the jury cannot ignore it even if instructed to do so, the Court nevertheless has held that Bruton error is subject to harmlessness analysis and may be held harmless based on other evidence admitted at Cite as 367 Or 292 (2020) 303 trial. Harrington v. California, 395 US 250, 254 , 89 S Ct 1726 , 23 L Ed 2d 284 (1969). In sum, the Supreme Court has rejected the notion of structural error in many circumstances that have involved violations of indisputably fundamental constitutional protections afforded to criminal defendants. The Court’s most recent substantial discussion of when an error is structural came in Weaver v. Massachusetts, ___ US ___, 137 S Ct 1899 , 198 L Ed 2d 420 (2017). In Weaver, the Court explained that it had held errors to be structural for at least three reasons. “First, an error has been deemed structural in some instances if the right at issue is not designed to protect the defendant from erroneous conviction but instead protects some other interest.” Id. An example given in Weaver is a defendant’s right to self-representation at trial: pro se representation typically makes a conviction more likely, not less, but wrongful denial of the right is a structural error because of its interference with “the fundamental legal principle that a defendant must be allowed to make his own choices about the proper way to protect his own liberty.” Id. “Second, an error has been deemed structural if the effects of the error are simply too hard to measure.” Id. The principal example given in Weaver is a defendant’s right to select his own retained counsel. Id. “Third, an error has been deemed structural if the error always results in fundamental unfairness”—for example, a denial of appointed counsel or the absence of a beyond-a-reasonable-doubt instruction. Id. Weaver does not hold that any of those conditions is sufficient to make an error structural. Neither, as Weaver acknowledges, does every example of structural error fall neatly into only one category. See id. (“In a particular case, more than one of these rationales may be part of the explanation for why an error is deemed to be structural.”). Rather, the purpose of that categorization, in Weaver itself, was simply to establish that “[a]n error can count as structural even if the error does not lead to fundamental unfairness in every case.” Id. Thus, although Weaver sets out important factors to consider, it 304 State v. Flores Ramos does not offer a clear rubric for evaluating whether an error is structural. In particular, we note that one of the bases for holding an error structural mentioned in Weaver—that the effects of the error are “simply too hard to measure,” ___ US at ___, 137 S Ct at 1908—often will have only a modest role to play in the analysis. Because the content of jury deliberations will remain unknown to the reviewing court—which can therefore never be certain about which path the jury took to its decision or what evidence jurors thought important— nearly all trial errors are capable of producing effects that are difficult to measure. Yet the Supreme Court has elsewhere recognized that many significant constitutional errors, despite having effects that are difficult to measure, are not structural. Referring to improper admissions of a defendant’s confession and violations of the Confrontation Clause, for example, the Court acknowledged that “[s]uch errors, no less than the failure to instruct on an element in violation of the right to a jury trial, infringe upon the jury’s factfinding role and affect the jury’s deliberative process in ways that are, strictly speaking, not readily calculable.” Neder, 527 US at 18 . Nevertheless, those errors are subject to a harmlessness analysis. Id. As another example, in Hedgpeth v. Pulido, 555 US 57, 58 , 129 S Ct 530 , 172 L Ed 2d 388 (2008), “the jury was instructed on alternative theories of guilt and may have relied on an invalid one” in convicting the defendant. Such an error, almost by definition, precludes any certainty about whether the error was harmless. Nevertheless, the Court held that the error was not structural, reasoning that “[a]n instructional error arising in the context of multiple theories of guilt no more vitiates all the jury’s findings than does omission or misstatement of an element of the offense when only one theory is submitted.” Id. at 61 (emphasis in original). As Weaver notes, in the context of the denial of a defendant’s right to select his or her own attorney, the Supreme Court did rely on the immeasurability of the effects of the error in concluding that the error was structural. United States v. Gonzalez-Lopez, 548 US 140, 150 , 126 S Ct 2557 , 165 L Ed 2d 409 (2006). But Gonzalez-Lopez appears to Cite as 367 Or 292 (2020) 305 be the only Supreme Court decision to deem an error structural based primarily on immeasurability, and the violation in that case did not involve the ordinary measurement difficulties attendant to any evidentiary or instructional error. Rather, as the Court explained, denial of a defendant’s counsel of choice could affect “investigation and discovery, development of the theory of defense, selection of the jury, presentation of the witnesses, and style of witness examination and jury argument,” as well as “whether and on what terms the defendant cooperates with the prosecution, plea bargains, or decides instead to go to trial.” Id. From the Supreme Court’s treatment of the issue in those cases, we conclude that the difficulty of measuring the effects of an error can support a determination that an error is structural, but it will generally weigh heavily only where the error “pervades the entire trial.” Id. With those broad contours of structural error in mind, we turn to the error at issue in this case. As we have explained, the reason that the Sixth Amendment forbids the jury instruction challenged here is because the jury instruction told the jury that it could do something that it constitutionally could not: return a guilty verdict without being unanimous. The state argues that, as a result, the instruction does not lead to fundamental unfairness in every case; it does so, at most, only when the jury returns a nonunanimous verdict. And, given that the jury can be (and here was) polled, the unfairness is easy to detect when it does occur. The state concludes that, because the error does not make every case fundamentally unfair and because the effect of the error is measurable, it is not structural; rather, it is instead subject to a harmlessness analysis. Defendant, on the other hand, offers several accounts of how the erroneous instruction leads to unfairness that is neither readily detectable nor limited to situations in which the jury returns a nonunanimous verdict. We examine each of those arguments in turn. B. Reasonable Doubt Defendant’s first and most straightforward argument is that telling a jury that it may convict a defendant 306 State v. Flores Ramos without being unanimous is tantamount to misinstructing the jury about the reasonable-doubt standard. Were that so, Sullivan, 508 US 275 , would require us to conclude that the error is structural. In this case, the jury was instructed that it could not convict defendant unless persuaded of his guilt beyond a reasonable doubt: “The defendant, Mr. Flores Ramos, is innocent unless and until Mr. Flores Ramos is proven guilty beyond a reasonable doubt. The burden is on the State and the State alone to prove the guilt of the defendant beyond a reasonable doubt. “Reasonable doubt is a doubt based on common sense and reason. Reasonable doubt is not an imaginary doubt. Reasonable doubt means an honest uncertainty as to the guilt of the defendant. “You must return a verdict of not guilty if, after careful and impartial consideration of all of the evidence in the case, you are not convinced to a moral certainty that the defendant is guilty.” Defendant did not object to those instructions, and he does not argue that they were wrong or that ordinarily they would be insufficient. Rather, he argues that “the nonunanimous jury instruction is structural error even in light of an otherwise adequate reasonable-doubt instruction.” He makes two arguments for that proposition. First, defendant contends that “[a] nonunanimousverdict instruction incorrectly informs the jury about the quantum of certitude necessary to find guilt beyond a reasonable doubt—83%.” We disagree with that characterization. An instruction that 10 votes out of 12 is sufficient to convict no more instructs the jury that “beyond a reasonable doubt” means “83% certainty” than a jury unanimity instruction implies that a juror must be 100% certain to convict. Defendant’s argument incorrectly conflates the percentage of votes required for a verdict with the degree of certainty that an individual juror must feel in order to conclude that the defendant is guilty beyond a reasonable doubt. Cite as 367 Or 292 (2020) 307 The subtler version of defendant’s argument is that the nonunanimous jury instruction improperly indicates to a juror that that juror may find guilt beyond a reasonable doubt despite the reasonable doubts of other jurors. As a result, defendant contends, jurors would misunderstand the nature of the “beyond a reasonable doubt” standard, which defendant understands to require acquittal if any reasonable juror could have a reasonable doubt. As defendant puts it, “[T]he court has effectively told the jury that the reasonable doubts of other jurors are irrelevant: the jury may find guilt beyond a reasonable doubt even if individual jurors do not agree.” The problem with defendant’s argument, as we understand it, is that it relies on a conception of reasonable doubt that the Supreme Court has expressly rejected. “What the factfinder must determine to return a verdict of guilty is prescribed by the Due Process Clause.” Sullivan, 508 US at 277 . Specifically, the Due Process Clause requires proof of guilt beyond a reasonable doubt. In re Winship, 397 US 358, 364 , 90 S Ct 1068 , 25 L Ed 2d 368 (1970). That is true regardless of whether the finder of fact is a judge or a jury. As the Court explained in Sullivan: “It would not satisfy the Sixth Amendment to have a jury determine that the defendant is probably guilty, and then leave it up to the judge to determine (as Winship requires) whether he is guilty beyond a reasonable doubt. In other words, the jury verdict required by the Sixth Amendment is a jury verdict of guilty beyond a reasonable doubt.” 508 US at 278 . As a result, both the Sixth Amendment and the Due Process Clause include an identical requirement of proof beyond a reasonable doubt. In Apodaca v. Oregon, 406 US 404 , 92 S Ct 1628 , 32 L Ed 2d 184 (1972), overruled by Ramos, 140 S Ct 1390 , the Supreme Court upheld a nonunanimous conviction against a challenge under the Sixth Amendment, as incorporated through the Due Process Clause of the Fourteenth Amendment. In a companion case, Johnson v. Louisiana, 406 US 356 , 92 S Ct 1620 , 32 L Ed 2d 152 (1972), the Court upheld nonunanimous convictions against an independent Due Process Clause challenge. The limited scope of Johnson had a somewhat technical reason behind it: the defendant 308 State v. Flores Ramos in Johnson had been convicted in a state court before the Court had issued its opinion in Duncan v. Louisiana, 391 US 145 , 88 S Ct 1444 , 20 L Ed 2d 491 (1968), which had incorporated the Sixth Amendment’s jury trial right to the states through the Fourteenth Amendment. Under DeStefano v. Woods, 392 US 631 , 88 S Ct 2093 , 20 L Ed 2d 1308 (1968), overruled by Griffith, 479 US 314 , the Sixth Amendment was not applicable to the cases tried before Duncan, even if those cases were still on direct appeal. Thus, Johnson did not address the Sixth Amendment, which did not apply to the defendant’s case; instead, it addressed only whether the Due Process Clause or the Equal Protection Clause of the Fourteenth Amendment required jury unanimity independent of the Sixth Amendment. One of the arguments advanced by the defendant in Johnson was that the nonunanimous verdict interfered with the Due Process Clause’s requirement of proof beyond a reasonable doubt. The defendant’s contention in Johnson is indistinguishable from the argument advanced by defendant in this case. The Supreme Court rejected that argument, explaining: “In our view disagreement of three jurors does not alone establish reasonable doubt, particularly when such a heavy majority of the jury, after having considered the dissenters’ views, remains convinced of guilt. That rational [jurors] disagree is not in itself equivalent to a failure of proof by the State, nor does it indicate infidelity to the reasonable-doubt standard.” Johnson, 406 US at 362 . That is, Johnson held that proof beyond a reasonable doubt does not require a conclusion that no reasonable juror could (or did) have a reasonable doubt. A juror who understands that he or she may believe a defendant guilty beyond a reasonable doubt even though other reasonable jurors may disagree properly understands the concept. As Johnson explained, that conclusion is consistent with numerous other applications of the reasonable-doubt standard: “Jury verdicts finding guilt beyond a reasonable doubt are regularly sustained even though the evidence was such that the jury would have been justified in having a reasonable Cite as 367 Or 292 (2020) 309 doubt; even though the trial judge might not have reached the same conclusion as the jury; and even though appellate judges are closely divided on the issue whether there was sufficient evidence to support a conviction. That want of jury unanimity is not to be equated with the existence of a reasonable doubt emerges even more clearly from the fact that when a jury in a federal court, which operates under the unanimity rule and is instructed to acquit a defendant if it has a reasonable doubt about his guilt, cannot agree unanimously upon a verdict, the defendant is not acquitted, but is merely given a new trial. If the doubt of a minority of jurors indicates the existence of a reasonable doubt, it would appear that a defendant should receive a directed verdict of acquittal rather than a retrial. We conclude, therefore, that verdicts rendered by nine out of 12 jurors are not automatically invalidated by the disagreement of the dissenting three. Appellant was not deprived of due process of law.” Johnson, 406 US at 362-63 (citations omitted). Of course, Ramos holds that even a substantial majority of jurors, properly satisfied beyond a reasonable doubt that the defendant is guilty, nevertheless cannot deliver a valid guilty verdict, but that is a distinct question from whether those jurors properly understand the concept of proof beyond a reasonable doubt. The Johnson reasonable-doubt holding remains good law after Ramos. Ramos addressed only whether the Sixth Amendment requires nonunanimous verdicts, a question that was not at issue in Johnson. And the holding in Ramos that the Sixth Amendment—because of the original meaning of the term “jury” that appears in its text—requires jury unanimity does not call into question the holding in Johnson that the concept of reasonable doubt does not, in and of itself, demand unanimity. In addition, the holding in Johnson on this point also has been relied on in at least one subsequent case, Tibbs v. Florida, 457 US 31 , 42 n 17, 102 S Ct 2211 , 72 L Ed 2d 652 (1982) (citing Johnson, 406 US at 362 , for the proposition that “[o]ur decisions also make clear that disagreements among jurors or judges do not themselves create a reasonable doubt of guilt”). We are bound by the holding in Johnson on the relationship between reasonable doubt and unanimity, and, therefore, we reject the 310 State v. Flores Ramos argument that defendant advances about their relationship. Moreover, even if we were not bound by it, the Court’s reasoning in Johnson on this point is persuasive. C. Effect of Nonunanimous Jury Instruction Next, defendant engages more directly with the state’s central contention: that no fundamental unfairness results when the jury returns a unanimous verdict, which can be discerned through polling. Defendant resists that conclusion in two related ways. First, he contends that the jury instruction that unanimity was not required prevented jurors from appreciating the significance of their individual decisions and that, as a result, jurors may have voted to convict without being convinced of defendant’s guilt. Second, he argues that the nonunanimous jury instruction affected the manner of deliberations. We consider each of those issues in turn. 1. Reduced responsibility Defendant takes the position that, because jurors were told that the votes of only 10 of them were sufficient for a conviction, if there were one or two remaining holdout jurors on any of the counts, those jurors may have felt a diminished sense of responsibility, knowing that their votes were not essential to the verdict. A sense of futility having been instilled, those jurors may have voted to convict defendant, perhaps simply to appease the majority or because of social pressure, even though they in fact retained reasonable doubts about his guilt. Had those jurors known what they should have been told—that even a single vote to acquit was enough to prevent a conviction—they might have refused to convict. Thus, defendant contends, his trial was fundamentally unfair, regardless of whether the jury reached a unanimous verdict. Even unanimous verdicts are tainted by the instruction. We disagree with defendant’s central contention, which is that the jury instruction permitting nonunanimous verdicts necessarily left holdout jurors with a diminished sense of responsibility for their votes, such that the trial was rendered fundamentally unfair. Even assuming that the erroneous instruction by itself may have a tendency to lower Cite as 367 Or 292 (2020) 311 the perceived stakes of the decision for some jurors, the jury received other instructions that made clear that, outvoted or no, jurors could not find the defendant guilty unless they were convinced of his guilt beyond a reasonable doubt. The trial court began its instructions to the jury with the following: “Members of the jury, it is your sole responsibility to make all of the decisions about the facts in this case. You must evaluate the evidence to determine how reliable or how believable that evidence is. When you make your decision about the facts, you must then apply the legal rules to those facts and reach your verdict. “Remember that your power to reach a verdict is not arbitrary. When I tell you what the law is on a particular subject or tell you how to evaluate certain evidence, you must follow these instructions.” Jurors were also instructed that “[i]t is your duty to weigh the evidence calmly and dispassionately and to decide this case on its merits. Do not allow bias, sympathy, or prejudice any place in your deliberations. Do not decide this case based on guesswork, conjecture, or speculation. Do not consider what sentence might be imposed by the Court if the defendant is found guilty.” And, after being instructed on reasonable doubt, jurors were told: “You must return a verdict of not guilty if, after careful and impartial consideration of all of the evidence in the case, you are not convinced to a moral certainty that the defendant is guilty.” The state argues that those instructions were sufficient to inform jurors that they could cast a guilty vote only if they concluded, based on the evidence, that defendant was guilty beyond a reasonable doubt, and so also would have told jurors that they could not vote to convict simply because they found themselves outnumbered. Defendant counters that all those instructions are ambiguous: “[T]he second-person use of the word ‘you’ in the reasonable doubt instruction reasonably refers to ‘you,’ the entire jury.” (Emphasis in original.) He asserts that jurors 312 State v. Flores Ramos were not told that they could not individually vote to convict unless they were convinced beyond a reasonable doubt of defendant’s guilt. Even assuming a certain level of ambiguity in those instructions, however, those instructions were given alongside an instruction that clearly emphasized jurors’ obligation to make individual decisions. Jurors were instructed to “keep in mind that each party is entitled to the considered decision of each juror.” And, before any of those other instructions, each juror swore or affirmed, as required by ORCP 57 E, “that they and each of them will well and truly try the matter in issue between the plaintiff and defendant, and a true verdict give according to the law and evidence as given them on the trial.” 4 (Emphasis added.) Taking the instructions as a whole, we do not think that jurors would have been left with any doubt that they were required to make an individual decision based on the evidence. For example, we see no basis for thinking that a juror, upon being told, “Do not decide this case based on guesswork, conjecture, or speculation,” would assume that the instruction applied only to the jury as a whole, but not to its members individually, and so would feel free to make his or her own decision based on a guess. Similarly, reasonable jurors, having been reminded that “each party is entitled to the considered decision of each juror,” would not interpret the instruction that “[y]ou must return a verdict of not guilty if * * * you are not convinced to a moral certainty that the defendant is guilty” to permit individual votes to be cast on some other standard. We therefore perceive no realistic possibility that jurors would understand their oath and the instructions as permitting them to cast a vote to convict defendant while still retaining a reasonable doubt about his guilt. With that conclusion in mind, defendant’s argument can succeed only if we assume that jurors may have disobeyed those other instructions. That is, defendant’s argument that the nonunanimous jury instruction leads to fundamental — 4 — The transcript reflects that the oath was administered, but it does not record the precise wording. Defendant makes no argument that the oath in this case was in any way defective. Cite as 367 Or 292 (2020) 313 unfairness, by creating the risk that a juror who would be inclined to acquit will “give up” too easily, requires us to assume that that juror will disregard the other instructions addressed to his or her individual responsibility. In evaluating whether an error requires reversal, the Supreme Court has repeatedly emphasized the presumption that “ ‘jurors, conscious of the gravity of their task, attend closely the particular language of the trial court’s instructions in a criminal case and strive to understand, make sense of, and follow the instructions given them.’ ” United States v. Olano, 507 US 725, 740 , 113 S Ct 1770 , 123 L Ed 2d 508 (1993) (quoting Francis v. Franklin, 471 US 307 , 324 n 9, 105 S Ct 1965 , 85 L Ed 2d 344 (1985)). In evaluating whether the instructional error that did occur here is such as to require reversal in every case, we must, like the Supreme Court, give great weight to “the almost invariable assumption of the law that jurors follow their instructions.” Richardson v. Marsh, 481 US 200, 206 , 107 S Ct 1702 , 95 L Ed 2d 176 (1987). In only a few circumstances has the Supreme Court found an exception to that rule. In Bruton, for example, as noted above, the Supreme Court held that it was unrealistic to expect jurors to obey an instruction to ignore a confession by a codefendant directly implicating the defendant when considering the question of the defendant’s guilt. But this is not a case of that type. Here, jurors not convinced of guilt beyond a reasonable doubt were simply required to report a vote of “not guilty,” even if they were outvoted. Jurors were not asked to perform the equivalent of “the mental gymnastics of considering an incriminating statement against only one of two defendants in a joint trial.” Frazier v. Cupp, 394 US 731, 735 , 89 S Ct 1420 , 22 L Ed 2d 684 (1969). There was no contradiction in the instructions, nor is there any reason to think that holdout jurors would face pressure to change their votes after the jury had already reached a verdict. Simply put, all 12 jurors, when polled, individually stated that they had found the defendant guilty beyond a reasonable doubt of the counts in dispute. We would have to speculate not to take them at their word. 314 State v. Flores Ramos Our rejection of defendant’s argument is consistent with, and likely compelled by, Supreme Court precedent. In Romano v. Oklahoma, 512 US 1 , 114 S Ct 2004 , 129 L Ed 2d 1 (1994), the defendant was sentenced to death by a jury that had been told that the defendant already had received a death sentence for a separate crime. The defendant argued that the imposition of a death sentence by a jury that had received that information violated the Due Process Clause because knowledge of the extant sentence would have diminished jurors’ sense of responsibility for their own sentencing decision. Id. at 6 . The Supreme Court rejected that argument for two reasons. First, the Court held that, “if the jurors followed the trial court’s instructions, which we presume they did, this evidence should have had little—if any—effect on their deliberations.” 512 US at 13 (citation omitted). Second, the Court explained: “Even assuming that the jury disregarded the trial court’s instructions and allowed the evidence of petitioner’s prior death sentence to influence its decision, it is impossible to know how this evidence might have affected the jury. It seems equally plausible that the evidence could have made the jurors more inclined to impose a death sentence, or it could have made them less inclined to do so. Either conclusion necessarily rests upon one’s intuition. To hold on the basis of this record that the admission of evidence relating to petitioner’s sentence in the [earlier murder prosecution] rendered petitioner’s sentencing proceeding for the [second murder case] fundamentally unfair would thus be an exercise in speculation, rather than reasoned judgment.” Id. at 13-14 . To conclude that the erroneous instruction in this case will always cause a diminished sense of responsibility in individual jurors, so as to render all trials fundamentally unfair, would strain against the Court’s Due Process Clause analysis in Romano. 2. Effect on deliberation That does not dispose of defendant’s alternative structural error contention, which is that the instruction that jurors could convict without being unanimous affected the process of jury deliberation. Defendant argues that, as a result of the instructional error, jurors in effect were not told “to engage in a unanimous-consensus deliberative model.” Cite as 367 Or 292 (2020) 315 As defendant puts it, permitting jurors to return nonunanimous verdicts “shifts the burden of persuading other jurors from majority jurors to minority jurors.” That is an unconvincing account of how an instruction that unanimity was required for guilty verdicts could have made a difference for counts where the jury did vote unanimously to convict. Whatever the number of votes necessary to reach a verdict, jurors in the minority will seek to persuade jurors in the majority, and jurors in the majority will seek to persuade jurors in the minority. The potential difference, when the verdict need not be unanimous, is that jurors in the majority need not persuade everybody to reach a verdict. But, while that might provide an account of why the instruction is unfair when the jury returns a nonunanimous guilty verdict, it fails to explain why the instruction makes the trial fundamentally unfair even if the jury returned a unanimous verdict. When the verdict is unanimous, either the jurors in the majority did successfully persuade any holdouts in favor of acquittal—that is, what defendant contends should have happened in fact did happen—or all jurors were persuaded of the defendant’s guilt before any discussion occurred. We see no fundamental unfairness so as to universally require reversal of unanimous verdicts. Amicus curiae the Criminal Justice Reform Clinic at Lewis & Clark Law School presents a more concrete argument about how the nonunanimous verdict instruction may have affected jury deliberations. Relying on social science research, the Clinic argues that juries told that they can reach a nonunanimous verdict are more likely to follow a deliberative process that is “verdict-driven rather than evidence-driven,” meaning that the jury votes sooner and more often, reaches a verdict more quickly, and spends comparatively less time discussing evidence. Below, we examine in more depth the question of how those asserted differences affect the harmless error analysis. At this point, though, the question is whether the error is structural, and we conclude that—even on the assumption that such differences do exist—the Clinic’s brief does not provide an account of why the erroneous jury instruction given in this case “cast[s] so much doubt on the 316 State v. Flores Ramos fairness of the trial process that, as a matter of law, [the error] can never be considered harmless.” Satterwhite v. Texas, 486 US 249, 256 , 108 S Ct 1792 , 100 L Ed 2d 284 (1988). As an initial matter, we note that the Supreme Court’s decision in Ramos did not require that jurors follow any particular deliberative model, nor is that an obvious implication of its holding. Neither defendant nor the Clinic develops an argument that the Sixth Amendment—or any other provision of the United States Constitution—requires that jurors deliberate in a particular manner. To the contrary, most courts to consider the question have rejected the argument that the Sixth Amendment requires a particular quality or quantity of deliberation before a conviction can be held valid. As one court put it, “It seems self-explanatory that ‘[n]o rule requires a jury to deliberate for any set length of time.’ ” United States v. Dolan, 120 F3d 856, 870 (8th Cir 1997) (quoting United States v. Penagaricano-Soler, 911 F2d 833, 846 n 15 (1st Cir 1990)). Challenges to verdicts based on the length of jury deliberations have been consistently rejected, even when the jury was out “only five to seven minutes” before returning a verdict. United States v. Brotherton, 427 F2d 1286, 1289 (8th Cir 1970); see also Wall v. United States, 384 F2d 758, 762 (10th Cir 1967) (upholding verdict where jury deliberated for one hour following an eight-day trial); Kimes v. United States, 242 F2d 99, 101 (5th Cir), cert den, 354 US 912 , 77 S Ct 1299 , 1 L Ed 2d 1429 (1957) (“we find nothing suspicious, questionable, or remarkable in the action of the jury in returning its verdict of guilty after deliberating only twenty minutes”); United States v. Anderson, 561 F2d 1301, 1303 (9th Cir 1977) (upholding verdict returned after “brief deliberation”); United States v. Burfoot, 899 F3d 326, 342 (4th Cir 2018) (upholding convictions where the jury deliberated for five hours after a fiveweek trial). Similarly, there are no cases purporting to regulate the frequency with which juries should vote on their way to reaching a verdict or the extent to which the evidence must be discussed. Finally, the Clinic does not argue that there is a oneto-one correlation between “verdict-driven” deliberations and instructions that jurors do not need to be unanimous Cite as 367 Or 292 (2020) 317 to convict. The Clinic’s claim, as we understand it, is that a jury instruction permitting nonunanimous verdicts makes “verdict-driven” deliberations more likely, not that “verdictdriven” deliberations occur only when juries are misinstructed on unanimity. At bottom, then, we are left with an argument that the instruction may have made the jury less likely to employ one constitutionally permissible style of deliberation and more likely to use a different, also constitutionally permissible, style of deliberation. That falls far short of the type of error after which “no criminal punishment may be regarded as fundamentally fair,” Clark, 478 US at 577-78 , so as to amount to structural error. D. Erosion of Public Confidence Finally, defendant argues that a nonunanimous jury instruction constitutes structural error because it “erodes public confidence in the jury-trial right.” However, the Supreme Court has emphasized that “the harmless-error doctrine is essential to preserve the ‘principle that the central purpose of a criminal trial is to decide the factual question of the defendant’s guilt or innocence, and promotes public respect for the criminal process by focusing on the underlying fairness of the trial rather than on the virtually inevitable presence of immaterial error.’ ” Fulminante, 499 US at 308 (quoting Delaware v. Van Arsdall, 475 US 673, 681 , 106 S Ct 143 , 189 L Ed 2d 674 (1986)). Thus, although constitutional error may tend to undermine public confidence, unwarranted reversals of criminal convictions also undermine the reliability of the adjudicative process, along with the public perception of it. Defendant draws an analogy to two other cases in which the Supreme Court has found structural error: Batson v. Kentucky, 476 US 79 , 106 S Ct 1712 , 90 L Ed 2d 69 (1986), and Vasquez v. Hillery, 474 US 254 , 106 S Ct 617 , 88 L Ed 2d 598 (1986). In Batson, the Supreme Court held that it violated the Equal Protection Clause for a prosecutor to exercise a peremptory strike against a juror on the basis of race. In Hillery, the Court reaffirmed that the Equal Protection Clause forbids the indictment of a defendant by a grand jury from which members of the defendant’s race have been excluded. In both cases, the error was held structural. 318 State v. Flores Ramos In Hillery, the Court explained that, “[w]hen constitutional error calls into question the objectivity of those charged with bringing a defendant to judgment, a reviewing court can neither indulge a presumption of regularity nor evaluate the resulting harm.” 474 US at 263 . In Batson, the Court held that race-based exclusion of jurors “undermine[s] public confidence in the fairness of our system of justice.” 476 US at 87 . Defendant argues that the same is true here. He points out that, in Ramos, the Supreme Court observed that the initial adoption of nonunanimous juries in Oregon had been motivated by racism: “Adopted in the 1930s, Oregon’s rule permitting nonunanimous verdicts can be similarly traced to the rise of the Ku Klux Klan and efforts to dilute ‘the influence of racial, ethnic, and religious minorities on Oregon juries.’ ” Ramos, 590 US at ___, 140 S Ct at 1394 (quoting State v. Williams, No. 15-CR-58698). Defendant contends that, because Oregon’s nonunanimous jury system was adopted in part for racist reasons, it should be held to undermine confidence in the criminal justice system just as in Batson. Several amici join defendant on this point, arguing that reversal of defendant’s convictions, and all others obtained under a system that permitted nonunanimous convictions, is necessary to restore the legitimacy of the criminal justice system. But there is little analogy between the constitutional violations that occurred in Batson and Hillery and the violation in this case. In this case, no juror was excluded on the basis of race. All jurors, regardless of race, unanimously found defendant guilty of the four counts in dispute. If the jury were permitted to convict a defendant without being unanimous, there undoubtedly would be some cases where the jury’s vote breaks down along racial or ethnic lines. But that does not explain why public confidence in unanimous verdicts—where that potential verifiably was not realized— should be undermined. Defendant’s analogy to decisions under the Equal Protection Clause fails for another reason as well. The Sixth Cite as 367 Or 292 (2020) 319 Amendment violation that occurred here—instructing the jury that it did not need to be unanimous to convict—does not depend on why Oregon first began using nonunanimous juries. The right to a unanimous verdict derives from the text and history of the Sixth Amendment and, as the Supreme Court explained in Ramos, “a jurisdiction adopting a nonunanimous jury rule even for benign reasons would still violate the Sixth Amendment.” 590 US at ___ n 44, 140 S Ct at 1401 n 44. We cannot conclude that the error is structural—that it always requires reversal, regardless of the circumstances under which it is given and the effect that it is likely to have—based on a historical circumstance that has no inherent link to the constitutional violation at issue. See Neder, 527 US at 14 (“Under our cases, a constitutional error is either structural or it is not.”). In all, defendant’s emphasis on the importance of unanimity to public confidence in the jury’s verdict only cements our view that the instructional error that occurred here was not the type of constitutional violation after which “a criminal trial cannot reliably serve its function as a vehicle for determination of guilt or innocence, and no criminal punishment may be regarded as fundamentally fair.” Clark, 478 US at 577-78 (citation omitted). The jury was not told that it needed to be unanimous, but—as to the four counts in dispute here—it did reach unanimous verdicts. Those verdicts represent the consensus of “a jury selected from a representative cross section of the entire community.” Ramos, 590 US at ___ n 47, 140 S Ct at 1402 n 47. Defendant’s trial before that “impartial adjudicator,” combined with his representation by counsel, gives rise to “a strong presumption that any other errors that may have occurred are subject to harmless-error analysis.” Clark, 478 US at 579 . For the reasons we have articulated above, defendant has not overcome that presumption.5 — 5 — Our conclusion on this point is consistent with the Supreme Court’s decision in Burch v. Louisiana, 441 US 130 , 132 n 4, 99 S Ct 1623 , 60 L Ed 2d 96 (1979), which affirmed the conviction of a defendant convicted unanimously by a six-person jury instructed that it could convict by a vote of five to one. The Court’s reasoning in Burch is somewhat obscure, and it is not clear whether the questions that we consider here were squarely presented in that case, so we have made our decision in this case without relying on Burch. 320 State v. Flores Ramos IV. HARMLESS ERROR When a federal constitutional error is not structural, the conviction can be affirmed only if the error “was harmless beyond a reasonable doubt.” Fulminante, 499 US at 307-08 . An error is harmless beyond a reasonable doubt if the reviewing court is satisfied “beyond a reasonable doubt that the error complained of did not contribute to the verdict obtained.” Chapman, 386 US at 24 . The state argues that a “nonunanimous verdict” instructional error is harmless whenever the jury, in fact, reached a unanimous verdict. Thus, the state argues, all that is needed to establish harmlessness is the jury poll showing unanimity. Defendant pushes back against that argument in several ways, contending that this court cannot find the error harmless as to any count in this case. We consider each of defendant’s arguments and, for the reasons that follow, reject them. A. Neder First, defendant argues that the state’s position is inconsistent with the Supreme Court’s decision in Neder. Relying on Neder, defendant argues that, whatever the poll shows, his convictions can be affirmed only if the record contains “uncontested and overwhelming evidence of guilt on every element.” (Emphasis in original.) In Neder, the trial court erred by failing to submit one of the elements of the offense to the jury at all, instead making its own decision that the state had satisfied that element beyond a reasonable doubt. The Court held that the conviction could nevertheless be affirmed if the error was harmless beyond a reasonable doubt, summarizing the applicable inquiry in the following manner: “Is it clear beyond a reasonable doubt that a rational jury would have found the defendant guilty absent the error?” Neder, 527 US at 18 . Applying that standard in Neder, the Court framed the question as being “whether the record contains evidence that could rationally lead to a contrary finding with respect to the omitted element.” Id. at 19 . The Court concluded, in that case, that there was no such evidence. Id. at 19-20 . Defendant argues that this court must conduct the same inquiry here, with respect to each element of every Cite as 367 Or 292 (2020) 321 charged offense. But that argument ignores the difference between the error in Neder and the error in this case. Under the Chapman standard, the overall question is whether the court can “conclude beyond a reasonable doubt that the jury verdict would have been the same absent the error.” Neder, 527 US at 19 . In Neder, the error lay in failing to submit an element to the jury at all, with the result that the jury never had an opportunity to decide it. That error could be held harmless only if the Court could be confident that the jury would have convicted the defendant even if it had considered the additional element. And the Court could have that confidence only if the “omitted element is supported by uncontroverted evidence.” Id. at 18 . In this case, by contrast, every element of each of the four disputed counts was submitted to the jury, and the poll shows that the jury unanimously had concluded that the state had proved every element beyond a reasonable doubt. The question is not whether a reasonable jury necessarily would reach the same conclusion; unlike the Court in Neder, we know that this jury in fact did so. The question in this case is whether we can be confident, beyond a reasonable doubt, that the jury would have reached the same conclusion had it been properly instructed as to unanimity—a conclusion that does not require overwhelming evidence on every element, if we can be satisfied that the poll was accurate and that the instruction did not have a significant impact on jury deliberations.6 B. Zolotoff Next, defendant argues that the state’s position is inconsistent with this court’s decision in State v. Zolotoff, 354 Or 711 , 320 P3d 561 (2014). Defendant argues that, under Zolotoff, an error in an instruction that otherwise would have provided “the jury with a legal distinction to apply during its deliberations” can be found harmless only — 6 — For the same reason, this case does not involve “first-guessing” a jury’s decision, which defendant argues is not permitted by the Oregon Constitution. The jury reached unanimous decisions on the disputed counts. The question is whether those decisions, which the jury did make, must be reversed because of the instructional error. We therefore reject defendant’s state constitutional argument. 322 State v. Flores Ramos if another instruction conveyed the same legal distinction to the jury. Defendant argues that “a proper unanimity instruction would have given the jury a significant legal distinction to consider when deliberating and assessing guilt,” although, perhaps because he reads Zolotoff as establishing a per se rule, he does not articulate a specific theory of how the instruction could have affected the jury’s deliberations. We disagree with defendant’s reading of Zolotoff. In Zolotoff, the defendant was convicted of possession of a weapon by an inmate. 354 Or at 713 . He had requested, but been denied, an instruction on the lesserincluded offense of attempted possession of a weapon by an inmate. Id. The state conceded that the failure to so instruct the jury had been error, but argued that the error was necessarily harmless because, even if the jury had been instructed on the lesser-included offense, it would also have been instructed, pursuant to ORS 136.460(2), that it could consider the lesser-included offense only after reaching a not guilty verdict on the greater-inclusive offense. 354 Or at 715-16 . Thus, the state’s reasoning ran, the jury never would have had cause to consider the attempt charge, even had it been so instructed, so the error could not have affected the verdict. We rejected that categorical argument. We first recognized that, as the state had argued, “there may be many instances in which an appellate court will be able to conclude from the evidence, the arguments, and the instructions that the jury would have reached the same verdict on the charged offense even if it also had received instruction on the lesser-included offense.” Zolotoff, 354 Or at 718-19 . But we held that “an error in failing to instruct on a lesserincluded offense will not always be harmless” because “[t]here may be circumstances in which the elements of the charged crime are clearer when they are viewed in contrast with the elements of a lesser-included offense.” Id. at 719 . Applying that reasoning to the case at hand, we explained, “That erroneously omitted instruction would have told the jurors that there was a legal distinction between taking a substantial step toward making the spoon into a weapon and completing the task. In other words, the definition of Cite as 367 Or 292 (2020) 323 the term ‘weapon’ told the jury what a weapon is, but it did not tell the jury that the spoon was not a weapon if it was an object that defendant was still in the process of making into a weapon. In this case, an instruction on the elements of the lesser-included offense of attempted possession of a weapon by an inmate would have been particularly helpful because, as the state concedes, there was evidence from which the jury could have found that the spoon was not a weapon and therefore that defendant did not actually possess a weapon; he only attempted to make the spoon into a weapon and possess it.” Id. at 720. Zolotoff did not, therefore, embrace a categorical rule that the omission of any instruction that might help the jury understand a legal distinction cannot be harmless. Rather, Zolotoff rejected the categorical rule proposed by the state in favor of a different approach, recognizing that an instruction on a lesser-included offense may help the jury better understand the elements of the greater-inclusive offense, and the absence of such an instruction may therefore have affected the verdict. But, as Zolotoff acknowledged, both of those conclusions turn on what the instruction would have helped the jury understand and the importance of the distinction to the case at hand. Here, even leaving aside the fact that Zolotoff did not involve an application of the federal harmlessness standard, Zolotoff is not especially pertinent. Defendant faults the instruction here for failing to inform the jury about the importance of unanimity and that that jury could return a guilty verdict only if it were unanimous. Obviously, the failure to impress upon the jurors that guilty verdicts needed to be unanimous was significant as to the single nonunanimous guilty verdict returned by the jury. But, insofar as the jury did return unanimous guilty verdicts on the other counts, defendant does not persuasively explain how instructing the jury on the necessity of a unanimous verdict would have affected the unanimous verdicts that they did return. As discussed above, jurors were given ample instruction on their duty with respect to their individual determinations of the defendant’s guilt, and they are presumed to have followed those instructions. “Judicious application of 324 State v. Flores Ramos the harmless-error rule does not require that we indulge assumptions of irrational jury behavior when a perfectly rational explanation for the jury’s verdict, completely consistent with the judge’s instructions, stares us in the face.” Schneble v. Florida, 405 US 427, 431-32 , 92 S Ct 1056 , 31 L Ed 2d 340 (1972). C. The Jury Poll Defendant also challenges the sufficiency of the poll of the jury. Insofar as defendant’s argument is that the poll, in and of itself, does not establish that the instruction had no effect on the jury’s deliberations, we agree. But, to the extent that defendant contends that the poll was insufficient to establish whether the jury in fact was unanimous, we disagree. Defendant suggests several ways in which a poll may fail to capture how jurors, in fact, voted: the jury may not have understood the use of words like “unanimous,” jurors had no legally significant reason to “record a unanimous verdict,” and jurors may simply raise their hands when put on the spot by a poll. But most of those concerns do not apply to this case. Here, the trial court, count-by-count, asked all jurors who voted “guilty” to raise their hands. No juror could have misunderstood that simple instruction; the poll itself gave jurors a reason—and a duty—to record their votes; and defendant suggests no basis for thinking that any juror would have given a false answer. More broadly, we are skeptical that jurors would not understand the word “unanimous” or that jurors, however polled, would not respond honestly. See United States v. Poole, 545 F3d 916, 921 (10th Cir 2008) (rejecting an argument that jurors would not have understood the trial court’s use of the word “nullity”). D. Effect on Deliberations With those arguments addressed, we turn to the argument made by the Clinic as amicus: that the instruction permitting nonunanimous guilty verdicts may have affected deliberations, and so cannot be held to be harmless. Although we considered the Clinic’s arguments above in the context of whether the instructional error was structural, there we dealt only with the question whether any potential Cite as 367 Or 292 (2020) 325 difference in deliberation resulting from the instruction would make the trial fundamentally unfair. In the harmless error context, the question before us is whether any difference in the style of deliberation could have made a difference to the result in this case. On that question, the fact that no particular deliberative style is constitutionally required is not dispositive. We are not able to approach this question entirely as a matter of first impression. In Johnson, when considering the argument that a nonunanimous verdict violated the Due Process Clause, the Supreme Court considered and rejected the defendant’s contention that a lack of unanimity indicated that the jurors voting to convict could not have conscientiously voted to convict. The Court gave the following reasons for its rejection of the argument: “Appellant, in effect, asks us to assume that, when minority jurors express sincere doubts about guilt, their fellow jurors will nevertheless ignore them and vote to convict even if deliberation has not been exhausted and minority jurors have grounds for acquittal which, if pursued, might persuade members of the majority to acquit. But the mere fact that three jurors voted to acquit does not in itself demonstrate that, had the nine jurors of the majority attended further to reason and the evidence, all or one of them would have developed a reasonable doubt about guilt. We have no grounds for believing that majority jurors, aware of their responsibility and power over the liberty of the defendant, would simply refuse to listen to arguments presented to them in favor of acquittal, terminate discussion, and render a verdict. On the contrary it is far more likely that a juror presenting reasoned argument in favor of acquittal would either have his arguments answered or would carry enough other jurors with him to prevent conviction. A majority will cease discussion and outvote a minority only after reasoned discussion has ceased to have persuasive effect or to serve any other purpose—when a minority, that is, continues to insist upon acquittal without having persuasive reasons in support of its position. At that juncture there is no basis for denigrating the vote of so large a majority of the jury or for refusing to accept their decision as being, at least in their minds, beyond a reasonable doubt. * * * Appellant offers no evidence that majority jurors simply ignore the reasonable doubts of their colleagues or otherwise act irresponsibly 326 State v. Flores Ramos in casting their votes in favor of conviction, and before we alter our own longstanding perceptions about jury behavior and overturn a considered legislative judgment that unanimity is not essential to reasoned jury verdicts, we must have some basis for doing so other than unsupported assumptions.” Johnson, 406 US at 361-62 (emphasis added). In Johnson, the Supreme Court appeared to assume that, even when the jury was in fact not unanimous, it still would have deliberated with the same care and to the same extent as if unanimity were required. A fortiori, the same presumption would appear to extend to juries that, while instructed that they could return a nonunanimous guilty verdict, nevertheless did reach unanimity. As noted above, Ramos did not address the Due Process Clause arguments considered in Johnson, and it did not overrule the majority opinion in Johnson. The reasoning of Ramos, based on text and history, does not call into question the reasoning of Johnson. And, though Johnson concerned whether a nonunanimous verdict violated the Due Process Clause, and the question here concerns the harmlessness of an error that did occur, the factual assumption in Johnson is relevant to both. Johnson therefore still binds us. Nevertheless, we read Johnson to establish only a rebuttable presumption; Johnson faulted the defendant for failing to rebut it, but it did not hold that nothing could. In this case, the Clinic argues that social science research, post-dating Johnson, demonstrates that instructions that jurors need not be unanimous do affect deliberations. Principally, the clinic relies on a study documented in Reid Hastie et al, Inside the Jury (1983).7 In the Hastie study, 69 mock juries, drawn from actual jury pools, were asked to render a verdict after watching a taped reenactment of a real trial. Id. at 45-55, 60. A third of the mock juries needed to be unanimous to reach any verdict, another third could reach any verdict by a 10-to-two vote, and the final third could reach any verdict by an eight-to-four vote. Id. at 60. In analyzing the results, the researchers looked at — 7 — The brief cites multiple other sources; however, many of those sources refer back to the Hastie study on the pertinent point. Cite as 367 Or 292 (2020) 327 when the jury first took an internal vote. Juries that polled themselves within 10 minutes were labeled “verdict-driven.” Id. at 164. When the first ballot took place after at least 40 minutes of deliberation, the jury was labeled “evidencedriven.” Id. “Evidence-driven” juries ended up deliberating for longer than “verdict-driven” juries and the deliberations involved more connections between facts and legal issues. Id. Based on the Hastie study, the Clinic argues that an instruction that jurors do not need to be unanimous “leads to the likelihood that deliberations are verdict-driven rather than evidence-driven,” thus producing less reliable (and, as pertinent here, different) results. Even assuming the validity of the Hastie study, and that it would be appropriate to accord dispositive weight to a single study, there are three reasons why it does not lend much support to a conclusion that deliberations in this case were affected by the erroneous jury instruction. First, the study found only a weak correlation between unanimity requirements and whether a jury was “evidence-driven,” and it is not clear whether the result was statistically significant. See id. at 173 (“majority rule juries are slightly likelier to adopt a verdict-driven deliberation style in contrast to the evidence-driven style”). Second, the Clinic’s theory of how the erroneous jury instruction was not harmless is not—and cannot be— just that the instruction may have affected deliberations. Rather, it is that the potentially altered deliberations could in turn have affected the jury’s verdicts. But the Hastie study found “no relationship between [deliberation] style and final verdict.” Id. at 165. Third, the Hastie study did not examine juries, like the jury in this case, that returned a unanimous verdict despite being instructed that unanimity was not required (it is not clear that any of the mock juries reached such a result), and so it sheds little light on how those juries deliberated or whether their unanimous verdicts differed in any way from those rendered by juries that were instructed that unanimity was required. To be sure, some research shows—contrary to the Supreme Court’s presumption in Johnson—that juries that 328 State v. Flores Ramos return a nonunanimous guilty verdict may not have given full consideration to the views of the outvoted jurors. See Brief of Law Professors and Social Scientists as Amici Curiae in Support of Petitioner at 6-9, Ramos v. Louisiana, 590 US ___ , 140 S Ct 1390 . In those cases, the nonunanimity instruction may well exert an influence on both deliberations and the verdict, and for that reason the research cited by the Clinic supports our decision to reverse the one nonunanimous verdict in defendant’s case. But the same does not appear to hold when jurors do, despite not being obligated to, reach a unanimous verdict. The fact that the verdict is unanimous provides some assurance, in and of itself, that no juror was ignored and that all jurors’ reasonable doubts as to those counts were resolved. Neither the social science research that has been offered, nor common sense, calls that conclusion into question, much less overcomes the presumption articulated in Johnson. We therefore conclude that, though slight differences in deliberative process may have occurred had the jury been properly instructed, those potential differences do not prevent us from concluding that the result was not affected and that the error was harmless beyond a reasonable doubt. E. Mixed Verdicts Defendant also advances a narrower argument— that, even if the jury’s unanimous verdicts were not directly affected by the erroneous jury instruction, those verdicts could still have been indirectly affected. Defendant argues that “it is certain that the instructional error affected deliberations because the jury was not 12-0 on every count.” (Emphasis in original.) That is, had the jury been properly instructed, it would have continued deliberating past the point at which it returned its verdict on the attempted firstdegree rape charge, because two jurors still favored acquittal on that charge. The ultimate question in this case, however, is not whether further deliberation on the attempted rape count could have led to a different result as to that count, but whether we can conclude, beyond a reasonable doubt, that the jury’s decisions on the other counts were unaffected. We know from the jury poll that, as to the other four counts, the jury—including Cite as 367 Or 292 (2020) 329 the two jurors who would have acquitted defendant on the attempted first-degree rape count—unanimously agreed that defendant was guilty beyond a reasonable doubt. That fact lends strong support to a conclusion that the instructional error was harmless as to the unanimous verdicts. As the Supreme Court has explained, the harmless error analysis proceeds on the assumption “that the jury considered all the evidence bearing on the issue in question before it made the findings on which the verdict rested,” except in cases where the instructions precluded the jury from doing so. Yates v. Evatt, 500 US 391, 405-06 , 111 S Ct 1884 , 114 L Ed 2d 432 (1991), overruled in part on other grounds by McGuire, 502 US 62 . Here, the jury was properly instructed on the elements of each count, the beyond-areasonable-doubt standard, its “duty to weigh the evidence calmly and dispassionately,” and its obligation to “return a verdict of not guilty if, after careful and impartial consideration of all the evidence in the case, you are not convinced to a moral certainty that the defendant is guilty.” Giving appropriate weight to the “almost invariable assumption of the law that jurors follow their instructions,” Marsh, 481 US at 206 , the fact that the jury returned unanimous verdicts on four counts tells us that each juror, after considering all of the evidence, was convinced beyond a reasonable doubt of the defendant’s guilt on those counts. We see no nonspeculative basis for supposing that further deliberation on those counts, based on the same evidence and among jurors who already had unanimously agreed that defendant was guilty, would have led jurors to change their minds. And defendant’s argument requires even more—a supposition that further deliberation on a different count would have shaken jurors’ confidence in the unanimous verdicts that they had already reached. The abstract possibility that a juror could have changed his or her mind after further deliberation is insufficient to prevent us from concluding that the instructional error was harmless beyond a reasonable doubt. The Supreme Court addressed a similar argument in Harrington, 395 US 250 . In that case, the Court considered whether a violation of the Bruton rule—the introduction of two codefendant 330 State v. Flores Ramos confessions implicating the defendant at a joint trial—was harmless. Id. at 252 . The defendant had argued that the Court “must reverse if [the Court] can imagine a single juror whose mind might have been made up because of [the codefendants’] confessions and who otherwise would have remained in doubt and unconvinced.” Id. at 254 . But the Court rejected that interpretation of the Chapman standard: “We of course do not know the jurors who sat. Our judgment must be based on our own reading of the record and on what seems to us to have been the probable impact of the two confessions on the minds of an average jury.” Id. Thus, even if we can imagine a juror changing his or her mind because of further deliberations on a different charge, that merely conceivable possibility, though significant in the double jeopardy context, does not preclude us from finding that the error is harmless beyond a reasonable doubt. “To set a barrier so high that it could never be surmounted would justify the very criticism that spawned the harmlesserror doctrine in the first place[.]” Neder, 527 US at 18 . Defendant cites no authority for the proposition that a possibility so slim of a different result precludes a finding that an error was harmless beyond a reasonable doubt.8 The absence of authority on that point is notable, — 8 — The only decision that could be read to lend support to defendant’s position is Blueford v. Arkansas, 566 US 599, 607-08 , 132 S Ct 2044 , 182 L Ed 2d 937 (2012), a case that did not involve a harmless error question. In Blueford, the defendant was charged with capital murder and several lesser-included homicide offenses. Id. at 602 . The jury was instructed that it could consider each lesserincluded offense only after concluding that the defendant was not guilty of all greater-inclusive offenses. Id. After several hours of deliberation, and a reported deadlock, the foreman reported that the jury had unanimously voted to acquit the defendant of capital murder and first-degree murder but was deadlocked on manslaughter. Id. at 603-04 . The trial court had the jury deliberate for another half hour and ultimately declared a mistrial, discharging the jury without any further polling or verdict. Id. at 604 . The defendant argued to the Supreme Court that the Double Jeopardy Clause barred his retrial for capital murder and first-degree murder, because he had been acquitted on those charges in the first trial. The Court disagreed, holding that the poll conducted by the trial court lacked the finality necessary to constitute a verdict of acquittal. Id. at 606 . The Court explained that the jury “was free to reconsider a greater offense, even after considering a lesser one” and that one or more jurors could have reconsidered their views on the greater-inclusive offenses after further deliberation about the manslaughter charge. Id. at 607 . But the question before the Court in Blueford was meaningfully different from the question before us in this case. In Blueford, the question of finality for double jeopardy purposes turned only on whether the jury could have reconsidered its Cite as 367 Or 292 (2020) 331 because the possibility of extended deliberations on a different offense presents a situation no different from any time that an error affects one count in a multicount case. For example, if evidence relevant to only one count is erroneously admitted against the defendant, it will be possible that deliberations as to that count would have been prolonged had the evidence been properly excluded. Similarly, an instructional error as to one count—misdescribing or omitting an element, for example—may shorten deliberations on that count. Defendant’s theory would be just as applicable in those cases as it would be here, as any further deliberation on any count would bring into play the abstract possibility that a juror could change his or her mind about a different charge. In effect, defendant’s position appears to be that any time reversible error is found as to one count, all other convictions must be reversed, unless, perhaps, they are supported by overwhelming evidence. Defendant’s argument would require a substantial break from past practice. Although we have never expressly considered defendant’s argument before, we have sustained convictions in several cases in which defendant’s position would have required reversal. For example, in State v. Boots, 308 Or 371, 374-75 , 780 P2d 725 (1989), two theories of aggravated murder were submitted to the jury, and the jury was instructed that it did not need to unanimously agree on a theory of aggravation to convict the defendant of aggravated murder. We held that that failure to require unanimity on the elements of the crime violated Article I, section 11. Id. at 377 . However, we reversed only the defendant’s conviction on aggravated murder, permitting the state the option of retaining the murder conviction, as to which the jury had necessarily reached unanimous agreement. Id. at 381 . The likelihood that a properly instructed jury would have deliberated longer, and could conceivably have reached a different result on the murder conviction, did not feature in the analysis. As we explained, in a second appeal after our remand, “an error-free conviction of a criminal offense need not be retried even though an appellate court has ordered view. Thus, a purely theoretical possibility that a single juror could have reconsidered her view about a different count was enough to prevent the judge’s poll from representing a final verdict. 332 State v. Flores Ramos a retrial of a greater offense of which the lesser offense is a lesser-included offense.” State v. Boots, 315 Or 572, 577 , 848 P2d 76 (1993). Similarly, in State v. Lotches, 331 Or 455 , 17 P3d 1045 (2000), the trial court erred in failing to instruct the jury about the need for unanimity as to the basis for three aggravated murder convictions. We determined that, as to two of the counts, the error was not harmless because the jury may not have been unanimous as to the basis for each conviction. Id. at 470-71 . With respect to the third count, however, we determined that the error was harmless because a different verdict revealed that jury necessarily did agree on the basis for that conviction. Id. at 471-72 . Although a properly instructed jury may well have deliberated longer on the other two aggravated murder counts, we did not hold that those errors required reversal of the third count or of any of defendant’s other convictions. Id. at 472 . Boots and Lotches admittedly were not decided under the “harmless beyond a reasonable doubt” standard applicable to federal constitutional violations, but we see no indication that that standard must be applied any differently. In United States v. Russell, 134 F3d 171 (3d Cir 1998), for example, the defendant was convicted of conducting a continuing criminal enterprise (CCE) and of conspiracy to distribute controlled substances. The jury was instructed that, to return a conviction on the CCE count, it needed to unanimously find that defendant participated in at least three violations of federal drug laws but was not told that it needed to unanimously agree on which violations occurred. Id. at 177 . The court held that the defendant’s right to jury unanimity had been violated and held, under the Chapman standard, that the error was not harmless as to the CCE count. Id. at 182 . But the court nonetheless affirmed defendant’s conspiracy conviction. Id. at 184 . See also State v. Charboneau, 323 Or 38, 51 , 913 P2d 308 (1996) (finding errors harmless as to some counts but not others under the Chapman standard). And in Benton v. Maryland, 395 US 784 , 89 S Ct 2056 , 23 L Ed 2d 707 (1969), the Supreme Court considered something of the reverse situation. In that case, the Cite as 367 Or 292 (2020) 333 defendant had been convicted of burglary and acquitted of larceny at a single trial. Id. at 785. After an appeal, the defendant’s burglary conviction was reversed, and the state retried him—for both burglary and larceny. Id. at 786. The Court held that retrying the defendant for larceny violated the Double Jeopardy Clause, as the defendant had already been tried and acquitted of that offense. Id. at 796. The defendant also sought reversal of his burglary charge, arguing that “some evidence, inadmissible under state law in a trial for burglary alone, was introduced in the joint trial for both burglary and larceny, and that the jury was prejudiced by this evidence.” Id. at 797. The Court did not adopt a per se rule that the mere submission of the larceny offense to the jury, by resulting in additional deliberation on a related topic, could have affected the verdict on the burglary offense. Rather, the Court concluded that “[i]t is not obvious on the face of the record that the burglary conviction was affected by the double jeopardy violation” and remanded the case to consider whether the larceny charge had led to consideration of additional evidence. Id. at 798. Thus, the Court necessarily found that the abstract possibility of an effect on deliberation was insufficient to preclude the finding of harmlessness beyond a reasonable doubt and indicated that only a direct effect on the verdict would suffice to require reversal. While none of those decisions expressly considered the argument that defendant advances here, they demonstrate that the approach to harmless error that defendant would have us adopt—a view that any change that would have lengthened jury deliberations on one count reasonably could have affected the verdict on any count—would be inconsistent with prior practice, in this court and in others. For the reasons given above, the possibility that the jury would have reached a different result on the unanimous counts because of further deliberation on the attempted rape count is too remote to persuade us that the error that occurred in this case was not harmless beyond a reasonable doubt. V. CONCLUSION Because the jury failed to reach a unanimous guilty verdict on count three, attempted first-degree rape, we reverse 334 State v. Flores Ramos defendant’s judgment of conviction as it pertains to that crime. However, as to the unanimous guilty verdicts on all other counts, we conclude that the trial court’s instruction to the jury that it could return a nonunanimous verdict did not amount to a structural error and was harmless beyond a reasonable doubt. We therefore affirm the judgment as to defendant’s other convictions. The decision of the Court of Appeals is affirmed in part and reversed in part. The judgment of the circuit court is affirmed in part and reversed in part, and the case is remanded to the circuit court for further proceedings.