State v. Jackson

498 P.3d 788, 368 Or. 705

• Court: Oregon Supreme Court
• Decided: November 17, 2021
• Docket: S067622
• Status: Published

Opinion

Argued and submitted November 17, 2020, order of circuit court affirmed November 17, 2021

STATE OF OREGON, Appellant, v. HOMER LEE JACKSON III, aka Homer Jackson, aka Homer Lee Jackson, Respondent. (CC 15CR46257) (SC S067622) 498 P3d 788

Defendant was charged with the murder of four victims. DNA consistent with defendant’s was found at the scene of each crime. Before trial, the state moved to cross-admit the crime scene evidence, arguing that the evidence from all four crime scenes, including the DNA evidence, was relevant to each of the four charged crimes under the doctrine of chances. The trial court denied the motion. The state filed a direct interlocutory appeal under ORS 138.045 to challenge the trial court’s order. Defendant challenged this court’s jurisdiction to hear the state’s interlocutory appeal under ORS 138.045(1)(d), which provides the state with a right to interlocutory appeal of pretrial orders “suppressing evidence.” Held: (1) Defendant did not demonstrate that the court clearly erred when, in prior decisions, it had interpreted the phrase “suppressing evidence” broadly to include a pretrial order that excludes evidence on any grounds; (2) the statistical reasoning used by the doctrine of chances is insufficient to establish the rele- vance of the other crime scene evidence; instead, the doctrine’s statistical reason- ing provides only an intermediate factual inference about the entire set of crime scene evidence that must be linked to the charged crime through an additional intermediate inference or theory of relevance; and (3) the state failed to link its proposed use of the other crime scene evidence to the fact it wished to prove at trial in a way that does not rely on a prohibited “bad character” inference. The order of the circuit court is affirmed.

En Banc On appeal from an order of the Multnomah County Circuit Court under ORS 138.045(2) and ORAP 12.07.* Marc D. Brown, Deputy Public Defender, Office of Public Defense Services, Salem, argued the cause and filed the briefs for respondent. Also on the briefs was Ernest G. Lannet, Chief Defender, Criminal Appellate Section. ______________ * Michael A. Greenlick, Judge. 706 State v. Jackson

David B. Thompson, Assistant Attorney General, Salem, argued the cause and filed the briefs for appellant. Also on the briefs were Ellen F. Rosenblum, Attorney General, and Benjamin Gutman, Solicitor General. NAKAMOTO, J. The order of the circuit court is affirmed. Garrett, J., concurred and filed an opinion in which Balmer, J., joined. Cite as 368 Or 705 (2021) 707

NAKAMOTO, J. The state charged defendant with the murder of four victims, who were killed over the span of 10 years. DNA consistent with defendant’s was found at the scene of each crime. Before trial, the state moved to cross-admit the crime scene evidence, arguing that the evidence from all four crime scenes, including the DNA evidence, was rele- vant to each of the four charged crimes. In support of that argument, the state relied on the doctrine of chances, which is, broadly speaking, the idea that repeated incidents of rare and similar events are unlikely to be explained by coinci- dence or random chance. The trial court denied the state’s motion to cross-admit the crime scene evidence. The state seeks direct interlocutory review of the trial court’s order, contending that the trial court erred in excluding the evidence from the other three crime scenes from defendant’s trial for one of the murders. The state initially argues (1) that the doctrine of chances supplies a theory of relevance for the crime scene evidence concerning the presence of defendant’s DNA near the bodies of three other murdered women and (2) that the doctrine does not depend on prohibited inferences about defendant’s bad char- acter and resultant propensity to commit criminal acts. In accordance with the Oregon Evidence Code, we initially hold that the doctrine of chances, standing alone, is insuffi- cient to make the other crime scene evidence relevant to any charged crime. The doctrine of chances instead provides an intermediate factual inference about the entire set of crime scene evidence that must be linked to the state’s proof of a fact material to the charged crime through a separate the- ory of relevance. In supplemental briefing, the state offers an alter- native. The state argues that, even if the doctrine of chances does not by itself supply the basis for the relevance of the DNA crime scene evidence, that evidence nonetheless is rele- vant to facts in its case by articulating a chain of inferences, ending with “defendant was the killer in each murder.” And in the state’s view, neither the penultimate inference nor any intermediate inference relies on prohibited “bad charac- ter” inferences, in violation of OEC 404. Keeping the state’s 708 State v. Jackson

articulated purpose and chain of reasoning for introducing the DNA evidence from all four crime scenes at the forefront and again applying requirements of the Oregon Evidence Code, we further hold that the state, stretching the doctrine of chances beyond its limits, fails to link its proposed use of the other crime scene evidence to the fact it wishes to prove at trial in a way that does not rely on a prohibited “bad char- acter” inference. Because we reject the state’s alternative argument as well, we affirm the trial court’s order excluding the evidence. I. BACKGROUND In a single indictment, the state charged defen- dant in 2019 with 15 counts of first-degree murder for the deaths of four Black female victims: TH, AA, LW, and LT. Each victim had been engaging in prostitution in northeast Portland. One victim was a juvenile, while the other three were young adults, aged 29 or younger. The four murders occurred from 1983 to 1993: TH and AA were killed in 1983, LW was killed in 1987, and LT was killed in 1993. Defendant’s DNA, or else DNA consistent with defendant’s, was found at each of the four crime scenes, all located in north or northeast Portland. TH was found par- tially submerged in a slough near Delta Park. Defendant’s DNA was on a belt left near TH’s body. The state contends that the belt belonged to TH and had been used as a liga- ture by her assailant. The DNA on the belt was consistent with defendant’s DNA and approximately 1 in 1.32 million African Americans. AA was found in a room inside an abandoned north- east Portland house. In that room were two burnt matches and two cigarette butts. One cigarette butt had defendant’s DNA on it. The odds that the DNA would match someone other than defendant are less than 1 in 10 billion. The other cigarette butt had DNA on it matching AA’s DNA. Defendant’s fingerprint was also found on a cabinet door in the same room where AA’s body was found. The cabinet door had been removed, and AA’s bloody sock was lying against it. LW was found in an empty lot near a pedestrian overpass in north Portland. Defendant’s DNA was found in Cite as 368 Or 705 (2021) 709

fingernail scrapings taken from LW. The odds that the DNA would match someone other than defendant are less than 1 in 10 billion. The state offered evidence that defendant’s DNA was the “predominant” or “major profile” from those scrapings, meaning that there was more of defendant’s DNA under LW’s fingernails than LW’s own DNA. According to the state’s expert, the fact that defendant’s DNA was “pre- dominant” suggests that defendant had contact with LW within about five hours before her death. Finally, LT was found near the same pedestrian overpass as LW, though six years later. In a manner unlike the other victims, LT had been brutalized by her assailant. That included a bite mark on her nipple. Forensic investiga- tors swabbed the bite mark, which revealed a small amount of DNA. From that small amount of DNA, investigators were able to develop a profile that is consistent with defendant’s DNA and the DNA of about 1 in 3,896 males. The state moved to cross-admit the evidence from all the crime scenes as part of its proof for the murder charges involving each victim. The state argued that the “other acts evidence”—the presence of defendant’s DNA and the similarities among the crimes—was relevant to proving that defendant committed each charged crime. For example, in attempting to prove that defendant had killed TH (the charged crime), the state sought to admit evidence of the AA, LW, and LT murders (the other crimes). And, in attempting to prove that defendant had killed AA (the charged crime), the state sought to admit evidence of the TH, LW, and LT murders (the other crimes). The state made the same argu- ments regarding the LW and LT murders. The state anticipated that defendant would argue that, as to each charged crime, his DNA ended up at the crime scene as the result of random chance—most likely, that he frequently employed prostitutes and, as a result, might have had contact with the victim for reasons unrelated to mur- der. The state maintained that, as to each charged crime, the DNA evidence at the other crime scenes was relevant to rebutting that anticipated defense. According to the state, the doctrine of chances—which it describes as “a theory of logical relevance * * * that avoids OEC 404(3)’s character 710 State v. Jackson

prohibition” and that “does not depend on character infer- ence”—established that the DNA evidence at all the crime scenes was relevant to defendant’s anticipated argument. Defendant opposed the motion on numerous grounds and further argued that, because the evidence was not cross-admissible, presenting the evidence of each crime to the same jury would result in substantial prejudice. On that basis, defendant moved to sever the cases for trial. The trial court held a three-day hearing on the state’s motion in limine to admit “other acts” evidence and on defendant’s motion to sever the cases. The trial court received a copy of documents that the state had disclosed during discovery regarding the evidence from each crime scene described above, and the court heard testimony. To rebut the state’s contention that the murders were very similar and showed a modus operandi, defendant presented testimony from a clinical and forensic psychologist.1 And the state presented testimony directed to the doctrine of chances from a scientist with the Oregon State Police labo- ratory’s DNA unit, who addressed the specific DNA evidence found at the scenes, and from a professor of criminology. Based on research and statistics, the professor testified that the chances of a prostitute being murdered within any given 24-hour period is rare: 1 in 160,000. The professor also tes- tified that the chances of a person’s DNA being present on a given day at the scene where a murdered prostitute is found is also rare, with the estimated chances varying depend- ing on how often the person employs prostitutes. The state argued that the infrequency of prostitute homicide estab- lishes the implausibility of random chance explaining defen- dant’s DNA being at the scenes of four separate prostitute homicides. Following the hearing, the trial court denied the state’s motion to cross-admit the crime scene evidence on three grounds. First, the trial court concluded that the state had failed to establish the foundational requirements nec- essary to support doctrine-of-chances reasoning—namely, 1 The trial court rejected the state’s theory that the evidence from all the crime scenes was cross-admissible to prove a modus operandi that identified defendant as the killer. That issue is not on appeal. Cite as 368 Or 705 (2021) 711

that the events were sufficiently similar and sufficiently infrequent. The trial court explained that the state’s evi- dence supporting its doctrine-of-chances theory “invites jurors to convict based on speculation and conjecture.” Second, the trial court concluded that, even if the state could provide the necessary evidentiary support for the doctrine of chances, the state was misusing the doctrine. The trial court explained that, based on this court’s case law, the doctrine of chances may be used only to prove that certain conduct was performed intentionally, rather than inadvertently, citing State v. Tena, 362 Or 514, 412 P3d 175 (2018). The trial court held that the state was improperly attempting to use the doctrine to establish defendant’s iden- tity as the murderer, rather than to prove that an act was committed intentionally. Third, the trial court determined that, because the doctrine of chances was unavailable as a matter of law, the other crime scene evidence could be relevant to the charged crime based only on the tendency of the other crime scene evidence to establish defendant’s character and propensity to commit the charged act, a theory of relevance that is barred by OEC 404(3). The trial court concluded that the other crime scene “evidence only has persuasive force if it means that the defendant committed the charged acts. If it is not admissible to prove identity, then it is inadmissible character evidence.” The trial court granted defendant’s motion to sever the cases for trial. That ruling is not at issue on appeal.2 Instead, the state seeks direct interlocutory review in this court under ORS 138.045 to challenge the trial court’s decision not to cross-admit the evidence from all the crime scenes in each murder trial. The state contends that the trial court erred when it denied the state’s motion to cross-admit the evidence, arguing that the evidence from each crime scene is relevant for nonpropensity purposes under the doctrine of chances. In addition to opposing that argument, defendant also

2 In its opening brief, the state also assigned error to the trial court’s deci- sion to sever the cases for trial but withdrew that assignment of error in its reply. 712 State v. Jackson

contends that this court lacks jurisdiction to resolve that question because the trial court’s order is not appealable under ORS 138.045. We begin with defendant’s jurisdic- tional argument. II. JURISDICTION Defendant challenges this court’s jurisdiction to hear the state’s interlocutory appeal. The right to appeal is statutory. See State v. K. P., 324 Or 1, 4, 921 P2d 380 (1996) (“[T]he right of appeal must be conferred by a statute. Without such a statute, there is no jurisdiction to consider an attempted appeal.” (Internal citation omitted.)). As statutory support for its appeal, the state relies on ORS 138.045(1)(d) (providing for interlocutory appeal to the Court of Appeals of a pretrial order “suppressing evidence”) and ORS 138.045(2) (providing for interlocutory appeal of an order in paragraph (1)(d) to the Supreme Court when the defendant is charged with murder). Together, those provisions allow the state to seek direct interlocutory appeal to this court in murder cases when the trial court has issued a pretrial order “sup- pressing evidence.” The question is whether the trial court’s order is one that suppresses evidence. For decades, this court has broadly interpreted the phrase “suppressing evidence,” as it is used in the interlocu- tory appeal statute, to include a pretrial order that excludes evidence on any grounds. See State v. Hess, 342 Or 647, 654, 159 P3d 309 (2007) (“The trial court has excluded evidence that the state desires to introduce at trial. That is all that ORS 138.060(1)(c)[, the predecessor to ORS 138.045(1)(d),] requires.”); see also State v. Koennecke, 274 Or 169, 172-73, 545 P2d 127 (1976) (holding that the state could seek inter- locutory review of a trial court order excluding evidence on nonconstitutional grounds). The state argues that the trial court’s order constitutes “suppressing evidence,” under that broad definition of the phrase, because the order excludes the evidence from each crime scene for the purpose of prov- ing defendant’s guilt as to each of the other crimes. Defendant does not disagree with the conclusion that the trial court’s order in this case falls within that broad definition. Instead, defendant argues that this court’s prior decisions have erred in interpreting the phrase “suppressing Cite as 368 Or 705 (2021) 713

evidence” so broadly. Under defendant’s argument, “sup- pressing evidence” is a legal term of art that has a narrower definition than “excluding evidence.” Defendant contends that whether an order suppresses evidence, rather than merely excludes evidence, depends on the grounds that the trial court relies on in concluding that the evidence should not be presented the jury. According to defendant, an order “suppressing evidence” refers only to a trial court’s ruling that evidence may not be presented to the jury on the ground that the collection of the evidence violated the defendant’s constitutional rights. The trial court’s order in this case would fall outside that definition because the trial court excluded the other crime scene evidence based on the Oregon Evidence Code, and not based on any constitutional violations. Defendant, therefore, asks us to overrule our prior decisions, narrowly interpret the phrase “suppressing evidence” as it is used in the interlocutory appeal statute, and conclude that statute does not provide the state with a right to interlocutory review of the trial court’s order in this case. The question whether to overrule one of our prior decisions is a question of stare decisis. “[S]tare decisis is a prudential doctrine that is defined by the competing needs for stability and flexibility in Oregon law.” Farmers Ins. Co. v. Mowry, 350 Or 686, 697-98, 261 P3d 1 (2011). It “requires that we begin with the assumption that issues considered in our prior cases are correctly decided.” Brownstone Homes Condo. Assn. v. Brownstone Forest Hts., 358 Or 223, 236, 363 P3d 467 (2015) (internal quotation marks and citation omitted). And it “means that the party seeking to change a precedent must assume responsibility for affirmatively per- suading us that we should abandon that precedent.” State v. Ciancanelli, 339 Or 282, 290, 121 P3d 613 (2005). We are not persuaded to abandon the broad inter- pretation that this court previously adopted. Defendant’s argument depends on establishing that “suppressing evi- dence” is a narrow legal term of art. Although defendant’s understanding of that term is consistent with its current usage, defendant has not demonstrated that the term had an established legal meaning in 1969, when the legislature adopted the relevant statutory provision. 714 State v. Jackson

For example, to support his narrow legal defini- tion of the term “suppressing evidence,” defendant cites the 1979 edition of Black’s Law Dictionary as providing defini- tions that are consistent with his narrow interpretation. “Suppression of evidence” is defined as “[t]he ruling of the trial judge to the effect that evidence sought to be admit- ted should be excluded because it was illegally acquired.” Black’s Law Dictionary 1291 (5th ed 1979). And “motion to suppress” is also defined as being based on illegal, typically unconstitutional, state action: “Device used to eliminate from the trial of a criminal case evidence which has been secured illegally, generally in vio- lation of the Fourth Amendment (search and seizure), the Fifth Amendment (privilege against self incrimination), or the Sixth Amendment (right to assistance of counsel, right of confrontation etc.), of U.S. Constitution.” Id. at 914. The problem for defendant is that, although those definitions appear in the 1979 edition of Black’s Law Dictionary, the legislature adopted the relevant statutory provision 10 years earlier, in 1969. Or Laws 1969, ch 529, § 1. And it is not clear that “suppressing evidence” had that clearly defined legal meaning at the time that the provision was adopted. The 1968 edition of Black’s Law Dictionary contains no definitions for the terms “suppression of evi- dence” and “motion to suppress.” And the term “suppress” is defined, but it is given a broad ordinary meaning, rather than the narrow legal meaning that defendant ascribes to it: “To put a stop to a thing actually existing; to prohibit, put down, to prevent, subdue, or end by force.” Black’s Law Dictionary 1609 (4th rev ed 1968). Even the 1979 edition of Black’s Law Dictionary that defendant relies on demonstrates the continuing devel- opment of the term “suppression.” Although that edition defines “suppression of evidence” and “motion to suppress” as excluding evidence that was illegally acquired, the defini- tion of “suppress” suggests that the term could also be used to describe evidence excluded on other grounds—specifically, relevance: “To ‘suppress evidence’ is to keep it from being used in a trial by showing that it was either gathered Cite as 368 Or 705 (2021) 715

illegally or that it is irrelevant.” Black’s Law Dictionary at 1291 (5th ed 1979). No legislative history further informs our analysis. Recordings from the legislative committee hearings are unavailable. And the minutes from those hearings and the documents considered by the committee members do not illuminate what the legislature intended in referring to “suppressing evidence.” As a result, although defendant’s interpretation is plausible, defendant has not persuaded us that this court clearly erred when it previously concluded that the phrase “suppressing evidence,” as it is used in ORS 138.045(1)(d), includes orders that exclude evidence on any grounds. Because the trial court’s order at issue is a pretrial order that excludes evidence, this court has jurisdiction to con- sider the state’s direct interlocutory appeal of that order. III. DOCTRINE OF CHANCES The parties dispute whether the trial court erred in denying the state’s motion to cross-admit the crime scene evidence under the doctrine of chances. The parties’ argu- ments are framed by our prior decisions on the doctrine of chances. Because those decisions have not always provided clear guidance on the proper scope of the doctrine or how it works, and because the proper focus for analysis of the admissibility of evidence begins with the Oregon Evidence Code, it is useful to review the relevant rules of evidence that frame our analysis before considering our case law and the parties’ arguments. A. Evidentiary Framework To be admissible, evidence must be relevant. See OEC 402 (“Evidence which is not relevant is not admissi- ble.”). “Relevant evidence” is defined as “evidence having any tendency to make the existence of any fact that is of consequence to the determination of the action more proba- ble or less probable than it would be without the evidence.” OEC 401. Although all admissible evidence must be rele- vant, not all relevant evidence is admissible. The grounds 716 State v. Jackson

for excluding relevant evidence include a bar on admitting “[e]vidence of other crimes, wrongs or acts * * * to prove the character of a person in order to show that the person acted in conformity therewith.” OEC 404(3).3 In evidence law, “character” “ ‘means a person’s disposition or propensity to engage or not to engage in certain types of behavior.’ ” State v. Skillicorn, 367 Or 464, 475-76, 479 P3d 254 (2021) (quot- ing Laird C. Kirkpatrick, Oregon Evidence § 404.03, 213 (7th ed 2020)). Thus, uncharged misconduct may not be used “to argue that the defendant has either a general propensity to engage in misconduct or a specific propensity to engage in misconduct like the charged crime and, therefore, it is more likely that the defendant committed the charged crime.” 367 Or at 476; see also id. (“The prosecution may not use uncharged misconduct evidence to prove ‘that the defendant is either generally a criminal or more particularly a rapist or burglar.’ ” (Quoting Edward J. Imwinkelried, 1 Uncharged Misconduct Evidence § 2:19, 2-139 (2013).)). But as OEC 404(3) further provides, that limitation does not preclude the admission of “other acts” or uncharged misconduct evi- dence for noncharacter purposes, “such as proof of motive, opportunity, intent, preparation, plan, knowledge, identity, or absence of mistake or accident.” 4 The proponent of the evidence “has the burden of showing that the proffered evidence is relevant and proba- tive of some noncharacter purpose.” State v. Pitt, 352 Or 566, 576, 293 P3d 1002 (2012). Whether evidence has a nonchar- acter purpose is not determined solely by assessing whether the ultimate fact that the proponent seeks to prove is a fact

3 The parties and the trial court have addressed only OEC 404(3) and not OEC 404(4), which contains a rule about “other acts” evidence specifically for criminal actions, because OEC 404(4) does not apply in this case. The latest mur- der occurred in 1993, before subsection (4) was added to the rule. See Or Laws 1997, ch 313, § 29; State v. Shaw, 338 Or 586, 613, 113 P3d 898 (2005) (holding that to apply OEC 404(4) to a crime that predates the rule’s enactment would result in an ex post facto violation). 4 The concurrence contends that we are perpetuating a misconception that OEC 404(3) prohibits “propensity” reasoning rather than “character” reasoning. 368 Or at 733-34 (Garrett, J., concurring). We emphasize that this case concerns prohibited character-based reasoning. As we later discuss, the state’s second argument for admissibility of the evidence from the other crime scenes relies on character-based reasoning prohibited by OEC 404(3), a conclusion with which the concurrence agrees. Id. at 737-38 (Garrett, J., concurring). Cite as 368 Or 705 (2021) 717

about a person’s character or propensity to commit crimes. Instead, this court has repeatedly explained that evidence may also be inadmissible, regardless of the ultimate fact to be proved, whenever “ ‘the chain of logical relevance’ con- necting the evidence to the fact it is proffered to prove relies on ‘an inference relating to [a person’s] character or propen- sities.’ ” Skillicorn, 367 Or at 476 (quoting State v. Johnson, 340 Or 319, 338, 131 P3d 173 (2006)); see also id. at 483 (“A proponent should identify the logical path that it will be asking the factfinder to follow.”).

As a result, the admissibility determination requires courts to focus on the proponent’s theory of relevance that connects the evidence to the fact of consequence. See id. at 475 (“The proponent’s theory of relevance is critical.”). If the proponent’s “theory of relevance requires the factfinder to employ propensity reasoning”—to rely on an inference about the defendant’s bad character and resultant propensity to commit criminal acts—at any link in the chain of logical relevance, then the evidence is subject to the limits on char- acter evidence in OEC 404(3). Id. at 476.

As noted above, the state seeks to admit the crime scene evidence from the three other murders as relevant to proving that defendant killed the victim of the charged crime. Thus, the state has the burden of identifying a chain of logical relevance connecting the crime scene evidence from the other charged crimes, for example, the AA, LW, and LT murders, to the charges at issue at trial for one victim, the TH murder in this example, without relying on inferences about defendant’s bad character and resultant propensity to commit criminal acts.

The state acknowledges the bar on character evi- dence in OEC 404(3). The state, therefore, disclaims that it is arguing that defendant’s DNA being at three other crime scenes establishes that defendant likely killed some other victims, likely had a propensity to kill prostitutes, and likely acted on that propensity in killing the victim in the charged act. Instead, the state argues that the doctrine of chances provides the necessary logical relevance connecting the “other scenes” evidence to the charged crime, without 718 State v. Jackson

depending on an inference about defendant’s bad character and propensity to commit murder. B. Overview of the Doctrine of Chances The doctrine of chances is “based on the objective improbability of the recurrence of uncommon events.” Skillicorn, 367 Or at 484. Under the doctrine, “if the number of events in a series claimed to have the same uncommon cause exceeds the number that can reasonably be expected to have that cause, a factfinder can infer that not all of the events actually have that cause.” Id. The doctrine’s basic idea is that, when viewed in isolation, a particular event might plausibly be explained by accident or random chance. But when that event has hap- pened repeatedly and the defendant says that each event was the result of accident or random chance, then accident or random chance become less plausible explanations for the series of events: “ ‘The fortuitous coincidence becomes too abnormal, bizarre, implausible, unusual, or objectively improbable to be believed.’ ” Id. at 486 (quoting Imwinkelried, 1 Uncharged Misconduct Evidence at § 5:6, 5-29-30 (empha- sis omitted)). The series of events in this case is defendant’s DNA being at each crime scene. The state anticipates that defen- dant will argue that, in each case, his DNA was at the crime scene for reasons unrelated to the crime. For example, if defendant frequently employed prostitutes, he might have had contact with the victim before the crimes took place or might have had contact with someone else who carried and transferred defendant’s DNA to the scene. We discuss those kinds of explanations for his DNA being at each crime scene as constituting random chance. By that we mean that defen- dant’s DNA was at the crime scene for a reason other than because he killed the victim. The state does not dispute that an innocent person’s DNA can end up at a crime scene for a variety of reasons unrelated to the crime. In fact, at each crime scene in this case, the state found DNA from individuals whom the state contends were not involved with the crimes. The state asserts that the DNA of those individuals ended up at the crime scene Cite as 368 Or 705 (2021) 719

by chance. According to the state, defendant is unlike those individuals because his DNA was at all four crime scenes. In briefing to the trial court, the state argued that, although it might be plausible that defendant’s DNA could end up at one crime scene by chance, “it defies all common sense to believe that he was similarly unlucky three other times.” (Emphasis in original.) See also Edward J. Imwinkelried, An Evidentiary Paradox: Defending the Character Evidence Prohibition by Upholding a Non-Character Theory of Logical Relevance, the Doctrine of Chances, 40 U Rich L Rev 419, 423 (2006) (“Innocent persons sometimes accidentally become enmeshed in suspicious circumstances, but it is objectively unlikely that will happen over and over again by random chance.”).

The question, then, is the extent to which the doc- trine of chances, as the state proposes to use it in this case, renders the evidence from all the crime scenes relevant and admissible to prove that defendant is the person who mur- dered the victim in the charged crime at trial, without rely- ing on character-based reasoning prohibited by OEC 404(3). Unfortunately, our case law has done more to impede, rather than facilitate, attempts to answer that question.

As we next discuss, our case law before our recent decision in Skillicorn explains the state’s initial argument in this case. The state contends in its opening brief that the doctrine itself is sufficient to make evidence from the other crime scenes relevant and admissible to prove the identity of the killer in the charged crime: “As noted, other-acts evidence is admissible under OEC 404(3) if it is relevant for a noncharacter purpose. The doc- trine of chances—a theory of logical relevance—avoids OEC 404(3)’s character prohibition because it does not depend on a character inference. Under the doctrine, evidence of multiple similar events gives rise to the permissible inter- mediate inference that, based on objective improbability, the events cannot be explained as coincidence. Thus, if evidence is logically relevant under a doctrine-of-chances theory—i.e., it gives rise to the requisite intermediate inference of objective improbability of coincidence and it has any tendency to establish the existence or absence of 720 State v. Jackson

a material fact—the evidence is admissible under OEC 404(3) to prove any material issue in the case.” (Emphasis in original; internal citation and quotation marks omitted.) In other words, the state’s initial argument is that the doctrine of chances supplies a theory of relevance for evidence from the other murder scenes that is not based on the bad character of defendant and that using the doc- trine makes the evidence admissible under OEC 404(3) to prove any material issue, and not only intentionality. We refer to that argument in shorthand as a noncharacter the- ory of relevance. C. Case Law Concerning the Doctrine of Chances Our case law does not resolve the extent to which the doctrine of chances represents a noncharacter theory of relevance for evidence of other acts extrinsic to the charged act. Although this court has decided cases that purport to define and apply the doctrine of chances, most of those cases misconceive the nature of the doctrine. Our most recent decision on the topic, Skillicorn, represents a course correc- tion. But Skillicorn did not resolve the extent to which the doctrine of chances represents a noncharacter theory of rel- evance; the facts in that case did not implicate the doctrine of chances in the first place. We review the case law before Skillicorn here because, even in supplemental briefing sub- mitted after Skillicorn, the parties’ arguments rely on those prior decisions. This court’s first decision on the doctrine of chances is State v. Johns, 301 Or 535, 725 P2d 312 (1986). The defen- dant in Johns shot his wife and claimed that it happened acci- dentally. The state attempted to rebut that claim by offering evidence that the defendant had assaulted his previous wife with a gun years earlier. The defendant had been convicted of the earlier assault and did not dispute it. Id. at 537-42. This court affirmed the state’s use of the prior assault under what it purported to be the doctrine of chances, reason- ing that the more frequently a person undertakes similar conduct—such as injuring a spouse—the more likely that that conduct was undertaken intentionally. On the same day that it decided Johns, the court decided another case based on the doctrine of chances, Cite as 368 Or 705 (2021) 721

State v. Allen, 301 Or 569, 725 P2d 331 (1986). In Allen, the defendant was charged with hiring an acquaintance to com- mit arson, and the trial court allowed the state to present evidence that the defendant committed an earlier arson. Although the defendant had admitted to his participation in the earlier arson, he denied any involvement in the charged arson. This court affirmed the admission of the prior arson as an application of the doctrine of chances. Id. at 573-74 (discussing the doctrine of chances); id. at 577 (affirming admission of prior arson). The court next addressed the doctrine of chances in State v. Leistiko, 352 Or 172, 282 P3d 857, adh’d to as modified on recons, 352 Or 622, 292 P3d 522 (2012). The defendant was charged with the rape of three separate vic- tims. He claimed that all the sexual contact was consensual. The state did not seek to cross-admit the evidence of each charged crime; it instead offered testimony from a fourth victim who alleged similar, uncharged conduct against the defendant. The state argued that the fourth victim’s testi- mony was admissible under Johns to prove that the defen- dant “intended to forcibly compel the victims to have sexual intercourse with him.” Id. at 182 (internal quotation marks omitted). This court rejected that argument, explaining that, as applied in Johns, the doctrine of chances may be used only to rebut assertions that certain conduct—either undis- puted conduct or conduct already found by the jury—was performed intentionally, rather than accidentally. Id. at 185. According to the court, allowing the doctrine of chances to be used beyond that would pose “an unacceptable risk that the uncharged misconduct evidence is being admitted to prove the act, not the defendant’s mental state.” Id. at 186. Later decisions by this court followed Leistiko by limiting the doctrine of chances to proving that an act was committed intentionally, rather than inadvertently. In State v. Baughman, 361 Or 386, 393 P3d 1132 (2017), the defen- dant was charged with the sexual abuse of a minor, called B. The trial court allowed the state to offer testimony from a woman called A, who alleged that the defendant had sim- ilarly abused her when she was a minor. This court held 722 State v. Jackson

that A’s testimony could not be used to prove the defendant’s intent to abuse B under the doctrine of chances because the “defendant did not raise mistake or inadvertence as a defense.” Id. at 407. Finally, in Tena, which the trial court relied on below, the defendant was charged with assaulting his girl- friend. The defendant argued that his girlfriend tripped after they had had an argument. To rebut the defendant’s assertion that mere coincidence explained the proximity between the argument and his girlfriend’s injury, the state sought to introduce evidence that the defendant had twice previously abused girlfriends. The state argued that its use of the prior incidents was justified under the doctrine of chances. This court rejected the state’s argument by rely- ing on the limit established in Leistiko: “It is important to emphasize that the doctrine of chances applies only to explain whether or not an act that a defendant performed was performed intentionally. It does not apply when there is a dispute about whether the defendant performed the act at all.” Tena, 362 Or at 524. That was the state of our case law when this court recently decided Skillicorn, which represents a fundamental reconsideration of the doctrine of chances. The defendant in Skillicorn left his girlfriend’s house in his truck after a dispute. As the defendant was leaving, he hit the back of his girlfriend’s car and then hit another car belonging to a neighbor down the street. The defendant maintained that both crashes were accidental. The state sought to admit evidence of previous incidents in which the defendant had intentionally driven aggressively in the neighborhood. The state maintained that the evidence was admissible under Johns and the doctrine of chances to rebut the defendant’s claim of an accident. The trial court admitted the evidence. Skillicorn, 367 Or at 467-72. In rejecting the state’s argument and reversing the trial court’s admission of the evidence, the court held that Johns misconceives the doctrine of chances. The court noted that the doctrine of chances relies on the objective improb- ability of the recurrence of uncommon events, such as acci- dents. Id. at 484. Thus, “where the doctrine is used to prove Cite as 368 Or 705 (2021) 723

‘lack of accident,’ the application of the doctrine requires an assessment of the odds that all of the events in a series were accidental; therefore, it does not make sense to include events in the series that are known not to have that cause or explanation.” Id. at 489. Including intentional acts as part of the series does “not enable the factfinder to make any determination regarding whether the defendant has suf- fered more accidents than could reasonably be expected.” Id. As a result, it is a mistake to consider a defendant’s prior intentional conduct when assessing whether a charged act was accidental under the doctrine of chances. The court then pointed out that the analysis in Johns suffers from that mistake. As described above, the defendant in Johns was charged with shooting his second wife, which he claimed was accidental. Contrary to the court’s conclusion in Johns, evidence that the defendant intentionally assaulted his first wife could not be used under the doctrine of chances to establish that the defendant had experienced an objectively improbable number of accidental injuries to his spouses. There was only one claimed accident, not a recurring series of accidents. As a result, the court’s reasoning in Johns to admit the prior assault cannot properly be called doctrine-of- chances reasoning. See id. at 491 (“[T]his court in Johns described the doctrine of chances but did not properly apply it.”). Instead, the state in offering the evidence, and this court in affirming its admission, engaged in prohibited character- based reasoning: The defendant intentionally assaulted his first wife, and so it is likely that he intentionally shot his second wife. See id. (“The state had used propensity rea- soning, and this court followed suit.”). This court, therefore, “overrule[d] Johns to the extent that it holds that evidence of uncharged misconduct can be admitted under the doc- trine of chances for the purpose of arguing that, because the defendant engaged in deliberate conduct before, it is likely that he engaged in it again during the charged incident.” Id. at 493. The mistake that Skillicorn identified in Johns is a mistake that runs through our other cases, which relied on and applied Johns. None of those cases implicates the 724 State v. Jackson

doctrine of chances, properly conceived, because none of those cases asks a factfinder to assess the objective improb- ability of a series of chance events. As a result, those cases do not inform the scope of the doctrine of chances, properly conceived. See Ciancanelli, 339 Or at 290 (“Many decisions of this court serve as precedent in later decisions. Thus, dis- avowing one case may undermine the precedential signifi- cance of several others.”). D. Whether the Doctrine of Chances Is a Noncharacter Theory of Relevance that Makes “Other Acts” Evidence Admissible The state argues correctly that its proposed appli- cation of the doctrine of chances in this case does not suffer from the same mistake that the court made in Johns. The state notes that, in offering the other crime scene evidence, it intends the jury to assess the improbability that defen- dant’s DNA could be at four similar crime scenes by random chance. Thus, this case properly implicates the doctrine of chances. And, as previously noted, the state initially takes the position that, because it properly invokes the doctrine of chances with respect to the evidence from the other crime scenes, then the evidence is admissible to prove any mate- rial fact at issue at trial, and not only intentionality, with- out using character-based reasoning prohibited by OEC 404(3), including defendant’s identity as the murderer in any charged crime. In briefing its position, the state relies extensively on this court’s description of the doctrine of chances in Skillicorn. Skillicorn does not, however, resolve whether the application of the doctrine of chances to evidence of other acts supplies a noncharacter theory of relevance that ren- ders the evidence admissible to establish any material fact at trial. Although the court in Skillicorn raised that ques- tion, the court did not attempt to resolve it, because the par- ties in that case assumed that the doctrine of chances was a noncharacter theory of relevance.5 5 We explained in Skillicorn that the parties “regard the doctrine as a nonpropensity theory of relevance. There is a debate among commentators regarding whether the doctrine actually is a nonpropensity theory of relevance. But we do not understand defendant to Cite as 368 Or 705 (2021) 725

This case, therefore, allows us to address for the first time the extent to which the doctrine of chances, prop- erly conceived, is a noncharacter theory of relevance that is sufficient to make the other-acts evidence relevant to prove any material fact at issue at trial. Considering whether the doctrine of chances is a noncharacter theory of relevance entails two distinct questions: (1) whether the doctrine of chances relies on prohibited character reasoning and (2) whether applying the doctrine of chances to the evi- dence is sufficient, without relying on additional inferences, to make a fact material to the charged crime more or less probable. As to the first question, the state argues at length that the doctrine of chances does not rely on prohibited character-based reasoning. The state explains that estab- lishing the improbability that defendant’s DNA could be at four similar crime scenes by random chance does not rely on any inferences about defendant’s bad character; rather, that improbability is established statistically and probabi- listically merely by assessing the number of times a person’s DNA can be expected to be at the scene of a prostitute homi- cide by random chance and comparing that to the number of times defendant’s DNA has been at the scenes of a prostitute homicide. To meet that argument, defendant maintains, among other things, that the state has failed to satisfy the foundational requirements necessary to make that comparison—namely, the similarity and unusual frequency of the events in question. See Skillicorn, 367 Or at 487-88 (examining those foundational requirements); id. at 488 (noting that the proponent has the burden to satisfy those foundational requirements). However, we assume, without deciding, that the state has satisfied those foundational requirements, because we conclude that the state’s argu- ment fails on other grounds.

categorically challenge the doctrine of chances as a basis for the admission of uncharged misconduct evidence[.]” 367 Or at 474-75 (internal citation omitted). We reiterated later in the opinion that “the parties in this case assume that the doctrine can be used as a non- propensity theory of relevance to justify the admission of uncharged misconduct evidence.” Id. at 484 n 5. 726 State v. Jackson

Thus, we take as a given that the state has estab- lished the statistical improbability that defendant’s DNA could be at four crime scenes of murdered prostitutes by random chance. And, to the extent the doctrine of chances is limited to such statistical reasoning, the state is correct that it entails no character-based reasoning about defendant. But, as we will explain, limiting the doctrine of chances to such statistical reasoning also limits its relevance. To assess the sufficiency of the doctrine of chances in establishing the relevance of the other crime scene evidence, we ask whether the doctrine of chances directly makes a fact about the charged crime more probable without rely- ing on additional inferences that go beyond the doctrine’s statistical reasoning. If additional inferences are needed to make the evidence relevant to the charged crime, then those inferences must be identified and evaluated to ensure that they do not rely on prohibited character reasoning. The state erroneously assumes that the doctrine of chances may serve as a basis, in itself, for the relevance of the evidence from the other crime scenes. We conclude that the doctrine of chances, properly conceived, does not by itself constitute a theory of relevance; therefore, the mere application of the doctrine of chances is insufficient to make evidence relevant under OEC 401 and admissible in accor- dance with OEC 404(3). We return to first principles under the Oregon Evidence Code: Relevant evidence is evidence that has a tendency to make the existence of a fact of consequence more probable or less probable than it would be without the evidence, and the admissibility of evidence requires courts to focus on the proponent’s theory of relevance that connects the evidence to the fact of consequence. The inquiry is not abstract, such as whether the evidence could be relevant to the charged crime. The inquiry is grounded in the rea- son for which the proponent offers the evidence. For its ini- tial argument, the state’s offered reason is the doctrine of chances, and the state relies only on the doctrine of chances because it does not require character-based reasoning. We are therefore required to resolve whether the doctrine of chances is sufficient to make the other crime scene evidence Cite as 368 Or 705 (2021) 727

relevant to one particular murder. We conclude that it is not. In a nutshell, the state relies on the doctrine in ways that go beyond the limits of the doctrine’s statistical reasoning. The state contends that, by establishing that random chance is unlikely to explain defendant’s DNA at all four crime scenes considered as a group, it has decreased the likelihood that random chance can explain defendant’s DNA being at one particular crime scene. But establish- ing the improbability that random chance explains all the events in a series does not diminish the likelihood that ran- dom chance can explain any one event in the series. To elaborate, if we start with the proposition that there is a reasonable likelihood that one event might rea- sonably be explained by chance, then, under the doctrine of chances, that proposition about one event remains true, even as more events are added to the series. As noted above, the state argued to the trial court that chance might reasonably explain an innocent person’s DNA being at one crime scene. At the same time, the doctrine of chances recognizes that it is less likely that chance can explain two similar uncommon events happening by random chance, as compared to just one event. It is even less likely that three similar uncommon events will happen by random chance, as compared to two. And so on. Under the doctrine of chances, the likelihood of chance explaining all the events goes down as more events are added to the series. But, no matter how many events are added, the doctrine of chances, by itself, does not dimin- ish the original proposition that chance might reasonably explain one event in the series. The doctrine of chances, by itself and without additional inferences, never changes the underlying odds with which we started. In Skillicorn, we recognized that the doctrine of chances is of “ ‘limited probative value’ ” in attempting to dis- prove that random chance can explain any one event in the series. 367 Or at 487 (quoting Imwinkelried, 1 Uncharged Misconduct Evidence 4:1 at 4-34). While discussing the doc- trine in the context of proving intentionality, the court in Skillicorn explained: “ ‘The only direct inference from the doctrine of chances is that one or some of the incidents were 728 State v. Jackson

not accidents.’ [The doctrine] does not prove that any par- ticular incident was intentional, much less that they all were.” Id. (quoting Imwinkelried, 1 Uncharged Misconduct Evidence 4:1 at 4-34). Stated more generally, the only direct inference from the doctrine of chances is that chance is unlikely to explain one or some events in the series. In this case, as to each charged crime, the state must prove that crime beyond a reasonable doubt. See State v. Boots, 308 Or 371, 377, 780 P2d 725 (1989) (requiring members of the jury to agree on the grounds upon which the state proved charges of aggravated murder, holding that the jury “may agree on both [grounds], if both are proved beyond a reasonable doubt”). In asking the jury to convict defendant of one murder, for example the murder of TH, the state must prove beyond a reasonable doubt that defendant killed TH. By using the doctrine of chances to establish the improbability that random chance can explain defendant’s DNA at all four crime scenes, the state can claim only the inference that defendant’s DNA was at one or some of those four crime scenes for reasons other than random chance. When the state is trying to prove that defendant killed TH, the inference that the doctrine of chances provides adds no new information about the likelihood that random chance can explain defendant’s DNA at TH’s crime scene. The fact that defendant in this case has been charged with all four crimes in the series does not affect the analysis. Charging a defendant with multiple crimes in a single indictment, and even trying those separate crimes to the same jury, does not change the state’s evidentiary bur- den. The state is required to prove each crime on its own merits. It is not permitted to ask the jury to convict defen- dant of murdering TH based on the statistical likelihood that defendant committed at least one murder out of a group of four murders in which TH was one of the victims. To recap, the doctrine of chances represents a per- missible path to one factual inference: Based on the objec- tive improbability that uncommon events will happen repeatedly by chance, a factfinder may reasonably infer that one or some events in a group of unlikely events were not the result of chance. In this case, assuming that the Cite as 368 Or 705 (2021) 729

foundational requirements have been met, a factfinder may infer that it is improbable that random chance can explain defendant’s DNA being present at all four locations where a murdered prostitute was found. Although that fact may be inferred without relying on character-based reasoning, that fact is not relevant to the charged crime without additional inferences connecting the use of the doctrine of chances to the charged crime. Thus, whether the other crime scene evi- dence is relevant to the charged crime and does not depend on character-based reasoning depends on the remaining links in the chain of logical relevance. In other words, the doctrine of chances does not imbue the evidence with logical relevancy; relevancy “ ‘is not an inherent characteristic of any item of evidence but exists only as a relation between an item of evidence and a matter properly provable in the case.’ ” State v. Guzek, 322 Or 245, 251, 906 P2d 272 (1995) (quoting Legislative Commentary to OEC 401, reprinted in Laird C. Kirkpatrick, Oregon Evidence 104 (2d ed 1989)). Thus, although the doctrine may function to estab- lish one link in a larger chain of inferences that makes other-acts evidence relevant to a particular fact of conse- quence at trial, we reject the notion that, once the proponent shows that the doctrine of chances properly applies to other- acts evidence, that application of the doctrine necessarily makes the evidence logically relevant. Instead, whether “other acts” evidence is relevant without implicating prohib- ited character-based reasoning will depend on the facts of each case and the proponent’s theory of relevance.6 E. Whether the State Identified a Chain of Inferences that Established Relevance and Admissibility of the Evidence from All Four Murders in Accordance with OEC 404(3) Skillicorn was decided after the initial briefing and oral argument was complete in this case, and we asked the 6 The concurrence maintains that we mischaracterize the state’s argument as asserting, once the doctrine of chances is implicated, it makes evidence admis- sible for all purposes at trial. See 368 Or at 737 (Garrett, J., concurring). To be clear, we understand the state to argue that use of the doctrine of chances should not be limited to proving intentionality categorically, as we previously main- tained in Leistiko. Instead, the state argues that the doctrine of chances may be used to prove any type of noncharacter fact. In view of our holding in this case, we need not resolve whether use of the doctrine of chances should be categorically limited to proving certain types of facts. 730 State v. Jackson

parties to submit supplemental briefing to address its effect on this case. In view of this court’s reexamination of the doctrine of chances and reemphasis on the import of the pro- ponent’s theory of relevance in Skillicorn, the state lays out in supplemental briefing the chain of inferences that it con- tends (1) establishes that the evidence from all four crime scenes is relevant in each trial to prove that defendant was the person who committed the murder of each victim and (2) avoids character reasoning. Ultimately, we conclude that the state’s additional effort to establish the admissibility of the evidence fails. The state articulates the following chain of rele- vance for the admissibility of the DNA evidence from the three other murder scenes in any one trial for a charged murder: First, the jury can use the evidence to decide that it is statistically improbable for a defendant’s DNA to be at four crime scenes involving murdered prostitutes by random chance, which is an application of the doctrine of chances. As we have discussed, the statistical inference about a group of events through use of the doctrine of chances does not com- pel an additional inference about whether one event could be random chance; therefore, the proponent of the evidence must identify additional intermediate inferences connect- ing the use of the doctrine of chances to the charged act. Here, the state posits for its second and intermediate infer- ence that, because of the statistical improbability of chance explaining defendant’s DNA at all the crime scenes, defen- dant was likely present at each of the crime scenes when the murders were committed, including the charged murder. Third, the state asserts, the second inference that defendant was present for each of the murders “supports the ultimate reasonable inference that defendant was the killer in each murder,” including the charged murder. The state maintains that all three inferences “are based solely on the jury’s probability assessment” con- cerning defendant’s DNA at the four murder scenes, not on character-based reasoning. On its face, that chain of inferences does not rely on obvious propensity reasoning to explain relevance, such as asking the jury to conclude that, because defendant killed three other women, he is the type of person who kills women like the victim in the charged Cite as 368 Or 705 (2021) 731

crime. Thus, the state articulates a chain of inferences that avoids an explicit character-based theory of relevance. We first address and reject the state’s contention that it has established the logical relevance of the evidence from the other crime scenes based solely on the improbabil- ity that defendant’s DNA would be at all four crime scenes by random chance. The state’s first inference depends on the doctrine of chances: the improbability that random chance can explain defendant’s DNA at all four murders. But that fact does not directly increase the likelihood of the state’s second factual inference: that random chance cannot explain defendant’s DNA at any one of the four crime scenes. As we have explained, the doctrine of chances operates at the level of the entire group of events, and not the level of the indi- vidual events within the group. Ruling out random chance as the explanation for all four events in the group does not rule out or even diminish random chance as the explana- tion for any one of the four events. Each one of the charged acts is only one event in the group. At most, the doctrine of chances allows a factfinder to infer that random chance can- not explain one or some events in the group, without identi- fying which events can be explained by random chance and which cannot. The difficulty for the state, then, is the task of explaining the move from the first inference about the series of events to the subsequent inferences about only one of the charged crimes within the series when the doctrine of chances, properly understood, does not accomplish the task. What basis besides the doctrine of chances would justify the state’s proposed inference that defendant was present for each murder, including the charged crime? The state makes no attempt to answer that question other than by asserting that the subsequent inferences are “reasonable” in view of the first. Having rejected the doctrine of chances as the sole basis that connects the factual proposition derived by using the doctrine—that defendant was likely present at the time of one or some of the murders—to the second link in the state’s chain of inferences—that defendant was likely pres- ent at the time of any particular one of those murders, we 732 State v. Jackson

examine the state’s claim that it is reasonable to make that connection. By framing the key second factual inference in the state’s chain of inferences as being about defendant’s likely presence at each murder, rather than framing the fac- tual inference as being about defendant having likely com- mitted each murder, the state attempts to avoid character- based reasoning. However, we conclude that the state, while disclaiming use of character-based reasoning, requires such reasoning to connect the inferences. Considering the use of the evidence during defen- dant’s upcoming trial on charges relating to one of the mur- ders makes that conclusion evident. In a trial involving LT’s murder, for example, whether evidence of defendant’s DNA at three other crime scenes involving TH, AA, and LW is rel- evant to prove that defendant killed LT depends on whether the factfinder believes that defendant killed the other vic- tims. As the trial court put it, the other crime scene “evi- dence only has persuasive force if it means that the defen- dant committed the charged acts.” If a factfinder concluded that random chance explained defendant’s DNA being at the other three crime scenes, the presence of defendant’s DNA at those crime scenes would not matter for the purpose for which the state seeks to introduce the other-acts evidence, to prove that defendant was present at the time that LT was murdered and that he was the murderer. And because the persuasive force of the other crime scene evidence is based on establishing defendant’s character as a murderer, the state’s theory of relevance is barred by OEC 404(3). Looking at the issue from a different direction, the state already has acknowledged, correctly, that were it to claim that defendant killed three women before in an attempt to prove that defendant killed again and murdered LT, that would constitute character-based evidence that is barred under OEC 404(3). Thus, for example, if the state had defendant’s confession that he had killed TH, AA, and LW, it could not use the confession to prove that defendant murdered LT. The state in this case has attempted to intro- duce evidence extrinsic to the charged murder to establish a circumstantial case that defendant probably killed the three other women, but that evidence is no more admissible under OEC 404(3) in his trial for LT’s murder than it would Cite as 368 Or 705 (2021) 733

be if the state had offered direct evidence by confession that defendant actually killed the three other women. In sum, an Oregon court that is asked to admit other-acts evidence cannot simply look for the proponent’s identification of a noncharacter material fact that permits use of other acts as proof, such as the actor’s “motive, oppor- tunity, intent, preparation, plan, knowledge, identity,” OEC 404(3), and for some probative value of that evidence that is connected in any way to the identified purpose. Instead, the proponent must articulate the chain of inferences that makes the evidence relevant to that purpose and explain how that chain of inferences does not depend on the actor’s character, and the court must carefully examine that chain of inferences to determine whether the proponent has met its burden to establish that it is offering a theory of rele- vance for the evidence that does not depend on character- based reasoning prohibited under OEC 404(3). In this case, the state’s reliance on the doctrine of chances does not satisfy its burden of offering an intermedi- ate inference that logically connects its use of the doctrine of chances to the fact of consequence it wishes to prove— defendant’s identity as the murderer. Instead, the state has articulated a chain of inferences that ultimately requires the factfinder to depend on character-based reasoning to conclude that the crime scene evidence from three other murders supports the conclusion that defendant committed the murder of a fourth victim. Accordingly, we affirm the trial court’s exclusion of that evidence. The order of the circuit court is affirmed. Garrett, J., concurred and filed an opinion in which Balmer, J., joined. GARRETT, J., concurring. I write separately because, although I agree with the majority’s disposition of this appeal, I view its reasoning as problematic in some respects. First, the majority opinion perpetuates a long- standing misconception in this court’s case law regarding OEC 404(3). Specifically, the majority describes that rule as 734 State v. Jackson

prohibiting “propensity” reasoning. However, OEC 404(3) actually says the following: “Evidence of other crimes, wrongs[,] or acts is not admis- sible to prove the character of a person in order to show that the person acted in conformity therewith. It may, however, be admissible for other purposes, such as proof of motive, opportunity, intent, preparation, plan, knowledge, identity, or absence of mistake or accident.” (Emphasis added.) Notoriously, neither the Oregon Evidence Code nor the federal evidence code (which contains a sub- stantively identical rule, FRE 404(b)) provides a definition of “character.” That omission has spawned a small library of case law and commentary on the topic of what, exactly, the rule prohibits. “Legal scholars and courts have long struggled to define character in the law of evidence. Since the time of John Henry Wigmore, observers have lamented that ‘[t]he prohibition against “character evidence” is one of the great enigmas [of] the law of evidence.’ ” Barrett J. Anderson, Note, Recognizing Character: A New Perspective on Character Evidence, 121 Yale LJ 1912, 1919 (2012) (quot- ing John Henry Wigmore, 1A Evidence in Trials at Common Law § 54.1, 1150 (Peter Tillers ed., 1983)). This court has defined “character” as a person’s “tendency to act in a certain way in all [the] varying situ- ations of life,” State v. Marshall, 312 Or 367, 372, 823 P2d 961 (1991) (internal quotation marks omitted), or as a per- son’s “disposition or propensity to commit certain crimes, wrongs, or acts,” State v. Johns, 301 Or 535, 548, 725 P2d 312 (1986), overruled in part by State v. Skillicorn, 367 Or 464, 479 P3d 254 (2021).1 The same or similar definitions are used by other courts, but they are of dubious value. See Anderson, 121 Yale LJ at 1922 (noting the use of such defi- nitions in numerous jurisdictions: “At best, these definitions are too general, confusing, and vague. They do not distin- guish, for example, between a character trait on one hand,

1 Skillicorn overruled Johns “to the extent that it holds that evidence of uncharged misconduct can be admitted under the doctrine of chances for the pur- pose of arguing that, because the defendant engaged in deliberate conduct before, it is likely that he engaged in it again during the charged incident.” Skillicorn, 367 Or at 493. Skillicorn did not call the definition of “character evidence” in Johns into question. Cite as 368 Or 705 (2021) 735

and a person’s habit, mental disorder, or sexuality on the other.”). In addition, this court—again, like many others— has used the word “character” interchangeably with “pro- pensity.” The two terms have been equated to the point that we have routinely described OEC 404(3) as prohibiting “propensity evidence,” even though that is not what the rule says. For example, we have observed: “OEC 404(3) is an ‘inclusionary’ rule, as opposed to an ‘exclusionary’ rule, expressly stating that prior bad acts evidence may be admissible as long as it is relevant for any purpose other than to prove ‘propensity’—that is, to prove the character of a person, to show that the person acted in conformity with that character.” State v. Turnidge (S059155), 359 Or 364, 429, 374 P3d 853 (2016). Curiously, that passage from Turnidge (S059155), in describing what OEC 404(3) “expressly” states, puts quota- tion marks around the wrong word (propensity instead of character). More recently, in Skillicorn, we took the same approach: “ ‘ “Character” for purposes of evidence law means a per- son’s disposition or propensity to engage or not to engage in certain types of behavior.’ Thus, OEC 404(3) prohibits the use of uncharged misconduct evidence to prove that a per- son has a propensity to engage in certain types of behavior and that the person acted in conformance with that pro- pensity on a particular occasion. In short, it prohibits ‘pro- pensity evidence.’ ” 367 Or at 475-76 (quoting Laird C. Kirkpatrick, Oregon Evidence § 404.03, 213 (7th ed 2020)). Although the conflation of character with propen- sity is enmeshed in our jurisprudence, commentators have pointed out a problem: The concepts are not the same. Moreover, the case law clearly reflects that they are not the same, even while it has uncritically substituted one word for the other. See, e.g., Anderson, 121 Yale LJ at 1915-16 (“Defining character as simply someone’s propensity to act in a certain way does not distinguish between what is com- monly perceived as character and other propensity-based qualities that courts have recognized are not character, such 736 State v. Jackson

as habits, mental illnesses and genetic attributes, skills and abilities, or other traits of personality.” (Footnotes omitted.)); id. at 1915-16 nn 10-12 (collecting cases). “Character” is nor- mally understood to refer to more generalized traits, usu- ally with a moral quality, while “propensity” can mean par- ticularized tendencies that may or may not have any moral component. See, e.g., David P. Leonard, The New Wigmore: Evidence of Other Misconduct and Similar Events § 8.3, 552- 53 (2d ed 2019) (“Character is thought to be a generalized tendency to act in a particular way, caused by something internal to the actor that arises from that person’s moral bearing.”). As another commentator has explained: “Courts and commentators often carelessly say that it is ‘propensity reasoning’ that is banned by the standard evidence rule. * * * That is an overstatement; if the law is properly read, it is * * * only reasoning based on the propen- sity known as character that is banned. “ ‘Character’ properly describes just one kind of propen- sity—a propensity to repeat a general category of act, such as acts of violence, acts of dishonesty, etc. And by definition, character means a propensity to do acts that have a good or bad moral connotation. But there are other kinds of pro- pensities. Inferring present action from these others is not banned. For example, there are specific propensities, i.e. a tendency to do a certain thing in a specific way. This is not properly referred to as ‘character.’ ” Paul F. Rothstein, Comment: The Doctrine of Chances, Brides of the Bath and a Reply to Sean Sullivan, 14 Law, Probability & Risk 51, 61 (2015) (emphasis in original). “Modus operandi” evidence, for example, is gener- ally understood to be admissible to prove “identity” (a pur- pose expressly allowed by OEC 404(3)), on the theory that, where the charged conduct bears sufficiently unique charac- teristics, evidence that the defendant previously engaged in conduct with those same characteristics supports an infer- ence that the defendant is the person who committed the charged conduct. That logic involves what can only be a form of propensity reasoning—the idea that the accused has a habit, inclination, or proclivity for doing a very specific thing in a very specific way. Thus, if all “propensity” reasoning is banned by OEC 404(3), “modus operandi” evidence cannot Cite as 368 Or 705 (2021) 737

be legitimate. But it is legitimate, as courts have tradition- ally understood, including this one. See, e.g., State v. Pinnell, 311 Or 98, 110 & n 18, 806 P2d 110 (1991) (describing per- missible use of other acts evidence to show that the accused committed a “signature crime,” such as by leaving “the mark of Zorro”). The only way to resolve that contradiction is to conclude that OEC 404(3) means just what it says— it prohibits character reasoning, not propensity reasoning, which is different, often more specific, less value-laden, and nowhere mentioned in the rule. In short, this court should fundamentally recon- sider the nature of the OEC 404(3) prohibition. However, disentangling the concepts of character and propensity is a difficult task, and not one that the state has asked us to undertake here. Instead, the state argues that its theory of relevance does not require even propensity reasoning. Considering that argument on its own terms, I believe it fails, but for reasons more straightforward than the majority’s analysis. The majority characterizes this case as raising such general questions as “the extent to which the doctrine of chances represents a noncharacter theory of relevance for evidence of other acts extrinsic to the charged act,” 368 Or at 720, and whether the doctrine of chances “by itself consti- tute[s] a theory of relevance,” id. at 726. The majority then answers those questions at a very high level of abstraction. Those answers are both categorical and largely ungrounded in authority, posing a risk that the pronouncements in today’s opinion are unintentionally overbroad and may cause confu- sion in future cases. This court’s troubled history with the doctrine of chances—well described by the majority and in Skillicorn—counsels a less ambitious approach. Much of the majority’s discussion seems intended to reject an argument that I do not understand the state to be making: that the doctrine of chances is somehow a free- standing theory that, by itself, “imbue[s]” evidence with rel- evance, 368 Or at 729, such that the evidence becomes rel- evant and admissible to prove any material fact. Although the state has referred to the doctrine of chances as a theory of relevance, in context I understand the state’s position to 738 State v. Jackson

be that, when evidence is logically relevant, the doctrine of chances is a way of explaining why the theory of rele- vance does not rely on propensity reasoning. In this case, the state has argued that the “other acts” evidence is logi- cally relevant to prove defendant’s identity as the murderer (a purpose allowed by OEC 404(3)), and that, under the doc- trine of chances, that theory of relevance relies solely on probabilistic rather than propensity reasoning. But the state’s theory does not avoid propensity reasoning. That can be explained without going further to suggest how the doc- trine of chances might or might not work in other contexts. The state contends that the presence of defendant’s DNA at the other murder scenes supports an inference that the presence of his DNA at the charged murder scene was not a coincidence (more precisely, an inference that defen- dant was present at the murder). However, the inference that defendant was present at the charged murder scene relies on additional, unstated inferences: that (1) instead of coincidence, a better explanation is that the series of events has a common cause, which (2) can only be that defendant had an unusual tendency to be present for the violent deaths of young, Black prostitutes over the time period in question, which (3) increases the likelihood that defendant was pres- ent for the charged crime. Whether or not that chain of rea- soning involves any assumption about defendant’s “charac- ter,” it does involve an assumption about his propensity for engaging in certain behavior. Therefore, if one assumes for purposes of this case that OEC 404(3) prohibits propensity reasoning, the trial court’s ruling should be affirmed. Balmer, J., joins in this concurring opinion.