**271I. Introduction
Thirty-five years ago, Dr. Roland Summit, M.D., a clinical psychiatrist, identified five categories of behavior that were reportedly common in victims of *446child sexual abuse: secrecy; helplessness; entrapment and accommodation; delayed, conflicted, unconvincing disclosure; and retraction. Dr. Summit drew on various sources, including his own clinical practice, and asserted that the five behaviors comprised a syndrome -- the "Child Sexual Abuse Accommodation Syndrome" (CSAAS).
Courts across the nation embraced Dr. Summit's findings, which paved the way for experts to testify about the syndrome in criminal sex abuse trials. In 1993, this Court found that CSAAS evidence was sufficiently reliable to be admitted. State v. J.Q., 130 N.J. 554, 617 A.2d 1196 (1993).
**272In the decades since Dr. Summit's article first appeared, neither the American Psychiatric Association nor the American Psychological Association has recognized CSAAS. The syndrome does not appear in the Diagnostic and Statistical Manual of Mental Disorders (DSM-5), the mental health field's authoritative list of mental disorders. And the notion of a child abuse accommodation "syndrome" has been examined, critiqued, and undermined by a number of scientific studies.
Defendant challenged the CSAAS evidence introduced at trial. To better assess defendant's claim, we remanded the matter for a hearing before the trial court. Four experts testified at the hearing, and the parties introduced and discussed numerous scientific studies.
We rely heavily on the record developed at the hearing. Based on what is known today, it is no longer possible to conclude that CSAAS has a sufficiently reliable basis in science to be the subject of expert testimony. We find continued scientific support for only one aspect of the theory -- delayed disclosure -- because scientists generally accept that a significant percentage of children delay reporting sexual abuse.
We therefore hold that expert testimony about CSAAS in general, and its component behaviors other than delayed disclosure, may no longer be admitted at criminal trials. Evidence about delayed disclosure can be presented if it satisfies all parts of the applicable evidence rule. See N.J.R.E. 702. In particular, the State must show that the evidence is beyond the understanding of the average juror.
That decision will turn on the facts of each case. Here, because the victim gave straightforward reasons about why she delayed reporting abuse, the jury did not need help from an expert to evaluate her explanation. However, if a child cannot offer a rational explanation, expert testimony may help the jury understand the witness's behavior. We therefore ask the Committee on Model Jury Charges to develop an appropriate instruction on delayed disclosure.
**273In this appeal, there was overwhelming evidence of defendant's guilt. Among other things, the victim made an audio recording of an act of sexual abuse that took place several weeks before she spoke with the police. As a result, we find that the expert testimony about CSAAS introduced at trial was harmless, and we affirm defendant's convictions.
II. Facts and Procedural History
A. Facts
A Hudson County Grand Jury charged defendant J.L.G. in a nine-count indictment in 2012. After the trial court severed four of the counts and the State dismissed a fifth, defendant went to trial on the following charges: first-degree aggravated sexual assault, N.J.S.A. 2C:14-2(a) ; third-degree aggravated criminal sexual contact, N.J.S.A. 2C:14-3(a) ; second-degree endangering the welfare of a child, N.J.S.A. 2C:24-4(a) ; and third-degree witness tampering, N.J.S.A. 2C:28-5(a). Defendant's *447stepdaughter, whom we refer to by the fictitious name "Bonnie" to protect her identity, testified at trial about an escalating pattern of sexual abuse that defendant carried out against her for roughly eighteen months. We rely on the trial record to recount the facts.
Defendant began dating Bonnie's mother around 1996, when Bonnie was an infant. He moved in with them months later and assumed the role of Bonnie's father.
According to Bonnie, the sexual abuse began in 2011, when she was fourteen and defendant was about thirty-two. One day, defendant called Bonnie into the living room and showed her his exposed penis. Over the course of the next year and a half, defendant had Bonnie record him masturbating, ejaculated on her chest area, touched her, had her masturbate him, performed oral sex on her, digitally penetrated her, and had vaginal intercourse with her four or five times.
Bonnie said the sexual abuse occurred on a daily basis, always against her wishes. After each incident, defendant put money on her dresser. At one point, he gave her an iPhone. He also **274instructed her not to say anything and threatened her; defendant pointed a gun at Bonnie and threatened to hurt her, her mother, or her brother if word got out. Bonnie told no one about the abuse, which she found embarrassing.
A close friend of Bonnie's mother visited the family apartment one day in or around 2011 and found defendant lying on top of Bonnie. Although defendant wore jeans, the friend noticed that he had an erection. When Bonnie's mother heard about the incident, she questioned her daughter. With a knife in hand, the mother said she would kill defendant "if he's doing something." Bonnie was afraid her mother would follow through with the threat and denied any sexual activity. Although Bonnie claimed she wanted to tell her mother, she also did not "want her to do anything for her to get locked up."
In May or June of 2012, Bonnie used her iPhone to record the last episode of sexual abuse. She felt that she could no longer "stick it out" at home until age 18 and wanted to have proof when the abuse eventually came to light.
Bonnie testified that defendant performed oral sex on her and also tried to penetrate her on this occasion. The State introduced the audio recording at trial and played it during Bonnie's testimony. The jury heard very descriptive, at times graphic, language about sexual acts. At one point, Bonnie refused to follow defendant's directive about how to position herself. At another, defendant said, "I'm not [ejaculating] in you" and cursed at Bonnie. Throughout the recording, Bonnie repeatedly stated that "[t]his is going to be the last thing."
After an argument with defendant on June 13, 2012, Bonnie told her mother that defendant had "been raping [her] for the past year and a half." Bonnie added that she had proof and tried to play the iPhone recording. According to Bonnie's mother, defendant tried to take away the phone, turned "white as a ghost," and left the apartment after she decided to call the police.
**275Bonnie made a statement to the prosecutor's office and placed two phone calls to defendant under the guidance of detectives. During the recorded conversations, defendant repeatedly said he was "sorry," at Bonnie's urging, but he did not specify the reason or admit to any sexual misconduct. He also offered to give Bonnie money, pay for her phone, and "[e]ven buy [her] a car" -- after he asked her to withdraw the allegations.
*448Soon after, the police located defendant at a friend's house and arrested him. They also retrieved a pair of shorts that Bonnie claimed she wore during the last sexual encounter with defendant. The shorts contained a small amount of male DNA. Although defendant was not identified as the source, he could not be excluded as a possible contributor.
B. Trial
The State presented the above evidence at trial through various witnesses including Bonnie, her mother, the mother's friend, an expert on DNA, and a number of police officers.
Defendant did not testify. In his defense, counsel highlighted the absence of physical evidence in support of Bonnie's accusations. Counsel also challenged Bonnie's credibility in a number of ways: inconsistencies in her account of events; a pending criminal charge against Bonnie that arguably influenced her cooperation with the prosecutor's office; Bonnie's delay in reporting acts of abuse for more than a year after they allegedly began; poor performance and behavior problems at school; and other issues.
The defense did not dispute the authenticity of the recording Bonnie made on her own. Counsel told the jury in summation that the conversation should not have happened, and that defendant "pleads guilty" to the child endangerment charge as a result of the recording. Counsel argued, however, that the State presented no physical proof of penetration in the case.
Central to this appeal is the CSAAS evidence that surfaced at various points during trial. Defendant tried to bar the testimony in advance. In a written opinion, the trial court denied defendant's **276pretrial motion. The court found that the evidence was relevant because it would help the jury evaluate "the significance of the victim's delayed disclosure." In addition, the trial judge concluded that the probative value of the evidence was not outweighed by the risk of undue prejudice because the testimony would not be offered to prove defendant's guilt, would be subject to cross-examination, and would be accompanied by appropriate limiting instructions. The court also found that the testimony satisfied the admissibility requirements for scientific evidence under N.J.R.E. 702. In that regard, the trial judge relied on this Court's settled case law.
Dr. Lynn Taska, a clinical psychologist, testified as an expert on CSAAS. Immediately before, the trial court gave the jury detailed instructions about how to consider her testimony; the court's charge closely followed the model jury charge.
Dr. Taska has a Ph.D. in clinical psychology and specializes in the area of child sexual abuse. She has testified as an expert witness on CSAAS more than thirty times.
Dr. Taska testified before Bonnie did. Her testimony fills thirty pages in the trial transcript. At the outset, Dr. Taska stated that she knew nothing about the facts of the case. She also explained that CSAAS was not a diagnostic tool and did not offer proof that sexual abuse actually happened. Instead, it was "a description of a collection of behaviors" "meant to educate us about how children ... who have been sexually abused typically behave."
Dr. Taska described Dr. Summit's article on CSAAS and discussed in detail the five factors that comprise the syndrome. Throughout her testimony, she referred to various studies about CSAAS and summarized a number of them for the jury. She told the jury that, in her judgment, "there is an enormous body of literature supporting elements of" CSAAS. Dr. Taska critiqued two studies critical of CSAAS and *449countered that "there are hundreds of studies that support" it. **277The prosecution referenced Dr. Taska's testimony in summation and argued that "just because" child victims of sexual assault "don't report the abuse, that doesn't mean they shouldn't be believed." The prosecution also recounted Bonnie's reasons for not disclosing the abuse: defendant had threatened her with a gun, and she was scared of how her mother might react.
In the final instructions to the jury, the trial court again recited the model charge on CSAAS. The court stressed that Dr. Taska's testimony could not be considered as proof that defendant sexually abused Bonnie. Instead, the testimony could be considered to "explain[ ] certain behavior of the alleged victim of child sexual abuse.... The accommodation syndrome, if proven, may help explain why a sexually abused child may delay reporting." The full text of the model charge appears later.
C. Verdict and Appellate History
The jury convicted defendant of all four counts. Defendant was sentenced to an aggregate term of twenty-three years in prison, subject to an eighty-five percent period of parole ineligibility on the aggravated sexual assault count. See N.J.S.A. 2C:43-7.2. The court dismissed the severed charges at sentencing.
Among other arguments on appeal, defendant challenged the admissibility of the CSAAS testimony. The Appellate Division affirmed the convictions but vacated a penalty for the Sex Crime Victim Treatment Fund so that the trial court could assess the correct penalty amount on remand. As to the CSAAS issue, the panel appropriately observed that "[t]he admissibility of CSAAS expert testimony is well settled." The panel cited J.Q., 130 N.J. 554, 617 A.2d 1196, for support. The Appellate Division also noted that the trial court provided proper limiting instructions and that Dr. Taska acknowledged CSAAS evidence was not meant to determine whether sexual abuse occurred in any individual case. The panel concluded that the testimony was relevant to explain Bonnie's delayed disclosure and determine her credibility.
We granted defendant's petition for certification limited to a single issue:
**278whether the trial court properly denied defendant's motion to exclude the testimony of the State's expert regarding CSAAS, on the grounds that CSAAS testimony was irrelevant to defendant's trial, that its admission was unduly prejudicial to defendant, and that CSAAS testimony is not sufficiently reliable to meet the standard of N.J.R.E. 702.
[ 229 N.J. 606, 607, 164 A.3d 402 (2017).]
Because we lacked an adequate factual record to consider the issue, we remanded to the trial court for a hearing "to determine whether CSAAS evidence meets the reliability standard of N.J.R.E. 702, in light of recent scientific evidence." Ibid.
D. Remand Hearing
The Honorable Peter F. Bariso, Jr., A.J.S.C., presided over the hearing on remand. It was held in July 2017 and lasted four days. Four experts submitted reports and testified: Dr. Anthony D'Urso, Psy.D., and Dr. Thomas Lyon, J.D., Ph.D., for the State; and Dr. Charles Brainerd, Ph.D., and Dr. Maggie Bruck, Ph.D., for defendant. Dozens of exhibits were introduced, including multiple published scientific articles.
Dr. D'Urso is the section chief and supervising psychologist at the Audrey Hepburn Children's House, a legislatively mandated regional child abuse diagnostic center in the State. He is also an associate professor of psychology at Montclair State *450University, where he teaches courses in clinical and forensic psychology.
Dr. Lyon is a professor at the University of Southern California Gould School of Law, where he holds an endowed chair in law and psychology. His research is designed "to identify methods for interviewing children that maximize children's willingness to disclose negative events while minimizing suggestibility and error."
Dr. Brainerd is an experimental and developmental psychologist. He is a tenured professor at Cornell University in the College of Human Ecology, the Department of Human Development, and the Graduate Fields of Cognitive Science, Psychology, and Human Development. He is the Chair of the Department of Human Development and Director of the Laboratory of Memory and Neuroscience. He also directs Cornell's Dual Ph.D./J.D. Program **279and the Ph.D. Program in Law, Psychology, and Human Development.
Dr. Bruck is a part-time professor in the Division of Child and Adolescent Psychiatry, Department of Psychiatry and Behavioral Science, at Johns Hopkins University School of Medicine. She is also an adjunct professor of psychology at McGill University. She previously served as a full professor at both universities. Dr. Bruck specializes in cognitive and developmental psychology, with a particular focus on autobiographical memory and the capabilities of young children.
Judge Bariso granted leave to participate as amicus curiae to the American Civil Liberties Union of New Jersey (ACLU), the American Professional Society on the Abuse of Children, the Association of Criminal Defense Lawyers of New Jersey, the Attorney General of New Jersey, the County Prosecutors Association of New Jersey, the Last Resort Exoneration Project at Seton Hall University School of Law, and the New Jersey Association for Justice.
As discussed further below, Judge Bariso found that the State failed to show general acceptance of CSAAS in the relevant scientific community and concluded that there was consensus only as to delayed disclosure. He also found "great controversy within the scientific community" about "the tenets of CSAAS." Judge Bariso concluded that CSAAS evidence did not meet the standard for admissibility under N.J.R.E. 702.
To better understand the remand hearing and the parties' arguments, we first consider the appropriate legal framework to assess expert testimony in a criminal case and then provide an overview of CSAAS and its history.
III. Legal Framework
N.J.R.E. 702 governs the admission of expert testimony. The rule provides that "[i]f scientific ... knowledge will assist the trier of fact to understand the evidence or to determine a fact in issue, a witness qualified as an expert by knowledge, skill, experience, **280training, or education may testify thereto in the form of an opinion or otherwise." N.J.R.E. 702.
To satisfy the rule, the proponent of expert evidence must establish three things: (1) the subject matter of the testimony must be "beyond the ken of the average juror"; (2) the field of inquiry "must be at a state of the art such that an expert's testimony could be sufficiently reliable"; and (3) "the witness must have sufficient expertise to offer the" testimony. State v. Kelly, 97 N.J. 178, 208, 478 A.2d 364 (1984). Consistent with the Court's remand order, the hearing before Judge Bariso focused primarily on the reliability prong.
*451In criminal cases, this Court has continued to rely on the Frye standard to assess reliability. The test requires trial judges to determine whether the science underlying the proposed expert testimony has "gained general acceptance in the particular field in which it belongs." Frye v. United States, 293 F. 1013, 1014 (D.C. Cir. 1923) ; accord State v. Harvey, 151 N.J. 117, 170, 699 A.2d 596 (1997) ; see also State v. Torres, 183 N.J. 554, 568, 874 A.2d 1084 (2005).
The Last Resort Exoneration Project asks the Court to depart from Frye in criminal cases and adopt the Daubert test that federal courts use. Under Daubert, trial judges perform a "preliminary assessment of whether the reasoning or methodology underlying the testimony is scientifically valid" and "whether that reasoning or methodology properly can be applied to the facts in issue." Daubert v. Merrell Dow Pharms., Inc., 509 U.S. 579, 592-93, 113 S.Ct. 2786, 125 L.Ed.2d 469 (1993). Under Daubert, general acceptance can still "have a bearing on the inquiry" but "is not a necessary precondition" to admissibility. Id. at 594, 597, 113 S.Ct. 2786.
Neither party asks us to make this change in the law, and we regularly decline to reach issues raised only by amici. See State v. J.R., 227 N.J. 393, 421, 152 A.3d 180 (2017). We prefer to wait for a case in which the parties raise and litigate the issue.
**281In any event, there are three ways to establish general acceptance under Frye: expert testimony, authoritative scientific and legal writings, and judicial opinions. State v. Townsend, 186 N.J. 473, 491, 897 A.2d 316 (2006) ; Harvey, 151 N.J. at 170, 699 A.2d 596 ; Kelly, 97 N.J. at 210, 478 A.2d 364. Although we look for wide support within the relevant scientific community, complete agreement is not required for evidence to be admitted. Harvey, 151 N.J. at 171, 699 A.2d 596.
IV. Overview and History of CSAAS
CSAAS originated from the work of Dr. Roland Summit, a clinical psychiatrist and advocate for child victims of sexual abuse. In 1983, he published an article in the journal Child Abuse and Neglect titled "The Child Sexual Abuse Accommodation Syndrome" (Summit I).1
The purpose of Dr. Summit's paper was "to provide a vehicle for a more sensitive, more therapeutic response to legitimate victims of child sexual abuse and to invite more active, more effective clinical advocacy for the child within the family and within the systems of child protection and criminal justice." Summit I, at 179-80. Dr. Summit believed that the behavior of child sexual abuse victims is "self-camouflaging" and "self-stigmatizing." Id. at 179. Adult beliefs, meanwhile, "are dominated by an entrenched and self-protective mythology that passes for common sense." Id. at 178. Too often, adults found fault in the "emotionally distraught child" who accused a respectable adult. Ibid.
Dr. Summit proposed CSAAS to better understand and accept the child victim's position; to challenge adult prejudices; and to "offer[ ] the child a right to parity with adults in the struggle for credibility and advocacy." Id. at 179, 191.
Dr. Summit described the syndrome he outlined as "a common denominator of the most frequently observed" behaviors of child **282sexual abuse victims: secrecy; *452helplessness; entrapment and accommodation; delayed, conflicted, and unconvincing disclosure; and retraction. Id. at 180-81. He described the first two categories as "preconditions" to sexual abuse, and the remaining three as "sequential contingencies" of sexual assault. Id. at 181. According to Dr. Summit, each category contradicts "the most common assumptions of adults." Ibid.
As to secrecy, Dr. Summit observed that "[t]he average child never asks and never tells." Ibid. Contrary to what Dr. Summit asserted is the "general expectation that the victim would normally seek help, the majority of the victims in retrospective surveys had never told anyone during their childhood." Ibid. According to Dr. Summit, they feared reprisal and faced "an adult conspiracy of silence and disbelief" if they spoke up. Ibid.
About helplessness, Dr. Summit wrote that most victims of child sexual abuse face "an overpowering adult" who "is often in a trusted and apparently loving position." Id. at 182-83. That "only increases the imbalance of power and underscores the helplessness of the child." Id. at 183. "[T]he child victim is expected to forcibly resist, to cry for help and to attempt to escape," even though, "no matter what the circumstances, the child had no choice but to submit quietly." Ibid. Dr. Summit added that "[c]linical experience and expert testimony can provide advocacy for the child" and "translate the child's world into an adult-acceptable language." Ibid.
On accommodation, Dr. Summit wrote that if sexual abuse is not stopped after the first incident, "[t]he healthy, normal, emotionally resilient child will learn to accommodate to the reality of continuing sexual abuse." Id. at 184. Accommodation mechanisms vary widely, according to Dr. Summit. The "survival skills" of a sexually abused child can include "domestic martyrdom, splitting of reality, altered consciousness, hysterical phenomena, delinquency, sociopathy, projection of rage, even self-mutilation." Id. at 186. "An alternative accommodation pattern exists" as well, "in which the child succeeds in hiding any indications of conflict. Such a child **283may be unusually achieving and popular, eager to please both teachers and peers." Ibid. The spectrum of accommodation, thus, includes both "suicidal" and "perfectly well-adjusted" children. Id. at 187.
For delayed disclosure, Dr. Summit wrote that "[m]ost ongoing sexual abuse is never disclosed, at least not outside the immediate family." Id. at 186. If a "family conflict triggers disclosure, it is usually only after some years of continuing sexual abuse." Ibid. Dr. Summit also noted that victims "tend[ ] to remain silent until ... adolescence." Ibid. When a child of any age discloses abuse after a period of delay, however, she is unlikely to be believed. Ibid.
Finally, on retraction, Dr. Summit wrote that "[w]hatever a child says about sexual abuse, she is likely to reverse it." Id. at 188. "Unless there is special support for the child and immediate intervention to force responsibility on the father," he continued, "the girl will follow the 'normal' course and retract her complaint." Ibid. In Dr. Summit's view, "[t]his simple lie carries more credibility than the most explicit claims of incestuous entrapment" because "[i]t confirms adult expectations that children cannot be trusted." Ibid. It also teaches the authorities "not to believe rebellious children who try to use their sexual power to destroy well-meaning parents." Ibid.
Dr. Summit stated that CSAAS was based on "[c]linical study of large numbers of children and their parents in proven cases of sexual abuse." Id. at 179. He also drew on (1) "statistically validated assumptions *453regarding prevalence, age relationships and role characteristics of child sexual abuse," and (2) "correlations and observations that have emerged as self-evident within an extended network of child abuse treatment programs and self-help organizations."Id. at 180. Among them, he wrote, is the "maxim among child sexual abuse intervention counselors and investigators that children never fabricate the kinds of explicit sexual manipulations they divulge in complaints or interrogations." Id. at 191. Dr. Summit added that he tested the validity of CSAAS "over **284a period of four years in [his] practice, which specializes in community consultation to diverse clinical and para-clinical sexual abuse programs." Id. at 180.
Nine years later, in 1992, Dr. Summit published a follow-up paper (Summit II).2 Dr. Summit noted that CSAAS has been "both elevated as gospel and denounced as dangerous pseudoscience." Summit II at 153. He revisited the original piece and its origin to address "subsequent distortions that court misuse has imposed." Id. at 154.
Dr. Summit stressed that CSAAS is "not a laboratory hypothesis" or "study of a defined population"; it is a "summary of diverse clinical consulting experience." Id. at 156. "CSAAS is a clinical opinion, not a scientific instrument," he wrote. Ibid. Dr. Summit noted that he relied on his "own broad consulting experience throughout Los Angeles County" along with "personal discussions" with "national visionaries" to compile a list of factors that "were both most characteristic of child sexual abuse" and counterintuitive to adults. Id. at 154.
Dr. Summit acknowledged the "misunderstanding" that stemmed from his use "of the word syndrome." Id. at 157. "I might better have chosen a name like the Child Sexual Abuse Accommodation Pattern to avoid any pathological or diagnostic implications." Ibid.
The following year, in J.Q., this Court found that CSAAS had a "sufficiently reliable scientific basis" to be presented to a jury. 130 N.J. at 556, 617 A.2d 1196. The Court relied extensively on Dr. Summit's initial 1983 article, which it called "the most concise and seemingly most authoritative statement of CSAAS." Id. at 566-67, 617 A.2d 1196. The Court observed that the article stemmed from "a scientific study of child-sexual-abuse victims," id. at 567, 617 A.2d 1196, and described "the most frequently observed victim behaviors," id. at 568, 617 A.2d 1196 (quoting Summit I, at 180).
**285The Court recounted the bases for Dr. Summit's observations and detailed the five categories of behavior discussed above. Id. at 568-70, 617 A.2d 1196.
The Court in J.Q. noted that
the behavioral studies of CSAAS are designed not to provide certain evidence of guilt or innocence but rather to insure that all agencies, including the clinician, the offender, the family, and the criminal justice system, offer "the child a right to parity with adults in the struggle for credibility and advocacy."
Id. at 571, 617 A.2d 1196 (quoting Summit I, at 191).
CSAAS achieves those goals, the Court explained, "by providing a 'common language' for analysis and a more 'recognizable map' to the understanding of child abuse." Ibid. (quoting Summit I, at 191).
The Court next analyzed CSAAS in light of the standard for the admission of expert evidence under Kelly and other case law.
*454Id. at 572, 617 A.2d 1196. It concluded that "[t]here does not appear to be a dispute about acceptance within the scientific community of the clinical theory that CSAAS identifies or describes behavioral traits commonly found in child-abuse victims." Id. at 573, 617 A.2d 1196. For support, the Court cited a law review article by Professor John E. B. Myers and others,3 as well as a note by Chandra Lorraine Holmes.4
Myers and his co-authors devoted part of an extended article to CSAAS. See, e.g., Myers et al., at 66-69. They described CSAAS; noted that some professionals "misinterpreted" the article and believed "Summit had discovered a 'syndrome' that could diagnose sexual abuse"; and discussed how various state courts had responded to CSAAS evidence. Id. at 67-68. The Myers article, however, does not cite scientific studies or articles that support **286Dr. Summit's five-factor syndrome. The article does conclude that "[t]he accommodation syndrome has a place in the courtroom" because it "helps explain why many sexually abused children delay reporting their abuse, and why many children recant allegations ... and deny that anything occurred." Id. at 68.
Holmes concluded that "CSAAS is a misnomer" because of its use of the term "syndrome," Holmes, at 157, and recognized that "[t]he methods underlying CSAAS are imperfect," id. at 169. Nonetheless, "[b]ecause the need for this type of testimony is compelling," the author concluded that courts should limit but allow CSAAS testimony "to explain a child's inconsistent behavior." Ibid. Holmes cautioned that the testimony should be accompanied by appropriate limiting instructions. Id. at 168-69.
The J.Q. Court concluded that CSAAS is not "evidence of guilt or innocence" and cannot be used as direct proof that abuse occurred. 130 N.J. at 571, 574, 578, 617 A.2d 1196. Rather, the Court allowed its use to help juries understand "traits often found in children who have been abused," which might otherwise be counterintuitive. Id. at 582, 617 A.2d 1196. The Court added that judges are to hold pretrial hearings to vet the qualifications of proposed expert witnesses, upon a defendant's request, and give proper limiting instructions to the jury. Id. at 583-84, 617 A.2d 1196.
In response to J.Q., the Model Criminal Jury Charge Committee developed an instruction on CSAAS. The current version of the jury charge reads as follows:
The law recognizes that stereotypes about sexual assault complaints may lead some of you to question [complainant's] credibility based solely on the fact that [he/she] did not complain about the alleged abuse earlier. You may or may not conclude that his/her testimony is untruthful based only on his/her [silence/delayed disclosure] [CHOOSE APPLICABLE TERM]. You may consider the [silence/delayed disclosure] along with all other evidence including [complainant's] explanation for his/her silence/delayed disclosure in deciding how much weight, if any, to afford to complainant's testimony. You may also consider the expert testimony that explained that silence/delay is one of the many ways in which a child may respond to sexual abuse. Accordingly, your deliberations in this regard should be informed by the testimony presented concerning *455the child sexual abuse accommodation syndrome. **287You may recall evidence that (NAME) [failed to disclose, or recanted, or acted or failed to act in a way addressed by the Child Sexual Abuse Accommodation Syndrome]. In this respect, Dr. [A], Ph.D., testified on behalf of the State [and Dr. [B], Ph.D., testified on behalf of the defendant]. Both witnesses were qualified as experts as to the Child Sexual Abuse Accommodation Syndrome. You may only consider the testimony of these experts for a limited purpose, as I will explain.
You may not consider Dr. [A]'s testimony as offering proof that child sexual abuse occurred in this case. [Likewise, you may not consider Dr. [B]'s testimony as proof that child sexual abuse did not occur]. The Child Sexual Abuse Accommodation Syndrome is not a diagnostic device and cannot determine whether or not abuse occurred. It relates only to a pattern of behavior of the victim which may be present in some child sexual abuse cases. You may not consider expert testimony about the Accommodation Syndrome as proving whether abuse occurred or did not occur. Similarly, you may not consider that testimony as proving, in and of itself, that ______, the alleged victim here, was or was not truthful.
Dr. [A]'s testimony may be considered as explaining certain behavior of the alleged victim of child sexual abuse. As I just stated, that testimony may not be considered as proof that abuse did, or did not, occur. The Accommodation Syndrome, if proven, may help explain why a sexually abused child may [delay reporting and/or recant allegations of abuse and/or deny that any sexual abuse occurred].
To illustrate, in a burglary or theft case involving an adult property owner, if the owner did not report the crime for several years, your common sense might tell you that the delay reflected a lack of truthfulness on the part of the owner. In that case, no expert would be offered to explain the conduct of the victim, because that conduct is within the common experience and knowledge of most jurors.
Here, Dr. [A] testified that, in child sexual abuse matters, [SUMMARIZE TESTIMONY]. This testimony was admitted only to explain that the behavior of the alleged victim was not necessarily inconsistent with sexual abuse. [CHARGE, IF APPLICABLE: here, Dr. [B] testified that, in child sexual abuse matters, [SUMMARIZE TESTIMONY]. This testimony was admitted only to explain that the behavior of the victim was not necessarily consistent with sexual abuse.]
The weight to be given to Dr. [A]'s [or Dr. [B]'s] testimony is entirely up to you. You may give it great weight, or slight weight, or any weight in between, or you may in your discretion reject it entirely.
You may not consider the expert testimony as in any way proving that [defendant] committed, or did not commit, any particular act of abuse. Testimony as to the Accommodation Syndrome is offered only to explain certain behavior of an alleged victim of child sexual abuse.
[Model Jury Charges (Criminal), "Child Sexual Abuse Accommodation Syndrome" (rev. May 16, 2011) (footnotes omitted).]
The trial court in this case recited the model charge nearly verbatim -- both before Dr. Taska testified and before the jury began deliberating.
**288*456On a number of occasions since J.Q., the Court has restated and refined certain principles about the use of CSAAS testimony. In State v. P.H., for example, the Court reversed a conviction after the trial court combined the charge on CSAAS with the model charge on "fresh complaint." 178 N.J. 378, 383, 840 A.2d 808 (2004). The latter charge informed the jury that it could "not consider the child's failure to complain as evidence weighing against the credibility of the child." Ibid. The Court found that the two charges, in combination, could confuse the jury, and offered guidance to reconcile the concepts. Id. at 399, 840 A.2d 808.
The following year, in State v. R.B., the Court cautioned that CSAAS experts should not list specific attributes of the syndrome that may track the behavior of the child victim in the case. 183 N.J. 308, 327-28, 873 A.2d 511 (2005). The Court again reviewed expert testimony on CSAAS in State v. Schnabel and found no error in how it was presented. 196 N.J. 116, 133-34, 952 A.2d 452 (2008). In State v. W.B., the Court directed that CSAAS experts may not quantify the percentage of children who lie about sexual abuse. 205 N.J. 588, 613-14, 17 A.3d 187 (2011).
Just last term, in State v. J.R., the Court again observed that experts should not discuss behaviors exhibited by the alleged victim in the trial when they describe conduct by confirmed victims of child sexual abuse. 227 N.J. at 416, 152 A.3d 180. To avoid confusion, experts should also not invoke highly publicized child sexual abuse scandals in their testimony. Ibid. In addition, the Court observed that because CSAAS evidence is meant to counter inferences about a child's delayed disclosure, experts should not testify before the victim at trial, as a general rule. Id. at 416-17, 152 A.3d 180.
In none of those cases, however, did the Court reassess the scientific underpinning of CSAAS evidence.
Forty other states and the District of Columbia allow CSAAS testimony for some purpose. See King v. Commonwealth, 472 S.W.3d 523, 535 (Ky. 2015) (Abramson, J., dissenting) (collecting cases). Many, like New Jersey, allow experts to explain common **289traits of abuse victims in general. See id. at 535 & n.15. Some limit CSAAS testimony to cases in which "the victim exhibit[s] a specific trait of the syndrome." See id. at 535 & n.16. Others permit the testimony to rehabilitate a witness's credibility. See id. at 535 & n.17. The Eighth and Ninth Circuits also allow the evidence. See id. at 535 n.15.
A few states bar CSAAS evidence. See King, 472 S.W.3d at 528-30 (Kentucky) ; State v. Ballard, 855 S.W.2d 557, 562 (Tenn. 1993) ; see also Hadden v. State, 690 So.2d 573, 577 (Fla. 1997) (noting CSAAS had not "been found to be generally accepted"). The Second Circuit has similarly cast doubt on CSAAS evidence. Friedman v. Rehal, 618 F.3d 142, 157 n.9 (2d Cir. 2010) (calling evidence "highly controversial").
The Supreme Court of Kentucky in King explained that no party in Kentucky had ever tried "to establish the validity of the CSAAS theory under either" Frye or Daubert. 472 S.W.3d at 529. In its 2015 ruling, the Court observed that "[t]he validity of the theory was not self-evident in 1985 and it is not self-evident today." Id. at 530.
V. Scientific Review of CSAAS
The remand hearing provided an opportunity to test the principles underlying CSAAS in an adversarial setting. The hearing benefitted not only from the testimony of authoritative experts but also from scientific evidence that has developed in the more than twenty years since J.Q.
*457Throughout the proceedings, both counsel and the experts highlighted a number of relevant scientific studies and articles. We turn next to that body of evidence as it relates to CSAAS as a whole and to its five individual categories.
A. CSAAS in General
The evidence at the hearing identified a number of shortcomings about the concept of a child sexual abuse accommodation syndrome.
First, the label "syndrome" itself raises serious questions. Stedman's Medical Dictionary defines "syndrome" as "[t]he aggregate **290of symptoms and signs associated with any morbid process, together constituting the picture of the disease." 1888 (28th ed. 2006); see also Oxford English Dictionary (3d ed. 2009; updated June 2018) (defining "syndrome" as "[a] concurrence of several symptoms in a disease"). The "Child Sexual Abuse Accommodation Syndrome," though, despite its original description, is not a syndrome. It does not describe a disease, is "not a scientific instrument," and does not have "any pathological or diagnostic implications." Summit II, at 156-57. The term "syndrome," however, can suggest just the opposite -- even though the parties agree that CSAAS cannot determine or gauge how likely it is that sexual abuse has occurred.
One of the State's experts, Dr. Lyon, "deliberately refrained from using the term 'syndrome.' " Even Dr. Summit, in his 1992 follow-up article, acknowledged that he would have chosen a different label had he foreseen how "CSAAS" could be misunderstood. Summit II, at 157.
Second, CSAAS and its five component behaviors are not easy to define with precision. Dr. D'Urso testified that the "constructs" of CSAAS "overlap all the time" and that "many things ... can be subsumed under the five categories." See also O'Donohue & Benuto,5 at 23 ("There is considerable, but indeterminate, overlap among the five key constructs.").
Third, there is disagreement as to how the behaviors relate to one another. Dr. Summit described the five behaviors as a "logical pattern and sequence of interaction." Summit II, at 155. Yet Dr. D'Urso testified they are not "sequential" and indeed need not all occur.
Fourth, experts debated the import of the five behaviors at the hearing. Dr. Summit originally wrote that CSAAS represented "a common denominator of the most frequently observed" behaviors of sexually abused children. Summit I, at 180. On a somewhat **291different but related point, other experts have explained that "there are no gold standard psychological symptoms specific to sexual abuse." London et al. (2005),6 at 194. In fact, the five behaviors can also be found in non-abused children. That caused Judge Bariso to question CSAAS's value as an educational tool "if its core characteristics are potentially equally prevalent among non-abused children or children who have suffered other types of trauma."
Comments by other experts do not resolve the issue. Dr. Lyon noted that the categories of behavior could still be noteworthy "by comparing the relevant proportion in both the abused and non-abused population." Yet, according to Dr. Bruck, some CSAAS symptoms -- recantation and *458denial -- "may be more common in non-abused children than in abused children." See also Elliott & Briere,7 at 273 (finding no behaviors or symptoms that "reliably distinguish an abused child from a nonabused one").
Finally, it is important to note that CSAAS stems from observations made in clinical practice -- not systematic scientific study. Summit I, at 179-80. Dr. Summit's seminal 1983 article reflected the conclusions of a clinical psychiatrist and advocate for victims of child sexual abuse, and not a body of empirical data.
Clinical wisdom is valuable, but it must be examined with care and objectively tested. When Dr. Summit's article first appeared, though, it was not subject to peer review. Even today, Judge Bariso found that expert testimony reveals a "lack of data supporting CSAAS." As Judge Bariso also noted, CSAAS is not recognized in the Diagnostic and Statistical Manual of Mental Disorders and has not been accepted by the American Psychiatric **292Association, the American Psychological Association, or the American Psychological Society.
The record contains limited scientific support for a five-part syndrome. Dr. D'Urso defended CSAAS and asserted that it enjoys "general acceptance within the professional community." Others disagree. For example, Dr. Bruck testified that "there's empirical support that some children show some of [the CSAAS] symptoms some of the time, but that's about it."
Dr. Brainerd surveyed the scientific literature in his expert report. In his view, the authoritative articles by Drs. Bruck, Ceci, and London, discussed below, reveal a lack of empirical support for CSAAS -- and demonstrate a lack of consensus. Based on the literature and other expert opinions, Dr. Brainerd could "not see how any reasonable scientist could fail to conclude that CSAAS is very, very far from" having achieved "a general consensus" in the scientific community.
Another study outlined twenty-one problems with CSAAS -- including that it is vague and has not been scientifically tested -- and described it "as an exemplar of junk science" that "should not be used in any way in any context (particularly in legal settings, where impactful decisions are being made)." O'Donohue & Benuto, at 22-27; see also Zajac et al.,8 at 2. ("[T]here is no scientific evidence of a syndrome-like cluster of symptoms or patterns of disclosure among abused children.").
Based on the record, it does not appear that CSAAS's five-category theory has been tested and empirically validated as a whole. We now consider relevant evidence for each of the five behaviors.
**293B. Individual Components of CSAAS
1. Secrecy
Dr. Summit explained secrecy by noting that "[t]he average child never asks and never tells." Summit I, at 181. He noted that child abuse is a secretive reality for victims: "It happens only when the child is alone with the offending adult, and it must never be shared with anyone else." Ibid. Dr. Taska offered a number of reasons why children do not disclose abuse: obedience, *459threats, gifts and bribes, and fear, among other causes.
In short, victims keep abuse a secret by not talking about it. Delayed disclosure and false denials are other ways of keeping abuse secret. Because Dr. Summit did not consider them under the secrecy prong, we consider them separately below.
2. Helplessness
Dr. Summit identified helplessness as a "precondition[ ]" to abuse, not a behavior. Ibid. The concept appears to state the obvious -- that young children cannot defend themselves against adults. It does not present a scientific finding.
3. Accommodation
Accommodation refers to the coping mechanism by which a child adjusts to sexual abuse. Dr. Summit wrote that victims can accommodate abuse by "hiding any indications of conflict" and being "unusually achieving and popular" -- "the honor student or the captain of the football team," "eager to please both teachers and peers." Summit I, at 186-87. Yet they can also accommodate with "[p]athological dependency, self-punishment, self-mutilation, selective restructuring of reality and multiple personalities." Id. at 184.
As Dr. Taska explained at trial, those "mechanisms" include "the whole range of symptoms" from being "really good ... really obedient ... really good ... in school" to "self-injuring," "using drugs and alcohol," "acting out," "tortur[ing] animals," and "set[ting] fires."
**294Accommodation thus is not a discrete behavior. It encompasses all possible behaviors from the most resilient to the most self-destructive. Under that definition, which the record reflects, all victims fall under the broad construct in one way or another.
4. Delayed Disclosure
Evidence presented at the hearing confirmed that children often delay reporting sexual abuse. Judge Bariso found that delayed disclosure is generally accepted among the scientific community. The record supports that finding.
Indeed, there is consistent and long-standing support in the scientific literature that most child victims of sexual abuse delay disclosure. A 1994 study found that 74.9% of victims did not disclose abuse "within the year that it first occurred, and 17.8% ... waited more than 5 years." Elliott & Briere, at 268. A 2005 study that criticized aspects of CSAAS nonetheless found empirical support for the principle that delayed disclosure "is very common." London et al. (2005), at 220. In a 2008 follow-up to that study, Dr. London and her coauthors found that 55 to 69% of adults reported "that they never told anyone about the sexual abuse during childhood," a finding the authors found "remarkably consistent" across the studies they surveyed. London et al. 2008,9 at 31.
Other studies support that conclusion. See DiPietro et al.,10 at 134 ("[F]ew children purposefully disclose abuse during childhood ...."); Dubowitz et al.,11 at 691 *460("[O]nly 21% of the children **295assessed as moderately or highly likely to have been abused had fully disclosed such abuse and had examination results indicative of abuse."); Lyon, False Denials,12 at 42 (surveying studies that reveal low rates of disclosure and delays in disclosure).
Some studies found higher rates of disclosure among children who were older when they were first abused. See London et al. (2005), at 201-02 (discussing studies); see also Gordon & Jaudes,13 at 321 ("Preschool children are considered the group most underreported for sexual abuse.").
Studies in this area often rely on retrospective memory -- that is, researchers ask about claims of abuse years after the fact. Experts point to shortcomings with that approach. Dr. Brainerd, for example, noted that memories can be falsified between childhood and adulthood, and victims may simply forget that they disclosed earlier. See also London et al. (2008), at 34.
The ACLU contends that the scientific literature should be discounted because it does not expressly show that child sexual abuse causes delayed disclosure. But the case law in this area does not require such a showing in the setting of this matter. To establish reliability under Rule 702 and satisfy Frye's general acceptance test in criminal cases, the focus properly belongs on whether there is a consensus among scientists that a significant percentage of children who have actually been abused do, in fact, delay disclosure. See Harvey, 151 N.J. at 170, 699 A.2d 596 ; see also Torres, 183 N.J. at 568, 874 A.2d 1084 ; Kelly, 97 N.J. at 210, 478 A.2d 364.
5. Retraction
Retraction occurs when a victim truthfully discloses abuse and then recants. Dr. Summit believed that recantation is the norm: "Whatever a child says about sexual abuse, she is likely to reverse **296it." Summit I, at 188. According to Dr. D'Urso, however, recantation is "perhaps the most debated" of the five CSAAS factors.
A 1991 article reported that recantation is "a recognized phenomenon" in child sex abuse cases, "particularly cases of incest." Sorensen & Snow,14 at 14. According to a study done by the authors, children recanted in 22% of cases, and 92% of those children later reaffirmed their allegations. Id. at 11. Dr. London and her co-authors identified a number of questions about the study: (1) how the authors selected the sample cases, many of which were chosen from their private psychological practice; (2) what type of therapeutic methods the authors used to elicit disclosures; and (3) the "high probability" that the sample included children who were involved in now-discredited "satanic ritual abuse" cases.15 London et al. (2005), at 212-13.
Dr. London's 2005 article also surveyed seven studies done after 1990, which showed recantation rates that ranged from 4 to 27%. Id. at 216. "[T]he highest rates of recantation," according to the publication, came from "studies that ha[d] the least *461certain diagnoses of sexual abuse." Ibid. A 2008 update that reviewed ten studies reached the same conclusion. London et al. (2008), at 35-36. The authors concluded that recantation is uncommon and occurs in only a minority of cases, id. at 35; London et al. (2005), at 217, and that higher rates may stem from "highly suggestive interview methods" and "studies of children who made allegations of ritualistic abuse," London et al. (2008), at 44.
A 2007 study of 257 substantiated cases of child sexual abuse found a recantation rate of 23.1%. Malloy et al.,16 at 165. The study **297stressed the need to consider a "range of factors that may influence recantation," including familial pressures and whether a sample involved substantiated cases of abuse. Id. at 167.
A more recent publication concluded that retraction "appear[s] to be the exception rather than the rule." Zajac et al., at 4 (published in 2013); see also McGuire & London,17 at 187 (recognizing dispute about rate of recantation but concluding "recantation is extremely rare").
Disagreement among the experts reflected the ongoing debate. Dr. D'Urso testified that Dr. Summit "overstated the frequency of recantation." In his expert report, Dr. D'Urso also acknowledged that "[r]etraction rates vary among researchers." Dr. Lyon co-authored the study that found a 23% recantation rate. Malloy et al., at 165. Dr. Brainerd observed that children who are suspected of being abuse victims in prospective interview studies "rarely recant." Dr. Bruck, who co-authored a number of studies in the record, wrote in her expert report that only a "small proportion" of victims recant. She also found that "children are more likely to take back a false accusation than a true" one.
In short, evidence presented at the hearing revealed that only a minority of victims recant truthful allegations of abuse and that experts do not agree on the rate of recantation.
C. Denial
The hearing also addressed the extent to which children falsely deny abuse when they are questioned. Denial is arguably a sub-set of "secrecy"; it is a way to maintain a secret. Recantation can also be considered a form of denial. Because Dr. Summit did not analyze denial under CSAAS, we briefly consider the topic on its own.
Denial is another heavily debated subject. Some experts believe that victims often deny abuse. The 1991 study by Sorensen and **298Snow found that 72% of children who eventually disclosed sexual abuse denied the abuse when they were first questioned. Sorensen & Snow, at 14. For reasons noted earlier, however, experts have questioned the reliability of the study. See London et al. (2005), at 212-13.
Dr. Lyon reviewed 21 studies of children with gonorrhea conducted from 1965 to 1993 and found "significant rates of false denials." Lyon, False Denials, at 43. He noted that the rate of disclosure, excluding children younger than three years old, was 42% -- or a corresponding rate of denial of 58%. Id. at 48. Many victims disclosed their abuse incrementally, sometimes only after multiple interviews. Id. at 53-54.
In his expert report, Dr. Lyon conceded that children with sexually transmitted diseases are likely not representative of *462abused children in general. In addition, the relatively dated studies involved -- most from the 1960s and 1970s -- did not follow a standard interview practice, and suggestive questioning may have affected the results. See id. at 52. Moreover, Dr. Lyon's expert report adjusted for studies that included teenagers, who might "have acquired gonorrhea through consensual sex." Id. at 48 n.2. That increased the rate of disclosure to 53% and thus decreased the rate of denial.
Dr. Lyon also described studies his lab conducted in which children who were four to nine years old "play[ed] with a friendly stranger," and "two toys appear[ed] to break in the children's hands" while they were playing. The stranger then asked the child "to keep the breakage a secret." Dr. Lyon found that roughly one-third of the children kept the secret, even after they were asked directly about the breaking. The State, however, acknowledged that the toy study, while relevant, is not a perfect analogy to sexual abuse cases.
Other experts concluded that most children do not deny abuse. Dr. London's 2005 article examined 17 studies and found that "when directly questioned in a formal setting, only a small percentage of abused children" deny abuse. London et al. (2005), at **299217, 220. Most disclose in the first or second interview. Id. at 217. In her expert report, Dr. Bruck, a co-author of the London studies, noted that rates of denial "in assessment interviews were highly variable" and ranged from 4 to 76%. See London et al. (2008), at 35. In her judgment, weaker studies -- that involved children later found to have made false allegations or children who may have been subjected to suggestive techniques, for example -- produced higher rates of denial. Studies with better methodologies produced low rates. See also Zajac et al., at 3-4 ("[D]enial ... during formal investigation appear[s] to be the exception ...."); McGuire & London, at 180-81 (reviewing studies and noting that "majority of abused children who come before authorities will disclose when questioned directly").
Dr. Brainerd also noted in his report that children who are suspected victims of sexual abuse "overwhelmingly disclose that they have been abused" when they participate in forensic interviews.
There does not appear to be a consensus among the experts, or in the scientific literature, on the subject of false denials.
VI. Parties' Arguments
The State argues that CSAAS is scientifically reliable and that expert testimony about it should continue to be allowed in child sexual abuse cases. The State submits that the scientific community generally accepts that delay, denial, and recantation occur. According to the State, CSAAS evidence is not meant to show "how all or even a majority of child sexual abuse victims behave"; it instead explains how the behavior of victims "is not inconsistent with having been molested." The "value" of CSAAS testimony, the State contends, is "to explain a child's sometimes counterintuitive post-abuse behavior or rehabilitate a child whose credibility has been attacked" on account of delayed disclosure, denial, or recantation.
In the State's view, disagreements about CSAAS can be highlighted at trial through cross-examination. The State disagrees **300with the notion that jurors understand how victims of child sexual abuse behave.
Various amici support the State's arguments. The Attorney General maintains that the hearing produced ample evidence that CSAAS is generally accepted within the scientific community. "All that matters," the Attorney General contends, "is that there is reliable evidence that some *463sexually abused children delay and recant." CSAAS is thus not only appropriate, according to the Attorney General, but also necessary to dispel misconceptions and common myths. The American Professional Society on the Abuse of Children and the County Prosecutors Association of New Jersey ably amplify those arguments.
Defendant argues that CSAAS evidence is wholly inadmissible under Rule 702 because CSAAS is unreliable and not generally accepted within the relevant scientific community. Even the State's experts, defendant submits, concede that there is a genuine dispute about CSAAS evidence among scientists. Defendant asserts that the State cannot satisfy its burden by claiming that CSAAS evidence simply "educates" the jury.
Defendant offers an independent reason to exclude CSAAS evidence as well. He claims that what CSAAS purports to explain is not beyond the ken of the average juror. Defendant contends that juror beliefs about child sexual abuse are more favorable to the State than what is supported by scientific findings. Defendant also argues that CSAAS is inadmissible because it improperly intrudes on the jury's role as factfinder. Finally, defendant argues that the admission of CSAAS testimony in his case was not harmless and requires a new trial.
The ACLU contends that CSAAS is not a scientific concept and that no part of Dr. Summit's theory satisfies Rule 702. The ACLU also argues that CSAAS evidence is unduly prejudicial. The Association of Criminal Defense Lawyers adds that trial courts should not use the model jury charge on CSAAS.
**301As noted earlier, the Last Resort Exoneration Project urged the Court to adopt the Daubert test in criminal cases. The New Jersey Association for Justice submits that the Daubert criteria are helpful to assess reliability. Both groups argue that CSAAS is not reliable under Daubert's principles.
VII. Legal Conclusions
We consider the evidence introduced at the remand hearing under N.J.R.E. 702. Courts generally defer to a trial court's credibility findings about the testimony of expert witnesses. See State v. Henderson, 208 N.J. 208, 247, 27 A.3d 872 (2011). We accept the court's factual "findings to the extent that they are supported by substantial credible evidence in the record" but "owe no particular deference to the" court's legal conclusions. State v. Chun, 194 N.J. 54, 93, 943 A.2d 114 (2008).
Whether expert testimony is sufficiently reliable to be admissible under N.J.R.E. 702 is a legal question we review de novo. See Harvey, 151 N.J. at 167, 699 A.2d 596. To assess the reliability prong of the rule, we apply the Frye test and consider whether CSAAS has achieved general acceptance in the scientific community. See Frye, 293 F. at 1014 ; see also Torres, 183 N.J. at 568, 874 A.2d 1084 ; Harvey, 151 N.J. at 170, 699 A.2d 596 ; Kelly, 97 N.J. at 210, 478 A.2d 364.
The evidence presented at the remand hearing answers that question. Judge Bariso found that the State had not met its burden to demonstrate general acceptance of CSAAS as a whole, and that there is consensus for only one type of behavior -- delayed disclosure. We agree.
Based on the expert testimony and body of evidence from the hearing, which are summarized above, the record sustains the following findings. There is limited scientific support for the overall five-part syndrome known as CSAAS. In fact, CSAAS does not describe a "syndrome." See Stedman's Medical Dictionary 1888. As described more fully in section V.A., *464CSAAS's component behaviors are neither precise nor specific to victims of sexual **302abuse. Also, the original study -- based on clinical observations -- has not been empirically validated as a whole. Studies in the record as well as testimony at the hearing confirm that CSAAS as a whole does not enjoy general acceptance within the scientific community.
As to the five individual features, we find consistent and long-standing support in the scientific literature and among experts only for the proposition that a significant percentage of victims of child sexual abuse delay disclosure.
None of the other features that comprise CSAAS have achieved sufficient acceptance in the scientific community to be considered reliable evidence under Rule 702. Two features, in fact, do not describe behavioral traits. Dr. Summit called secrecy and helplessness "preconditions" to sexual abuse. Summit I, at 181. The notion of secrecy overlaps with delayed disclosure, which is considered below. Helplessness is plainly not a behavior; it states the obvious fact that young children cannot defend themselves against adults.
Accommodation encompasses the full spectrum of behaviors -- from well-adjusted young adults who are high-achievers to self-destructive children who are failing. The theory is not based on scientific data; it describes the straightforward reality that all child victims cope with sexual abuse in one way or another. That is undoubtedly true, but by defining "accommodation" so broadly that it includes all forms of behavior, the category both proves too much and very little at the same time.
For the reasons discussed in sections V.B.5. and V.C., we do not find general acceptance in the scientific community about either retraction or false denial. Experts vary widely in their views about how often victims of child sexual abuse retract allegations or falsely deny them. Some conclude that the behaviors rarely occur; others find that roughly one-fourth of victims recant and as many as three-fourths falsely deny abuse. The hearing identified a number of concerns about studies that found higher rates. In any event, the record reveals a lack of consensus about the prevalence of recantation and false denial among victims of abuse. See also **303Kelly, 97 N.J. at 210, 478 A.2d 364 (noting that reliability turns on ability of expert's "mode of analysis ... to produce uniform and reasonably reliable results" grounded in science).
To satisfy the reliability prong of Rule 702, it is not enough to state that certain behaviors can be observed in some victims some of the time -- or that those behaviors are not inconsistent with abuse. That is not the type of "state of the art" evidence the case law requires. See id. at 208, 478 A.2d 364.
Nor is it sufficient for the State to claim that expert evidence is being admitted only to educate jurors and dispel misconceptions, and not as a diagnostic or predictive tool. The underlying claims of the syndrome must themselves be reliable to be admitted under Rule 702. In short, the expert testimony must satisfy Frye's reliability test and find general acceptance in the scientific community.
Because evidence about CSAAS as a whole and the above four categories does not satisfy that standard, experts may not present evidence on those topics at trial. Like the trial court, we reach a different conclusion about delayed disclosure. As discussed in section V.B.4., both experts in the field and the relevant scientific literature confirm that children often delay reporting sexual abuse. The authoritative studies and the testimony at the *465hearing reveal a consensus within the scientific community on this point: most child victims delay disclosing acts of sexual abuse.
To be clear, when the other prongs of Rule 702 are met, the State may present expert evidence on delayed disclosure among victims of child sexual abuse -- and only that evidence -- to a jury. Trial judges must exercise care to limit the testimony and bar any reference to "CSAAS," an abuse "syndrome," other CSAAS "behaviors" aside from delayed disclosure, or causes for delayed disclosure. The testimony should not stray from explaining that delayed disclosure commonly occurs among victims of child sexual abuse, and offering a basis for that conclusion.
**304When expert evidence on delay is introduced, trial courts should provide appropriate limiting instructions to the jury -- both before an expert witness testifies and as part of the court's final charge. We ask the Committee on Model Criminal Jury Charges to draft appropriate instructions limited to delayed disclosure as soon as practicable. We also invite the parties and amici to submit proposed charges to the Committee.
Jury instructions should explain that delay is not necessarily inconsistent with abuse. Evidence about delay is not proof of abuse; nor is it proof that the victim testified truthfully. But it can dispel misconceptions about delayed reporting and may be considered in assessing a witness's credibility. Parts of the current charge on battered woman syndrome,18 as well as the charge on CSAAS, may help inform the Committee. See Model Jury Charges (Criminal), "Battered Woman Syndrome -- Purposes Other Than Defenses" (June 4, 2007); Model Jury Charges (Criminal), "Battered Woman Syndrome -- Defenses" (June 4, 2007); Model Jury Charges (Criminal), "Child Sexual Abuse Accommodation Syndrome."
VIII. Rule 702's "Beyond-the-Ken" Requirement
Proponents of expert evidence on delayed disclosure have to satisfy all three parts of Rule 702. They must not only present sufficient proof that the evidence is reliable, and that the witness has sufficient expertise, but must also show that the testimony "concern[s] a subject matter that is beyond the ken of the average juror." Kelly, 97 N.J. at 208, 478 A.2d 364. As the Rule states, expert testimony may be admitted if it "will assist the trier of fact to understand the evidence or to determine a fact in issue." N.J.R.E. 702.
**305Under the Rule, expert testimony is not appropriate to explain what a jury can understand by itself. See State v. Cain, 224 N.J. 410, 427, 133 A.3d 619 (2016). Matters "within the competence of the jury" are for the collective wisdom of the jury to assess. See State v. Sowell, 213 N.J. 89, 99, 61 A.3d 882 (2013). By contrast, issues that are beyond the understanding of the average juror may call for expert evidence. Trial judges, as gatekeepers, decide that threshold question. Ibid.
Whether a victim's delayed disclosure is beyond the ken of the average juror will depend on the facts of the case. If a child witness cannot offer a rational explanation for the delay in disclosing abuse -- which may happen during the pretrial investigative phase or on the witness stand -- expert evidence may be admitted *466to help the jury understand the child's behavior. In this context, we do not accept that jurors can interpret and understand an explanation that is not offered.
On the other hand, a young teenager's explanation from the witness stand may fall within the ken of the average juror and might be assessed without expert testimony. In this case, Bonnie gave sound reasons for the delay. She testified that she did not tell her mother about the abuse sooner because (a) defendant threatened her with a gun, (b) she was embarrassed by the degrading experiences, and (c) she feared that her mother would kill defendant and be sent to prison, based on how her mother reacted when a friend spotted defendant lying on top of Bonnie on a prior occasion.
No juror needed help from an expert to understand and evaluate Bonnie's testimony. In a case like this one, expert testimony is not called for to "assist the trier of fact." See N.J.R.E. 702. Counsel, of course, may still argue to the jury about a victim's delayed disclosure based on the evidence, reasonable inferences that can be drawn from it, and common sense.
**306IX. Application
The jury in this case heard about CSAAS at various points in the trial. In addition to Dr. Taska's detailed testimony, counsel referred to CSAAS in opening argument and summation, and the trial court used the model jury charge to instruct the jury twice on CSAAS evidence.
Applying the above findings to this case, it was error to admit testimony about CSAAS -- both as to the theory in general and the behaviors that are not generally accepted by the scientific community. We also have serious concerns about the admissibility of expert testimony on delayed disclosure in this case because Bonnie, a teenager, gave reasons for the delay that were not beyond the ken of the average juror.
Nonetheless, we find those errors harmless in light of the overwhelming evidence of defendant's guilt. An error is harmless unless, in light of the record as a whole, there is a "possibility that it led to an unjust verdict" -- that is, a possibility "sufficient to raise a reasonable doubt" that "the error led the jury to a result it otherwise might not have reached." State v. Macon, 57 N.J. 325, 335-36, 273 A.2d 1 (1971).
The jury heard powerful evidence of defendant's guilt on all four counts for which he was convicted. The victim testified about a series of escalating acts of sexual abuse. She explained that she used her iPhone to record an encounter during which defendant sexually abused her. The recording corroborated her testimony. Explicit and disturbing language captured on the recording -- in words defendant admits were his own -- graphically confirm the victim's description of an act of sexual abuse by defendant. Law enforcement also monitored phone conversations between the victim and defendant in which he offered her money and other items after asking her to retract her accusations. The jury heard as well from a friend of the victim's mother who once visited the family apartment and found defendant lying on top of the victim, clothed but noticeably erect.
**307Count One of the indictment charged first-degree aggravated assault and alleged that defendant committed "an act of sexual penetration." Count Two charged third-degree aggravated criminal sexual contact and accused defendant of committing "an act of sexual contact." In light of the evidence presented -- in particular, the testimony of the victim combined with a graphic audio recording of an act of sexual *467abuse -- we do not find a sufficient possibility that the admission of CSAAS evidence led the jury to an unjust verdict or one it might otherwise not have reached. Ibid. We note that the recording captured an event that took place only weeks before the victim revealed the alleged abuse to her mother and the police. That fact lessened the impact of the CSAAS evidence in this case.
During closing argument, defense counsel conceded that there was sufficient evidence to support Count Three, second-degree endangering the welfare of a child. Counsel's strategic concession was no doubt influenced by the strength of the evidence, including the iPhone recording.
Finally, the CSAAS evidence had little if any bearing on the third-degree witness tampering charge. To prove Count Nine, the State relied not only on Bonnie's testimony but also on the recorded phone conversations in which defendant offered her money and gifts to withdraw her accusations. CSAAS evidence did not likely affect that conviction.
For all of those reasons, we find that the admission of the CSAAS evidence in this case was harmless.
X. Judgment
Allegations of child sexual abuse are particularly sensitive and can be difficult to prove. They also carry a powerful stigma and can be hard to defend against.
In this and all areas, the Judiciary must ensure that proceedings are fair to both the accused and the victim. Trial judges partly fulfill that responsibility by serving as a gatekeeper. In that **308role, they must assess whether expert testimony is sufficiently reliable before it can be presented to a jury.
In 1983, the concept of a Child Sexual Abuse Accommodation Syndrome was introduced in Dr. Summit's article. This Court adopted CSAAS in 1993, like most courts in the nation. Today, we have the benefit of more critical and thorough scientific analysis of CSAAS, which cautions against its continued use. Based on the record before the Court, we conclude that CSAAS does not satisfy a basic standard of admissibility -- reliability -- because it is not generally accepted by the scientific community. Expert testimony about CSAAS therefore may no longer be presented to juries. Only testimony about delayed disclosure can be admitted in appropriate cases. In this appeal, however, the evidence was harmless in light of overwhelming proof of defendant's guilt.
For all of those reasons, we modify the judgment of the Appellate Division and affirm defendant's convictions.
JUSTICES LaVECCHIA, ALBIN, PATTERSON, FERNANDEZ-VINA, SOLOMON, and TIMPONE join in CHIEF JUSTICE RABNER's opinion.