WALTERS, J.

This is an automatic and direct review pursuant to ORS 138.012(1) of sentences of death imposed on defendant after a "penalty phase" trial.1 Defendant raises 33 assignments of error. Five of them merit discussion, specifically, defendant's claims that (1) the adoption of Article I, section 40, of the Oregon Constitution, which, provides that, in specified circumstances, death shall be the penalty for aggravated murder, violated the "separate vote" requirement of Article XVII, section 1, of the Oregon Constitution; (2) the so-called "third question," which asks whether defendant's conduct was "unreasonable in response to the provocation, if any, by the deceased," is constitutionally infirm; (3) the trial court erred by refusing to admit evidence relevant to the third question or failing to pose the question to the jury; (4) the trial court erred in empanelling an "anonymous" jury; and (5) the trial court erred in allowing the state's expert to testify about defendant's consensual homosexual experience as a teenager.

We discuss but reject defendant's arguments as to three of those assignments of error. We conclude that his position as to the fourth and fifth are well-taken. We conclude that the trial court erred in empanelling an "anonymous" jury without finding, as required by this court's decision in State v. Sundberg, 349 Or. 608, 247 P.3d 1213 (2011), that there were strong and particular grounds for believing that the jurors' identities needed to be protected. Because that error was not harmless, we vacate the sentences of death and remand to the circuit court for a new penalty-phase trial.2 We also conclude that the trial court erred in denying defendant's motion to exclude evidence of defendant's homosexual experiences as a teenager.

A. Defendant's challenge to Article I, section 40, of the Oregon Constitution.

The first claim that we discuss is defendant's "separate vote" challenge to the validity of Article I, section 40, of the Oregon Constitution, a provision that was adopted by the people in 1984 as "Ballot Measure 6." Article I, section 40, provides:

Defendant contends that Article I, section 40, was void ab initio because it was adopted in a manner that violated the "separate vote" requirement, set out at Article XVII, section 1, of the Oregon Constitution, for constitutional amendments submitted to the people. Defendant further contends that, because Article I, section 40, was void when adopted, the "entire statutory `Oregon Death Penalty Scheme' (i.e., all substantive and procedural statutes purporting to authorize and implement capital punishment in Oregon * * *)"3 also is void because it is "dependent for constitutionality" on Article I, section 40.

Article XVII, section 1, of the Oregon Constitution provides, in part:

That "separate vote" requirement is "aimed at ensuring that the voters are able to express their will in one vote as to only one constitutional change." Armatta v. Kitzhaber, 327 Or. 250, 269, 959 P.2d 49 (1998).

Armatta is the seminal case on the separate-vote requirement. There, the court compared the wording, historical development, and case law surrounding Article XVII, section 1, with the "single-subject" requirement for initiated measures, set out at Article IV, section 1(2)(d),4 and concluded that the two constitutional provisions were different in a number of respects. Most significantly, the court observed that, whereas the single-subject requirement focuses on the content of a proposed statute or amendment, the separate-vote requirement focuses on the form of submission of an amendment and the potential changes to the existing constitution that the amendment proposes. Id. at 274, 959 P.2d 49. Based on that analysis, the court concluded that, when faced with a claim that a proposed constitutional amendment offends Article XVII, section 1, the proper inquiry is "whether, if adopted, the proposal would make two or more changes to the constitution that are substantive and that are not closely related." Id. at 277, 959 P.2d 49.

The analysis in Armatta, and in other separate-vote cases that have followed, establishes two principles that are important to determining whether a ballot measure makes "two or more changes" to the Oregon Constitution that require separate votes. First, if a measure proposes to add new matter to the constitution, the measure proposes at least one constitutional change. Lehman v. Bradbury, 333 Or. 231, 242-43, 37 P.3d 989 (2002); Armatta, 327 Or. at 277-78, 959 P.2d 49. Second, if a measure has the effect of modifying an existing constitutional provision, it proposes at least one additional change to the constitution, whether that effect is express or implicit. Meyer v. Bradbury, 341 Or. 288, 297, 142 P.3d 1031 (2006); Lehman, 333 Or. at 243, 37 P.3d 989; Armatta, 327 Or. at 278-80, 959 P.2d 49.

Applying those principles, defendant asserts that Measure 6 proposed to change the Oregon Constitution in 15 separate ways. The state acknowledges three of those changes: (1) the addition of a new constitutional requirement that a person convicted of aggravated murder be sentenced to death or life imprisonment; (2) an amendment to former Article I, section 15 — specifically, an exemption from section 15's then-existing5 admonition that criminal penalties must be based on principles of reformation rather than vindictive justice; and (3) an amendment to Article I, section 16, specifically an exemption from section 16's prohibition on cruel and unusual punishment and requirement of proportionate punishment.6 We agree that Article I, section 40, proposed at least those three changes, but, before we analyze whether Measure 6 was invalid because it did not permit a separate vote on each of those changes, we turn to the question whether, as defendant contends, that measure also proposed additional changes to the constitution.

We begin with defendant's assertion that Measure 6 proposed not one, but two, new additions to the constitution. Defendant argues that that is so because the measure declares that (1) the sentence for aggravated murder in certain specified circumstances is death, and (2) the sentence for aggravated murder in other circumstances is life imprisonment. We disagree. The measure provides that, "upon unanimous affirmative jury findings as provided by law," death is the penalty for aggravated murder. The measure then specifies the penalty that must be imposed when the jury does not make those findings. The latter declaration merely completes the sentencing construct that the former declaration describes; it does not propose to add a separate constitutional provision.

Defendant also contends that Measure 6 proposed a separate amendment to the constitution by "permitting" the death penalty. Although defendant does not explain that contention, we assume that it relates to defendant's claim in a different assignment of error that, at the time that Measure 6 was considered, the constitution contained an actual, albeit implied, prohibition on the death penalty. Defendant's theory, in a nutshell, is that the 1964 repeal of two constitutional provisions that authorized the death penalty revived, sub silentio, a preceding (1920) provision that had abolished the death penalty.7 Defendant suggests that Measure 6 separately proposed to amend that prohibition by declaring that the penalty for aggravated murder was, in certain circumstances, death.

Defendant's theory depends on our agreement with his idea that a constitutional provision that is repealed by a later constitutional amendment is revived when the amendment is itself repealed. That idea is contrary to this court's ordinary method of constitutional interpretation, as described in ORS 174.090,8 and we reject it. It follows that, although Measure 6 proposed the addition of a provision to the constitution that "permitted" the death penalty for aggravated murder, it did not, as defendant seems to suggest, also propose the revival of an earlier prohibition on the death penalty and separately amend it.

Next, defendant contends that by using the introductory phrase "notwithstanding sections 15 and 16 of this Article," Measure 6 proposed ten separate constitutional changes. First, defendant argues, the measure effectively proposed to amend the directive in Article I, section 16, that, "[i]n all criminal cases whatever, the jury shall have the right to determine the law, and the facts under the direction of the Court as to the law, and the right of new trial, as in civil case." Based on a reference to that directive in a dissenting opinion in State v. Boots, 315 Or. 572, 592-93, 848 P.2d 76, cert. den. 510 U.S. 1013, 114 S.Ct. 606, 126 L.Ed.2d 571 (1993), defendant argues that section 16 gives juries the authority to acquit criminal defendants regardless of the overwhelming nature of the evidence and that, as extended to the death penalty, it allows juries to "acquit" capital defendants of the death penalty, even as against the evidence. This court has not determined that the cited portion of Article I, section 16, operates in the manner that defendant suggests, but even if it does, there is no logical basis for concluding that the "notwithstanding" clause of Measure 6 interferes with that aspect of Article I, section 16. To the contrary, Measure 6 expressly acknowledges the role of the jury and prescribes the death penalty only "upon unanimous jury findings." Measure 6 does not limit a jury's authority to make the findings that it deems appropriate or to decide that a defendant should not be sentenced to death.

Defendant calculates nine other proposed changes to sections 15 and 16 as follows. Defendant first points out that sections 15 and 16 express three "penal principles" that he contends are entirely separate from one another — the ban on "vindictive justice" expressed in Article I, sections 15, the ban on "cruel and unusual punishment" expressed in Article I, section 16, and the requirement that "penalties [be] proportioned to the offense," also expressed in Article I, section 16. Second, defendant argues, sections 15 and 16 permit three different types of challenges to the death penalty — challenges to the death penalty per se, challenges to the "procedures for imposing a death sentence," and challenges to the method and manner of execution. In defendant's view, the measure precludes a defendant from using all three "penal principles" as a basis for all three types of legal challenges and, in doing so, makes a total of nine changes to the constitution.

Defendant is correct that the text of sections 15 and 16 includes three textually separate provisions — a ban on "vindictive justice," a ban on "cruel and unusual punishment" and a requirement that "penalties [be] proportioned to the offense." Defendant also is correct that those separate provisions may serve as the basis for at least three different types of constitutional challenges — facial challenges to legislatively imposed sanctions, challenges to the procedures prescribed or followed in the imposition of sanctions, and challenges to the application of sanctions in particular circumstances. However, defendant is incorrect that the measure prohibits each of the nine challenges that he posits. As the state correctly explains, "the `notwithstanding clause' can relate only to what * * * follows it in the text," i.e., a declaration that, in certain specified circumstances, death is the sentence for aggravated murder. Matters that are not included within the text that follows the "notwithstanding" clause — for example, issues pertaining to the procedure followed in imposing that sanction, or to the manner and method of execution — are not affected by that clause. Thus, the measure changes sections 15 and 16 by eliminating any constitutional barriers that those sections potentially posed to death as a sanction for aggravated murder; it does not render sections 15 or 16 otherwise inoperable. In other words, the text of the measure appears to preclude challenges to the death penalty as a sanction for aggravated murder; it does not preclude other challenges under Article 1, section 15 and 16.

As a result, it is not surprising that defendant does not rely on the text of the measure for his argument that the "notwithstanding" clause has a more expansive effect. Rather, he argues that application of ordinary interpretive principles, including, particularly, the idea that this court's interpretations of a constitutional text provide context for interpretation of its terms, lead inevitably to the conclusion that Measure 6 precludes any and all constitutional challenges based on the relevant provisions of Article I, sections 15 and 16.9 Defendant particularly relies on Clark v. Paulus, 295 Or. 673, 677, 669 P.2d 794 (1983), in which this court broadly described Measure 6 as exempting the death penalty from the constitutional guarantees "embodied" in Article I, sections 15 and 16, and State v. Wagner, 305 Or. 115, 139, 752 P.2d 1136 (1988), vac'd and remanded 492 U.S. 914, 109 S.Ct. 3235, 106 L.Ed.2d 583 (1989), in which the court, after examining the "notwithstanding" clause, stated in very expansive terms, that, in light of the "notwithstanding" clause of Measure 6, in capital cases the court would not

Defendant's theory fails. Although it is true that, in some cases, the court has described the clause's effect in broad terms, a careful examination of those cases, and all other cases in which challenges to a death sentence under Article I, sections 15 or 16, were at play, reveals that the court has never applied Measure 6 to bar all challenges that rely on those sections.10 In short, the interpretation of Measure 6 upon which defendant's theory is premised has no textual or contextual support.

Finally, defendant contends that other short phrases in the body of Measure 6 proposed two changes to Article I, section 11, of the Oregon Constitution.11 Defendant first notes that Article I, section 11, permits criminal defendants to be found guilty by non-unanimous verdicts, with an exception requiring unanimous verdicts in cases of first degree murder. Defendant suggests that, by requiring "unanimous jury findings," Measure 6 adds a second exception to Article I, section 11, requiring unanimous verdicts in death penalty cases. However, Article I, section 11, prescribes only the number of jurors who must agree on "verdicts" of "guilty or not guilty," whereas the "unanimous affirmative jury findings" requirement of Measure 6 applies to the findings necessary to support the imposition of the death penalty after a jury has reached a guilty verdict. Moreover, to the extent that Measure 6 may be interpreted as requiring unanimous jury findings to support a guilty or not guilty verdict, it is consistent with, and does not change, the requirement in Article I, section 11, that "verdicts of guilty of first degree murder [(the historical analog of aggravated murder)] [be] found only by a unanimous verdict."

Defendant also appears to suggest that the right of criminal defendants to jury trial, also guaranteed by Article I, section 11, is separately affected by Measure 6's reference to jury findings "as provided by law." But defendant does not show, and this court has no reason to believe, that that reference in any way alters the jury trial guarantee afforded by Article I, section 11. The measure does not explicitly or implicitly amend Article I, section 11.

We conclude that sections 15 and 16 of Article I include three textually separate provisions — a ban on "vindictive punishment," a ban on "cruel and unusual punishment" and a requirement that "penalties [be] proportioned to the offense" — that are implicated by Measure 6. We further conclude that, insofar as Measure 6 eliminated the barriers to the imposition of the death penalty posed by each of those provisions, it made three, not two, separate changes to the Oregon Constitution. As the parties acknowledge, the measure also made a fourth constitutional change, namely, creating the new constitutional requirement that persons convicted of aggravated murder be sentenced to death or life imprisonment. Those four changes are all substantive, and the parties do not argue otherwise.

That leaves us to determine whether the four identified changes are "closely-related." In Lehman v. Bradbury, 333 Or. 231, 246, 37 P.3d 989 (2002), the court described the "closely-related" inquiry in the following terms:

In Swett v. Bradbury, 333 Or. 597, 43 P.3d 1094 (2002), the court observed that that statement was "descriptive not prescriptive," and that it was "equally valid analytically to start the inquiry by focusing on the changes themselves." Id. at 607, 43 P.3d 1094. For reasons that will become clear, we choose that option.

One thing that should be immediately obvious about Measure 6 is that it contains only one provision and proposes to do only one thing — prescribe the penalty for aggravated murder. All of the other changes that Measure 6 effects are directed at eliminating the potential constitutional barriers to the imposition of that penalty posed by Article I, sections 15 and 16. Said another way, the three changes that the measure makes to Article I, sections 15 and 16, are necessary corollaries to the new provision that permits the imposition of the death penalty.

In that way, Measure 6 is very different from the measures that the court considered in Armatta and in the "separate vote" cases it decided after Armatta. Each of those measures contained more than one provision, and it would have been possible for voters to separately decide (1) which of many procedural rights they wished to grant to victims (Armatta);12 (2) whether to impose term limits for state and/or federal offices (Lehman); (3) whether to require disclosure of campaign contributions and/or impose a requirement that signature gatherers for initiative petitions be registered Oregon voters (Swett); (4) whether to require payment of just compensation for the financial impact of all regulations on private real property and/or except certain kinds of uses from that compensation requirement (League of Oregon Cities v. State of Oregon, 334 Or. 645, 56 P.3d 892 (2002)); (5) which of a number of different substantive and procedural protections to provide to persons whose property is subject to forfeiture, and how to use forfeiture proceeds (Lincoln Interagency Narcotics Team v. Kitzhaber, 341 Or. 496, 145 P.3d 151 (2006));13 and (6) whether to permit the legislature to prohibit or limit campaign contributions and expenditures and, if so, whether to require more than majority approval for such legislation (Meyer v. Bradbury, 341 Or. 288, 142 P.3d 1031 (2006)).14

Defendant argues that the same is true of Measure 6 — that a voter conceivably could favor one or more of the changes that Measure 6 effects and oppose others. For example, defendant suggests, a voter could support the measure's directive that the penalty for aggravated murder is death, but oppose excepting that directive from the relevant provisions of Article I, sections 15 or 16. But, in so contending, defendant fails to recognize the limits of the measure's effect on those sections. As we have explained, Measure 6 ensures that Article I, sections 15 and 16, will not stand as barriers to imposition of the death penalty. It does not otherwise permit that penalty to be imposed in violation of those sections — for example, by methods that are cruel and unusual. Because the measure did not propose to eliminate all or any of the protections afforded by sections 15 and 16, it was not possible for voters to separately decide whether they wished to do so. A voter who favored death as a penalty for aggravated murder could not achieve that objective without also favoring removal of potential barriers to imposition of that penalty, specifically those found in Article I, Sections 15 and 16.

So understood, the four separate and substantive changes that Measure 6 made to the Oregon Constitution are "closely related." It is not simply that those separate constitutional changes are bound by a shared goal or subject matter — a relationship that this court concluded, in Swett, was insufficiently close to pass muster under Article XVII, section 1. 333 Or. at 609, 43 P.3d 1094. Rather, those changes were necessary to imposition of death as a penalty for aggravated murder in this state.

As articulated in the passage from Lehman quoted above, 333 Or. at 246, 37 P.3d 989, the next step in the separate vote analysis would require that we consider whether the constitutional provisions that Measure 6 affects are closely related. However, it should be clear from the foregoing discussion that that step is not relevant here. Because the changes that Measure 6 proposed to make to the existing constitution were necessary to give effect to the death penalty scheme that is at the measure's heart, a separate vote on whether those effects should logically follow was not possible. Therefore, whether the existing constitutional provisions that were affected by the new constitutional provision were themselves closely related is an unnecessary inquiry. Where, as here, a measure contains only one new provision and the changes that the measure makes to existing provisions are only those necessary to effectuate that provision, the only conclusion that we can reach is that those necessary changes are closely related.

We conclude that, although Measure 6 made more than one substantive change to the Oregon Constitution, those changes were closely related. It follows that Measure 6 did not violate the "separate vote" requirement of Article XVII, section 1, and that Article I, section 40, is not, as defendant asserts, "void ab initio." It also follows that defendant's challenge to the validity of Oregon death penalty scheme, which depends on that assertion, is without merit. That statutory scheme remains in effect.

B. The "third question," ORS 163.150(1)(b)(C).

The next assignments of error that we consider are those that concern the so-called "third question," ORS 163.150(1)(b)(C). The "third question" is one of four questions that are submitted to the jury at the close of a penalty-phase trial, the jury's answers to which determine whether a death sentence will be imposed. The question and the relevant statutory context are as follows:

ORS 163.150(1)(b)(C).

In one of his assignments of error, defendant contends that the inclusion of the "third question" in Oregon's statutory death penalty scheme renders it unconstitutional. In another, defendant contends that the trial court erred in failing to pose the "third question" to the jury. Defendant's positions on those issues are interwoven and somewhat inconsistent. We discuss the latter issue first to give context to the former.

In his challenge to the rulings of the trial court, defendant claims that the trial court erred by (1) failing to instruct the jury on the third question, and (2) excluding evidence related to the third question. The state contends that neither of those claims were preserved.

We agree with the state. At no point in the proceedings did defendant ask that the court instruct the jury on the third question or attempt to offer evidence relating to provocation. Defendant argues, however, that those kinds of direct actions were not required, and that he adequately preserved his present claims by (1) resisting the state's repeated requests that he acknowledge having waived his right to request an instruction on the third question or offer provocation evidence; and (2) attempting to explain to the trial court his various reasons, most of which had to do with the meaning and constitutional implications of ORS 163.150(1)(b)(C), for not seeking an instruction on the third question. Neither of those actions was sufficient to preserve this assignment of error.

As far as we are able to tell, the trial court never questioned defendant's right to offer evidence or request that the court instruct the jury on the "third question." Although defendant insists that, early in the proceedings, the trial court ruled that no evidence going to the issue of provocation "could" be offered by either party, the record suggests that the trial court was responding to what it perceived to be the parties' positions on the matter, and would have considered admitting evidence on provocation if defendant at any point had stated that that was his wish. Instead, defendant repeatedly told the court that he did not intend to offer provocation evidence and, although defendant explained that he had well-considered reasons for declining to seek a third question instruction, the fact is that he repeatedly told the court that he was not seeking such an instruction. In light of those statements, defendant's present claims that the court erred in failing to admit evidence or instruct the jury on the third question are not preserved.

Defendant did, however, preserve his arguments that the "third question" renders the death penalty statutes facially invalid, and we turn to that assignment of error. We begin with defendant's contention that the "third question" is facially vague, and thereby violates the fair notice requirement of the federal Due Process Clause, the equal privileges and immunities provision of Article I, section 20, of the Oregon Constitution, and the ban on ex post facto laws found in Article I, section 21. Defendant mentions in passing the statutory terms "reasonable" and "provocation" as contributing to that vagueness, but concentrates his arguments on the phrase "if raised by the evidence." He contends that that phrase is problematic because it offers no guidance on a myriad of questions: whether, before the trial court poses the third question, evidence must be offered, and, if so, how much, or whether the trial court must pose the third question in any event; when the trial court must make the determination of whether to pose the third question; whether a court's determination that it will not pose the third question precludes the parties from presenting evidence that is directed at the question; and whether the court is even permitted to decide whether to pose the third question or must leave it to the jury to decide whether the reasonableness of defendant's conduct has been "raised by the evidence."

Defendant misapprehends what it means for a statute to be unconstitutionally vague. A statute is unconstitutionally vague if it gives the police, the prosecutor, or the court, uncontrolled or unbridled discretion to punish defendants or to decide what is prohibited, or fails to inform persons subject to it of what conduct on their part will render them liable. State v. Illig-Renn, 341 Or. 228, 238-42, 142 P.3d 62 (2006). Vagueness concerns typically arise when a statute contains terms that are so indeterminate or standardless that they leave questions about its application to the ad hoc judgments of judge, jury, or police. Id. The uncertainties that defendant identifies in this court do not raise those kinds of concerns. Although defendant raises legitimate questions about the circumstances in which a court must pose the third question, those questions are no different from questions that frequently arise about the meaning of many statutes. The fact that a court may be required to interpret a statute does not necessarily make it unconstitutionally vague. As discussed above, defendant had the opportunity, at trial, to press his interpretation of the death penalty statutes and to ask the court to conduct his penalty-phase trial in accord.15 If defendant had done so, and if the trial court had disagreed, the appellate process would have provided defendant with a means of assuring that his penalty phase trial was lawfully conducted. Although questions may remain as to the correct interpretation of the statutory provisions pertaining to the "third question," those questions do not raise the specter of unbridled discretion that renders a statute constitutionally invalid.

Defendant also contends that, if ORS 163.150(1)(b)(C) permits or requires the trial court to withhold the third question from the jury on the court's finding that the issue of provocation is not raised by the evidence, it violates the clear import of Article I, section 11, of the Oregon Constitution16 — the provision that the accused in a capital case may not be tried by the court alone. As we will explain, that argument is too abstract and hypothetical to be a proper subject for this court's jurisdiction.

Defendant's argument is based on the contention that a proper interpretation of ORS 163.150(1)(b)(C) requires the court to submit the "third question" to the jury in every case. Under that interpretation, the jury, not the court, must decide whether the issue of provocation is raised by the evidence and, if so, whether defendant's response was unreasonable. However, as explained above, although defendant informed the trial court that he did not intend to offer evidence of provocation, he never took the position that the court was, nevertheless, required to submit the "third question" to the jury. In fact, defendant told the court that he was not seeking such an instruction. Consequently, the trial court was not called on to interpret the statute and did not give it the interpretation that defendant claims would make it unconstitutional. As the case comes before us, we understand defendant to ask us to decide whether, if the trial court had interpreted the statute to permit the court to withhold the third question from the jury, that interpretation would have been unconstitutional. That is an undertaking in which we will not engage. This court simply does not decide hypothetical or abstract questions. See Strunk v. PERB, 338 Or. 145, 154, 108 P.3d 1058 (2005) ("to be justiciable, controversy must involve present facts as opposed to a dispute which is based on future events or a hypothetical issue").

Defendant's contention that the third question violates Article I, section 16, of the Oregon Constitution, and the Eighth and Fourteenth Amendments to the United States Constitution, because it makes the death penalty statutes irrational or "vertically disproportionate," is similarly problematic. Defendant asserts that if a jury were to answer the "third question" in the negative — that the conduct of a defendant in killing the deceased was not unreasonable in response to the provocation, if any, by the deceased — then it would be irrational and disproportionate for the defendant to be convicted of, and suffer the penalties associated with, aggravated murder. In other words, defendant contends that a negative answer to the third question would constitute a jury finding that the defendant had acted reasonably in response to provocation by the deceased, and would preclude a guilty verdict on the aggravated murder charge. To impose the punishment for aggravated murder in that circumstance would, in defendant's view, be arbitrary and "vertically disproportionate."

As is obvious, however, that circumstance did not occur here. Defendant told the court that he did not intend to offer provocation evidence and that he was not seeking an instruction on the "third question." As a result, the trial court did not submit that question to the jury. Whether a negative answer to that question could, when relevant and submitted, result in disproportionate punishment is an abstract question that we will not answer.

In summary, we reject, in their current posture, each of defendant's arguments that the statutory provision that sets out the third question is constitutionally invalid.

C. Defendant's challenge to use of "anonymous jury" procedures.

We next address defendant's claim that the trial court violated various constitutional provisions, including Article I, sections 10 and 11, of the Oregon Constitution, and the Sixth Amendment to the United States Constitution,17 by empanelling an "anonymous jury" to hear and decide whether defendant should be subjected to the death penalty. We first consider defendant's contention that the empanelling of an "anonymous jury" denied him a fair trial and an impartial jury, in violation of Article I, section 11, of the Oregon Constitution, under this court's recent opinion in State v. Sundberg, 349 Or. 608, 247 P.3d 1213 (2011).18

In Sundberg, the trial court denied a criminal defendant's motion for a new trial, which the defendant sought on the ground that an "anonymous jury selection procedure" that the trial court had employed over the defendant's objection was an "irregularity" that denied him a fair trial and an impartial jury, in violation of Article I, section 11. In Sundberg, the issue arose out of the trial court's decision that the parties would not be provided with, and would not be permitted to learn through the voir dire process, the names, addresses or employers of the potential jurors. When the defendant objected to the court's insistence on an "anonymous jury," the court overruled the objection, explaining that its decision reflected a circuit-wide response to concerns expressed over the years by jurors about having their names known to the litigants. 349 Or. at 610-11, 247 P.3d 1213. As it turned out, however, a significant number of the jurors who underwent voir dire under the "anonymous jury" procedure in the defendant's case had also undergone voir dire in a different case on the same day, before a judge that did not employ the "anonymous jury" procedure. After the defendant was found guilty, he moved for a new trial, arguing that, particularly when some of the jurors had been exposed to a non-anonymous voir dire procedure, the use of the anonymous procedure would have suggested to the jurors that defendant might pose some sort of danger to them and, thus, might have affected the jurors' ability to be impartial. The trial court denied the motion. Id. at 612-13, 247 P.3d 1213.

On review, the defendant asked this court to apply the rule that many federal and state courts have applied — that anonymous juries are permitted only when there are strong grounds for believing that the jury needs the protection of anonymity and the trial court takes reasonable precautions to minimize the prejudicial effects on the defendant and to ensure the protection of the defendant's constitutional rights. The defendant argued that, under that rule, the decision to employ an anonymous jury in his case was error, because there had been no finding of any need to protect jurors and no attempt to minimize possible prejudice to him.

In considering the question the defendant raised, the court first observed that, while trial courts have inherent authority to empanel an anonymous jury, that authority must be exercised in a way that does not violate defendant's constitutional rights, including his or her right under Article I, section 11, of the Oregon Constitution to "trial by an impartial jury." The court next determined that anonymity procedures could affect a defendant's right to an impartial jury in two ways:

Sundberg, 349 Or. at 620 [247 P.3d 1213]. With respect to the first effect, the court noted that, although knowledge of names, addresses and employers is not always necessary to achieve an impartial jury, such knowledge is helpful and in some cases, may be crucial ("If a McCoy is on trial, she will want to know if any of the prospective jurors are Hatfields"). Id. at 621 [247 P.3d 1213]. With respect to the second effect, the court observed that anonymous juries have not been the norm in Oregon and elsewhere and, as such, might suggest dangerousness (and, thus, guilt) to prospective jurors. Id. In light of those potential effects, the court concluded that that the approach that the defendant had proposed was essentially the correct one, and the court adopted the rule as expressed in one of the federal cases that the defendant had cited:

Id. (quoting U.S v. Paccione, 949 F.2d 1183, 1192 (2d Cir.1991), cert. den. 505 U.S. 1220, 112 S.Ct. 3029, 120 L.Ed.2d 900 (1992)).

The court explained that, to satisfy the first part of that test, the trial court must determine that the circumstances of the particular trial provide sufficient grounds to believe that the jurors need the protection of anonymity. The court noted that some courts had identified a specific group of factors to be considered in making that determination, but it declined to endorse any particular list of factors. The court added, however, that the determination "must be made on the facts of each case — and not on the basis of a generalized desire to protect the anonymity of all jurors in all cases in the interest of juror privacy." Sundberg, 349 Or. at 622, 247 P.3d 1213. The court went on to explain that, "[i]f grounds exist to empanel an anonymous jury, then the trial court may do so, but it must take reasonable precautions to ensure that defendant's right to an impartial jury is protected." Id. Depending on the particular circumstances, such precautions might include permitting extensive voir dire of each juror regarding potentially relevant matters and providing a "plausible and nonprejudicial" explanation to jurors for not disclosing their identities. Id. at 623, 247 P.3d 1213.

Applying the foregoing analysis to the circumstances that were at issue, the court concluded that, because the trial court had not made any finding that the circumstances of the particular case would support a need to protect the jurors' identities, the use of an anonymous jury was error. The court further concluded that the error was not harmless, because, in the circumstances, "the unexplained use of an anonymous jury created too great a risk that the jury may have believed that defendant was dangerous, and, therefore, that he was more likely to be guilty." Id. at 625, 247 P.3d 1213.

Defendant argues that Sundberg controls the present case. He contends that, as in Sundberg, the trial court erred in deciding to use an anonymous jury when it had not determined, and had no grounds for determining that there were "strong reasons" for believing that the jury needed the protection of anonymity that were particular to defendant's case. In response, the state argues that the rule from Sundberg is inapplicable for a number of reasons — including that the jury in this case was not an "anonymous jury" within the meaning of Sundberg. The state notes, in that regard, that, although the trial judge in the present case did not permit prospective juror's names to be used in open court and gave the jurors the option of withholding certain other identifying information requested on juror questionnaires, the court provided the attorneys with a list of the jurors' names and addresses and told them that, assuming the information became relevant, they could question individual jurors about other information that the jurors may have withheld in responding to their juror questionnaires. Furthermore, the state notes, the trial court told the jurors that they had been given the option of withholding identifying information "for no other reason than to help protect your anonymity from the public and the press." The state contends that those circumstances stand in stark contrast to the circumstances in Sundberg, where the court prevented even the parties' attorneys from discovering the jurors' names, addresses, and employers, and where the trial court gave the jury no explanation of its reason for requiring anonymity. The state adds that, to the extent that defendant wishes to argue that the jury in this case was "anonymous," as that term is used in the Sundberg analysis, because he, as opposed to his attorney, was precluded from learning the jurors' identities, he is precluded from doing so by his failure to argue that particular nuance before the trial court.

The facts material to those arguments are as follows. Defendant's penalty phase proceeding was scheduled to begin early in 2006, and two orientation meetings for prospective jurors were scheduled for January 17, 2006. During the months that preceded the scheduled meetings, counsel for the parties met a number of times in the chambers of the judge who would be presiding in the trial to discuss what should be included in the juror questionnaire. All eventually agreed to a final questionnaire that included, among other things, questions about the juror's name, address, and employer, the name and employer of the juror's spouse, the names, ages, and occupations of juror's children, the names, ages, and occupations of the juror's siblings, and whether the juror, or any of the juror's friends or family members were or had been involved in certain organizations or activities.

On January 13, 2006, at an off-the-record meeting with the parties' counsel on a different matter, the judge announced that, in filling out the juror questionnaire, prospective jurors would not be required to reveal identifying information about themselves or any other person. It was decided (it is not clear by whom and by what process) that prospective jurors would be given a list of the questions on the juror questionnaire that would be considered optional. As a result of that decision, when juror questionnaires were distributed to the jury venire at the January 17 orientation meetings, they were accompanied by a separate instruction sheet that advised the prospective jurors that they were not required to reveal, in certain designated questions, their names or addresses, the names of their spouses or children or anyone else, or the names of their employers. The trial court also orally advised the prospective jurors that they did not have to answer the designated questions and that the option was being provided to protect jurors' identities. Although the trial court disputed that it had done so, defense counsel understood that the court had informed the jurors that the questionnaires would be provided only to the lawyers and not to anyone else.

Defendant objected to the court's decision to thus "anonymize" the jury panel and moved for dismissal of the jury venire and an evidentiary hearing. In the motion, defendant argued that the trial court's instructions regarding the jury questionnaires might result in a biased jury because it might lead jurors to infer that their anonymity was required because defendant was dangerous. Defendant also argued that permitting jurors to withhold complete information would affect defendant's ability to conduct voir dire and to intelligently utilize peremptory challenges, implicating his right to an impartial jury. Defendant argued that the only remedy that could cure the resulting prejudice was dismissal and replacement of the jury venire.

The requested evidentiary hearing was held on January 19, 2006. Through testimony from the jury coordinator and presiding judge for the county, defendant established that, some months before, the county's judges had informally, and not necessarily unanimously, adopted a policy that, in every case, juror names would not be used in open court. The purpose of the policy was to ensure the privacy of jurors.

After defendant elicited that evidence, he reiterated his position that the potential jurors should not have been told that they could omit identifying information from their questionnaires, concentrating his oral arguments on a concern that the anonymous jury procedure could send a message to the jurors that there was reason to fear defendant. The trial court attempted to respond to that concern by instructing the state to submit a draft of a curative instruction to the defense. The court said that defendant would "sign off on it, and then we'll proceed." However, defense counsel informed the court that defendant would not sign off on a curative instruction because "our position is that no curative instruction is going to fix the taint."

In an apparent response to defendant's contention that the provision of the anonymity option would affect his ability to conduct voir dire effectively, the trial court observed that many of the jurors had chosen to answer all of the questions. The court also advised the parties that, "if, for some reason, any of this information, what is left blank on the questionnaires, becomes somehow relevant during the voir dire process, we'll deal with that on a case-by-case basis." Finally, the court pointed out that the names of jurors would be supplied to the attorneys "with the instructions that those were not to be released to any third parties." The court explained that, by "third parties," it meant "people who are not employees or agents of the attorneys," and added that "we should probably have a protective order regarding that."

Later in the same hearing, the prosecutor presented a draft protective order and the court inquired whether defense counsel was comfortable with it. Counsel responded that "as a substitute measure for getting what I actually want, I don't have any objection to the form of the order." The order, which the judge ultimately signed, provided that the jurors' names and addresses, which would be provided to counsel, "may not be viewed or otherwise used except in connection with representation of a party." It also specifically provided that the names and addresses "may [not] be provided or viewed in any way by the defendant."

Some days later, as voir dire was prepared to commence, the parties were provided with the judge's "opening remarks" to the jury, which included the anticipated curative instruction. It read:

However, in giving its opening remarks to individual prospective jurors, the court did not always repeat that instruction as drafted. The court separately addressed each juror and, more often than not, told the juror that "the attorneys," rather than "the parties," had been provided with the juror's name. In fact, of the 12 jurors who were selected to hear defendant's case, only four were told that "the parties" had been provided with their names. The rest heard that "the attorneys" had been provided with that information.

A few days into voir dire, counsel for defendant filed a "renewed objection to the "anonymization" of juror data and identities and demand for dismissal of tainted jurors" — specifically arguing both that the selection procedure implied to the jury that defendant was presently dangerous and that the procedure prevented counsel from eliciting information about potential jurors that was essential to the intelligent and meaningful exercise of peremptory challenges. The court denied the "renewed objections" motion.

As noted, defendant contends that, in taking the described actions, the trial court empanelled an anonymous jury in violation of this court's decision in Sundberg. The state responds that the anonymity with which Sundberg is concerned is confined to circumstances that exactly mirror the facts of that case — a trial court's prohibition on the parties and their attorneys learning jurors' names or identifying information, and that here the court disclosed the names of jurors to defense counsel. Moreover, the state contends, we may not consider the fact that defendant was prohibited from obtaining that information, because defendant did not specifically object to that prohibition below.

In Sundberg, the defendant articulated, as the basis for his objection at trial to the anonymous juror procedure, that the procedure would preclude him from conducting "adequate voir dire." The defendant did not argue until after the jury returned a guilty verdict (in a post-trial motion), that that procedure was improper because it may have affected the way that the jurors perceived him and, therefore, their impartiality. However, the defendant's failure to timely state that second concern did not prevent this court from considering, in its analysis, the effect that the trial court's procedure may have had on jurors' perception of the defendant.

In this case, defendant voiced the same two concerns about the "anonymous" jury procedure that the defendant in Sundberg voiced and that the court considered in that case. Although defendant focused in oral argument primarily on the effect that the trial court's selection procedure could have on the jurors' ability to be impartial and stated his objection that that procedure interfered with his ability to conduct voir dire only in writing and in general terms, we do not think that defendant's choices confine this court's analysis. As in Sundberg, we will consider the full effect of the procedure that the trial court imposed in deciding whether it met constitutional requirements.

Important to our willingness to do so is the fact that defendant did alert the trial court in written memoranda that he objected to its jury selection procedure on the basis that that procedure would affect his ability to conduct voir dire. Before the off-the-record meeting on January 13, the parties had agreed on a juror questionnaire that asked jurors to disclose identifying information about themselves, their families and employers. The parties apparently understood, at that point, that juror numbers, instead of names, would be used in open court, but that defense counsel and defendant would have access to the juror questionnaires and all of the information that jurors provided in response, including their names and addresses. Defendant did not object to that procedure. At the January 13 off-the-record meeting, however, the court announced that prospective jurors would be permitted the option of refusing to provide certain of the information requested in the questionnaire. Defense counsel understood the judge as also instructing the jurors at that meeting that their questionnaires would be provided to the attorneys "and not to anyone else," which defense counsel took to mean "and not to defendant." Thus, when defense counsel filed her motion for an evidentiary hearing on the court's decision to "anonymize" the jury, it was her belief that the procedure that the court was imposing would permit her but not defendant to learn the information on the juror questionnaires. In the memorandum that counsel filed in support of that motion, counsel objected that the court's withholding of complete information would "affect defendant's ability to conduct voir dire" and to intelligently utilize peremptory challenges. Counsel thereby objected generally to the procedure that she understood the trial court to have imposed and did not limit her objection to any particular aspect of that procedure.

Although the trial court disputed that it had told the jurors that the juror questionnaires would be provided to the attorneys and "not to anyone else," the trial court knew when it considered that memorandum that the procedure that it was imposing would in fact permit defense counsel to learn the names and addresses of the jurors, but preclude defendant from doing so. At the hearing on the motion, the court explained that defense counsel would be provided with juror names but that those names were not to be released to "third parties." The court explained that, by third parties, it meant "people who are not employees or agents of the attorneys." When the trial court entered its protective order, it specifically prohibited counsel from releasing juror names and addresses to defendant. Thus, the trial court adopted a procedure that expressly precluded defendant from learning the names and addresses of jurors, and defendant had objected generally and in writing to that procedure on the basis that it would affect his ability to conduct voir dire.

It is true, as the state observes, that defendant focused much of his oral argument on his contention that, by permitting jurors the option of withholding certain information in responding to the juror questionnaires, the court had tainted their view of defendant and his dangerousness. It also is true that the court and the parties extensively discussed the extent of that taint and the "curative instruction" that the state proposed to address it. But at no time during that discussion did defendant abandon his position, conveyed in his written motion, that the procedure that the court had announced also had the effect of preventing him from obtaining the information that he needed in voir dire. Defendant did not waive that argument when counsel stated that she did not have any objection to the form of the protective order that prohibited counsel form sharing juror names with defendant. Instead, defendant specifically renewed that argument in his written, renewed objection to "anonymization" of juror data. The fact that defendant concentrated, in oral argument, on the effect of the court's procedures on the jurors' impartiality does not preclude this court from considering the fact that the trial court entered an order precluding defendant from learning the identity of the jurors who were charged with meting out his penalty. We will therefore include that fact in our analysis of whether the selection procedure that the trial court imposed was sufficiently different from that used in Sundberg that its holding is inapplicable here. We turn to that question now.

As noted, the state argues that, under the procedure the trial court used in this case, the jury was not "anonymous" in the Sundberg sense — first, because the trial court disclosed juror identifying information to counsel, and, second, because the court informed jurors that the procedure was for the sole purpose of protecting their privacy from the public. We disagree. Jurors may be "anonymous" in different ways — from the defendant's perspective, because the defendant does not know their identifying information, from counsel's perspective, because counsel does not know their identifying information, and from their own perspective, because they understand that identifying information that they ordinarily would be required to provide may be withheld, and, in any event, will not be provided to the defendant. In Sundberg, the court discussed the concerns that arise when a jury is anonymous from two of those perspectives — the defendant's perspective and the jurors' own perspective. The court described the problem that arises when juror are anonymous from the defendant's perspective by observing that "[i]f McCoy is on trial, she will want to know if any of the prospective jurors are Hatfields." Providing jurors' names and addresses to counsel may mitigate the potential harm that flows from thus anonymizing the jury, but it cannot completely eliminate it: That is so because, in colloquial terms, the defendant may be the only person who is aware of the dispute between the Hatfields and the McCoys.

In Sundberg, the problem that the court identified when considering anonymity from the jurors' perspective was that anonymous juries have not been the norm in Oregon, and a deviation from the norm of full disclosure might suggest the defendant's dangerousness. The court explained that, when jurors learned that they would remain anonymous and were aware that such anonymity deviated from Oregon's norm of full disclosure, they could conclude that their privacy was being protected because the defendant was dangerous. In this case, the jurors also would have been aware that their ability to remain anonymous was a deviation from the norm. We can infer that awareness from the fact that the jurors received pre-printed questionnaires that requested their names and other personal information, but were then told, both orally and in writing, that they had the option of withholding that information.

It is, of course, true that the trial court made an effort to explain that there was a non-prejudicial reason for the deviation — that they were being permitted to withhold information "for no other reason than to protect their anonymity from the public and the press." However, the court also made a statement that the jurors could understand as an indication that their identities were being protected from defendant. The court told eight of the 12 jurors that it would furnish their names, not to "the parties," but to "the attorneys." Under the circumstances, we do not think that the trial court's explanation for permitting anonymity negated the risk to juror impartiality that the unusual anonymity procedure posed.

We conclude that, in spite of the differences that the state has noted, the procedure that the trial court followed in this case gave rise to the same risks that the court identified in Sundberg. And, because those risks are the same, the rule of Sundberg is applicable. According to that rule, a court may empanel an anonymous jury "only when the trial court finds that the circumstances of a particular case justify that practice and takes steps to mitigate any prejudice to defendant." Sundberg, 349 Or. at 624, 247 P.3d 1213. It may not empanel an anonymous jury based on "a generalized desire to protect the anonymity of all jurors in all cases in the interest of juror privacy." Id. at 622, 247 P.3d 1213.

In the present case, the trial court did not make the required findings. Instead, the court explained that it was merely carrying out a county-wide policy of protecting the identities of jurors — seemingly relying on the very kind of "generalized desire to protect the anonymity of all jurors in all cases" that this court suggested was inadequate in Sundberg. Id. At no point did the trial court state that the circumstances of the particular case provided grounds for believing that the jurors needed the protection of anonymity — even when the scope and bona fides of the "policy" were questioned. In fact, when counsel for the state suggested that the procedures could be explained to the jury in terms of concerns about publicity, the trial court responded emphatically that "[t]he publicity surrounding this case has nothing to do with * * * what I'm doing by way of anonymity."

The state points out that, by instructing the prospective jurors that the procedures had been used to protect their anonymity "from the public and the press," the trial court took the very kind of "reasonable precautions to ensure that the defendant's right to an impartial jury is protected" that the Sundberg rule demands. Id. at 622, 247 P.3d 1213. But, under Sundberg, the requirement that the trial court take "reasonable precautions" is necessary but secondary: It does not absolve the court from the rule's primary requirement — that it determine that particular circumstances of the case provide grounds for believing that the jurors need the protection of anonymity.20 The trial court did not make the required determination. Under Sundberg, the use of an anonymous jury in the absence of such a determination was error.

Still, the error is not a ground for a new penalty phase trial if it is found to be harmless, i.e., if we determine that there is little likelihood that it affected the verdict. State v. Davis, 336 Or. 19, 32, 77 P.3d 1111 (2003). The state contends that there is little likelihood that the anonymity procedures used in this case affected the jurors' decision-making, because the trial court instructed the jurors that they were being used solely to protect the juror's privacy from the press and public. The state also argues that, even if the anonymity procedures may have conveyed some slight implication of dangerousness, it would have been insignificant in the context of the overwhelming evidence that the jury had heard about defendant's horrific crimes, and for that reason would have been unlikely to influence the jury's ultimate decision on the fourth question — whether "there is a probability that [] defendant would commit criminal acts of violence that would constitute a continuing threat to society." ORS 163.150(1)(b)(B). Finally, the state argues that, because defendant's lawyers had all of the jurors' names and were permitted to inquire about the information that the jurors withheld if it became relevant, the anonymity procedures could not have hampered defendant's ability to conduct effective voir dire.

For the reasons we have already stated, we are not persuaded. The trial court did not decide that this case was different from others nor explain to the jurors why this case was different from others — leaving the jurors to speculate that, perhaps, in this particular case, their disclosure of personal information might expose them to risks that would not be present in other cases. We also cannot ignore the fact that eight of the jurors were instructed that "the attorneys," as opposed to "the parties," had been provided with their names. Jurors who heard that the trial court had given their names to "the attorneys" would have assumed, correctly, that only the attorneys — and not defendant — had been provided with their names. That detail could have undermined the trial court's general explanation of the procedure as protecting jurors' anonymity from the "press and public," and focused the jurors on whether this defendant was, perhaps, more dangerous than defendants in other cases. And the concerns about juror impartiality that arise in connection with that kind of speculation by jurors would only be amplified in this context — a penalty phase proceeding where defendant's "future dangerousness" was specifically at issue. ORS 163.150(1)(b)(B).

We also are unpersuaded by the state's suggestion that, in the context of the evidence that the jurors would see and hear about the gruesome murders that defendant had committed, the "small" implication of defendant's dangerousness that might be associated with the trial court's anonymity procedures could have little force or effect on the jurors' ultimate assessment of defendant's future dangerousness. Evidence of defendant's obvious past dangerousness to his past victims certainly speaks to defendant's future dangerousness at some level, but a suggestion that defendant constitutes a present or future danger to jurors and their families cannot be dismissed as comparatively negligible when the question under consideration is "probability that the defendant would commit criminal acts of violence that would constitute a continuing threat to society."

Finally, we are not convinced that the anonymous jury procedures were harmless with respect to defendant's ability to use voir dire effectively. As explained, without being able to share juror's names with a defendant, counsel may not have the information necessary to make effective use of that information. Here, defendant may have been the only person who could connect a name with a potential bias. Moreover, even leaving aside the question of whether defendant's lawyers' knowledge of those names is an adequate substitute for defendant's knowledge, the fact remains that the anonymity procedures that the trial court imposed prevented even the lawyers from knowing other potentially relevant information (i.e., names of spouses, children, employers) about the jurors who eventually were seated. Here, three of the jurors who sat on defendant's case declined to answer at least one of the other questions that the trial court presented as "optional."21

The state points out that the trial court instructed the lawyers that, in their oral examinations of jurors, they were free to inquire into any matters that a juror had omitted from his or her questionnaire, "should those issues become relevant." The state contends that, if the lawyers declined to pursue that option, it was their decision, and not the court's procedure, that prevented them from obtaining the omitted information.

The state's argument misapprehends the utility of the requested information. The information that the jurors were permitted to withhold — about spouses, children, workplaces, etc. — were included in the questionnaires because the parties and the court had agreed that they should be — with the implication that answers to those questions would be relevant or could suggest important avenues for further relevant inquiry. In other words, the jurors' answers to the questions were, by agreement, either inherently relevant or were triggers for further relevant questioning. When a juror withheld information that may have alerted a party that further questioning might be fruitful, that questioning could not occur: To return to the example that the court used in Sundberg, without knowing that a juror is named Hatfield, a defendant may not think to ask the juror if she was involved in the dispute with the McCoys and certainly could not demonstrate to the court that the name of the juror had become relevant. Thus, the argument that the state interposes is circular and, ultimately, empty. Without the background information that the answers would have provided, defendant could not meet the trial court's standard for further inquiry, because he could not know and could not show that that any particular line of inquiry was relevant.

In summary, the trial court and the parties designed a voir dire process that was intended to provide the parties with the information that all agreed was necessary to effective voir dire. The trial court then changed that process and precluded defendant from learning the jurors' names and precluded counsel and defendant from learning other potentially relevant information about potential jurors, including several jurors who were seated on the jury and who heard defendant's case. The fact that defendant's lawyers had access to the jurors' names and addresses and were permitted to inquire into other withheld information "should those issues become relevant," did not cure the problem that the change in process created.

Under these circumstances, we cannot say that there was little likelihood that the error in employing anonymous jury procedures, without finding strong and particular grounds to do so, affected the verdict. Because the error was not harmless, defendant's sentence of death must be vacated.

D. Defendant's past homosexual experience.

The last assignment of error that we discuss is defendant's claim that the trial court erred in permitting a psychiatrist to testify about a homosexual relationship that defendant had had in his youth. In defendant's previous penalty phase trials, a psychiatrist testified about that relationship. At the outset of this penalty phase trial, the defendant asked the court to exclude that same material on grounds of irrelevance and unfair prejudice,22 citing Beam v. Paskett, 3 F.3d 1301 (9th Cir.1993), cert. den. 511 U.S. 1060, 114 S.Ct. 1631, 128 L.Ed.2d 354 (1994). In Beam, the court held that evidence of a defendant's nonviolent, consensual homosexual conduct was not admissible to prove that defendant would be a "continuing threat" to society, when the state had not presented "evidence demonstrating a close link between that history and the defendant's future dangerousness." Id. at 1309.

We agree with the court's statement in Beam that

Id.

In this case, the state failed to introduce evidence demonstrating a link between defendant's homosexual experience as a teenager and his future dangerousness. The state theorizes that such a link exists because defendant tortured and murdered his female victims to satisfy sexual urges, and that defendant's youthful homosexual experience shows that those violent sexual urges, and the danger that defendant would act on them, would continue to exist even in a prison setting where defendant only would be interacting with males. But the state's theory is just that — a theory. The state did not present evidence that sexually obsessive and violent persons in general easily transfer their deviant urges and behaviors from one sex to another — or that defendant, himself, was susceptible to that kind of adjustment. Neither did the state present evidence that, regardless of defendant's apparent heterosexuality as an adult, his homosexual experience in youth indicates an underlying flexibility in sexual preferences. In the absence of factual evidence establishing such links, or some more direct link between homosexual activity in youth and sexual violence toward males as an adult, the relevance under OEC 401 of defendant's teenage experience to the issue of his future dangerousness is questionable and, under OEC 403, the danger of its unfair prejudice certainly would outweigh any slight relevance that it may have.

We agree with defendant that the trial court erred in allowing the state to present this evidence. We need not, however, decide if this error, alone, requires reversal. Because we already have identified one error in the proceedings below that was sufficient to require reversal and remand, we need not proceed with a harmless error analysis.

The sentence of death is vacated, and the case is remanded to the circuit court for further proceedings.