¶1 Kevin Coe challenges a Court of Appeals decision affirming his 2008 commitment as a sexually violent predator (SVP). He is primarily challenging the trial court’s decision to admit evidence of several unadjudicated sexual offenses. He is also challenging the trial court’s decision to allow an expert psychologist to rely on that evidence. In addition to these challenges, Coe is claiming his trial counsel was ineffective, that his due process confrontation rights were violated, and that he is entitled to *488a new trial under the cumulative error doctrine. Finding no reversible error in any of Coe’s claims, we affirm his commitment.
FACTUAL AND PROCEDURAL HISTORY
¶2 Coe was originally convicted in 1981 of four counts of first degree rape. State v. Coe, 101 Wn.2d 772, 774, 684 P.2d 668 (1984) (Coe I). But those convictions were overturned on appeal primarily because some of the victims identified Coe only after being hypnotized. Id. at 785-86, 788-89. At Coe’s second trial in 1985, Coe was convicted of three counts of first degree rape. State v. Coe, 109 Wn.2d 832, 834, 750 P.2d 208 (1988) (Coe II). Again, Coe appealed and this court reversed two of the three counts because of the “admission of posthypnotic identification testimony.” Id. at 850. Ultimately, Coe’s 1985 conviction for the first degree rape of Julie H. was his sole conviction, and he was sentenced to 25 years.
¶3 On August 30, 2006, the State filed a petition seeking to have Coe committed as an SVP pursuant to chapter 71.09 RCW. During the trial, the State sought to link Coe to 40 unadjudicated sexual offenses. These 40 offenses included both rapes and indecent exposure incidents. The trial court admitted 36 of these offenses after finding by a preponderance of the evidence that Coe was the offender.
¶4 The State proved these offenses through multiple sources. For example, it relied on Dr. Robert Keppel, the State’s “signature analysis” expert, who linked Coe to 18 rapes, including the Julie H. rape. Additionally, it relied on statistical results from the Homicide Investigation Tracking System (HITS) database, which linked Coe to 13 rapes that were admitted at trial. Dr. Keppel corroborated his signature analysis with the HITS results but did not rely on them. The State also had several victims identified by the signature analysis and HITS results testify at the SVP trial.
*489¶5 Further, because the above evidence alone does not prove SVP status, see RCW 71.09.020(18), the State’s psychologist, Dr. Amy Phenix, testified that Coe suffered from the following mental abnormalities: (1) paraphilia, not otherwise specified (NOS), nonconsenting females, with sadistic traits; (2) paraphilia NOS, urophilia, and coprophilia; and (3) exhibitionism. Additionally, she testified that Coe had a personality disorder NOS, with narcissistic and antisocial traits.
¶6 Her opinion was based on her review of over 74,000 pages of material, which included the other offenses. She considered identifications by Coe’s victims, blood typing evidence, and Coe’s own admission to two offenses. She also incorporated the signature analysis and HITS results into her diagnosis.
¶7 On October 15,2008, after a month-long trial, the jury found Coe was an SVP. The trial court then ordered Coe’s civil commitment. Coe appealed, and the Court of Appeals affirmed the trial court. He then petitioned for review, which we granted. In re Det. of Coe, 172 Wn.2d 1001, 258 P.3d 685 (2011).
ISSUES
¶8 1. Did Coe receive ineffective assistance of counsel when his trial counsel failed to offer a jury instruction defining “personality disorder”?
¶9 2. Was the signature identified by Dr. Keppel in his signature analysis sufficiently unique to identify Coe as the offender under ER 404(b)?
¶10 3. Were the HITS results properly admitted by the trial court?
¶11 4. Did the trial court err in allowing seven victims, identified by the signature analysis and HITS evidence, to testify at trial?
¶12 5. During the trial, the State’s expert psychologist relied on several offenses where the victims were unavail*490able to testify, and the expert disclosed those offenses to the jury. Does Coe have a due process right to confront those victims?
¶13 6. May the State’s expert psychologist rely on the signature analysis and HITS results in determining that Coe was an SVP?
¶14 7. May the State’s expert witness, in explaining the basis for her expert conclusions, disclose unadjudicated rapes that were not substantively admitted?
¶15 8. Is Coe entitled to a new trial under the cumulative error doctrine?
ANALYSIS
¶16 Committing an individual as an SVP under chapter 71.09 RCW requires the State to show beyond a reasonable doubt that the individual is an SVP. In re Det. of Post, 170 Wn.2d 302, 309-10, 241 P.3d 1234 (2010). An SVP is an individual “who has been convicted of or charged with a crime of sexual violence and who suffers from a mental abnormality or personality disorder which makes the person likely to engage in predatory acts of sexual violence if not confined in a secure facility.” RCW 71.09.020(18) (emphasis added).
¶17 We address each of Coe’s challenges to his commitment in turn.
1. Ineffective Assistance of Counsel
¶18 Coe first claims his trial counsel was ineffective for failing to offer a jury instruction defining “personality disorder.” To establish ineffective assistance of counsel, a party must prove (1) that counsel’s performance was deficient and (2) that it resulted in prejudice. State v. Grier, 171 Wn.2d 17, 32-34, 246 P.3d 1260 (2011). To establish deficiency, an appellant “must overcome ‘a strong presumption that counsel’s performance was reasonable’ ” in order to prove that performance fell “ ‘below an objective standard of *491reasonableness.’ ” Id. at 33 (quoting State v. Kyllo, 166 Wn.2d 856, 862, 215 P.3d 177 (2009); Strickland v. Washington, 466 U.S. 668, 688, 104 S. Ct. 2052, 80 L. Ed. 2d 674 (1984)).
¶19 Here, Coe’s ineffective assistance of counsel claim fails because he cannot show that his counsel’s performance was deficient. At the time of his trial, then-controlling authority stated that defining “personality disorder” was unnecessary. See In re Det. of Twining, 77 Wn. App. 882, 895-96, 894 P.2d 1331 (1995), overruled by In re Det. of Pouncy, 168 Wn.2d 382, 391, 229 P.3d 678 (2010). It was not until more recently, after Coe’s trial, that we overruled Twining and held that a trial court’s failure to instruct the jury on the term “personality disorder” was error. Pouncy, 168 Wn.2d at 385. Pouncy, however, does not affect the analysis as we make “ ‘every effort ... to eliminate the distorting effects of hindsight, to reconstruct the circumstances of counsel’s challenged conduct, and to evaluate the conduct from counsel’s perspective at the time.’ ” Grier, 171 Wn.2d at 34 (quoting Strickland, 466 U.S. at 689).
¶20 Thus, Coe’s counsel had no reason to believe, at the time of trial, that he should have requested a jury instruction for the term “personality disorder.” It is difficult to imagine exactly how Coe’s counsel was deficient when then-controlling authority stated an instruction was not necessary. Consequently, Coe fails to establish deficient performance and his ineffective assistance of counsel claim necessarily fails.
2. Admitting Dr. Keppel’s Signature Analysis
¶21 Next, Coe challenges the trial court’s decision to admit Dr. Keppel’s signature analysis under ER 404(b). Coe contends that the signature identified by Dr. Keppel was too common and that the identified signature was not fully present in several of the rapes. We disagree. Certainly, the decision to admit Dr. Keppel’s signature analysis is debat*492able, but under the unique facts of this case, namely the expert opinion about the signature’s uniqueness, we cannot say the trial court abused its discretion.
¶22 We review the trial court’s decision to admit the evidence for abuse of discretion. State v. Foxhoven, 161 Wn.2d 168, 174, 163 P.3d 786 (2007). “ ‘Discretion is abused if it is exercised on untenable grounds or for untenable reasons.’ ” Id. (quoting State v. Vy Thang, 145 Wn.2d 630, 642, 41 P.3d 1159 (2002)).
a. Factual Basis for the Signature Analysis
¶23 Dr. Keppel’s signature analysis identified a combination of ritualistic behaviors and modus operandi that allowed him to identify Coe as the perpetrator of 18 rapes. According to Dr. Keppel, ritual behaviors are those behaviors that are unnecessary to complete a crime but express the offender’s underlying motivation. In contrast, modus operandi includes all the behaviors necessary to complete the actual crime.1
¶24 Dr. Keppel identified the following five behaviors as Coe’s ritualistic signature: “1) intimidation, 2) co-opting the victim into compliance, 3) the rapist undoing his own clothing, 4) the necessity of sexual intercourse and/or ejaculation, and 5) the need for questioning and engaging in conversation of the victim.” Clerk’s Papers (CP) at 4429. As for modus operandi, Dr. Keppel identified the following: (1) the assaults occurred in the South Hill area of Spokane, (2) the assaults were during the early morning hours or late at night, (3) the victims’ ages varied from 14 to 51 years old, and (4) the victims were either walking or jogging before being attacked.
¶25 To create the signature, Dr. Keppel first analyzed Julie H.’s rape. At this time, Dr. Keppel specifically did not examine any other rapes Coe was suspected of committing because he did not want to bias his analysis. Coe’s signature *493in Julie H.’s case (October 1980) appeared basically as follows: (1) surprised victim, grabbed from behind and placed gloved fingers in her mouth, threw the victim to the ground, used aggressive language, threatened her with an unseen knife, and threatened to kill her if she called the police; (2) told victim to take off her clothing; (3) rapist unzipped his own pants; (4) sexual acts included masturbation, fondling, vaginal penetration, and ejaculation; and (5) rapist talked continuously and asked victim about her sex experiences.
¶26 After identifying a signature, Dr. Keppel analyzed the 5 other rapes that Coe was initially charged with. He then applied that signature to 54 other offenses, which led to the other 12 rapes, for a total of 18 rapes.
b. Admissibility of the Signature Analysis
¶27 Under ER 404(b), a trial court can admit evidence of other crimes or wrongs only if it “ ‘(1) find[s] by a preponderance of the evidence that the misconduct occurred, (2) identif [ies] the purpose for which the evidence is sought to be introduced, (3) determine [s] whether the evidence is relevant to prove an element of the crime charged, and (4) weights] the probative value against the prejudicial effect.’ ” Foxhoven, 161 Wn.2d at 175 (quoting Thang, 145 Wn.2d at 642). Here, Coe is challenging only the third prong: whether the signature is unique enough to identify him as the offender such that the crimes are relevant.
¶28 A signature must be so unique that commission of one crime creates a high probability that the accused also committed the other crimes. Id. at 176. This uniqueness is more than a mere similarity; it must be like a signature. Id. A court must “look at all the circumstances of a crime and the perpetrator’s acts to try to distinguish a type of signature.” Id. at 179. This includes both the similarities and dissimilarities. See id. at 176, 178-79.
¶29 Courts look to various factors when determining if a signature exists. See Thang, 145 Wn.2d at 643-44. Such *494factors can include geographical and temporal proximity. Id. at 643 (citing State v. Russell, 125 Wn.2d 24, 68, 882 P.2d 747 (1994)). Essentially, a court examines the method in which the crime is committed and determines if the method is sufficiently unique. See, e.g., id. at 643-45; Russell, 125 Wn.2d at 68. This does not require each feature to be unique. Thang, 145 Wn.2d at 644. Seemingly common features can combine to create a unique signature, “especially when combined with a lack of dissimilarities.” Id. (citing State v. Jenkins, 53 Wn. App. 228, 237, 766 P.2d 499 (1989)).
¶30 For example, in Jenkins, the following similarities between two burglaries were sufficient for admission: (1) offender used a pipe wrench to open the door, (2) offender drove a brown Camaro, (3) offender burglarized multiapartment complexes with a partner, and (4) offender entered only ground floor units. 53 Wn. App. at 237. These seemingly common features, coupled with lack of dissimilarities, were enough for the court to conclude that it was highly probable the crimes were committed by the same person. Id. When there are too many dissimilarities, the crime should be excluded. Thang, 145 Wn.2d at 645. In Thang, the two burglaries at issue occurred 18 months apart in completely different parts of the state. Id. The offender in each crime stole similar items from elderly victims in both crimes, but the offenders entered and exited the homes in different manners. Id. The offender in each crime also fled the crimes differently, one on foot and one by car. Id. Moreover, one victim was killed by repeated kicking while the other victim was kicked only three times. Id 2 Considering these dissimilarities, the trial court abused its discretion when it admitted the evidence. Id.
¶31 In this case, Dr. Keppel identified five ritualistic behaviors as Coe’s signature: (1) intimidating the victim, (2) *495co-opting victim compliance, (3) removing his own clothing, (4) needing intercourse and/or ejaculation, and (5) engaging in conversation. Coe challenges Dr. Keppel’s reliance on these behaviors, stating that “four of [the behaviors] are ordinary incidents of rape.” Opening Br. of Appellant at 25. Coe is correct that several of the behaviors appear exceedingly common, but the signature identified by Dr. Keppel is more distinct than these five general labels indicate. The distinctiveness of Coe’s signature becomes apparent after examining each of the 17 other rapes as Dr. Keppel did in his report. The Court of Appeals summarized Dr. Keppel’s findings as follows:
Jean C. (4/1978): (1) surprised victim, threatened to hurt her if she screamed, put gloved hand in her mouth, used aggressive and offensive words; (2) low level of violence; (3) rapist unzipped his own pants; (4) vaginal penetration and ejaculation; (5) asked victim if she could urinate or defecate, asked personal questions, described sex acts.
Shelly H. (9/79): (1) surprised victim, tried to put hand in her mouth but she bit him, knocked her to ground and hit her, told her he was going to degrade her, told her to perform fellatio right or he would kill her, told her he knew where she worked and lived and would return if she resisted; (2) told victim to take off her clothing; (3) rapist removed his own clothes; (4) masturbation, partial penetration with ejaculation; (5) asked victim to urinate on him, asked about boyfriend and masturbation, talked about victim’s career.
Paige K. (12/79): (1) grabbed victim and pushed thumb down her throat, knocked her to ground, told her he had a knife and would hurt her if she screamed; (2) low level of violence; (3) no record whether rapist removed his clothes; (4) masturbation, vaginal penetration, external ejaculation; (5) rapist talked continuously, asked victim about her sex experiences and personal details.
Joanne T. (12/79): (1) grabbed victim by covering her mouth, pushed hand in her mouth to prevent screaming, threatened her and told her he would get a knife and come back; (2) low level of violence, told victim to undress; (3) rapist took off his *496own clothes; (4) masturbation, external ejaculation; (5) asked personal questions and used offensive language.
Dorcas T. (12/79): (1) threatened victim with an unseen knife if she screamed, used aggressive and offensive words; (2) low level of violence; (3) no record whether rapist removed his own clothes; (4) masturbation, attempted penetration, external ejaculation; (5) asked personal and offensive questions.
Darria L. (2/80): (1) rapist confronted victim in street while displaying large dildo and said he needed sex, threatened her; (2) told victim to remove her clothes; (3) rapist removed his own clothes; (4) masturbation, attempted penetration, external ejaculation; (5) asked personal and offensive questions.
Mary L. (3/80): (1) grabbed victim from behind, threatened her with unseen knife, used aggressive and offensive words; (2) told victim to remove her clothes, low level of violence; (3) rapist removed his own clothes; (4) penetration with ejaculation; (5) used aggressive and offensive language and asked personal questions.
Margaret D. (4/80): (1) grabbed victim from behind, threatened her with unseen knife, threw her to ground, used aggressive and offensive words, said he knew where she lived; (2) told victim to remove her clothes, low level of violence; (3) rapist removed his own clothes; (4) penetration with ejaculation; (5) asked personal and offensive questions.
[Elizabjeth A. (5/80): (1) grabbed victim by the neck and dragged her, exhibited a knife, used aggressive and offensive words, threatened to kill the victim if she did not shut up; (2) told victim to comply and she would not get hurt; (3) no record whether rapist removed his own clothes; (4) masturbation, penetration, ejaculation; (5) conversation, including offensive sexual statements.
Teresa K. (6/80): (1) grabbed victim by the neck and threw her to ground, forced fingers in victim’s mouth and choked her, threatened her with unseen knife, told her he knew where she lived and would kill her if she reported the rape; (2) told victim to remove her clothes, low level of violence; (3) rapist removed his own clothes; (4) penetration and ejaculation; (5) asked personal questions.
Sherry J. (7/80): (1) grabbed victim from behind and threw her to ground, threatened her with unseen knife, used aggres*497sive and offensive words such as “dump a load,” threatened to return and kill victim if reported; (2) told victim to remove her clothes, low level of violence; (3) no record whether rapist removed his own clothes; (4) penetration and ejaculation; (5) asked personal and offensive questions.
Gretchen C. (8/80): (1) grabbed victim by the throat and threw her to ground, threatened with unseen knife, threatened to kill victim if she looked at him, used aggressive and offensive language, including asking victim to urinate and defecate; (2) told victim to remove her clothes, low level of violence; (3) no record whether rapist removed his own clothes; (4) penetration; (5) asked personal questions.
Sherry S. (8/80): (1) grabbed victim around her back, forced his fingers down her throat and warned her not to bite, threatened her with unseen knife, used aggressive and offensive words, told her he knew where she lived; (2) told victim to remove her clothes, low level of violence; (3) no record whether rapist removed his own clothes; (4) penetration and ejaculation; (5) asked personal questions and made offensive statements.
Jennifer C. (11/80): (1) grabbed victim around the neck, threatened her with unseen knife, threw her to ground, used aggressive and offensive words; (2) told victim to remove her clothes, low level of violence; (3) no record whether rapist removed his own clothes; (4) masturbation, external ejaculation, vaginal penetration with fingers; (5) asked personal and offensive questions.
Cheri H. (12/80): (1) grabbed victim from behind, forced fingers down her throat, threatened her with unseen knife, warned her not to call police; (2) told victim to walk to wooded area and to take off clothes, low level of violence; (3) rapist unzipped his own pants; (4) penetration with ejaculation; (5) asked personal and offensive questions.
Mary S. (2/81): (1) grabbed victim and pulled to ground, tried to force fingers into her mouth, threatened with unseen knife, covered her face with his hands and her hair, hit victim when she struggled; (2) told victim she would not be hurt if she did what he said, relatively low level of violence; (3) no record whether rapist removed his own clothes; (4) penetrated vagina with fingers, external ejaculation; (5) personal and offensive questions and statements.
*498Diane F. (2/81): (1) grabbed victim by the mouth, forced his fingers into her mouth, threatened her with unseen knife, used aggressive and offensive words; (2) told victim to remove her clothes, low level of violence; (3) rapist unzipped his own pants; (4) masturbation and penetration; (5) asked personal and offensive questions.
In re Det. of Coe, 160 Wn. App. 809, 821-24, 250 P.3d 1056 (2011).
¶32 This case is more like Jenkins than Thang. Importantly, and unlike in Thang, an expert witness identified the signature here. Allowing experts to testify about how certain behaviors can combine to form a unique signature has been previously approved by various appellate courts. See, e.g., United States v. Trenkler, 61 F.3d 45, 54 (1st Cir. 1995) (admitting expert testimony about how two bombs were similar); Russell, 125 Wn.2d at 69-70 (allowing expert to testify about whether certain behaviors were common or rare); State v. Fortin, 189 N.J. 579, 599 n.11, 917 A.2d 746 (2007) (noting that facial beatings, anal injuries, and manual strangulation are seemingly common to sexual assault and that the State had not provided an expert who testified to the contrary).
¶33 Here, Dr. Keppel’s opinion was based on his years of experience dealing with over 2,000 sex offenses. He examined over 50 rapes using a “linkage analysis” to identify 18 rapes that exhibited the ritualistic signature.3 Having expert testimony that the signature is unique distinguishes this case from Thang because it creates an identifiable reason for qualifying the signature as unique.
¶34 Coe contends that even if the signature was unique, not all five behaviors were present in each rape. Dr. Keppel readily admitted that a behavior may be missing from some *499of the rapes.4 He maintained that although one or two behaviors might be missing from any given rape, the strength of the remaining behaviors sufficiently demonstrated the presence of Coe’s signature. Further, he explained that a factor could be missing merely because the police report failed to mention it but that did not mean that the factor was not present. Given the strength of Dr. Keppel’s explanation and expert opinion that the signature is still present, we are not persuaded by Coe’s argument.
¶35 Additionally, the existence of some dissimilarities in the crimes is not dispositive. See Russell, 125 Wn.2d at 30-36, 67-68. In Russell, the trial court cross-admitted evidence of three murders where the victims were posed nude with the aid of props after being violently murdered and sexually assaulted. Id. at 67-68. There were significant differences in the victims’ posed positions, in the props that were used, and in the location of the bodies. See id. at 72, 30-36. But because posing victims with props after sexually assaulting them was sufficiently unique, the court upheld admitting the evidence. Id. at 72-73. Similarly, the mere existence of some dissimilarities between the 18 rapes does not mandate exclusion when Dr. Keppel has testified as to the signature’s uniqueness.
¶36 Here, the dissimilarities between rapes include (1) victims’ ages varied from 14 to 51 years old, (2) the victims provided varying descriptions of the rapist’s physical appearance, (3) some victims were jogging while others were walking or getting off the bus, and (4) rapist requested money sometimes. But some dissimilarities are expected considering the sheer number of crimes and the fact that ritualistic behaviors can evolve over time. CP at 4416. The varying physical descriptions are explained by the fact that Coe’s physical appearance did vary. He gained and lost *500weight due to occasional fasting. Also, Dr. Keppel testified that ritual behaviors can “express themselves differently over a series of offenses either due to the refinement and more complete reflection of their underlying intent or fantasy substrate, or through an addition of unexpectedly arousing aspects of a prior offense.” Id.
¶37 Coe also attacks the distinctiveness of the signature, but even if there is “a slight deficit in distinctiveness [that] can be overcome by the sheer number of crimes.” David P. Leonard, The New Wigmore: Evidence of Other Misconduct and Similar Events § 13.7.2, at 737 (2009) (citing United States v. Levi, 310 U.S. App. D.C. 152, 45 F.3d 453 (1995)). Attributing some significance to the number of crimes also makes sense considering violent sexual assaults in such a limited geographic area are not extremely common. Cf. Trenkler, 61 F.3d at 55 (attributing significance to the fact that both crimes involved bombings, which are in and of themselves distinctive crimes). The signature here, like the signature in Jenkins, is present in enough crimes that the jury is entitled to believe the crimes are linked.
¶38 Considering the unique signature that the combination of the five ritual behaviors creates, along with the modus operandi evidence, there was sufficient evidence for the trial court to admit the signature analysis. The mere existence of dissimilarities between the crimes is not dispositive. See Russell, 125 Wn.2d at 67-68; Foxhoven, 161 Wn.2d at 176.
¶39 The Court of Appeals reached the same conclusion but committed two errors in doing so. We address these errors to dispel any confusion that may exist in proving identity under ER 404(b). First, the Court of Appeals claimed that all dissimilarities go to weight and not admissibility under ER 404(b). Coe, 160 Wn. App. at 825 (citing Foxhoven, 161 Wn.2d at 178-79). That legal principle, however, does not apply in this case; it applies only in cases, like Foxhoven, where the offender literally signs or marks his name. See 161 Wn.2d at 179. Absent a literal signature, the court “look[s] at *501all the circumstances of a crime and the perpetrator’s acts to try to distinguish a type of signature.” Id.
¶40 Second, the Court of Appeals erroneously stated that a lower level of similarity is required “when crimes share a ritualistic quality.” Coe, 160 Wn. App. at 825 (citing State v. Fualaau, 155 Wn. App. 347, 357, 228 P.3d 771, review denied, 169 Wn.2d 1023 (2010), and cert. denied, 131 S. Ct. 1786 (2011)). Fualaau cites Russell as the progenitor of this legal principle, 155 Wn. App. at 357-58 (citing Russell, 125 Wn.2d at 67-68), but Russell states no such thing. Rather, Russell used an ordinary identity analysis comparing all the circumstances of the crimes to find a unique signature. Russell, 125 Wn.2d at 67-68. Fualaau also fails to identify any reason independent from Russell for treating ritual signatures differently from ordinary signatures. Thus, the Court of Appeals erred in stating a lower level of similarity was required for the 17 rapes simply because the signature was based on ritualistic factors.
¶41 Regardless, these errors are not present in the trial court’s reasoning. Ultimately, the trial court’s decision to admit the evidence may be debatable, but the trial court did not abuse its discretion. We therefore accept the evidence of the expert who has “many years of experience in evaluating the similarities of crimes and determining whether they should be ascribed to a single person.” Leonard, supra, § 13.8.2, at 764. We affirm the trial court’s decision to admit the signature analysis notwithstanding the Court of Appeals’ errors.
3. HITS Evidence
¶42 The trial court admitted the HITS results into evidence over Coe’s objections. The HITS database is an investigatory tool that allows law enforcement agencies to search using signature details for similar crimes in a specific area. The HITS database contains two separate databases: one for sexual assaults and one for homicides. *502The HITS sexual assault database contains information on 8,100 sexual assaults. The information is gathered by investigators who typically contact law enforcement agencies across the state. About 70 rape cases are entered each year. This number does not reflect the total number of rapes that occur. The database also contains information from many other states, but the investigators do not actively seek it out. Instead, the investigators enter cases from other states only when an agency requests a case be entered. Despite these outside submissions, 90 percent are from Washington.
¶43 In Coe’s case, the State’s HITS expert, Tamara Matheny, searched the database initially using seven criteria supplied by an assistant attorney general. These seven criteria were (1) offender was white, (2) offender was male, (3) offender was a stranger to the victim, (4) the initial contact site was outdoors, (5) the initial contact site was the site of the assault, (6) force was immediately used on the victim, and (7) offender questioned the victim either moderately or excessively. Matheny’s search returned only 26 cases out of the 8,100, which represented 0.30 percent of the database. She then narrowed the results further by adding the additional criteria: (8) a weapon was used, and (9) that weapon was a cutting or stabbing weapon. This reduced the results to 16 cases or 0.19 percent of the database. Finally, Matheny added one last criterion: (10) use of the weapon was implied. Matheny’s query now returned 14 cases, which represented 0.17 percent of the database. The trial court ultimately admitted 13 of these cases.
¶44 Coe raises three arguments against admitting these 13 cases, arguing that (1) the criteria were insufficiently unique to be admitted under ER 404(b), (2) the HITS database contained inadmissible hearsay, and (3) HITS is otherwise unreliable and misleading. Like the signature analysis above, this claim is reviewed for abuse of discretion because it involves the proper admission of evidence. Foxhoven, 161 Wn.2d at 174.
*503 a. Unique Signature
¶45 Coe first challenges whether the search criteria were sufficiently unique to create a signature under ER 404(b). The Court of Appeals dismissed this argument, stating that common features can create a signature when combined. Coe, 160 Wn. App. at 828-29. This is correct, but additional analysis is necessary.
¶46 As stated above, evidence of other crimes is admissible if a unique signature exists such that it marks it as the handiwork of the accused. Foxhoven, 161 Wn.2d at 176. Whether a signature exists depends on all the circumstances surrounding the crime, including dissimilarities. Id. at 176, 178-79. The main concern, as evidenced by every case dealing with identity, is that the signature be unique. See, e.g., id. at 176; Thang, 145 Wn.2d at 643.
¶47 The 10 features used as search criteria in this case meet this exacting standard. The fact that the HITS search produced only 14 cases out of over 8,000 possible cases is evidence of the features’ uniqueness.5 Granted, the search criteria seem exceedingly common when taken individually, but, as noted in Thang, common features can combine to create a unique signature if the dissimilarities do not indicate otherwise.6 In this case, Matheny did not consider any dissimilarities when she conducted the search. The Court of Appeals’ failure to note this necessary step was error. But the trial court did consider dissimilarities when it reviewed each crime for admission. See CP at 890-96. Thus, the trial court did not abuse its discretion in concluding that the 10 criteria identify a sufficiently unique signature.
*504 b. Hearsay
¶48 Coe next argues that the HITS results could not be substantively admitted because they were based on inadmissible hearsay. The State counters that Coe failed to preserve the hearsay argument. ER 103(a)(1) requires an objection to state the specific grounds for objection unless it is readily apparent from the circumstances. Here, during a pretrial motion to exclude the HITS evidence, Coe argued that “the HITS . .. database contains information based on multiple layers of hearsay.” CP at 3999. He clearly raised the hearsay argument and gave the trial court an opportunity to address it, thus preserving the argument.
¶49 Because the argument was preserved, we must determine whether the HITS evidence is hearsay. Hearsay is an out-of-court statement offered for the truth of the matter asserted. ER 801(c). While we have never specifically addressed whether the HITS database is admissible, other jurisdictions have denied admission of similar databases, reasoning they contain inadmissible hearsay. People v. Hernandez, 55 Cal. App. 4th 225, 240-41, 63 Cal. Rptr. 2d 769 (1997) (database contained multiple levels of hearsay in the form of police reports that included victims’ observations); Fortin, 189 N.J. at 604-06.
¶50 The Court of Appeals ultimately upheld admission of the HITS evidence under ER 703, which authorizes an expert to rely on otherwise inadmissible hearsay. Coe, 160 Wn. App. at 829. The problem is that the trial court also admitted the HITS evidence for the substantive purpose of proving identity. See Grp. Health Coop. of Puget Sound, Inc. v. Dep’t of Revenue, 106 Wn.2d 391, 399-400, 722 P.2d 787 (1986) (allowing admission of otherwise inadmissible hearsay to show basis for expert’s opinion but not as proof of the facts). In light of the substantive admission, the Court of Appeals erred.
¶51 The State attempts to justify the HITS evidence’s substantive admission under either the business *505record or public record exception to the hearsay rule.7 Neither exception applies. The business record exception generally applies to objective records of a regularly recorded activity and not those “reflecting the exercise of skill, judgment, and discretion.” 5C Karl B. Tegland, Washington Practice: Evidence Law and Practice § 803.37 (5th ed. 2007) (citing State v. Hopkins, 134 Wn. App. 780, 142 P.3d 1104 (2006)); RCW 5.45.020.
¶52 Even assuming the HITS investigators’ actions satisfied the exception,8 the police reports that the database relies on do not. Police reports are a subjective summary of the officer’s investigation, rendering them inadmissible. State v. Hines, 87 Wn. App. 98, 101-02, 941 P.2d 9 (1997). Moreover, the victims’ statements that make up the reports are an additional level of hearsay. Where, as here, multiple levels of hearsay are involved, each level must meet an exception. ER 805.
¶53 The public record exception does not apply for the same reason. This exception applies to records on file at a department of this state that have been certified by the custodian of that record. RCW 5.44.040. Similar to the business record exception, the exception applies to objective records that are not based on judgment or discretion. State v. Monson, 113 Wn.2d 833, 839, 784 P.2d 485 (1989); Hines, 87 Wn. App. at 101. Further, the same double hearsay also exists, thus providing yet another reason the public record exception does not apply.
¶54 The State completely ignores the double hearsay that bars application of both exceptions. Without an applicable exception to the underlying layers of hearsay — the *506police officers’ observations and the victims’ statements— the HITS evidence is inadmissible. See ER 805. Accordingly, neither exception applies.
¶55 The State raises a final independent argument that we should follow cases from the Eighth Circuit Court of Appeals approving the admission of computer printouts into evidence despite hearsay concerns. See United States v. Smith, 973 F.2d 603 (8th Cir. 1992); United States v. Enterline, 894 F.2d 287 (8th Cir. 1990). Those cases, however, failed to address the double hearsay contained in the printouts, United States v. Sallins, 993 F.2d 344 (3d Cir. 1993), which renders the cases unpersuasive. Further, the Eighth Circuit’s reasoning has not been adopted by Washington and it seems contradictory to the existing jurisprudence in Hines and Monson. Accordingly, we reject the State’s arguments and hold that the HITS database contains inadmissible hearsay preventing substantive admission.
c. Unreliable or Misleading
¶56 The final argument Coe raises regarding the HITS evidence is that beyond being unreliable as hearsay, it is generally unreliable and misleading. This part of Coe’s argument really contains two separate claims. First, he claims that the trial court improperly shifted the burden of proof and required Coe to prove the HITS database was unreliable. Second, he claims that the HITS database is unrepresentative of the number of rapes that occur in Washington. We reject both arguments.
¶57 First, the trial court never shifted the burden of proof. During a pretrial evidentiary hearing, Coe argued that the HITS database did not contain well-founded statistics. The trial court interrupted Coe and stated, “[W]hat you believe isn’t important, it’s what you demonstrated to me through evidence. And I don’t have any evidence whatsoever on the statistical validity of the studies____[C]onfine your arguments to what evidence you provided to the *507Court.” 23 Verbatim Report of Proceedings (VRP) at 4035-36. Coe misunderstands the trial court’s statement. The State had just presented three witnesses that testified about the HITS database’s relevance and reliability. Coe then wished to rebut those witnesses’ testimony but failed to introduce supporting evidence. That is why the trial court told Coe to limit his arguments to the facts. The court did not shift the burden; the initial burden had already been met.
¶58 Second, Coe fails to demonstrate that the database is unrepresentative. The HITS database is “nothing more than [a] sophisticated record-keeping system [ ].” Russell, 125 Wn.2d at 70. And there is nothing prohibiting the use of “well-founded statistics to establish some fact that will be useful to the trier of fact.” Id.9 Even if the HITS database does not contain every rape that has occurred in Washington, that does not make the database statistically unreliable. The database mostly contains stranger-on-stranger rapes that involve some form of penetration. Because the State is looking to identify a unique signature that belongs to a rapist who attacks only strangers, it makes sense to search crimes involving only those types of victims. If the State can identify a unique signature in such a database, that makes the identified signature all the more relevant.
¶59 Coe’s reliance on Hernandez to prove otherwise is misplaced. The Hernandez court primarily rejected the database because of hearsay concerns, not general reliability concerns. 55 Cal. App. 4th at 240-41. And to the extent the Hernandez court did address general reliability, the case is distinguishable. In Hernandez, the results were used to show that no similar crimes occurred either before the defendant arrived in the area or after the defendant was arrested. Id. at 228. By contrast, the HITS results were *508used to show that Coe’s signature was present in several offenses, not that it was absent from HITS. In other words, the HITS results are based on the crimes actually contained in the database, thus giving an evidentiary foundation for using the database.
¶60 Given our prior approval of the HITS database, the trial court did not abuse its discretion in concluding the HITS database was sufficiently representative to be reliable. Regardless, the trial court still abused its discretion in substantively admitting the HITS results because of the inadmissible hearsay.
d. Prejudice
¶61 Although the trial court improperly admitted the HITS evidence, the admission did not prejudice Coe. An evidentiary error “is prejudicial if, ‘within reasonable probabilities, had the error not occurred, the outcome of the trial would have been materially affected.’ ” State v. Neal, 144 Wn.2d 600, 611, 30 P.3d 1255 (2001) (quoting State v. Smith, 106 Wn.2d 772, 780, 725 P.2d 951 (1986)). Here, all of the 13 admitted rapes identified by HITS were also identified by Dr. Keppel as containing Coe’s signature. Moreover, as discussed below, Dr. Phenix properly relied on the HITS results in forming her expert opinion that Coe was an SVP. Evidence that Coe suffers from a mental abnormality or personality disorder that makes him likely to reoffend is central to a commitment hearing like this one, where a charged crime of sexual violence has already been proved. See Post, 170 Wn.2d at 309-10. Considering that all of the rapes identified by the HITS searches were also identified by the signature analysis and that Dr. Phenix was still permitted to rely on the HITS results in forming her opinion, any error was harmless.
4. Victim Testimony
¶62 The trial court allowed seven victims identified by the signature analysis and the HITS evidence to testify at *509the trial. Coe contends the trial court erred in allowing these victims to testify because the signature analysis and HITS evidence were inadmissible and the testimony’s relevance hinged on admission of that evidence. As stated above, the signature analysis was properly admitted, and it identified all of the victims that testified. Their testimony was therefore relevant. See In re Det. of Turay, 139 Wn.2d 379, 401, 986 P.2d 790 (1999) (holding that prior victim testimony is relevant in an SVP proceeding). The trial court, therefore, properly allowed the victims to testify.
5. Due Process Challenge
¶63 Coe also challenges the admission of evidence from five other sexual assaults where the victims were unavailable to testify.10 During the trial Dr. Phenix testified that she relied on the reports of these five assaults, which included the victims’ statements, in determining that Coe was an SVP. Coe was unable to confront or cross-examine the five victims of these assaults because they were unavailable for deposition or for trial. He alleges this inability violated his due process rights. We disagree.
¶64 We have previously held that a defendant in an SVP proceeding has no right to confront witnesses, either in trial or in deposition. In re Det. of Stout, 159 Wn.2d 357, 368-74, 150 P.3d 86 (2007). In Stout, the detainee wished to confront a victim who gave two telephonic depositions and one filmed deposition. Id. at 368. This court applied the Mathews11 balancing test and concluded that Stout had no due process right to confront witnesses. Id. at 370-74. The court noted that “although SVP commitment proceedings include many of the same protections as a criminal trial, SVP commitment proceedings are not criminal proceedings.” Id. at 369.
*510¶65 Coe attempts to limit Stout to its facts because the detainee there at least had an opportunity to cross-examine the witness, which Coe never did.12 Coe believes this fact tips the Mathews balancing test in his favor. We disagree. The Mathews test balances “(1) the private interest affected; (2) the risk of erroneous deprivation of that interest through existing procedures and the probable value, if any, of additional procedural safeguards; and (3) the governmental interest, including costs and administrative burdens of additional procedures.” Id. at 370 (citing Mathews, 424 U.S. at 335).
¶66 The first factor clearly favors Coe because he has a significant interest in his physical liberty. See id. But the remaining two factors still favor the State. In Stout, the second factor favored the State because significant procedural safeguards exist in an SVP proceeding. Id. at 370-71. Specifically, the detainee has an initial right to cross-examine witnesses during the probable cause hearing, and throughout the proceedings he has a right to counsel, to present evidence, and to view the petitions and reports on file. Id. at 370. Further, the detainee, throughout the entire proceeding, has a right to counsel, to a jury trial, and to have the charges proved beyond a reasonable doubt. Id. at 370-71. Finally, the jury verdict must be unanimous in an SVP proceeding. Id.
¶67 These significant safeguards led the Stout court to conclude that creating a confrontation right would add little value. Id. at 371. In Stout, there was no concern over the testimony’s veracity as it was taken under oath, and the detainee had an opportunity to review the videotaped testimony, thereby allowing him the opportunity to impeach the witness. Id. Those same facts do not exist here, and Coe believes that is enough to change the outcome of the Mathews test. Here, Dr. Phenix testified that she incorpo*511rated these five offenses into her evaluation of Coe. There was not an opportunity for the jury to evaluate these witnesses, like the jury could with the witness in Stout.
¶68 Regardless, the same statutory safeguards exist. Coe had a right to counsel, to a jury trial, and to a unanimous verdict. “Most importantly, at trial the State carries the burden of proof beyond a reasonable doubt.” Id. at 370-71. His inability to cross-examine the five victims does not reduce the effectiveness of the current statutory procedural safeguards. We stated the safeguards were sufficient in Stout, and the facts of this case do not change the analysis now. If Coe believed the facts as related by Dr. Phenix were untrue, nothing prevented him from offering rebuttal testimony about those facts or cross-examining Dr. Phenix.
¶69 Additionally, the evidence was never admitted substantively, which favors the State under the second prong. The evidence was admitted only to show the underlying basis for Dr. Phenix’s opinion. ER 705 allows otherwise inadmissible evidence to be admitted for this purpose so long as it is not being offered for the truth of the matter asserted. Grp. Health Coop. of Puget Sound, Inc., 106 Wn.2d at 399. This reduces the probable value of requiring an opportunity for confrontation. The cases Coe relies on to suggest otherwise are all criminal cases involving the Sixth Amendment to the United States Constitution. See Reply Br. of Appellant at 29 (citing Crawford v. Washington, 541 U.S. 36, 61, 124 S. Ct. 1354, 158 L. Ed. 2d 177 (2004); Melendez-Diaz v. Massachusetts, 557 U.S. 305, 321, 129 S. Ct. 2527, 174 L. Ed. 2d 314 (2009); People v. Goldstein, 6 N.Y.3d 119, 122, 843 N.E.2d 727, 810 N.Y.S.2d 100 (2005)). Such cases have little relevance when considering the due process concerns outside a criminal trial. Cf. Stout, 159 Wn.2d at 372-73 (emphasizing that few cases outside of a criminal trial context have supported a due process right to confront a witness). Consequently, the second Mathews factor favors the State.
¶70 The third Mathews factor also favors the State. Id. at 371. To begin, the State has a significant interest in prevent*512ing an SVP from reoffending. Id. The State also “has an interest in streamlining commitment procedures [to] avoid [ ] the heavy financial burden that would” accompany live testimony. Id. The additional financial burden is unjustifiable considering the marginal protection that an additional confrontation right would provide to the detainee’s liberty interest. Id. at 372. As a result, we hold that Coe had no due process right to confront these victims.
6. Dr. Phenix’s Reliance on the Signature Analysis and HITS Evidence
¶71 Dr. Phenix believed, in her professional opinion, that Coe was an SVP. In making this determination, she relied on the signature analysis and HITS evidence in addition to her independent review of over 74,000 pages of material. Coe objected to her reliance on the signature analysis and HITS evidence, arguing that it was unreliable and irrelevant. The trial court held the underlying evidence was admissible and, consequently, allowed Dr. Phenix to rely on it.
¶72 We affirm the trial court’s decision in regard to the signature analysis because it was properly admitted.13 As for the HITS results, we affirm the trial court’s decision because an expert can rely on inadmissible hearsay. See ER 703; In re Det. of Marshall, 156 Wn.2d 150, 161-62, 125 P.3d 111 (2005).14 In Marshall, this court held that Dr. Phenix properly relied on records, which were inadmissible hearsay, to form her opinion that the accused was an SVP. 156 Wn.2d at 162-63. Her reliance was proper because the records were typically relied on by experts in her field. Id. at 162. Moreover, Dr. Phenix could relate the inadmissible hearsay to the jury so long as she was merely *513explaining the underlying basis for her expert opinion. Id. at 162-63.
f 73 The HITS evidence at issue here is no different than the records at issue in Marshall. In both cases, Dr. Phenix relied on inadmissible hearsay to form her opinion, and in both cases, she related those facts to the jury to explain the underlying basis for her expert opinion.15
¶74 Furthermore, Dr. Phenix is not the first expert to have relied on the HITS database to support her independent opinion. See Russell, 125 Wn.2d at 69-70. Here, Dr. Phenix independently reviewed 74,000 pages of documents and personally evaluated Coe in forming her opinion that Coe was an SVP. Using the HITS database to support this independent opinion is similar to how the experts in Russell supported their independent opinions with the HITS database. We therefore affirm the trial court’s decision to allow Dr. Phenix to rely on the HITS evidence.
7. Dr. Phenix’s Disclosure of Unadjudicated Rapes
¶75 In conducting her evaluation, Dr. Phenix also relied on 33 sexual offenses, and at trial she disclosed 20 of these to the jury. Coe challenges the trial court’s decision to allow Dr. Phenix to disclose those 20 offenses to the jury. He does not, however, challenge her reliance on them.
¶76 A trial court may allow an expert to reveal the underlying basis for her opinion if doing so will help the jury understand the expert’s opinion. ER 705; Marshall, 156 Wn.2d at 163. The disclosure is permissible even if the information would be inadmissible as substantive evidence. Marshall, 156 Wn.2d at 163. For example, in Marshall, Dr. Phenix disclosed a number of facts that would have been inadmissible as substantive evidence but that were admissible to demonstrate the basis for her opinion. Id. The trial court need only give an appropriate limiting instruction *514explaining that the jury is not to consider this revealed information as substantive evidence. Id. (citing 5B Karl B. Tegland, Washington Practice: Evidence Law and Practice §§ 705.4, 705.5 (4th ed. 1999)).
¶77 Here, Dr. Phenix’s disclosure of the unadjudicated offenses is no different from her disclosure of the otherwise inadmissible facts in Marshall. She relied on these unadjudicated offenses, and she testified that it is common for experts in SVP proceedings to do so. Additionally, the trial court gave an appropriate limiting instruction, which in relevant part states:
Dr. Phenix is about to testify regarding the factual bases of her opinion. You may consider this testimony only in deciding what credibility and weight should be given to the opinions of Dr. Phenix. You may not consider it as evidence that the information relied upon by the witness is true or that the evidence described actually occurred.
17 VRP at 3086. Accordingly, the trial court did not err in allowing Dr. Phenix to disclose these unadjudicated offenses.
¶78 On review, Coe seems to challenge the idea that a limiting instruction could ever prevent a jury from considering the disclosed facts as evidence. Opening Br. of Appellant at 78-79 (“[G]iven the sheer amount of evidence offered through Phenix, the likelihood that the jury would maintain this distinction and disregard the underlying information for its truth seems remote”). But as the Court of Appeals correctly stated, “The jury is presumed to follow the court’s instructions.” Coe, 160 Wn. App. at 837 (citing State v. Yates, 161 Wn.2d 714, 763, 168 P.3d 359 (2007)). Coe cites to Bruton v. United States, 391 U.S. 123, 135-36, 88 S. Ct. 1620, 20 L. Ed. 2d 476 (1968), where the Court held that a jury cannot be expected to ignore the confession of a nontestifying codefendant that expressly implicates the defendant. However, Bruton involves a narrow exception to the general rule that juries follow instructions. Richardson *515v. Marsh, 481 U.S. 200, 207-08, 107 S. Ct. 1702, 95 L. Ed. 2d 176 (1987). That exception does not exist here.
¶79 There is nothing to differentiate this case from Marshall where we already condoned using a limiting instruction in this context. Therefore, the trial court did not abuse its discretion in allowing disclosure because it gave an appropriate limiting instruction and because Dr. Phenix reasonably relied on these offenses in forming her opinion.
8. Cumulative Error
¶80 Finally, Coe argues the alleged errors resulted in reversible cumulative error. The cumulative error doctrine applies where a combination of trial errors denies the accused a fair trial even where any one of the errors, taken individually, may not justify reversal. State v. Greiff, 141 Wn.2d 910, 929, 10 P.3d 390 (2000). Considering the fact that the trial court’s sole error, substantively admitting the HITS evidence, was harmless, we hold that Coe was not deprived of a fair trial.
CONCLUSION
¶81 In sum, we affirm the result reached by the Court of Appeals and hold that Coe establishes no reversible error. We note, however, that substantively admitting the HITS evidence was error because the database was based on inadmissible hearsay. Regardless, Coe fails to establish prejudice as each crime identified by the HITS evidence was also independently identified by Dr. Keppel’s signature analysis. While we may disapprove of the Court of Appeals’ reasoning as to why the signature analysis was admissible, we agree that the trial court’s decision to admit the signature analysis under ER 404(b) was proper under the unique facts of this case. Finding no reversible error, we affirm Coe’s commitment order.
Madsen, C.J., and C. Johnson, Fairhurst, J.M. Johnson, Stephens, Wiggins, and González, JJ., concur.These are the expert’s definition of the terms, not legal definitions.
The offender also allegedly remarked that “ ‘the bitch is dead’ ” during both crimes but that statement had limited relevance. Thang, 145 Wn.2d at 645, 641 (witness could not remember who made the “bitch” comment during the incident).
Coe does not dispute Dr. Keppel’s status as an expert in this area. Nor does Coe challenge the use of “linkage analysis.”
The State claims that Coe failed to preserve this partial signature argument, but Coe raised these differences when the trial court heard the motions. Accordingly, the error was preserved regardless of whether the specific objection was made. See ER 103(a)(1) (objection is sufficient if the specific ground is apparent from the context).
Coe also challenges whether statistics can ever be used to show a signature but provides no authority to support such a broad assertion. Accordingly, we reject it.
Although not an independent reason to admit the HITS results, the 13 rapes admitted by the HITS results were also identified by Dr. Keppel’s signature analysis.
The State also argues that Russell is dispositive on the issue of HITS admissibility. But this court did not consider a hearsay objection in Russell, and the HITS results were not substantively admitted. See 125 Wn.2d at 70.
Arguably, their actions do not as the data was not entered pursuant to a standardized process but in preparation for this litigation. See 23 Verbatim Report of Proceedings at 3929-33 (Matheny received the 24 cases, including the 6 cases Coe was originally charged with, for input into the database around the time the State petitioned for Coe’s commitment).
Coe attempts to distinguish Russell because the HITS results were not admitted substantively in that case. But Coe fails to state exactly how this distinction impacts the reliability of the HITS database.
Dr. Phenix relied on testimony from Daria L. (two rapes), Diana A. (rape), Valerie L. (attempted rape), Mary O. (indecent exposure), and Claudia H. (indecent exposure).
Mathews v. Eldridge, 424 U.S. 319, 334-35, 96 S. Ct. 893, 47 L. Ed. 2d 18 (1976).
Cross-examination in Stout meant that the detainee could view the depositions and point out inconsistencies. 159 Wn.2d at 371.
We accept Dr. Phenix’s uncontroverted testimony that experts in her field typically rely upon a signature analysis.
Dr. Phenix’s status as an expert in this field is undisputed. Cf. Marshall, 156 Wn.2d at 160.
Again, we accept Dr. Phenix’s uncontroverted testimony that experts in her field would rely upon HITS evidence.