BAD INVESTIGATIONS CAN LEAD TO EXONERATIONS

What 339 Child Sexual Abuse Exonerations Can Teach Investigators

Child sexual abuse investigations present professionals with an unusually difficult responsibility.

We must take children's allegations seriously. We know that many children delay disclosure. We know that victims sometimes recant truthful allegations. We know that sexual abuse frequently occurs within families and other trusted relationships. We also know that many legitimate cases contain little or no physical evidence.

At the same time, investigators have another responsibility that cannot be separated from the first.

We must determine what actually happened.

A new study published in Frontiers in Psychology provides an important reminder of why those two responsibilities must coexist.

In March 2026, Emily Denne, Colleen Sullivan, and Tess Neal published "Characteristics of Exonerated Cases of Child Sexual Abuse." The researchers examined all 339 known child sexual abuse exonerations contained in the National Registry of Exonerations as of December 2025.

Their findings should get the attention of anyone who investigates, evaluates, prosecutes, or provides expert testimony in child sexual abuse cases.

The central lesson is uncomfortable, but important:

Characteristics commonly associated with legitimate child sexual abuse allegations can also appear in cases that ultimately result in exoneration. That means characteristics are not conclusions.

THE PROBLEM WITH "CONSISTENT WITH"

Professionals who work in child abuse frequently encounter certain characteristics of disclosure.

A child may wait months or years before telling someone what happened. A child may initially disclose abuse and later recant. The alleged offender may be a parent, caregiver, relative, coach, teacher, or another trusted adult. The child may describe abuse occurring repeatedly over time.

None of those circumstances should automatically make us suspicious of the child. Research has established that many genuine victims of child sexual abuse delay disclosure, struggle to describe what happened, maintain relationships with offenders, or later retract truthful allegations.

But Denne and colleagues demonstrate why we must be careful about what we do with that knowledge.

Among the 339 exonerated cases they examined, 49% involved victim recantation. Delayed disclosure appeared in 47.5%. Allegations of repeated abuse appeared in 47.2%. An immediate family member or caregiver was the alleged perpetrator in 47.2%. About 68% involved a single alleged victim.

Those characteristics look remarkably familiar, and that creates an important reasoning problem.

If delayed disclosure is common among genuine victims, we can reasonably say that delayed disclosure is consistent with sexual abuse. We cannot then say that delayed disclosure is evidence that sexual abuse occurred. Those are different propositions.

The same applies to recantation. A genuine victim may recant. Therefore, recantation does not establish that the original allegation was false. But we cannot reverse that reasoning and conclude that a recantation somehow strengthens the original allegation either. We must investigate why the child recanted.

This distinction applies throughout child sexual abuse investigations. A characteristic may be consistent with abuse without proving abuse.

THE AUTHORS FOUND SUBSTANTIAL OVERLAP

Denne and colleagues ultimately reached an important conclusion. Characteristics of exonerated cases looked similar in many respects to characteristics found in substantiated child sexual abuse cases.

That should cause us to reconsider how much weight we place on individual "indicators." Delayed disclosure is not a truth test. Recantation is not a truth test. Behavioral changes are not a truth test. A custody dispute is not a truth test. Repeated allegations are not a truth test. A child's emotional presentation is not a truth test. Even a well-conducted forensic interview is not a truth test.

Each can provide important information. Each can generate additional investigative questions. Some may substantially affect our interpretation of the evidence.

But none gives us permission to stop investigating.

That is why the better question is often not: "Is this consistent with child sexual abuse?" The better question is: "What evidence independently establishes what happened?"

28.3% HAD CORROBORATIVE EVIDENCE!

Perhaps the most fascinating finding in the study involves corroboration. The researchers found what they classified as corroborative evidence in 28.3% of the exonerated cases. Of the 96 cases containing corroborative evidence, 91 contained medical evidence and 16 contained behavioral evidence, with some cases containing both. Medical findings included physical trauma, sexually transmitted infections, and other findings interpreted as evidence of abuse.

At first glance, that sounds contradictory.If there was corroborating evidence, how did these cases result in exoneration? The answer teaches an extremely important investigative lesson: Corroboration must corroborate the specific proposition we are trying to prove.

Evidence that establishes that something happened to a child does not necessarily establish who caused it. A sexually transmitted infection may provide evidence of sexual contact. It does not necessarily identify the offender. Physical trauma may support that an injury occurred. It does not necessarily establish who inflicted it. A text message may establish communication. It does not necessarily establish sexual abuse. Behavioral changes may establish that a child is experiencing distress. They do not necessarily establish what caused that distress.

This distinction becomes especially important because evidence that supports one portion of an allegation can psychologically strengthen our confidence in everything else the child reported. That is dangerous reasoning.

Investigators must continually ask: What exactly does this evidence establish? What does it not establish? And is there another reasonable explanation for it?

CORROBORATION IS NOT SIMPLY FINDING SOMETHING THAT AGREES

This is where the concept of corroboration itself deserves closer examination. Imagine a child reports that a particular person sexually assaulted them. A medical examination identifies an injury consistent with sexual contact. The medical finding may corroborate part of the child's account. But it may not identify the offender. Now imagine investigators discover that the suspect had access to the child during the period in question. That corroborates opportunity. It still does not independently establish that the suspect committed the offense. Then investigators recover messages showing that the suspect communicated privately with the child. That establishes communication. What did those messages actually say?

Good investigations build evidence proposition by proposition. Did an event occur? When could it have occurred? Where could it have occurred? Did the accused person have access? Can the child's description of the location be verified? Can communications be recovered? Did the child describe something that could only reasonably have been known through experience? Were there contemporaneous statements? Do digital records support or contradict the timeline? Are there photographs, videos, location records, witnesses, receipts, employment records, school records, or other independent sources capable of testing portions of the account?

Corroboration becomes stronger when independent sources converge on the same proposition. The goal should never be to collect evidence that merely makes the allegation look more believable. The goal is to test it.

SUGGESTIVE QUESTIONING STILL MATTERS

Suggestive questioning appeared in 21.8% of the exonerated cases. Custody disputes appeared in 19.5%, while satanic ritual or hysteria-related allegations appeared in 15%. But the historical pattern may be even more important.

Denne and colleagues found that several of the more obvious problematic features became less common over time. Suggestive questioning declined considerably. Satanic ritual abuse cases largely disappeared. Recantations also became less common among more recent exonerations.

That is encouraging!

Our understanding of child suggestibility has improved. Forensic interviewing protocols have improved. Professionals increasingly understand the dangers of repeated, coercive, leading, and highly suggestive questioning.

But improvement creates a different challenge. The mistakes may become harder to see. Modern problematic cases may not contain hours of obviously coercive interviews or extraordinary allegations produced through repeated suggestion. The problems leading to an erroneous conclusion may now be subtler.

That means we cannot measure the quality of an investigation simply by asking whether the forensic interview was conducted correctly. A good forensic interview is essential. But a good forensic interview is the beginning of an investigation, not the end of one.

A CLEAN INTERVIEW DOES NOT PROVE THE ALLEGATION

Suppose a child completes a contemporary forensic interview. The interviewer primarily uses open-ended prompts. There is no obvious coercion. There is no evidence of extensive repeated questioning. The child provides a coherent account. The allegations are plausible. There are no bizarre claims.

Those are all positive things! They tell us that the child's account may have been obtained under conditions designed to reduce contamination and maximize accurate recall. But... they do not establish that the allegation is true.

A forensic interview is not a lie detector. Its purpose is to obtain the child's account in a developmentally appropriate and legally defensible manner while reducing unnecessary contamination. Once we have that account, investigators have something extremely valuable. Investigative leads!

THE STRONGEST DETAILS CREATE TESTABLE QUESTIONS

A child's statement becomes particularly valuable when it gives investigators information they can independently test.

"He took me to this house." - Can we establish that the child and suspect were there?

"This happened when my mother was working." - Can we obtain her employment records?

"He messaged me afterward." - Can we recover the messages?

"He locked the bedroom door." - Does the door have a lock?

"He gave me a particular gift." - Can we locate it? Can we determine when it was purchased?

"This happened after practice." - Can we reconstruct the practice schedule and transportation?

"I told my friend afterward." - What does the friend independently remember the child saying?

Each answer begins to build or challenge the allegation. Sometimes the evidence supports the child's account. Sometimes it contradicts part of it. Sometimes it identifies another possible explanation. And sometimes investigators discover evidence even stronger than the original disclosure. That is what investigation is supposed to do.

FALSE ALLEGATIONS SHOULD NOT MAKE US LESS WILLING TO BELIEVE CHILDREN

There is an obvious danger in discussing research involving wrongful convictions and false accusations. Someone will inevitably misuse it to argue that allegations of child sexual abuse cannot be trusted. That is not what this research establishes.

Denne and colleagues examined 339 convictions that ultimately resulted in exoneration. This is a highly selected population. The study cannot tell us what percentage of child sexual abuse allegations are false. Nor does an exoneration necessarily establish that researchers independently proved the underlying allegation was fabricated. The authors explicitly recognize the complexity of drawing that conclusion from exoneration data.

The study answers a different question.

What characteristics appear in child sexual abuse cases where convictions were later overturned or defendants were otherwise exonerated?

That question matters because we can learn from failure. The existence of false or mistaken allegations should not make us less willing to listen to children. It should make us better investigators.

THE ANSWER TO FALSE ALLEGATIONS IS NOT DISBELIEF

There are two serious errors professionals can make in these cases. The first is beginning with the assumption that the child must be telling the truth and investigating primarily to prove the allegation. That creates confirmation bias! Contradictory evidence becomes something to explain away rather than something to investigate.

The second error is beginning with suspicion toward the child and searching for reasons the allegation cannot be trusted. That approach can cause professionals to miss genuine abuse and abandon children who desperately need intervention. Neither approach is acceptable.

The answer to false allegations is not disbelief. The answer is better investigation.

Listen carefully to the child. Interview appropriately. Preserve the child's account. Identify testable details. Seek independent evidence. Follow information that supports the allegation. Follow information that challenges it. Document both. Then go where the evidence leads.

THAT IS WHY NCACIA BUILT CSAC AND SISVO THE WAY WE DID

This research also reinforces why the National Crimes Against Children Investigators Association approaches child sexual abuse training differently.

Our Child Sexual Abuse Course (CSAC), and our Strategic Interviewing and Sexual and Violent Offenders course (SISVO), are built around a basic principle:

Successful child abuse investigation and protection against wrongful accusation are not competing goals.

They depend on many of the same investigative practices!

A careful investigator who understands disclosure dynamics is better equipped to recognize legitimate victims who delay reporting, minimize what happened, recant, maintain contact with an offender, or struggle to provide a perfectly chronological account.

That knowledge can prevent investigators from incorrectly dismissing legitimate victims.

But investigators also need to understand suggestibility, contamination, source-monitoring problems, confirmation bias, competing hypotheses, corroboration, statement development, and the limits of behavioral indicators.

That knowledge can prevent investigators from incorrectly strengthening an allegation that the evidence does not support.

CSAC is designed around that balance. Investigators learn how children disclose abuse, how trauma and development can affect reporting, how forensic interviews should fit into the larger investigation, and how to move from a child's statement into a structured corroboration process.

The goal is not simply to decide whether the child "sounds credible." The goal is to investigate the allegation.

SISVO addresses the case from another critical direction. Investigators must be able to interview accused offenders effectively without treating a denial as proof of guilt or innocence. They need to recognize denial, minimization, mitigation, grooming narratives, rationalizations, opportunity explanations, and attempts to manipulate how behavior is interpreted.

But those techniques must remain evidence-based!

The purpose of a suspect interview should not simply be to obtain a confession. It should be to obtain a statement that can be investigated. What does the suspect admit? What do they deny? Where do they say they were? What access do they acknowledge? What communications do they explain? What relationship do they describe? What alternative explanation do they provide? Which parts of that statement can investigators independently verify?

A truthful denial should survive competent investigation. A deceptive account should create contradictions, inconsistencies, unexplained evidence, or investigative leads that allow professionals to test it.

That is why CSAC and SISVO complement one another.

One strengthens how investigators understand and investigate the child's account. The other strengthens how investigators evaluate and investigate the accused person's account. Both ultimately lead back to the same place. Evidence!

THE STANDARD SHOULD BE THE SAME REGARDLESS OF WHERE THE EVIDENCE LEADS

A successful child sexual abuse investigation should never be defined simply by whether an arrest was made. Sometimes success means building a strong, corroborated case against a legitimate offender and giving a prosecutor the evidence necessary to hold that person accountable. Sometimes success means discovering that the available evidence does not support the original allegation. And occasionally, success means discovering evidence that affirmatively points somewhere investigators did not initially expect.

All three require professional courage. Investigators should want evidence that confirms their theory. They should also want evidence capable of proving their theory wrong. That is how we reduce wrongful accusations without abandoning genuine victims. It is also how we build stronger cases against actual offenders.

The lesson from Denne, Sullivan, and Neal's research is not that investigators should trust children less. It is that we should investigate better.

Delayed disclosure may be meaningful. Recantation may be meaningful. Behavioral changes may be meaningful. Physical findings may be meaningful. A well-conducted forensic interview may be extremely important. But characteristics are not conclusions. Our responsibility is to determine what happened. And the best protection for an abused child and an innocent suspect is ultimately the same thing:

A careful, unbiased, corroboration-based investigation.