Snap Shot
- The question: How accurately do IPV risk instruments such as the Danger Assessment, Lethality Screen, ODARA, and SARA predict further partner violence, and how should results be used in court?
- What the research shows: The instruments do better than chance but perform in the modest range; the largest meta-analysis (50 studies) found an overall AUC of .65, with actuarial tools averaging .66 and structured professional judgment tools .58. In a prospective study of 254 participants recruited at police-involved incidents, the Lethality Screen had sensitivity of 92 to 93 percent for near-lethal and severe violence but specificity of 21 percent, so most participants who were not seriously harmed during follow-up had nonetheless screened positive. Victims’ own perceptions of risk carry real information.
- What it does not show: A risk score does not show whether a charged act occurred. A high-risk category identifies a group with an elevated rate of reoffending; whether that rate exceeds 50 percent depends on the instrument, the outcome, the follow-up period, and the match between the case and the norm sample, and norm-table probabilities may not transfer to another population. Validation for female perpetrators, male victims, and same-sex couples is thin. The instruments are suited to structuring safety planning and risk management rather than to predicting an individual’s behavior.
- The law: Risk testimony is screened under Rule 702 (Daubert, Kumho), and Barefoot leaves dangerousness predictions to cross-examination. At the guilt phase, scores risk the profile objection (Ketchner, Haskie). Arizona release decisions may consider a lethality assessment (A.R.S. § 13-3967), but categorically denying bail for a charge not inherently predictive of dangerousness violates due process (Simpson v. Miller).
- Issues for each side: Prosecutors should not describe a screening result as a probability of lethal violence or apply it outside its validated population; defense counsel should examine administration, version, and fit to the decision, without treating the instruments as having no value.
Intimate partner violence (IPV) risk assessment is the structured process of estimating whether a person who has been violent toward a partner is likely to be violent again, how serious that violence might be, and what might reduce the danger. A small family of published instruments now dominates the field: the Danger Assessment and its short police version, the Lethality Screen; the Ontario Domestic Assault Risk Assessment and the Domestic Violence Risk Appraisal Guide; the Spousal Assault Risk Assessment Guide and its brief police form, the B-SAFER; and the Domestic Violence Screening Instrument. Results from these tools now appear in police reports, pretrial services reports, protective-order hearings, presentence reports, family-court evaluations, and expert testimony.
This paper explains the two main approaches, describes the major instruments, translates accuracy statistics into plain language, summarizes the meta-analytic evidence through 2026, and reviews the legal settings in which risk evidence is offered.
What Risk Assessment Is and Is Not
A risk assessment answers a forward-looking question: how likely is further violence, and under what conditions? It does not answer the trial question of whether a particular act occurred. A high score does not show that an alleged assault happened, and a low score does not show that it did not. The instruments were validated on people already identified as having committed IPV, and most require an established prior assault before they can be scored.
The predicted outcome also varies: new police-recorded assault on a partner (reassault), severe or near-lethal violence, or any new charge. A tool that predicts one outcome may predict another poorly, so the first question about any score is: a prediction of what, over what period, in what population?
Two Approaches: Actuarial Tools and Structured Professional Judgment
The instruments fall into two families. Actuarial instruments assign points to a fixed set of items that were selected statistically because they predicted recidivism in a development sample. The points are summed and the total is linked to the reoffense rate observed among people with similar scores. The evaluator’s discretion is deliberately limited: the same facts should yield the same score regardless of who does the scoring. The ODARA, DVRAG, and DVSI are actuarial.
Structured professional judgment (SPJ) instruments provide a research-based list of risk factors with rating guidance. Rather than adding up a score, the evaluator weighs the factors present, considers how they interact, and reaches a categorical judgment (typically low, moderate, or high priority) with a management plan. The SARA, SARA-V3, and B-SAFER are SPJ instruments. The Danger Assessment sits between the two: it produces a weighted numerical score but is administered with the victim and is often used to guide safety planning.
Actuarial proponents point to the transparency of a fixed score; SPJ proponents respond that fixed scores miss unusual case features and dynamic factors. In IPV, the largest meta-analysis found modestly higher accuracy for actuarial tools (van der Put et al., 2019).
The Principal Instruments
The Danger Assessment and the Lethality Screen
The Danger Assessment (DA), developed by Jacquelyn Campbell, is completed with the victim. A calendar of the past year’s abuse incidents, intended to aid recall, is followed by 20 yes-or-no questions about risk factors such as threats with a weapon, strangulation, forced sex, extreme jealousy, the partner’s unemployment, and recent separation. In its 2009 revision, items were weighted using odds ratios from an 11-city case-control study comparing 310 intimate partner femicide cases with 324 abused women (the 2003 report from this research had analyzed 220 femicide victims and 343 abused women; Campbell et al., 2003), and scores were grouped into four levels: variable, increased, severe, and extreme danger (Campbell et al., 2009). Tested on an independent sample of 194 attempted femicides compared with abused controls, the weighted DA distinguished the groups with an area under the curve reported as .90, a retrospective comparison of cases and controls rather than a prospective prediction.
That design matters. A retrospective case-control comparison shows that DA risk factors were far more common in the most dangerous relationships; it does not give the probability that a particular woman with a particular score will be killed. Used prospectively to predict reassault, the DA’s accuracy is more modest (Messing & Thaller, 2013).
The Lethality Screen is an 11-item version of the DA built for officers at the scene and is the core of the Lethality Assessment Program (LAP), in which a “high-danger” screen prompts an immediate connection to a domestic violence hotline. In a prospective study of 254 participants recruited at the scene of police-involved incidents in one Southwestern state, the screen had high sensitivity (92 to 93 percent) and high negative predictive value for near-lethal and severe violence, but its specificity was only 21 percent (Messing et al., 2017). In other words, it rarely missed participants who were later seriously harmed, but most participants who were not seriously harmed during follow-up had nonetheless screened as high-danger. That tradeoff was deliberate for a tool meant to route victims to services, and a high-danger screen is not a prediction that lethal violence is likely. A quasi-experimental evaluation found that women in the LAP condition took more protective actions and reported somewhat less violence at follow-up than comparison women, although the authors described the evidence as emerging rather than established (Messing et al., 2015).
The ODARA and the DVRAG
The Ontario Domestic Assault Risk Assessment (ODARA) is a 13-item actuarial tool developed from police and criminal records on 589 men who had assaulted a female partner (Hilton et al., 2004). Items include prior domestic and nondomestic incidents, threats and confinement during the index assault, the victim’s concern about future assaults, substance abuse, and assault during pregnancy; it can be scored from police files.
The Domestic Violence Risk Appraisal Guide (DVRAG) adds a score from the Hare Psychopathy Checklist-Revised (PCL-R) to ODARA-type items, yielding 14 items (Hilton et al., 2008). It requires a trained clinician and is intended for correctional and forensic settings.
In an Australian police sample the ODARA predicted physical reassault with an AUC of .68, but only 23 percent of family violence cases met its inclusion criteria (male-to-female assault in a cohabiting relationship) (Lauria et al., 2017). The first published U.S. validation, using 356 New York police cases, found significant but smaller accuracy (AUCs of about .59 to .63) and poor calibration: the recidivism rates attached to each score in the Canadian norms did not match the rates observed in the U.S. sample (Radatz & Hilton, 2022). That distinction between ranking (discrimination) and absolute probability (calibration) is taken up below.
The SARA, the SARA-V3, and the B-SAFER
The Spousal Assault Risk Assessment Guide (SARA), first published in 1995, is the most widely researched SPJ instrument for IPV. The original version has 20 items in four domains (criminal history, psychosocial adjustment, spousal assault history, and the current offense), and evaluators combine item ratings into a summary judgment of low, moderate, or high risk (Kropp & Hart, 2000). A third version, the SARA-V3, was published in 2015. It is organized into 24 factors in three sections (the nature of the IPV, perpetrator risk factors, and victim vulnerability factors), rates each factor for presence in the past year and earlier, and ends with three summary judgments: case prioritization, risk of serious physical harm, and imminence (Allard et al., 2024). A systematic review of the SARA’s psychometric properties found weighted average AUCs of about .63 for IPV recidivism using the original version and cautioned that, at the time of the review, there was almost no predictive research on the SARA-V3 (Allard et al., 2024). The first Australian study of SARA-V3 summary opinions, published in 2026, found excellent agreement among raters on total scores but little ability of the summary opinions to distinguish men who were later charged with IPV from those who were not (AUCs of .51 to .61) (Bryce et al., 2026). Because that is a single file-based study, it is a reason for caution rather than a final verdict.
The Brief Spousal Assault Form for the Evaluation of Risk (B-SAFER) is a shortened SARA-derived SPJ tool for police that includes victim vulnerability factors. In a Swedish police sample, B-SAFER total scores and officers’ overall risk ratings predicted recidivism with AUCs of about .70 and .69 (Storey et al., 2014).
The DVSI and the DVSI-R
The Domestic Violence Screening Instrument (DVSI) was developed by Colorado’s judicial branch for pretrial and probation use and is scored from criminal-justice records. The Colorado version has 12 items, such as prior arrests, protection-order violations, weapon use, and recent separation; a Colorado validation found that it predicted reoffending (Williams & Houghton, 2004). Connecticut’s revised version, the DVSI-R, has been tested in large statewide samples, including 29,317 family violence defendants (Stansfield & Williams, 2014). Published descriptions of the DVSI-R’s item count vary, so the current form in use should be consulted directly.
How Accurate Are the Instruments?
Accuracy is usually reported as the area under the receiver operating characteristic curve, or AUC. In plain terms, the AUC is the probability that a randomly chosen person who did reoffend scored higher on the instrument than a randomly chosen person who did not. An AUC of .50 is a coin flip; 1.00 is perfect sorting. The AUC measures ranking only: it does not say what proportion of high scorers will reoffend.
By that measure, IPV instruments perform in the modest range. Messing and Thaller (2013) found average AUCs of .67 for the ODARA, .63 for the SARA, .62 for the DA, .58 for the DVSI, and .54 for the Kingston Screening Instrument, and noted that fewer than half of the validation analyses administered the instrument as designed. The largest meta-analysis to date pooled 50 studies, 39 instruments, and 68,855 cases and found an overall AUC of .65; actuarial tools averaged .66 and SPJ tools .58, and predicting the first onset of violence was more accurate than predicting its recurrence (van der Put et al., 2019). A systematic review of 43 studies of 18 tools reached a complementary conclusion: most validation research was conducted by researchers coding files rather than by practitioners in real-world settings, few studies addressed feasibility, and reporting of perpetrator and victim gender was often unclear (Graham et al., 2021).
An earlier Canadian meta-analysis found that IPV-specific instruments, general violence instruments, structured judgment, and victims’ own judgments all showed moderate, broadly similar accuracy (Hanson et al., 2007). Across reviews, the instruments do better than chance, but none sorts people with the precision a score might suggest. These instruments discriminate better than chance, with modest accuracy, and are suited to structuring safety planning and risk management rather than to predicting an individual’s behavior.
Base Rates, False Positives, and Calibration
The most common error in using risk evidence is to treat a high-risk classification as meaning that violence is more likely than not. Whether it does depends on the base rate, which is the proportion of people in the population who actually reoffend during the follow-up period. When the base rate is low, even a reasonably accurate tool will flag many people who do not go on to reoffend. A broad meta-analysis of violence risk instruments across settings found a median positive predictive value of .41 and a median negative predictive value of .91 for violent offending (Fazel et al., 2012). Put simply, the tools were much better at identifying people unlikely to offend than at confirming that a flagged person would.
The Lethality Screen illustrates the point. With a specificity of about 21 percent in its validation study, roughly four of every five participants who did not go on to experience severe or near-lethal violence were nonetheless screened as high-danger (Messing et al., 2017). That is not a defect for a screening tool meant to connect victims with services, but it is a serious problem if the screen result is presented in court as evidence that a particular person is likely to kill.
Calibration is separate from ranking. An instrument can rank people correctly yet attach the wrong probabilities to scores in a population with a different base rate, as the U.S. ODARA study found (Radatz & Hilton, 2022). Probabilities from a manual’s norm tables are only as good as the match between the case and the norm sample.
Victim Perception as a Predictor
Several prospective studies show that victims’ own predictions of reassault carry real information. In a sample of 177 women, survivors’ predictions were associated with severe reassault over four months and added to the accuracy of established risk factors (Weisz et al., 2000). In a larger multisite study with a 15-month follow-up, women’s perceptions of risk alone predicted repeated reassault about as well as some structured instruments and improved prediction when combined with risk factors (Heckert & Gondolf, 2004). Among 728 women interviewed after their partners’ arrests, those who perceived high risk despite a low actuarial score tended to report dynamic factors, such as escalating violence, that the tool did not capture (Connor-Smith et al., 2011).
Victim perception is not infallible; some victims underestimate danger and some overestimate it. All three studies sampled women, so whether men’s perceptions of risk from female partners are similarly informative has not been tested. The research supports treating the victim’s judgment as one valuable source of information, best combined with structured assessment, rather than as either decisive or irrelevant.
Validity Across Gender, Culture, and Setting
Nearly all IPV instruments were developed on men who assaulted female partners. Evidence for their use with women who perpetrate IPV, with men who are victimized, and with same-sex couples is thin. The ODARA was tested in a small sample of 30 incarcerated women; it predicted IPV recidivism, but the observed recidivism rates differed from male norms, and the confidence interval around the AUC was very wide (Hilton et al., 2014). In 699 women identified by Australian police as perpetrators, a police actuarial screening tool and a structured professional-judgment guide showed acceptable predictive validity, whereas a lethality screen adapted from male samples performed poorly; the authors called the results preliminary (Nazarewicz et al., 2026). In 867 Swedish police B-SAFER assessments, victim-vulnerability factors were linked to officers’ risk judgments and management recommendations in male-to-female cases, but in female-to-male cases they were more often omitted and showed no consistent association (Storey & Strand, 2016). In Connecticut’s DVSI-R sample (70% male), the timing of reoffending differed significantly by gender (Stansfield & Williams, 2014). The DA and Lethality Screen were built around female victims of male partners and are not validated for other configurations, and the homicide meta-analysis cited below covered only male perpetrators and female victims (Spencer & Stith, 2020). On the other side, only 9 of 44 comparable risk markers for physical IPV differed in strength between men and women in a meta-analysis of 503 studies (Spencer et al., 2022), so many items may transfer across sexes even where norms and cutoffs do not. Graham et al. (2021) found that many validation studies did not clearly report the gender of perpetrators and victims at all.
Cross-cultural evidence is similarly limited. Validation samples have come mainly from Canada, the United States, Sweden, and Australia, with few results reported by race, ethnicity, or immigration status, although the DA has been adapted for immigrant women (Messing et al., 2013). Setting matters too: tools validated by trained researchers working from complete files may perform differently when completed quickly at a scene (Graham et al., 2021). A score for a person or setting unlike the development sample rests on an untested assumption of generalizability.
Limits of the Evidence
Most outcome data come from new arrests or charges, which undercount unreported violence. An intervention prompted by a result, such as detention, may prevent the predicted violence and make a correct prediction look like a false positive. Homicide is too rare to be predicted prospectively with useful precision. Much validation research has been conducted by instrument developers, and independent replications sometimes report lower accuracy.
None of this means the instruments lack value. Structured assessment organizes information and directs attention to empirically supported risk factors such as prior violence, strangulation, threats with weapons, and firearm access, which a meta-analysis identified among the strongest correlates of intimate partner homicide (Spencer & Stith, 2020). The limits concern how precisely results can be stated, not whether the factors matter.
Questions to Ask About This Research
When a high-danger Lethality Screen result is offered at a release hearing, ask what the base rate is and what a positive result means: with sensitivity of 92 to 93 percent but specificity of 21 percent (Messing et al., 2017), most participants who were not seriously harmed during follow-up had screened positive, so the screen identifies a group for services, not a probability that this person will kill. When a single study, such as the SARA-V3 summary-opinion AUCs of .51 to .61 (Bryce et al., 2026), is used to argue that an instrument has no value, ask whether it has been replicated, and compare the modest but better-than-chance meta-analytic AUC of .65 (van der Put et al., 2019). When norm-table probabilities are applied to a U.S. case or a woman who perpetrates IPV, ask whether they apply to this person, given the poor calibration found by Radatz and Hilton (2022) and the sample of 30 women in Hilton et al. (2014). These questions are explained in “Reading Research Claims: A Guide for Attorneys” on this site.
Risk Assessment and the Law
Admissibility of Risk Testimony
When a risk assessment is offered through an expert, admissibility is governed in federal and military courts by Federal Rule of Evidence 702 and Military Rule of Evidence 702, interpreted under Daubert v. Merrell Dow Pharmaceuticals, Inc. (1993) and extended to all expert testimony, including clinical and experience-based opinion, by Kumho Tire Co. v. Carmichael (1999). The 2023 amendment to Rule 702 clarified that the proponent must show by a preponderance of the evidence that the testimony rests on sufficient facts or data, reflects reliable methods, and applies those methods reliably to the case. Arizona Rule of Evidence 702 was amended in 2012 to follow the federal Daubert standard and again effective January 1, 2024 (Order R-23-0004) to conform to the 2023 federal amendment; confirm current text. Relevant questions include whether the instrument was validated for the population and outcome at issue, whether it was administered as designed, what its known error rates are, and whether the expert’s conclusion goes beyond what the instrument can support.
In Barefoot v. Estelle (1983), the Supreme Court declined to bar dangerousness predictions as inherently unreliable, treating cross-examination and opposing experts as the check. Courts also distinguish forecasting future conduct from proving past conduct. In State v. Ketchner (2014), the Arizona Supreme Court found error where general domestic violence expert testimony invited jurors to infer guilt from the defendant’s resemblance to a profile; in State v. Haskie (2017), it upheld “cold” expert testimony about victim behavior while cautioning courts to screen such evidence under Rules 401, 403, and 702. Risk instruments offered at the guilt phase are vulnerable to the profile objection, because their items describe people who reoffend, not evidence that a charged act occurred.
Release and Bail
Arizona’s release statute directs the judicial officer, in setting conditions, to consider among other factors the views of the victim, evidence of danger to others, and the results of a risk or lethality assessment presented in a domestic violence case (A.R.S. § 13-3967(B)); it also authorizes a prohibition on possessing deadly weapons. In Simpson v. Miller (2017), the Arizona Supreme Court held that categorically denying bail to everyone charged with sexual conduct with a minor under age fifteen violated due process because that charge is not inherently predictive of future dangerousness; for such offenses, detention requires a case-specific determination. Structured risk information can inform that individualized inquiry, but the instruments’ modest accuracy and the screen-versus-prediction distinction discussed above bear directly on how much weight a score deserves.
Arizona defines domestic violence by the parties’ relationship (A.R.S. § 13-3601) and treats strangulation in a domestic relationship as aggravated assault (A.R.S. § 13-1204(B)). Strangulation is also a DA and Lethality Screen item, so one fact may drive both the charge and the score.
Sentencing and Supervision
At sentencing, risk results appear in presentence reports and may inform supervision and treatment, consistent with the goal of protecting the public (18 U.S.C. § 3553(a)(2)(C) in federal court). In State v. Loomis (2016), the Wisconsin Supreme Court allowed consideration of a general actuarial risk score but required advisements about its limitations and held that it could not be determinative. In military practice, IPV offenses charged under Article 128b of the Uniform Code of Military Justice (10 U.S.C. § 928b) raise the same questions when risk evidence is offered in sentencing.
Firearms
Access to a firearm is among the strongest risk factors for intimate partner homicide (Spencer & Stith, 2020), and firearms law intersects with risk assessment at several points. Federal law prohibits firearm possession by persons subject to qualifying domestic violence protective orders (18 U.S.C. § 922(g)(8)) and by persons convicted of misdemeanor crimes of domestic violence (18 U.S.C. § 922(g)(9)). In United States v. Rahimi (2024), the Supreme Court rejected a facial Second Amendment challenge to § 922(g)(8), holding that an individual whom a court has found to pose a credible threat to the physical safety of another may be temporarily disarmed; the Court did not decide the provision’s validity as applied to orders resting only on the explicit-prohibition prong. The statute does not require a formal risk instrument, but evidence about threats, prior violence, and weapon use, the same facts that drive risk scores, often bears on the credible-threat finding.
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 is best positioned 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
Prosecutors most often encounter risk results at charging and at release hearings, frequently in the form of a Lethality Screen completed by an officer or a DA completed by an advocate. The research supports treating these results as structured evidence that relevant risk factors are present, and the risk factors themselves, such as strangulation, threats with weapons, and firearm access, are among the best supported in the literature. Errors from this vantage point tend to arise when a screening result is described as a probability of lethal violence, when a score obtained for safety planning is offered at the guilt phase, where it can draw a profile objection under Ketchner, or when an instrument is applied to a case outside its validated population, such as a female perpetrator or a same-sex couple.
It is also useful to know what each instrument was built to do. The Lethality Screen was designed to miss few dangerous cases at the cost of flagging many that do not escalate; the ODARA predicts police-recorded reassault, not homicide; and SPJ ratings are judgments of case priority, not percentages. Presenting results in those terms, alongside the underlying facts and the victim’s account of danger, keeps the evidence within what the research supports.
Special Issues for Defense
Defense counsel most often encounter risk results in release, sentencing, and protective-order settings, where a single score or category can carry considerable weight. The research supports close attention to how the instrument was administered, whether the scorer was trained, which version was used, what information was available, and whether the outcome the tool predicts matches the decision being made. Errors from this vantage point tend to arise when the instruments are treated as having no value at all; the meta-analytic evidence shows that they predict better than chance, with accuracy that one earlier meta-analysis (Hanson et al., 2007) found broadly similar to other structured approaches, and that several of their core items reflect well-established risk factors.
It is also useful to understand base rates and calibration. A high-risk category identifies a group with an elevated rate of reoffense; whether that rate exceeds 50 percent depends on the instrument, the outcome, the follow-up period, and the match between the case and the norm sample, so a category does not by itself show that a particular person is more likely than not to reoffend, and probability tables derived from one population may not transfer to another. Questions about these issues, and about the absence of validation for particular populations such as women who perpetrate IPV, are grounded in the published literature. Framing them in those terms, and acknowledging the risk factors that genuinely are present, keeps the inquiry focused on what the score can and cannot establish.
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
Allard, V., Higgs, T., & Slight, M. (2024). Psychometric properties of the Spousal Assault Risk Assessment from samples of people having perpetrated intimate partner violence. Trauma, Violence, & Abuse, 25(5), 3777–3795. https://doi.org/10.1177/15248380241262275
Bryce, C., Riachi, M., Spivak, B., & McEwan, T. E. (2026). Interrater reliability and predictive validity of the Spousal Assault Risk Assessment – Version 3 (SARA-V3). Assessment. Advance online publication. https://doi.org/10.1177/10731911261469030
Campbell, J. C., Webster, D., Koziol-McLain, J., Block, C., Campbell, D., Curry, M. A., Gary, F., Glass, N., McFarlane, J., Sachs, C., Sharps, P., Ulrich, Y., Wilt, S. A., Manganello, J., Xu, X., Schollenberger, J., Frye, V., & Laughon, K. (2003). Risk factors for femicide in abusive relationships: Results from a multisite case control study. American Journal of Public Health, 93(7), 1089–1097. https://doi.org/10.2105/AJPH.93.7.1089
Campbell, J. C., Webster, D. W., & Glass, N. (2009). The Danger Assessment: Validation of a lethality risk assessment instrument for intimate partner femicide. Journal of Interpersonal Violence, 24(4), 653–674. https://doi.org/10.1177/0886260508317180
Connor-Smith, J. K., Henning, K., Moore, S., & Holdford, R. (2011). Risk assessments by female victims of intimate partner violence: Predictors of risk perceptions and comparison to an actuarial measure. Journal of Interpersonal Violence, 26(12), 2517–2550. https://doi.org/10.1177/0886260510383024
Fazel, S., Singh, J. P., Doll, H., & Grann, M. (2012). Use of risk assessment instruments to predict violence and antisocial behaviour in 73 samples involving 24 827 people: Systematic review and meta-analysis. BMJ, 345, e4692. https://doi.org/10.1136/bmj.e4692
Graham, L. M., Sahay, K. M., Rizo, C. F., Messing, J. T., & Macy, R. J. (2021). The validity and reliability of available intimate partner homicide and reassault risk assessment tools: A systematic review. Trauma, Violence, & Abuse, 22(1), 18–40. https://doi.org/10.1177/1524838018821952
Hanson, R. K., Helmus, L., & Bourgon, G. (2007). The validity of risk assessments for intimate partner violence: A meta-analysis. Public Safety Canada.
Heckert, D. A., & Gondolf, E. W. (2004). Battered women’s perceptions of risk versus risk factors and instruments in predicting repeat reassault. Journal of Interpersonal Violence, 19(7), 778–800. https://doi.org/10.1177/0886260504265619
Hilton, N. Z., Harris, G. T., Rice, M. E., Houghton, R. E., & Eke, A. W. (2008). An indepth actuarial assessment for wife assault recidivism: The Domestic Violence Risk Appraisal Guide. Law and Human Behavior, 32(2), 150–163. https://doi.org/10.1007/s10979-007-9088-6
Hilton, N. Z., Harris, G. T., Rice, M. E., Lang, C., Cormier, C. A., & Lines, K. J. (2004). A brief actuarial assessment for the prediction of wife assault recidivism: The Ontario Domestic Assault Risk Assessment. Psychological Assessment, 16(3), 267–275. https://doi.org/10.1037/1040-3590.16.3.267
Hilton, N. Z., Popham, S., Lang, C., & Harris, G. T. (2014). Preliminary validation of the ODARA for female intimate partner violence offenders. Partner Abuse, 5(2), 189–203. https://doi.org/10.1891/1946-6560.5.2.189
Kropp, P. R., & Hart, S. D. (2000). The Spousal Assault Risk Assessment (SARA) Guide: Reliability and validity in adult male offenders. Law and Human Behavior, 24(1), 101–118. https://doi.org/10.1023/A:1005430904495
Lauria, I., McEwan, T. E., Luebbers, S., Simmons, M., & Ogloff, J. R. P. (2017). Evaluating the Ontario Domestic Assault Risk Assessment in an Australian frontline police setting. Criminal Justice and Behavior, 44(12), 1545–1558. https://doi.org/10.1177/0093854817738280
Messing, J. T., Amanor-Boadu, Y., Cavanaugh, C. E., Glass, N. E., & Campbell, J. C. (2013). Culturally competent intimate partner violence risk assessment: Adapting the Danger Assessment for immigrant women. Social Work Research, 37(3), 263–275. https://doi.org/10.1093/swr/svt019
Messing, J. T., Campbell, J., Sullivan Wilson, J., Brown, S., & Patchell, B. (2017). The Lethality Screen: The predictive validity of an intimate partner violence risk assessment for use by first responders. Journal of Interpersonal Violence, 32(2), 205–226. https://doi.org/10.1177/0886260515585540
Messing, J. T., Campbell, J., Webster, D. W., Brown, S., Patchell, B., & Wilson, J. S. (2015). The Oklahoma Lethality Assessment Study: A quasi-experimental evaluation of the Lethality Assessment Program. Social Service Review, 89(3), 499–530. https://doi.org/10.1086/683194
Messing, J. T., & Thaller, J. (2013). The average predictive validity of intimate partner violence risk assessment instruments. Journal of Interpersonal Violence, 28(7), 1537–1558. https://doi.org/10.1177/0886260512468250
Nazarewicz, J., Trood, M. D., McEwan, T. E., Strand, S., Luebbers, S., & Spivak, B. L. (2026). Assessing risk among women who perpetrate intimate partner abuse. Psychology, Crime & Law, 32(2), 367–393. https://doi.org/10.1080/1068316X.2024.2369239
Radatz, D. L., & Hilton, N. Z. (2022). The Ontario Domestic Assault Risk Assessment: Predicting violence among men with a police record of intimate partner violence in the United States. Criminal Justice and Behavior, 49(3), 371–388. https://doi.org/10.1177/00938548211035816
Spencer, C. M., & Stith, S. M. (2020). Risk factors for male perpetration and female victimization of intimate partner homicide: A meta-analysis. Trauma, Violence, & Abuse, 21(3), 527–540. https://doi.org/10.1177/1524838018781101
Spencer, C. M., Stith, S. M., & Cafferky, B. (2022). What puts individuals at risk for physical intimate partner violence perpetration? A meta-analysis examining risk markers for men and women. Trauma, Violence, & Abuse, 23(1), 36–51. https://doi.org/10.1177/1524838020925776
Stansfield, R., & Williams, K. R. (2014). Predicting family violence recidivism using the DVSI-R: Integrating survival analysis and perpetrator characteristics. Criminal Justice and Behavior, 41(2), 163–180. https://doi.org/10.1177/0093854813500776
Storey, J. E., Kropp, P. R., Hart, S. D., Belfrage, H., & Strand, S. (2014). Assessment and management of risk for intimate partner violence by police officers using the Brief Spousal Assault Form for the Evaluation of Risk. Criminal Justice and Behavior, 41(2), 256–271. https://doi.org/10.1177/0093854813503960
Storey, J. E., & Strand, S. (2016). The influence of victim vulnerability and gender on police officers’ assessment of intimate partner violence risk. Journal of Family Violence, 32(1), 125–134. https://doi.org/10.1007/s10896-016-9905-0
van der Put, C. E., Gubbels, J., & Assink, M. (2019). Predicting domestic violence: A meta-analysis on the predictive validity of risk assessment tools. Aggression and Violent Behavior, 47, 100–116. https://doi.org/10.1016/j.avb.2019.03.008
Weisz, A. N., Tolman, R. M., & Saunders, D. G. (2000). Assessing the risk of severe domestic violence: The importance of survivors’ predictions. Journal of Interpersonal Violence, 15(1), 75–90. https://doi.org/10.1177/088626000015001006
Williams, K. R., & Houghton, A. B. (2004). Assessing the risk of domestic violence reoffending: A validation study. Law and Human Behavior, 28(4), 437–455. https://doi.org/10.1023/B:LAHU.0000039334.59297.f0
Legal Authorities
Barefoot v. Estelle, 463 U.S. 880 (1983).
Daubert v. Merrell Dow Pharmaceuticals, Inc., 509 U.S. 579 (1993).
Kumho Tire Co. v. Carmichael, 526 U.S. 137 (1999).
Simpson v. Miller, 241 Ariz. 341, 387 P.3d 1270 (2017).
State v. Haskie, 242 Ariz. 582 (2017).
State v. Ketchner, 236 Ariz. 262, 339 P.3d 645 (2014).
State v. Loomis, 2016 WI 68, 371 Wis. 2d 235, 881 N.W.2d 749.
United States v. Rahimi, 602 U.S. 680 (2024).
10 U.S.C. § 928b (Article 128b, Uniform Code of Military Justice).
18 U.S.C. § 922(g)(8)–(9).
18 U.S.C. § 3553(a).
Ariz. Rev. Stat. § 13-1204(B).
Ariz. Rev. Stat. § 13-3601.
Ariz. Rev. Stat. § 13-3967.
Ariz. Sup. Ct. Order R-23-0004 (2023) (amending Ariz. R. Evid. 702, effective Jan. 1, 2024).
Ariz. R. Evid. 702.
Fed. R. Evid. 702.
Mil. R. Evid. 702.
Related topic papers: The HCR-20 V3 in Family Violence Cases · Predominant-Aggressor Determinations and Arrest · Protective Orders and Custody · Expert Testimony on Intimate Partner Violence · Intimate Partner Violence Against Men