Here’s the scenario, in your trial the counterintuitive expert testifies she has forensically interviewed over 21,000 children.

Trick question: “How many of the 21,000 children had unproven allegations?”

Answer: “All of them.”

When the counter-intuitive expert describes what she has learned from thousands of forensic child interviews, she is potentially misleading the jury. Correctly understood none of the children have been confirmed victims of abuse at the time of their interview. Science shows that 68% of children’s’ claims of sexual abuse are ultimately determined to be unfounded or false (Carnes, 2001, described below). So the expert who describes certain behaviors seen in children suspected of abuse is statistically more likely to be describing behaviors of children who make unfounded or false reports.

Null hypothesis

(NH; symbol: H₀) a statement that a study will find no meaningful differences between the groups or conditions under investigation, such that there is no relationship among the variables of interest and that any variation in observed data is the result of chance or random processes. For example, if a researcher is investigating a new technique to improve the skills of children who have difficulty reading, the null hypothesis would predict no difference between the average reading performances of those children who receive the intervention and those who do not. The NH is contrasted with the alternative hypothesis, which is a prediction of a significant finding (e.g., a significant difference between sample means, a correlation that is significantly different from zero). Statistical procedures are applied to research data in an attempt to disprove or reject the NH at a predetermined significance level. See hypothesis testing. APA Dictionary of Psychology

Studies on Unproven and False Child Abuse Claims

In her dissertation, Dr. Wendy Dutton cautions:

“One of the best methods for approximating the scope of CSA has been retrospective surveys of randomly selected adult populations. However, the surveys conducted to date show considerable variability, best explained by differences in research methodology (London et al., 2005). The survey population, the method of administration, questions asked, and definitions of sexual abuse appeared to influence the incidence data reported (Finkelhor, 1994a). The accuracy of survey research has also been impacted by the ability or willingness of individuals to truthfully report their sexual abuse histories (Finkelhore, 1994a).”

Gender differences in children’s disclosures and legal narratives of sexual abuse.Wendy A. Dutton. A dissertation presented in partial fulfillment of the requirements for the degree doctor of philosophy.  Approved April 2011, Madelaine Adelman, Chair. Arizona State University.  Page 14.

BOTTOM LINE
Multiple studies describe unfounded/unproven and false rates of child sexual abuse claims.  These are further described below in the DEEP DIVE. Bear in mind that unfounded/unproven is different than false. Often counterintuitive experts will avoid discussing unfounded/unproven rates and focus on the rates of “false”, which is actually misleading. Don’t fall for their misdirection.

A significant study (Carnes, 2001) followed forensic child interviews in 18 states for two years. After repeated forensic interviews of 147 children, 55.5% of the children’s sexual abuse claims were determined to be unfounded or false. 64 children (45.5%) were determined to have credible disclosures and be forwarded for prosecution. But of the cases that were forwarded for prosecution, only 73% (47 children) were ultimately found to be proven in a court of law. Bottom Line – of all 147 allegations, only 32% were found to be proven. The other 68% were determined to be unfounded or false. 

So when the  forensic child interviewer says they are basing their testimony upon their hundreds of forensic child interviews, over two-thirds of their experience actually reflects unfounded / false claims of abuse. Their behavioral observations are more likely to reflect an unfounded or false allegation of abuse. This is an incredibly important distinction.

Here are important studies. 

A) Connie Nicholas Carnes, et al. Extended forensic evaluation when sexual abuse is suspected: a multisite field study. Child Maltreatment (2001) 6: 230 

TIP #1: Dr. Wendy Dutton cites the Carnes study in references for court testimony (as of March 5, 2022). She also cited this study in her 2011 dissertation. For years she has also cited her dissertation as a basis for her testimony. By both citations, she has based her testimony in part of on the the findings of Carnes. 

TIP #2: The judge won’t allow for the mention of percentages in the trial phase. But it is allowed in your motion to suppress / limit or a Daubert challenge. In the trial phase you can try phrases like “a majority of investigated cases” or “a significant portion of investigated cases.” 

B) 2.98 million American children underwent a Child Protective Services investigation (or alternative CPS response) for an allegation of child abuse or neglect in 2010.  One-fifth of the claims met the legal requirements of abuse. In the other four-fifths of the claims, the investigation determined that there was not sufficient evidence under state law to conclude or suspect that the child was maltreated or at-risk of being maltreated.

Administration for Children and Families. Child Maltreatment 2010. Washington DC. Department of Health and Human Services. Page 6. http://www.acf.hhs.gov/programs/cb/pubs/cm10/index.htm

Administration for Children and Families. Child Maltreatment 2010. Washington DC. Department of Health and Human Services. Table 3-2. http://www.acf.hhs.gov/programs/cb/pubs/cm10/index.htm

Administration for Children and Families. Child Maltreatment 2010. Washington DC. Department of Health and Human Services. Pages 7-8. http://www.acf.hhs.gov/programs/cb/pubs/cm10/index.htm

TIP: Keep in mind this is a combination of physical abuse and sexual abuse claims (not just physical). 

C) Mikkelsen (1992) cites a large retrospective review involving all 576 complaints of sexual abuse made to the Denver, Colorado, Department of Social Services in 1983, Jones and McGraw utilized the innovative strategy of breaking away from a strict Yes/No dichotomy and put forward four categories that capture the ambiguity of many of these situations. Their four categories and the frequency of each are as follows: confirmed = 53%; suspicious but not conclusive = 17%; insufficient evidence to reach a conclusion = 24%; and, definite false allegations = 2%. (43% were unfounded or false). 
False sexual-abuse allegations by children and adolescents: Contextual factors and clinical subtypes. Mikkelsen, et al. (October 1992). American journal of psychotherapy, vol. XLVI, No. 4.

Another important variable that was analyzed in Mikkelsen’s review was the age of the child as relates to the frequency of false allegations. They cite the research summarized by Everson and Boat that showed the relative frequencies for different age groups concerning false allegations. [N for each group indicated in ()]: 1 to 2.9 years of age (N = 124) 6%; 3 to 5.9 years of age (N = 301) 7%; 6 to 11.9 years of age (N = 414) 4.3%; and 12 to 17.9 years of age (N = 410) 8.0%. Thus, it would appear that the risk of an allegation being false is greater with older children. Mikkelsen, et al. (1992)

TIP: Counterintuitive experts normally do not cite Mikkelsen. It is important that this study be part of your cross. Consider providing the expert the Mikkelsen study during your initial interview prior to trial.

DEEP DIVE

Rate of Unproven and False Reports

Research on child sexual abuse typically involves anecdotal surveys about reactions to self-reported sexual victimization in the past.  And herein lies one of the fundamental flaws in the research – there is no accounting for subjects who are mistaken or lying about sexual assault (see the article on Discriminant Validity).  So, in your motion to suppress / limit or Daubert challenge, consider working to establish research findings for what percent of sexual assault claims are proven, unproven or false (the judge should allow percentages to be discussed in pretrial motions or Daubert, but not in the trial phase. 

We begin with a basic question: “What percent of sexual assault claims have been found to be unproven or false?”  Unproven is different than false. 

A) “What percent of sexual assault claims have been found to be unproven?”

This is intuitively the best way to ask the question.  Law enforcement investigates a suspected crime and gives their findings to prosecutors, who then determine whether charges are likely to remain unfounded or can be proven.

B) “What percent of sexual assault claims have been found to be false?”

The American legal system normally does not determine if a complaining witness gave a false report. Most often, the legal system accepts the finding of  “guilty” or “not guilty” and moves on to other cases. Very rarely does an investigating agency determine that a claim was false. 

The American legal system doesn’t seek to prove the negative – i.e. a jury does not determine “a defendant is innocent of the crime in question.”  Our legal system considers the accused as innocent until proven guilty and at the end of a trial might be found not guilty of the specifications. The judge or jury does not determine the defendant is innocent and the charges were “false”. The charges remain “unproven”. This is an important “slight-of-hand” that CSA experts will play – they will describe low “false” rates and fail to mention the much larger “unproven/unfounded” rates.

It is difficult to estimate the proportion of true-to-unproven-to-false accusations, due in part to inconsistent definitions of what constitutes an unproven or false accusation (Saunders, 2012).  But research has discovered unproven or false reports occur on a regular basis. Here’s some of the more prominent studies. 

Establishing ground truth in studies of child sexual abuse.

Steven W. Horowitz, Michael E. Lamb, Phillip W. Esplin, Tascha D. Boychuk, Lisa Reiter-Lavery, and Orit Krippen.

Expert Evidence, January 1995. 4(2), 42-51.

Abstract:

Reports of field research on child sexual abuse can be difficult to interpret when the criteria for inclusion of subjects in abused, non-abused, and false allegation groups are not specified. Since no single reliable and valid method exists for establishing what really did or did not occur to alleged victims (ground truth), multiple, convergent, and explicitly stated criteria are required. In this paper, examples from published field research are used to illustrate the advantages of specifying such criteria. Nine sources of information that could contribute to a decision about ground truth in child abuse cases are then described. Procedures for assessing the validity of these sources of information are then suggested. 

Three general categories merit differentiation in research on sexually abused children; abused children, non-abused children about whom no allegations have been made, and non-abused children about whom false allegations have been made. Unfortunately, although most researchers have sought to compare children in the first two groups, they have not always excluded children in the third group. Differentiation between the last two (nonabused) groups is necessary because false allegations may lead to invasive and disruptive investigations that themselves affect children’s behavior and performance.  (p. 43)

If only a small percentage of cases can be absolutely confirmed, studies that deal only with such cases risk threats to external validity, because the cases included are so unrepresentative . . . (p. 43)

The authors list nine sources of information for validation of child sexual abuse.

1. Medical evidence (certainty, consistency, quality)
2. Confessions (source of information, richness, motivation)
3. Witness statements (verification of central or peripheral details, motivation)
4. Coaching admissions (motivation)
5. Serial victim statements (drift of core and peripheral details)
6. Recantations (richness, disruption, pressure)
7. Polygraph examination (type of test, uniformity of results, source of information)
8. Physical evidence
9. Statement analysis

Regarding Serial Victim Statements:

Alleged victims are often interviewed many times, and the consistency of these repeated statements can be evaluated. Some peripheral details should change, disappear, and reappear in the different statements, but core elements of the statements should not vary (Christianson, 1989; Undeutsch, 1989). If the most central elements of the statement vary, it may indicate that the accusation is false and that the child is changing the story to satisfy the demands of different interviewers.

Summary:

When researchers do not specify how ground truth was established or when they depend on a single index to assign cases to abused, non-abused, and the false allegation groups, they risk confusion rather than insight. (p. 49)

Note that Dr. Wendy Dutton includes this study in her dissertation citations and is part of the basis of her testimony. 

The following is a compilation of research regarding false child abuse claims: 

• A study of 576 child sexual abuse allegations made to the Denver Department of Social Services found that 3% were unreliable accounts and 8% were purely fictitious. Clinical features relevant to the fictitious accounts included: lack of emotion and an absence of coercion and threat in the recount, Jones and McGraw (1987).

Jones & McGraw, Reliable and fictitious accounts of sexual abuse to children. Journal of Interpersonal violence, Vol 2(1), Mar 1987, pp. 27-45

• One study of 410 children referred for suspected child sexual abuse, 197 (48%) were confirmed abused, 63 (15%) were determined to be false allegations.

Anthony and Watkeys. False allegations in child sexual abuse: The pattern of referral in an area where reporting is not mandatory. Children & Society, Vol 5(2), Sum 1991, pp. 111-122.

• After repeated forensic interviews 55.5% of children’s sexual abuse claims were determined to be unfounded or false.

Connie Nicholas Carnes, et al. Extended forensic evaluation when sexual abuse is suspected: a multisite field study. Child Maltreatment (2001) 6: 230

• 2.98 million American children underwent a Child Protective Services investigation (or alternative CPS response) for an allegation of child abuse or neglect in 2010.  One-fifth of the claims met the legal requirements of abuse. In the other four-fifths of the claims, the investigation determined that there was not sufficient evidence under state law to conclude or suspect that the child was maltreated or at-risk of being maltreated.

Administration for Children and Families. Child Maltreatment 2010. Washington DC. Department of Health and Human Services. Page 6. http://www.acf.hhs.gov/programs/cb/pubs/cm10/index.htm

Administration for Children and Families. Child Maltreatment 2010. Washington DC. Department of Health and Human Services. Table 3-2. http://www.acf.hhs.gov/programs/cb/pubs/cm10/index.htm

Administration for Children and Families. Child Maltreatment 2010. Washington DC. Department of Health and Human Services. Pages 7-8. http://www.acf.hhs.gov/programs/cb/pubs/cm10/index.htm

• A review of the literature concerning systematic research into the frequency of false allegations of sexual abuse revealed rates ranging from two to eight percent of referrals to child abuse clinics, six percent of emergency room referrals and substantially higher rates of 36.4% to 55.5% for the special circumstance of allegations arising out of the context of custody disputes.

False sexual-abuse allegations by children and adolescents: Contextual factors and clinical subtypes. Mikkelsen, Edwin J., et al., American Journal of Psychotherapy, Vol 46(4), Oct, 1992. Boundaries, behavior, and sexual misconduct: Current issues and the medicolegal interface. pp. 556-570.

• In a large retrospective review involving all 576 complaints of sexual abuse made to the Denver, Colorado, Department of Social Services in 1983, Jones and McGraw ^[8] utilized the innovative strategy of breaking away from a strict Yes/No dichotomy and put forward four categories that capture the ambiguity of many of these situations. Their four categories and the frequency of each are as follows: confirmed–53%; suspicious but not conclusive–17%; insufficient evidence to reach a conclusion–24%; and, definite false allegations–2%. (43% were unfounded) Mikkelsen, et al.

• An important variable that has been analyzed in this research is the age of the child as it relates to the frequency of false allegations. This research as summarized by Everson and Boat indicates that the relative frequencies for different age groups is as follows [N for each group indicated in ()]: 1 to 2.9 years of age (N = 124) 6%; 3 to 5.9 years of age (N = 301) 7%; 6 to 11.9 years of age (N = 414) 4.3%; and 12 to 17.9 years of age (N = 410) 8.0%. Thus, it would appear that the risk of an allegation being false is greater with older children. Mikkelsen, et al.

Dr. Wendy Dutton’s description of her testimony (February 2022):

There are two types of false allegations:  erroneous and malicious.  Erroneous allegations are the more common of the two.  Erroneous allegations are reports that are made in good faith.  Upon investigation, erroneous allegations are either determined to have an alternate cause, or there is insufficient evidence to support that abuse occurred.  Malicious false reports are consciously made reports in an effort to achieve a goal or secondary gain. Malicious false reports tend to be rare.

Studies she relies upon:

Faller, 1998; Jones & McGraw, 1987; Oates et al., 2000; Thoennes & Tjaden, 1990; Trocme & Bala, 2005.

Extended forensic evaluation when sexual abuse is suspected: a multisite field study.

Connie Nicholas Carnes, et al. Child Maltreatment (2001) 6: 230

Abstract

A subset of children referred due to suspected sexual abuse require more than one interview for professionals to reach an opinion about the veracity of allegations. The National Children’s Advocacy Center’s forensic evaluation model was designed for that specific group of children. The multisite study of the model reported here followed a 2-year pilot study. Professionals in 12 states adopted the model and collected data for 2 years on a total of 147 participants. In 44.5% of the cases, a credible disclosure was obtained, with 73% of these cases supported in the legal system. The forensic evaluation procedure yielded clear information to be used in child protection and prosecutory decisions in 64% of the cases (combining cases with credible disclosures and abuse unlikely findings). Finally, the study examined the effects of the length of the evaluation and of the case and child characteristics on evaluation outcomes.

147 children total
64 children had credible disclosures (147 x 45.5% = 64 children)
47 of the credible disclosures supported in the legal system. (64 x 73% = 47 children)
Of all 147 allegations, only 47 (32%) were found to be proven. The other 68% were determined to by unfounded or false. 

• This study was designed to test the efficacy of an extended forensic evaluation model that was designed as a fact-finding procedure for children who, if they were abused, are reluctant to disclose or who are too young or frightened to give clear details in a single interview of whether abuse actually occurred. (Carnes, p. 231)
• The extended forensic evaluation model exam­ined in this study spreads the phases of a forensic interview with the child over time. The first session in the model is devoted to gathering relevant informa­tion from nonoffending caregivers without the child present. The model incorporates social, behavioral, and clinical assessment of the child to determine any treatment needs. It has been designed as a forensi­cally sound fact-finding procedure that assists the investigation and as a clinically sound assessment pro­cedure that serves as a foundation for planning any needed treatment. (Carnes, p. 231)
• This study had several goals. First, it sought to assess the model’s efficacy as a fact-finding procedure and to examine its usefulness in the legal system in multiple jurisdictions (in a 2-year pilot project in Huntsville, Alabama [ Carnes, Wilson, & Nelson-Gardell, 1999], disclosures obtained through the evaluation model were supported in court proceedings in 71 % of the cases). Second, recognizing that some abused chil­dren may require more than one opportunity to dis­close, the project was designed to examine that pro­cess and to explore that possibility. Finally, the study explored the effects of case variables on outcomes, including child factors such as age, race, and gender and case factors such as custody disputes, abuse history of the nonoffending caregiver, and domestic violence. (Carnes, p. 232)
• Participants were children who came to the atten­tion of participating agencies due to suspicions of sex­ual abuse and who did not make clear statements to refute or confirm those suspicions during a single investigative interview. Children who came to the attention of participating agencies were eligible for participation when the child did not disclose abuse to investigators but exhibited behaviors or other indica­tors strongly suggestive of victimization ( e.g., develop­mentally inappropriate sexualized behaviors or diag­nostic medical evidence), when the child did not disclose abuse to investigators but had allegedly dis­closed to some other person, or when the child made a statement in an initial interview that was unclear or suggestive of abuse but lacked the clarity necessary to rule out alternative hypotheses or to make prosecuto­rial and/ or child protective decisions. (Carnes, p. 232)
• Evaluators who participated in the study were all mental health professionals. The researchers pro­vided a 2-day training to these individuals on the forensic evaluation model and on the data collection process for the project. At the training, each step of the model was discussed in detail, and specific tools and techniques were provided to help standardize the procedures to be used. A 225-page manual on the extended forensic evaluation model was provided to evaluators. This manual also described the proce­dures in detail and contained worksheets and guide­lines for each stage of the model. The free provision of the training to evaluators served as an incentive for participation in the project. The training was made available on several different occasions to those evaluators who expressed an interest in participating in the project. Most evaluators were employed by children’s advocacy centers, and some were drawn from other types of agencies. Participating agencies are listed in the acknowledgements for this article. (Carnes, p. 232)
• The training process continued beyond the formal 2-day training session. The authors fielded phone calls throughout the 2-year process to provide technical assistance to the evaluators. Written updates were provided to all participants as issues emerged about various details of the model, so that all could have the benefit of the same technical assistance. Two update meetings were held for participants during the data collection phase. These meetings provided further training on the model and data collection procedures. (Carnes, p. 232)
• The investigators used an experimental comparison group design applying random assignment to treatment conditions. When an eligible participant came to the various agencies for forensic evaluation, random assignment to either a four- or eight-session format of the model was made. The four-session model involved an interview with a nonoffending caregiver and then three sessions with the child. The eight-session model involved an interview with a nonoffending caregiver and seven sessions with the child. (Carnes, p. 232)
• The original sample consisted of 14 7 children from the age of 2 through 17 (M = 6.26). Females made up 68.7% of the sample; 104 (70.7%) were Cau­casian, 26 (17.7%) were African American, and 17 (11.5%) were other or unknown ethnicity. Twenty­two professionals at 18 agencies contributed data. Two agencies were located in the western United States, two in the Midwest, three in the Northeast, and the rest in the Southeast. (Carnes, p. 233)
• Outcome categorization and interrater procedure. The dependent variable or outcome of the evaluation pro­cess was measured at the nominal level and was a re­sult of a decision-making process regarding the veracity of the allegations that precipitated the evalua­tion. A desk guide designed to critically assess the credibility of children’s statements was used to help evaluators categorize the evaluation outcomes. The guide included categories such as details disclosed, developmental appropriateness, affective observa­tions, behavioral checklist results, corroborative in­formation, and assessment of possible alternative explanations and motivational factors (Carnes et al., 1999). At the end of the guide was a list of outcomes from which the evaluator chose after consideration of all the factors embedded in the guide. The guide is not an empirically normed scale, but all of its ele­ments were drawn from the literature on credibility assessment. It does not have known psychometric properties; however, it served to assure that the evalu­ators were using similar empirically based criteria to categorize outcomes. (Carnes, p. 233)
• From the original sample of 14 7 cases, 10 were eliminated prior to statistical analysis through an interrater review process. Two of the authors in­dependently reviewed every case to determine whether the evaluator had provided adequate docu­mentation (via the desk guide) to support the conclu­sion. Approximately 20 cases were identified for fur­ther review. Then all four authors examined those cases more closely and made independent decision,􀀊 regarding the adequacy of the documentation to sup­port the conclusion. Then the raters discussed their opinions and made a decision on each case. Thus, 10 cases were eliminated prior to statistical analysis due to inadequate documentation to support the conclusion. (Carnes, p. 233)
• Abuse likely cases. As discussed above, 10 of the 14 7 original cases were eliminated due to inadequate doc­umentation. In 61 (44.5%) of the 137 valid cases in the sample, evaluators were able to gain credible disclosures from children that could help investigators continue the investigation and could serve as direct evidence for triers of fact. For these cases, the evaluator believed the child’s statements to be credible, based on the factors in the critical analysis guide. (Carnes, pp. 234-235)
• Abuse unlikely cases. There was a wide variety of cases in the abuse unlikely category (n=26, 19%). Of these, 5 children appeared to have been exposed to an overly sexualized environment but did not appear to have been overtly abused. Although no criminal act was identified, these children and families were clearly in need of services. In 4 cases (3% of the sample), it appeared evident an adult had initially coached a child to make a false statement of abuse. A variety of other circumstances existed in this category, including experimentation with age mates, impulsivity and boundary issues, and initial misinter­pretation of innocuous activities. Some of the cases were in need of continued intervention, although there was insufficient evidence to conclude that abuse had occurred. (Carnes, pp. 234-235)
• Outcome unclear. This group of cases ( n = 50, 36%) remained unresolved. In 13 cases, it appeared diffi­cult for evaluators to confidently confirm or rule out suspicions of sexual abuse due to multiple forms of family dysfunction and abuse. Another 10 unclear cases were those in which diagnostic medical evi­dence existed but there was no statement from the child (6 of these 10 cases were reported to authorities by medical personnel). These cases were particularly concerning due to the risk represented by the physi­cal condition. Sexualized behaviors were the primary concern in 7 of these unclear cases, but the children made no verbal disclosure. Developmental limita­tions inhibited resolving 6 of the cases involving chil­dren who ranged from 3 to 4.5 years of age. Of these children, 4 made statements that suggested sexual abuse, however, due to limited language and cogni­tive development, the evaluators were unable to form an opinion about their credibility. (Carnes, pp. 234-235)

LIMITATIONS

• Research in the field of child maltreatment pre­sents many challenges. This study was no exception. Although the evaluation procedures were discussed in training, described in a manual, and clarified through various in-person and written follow-up pro­cedures, there is still a lack of strong independent information confirming that the interviewers fol­lowed the procedures. Although the reported spread of the various evaluation techniques across the evalua­tion time span suggests they followed the procedures, no independent corroboration exists, leaving the question open about whether the recommended extended forensic evaluation process was followed. Future research needs to address this issue in its design. (Carnes, pp. 236)
• As was pointed out earlier in this article, it is impos­sible to know ground truth. This is another major lim­itation of the study. Although legal corroboration exists for a majority of the abuse likely cases, we can­not know for sure how often the interviewers’ judg­ment coincided with the truth of the situation.
• Another limitation unique to this study has to do with the method of data collection. Almost all of the data collected for the study were filtered through the perceptions of the evaluators. This limits the reliabil­ity of many of the variable measurements. (Carnes, pp. 236)
• Clearly, a single-interview model would leave some children at risk that were not pre­pared or able to describe the abuse in the first inter­view. In this study, 44.5% of the children were assessed in the abuse likely classification, and that clas­sification was corroborated in 88% of the cases for which such information was available. None of these children would have been protected nor the person who abused them held accountable with a single­interview model. (Carnes, pp. 236)