Snap Shot
- The question: How well do the instruments used to assess sexual reoffense risk in adolescents work, why are adult actuarial tools not used for most adolescents, and what should an attorney expect a sound adolescent risk assessment to say and not say?
- What the research shows: In a meta-analysis of 33 studies of 6,196 male adolescents, total scores on the J-SOAP-II, ERASOR, J-SORRAT-II, and Static-99 each predicted sexual reoffending with modest accuracy (aggregated AUCs of about .64 to .67), with no tool clearly better than the others (Viljoen et al., 2012). Adolescent tools include dynamic factors that can guide treatment, and research on protective factors and on the strengths-based PROFESOR is emerging.
- What it does not show: These tools do not predict reoffending precisely enough to support long-term decisions on their own, have no validated absolute recidivism estimates, and have little research with girls. The PROFESOR has initial reliability data but no published predictive-validity studies were located. No score shows that a particular adolescent will or will not reoffend.
- The law: Risk opinions must meet Fed. R. Evid. 702, Mil. R. Evid. 702, and Ariz. R. Evid. 702 as amended in 2023 and 2024. They bear on juvenile disposition, transfer (A.R.S. § 8-327), discretionary juvenile registration (A.R.S. § 13-3821(D)), and, in courts-martial, rehabilitative-potential evidence, which must rest on the accused’s personal circumstances (R.C.M. 1001(b)(5)).
- Issues for each side: Prosecutors can point to the moderate benefit of structured tools over unstructured judgment and to the factors associated with reoffending, but should not present scores as probabilities; defense counsel can show the tools’ modest accuracy and short time frame, but should not argue that modest accuracy means a high-risk rating carries no information.
When an adolescent has sexually offended, courts, probation officers, and treatment providers want to know how likely it is that he or she will do so again and what would reduce that likelihood. Structured instruments developed for adolescents are now a standard part of those evaluations. Their results can influence disposition, placement, transfer to adult court, registration, and the conditions of supervision. Attorneys on both sides therefore need to understand what these instruments measure, how well they predict, and where their limits lie.
This paper describes the three instruments most often discussed in adolescent sexual risk assessment, the meta-analytic evidence on their accuracy, why adult actuarial tools are generally not used with adolescents, the time-limited nature of adolescent risk estimates, and the emerging research on dynamic and protective factors. It does not reproduce or paraphrase the items, coding rules, or manuals of any instrument. Recidivism base rates, which every risk estimate must be read against, are discussed in “Sexual Recidivism Among Adolescents Who Sexually Offend.”
The Instruments
The Estimate of Risk of Adolescent Sexual Offense Recidivism (ERASOR) is a structured professional judgment guide for adolescents aged 12 to 18 who have previously committed a sexual assault (Worling & Curwen, 2001). It organizes 25 risk factors, both historical and potentially changeable, into five categories, and the evaluator reaches an overall judgment of low, moderate, or high risk rather than summing a score into a probability. The manual directs evaluators to tell the audience about the scientific limits of the estimate, to justify it by reference to the factors present, to make it as specific as possible, to describe circumstances that could raise risk in the short term, and to recommend strategies for managing it.
The Juvenile Sex Offender Assessment Protocol-II (J-SOAP-II) is a 28-item checklist designed for boys aged 12 to 18 who have been adjudicated for sexual offenses, as well as boys with a history of sexually coercive behavior who have not been adjudicated (Prentky & Righthand, 2003). It has two historical scales, one reflecting sexual drive and preoccupation and one reflecting impulsive and antisocial behavior, and two dynamic scales, one addressing intervention-related factors and one addressing community stability and adjustment. Its authors describe it as an experimental scale that is not actuarial, that provides no cutoff scores or recidivism estimates, and that should not be used in isolation from a comprehensive assessment.
The Protective + Risk Observations For Eliminating Sexual Offense Recidivism (PROFESOR) is a checklist of protective factors designed for people aged 12 to 25 who have committed sexual offenses (Worling, 2017). It contains 20 factors, each rated on a continuum from protective to risk, all of them potentially changeable and rated on current functioning, and it is intended primarily to guide intervention and to measure change over time rather than to predict recidivism. Its published rationale and an initial study of reliability appeared recently (Worling & Minske, 2025); no published study of its predictive validity was located.
Other instruments are used as well, including the Juvenile Sexual Offense Recidivism Risk Assessment Tool-II (J-SORRAT-II) and general juvenile risk tools such as the Structured Assessment of Violence Risk in Youth (SAVRY), which address general and violent rather than sexual reoffending.
How Well the Tools Predict
The central evidence is a meta-analysis that pooled 33 studies, published and unpublished, of 6,196 adolescent boys who had sexually offended (Viljoen et al., 2012). Total scores on the J-SOAP-II, ERASOR, J-SORRAT-II, and Static-99 each significantly predicted sexual reoffending. Aggregated correlations ranged from .12 to .20 and aggregated areas under the curve (AUCs) from about .64 to .67, which the authors described as moderate; an AUC of .50 represents chance, and an AUC of .67 means that a randomly chosen adolescent who reoffended would score higher than a randomly chosen adolescent who did not about two-thirds of the time. The tools did not differ significantly from one another, the ERASOR’s structured clinical judgment did not differ significantly from its total score, and heterogeneity across studies was often moderate to high. The J-SOAP-II scale reflecting sexual drive and preoccupation predicted sexual but not general reoffending, while the scale reflecting impulsive and antisocial behavior predicted both, suggesting that sexual reoffending reflects both sexual and general antisocial factors.
The authors concluded that the tools offer clear benefits over unstructured clinical judgment but that their accuracy may be insufficient for decisions requiring a high degree of precision, naming civil commitment and lifetime registration as examples (Viljoen et al., 2012). Nearly all of the studies coded the tools from files after the fact and measured reoffending through official records, and some adapted the tools. A systematic review of 19 studies of six instruments reached a similar conclusion: none showed unequivocally positive results, specialized tools predicted sexual recidivism better than general youth tools, and, because adolescents change rapidly, long-term restrictions should not be imposed on the basis of a risk assessment alone (Hempel et al., 2013).
Individual studies show both promise and inconsistency. In a prospective study of 191 adolescent males assessed by clinicians at a specialized program, 9.4% were charged with a new sexual offense over a mean follow-up of 3.66 years (Worling et al., 2012). Over that full period, the ERASOR total score (AUC = .72) and the number of risk factors present (AUC = .73) predicted sexual reoffending, but the clinical judgment rating did not (AUC = .61). Over a shorter interval of up to 2.5 years, all three predicted strongly, but that analysis involved only 70 adolescents and six recidivists, so its estimates are imprecise. A review of nine studies of the J-SOAP-II, together with new data on 73 boys, found that the impulsive and antisocial scale performed well across nearly all psychometric properties, including prediction, while the sexual drive and preoccupation scale did not predict concerning sexual behavior, consistent with seven earlier studies that found no prediction of sexual recidivism (Fanniff & Letourneau, 2012). These mixed results are one reason the tools’ authors caution against reliance on any single scale.
A practical limit follows from low base rates. When few adolescents reoffend, even a tool with moderate accuracy will place many adolescents who will not reoffend in higher-risk groups. In a seven-year study of adolescent males charged with serious offenses, 7.87% of 127 adolescents who had committed sexual offenses were later arrested for a sexual offense, compared with 2.84% of 1,021 who had committed other offenses; general recidivism was similar in the two groups, and most risk factors for general delinquency operated similarly (Fanniff et al., 2017). Because the second group was much larger, most of the later sexual arrests in that sample were of adolescents with no prior sexual offense, a reminder that risk tools applied only to adolescents already identified will not account for most new sexual offending.
Why Adult Actuarial Tools Are Generally Not Used
The Static-99 performed about as well as the adolescent tools in the meta-analysis, but its accuracy with adolescents was lower than in adult samples, and the authors noted that adolescents automatically receive points on items reflecting young age and relationship history, so the tool may overestimate their risk (Viljoen et al., 2012). The Static-99R coding rules (revised 2016; coding rules are updated from time to time, so the current version should be confirmed) state that the instrument is not recommended for those younger than 18 at release; may be used, with caution, only when the person was 17 at the offense, was released at 18 or older, and committed an offense resembling typical adult offenses; and that its recidivism estimates should never be applied to someone whose last known sexual offense occurred before 17 (Phenix et al., 2017). The coding rules explain that adolescents who sexually offend are not simply younger versions of adults and that no study has tested whether the instrument’s estimates are calibrated for them. The rules also state that the instrument is not recommended for females.
The same reasoning matters in courts-martial and in federal and state adult courts, where a young adult may be evaluated for conduct that occurred at 16 or 17, or for offenses committed shortly after turning 18. Whether an adult instrument or an adolescent instrument fits depends on the person’s age at the offense and at release and on the nature of the offense, and an evaluator should explain the choice.
Time-Limited Estimates and Dynamic Factors
Adolescent risk estimates are short-lived by design. The ERASOR manual states that most of the research supporting its factors used follow-up periods of three years or less, that adolescence brings rapid social, physical, familial, and sexual change, and that estimates should be repeated after a fixed interval, such as two years, or after significant change in a risk factor (Worling & Curwen, 2001). The prospective ERASOR study likewise recommended periodic reassessment (Worling et al., 2012). Because the ERASOR manual and most supporting studies rely on follow-up periods of three years or less, and no study located tested how quickly accuracy declines, a rating made at disposition should be repeated at a fixed interval or after significant change and used cautiously for later decisions, such as a registration review, a transfer hearing in another case, or an adult sentencing.
Because the ERASOR and J-SOAP-II include potentially changeable factors, they can identify treatment and supervision targets, which is often their most useful function (Viljoen et al., 2012). Research on factors that protect against reoffending is newer. In a prospective study of 81 adolescent males, the ERASOR predicted sexual recidivism over an average 3.5-year follow-up, and a parent-rated measure of emotional strengths was associated with lower sexual recidivism, although it did not add to the prediction provided by the ERASOR; school functioning was associated with lower nonsexual recidivism (Worling & Langton, 2015). Strengths-based instruments such as the PROFESOR rest on this line of work, but whether rating protective factors improves prediction for adolescents has not been established.
Risk tools are also used in juvenile justice generally, beyond sexual offending. A systematic review of 22 studies, 13 of them involving adolescents in the juvenile system, found that adopting risk assessment tools was associated with small reductions in restrictive placements and in recidivism, but those findings were no longer significant once studies at high risk of bias were removed, and their effects on racial and ethnic disparities were unclear (Viljoen et al., 2019).
Limits of the Evidence
The research base is small compared with the adult literature, and most studies are retrospective, coded from files, and measured against official records that undercount reoffending. Samples are almost entirely male: the Viljoen et al. (2012) meta-analysis included only males, the J-SOAP-II was designed for boys, and the prospective ERASOR study excluded the five girls assessed during its follow-up because meaningful comparisons were not possible (Worling et al., 2012). There is therefore little basis for applying these tools to girls. Some studies were conducted by the tools’ developers or at the sites where they were developed; developer involvement did not significantly moderate results in the meta-analysis, but results vary widely across studies. No adolescent instrument provides validated absolute recidivism estimates, and the PROFESOR’s predictive validity is untested. Finally, a tool’s accuracy is a property of a group of assessments: a score cannot by itself establish what a particular adolescent will do.
Questions to Ask About This Research
When an aggregated AUC of about .67 (Viljoen et al., 2012) is offered as showing that a tool works, ask what the number measures: how well scores rank reoffenders above nonreoffenders across a group, not the likelihood that this adolescent will reoffend. When the strong short-interval results from Worling et al. (2012) are cited, ask how many people were studied: six recidivists among 70 adolescents. When a Static-99R score is offered for someone who offended at 15 or 16, ask whether the tool applies, given its coding rules (Phenix et al., 2017). When a risk rating is offered years after it was made, ask whether the conclusion still follows, given the instruments’ own time limits. When a PROFESOR summary is offered as a risk prediction, ask whether that use has been validated. These questions are explained in “Reading Research Claims: A Guide for Attorneys” on this site.
Adolescent Risk Assessment and the Law
Federal and Military Standards
Expert risk opinions must satisfy Federal Rule of Evidence 702, amended December 1, 2023, to require the proponent to show by a preponderance of the evidence that the opinion rests on sufficient facts or data, is the product of reliable principles and methods, and reflects a reliable application of those methods to the case (Daubert v. Merrell Dow Pharmaceuticals, Inc., 1993); under Kumho Tire Co. v. Carmichael (1999), that gatekeeping extends to all expert testimony, including structured professional judgment. For adolescent risk instruments, the reliability questions are concrete: whether the tool was designed and validated for someone of the person’s age and sex, whether it was scored by its manual, how old the assessment is, and whether the evaluator presented a structured judgment as if it were a statistical probability. In federal juvenile transfer proceedings, the court considers, among other factors, the juvenile’s present intellectual development and psychological maturity, past treatment efforts, and available programs (18 U.S.C. § 5032), and a current, properly qualified risk assessment can inform each.
In courts-martial, where accused service members may be 17 or 18, an evaluator may be asked to assess conduct from late adolescence. The prosecution’s opinion evidence on rehabilitative potential must relate to the accused’s personal circumstances, and the severity or nature of the offense may not be its principal basis (R.C.M. 1001(b)(5)); the defense may offer risk and treatment evidence in mitigation under R.C.M. 1001(d). A risk assessment that applies a tool outside its intended population, or relies chiefly on the offense, is open to challenge on those grounds. The 2023 federal amendment to Rule 702 took effect in Military Rule of Evidence 702 in 2025 (Joint Service Committee on Military Justice, 2025); practitioners should confirm current text. The R.C.M. provisions described here are from the 2024 Manual for Courts-Martial.
Arizona
Arizona Rule of Evidence 702 was amended, effective January 1, 2024, to conform to the 2023 federal amendment (Order R-23-0004). Risk assessments commonly inform juvenile disposition and supervision, the public-safety and rehabilitation findings required for transfer (A.R.S. §§ 8-327, 13-504), and the juvenile court’s discretionary decision whether to require registration of a juvenile adjudicated for a listed offense, which ends at 25 and may be terminated on successful completion of probation (A.R.S. § 13-3821(D), (F), (G)). A 2019 report to the Arizona Supreme Court’s Juvenile Justice Services Division recommended that dispositions be driven by assessment rather than by offense alone; that recommendation is described here from an unpublished, non-peer-reviewed review of the report (Caldwell, 2020). On the reliability of the specialized tools, the peer-reviewed systematic review by Hempel et al. (2013) found that none showed unequivocally positive results.
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
Viljoen et al. (2012) concluded that structured tools offer benefits over unstructured clinical judgment, though their accuracy is moderate and may be insufficient for decisions requiring high precision, and the evidence base remains small; the research also supports attention to the factors associated with sexual reoffending, including sexual interest in younger children, persistent sexual offending, and impulsive and antisocial behavior. High-risk ratings are associated with higher detected reoffending at the group level, and the dynamic factors identified can justify specific supervision and treatment conditions. Prosecutors are often best positioned to supply complete offense, victim, and supervision records, without which these instruments cannot be scored properly.
The research also counsels caution. The tools’ accuracy is modest, they provide no validated probabilities, and their estimates are time-limited, so a rating made at disposition should not be presented as a measure of an adolescent’s risk years later. Adult actuarial instruments, including the Static-99R, should not be applied outside the limits their own coding rules set, and testimony that converts a high rating into a statement that the adolescent will reoffend goes beyond the evidence and is open to challenge under Rule 702.
Special Issues for Defense
The research supports scrutiny of how an adolescent risk assessment was done: whether the instrument fits the client’s age and sex, whether it was scored by its manual from adequate information, whether its time limits have run, and whether the evaluator acknowledged the tools’ modest accuracy and lack of absolute estimates. Defense counsel are often best positioned to supply evidence of change, treatment progress, school functioning, family support, and other protective factors, and to request reassessment when circumstances have changed.
The research also counsels caution. Modest accuracy is not the same as no accuracy: higher ratings are associated with higher detected reoffending, and factors such as sexual interest in younger children and persistent offending are supported predictors. Low base rates do not show that a particular adolescent is low risk. Protective-factor ratings, including the PROFESOR, have not been shown to improve prediction, and presenting them as evidence that risk is negligible goes beyond the research.
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
Caldwell, M. F. (2020, March 4). Review of the final report: Enhancing court and probation practices for cases involving juveniles who commit sexually abusive behaviors [Unpublished report]. Department of Psychology, University of Wisconsin–Madison.
Fanniff, A. M., & Letourneau, E. J. (2012). Another piece of the puzzle: Psychometric properties of the J-SOAP-II. Sexual Abuse, 24(4), 378–408. https://doi.org/10.1177/1079063211431842
Fanniff, A. M., Schubert, C. A., Mulvey, E. P., Iselin, A.-M. R., & Piquero, A. R. (2017). Risk and outcomes: Are adolescents charged with sex offenses different from other adolescent offenders? Journal of Youth and Adolescence, 46(7), 1394–1423. https://doi.org/10.1007/s10964-016-0536-9
Hempel, I., Buck, N., Cima, M., & van Marle, H. (2013). Review of risk assessment instruments for juvenile sex offenders: What is next? International Journal of Offender Therapy and Comparative Criminology, 57(2), 208–228. https://doi.org/10.1177/0306624X11428315
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
Phenix, A., Fernandez, Y., Harris, A. J. R., Helmus, M., Hanson, R. K., & Thornton, D. (2017). Static-99R coding rules: Revised—2016 (Research Report 2017-R012). Public Safety Canada.
Prentky, R., & Righthand, S. (2003). Juvenile Sex Offender Assessment Protocol-II (J-SOAP-II) manual (NCJ 202316). Office of Juvenile Justice and Delinquency Prevention.
Viljoen, J. L., Jonnson, M. R., Cochrane, D. M., Vargen, L. M., & Vincent, G. M. (2019). Impact of risk assessment instruments on rates of pretrial detention, postconviction placements, and release: A systematic review and meta-analysis. Law and Human Behavior, 43(5), 397–420. https://doi.org/10.1037/lhb0000344
Viljoen, J. L., Mordell, S., & Beneteau, J. L. (2012). Prediction of adolescent sexual reoffending: A meta-analysis of the J-SOAP-II, ERASOR, J-SORRAT-II, and Static-99. Law and Human Behavior, 36(5), 423–438. https://doi.org/10.1037/h0093938
Worling, J. R. (2017). Protective + Risk Observations For Eliminating Sexual Offense Recidivism (PROFESOR) [Structured checklist].
Worling, J. R., Bookalam, D., & Litteljohn, A. (2012). Prospective validity of the Estimate of Risk of Adolescent Sexual Offense Recidivism (ERASOR). Sexual Abuse, 24(3), 203–223. https://doi.org/10.1177/1079063211407080
Worling, J. R., & Curwen, T. (2001). The “ERASOR”: Estimate of Risk of Adolescent Sexual Offense Recidivism (Version 2.0). Sexual Abuse: Family Education & Treatment Program, Thistletown Regional Centre.
Worling, J. R., & Langton, C. M. (2015). A prospective investigation of factors that predict desistance from recidivism for adolescents who have sexually offended. Sexual Abuse, 27(1), 127–142. https://doi.org/10.1177/1079063214549260
Worling, J. R., & Minske, P. (2025). Protective + Risk Observations For Eliminating Sexual Offense Recidivism (PROFESOR): Rationale for development and an initial investigation of reliability. Journal of Sexual Aggression, 31(1), 134–147. https://doi.org/10.1080/13552600.2024.2332434
Legal Authorities
Daubert v. Merrell Dow Pharmaceuticals, Inc., 509 U.S. 579 (1993).
Kumho Tire Co. v. Carmichael, 526 U.S. 137 (1999).
18 U.S.C. § 5032.
Ariz. Rev. Stat. Ann. §§ 8-327, 13-504, 13-3821(D), (F), (G).
Fed. R. Evid. 702.
Mil. R. Evid. 702.
Ariz. R. Evid. 702.
Rule for Courts-Martial 1001(b)(5), (d).
Related topic papers: Sexual Recidivism Among Adolescents Who Sexually Offend · Risk Assessment Tools Outside Their Normative Samples · Normative and Problematic Sexual Behavior in Children and Adolescents · Juvenile Transfer and Competency in Sex Offense Cases · Young Adults Aged 18 to 25 · Treatment for Adolescents Who Sexually Offend
Related topics from Arizona Forensic Services: Psychosexual Risk Evaluations · Courts-Martial