Sex, Lies, and Statistics: Inferences From the Child Sexual Abuse Accommodation Syndrome

Traditions and Trends

Courts have admitted testimony about the effects of psychological trauma with the expectation that witnesses will be subject to cross-examination.¹ Criminal and civil courts have admitted testimony about posttraumatic stress disorder (PTSD), a diagnostic category that incorporates causation (i.e., exposure to a traumatic event) as a diagnostic criterion.² Testimony has been admitted to explain common behavioral health sequelae of sexual assault including severe anxiety, agitation, and crying spells characteristic of rape trauma syndrome (RTS).³ Criminal and civil courts have also admitted testimony that explains counterintuitive behaviors, such as the delayed and often unconvincing disclosure that is characteristic of child sexual abuse accommodation syndrome (CSAAS).⁴ A clinician testifying that an evaluee has one of these conditions and that the only explanation for it is the criminal conduct of the defendant would tend to prejudice a jury.⁵ For this reason, such testimony is admitted only to explain victim behavior and is inadmissible on the issue of guilt.^6,7 Similarly, in syndromes reported by some criminal defendants, including battered spouse syndrome,⁸ where victim and defendant roles have been conflated, expert witness testimony may be helpful in providing a context for violent behavior that is arguably self-defense. Overall, however, evidence of syndromes in court proceedings has been criticized as a major source of confusion, especially in sexual assault cases.⁹ This problem is particularly true of child sexual abuse and CSAAS testimony. Although often helpful to fact finders, descriptions of syndromal types may lack validity unless coupled with clinical examinations. It has been pointed out, for example, that social science testimony on types of individuals who give false confessions (nomothetic inference) may not be as persuasive as testimony based on personal examination of a person (idiographic inference).^10,11 That is, while understanding theoretically why some suspects confess falsely, the testimony will lack weight unless it is linked to clinical data. On the other hand, as we will see, diagnoses of informal syndromes in victims run the risk of misleading juries.

Historically, victim testimony itself has vexed both psychiatry and justice. Twentieth-century jurisprudence evinced strong suspicion both of children's testimony generally and of rape complainants in particular.^12,–,14 There are numerous uses of statistical inferences in criminal and civil cases, ranging from DNA fingerprinting to cheating on examinations.¹⁵ Statistical inferences in psychiatry have become an important basis for the prediction and interpretation of behavior, with actuarial data gaining momentum.^16,17 Increasing attention has been paid to statistical information in The Journal.^18,–,20 Whereas actuarial tools and statistical inferences may enhance, or be superior to, clinical judgment, when it comes to human behavior, the gold standard is still testimony based on clinically derived data. In this article, we review CSAAS by way of a recent New Jersey Supreme Court opinion on the extent to which a purely statistical argument can be used to assess a child victim's credibility.

Child Sexual Abuse Accommodation Syndrome

CSAAS was first described by Summit⁴ as a way to explain why some sexually abused children do not appear to have traumatic symptomatology and how their coping behavior may be misunderstood and misinterpreted by caregivers, law enforcement, and the courts. Summit recognized that disbelief and rejection of the child's experience are destructive to development: “At a time when the child most needs love, endorsement and exculpation, the unprepared parent typically responds with horror, rejection and blame” (Ref. 4, p 179). Yet, the abused child must adapt in the face of adversity to devote cognitive and socioemotional resources to developing typical competencies. Summit explained the consequences: The accommodation process intrinsic to the world of child sexual abuse inspires prejudice and rejection in any adult who chooses to remain aloof from the helplessness and pain of the child's dilemma or who expects that a child should behave in accordance with adult concepts of self-determinism and autonomous, rational choices [Ref. 4, p 179]. Summit's goal was to enhance our understanding of victims, to give them a voice, and to provide a context for understanding their coping behavior within the family and systems of child protection and criminal justice.

His schema lists five facets of CSAAS: secrecy; helplessness; entrapment and accommodation; delayed, conflicted, and unconvincing disclosure; and retraction. The first two are preconditions of sexual abuse, and the remaining three are sequential contingencies. Preconditions are understood to set the stage for the initiation and continuation of sexual abuse. Secrecy, the first precondition, is an intrinsic characteristic of child sexual abuse, as it virtually always occurs when perpetrators are alone with their child victims. Helplessness, the second CSAAS aspect, refers to the power imbalance between children and adult perpetrators and is a factor in both the initiation of sexual assault and maintenance of secrecy. Perpetrators may be in control of material resources, have influence on significant persons in children's lives, or have the power to make decisions affecting their victims. Secrecy and helplessness may also be enforced through explicit threats of violence to child victims, family members, and even pets. In addition to being compelled to maintain secrecy, children may experience a range of emotions, including embarrassment, shame, feelings of responsibility for the abuse, and fear of not being believed and of punishment and retaliation. Disclosure may also be difficult when children perceive that the environment would not be supportive (e.g., high levels of family stress or dysfunction).

Children as Victims and Witnesses

The forces militating against the victim may be enormous and help to maintain the secrecy of child sexual abuse. A helpless, isolated child, feeling threatened or emotionally bullied, would hardly be in a position to undertake the role of accuser. As noted, the first two aspects of CSAAS are preconditions that are the context for the sequential contingencies of entrapment and accommodation, delayed disclosure, and retracting allegations. Entrapment and accommodation occur when child victims are unable to breach the secrecy of sexual abuse because of their relative powerlessness and coerced participation. Children subjected to repeated sexual assaults and increasing sexual demands and who have no means to end the situation must eventually organize and make sense of their experiences to survive and restore cognitive equilibrium. The secrecy of child sexual abuse, the relative helplessness of the victims, children's eventual accommodation, and any consequent psychopathology are understood to be factors in delayed disclosure and unwillingness to disclose at all. Despite the relationship of abuse and onset of psychopathology, the presence of psychiatric conditions may not be presented as evidence in the prosecution of a criminal defendant.

Even when children disclose abuse under this prohibitive set of circumstances, investigations are typically hampered by additional challenges. Many types of sexual abuse do not leave physical evidence, and few forensic medical examinations of child victims are conclusive for abuse.²¹ The absence of confirming medical findings or corroborative witnesses leads investigators to rely on the testimony of victims and defendants.^22,23 Younger children and those with developmental delays may have limited cognitive and linguistic ability to recognize the abuse clearly and report it coherently. Such limitations, combined with delayed or incremental disclosures, may raise credibility concerns. However, to bar their testimony as incompetent, per se, would be another obstacle to prosecution.

Children are often reluctant to disclose abuse committed by individuals who are known, trusted, and loved.²⁴ This set of conditions also leads to challenges to children's credibility. Summit illustrates: It is sad to hear children attacked by attorneys and discredited by juries because they claimed to be molested yet admitted they had made no protest nor outcry. The point to emphasize here is not so much the miscarriage of justice as the continuing assault on the child. If the child's testimony is rejected in court, there is more likely to be a rejection by the mother and other relatives who may be eager to restore trust in the accused adult and to brand the child as malicious [Ref. 4, p 183]. Summit notes that children who perceive that help is not forthcoming from family members, child protective services, or the courts, may retract their allegations. High levels of family stress that typically follow disclosure and invalidating responses by nonoffending caregivers may result in further coercion of child victims, explicitly or implicitly, into submitting to adult demands by recanting allegations and continuing the victimization. In this way, secrecy is maintained, and the children's continued helplessness, entrapment, and accommodation are perpetuated.

Developments in CSAAS

CSAAS, as an empirically derived nondiagnostic description of behavior, has enjoyed wide acceptance within criminal justice. Summit's argument for validity included statistically supported assumptions emerging from clinical work; four years of testing in the author's practice; strong endorsements from victims, offenders, and family members; and consensus derived from hundreds of training symposia. Still, after about 30 years of field validation, courts struggle with balancing testimony on CSAAS-related behaviors against the implication that the child was, in fact, abused.⁵ In a follow-up to his 1983 article, Summit²⁵ observed that prosecutors had used evidence of CSAAS as proof that children had been abused. He reiterated that CSAAS was not intended to be a diagnostic tool. Since he first introduced and later clarified the use of the CSAAS model, a body of research has emerged examining in more detail the characteristics of children's disclosure.

Subsequent research supports Summit's original observation that child victims often maintain secrecy and delay disclosure or do not disclose at all during childhood. London and colleagues²⁶ reviewed 11 studies of adults who had been sexually abused as children. Across the studies, 60 to 70 percent of adult participants did not disclose abuse during childhood. In another review,²⁷ they found that across the studies, 55 to 69 percent of adult survivors of child sexual abuse said that they did not disclose the abuse during childhood. Disclosure was delayed from one month to over five years. In other reviews examining the findings of nationally representative surveys, 28 to 31.7 percent of participants reporting histories of child sexual abuse had not disclosed until the survey interviews, and 28 to 47 percent reported delays ranging from over one month to five years.^28,29 These findings confirm that children frequently delay reporting or do not report sexual abuse and that official crime and child protective services statistics are likely to underestimate abuse prevalence. These aggregate findings support a role for expert testimony about CSAAS, to provide a context for jurors in assessing victims' behavior.

The literature is less clear in its support of Summit's position that children's disclosures often appear unreliable and that children frequently recant. For example, London and colleagues²⁷ found a 4 to 27 percent rate of recanted allegations in confirmed cases of child sexual abuse across 10 studies, markedly lower than that reported for delayed disclosure. However, one study of 102 assaults videotaped by the offenders and examined alongside the victims' statements produced results consistent with Summit's findings.³⁰ In this study, the severity ratings of the children's interviews were significantly lower than the ratings derived from the videos. That is, the confirmed victims significantly minimized or denied sexual abuse. Thus, there is some support for juries' hearing from expert witnesses on varieties of accommodation.

The literature confirms demographically-based inferences on the likelihood of victim reporting. For example, males are less likely than females to disclose abuse, because of a range of factors, such as concerns about being identified as homosexual (when the perpetrator is male) and the cultural belief that males cannot be victims (when the perpetrator is female).^31,–,33 Children who are older at the time of the forensic interview may be more likely than younger children to disclose abuse, in part because of differences in verbal and cognitive development.^30,33 Family characteristics also correlate with children's disclosures. Studies of adults who had been sexually abused as children by a parent or relative found lower rates of disclosure than in children who were abused by outsiders.^14,27 Children's fear of physical harm, expectations of negative emotional experiences, concerns that the perpetrator would be incarcerated, and perceptions of low maternal support have been associated with lower levels of disclosure.^33,34 Similarly, recanting allegations may be more likely when children have been sexually abused by a family member, when the nonoffending caregiver was unsupportive, or when the victim was younger than 10.³⁵

Admissibility of Testimony

Given the mixed findings on children's disclosure in the research literature and the questionable use of CSAAS in some criminal proceedings, testimony on CSAAS has been subject to evidentiary challenges, either to its general acceptance (Frye test) or replicability (Daubert test). Under the Federal Rules of Evidence and the states' rules that shadow them, Rule 703 is invoked to determine the reliability of CSAAS testimony.³⁶ That is, if there were clinical validity to CSAAS, it would not necessarily be helpful in adjudicating the guilt or innocence of a criminal defendant. Under the rule, the probative value of the testimony would have to outweigh its prejudicial effect. As Flint observes, “Even assuming that CSAAS could definitively indicate that a child was abused, it still says nothing about when the child was abused or by whom” (Ref. 5, p 177). Hence, while CSAAS testimony is often essential in helping courts to understand children's responses to sexual abuse, judges also show hesitation in permitting juries to consider the syndrome. Our impression was that although such testimony would rarely be barred by rule, its scope and implications have often been placed on a short tether by trial courts.

Various jurisdictions have ruled on the admissibility of CSAAS evidence. In Lowe v. Walker,³⁷ Mr. Lowe petitioned for a writ of habeas corpus based on violation of his right to due process. He argued that despite cautionary remarks given to the jury, CSAAS evidence admitted during trial was junk science and not in the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition, Text Revision (DSM-IV-TR).³⁸ The California appellate court rejected Mr. Lowe's petition affirming that CSAAS may be admitted to explain the behavior of child victims. In another California appeal, People v. Wells,³⁹ the court affirmed the admissibility of CSAAS evidence to explain children's behavior, while barring testimony to the effect that the child did not look traumatized on the video. In W.R.C. v. State,⁴⁰ the defendant unsuccessfully appealed his conviction based on lack of consensus in the scientific community regarding children's responses to abuse. In this case, the victim waited 10 years to disclose, and the prosecution expert testified as to the frequency of delayed disclosure among child victims. These cases support the conservative use of CSAAS to explain children's reactions to abuse and to address common misunderstandings about how they respond. This application was underscored in a 2010 California opinion about whether the expert could opine about the frequency of false allegations by children.⁴¹ Besides addressing the need not to conflate the issues of CSAAS and false reports, the court pointed out that, since the expert's use of an imperfect statistic did not contaminate the question of guilt or innocence, it was permissible.

Legislatures have enacted statutes permitting expert testimony on sexual assault matters, presumably to enable prosecution. In Pennsylvania, within days after the guilty verdict in the 2012 Jerry Sandusky (Penn State) trial, the governor signed into law such a bill,⁴² making Pennsylvania the last jurisdiction to do so. The law gives potential witnesses the ability to … assist the trier of fact in understanding the dynamics of sexual violence, victim responses to sexual violence and the impact of sexual violence on victims during and after being assaulted. If qualified as an expert, the witness may testify to facts and opinions regarding specific types of victim behaviors. The witness' opinion regarding the credibility of any other witness, including the victim, shall not be admissible [Ref. 42].

State of New Jersey v. W.B.: Misleading Statistics?

In 2011, the Supreme Court of New Jersey announced opinions in a complex matter involving, among other things, the admissibility of CSAAS testimony. In State v. W.B.,⁴³ the question was not about the overall admissibility of CSAAS evidence; rather, it was about the use of a statistical inference to suggest that the victim was telling the truth. Relying on prior decisions, the court said: Simply stated, CSAAS cannot be used as probative testimony of the existence of sexual abuse in a particular case. Therefore, introduction of such testimony will be upheld so long as the expert does not attempt to “connect the dots” between the particular child's behavior and the syndrome, or opine whether the particular child was abused [Ref. 43, p 200, citations omitted]. New Jersey had dealt with the admissibility of CSAAS several times. The benchmark for inadmissible testimony was set in State v. J.Q,⁴⁴ in which the high court condemned expert testimony that directly inculpated the defendant: “The final question to the witness was: ‘Doctor, based on your examination of the girls can you give this jury your expert opinion as to whether or not both were sexually abused?’ Answer: ‘I believe that they were sexually abused’” (Ref. 44, p 1208). Following J.Q., New Jersey courts have not permitted experts to connect the dots between characteristics of victims and guilt of defendants, restricting testimony to rehabilitating victims' testimony.⁴⁵

If jurors are to hear CSAAS testimony, how are they to regard it? In State v. P.H.⁴⁶ New Jersey's supreme court ruled that a defendant had a right to have the jury fully weigh the victim's credibility. As it was, at trial the jurors had been instructed both on CSAAS testimony and on delay in testimony, which was confusing. In the latter instance, by following the judge's charge, the jurors would have been barred from regarding a delay in reporting abuse as diminishing the victim's credibility. Then, in 2008, the same court ruled on a related question. In State v. Schnabel ⁴⁷ the state used CSAAS testimony to help explain a delay in reporting. The defendant was aware that the two teenaged victims had experienced prior sexual abuse at the hands of their brother. Accordingly, he asked for an exception to the state's Rape Shield Law (whereby evidence of the victim's previous sexual conduct is presumed inadmissible at trial), but his request was denied by the trial and appeals courts. The high court found that the Rape Shield Law did not apply and that the defendant's argument that CSAAS could have begun with the prior abuse compelled the jury to consider it. Thus, before considering the issues in W.B., the court had attempted to fashion a climate in which the victim's behavior and the defendant's guilt could be assessed with minimal cross-contamination.

In the W.B. case, the alleged perpetrator, W.B., the 16-year-old female victim's stepfather, was accused by the victim of sexually assaulting her at age 14. The child, D.L., reported that her cousin had also assaulted her. She signed a sworn statement, and W.B. confessed. At trial, D.L. recanted, saying she made a false statement because the defendant and D.L.'s mother did not approve of her relationship with her former boyfriend and wanted to interfere. The issues at trial were complex and included the victim's prior inconsistent statements, evidence of CSAAS, testimony by the victim's boyfriend, and the defendant's confession.

The prosecution called a psychologist, Dr. Richard Coco, to explain CSAAS to the jury. Testifying that he had no opinion about whether CSAAS applied to D.L., he correctly asserted that the presence of the syndrome is neither a diagnosis nor an indication that sexual abuse occurred. Dr. Coco had neither interviewed the victim nor read her statements. Thus, it was never his intention to diagnose sexual abuse. In his testimony, the psychologist asserted that only 5 to 10 percent of sexually abused children lied about it and that those who falsely reported abuse tended to be younger. He added: “And in many ways they actually under report what actually happens to them, in terms of the details of the actual abuse incident” (Ref. 43, p 201). He inferred from the literature that younger children were less likely to report abuse accurately and consistently. Accordingly, the defense attorney on re-cross asked for statistics on 16-year-olds. Dr. Coco could not supply a specific number, but responded that … an adolescent has more ability to do things with information than a younger child does so that they would be more capable of fabricating something than a younger child would…. Adolescents have more ability to fabricate stories, but that doesn't necessarily mean that they do it more frequently in regards to sexual abuse [Ref. 43, pp 201–2]. The testimony appears to say that, while younger children cannot be relied on to report abuse, older children are more in control of the narrative and would have the capacity to fabricate abuse. In neither case, however, would these generalizations apply to a specific person, and Dr. Coco was careful to avoid saying so.

The testimony was permitted and the defendant was convicted. W.B. appealed and his conviction was upheld.⁴⁸ His arguments included that his statement, given in the middle of the night while he was intoxicated and deprived of sleep and food, was coerced. The appellate court rejected W.B.'s assertions, finding that the police acted in accordance with common practice and that the Miranda waiver was properly executed. The Supreme Court of New Jersey heard various arguments including the admissibility of Dr. Coco's CSAAS testimony and whether it was harmful to the defendant. The court's opinion relied on its prior decisions on the admissibility of CSAAS, notably State v. P.H.,⁴⁶ the basis for New Jersey's Model Jury Instruction on CSAAS, which was read to the jury in the instant case (see the Appendix). Though the majority opinion concluded that the introduction of Dr. Coco's testimony was harmless error, it was clear that the testimony was inadmissible: Dr. Coco's testimony included an assertion that only 5–10 percent of children exhibiting CSAAS symptoms lie about sexual abuse. Such testimony creates an inference that D.L. told the truth in her original accusation, despite her motives to fabricate the allegations, and notwithstanding her trial testimony recanting them. Certainly, that is not the purpose of CSAAS testimony or the reason for its admission. Even Dr. Coco so acknowledged. Accordingly, we hold that expert testimony about the statistical credibility of victim-witnesses is inadmissible. Statistical information quantifying the number or percentage of abuse victims who lie deprives the jury of its right and duty to decide the question of credibility of the victim based on evidence relating to the particular victim and the particular facts of the case. Any CSAAS expert testimony beyond its permissible, limited scope cannot be tolerated [Ref. 43, p 202]. Justice Albin's dissent argued, among other things, that the introduction of Dr. Coco's testimony was not harmless and that the defendant should have been granted a new trial. Commenting directly on how the testimony could have led to the defendant's conviction, he said: The jury was permitted to convict defendant based on a simple syllogism totally unrelated to the evidence: if ninety to ninety-five percent of sexual-abuse complainants tell the truth, then D.L. by the laws of statistical probability must have been telling the truth when she reported the sexual abuse to the police; and if D.L. was therefore truthful, then defendant must be lying and guilty of the crimes charged. The defense loudly objected; however, no correction was made by the trial court. The jury was never told that it could not draw the obvious damning inferences that flowed from Dr. Coco's use of statistics to bolster D.L.'s initial complaint. The majority concedes that this testimony was improper and did not fall within the realm of CSAAS evidence, but claims that the error was harmless [Ref. 43, p 210].

Commenting on other instances in which witnesses exceeded their proper limits, Justice Albin noted that, in a 1993 case,⁴⁴ the court … disapproved of a CSAAS expert touting the credibility of an alleged sex victim whom the expert had interviewed. Opining on the alleged victims' credibility was an error that went to the heart of the very integrity of the proceedings, and therefore was not deemed harmless. How much worse in this case where the expert—based on mere statistics—placed his authoritative imprimatur on the credibility of D.L. [Ref. 43, p 210, citations omitted].