Skip to content
Search resources Open menu

Adult-CERT

Fit and Generalizability: Matching the Research to the Case

Published

Snap Shot

  • The question: Do the studies an expert cites describe people and circumstances like those in the case, and when does a mismatch matter?
  • What fit requires: Under Daubert, Joiner, and amended Rule 702(d), each opinion must reflect a reliable application of the research to the facts, and a court may exclude an opinion with “too great an analytical gap” between data and conclusion. Even cold testimony must help the factfinder.
  • How samples differ: Much of the research comes from college women or help-seeking samples; men are substantially underrepresented, although the 2025 military survey estimated comparable numbers of men and women experiencing unwanted sexual contact. Relationship, alcohol, time since the assault, definitions, setting, and date also matter.
  • What it does not show: A mismatch matters only if it plausibly affects the behavior at issue, and thin research on a population does not mean the behavior is unusual there. No study, however well it fits, can say whether a particular complainant was assaulted.
  • The law: Houser requires attention to basis and legal relevance, and Flesher reversed where reliability and scope were not adequately examined; Haskie cautions against categorical claims without empirical support. Disclosure rules (Fed. R. Crim. P. 16; Ariz. R. Crim. P. 15.1, 15.2) and Rule 803(18) allow parties to obtain and use the studies.
  • Issues for each side: Prosecutors should present findings with their populations and select studies that fit; defense counsel should obtain and read the cited studies and raise only mismatches that matter, applying the same questions to defense experts’ research.

Counterintuitive behavior testimony rests on research. An expert who tells a factfinder that freezing, delayed disclosure, or continued contact with the accused is common among people who have been sexually assaulted is summarizing studies, and the value of that summary depends on whether the studies describe people and circumstances like those in the case. Sexual assault is common and underreported, responses to it vary widely, and the behaviors addressed by the testimony genuinely occur in many true cases. But “common” is always common among someone: college women surveyed about past experiences, patients at a hospital clinic, adolescents seeking help at a specialized center, or veterans in treatment. Whether a finding from one of those groups tells a factfinder anything useful about a different person in a different situation is the question of generalizability.

The law asks the same question under the name “fit.” Since Daubert v. Merrell Dow Pharmaceuticals, Inc. (1993), courts have required that expert testimony be not only reliable in the abstract but connected to the facts of the case, and the 2023 amendment to Federal Rule of Evidence 702 reinforced that each opinion must reflect a reliable application of the expert’s methods to those facts. Fit problems are not unique to counterintuitive behavior testimony or to either side. They arise whenever any expert, for the prosecution or the defense, relies on studies whose participants or conditions differ in important ways from the case at hand.

This paper explains the legal doctrine of fit, describes the main ways research samples in this field differ from one another and from typical cases, offers an even-handed set of questions for evaluating whether a cited study fits, and discusses the practical importance of obtaining and reading an expert’s sources before trial.

Fit as a Legal Requirement

Daubert held that the trial judge must ensure that expert testimony is both reliable and relevant, and it described the relevance requirement as one of “fit”: the testimony must be sufficiently tied to the facts of the case that it will help the factfinder resolve a disputed issue. Scientific validity for one purpose, the Court noted, is not necessarily validity for another. In General Electric Co. v. Joiner (1997), the Court applied that principle and held that a trial judge did not abuse his discretion in excluding an opinion that rested on studies too different from the case to support it. The expert relied in part on animal research in which infant mice received massive injections of a chemical, and on human epidemiological studies that did not find the association the expert claimed or involved different exposures. The Court explained that nothing requires a court to admit an opinion connected to the data only by the expert’s own assertion, and that a court may conclude there is “too great an analytical gap between the data and the opinion proffered.” Kumho Tire Co. v. Carmichael (1999) extended these gatekeeping obligations to all expert testimony, including testimony based on clinical or professional experience.

Rule 702 was amended in 2000 to require that testimony rest on sufficient facts or data, be the product of reliable principles and methods, and reflect reliable application of those methods to the facts of the case. The amendment effective December 1, 2023, added that the proponent must establish each requirement by a preponderance of the evidence and revised the final element to require that each opinion “reflects a reliable application” of the principles and methods. The Advisory Committee’s note explains that some courts had wrongly treated questions about the sufficiency of an expert’s basis and the application of the methodology as matters of weight for the jury, and it emphasizes that an expert’s opinion must stay within the bounds of what can be concluded from a reliable application of the methodology. For counterintuitive behavior testimony, that is a fit requirement: a general statement drawn from studies of one population must be one that those studies can actually support when applied to the circumstances of the case.

For educational or “cold” testimony, fit operates somewhat differently. The expert typically does not apply research to the complainant at all, and courts have permitted such testimony on the ground that it helps the factfinder understand behavior at issue in the case. The testimony still must be helpful, and its helpfulness depends on whether the research describes behavior and circumstances resembling those the factfinder must evaluate. Testimony about continued contact between dating partners, for example, may help a factfinder evaluate continued contact between dating partners; whether it helps evaluate a different relationship is a question the research itself may or may not answer.

How Research Samples Differ

Several features of research samples recur in this literature, and each can affect how far a finding generalizes. None of them makes a study wrong; each defines the population to which its findings most directly apply.

Who Was Studied

Much research on responses to sexual assault comes from samples of college women, who are accessible to researchers and among whom assault is common. Other studies recruit people who sought help: patients at emergency clinics for sexual assault, clients of rape crisis services, or veterans in treatment for military sexual trauma. Each recruitment route selects particular people. Help-seeking samples include only those who sought care, often soon after the event, and may overrepresent more severe assaults or more distressed survivors. The Swedish study that found tonic immobility in about 70% of women seen at an emergency clinic for sexual assault reported that 317 of 1,047 eligible women (about 30%) enrolled and 298 were analyzed, that more than a third were lost to follow-up, and that no data were available on women who did not seek care (Möller et al., 2017). The authors themselves identified these limits. The Dutch study of delayed disclosure drew on 323 help-seeking adolescent and young adult women, ages 12 to 25, and excluded those with a history of chronic childhood sexual abuse (Bicanic et al., 2015). These are well-conducted studies, and their samples describe whom their findings most directly concern. Samples drawn from military cases are rare. One study coded victim behavior recorded in the forensic case files of 58 service members charged with sexual assault and referred for forensic evaluation or consultation: 12.0% of victims were male, 37.9% reported within a month, forceful resistance occurred in 15.5% of cases, and physical injury was absent in 96.5% (Umbrasas, 2022). That sample is closer to a court-martial than a campus survey is, but it includes only charged cases referred for evaluation, relies on what the files recorded, and cannot compare accurate with inaccurate reports.

Gender

Men are substantially underrepresented in research on responses to sexual assault, although in the military the estimated number of men who experience unwanted sexual contact each year is comparable to the number of women; the Department of War’s 2025 survey estimated 10,632 men and 9,860 women (U.S. Department of War, 2026, p. 4). Reviews of research on adult male victims have noted wide variation in prevalence estimates, inconsistent definitions, and a far smaller evidence base than exists for women (Peterson et al., 2011). Coxell and King (2010) reviewed what is known about the prevalence of adult male sexual assault, re-victimization, and the tonic immobility response in men. In a mixed-methods study of veterans in treatment for military sexual trauma, O’Brien, Keith, and Shoemaker (2015) described how beliefs about male rape, intensified by aspects of military culture, can delay disclosure and impede recovery among male survivors. This research supports the proposition that male victims also delay, minimize, and may freeze, and it suggests that the reasons can differ. It also shows that a figure derived from female samples should not be presented as if it described men without saying so.

Relationship Between the Parties

Responses to assault vary with the relationship between the people involved. In a sample of 254 college women assaulted by a friend, a casual dating partner, or a steady dating partner, about three-quarters of the relationships continued in some form after the assault, and continuation was more likely when the perpetrator was a steady dating partner than a friend or casual partner (Edwards et al., 2012). In a survey of college students, about 23% of both female and male respondents who reported a sexual assault said they later had intercourse with the assailant on at least one occasion (Ellis et al., 2009). These findings are directly relevant to cases involving similar relationships. Their application to strangers, coworkers, or people who met once is less direct, because the reasons for continued contact, such as shared social networks, emotional attachment, or a desire to restore a prior relationship, depend on what relationship existed.

Alcohol

Alcohol is involved in a large share of sexual assaults, and its role is complex. A review of research from 2000 onward found that alcohol-involved assault is especially common among college women, that intoxication can affect perception, risk recognition, and self-protective responses, and that post-assault drinking can be both a coping response and a risk factor for later victimization (Lorenz & Ullman, 2016). Intoxication also affects memory, which bears on inconsistency and incompleteness in later accounts. Studies that did not measure alcohol, or that combined intoxicated and sober participants, may not describe either group well. When alcohol is central to a case, the relevant question is whether the cited research included, separated, or excluded alcohol-involved assaults.

Time Since the Assault and Method of Measurement

Some studies measure responses within days or weeks of an assault; others ask participants to recall events from years earlier. The Swedish tonic immobility study administered its questionnaire an average of about 19 days after the assault (Möller et al., 2017), while many survey studies ask about lifetime experiences. Recall over long intervals is subject to the ordinary limits of memory, and the meaning participants attach to events can change over time. Studies also differ in how they define assault, ranging from completed rape to any unwanted sexual contact, and in how they measure behaviors such as “delay” or “continued contact.” A finding about “unwanted sexual contact” in a broad survey may not translate directly to a specific charged offense, and a finding about disclosure “more than one week” after an event may not illuminate a disclosure months or years later. Survey categories are also the respondent’s classification, not a description of the event. In a think-aloud study, 54 university students completed a widely used victimization survey while describing their experiences aloud; participants sometimes found items confusing or read them in unintended ways, and comparing their answers with their descriptions identified both false positives and false negatives, with the survey likely underestimating rape and attempted rape, especially among women (Jeffrey & Senn, 2025). The sample was small and drawn from one university.

Setting and Recency

Reporting behavior depends on the setting. Military reporting options, including Restricted Reports, the involvement of victim advocates and special victims’ counsel, and the transfer of prosecutorial authority over covered offenses to the services’ Offices of Special Trial Counsel for offenses after December 27, 2023, create an environment that differs from a civilian college campus or a European clinic. Reporting environments also change over time, so studies from earlier decades may describe conditions that no longer apply. Older studies are not invalid for that reason, but their date is part of what determines how well they fit.

Evaluating Whether a Study Fits

The following questions apply to any study cited by any expert, whether offered to explain a complainant’s behavior or to challenge an account. The first concerns the population: who was studied, how they were recruited, and whether they resemble the people in the case in age, gender, and circumstances. The second concerns the relationship: whether the study addressed assaults by strangers, acquaintances, friends, or intimate partners, and whether it reported results separately by relationship type. The third concerns setting: whether participants were college students, clinic patients, treatment-seekers, military members, or a community sample, and in what country and period. The fourth concerns measurement: how the study defined assault and the behavior at issue, whether it relied on retrospective self-report, and how long after the event the behavior was measured. The fifth concerns alcohol and other case-specific factors: whether the study measured intoxication or other factors central to the case. The sixth concerns the comparison group: whether the study compared victims with anyone, and what that comparison can and cannot show. The seventh concerns recency and replication: when the study was conducted, whether its findings have been replicated, and whether later research qualifies them. Finally, it is worth asking what the authors themselves say about limitations; well-written studies usually state the populations to which their findings may not extend.

No single study will match every feature of a case, and a mismatch on one dimension does not make a study irrelevant. The question is whether the differences are ones that plausibly affect the behavior at issue. A study of continued contact among dating partners may say little about strangers, because relationship type is central to the behavior; the same study may apply reasonably well to dating partners of a different age, if nothing suggests that age changes the pattern. Convergent findings across several studies with different samples provide stronger support than a single study, because they suggest the pattern is not an artifact of one population.

Fit Applies to Every Expert

Fit problems can affect defense experts as readily as prosecution experts. A defense expert who cites laboratory research on memory conducted with sober undergraduates viewing brief film clips to explain inconsistencies in a complainant’s account faces the same question of whether those conditions resemble the case. So does an expert who relies on deception-detection research using laboratory lies, or on studies of police-classified false reports drawn from a single jurisdiction and period, or on research about intoxication that measured alcohol doses and tasks unlike the events at issue. A hypothetical illustrates the symmetry. Suppose a prosecution expert cites a study of college women assaulted by dating partners to explain continued contact between two adult coworkers, while a defense expert cites a study of memory in sober volunteers to explain gaps in the account of a complainant who was heavily intoxicated. Each expert has relied on genuine research, and each faces the same question: whether the research describes the circumstances to which the testimony is being applied.

Obtaining and Reading the Research

The most effective way to evaluate fit is to obtain the studies on which an expert relies and read them, including their methods and limitations sections. Federal Rule of Criminal Procedure 16, as amended effective December 1, 2022, requires that expert disclosures contain a complete statement of all opinions the party will elicit, the bases and reasons for them, and the witness’s qualifications, including publications from the previous ten years and prior testimony from the previous four; the defense’s reciprocal obligations follow the same pattern. Courts-martial apply their own discovery and production rules, and Arizona requires pretrial disclosure of expert witnesses under its criminal disclosure rules. Requesting the specific studies supporting each opinion, early enough to read them, allows each party to identify the populations studied, the definitions used, and any limitations the authors themselves acknowledged.

Reading the studies serves both sides. A proponent who reads the expert’s sources in advance can confirm that each frequency statement is accurate, that it is presented with its population, and that the testimony stays within what the studies show, which strengthens the foundation under Rule 702. An opponent can identify genuine mismatches and distinguish them from differences that do not matter. Under Federal Rule of Evidence 803(18) and its military and Arizona counterparts, statements in a learned treatise or periodical established as reliable may be read into evidence during an expert’s examination, although they may not be received as exhibits.

Limits of the Evidence

Generalizability is a matter of degree and judgment, not a test with a fixed answer. Researchers have not systematically tested how most counterintuitive behaviors vary across populations, relationship types, levels of intoxication, or settings, so it is often unknown whether a given difference between a study and a case matters. The absence of research on a particular population, such as adult male victims assaulted by acquaintances, does not mean that the behaviors do not occur in that population; it means the evidence about frequency is thinner. Conversely, the existence of many studies in one population does not establish that findings extend beyond it. No study, however well it fits, can tell a factfinder whether a particular complainant was assaulted.

Questions to Ask About This Research

When the Möller et al. (2017) finding of tonic immobility in about 70% of women at a Swedish emergency clinic is offered, ask who was studied and whether it applies to this person: 317 of 1,047 eligible women (about 30%) enrolled, more than a third were lost to follow-up, and women who did not seek care are not represented. When the Edwards et al. (2012) finding that about three-quarters of relationships continued after an assault is applied to a case, ask whether the relationship in the case resembles the friends and dating partners studied, since relationship type is central to that behavior. When the thin research on adult male victims is used to suggest that a behavior is unusual among men, ask how much evidence that inference rests on; a smaller evidence base means frequency is less well known, not that the behavior is rare, and O’Brien, Keith, and Shoemaker (2015) describe how beliefs about male rape can delay disclosure. When an expert relies on laboratory memory research with sober undergraduates viewing brief film clips, ask whether those conditions match those of a heavily intoxicated complainant. These questions are explained in “Reading Research Claims: A Guide for Attorneys” on this site.

Fit and the Law

Federal and Military Standards

Daubert, Joiner, and Kumho Tire establish that a trial court must ensure that an expert’s opinion is connected to reliable data and to the facts of the case, and Rule 702(d), as amended in 2023, requires that each opinion reflect a reliable application of the expert’s principles and methods to those facts, established by a preponderance of the evidence. Federal Rule of Criminal Procedure 16(a)(1)(G) and (b)(1)(C) govern expert disclosure in federal criminal cases. In courts-martial, Military Rule of Evidence 702 tracks the federal rule, and under Military Rule of Evidence 1102 amendments to the Federal Rules generally apply eighteen months after their effective date unless the President directs otherwise. The December 1, 2023 amendment to Federal Rule 702 accordingly became part of Military Rule of Evidence 702 in 2025: the proponent of expert testimony in a court-martial must now show the military judge that it is more likely than not that each admissibility requirement is met, so gaps in an expert’s basis or in the application of the method are questions of admissibility rather than of weight for the members (Joint Service Committee on Military Justice, 2025). United States v. Houser (1993) requires the military judge to consider the basis for the testimony and its legal relevance, and United States v. Flesher (2014) reversed where the military judge did not adequately examine the expert’s qualifications, the scope of her testimony, or its reliability. Military Rule of Evidence 803(18) permits use of learned treatises on examination of an expert.

Arizona

Arizona Rule of Evidence 702 adopted the federal standard in 2012 and was amended effective January 1, 2024, to conform to the 2023 federal amendment. In State v. Salazar-Mercado (2014), the Arizona Supreme Court held that cold expert testimony need not apply general principles to the facts, but must still satisfy Rule 702’s requirements, including that it help the trier of fact, and must survive Rule 403. State v. Haskie (2017) cautioned trial courts to confine experts to matters within their expertise and not to permit broad, categorical statements about what is “common” or “usual” without empirical support, a direction that turns in part on whether the expert’s research supports the statement for the circumstances at issue. Arizona Rule of Evidence 803(18) permits the use of learned treatises, and Arizona Rule of Criminal Procedure 15.1 and 15.2 govern pretrial disclosure of expert witnesses by the State and the defense.

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 fit analysis supports counterintuitive behavior testimony that is matched to the case. Research on tonic immobility, delayed disclosure, and continued contact is substantial, and for many common case patterns, such as an assault by an acquaintance or dating partner followed by delayed reporting, there are studies whose populations resemble the people involved. Prosecutors can ensure that the expert identifies those studies, states the populations they describe, and acknowledges where the research is thinner, as it is for adult male victims. Testimony that presents findings with their populations, rather than as universal statements about “victims,” is more accurate and more likely to satisfy amended Rule 702 and Haskie.

The analysis also counsels caution about generalizing beyond the research. Citing a study whose participants differ from the case on a dimension central to the behavior, such as relationship type for continued contact, invites a challenge that can undermine otherwise sound testimony. Reviewing the expert’s sources before disclosure, and selecting those that fit, is the most effective way to avoid the analytical gap that Joiner describes.

Special Issues for Defense

The fit analysis supports a focused inquiry into whether the studies cited by an expert describe people and circumstances like those in the case. Defense counsel can request the specific studies supporting each opinion under the applicable disclosure rules, to read them with attention to sample, relationship, setting, measurement, alcohol, and recency, and to raise genuine mismatches before trial under Rule 702 or on cross-examination using the studies’ own stated limitations.

The analysis also counsels caution about overclaiming. A difference between a study and a case is significant only if it plausibly affects the behavior at issue, and the absence of research on a particular population does not show that the behavior is unusual in that population. Defense experts are subject to the same fit requirement, and research offered on memory, deception, intoxication, or false reports should be examined with the same questions. Group research on stress, alcohol, and discussion shows how memory can be affected; it cannot show that a particular complainant’s memory was. Challenges confined to differences that matter are more accurate.

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

Bicanic, I. A. E., Hehenkamp, L. M., van de Putte, E. M., van Wijk, A. J., & de Jongh, A. (2015). Predictors of delayed disclosure of rape in female adolescents and young adults. European Journal of Psychotraumatology, 6, Article 25883. https://doi.org/10.3402/ejpt.v6.25883

Coxell, A. W., & King, M. B. (2010). Adult male rape and sexual assault: Prevalence, re-victimisation and the tonic immobility response. Sexual and Relationship Therapy, 25(4), 372–379.

Edwards, K. M., Kearns, M. C., Gidycz, C. A., & Calhoun, K. S. (2012). Predictors of victim-perpetrator relationship stability following a sexual assault: A brief report. Violence and Victims, 27(1), 25–32. https://doi.org/10.1891/0886-6708.27.1.25

Ellis, L., Widmayer, A., & Palmer, C. T. (2009). Perpetrators of sexual assault continuing to have sex with their victims following the initial assault: Evidence for evolved reproductive strategies. International Journal of Offender Therapy and Comparative Criminology, 53(4), 454–463. https://doi.org/10.1177/0306624X08316711

Jeffrey, N. K., & Senn, C. Y. (2025). Gender differences in sexual violence victimization experiences and validity of victimization reports: A think-aloud study. Journal of Sex Research, 62(9), 1820–1836. https://doi.org/10.1080/00224499.2024.2397496

Joint Service Committee on Military Justice. (2025). Amendments to the Military Rules of Evidence since 1984 (G. E. Maggs, Ed.; 2025 ed.). https://jsc.defense.gov/Portals/99/MRE%20Amendments%202025.pdf

Lorenz, K., & Ullman, S. E. (2016). Alcohol and sexual assault victimization: Research findings and future directions. Aggression and Violent Behavior, 31, 82–94. https://doi.org/10.1016/j.avb.2016.08.001

Möller, A., Söndergaard, H. P., & Helström, L. (2017). Tonic immobility during sexual assault – a common reaction predicting post-traumatic stress disorder and severe depression. Acta Obstetricia et Gynecologica Scandinavica, 96(8), 932–938. https://doi.org/10.1111/aogs.13174

O’Brien, C., Keith, J., & Shoemaker, L. (2015). Don’t tell: Military culture and male rape. Psychological Services, 12(4), 357–365. https://doi.org/10.1037/ser0000049

Peterson, Z. D., Voller, E. K., Polusny, M. A., & Murdoch, M. (2011). Prevalence and consequences of adult sexual assault of men: Review of empirical findings and state of the literature. Clinical Psychology Review, 31(1), 1–24. https://doi.org/10.1016/j.cpr.2010.08.006

Umbrasas, K. (2022). Victim behavior in a sample of military sexual assault cases. Journal of Interpersonal Violence, 37(3–4), 1367–1383. https://doi.org/10.1177/0886260520922359

U.S. Department of War. (2026). Department of War annual report on sexual assault in the military: Fiscal year 2025. Sexual Assault Prevention and Response Office.

Legal Authorities

Daubert v. Merrell Dow Pharmaceuticals, Inc., 509 U.S. 579 (1993).

General Electric Co. v. Joiner, 522 U.S. 136 (1997).

Kumho Tire Co. v. Carmichael, 526 U.S. 137 (1999).

State v. Haskie, 242 Ariz. 582, 399 P.3d 657 (2017).

State v. Salazar-Mercado, 234 Ariz. 590, 325 P.3d 996 (2014).

United States v. Flesher, 73 M.J. 303 (C.A.A.F. 2014).

United States v. Houser, 36 M.J. 392 (C.M.A. 1993).

Fed. R. Crim. P. 16.

Fed. R. Evid. 403.

Fed. R. Evid. 702.

Fed. R. Evid. 803(18).

Ariz. R. Crim. P. 15.1.

Ariz. R. Crim. P. 15.2.

Ariz. R. Evid. 702.

Ariz. R. Evid. 803(18).

Mil. R. Evid. 702.

Mil. R. Evid. 803(18).

Mil. R. Evid. 1102.

Related topic papers: Behavioral Indicators, Base Rates, and the Limits of Inference · Variability, Confounds, and Individual Differences in Responses to Sexual Assault · Continued Contact with the Accused · Alcohol, Blackouts, and Memory in Sexual Assault Cases · Preparing for and Examining a Counterintuitive Behavior Expert

Related topics from Arizona Forensic Services: Alcohol Blackouts · Memory in the Courtroom · Courts-Martial