Snap Shot
- The question: When an adult sexual assault complainant recants or withdraws, does the recantation say anything about whether the assault happened?
- What the research shows: Most recantation research concerns domestic violence and child sexual abuse (23.1% of children recanted in one sample of substantiated dependency cases), and it does not transfer directly to adult acquaintance cases. Police-file research on adults found recantations both in cases researchers judged false and in cases judged not false, where fear or pressure appeared to motivate them.
- What it does not show: There is no validated recantation rate for adult complainants and no comparison of recantation after true and false reports. Recantation is non-specific: either statement may be true, and the research cannot say which. I located no support for the “up to 80 percent” figure in adult sexual assault research.
- The law: From December 1, 2026, a testifying witness’s unsworn prior inconsistent statement may be used substantively in federal court; the military rule keeps the oath requirement until amended. If the complainant does not testify, Crawford, Davis, and Giles govern. Experts may not vouch (Brooks, Mullins, Lindsey).
- Issues for each side: Prosecutors should label domestic violence and child research for what it is, explain its fit, and avoid the 80 percent figure; defense counsel should test fit and focus on the circumstances and corroboration of each statement, not argue that the recantation itself proves innocence.
Some complainants in adult sexual assault cases later say that the assault did not happen, or that it did not happen as first described. Others do not retract the allegation but decline to cooperate further, ask that the case be dropped, or become unavailable. When the prosecution proceeds, it may offer counterintuitive behavior testimony, meaning expert testimony explaining that behaviors jurors may not expect from a person who has been assaulted, here recantation or withdrawal, are common among people who have been. The defense, for its part, may argue that the recantation is the truth.
Recantation genuinely occurs after assaults that happened. Sexual assault is common and underreported, the criminal process is demanding, and complainants can face pressure from the accused, from family or friends, or from their own wish to put the event behind them. Recantation may also occur after reports that were mistaken or untrue, when the retraction is the accurate statement. The research question is therefore not whether recantation can follow a true report, which it can, but what the research actually shows about recantation in adult sexual assault cases and whether it helps a factfinder decide which account is true.
This paper explains where most recantation research comes from and why it does not transfer directly to adult sexual assault, reviews what is known about withdrawal and recantation in adult cases, addresses the frequently repeated “up to 80 percent” figure, and summarizes the law on prior inconsistent statements and confrontation. Recantation in ongoing abusive relationships is addressed more fully in the Intimate Partner Violence resource.
Where the Recantation Research Comes From
Most empirical research on recantation concerns two populations other than adult sexual assault complainants. The first is domestic violence. In a qualitative study of recorded jail telephone calls between 25 men detained for felony domestic violence and their female partners, recantation intentions developed after the men appealed to the women’s sympathy by describing their own suffering and minimized the abuse; once the women decided to recant, the couples jointly built a plan that redefined the event and assigned blame to the state for their separation (Bonomi et al., 2011). The study documents a mechanism by which a genuine victim may come to recant, but its small sample, its focus on detained defendants in established relationships, and its access to the parties’ private conversations all limit generalization.
The second is child sexual abuse. In a random sample of 257 substantiated cases in dependency court, 23.1% of children recanted at some point; recantation was more likely among younger children, children abused by a parent figure, and children whose nonoffending caregiver was unsupportive, and the authors found no support for the alternative explanation that recantations reflected the inclusion of false allegations (Malloy et al., 2007). A later matched study found that children were less likely to recant when family members expressed belief in the allegations and more likely to recant when family members expressed disbelief and when visits with the alleged perpetrator were recommended at the first hearing (Malloy et al., 2016).
Neither body of research transfers directly to adult sexual assault. Domestic violence recantation research concerns people living with or dependent on the accused, often with shared children, finances, and housing, and with a history of repeated violence. Child recantation research concerns children dependent on the adults around them in cases that child welfare authorities had substantiated. Many adult sexual assault cases involve acquaintances with no ongoing relationship, no dependency, and different pressures. An expert who relies on domestic violence or child research to explain an adult complainant’s recantation should say so plainly and explain why the research does or does not fit the circumstances of the case.
What Is Known About Adult Sexual Assault
The most relevant adult research comes from studies of case processing. In Los Angeles, researchers examining sexual assault cases reported to the police in 2008 found that recantation by the complainant was the strongest predictor of a detective’s decision to classify a case as unfounded, and estimated that 4.5% of reports were false (Spohn et al., 2014). The detailed report from that project shows why recantation and falsity are not the same thing. In a stratified random sample of 401 reports, the police had unfounded 81 cases, and 45 of those involved a recantation; within those 81 unfounded cases only, the researchers classified 55 as false reports under their own criteria, 5 as baseless but not false, 10 as not false, and 11 as ambiguous; and 8 of the 10 cases judged not false involved recantations that appeared to be motivated by fear, pressure, or a lack of interest in going forward (Spohn & Tellis, 2012). The research thus found recantations both in cases the researchers judged false and in cases they judged not to be false.
Non-cooperation is more common than outright recantation. Spohn and Tellis (2012) cite earlier studies finding that victims declined to cooperate with police in 36% of sexual assault cases in San Diego and in 42.7% of felony sexual assault reports in Tucson, and they found that the victim’s willingness to cooperate significantly influenced charging decisions. In England and Wales, analysis of a large representative sample of rape allegations reported to the London Metropolitan Police found that attrition was influenced by factors including what police and prosecutors perceived as evidence against the truthfulness of the allegation, such as inconsistencies in the account or a record of a previous false allegation (Hohl & Stanko, 2015). In three years of rape and serious sexual offense records from one English police force, victim withdrawal was more likely when the suspect was a partner or ex-partner and least likely when the suspect was a stranger (Davies et al., 2022); the abstract does not separate adult from child victims. Declining to proceed is not a recantation, and nothing in this research equates non-cooperation, an unfounded classification, or a case that is not prosecuted with a false report.
What the adult research supports is narrow but important: complainants withdraw from cases, and sometimes recant, for reasons that include fear, pressure, and the burdens of the process, and some recantations occur in cases that researchers judged not to be false. What it does not supply is a validated rate of recantation among adult sexual assault complainants, a comparison of recantation rates after true and false reports, or any way to tell from the recantation itself which statement is accurate.
Recantation Is Non-Specific
A recantation is a second statement that contradicts the first. Either statement may be true. If the original report was accurate, the recantation may reflect fear, pressure from the accused or others, a desire to avoid testifying, or relationship and practical considerations. If the original report was mistaken or false, the recantation may be the truth. Because recantation occurs in both circumstances, the fact of recantation does not distinguish between them, and expert testimony cannot supply the missing information. The circumstances of a recantation, such as when it occurred, who was present, what preceded it, whether it was prompted by contact with the accused, and whether either statement is corroborated, are evidence for the factfinder to weigh; they are not matters on which group research can render a verdict.
This non-specificity applies to both parties. Testimony that recantation is “consistent with” genuine victimization is also consistent with a truthful retraction, and argument that a recantation proves the original report was false is equally unsupported. Base rates do not resolve the question either: the commonly cited estimates of confirmed false reports from police files fall in roughly the 2% to 10% range, with a large unresolved remainder that is not evidence of falsity (Lisak et al., 2010; Spohn et al., 2014), and in the university study by Lisak and colleagues 44.9% of reports did not proceed to prosecution or disciplinary action and 13.9% had too little information to classify, against 8 of 136 (5.9%) coded false, but those figures describe reports in general, not the subset of cases in which a complainant has recanted.
A person’s own view of an experience can also change without pressure or a false report. In a study of 380 college students, 7.9% regarded their most recent sexual victimization less strongly as sexual assault at the time of the survey than they recalled doing one month afterward, and 36.1% more strongly; reasons included meaning-making, attributions, and uncertainty (Edwards et al., 2025). The study measured retrospective self-reported labeling, not retraction of factual allegations, and a shift in either direction does not show which view is accurate.
The “Up to 80 Percent” Figure
Some prosecution training materials state that up to 80% of victims recant, citing Meier (2006). That citation does not support the figure. Meier’s essay is a short commentary on the Supreme Court’s decision in Davis v. Washington and its implications for domestic violence prosecution, not a study of recantation rates. Figures in that range circulate in the domestic violence prosecution literature, where they may be estimates that combine recantation with refusal to cooperate, two different behaviors. I located no study that supports applying such a figure to adult sexual assault complainants, and the only systematic recantation rate cited in this paper, 23.1%, comes from children in substantiated dependency cases (Malloy et al., 2007). The figure should not be used in adult sexual assault cases, and testimony that relies on it is vulnerable to challenge under Rule 702.
Limits of the Evidence
The domestic violence research is small and qualitative, the child research concerns substantiated dependency cases, and the adult research consists largely of police-file studies in which recantation and non-cooperation are recorded by investigators rather than measured directly, and in which the truth is often unknown. Rates of non-cooperation vary widely across jurisdictions and definitions. No study compares recantation after true and false adult sexual assault reports. Group research therefore cannot tell a court why a particular complainant recanted or which of her statements is true.
Questions to Ask About This Research
When prosecution testimony cites the 23.1% recantation rate among 257 children in substantiated dependency cases (Malloy et al., 2007) or the jail calls of 25 men detained for domestic violence (Bonomi et al., 2011), counsel can ask who was studied and whether the findings apply to an adult complainant with no ongoing relationship with, or dependency on, the accused. When a witness offers the “up to 80 percent” figure, the question is how much evidence it rests on, since its cited source is a commentary rather than a study (Meier, 2006). When the defense points to the Los Angeles finding that recantation was the strongest predictor of an unfounded classification, alongside an estimated false-report rate of 4.5% among all reports to the department (Spohn et al., 2014), counsel can ask what was counted: in a sample of 401 reports, 81 had been unfounded, 45 of those involved a recantation, and researchers classified 55 of the 81 unfounded cases (not of all 401) as false, and 8 of the 10 judged not false involved recantations that appeared motivated by fear, pressure, or a lack of interest (Spohn & Tellis, 2012). Whether any group figure applies to this complainant’s recantation is a separate question. These questions are explained in “Reading Research Claims: A Guide for Attorneys” on this site.
Recantation and the Law
Federal and Military Standards
When a complainant testifies and recants, the original accusation may be offered as a prior inconsistent statement. Under Federal Rule of Evidence 801(d)(1)(A), such a statement has been usable as substantive evidence only if it was given under penalty of perjury at a proceeding or deposition; an amendment effective December 1, 2026, removes that requirement, so that a testifying witness’s unsworn prior inconsistent statement, such as an initial report to police, may be considered for its truth. Military Rule 801(d)(1)(A) retains the oath requirement until it is amended, and under Military Rule of Evidence 1102 federal amendments generally apply eighteen months after their effective date unless the President directs otherwise. Admitting a prior inconsistent statement of a witness who testifies and is subject to cross-examination does not violate the Confrontation Clause (California v. Green, 1970).
When a complainant does not testify, Crawford v. Washington (2004) bars testimonial hearsay unless the declarant is unavailable and the defendant had a prior opportunity for cross-examination. Statements made to meet an ongoing emergency are nontestimonial, while statements describing past events to investigating officers are testimonial (Davis v. Washington, 2006). The forfeiture-by-wrongdoing exception, reflected in Military Rule of Evidence 804(b)(6), applies only when the accused intended to prevent the witness from testifying (Giles v. California, 2008). A recanting complainant may also face potential exposure for a false statement, either the original report or the recantation, which can raise self-incrimination issues that affect whether and how the witness testifies.
Expert testimony offered to explain recantation must satisfy Rule 702 and, in courts-martial, the factors of United States v. Houser (1993). The Supreme Court has recognized that a court may exclude an opinion when there is “too great an analytical gap” between the data and the opinion (General Electric Co. v. Joiner, 1997), a principle that applies when an expert extends domestic violence or child research to an adult acquaintance case. Experts may not vouch, directly or indirectly, for the original account (United States v. Brooks, 2007; United States v. Mullins, 2010).
Arizona
In State v. Haskie (2017), the Arizona Supreme Court upheld cold expert testimony explaining why a domestic violence victim might recant and remain in an abusive relationship, while cautioning trial courts not to permit broad claims about what is common without empirical support. State v. Lindsey (1986) allows experts to describe general characteristics such as recantation among sexual abuse victims but bars opinions on the truthfulness of a witness or class of witnesses (149 Ariz. at 475). Arizona treats a testifying witness’s prior inconsistent statement as substantive evidence without an oath requirement, but State v. Allred (1982) held that Rule 403 limits its use: it is an abuse of discretion to use a prior inconsistent statement as a pretext for substantive use when the witness denies it, it is uncorroborated, and it is the only evidence of guilt. The court reversed one defendant’s conviction on that ground and affirmed the other’s, where the children admitted making their statements and other evidence showed that the crime occurred. Arizona’s Victims’ Bill of Rights also gives victims the right to refuse a defense interview or deposition (Ariz. Const. art. II, § 2.1(A)(5)), which can limit pretrial inquiry into the circumstances of a recantation.
Special Issues in Legal Practice
The principles described above apply with equal force to every party in a legal proceeding. The science operates the same way regardless of who is asking the questions, and the research summarized here does not by itself favor either side. What differs between the parties is not the science but the vantage point — the particular places where these issues tend to surface, and the particular errors each party may need to guard against. The two summaries below identify those recurring issues. They are offered for orientation, not as litigation strategy; in every case the controlling question is what the evidence in that case will, and will not, support, and the expert’s role is to inform the trier of fact rather than to advocate for either side.
Special Issues for Prosecutors
The research supports testimony that recantation and withdrawal can follow genuine assaults, that fear, pressure, and the burdens of the criminal process are documented reasons, and that police-file research has found recantations motivated by those reasons in cases researchers judged not to be false (Spohn & Tellis, 2012). The domestic violence and child research identifies concrete mechanisms, such as sympathy appeals and family disbelief, that may be relevant when the facts include an ongoing relationship or dependency. Prosecutors can keep that testimony accurate: labeling domestic violence and child research for what it is, explaining its fit or lack of fit, and avoiding the unsupported “up to 80 percent” figure. Haskie and the human-lie-detector cases mark the boundary against implying that the original account is the true one.
The research also counsels caution. Recantation is non-specific, and an expert asked whether recantations also follow false reports can accurately say that they do. When the complainant does not testify, the Confrontation Clause cases govern what can be admitted, and an expert cannot be used to relay the absent complainant’s testimonial statements. Under the amended Federal Rule 801(d)(1)(A), an unsworn initial report may become substantive evidence in federal court, but Rule 403 still permits a court to exclude evidence whose probative value is substantially outweighed by unfair prejudice, and Arizona’s Allred shows the risk of relying on such a statement as the only proof.
Special Issues for Defense
The research supports several legitimate points. Most recantation research concerns domestic violence and children and does not transfer directly to adult acquaintance cases; the adult research does not provide a recantation rate or show that recantations are usually false or usually true; and the “up to 80 percent” figure has no basis in adult sexual assault research. Defense counsel can test whether an expert’s sources fit the case, whether the expert has distinguished recantation from non-cooperation, and whether the testimony implies that the original account is true.
The research also counsels caution about the defense’s own arguments. The same police-file research that shows recantations after false reports also shows recantations, motivated by fear or pressure, in cases researchers judged not to be false, so an argument that the recantation itself proves innocence rests on the inference the defense rejects when the prosecution makes it. The circumstances of the recantation and the corroboration, or lack of it, for each statement present the most defensible ground for argument.
This paper is for educational purposes only and is not legal advice. The author is a psychologist, not an attorney; summaries of law are general and must be verified against current authority or with a licensed attorney.
References
Bonomi, A. E., Gangamma, R., Locke, C. R., Katafiasz, H., & Martin, D. (2011). “Meet me at the hill where we used to park”: Interpersonal processes associated with victim recantation. Social Science & Medicine, 73(7), 1054–1061. https://doi.org/10.1016/j.socscimed.2011.07.005
Davies, K., Spence, R., Cummings, E., Cross, M., & Horvath, M. A. H. (2022). Understanding sexual violence and factors related to police outcomes. Frontiers in Psychology, 13, Article 977318. https://doi.org/10.3389/fpsyg.2022.977318
Edwards, M. E., Howard Valdivia, R. L., Blayney, J. A., & Jaffe, A. E. (2025). Sexual assault labeling over time among gender and sexual identity groups: How and why survivors’ perceptions changed. Sex Roles, 91(4), Article 28. https://doi.org/10.1007/s11199-025-01576-4
Hohl, K., & Stanko, E. A. (2015). Complaints of rape and the criminal justice system: Fresh evidence on the attrition problem in England and Wales. European Journal of Criminology, 12(3), 324–341. https://doi.org/10.1177/1477370815571949
Lisak, D., Gardinier, L., Nicksa, S. C., & Cote, A. M. (2010). False allegations of sexual assault: An analysis of ten years of reported cases. Violence Against Women, 16(12), 1318–1334. https://doi.org/10.1177/1077801210387747
Malloy, L. C., Lyon, T. D., & Quas, J. A. (2007). Filial dependency and recantation of child sexual abuse allegations. Journal of the American Academy of Child & Adolescent Psychiatry, 46(2), 162–170. https://doi.org/10.1097/01.chi.0000246067.77953.f7
Malloy, L. C., Mugno, A. P., Rivard, J. R., Lyon, T. D., & Quas, J. A. (2016). Familial influences on recantation in substantiated child sexual abuse cases. Child Maltreatment, 21(3), 256–261. https://doi.org/10.1177/1077559516650936
Meier, J. S. (2006). Davis/Hammon, domestic violence, and the Supreme Court: The case for cautious optimism. Michigan Law Review First Impressions, 105, 22.
Spohn, C., & Tellis, K. (2012). Policing and prosecuting sexual assault in Los Angeles City and County: A collaborative study (Document No. 237582). National Institute of Justice. https://www.ojp.gov/pdffiles1/nij/grants/237582.pdf
Spohn, C., White, C., & Tellis, K. (2014). Unfounding sexual assault: Examining the decision to unfound and identifying false reports. Law & Society Review, 48(1), 161–192. https://doi.org/10.1111/lasr.12060
Legal Authorities
California v. Green, 399 U.S. 149 (1970).
Crawford v. Washington, 541 U.S. 36 (2004).
Davis v. Washington, 547 U.S. 813 (2006).
General Electric Co. v. Joiner, 522 U.S. 136 (1997).
Giles v. California, 554 U.S. 353 (2008).
State v. Allred, 134 Ariz. 274, 655 P.2d 1326 (1982).
State v. Haskie, 242 Ariz. 582, 399 P.3d 657 (2017).
State v. Lindsey, 149 Ariz. 472, 720 P.2d 73 (1986).
United States v. Brooks, 64 M.J. 325 (C.A.A.F. 2007).
United States v. Houser, 36 M.J. 392 (C.M.A. 1993).
United States v. Mullins, 69 M.J. 113 (C.A.A.F. 2010).
U.S. Const. amend. VI.
Ariz. Const. art. II, § 2.1.
Fed. R. Evid. 403.
Fed. R. Evid. 702.
Fed. R. Evid. 801(d)(1)(A) (as amended effective Dec. 1, 2026).
Ariz. R. Evid. 801(d)(1)(A).
Mil. R. Evid. 702.
Mil. R. Evid. 801(d)(1)(A).
Mil. R. Evid. 804(b)(6).
Mil. R. Evid. 1102.
Related topic papers: False, Unfounded, and Unresolved Reports: What the Research Shows · Behavioral Indicators, Base Rates, and the Limits of Inference · Fit and Generalizability: Matching the Research to the Case · Inconsistent and Incomplete Accounts · Continued Contact with the Accused
Related topics from Arizona Forensic Services: Child Abuse Allegations · Courts-Martial