Managing and Treating individuals who have been found not guilty by reason of insanity

Violence risk assessments did not predict long-term outcomes for insanity acquittees in the community but should be utilized for the development of treatment and management strategies. This is the bottom line of a recently published article in Law and Human Behavior. Below is a summary of the research and findings as well as a translation of this research into practice.

Featured Article | Law and Human Behavior | 2018, Vol. 42, No. 5, 403-412

Reconsidering Risk Assessment With Insanity Acquittees


Michael J. Vitacco, Augusta University
Elena Balduzzi, Psychiatric Secure Review Board, Portland, Oregon
Kimberly Rideout, Oregon State Hospital, Salem, Oregon
Shelly Banfe and Juliet Britton, Psychiatric Secure Review Board, Portland, Oregon


States continue to rely on conditional release (CR) as an effective and cost effective way to manage individuals found not guilty by reason of insanity (NGRI). Research has demonstrated that insanity acquittees returning to the community have low recidivism rates and moderately low revocation rates. This study followed 238 individuals found NGRI in Oregon who were evaluated with the Historical, Clinical, Risk-20 (HCR-20; Webster, Douglas, Eaves, & Hart, 1997) and placed in the community on CR. The majority of individuals on CR (n = 157, 66%) maintained their release throughout the entire follow-up period (between 4 and 9 years), but 81 (33.6%) had their release revoked during the follow up. In considering the efficacy of violence risk assessment in predicting CR outcome with NGRI acquittees, the HCR-20 was mostly unrelated to CR outcome. Only two items from the HCR-20, both from the Risk Management scale (exposure to destabilizers and stress) predicted revocation, but not imminence to CR revocation. This paper reconsiders how risk assessments are utilized with insanity acquittees and provides a roadmap for improving risk assessments with this unique population by relying on risk assessment results to plan effective interventions to reduce the likelihood of revocation and violence.


violence risk assessment, insanity acquittees, conditional release

Summary of the Research

“Individuals adjudicated not guilty by reason of insanity (NGRI) are most often committed to inpatient forensic hospitals for a period of evaluation and treatment, with a focus on treating mental illness and reducing future violent and antisocial behavior. In most instances, the goal of inpatient hospitalization is to prepare individuals for a return to their communities. However, individuals found NGRI are not granted unfettered access to the community; instead, they are typically required to participate with several mandated conditions (e.g., medication compliance, abstinence from drugs and alcohol, participate in treatment and/or other structured programming), which is why their return to the community is referred to as conditional release (CR). Failure to meet required conditions, or the commission of a new criminal act, can lead to revocation of CR and a return to a secure forensic hospital or jail” (p. 403).

“The current study evaluates a relatively large sample of individuals on CR from the state of Oregon. Consistent with recent research from a statewide sample in Georgia demonstrating the necessity of proper placement and community treatment for individuals on CR we propose the RM [Risk Management] scale from the HCR-20 will predict CR revocation. Revocation status can also include the commission of a new criminal offense. Further, it is our prediction that the other two HCR-20 scales (i.e., Historical and Clinical) will not be related to CR outcome. Additional research is needed on this subject given inconsistent results across state samples; we are specifically cognizant of the fact that research has been inconsistent on the effectiveness of these scales in predicting revocation with this specialized population. We also hypothesize that the total number of prior criminal charges will predict imminence to revocation. In addition, given that substance use can form the basis for revocation of CR, we predict a substance abuse diagnosis will be related to both revocation and imminence to failure” (p. 406).

“Two metrics are frequently measured when evaluating conditional program outcomes: rates of revocation and commission of new criminal behavior. The results of the current study found a relatively low revocation rate, as 66% of individuals were able to maintain their CR over the entirety of the follow-up period. This statistic is consistent with CR data from other statewide CR programs. Regarding criminal recidivism, only one person was arrested for a new crime, and this was a nonviolent offense. The results of the current study, along with other studies of NGRI patients on CR, stand in stark contrast to the most recent statistics on criminal recidivism generated by the National Institute of Justice (NIJ). NIJ statistics for the entire United States indicate 67.8% of individuals released from prison were rearrested after 3 years and over 75% were rearrested after 5 years. At a state level, data from the Oregon Recidivism Analysis found 48% of those on felony probation were rearrested within 3 years. In sum, although the revocation rate approximates the rearrest rate in Oregon, it remains lower than the recidivism rate” (p. 408-409).

Translating Research into Practice

“In considering why CR may be effective one should consider that studies have linked untreated mental illness with violent behavior, potentially as a result of its inverse relationship with treatment adherence. Conditions of release, which include mandated medication compliance and strict supervision, appear effective at preventing individuals on CR from reoffending. If an individuals’ mental health deteriorates or they begin to engage in substance use, their release can be revoked and they can be returned to a secure mental health facility. CR programs also provide treatment for alcohol and drug use, and substance use has been linked to problematic outcomes. The combination of mandated medication and supervision, which are hallmarks of CR programs, provide some safeguards against criminal behavior, including violence” (p. 409).

“The current findings indicate violence risk assessment results from the HCR-20 have very little association with CR outcome, including both prediction and imminence of revocation, for those individuals on CR. Both the logistic regression and survival analysis yielded nonsignificant results with the HCR-20 scales as the independent variables. Of note, the RM scale just reached significance. Several recent studies have questioned the predictive validity of violence risk assessments in predicting the CR outcomes of NGRI acquittees. Consistent with previous studies, results from the current study suggest that most of the items and scales often used to predict violent behavior have limited ability to predict CR outcome. Nevertheless, the findings from this study do not suggest that risk assessment is irrelevant for patients being considered for CR” (p. 409).

“To the contrary, empirically focused risk assessment should remain at the forefront of treatment, management, and ultimately, play a role in release decisions. Specifically, violence risk assessment should inform risk management practices for individuals found NGRI, including informing interventions in community settings. That said, it may be most efficacious to conduct risk assessments early in the hospitalization process of NGRI commitment such that early identification of risk factors can inform treatment planning tailored toward risk mitigation. Early violence risk assessments provide a baseline for comparing subsequent risk assessments. Evidence of a lack of change or progress on items measuring dynamic risk variables may indicate a need for adjustments in treatment delivery” (p. 409-410).

“In addition, the current results indicate that the majority of CR revocations were due to mental health deterioration or rules infractions. To that end, data from the RM scale may be helpful in guiding release decisions and planning for which treatments are needed in the community. RM items require clinicians to consider potential barriers to successful transitions. In this sample, items measuring exposure to destabilizers and stress both predicted revocation. The RM scale predicted CR outcomes in a statewide sample of forensic patients from Georgia. Community programs that offer intensive case management and assertive community treatment may wish to target relevant risk management factors, such as helping patients better manage the situations causing stress and/or reduce exposure to situations which have been shown to destabilize them” (p. 410).

“Data on imminence of CR revocation provide findings that allow for three insights into CR revocation. First, most revocations occurred relatively early on, indicating that individuals on CR are most at risk for difficulties after they are discharged from a secure inpatient setting. Second, the number of prior criminal offenses predicted earlier revocation, which is consistent with previous research on CR. With regard to this finding, we cannot be certain if the number of offenses and imminence to revocation is a function of being more likely to engage in behaviors that lead to revocation (e.g., substance abuse) or whether a subset of NGRI acquittees may be considered by community providers to be high-risk patients. Any change in mental status or increase in concerning behaviors, even if relatively slight, may trigger a perception that revocation is necessary to reduce the risk
of criminal behavior or violence” (p. 410).

“In summary, data on CR support the notion that current violence risk instruments may be valuable in identifying relevant treatment targets aimed at reducing violence but are less useful in identifying which individuals are likely to be revoked or identifying those most at risk for early revocation. These findings may be related to the nature of violence risk assessment instruments, whose primary purpose is to evaluate factors related to violence, not revocation. In CR samples, including this one, the majority of individuals are revoked for rule violations, not violent or criminal behavior. As such, violence risk measures, like the HCR-20, may not be well suited to predict CR revocation, but may be more appropriate for use in making release decisions” (p. 410).

Other Interesting Tidbits for Researchers and Clinicians

“The decision to revoke an NGRI acquittee’s CR is often the combination of multiple factors, not the least of which are the various individuals involved in setting in motion the process of revocation (e.g., case manager, PSRB liaison). This study did not collect data on the decision-making process. In some cases, the decision-making process which leads to revocation may be highly transparent (e.g., an individual makes overt threats of violence as a result of medication noncompliance and is returned to an inpatient forensic unit) or may be opaque (e.g., case manager believes the individual’s mental
health is deteriorating). Subsequent research could consider the decisions of case managers and the threshold for initiating CR revocation. Very little research has examined factors underlying the revocation process. Research examining the front end of the process has noted that many decision makers do not employ clear and exacting standards for making recommendations of release. As stated by McDermott et al. (2008), part of the difficulty with CR decisions ‘lies in the inherently ambiguous definitions of mental illness and dangerousness, both of which are necessary for the continued confinement of insanity acquittees’ (p. 329). Making these decisions less ambiguous should be a goal of researchers, clinicians, and policymakers” (p. 410-411).

Join the Discussion

As always, please join the discussion below if you have thoughts or comments to add!

Authored by Amanda Beltrani

Amanda Beltrani is a current doctoral student at Fairleigh Dickinson University. Her professional interests include forensic assessments, professional decision making, and cognitive biases.

Forensic Risk Assessments Must Account for Cultural Considerations

A Canadian court recently ruled that risk assessment instruments must take cultural considerations into account in order to be used properly [Note: This decision was recently overturned]. The ruling may have widespread implications on the use of existing measures in forensic evaluations. This is the bottom line of a recently published article in Psychology, Public Policy, and Law. Below is a summary of the research and findings as well as a translation of this research into practice.

Featured Article | Psychology, Public Policy, and Law | 2016, Vol. 22, No. 4, 427-438

Forensic Risk Assessment and Cultural Diversity: Contemporary Challenges and Future Directions


Stephanie M. Shepherd Swinburne University of Technology
Roberto Lewis-Fernandez Columbia University and New York State Psychiatric Institute, New York, New York


A Canadian Federal court recently impugned the administering of 5 risk assessment instruments with Canadian Aboriginal prisoners. The ramifications of the ruling for the field are notable given the universal employment of risk instruments with Indigenous offenders and patients. Effectively, forensic clinicians and researchers can no longer overlook the role of culture in risk assessment—a robust academic dialogue on this subject matter is consequently warranted. This article explores how culture can shape the entire risk assessment process; from instrument construction and validation, to risk marker sensitivity, symptom articulation, and client-clinician interaction. Future directions for cross-cultural assessment are discussed.


violence risk assessment, Ewert v. Canada, cross-cultural mental health, transcultural psychiatry, forensic psychology

Summary of the Research

“On September 18, 2015 a Canadian court (Ewert v. Canada, 2015) strongly cautioned the continued use of five risk instruments (Hare Psychopathy Checklist Revised [PCL-R], Violence Risk Appraisal Guide [V-RAG], Sex Offender Risk Appraisal Guide [SORAG], Static 99, Violence Risk Scale—Sex Offender [VRSSO]) by the Correctional Service Canada (CSC) with Aboriginal inmates. The instruments evaluate risk for future violence (VRAG), risk for sexual violence/offending (SORAG, Static 99, VRS-SO), and the presence of psychopathic traits (PCL-R). The court concluded that the research base supporting the psychometric properties of the five above instruments with Canadian Aboriginal prisoners was currently inadequate. Additionally, actuarial tests were accepted by the court as susceptible to cultural biases and were therefore viewed as unreliable predictors of reoffense for Aboriginal inmates. By using such instruments, the CSC was found to be in breach of statutory obligations necessitating cultural responsiveness” (p. 427).

“The decision by the Canadian Federal Court has potentially wide-reaching ramifications for forensic risk assessment. The judicial rationale for this finding could, in theory, extend to other jurisdictions (i.e., United States, Australia, New Zealand, and United Kingdom) where actuarial instruments are used with Indigenous offenders. The Ewert v. Canada decision underscored an enduring concern in forensic assessment—that risk assessment instruments may not be cross-culturally suitable in their current state. While the court decision will undoubtedly galvanize academic inquiry into cross-cultural risk assessment, a developing research base has outlined apprehensions over the use of existing risk instruments with minority populations” (p. 427).

“Since their inception into clinical practice, risk instruments have predominantly assumed one of two templates: an actuarial design which encompasses a finite suite of predetermined risk items drawn from a construction sample of offenders or patients; or a Structured Professional Judgment (SPJ) model that enables clinical discretion to be anchored from a concert of risk factors identified in the risk literature. While the actuarial model minimizes clinical discretion in favor of statistical forecasting, the SPJ model provides a scientifically substantiated platform complemented by a rater’s contextual oversight. Both brands of risk assessment have more or less demonstrated equivalent predictive capacities for future offending” (p. 428).

“The risk instrument validation literature is extensive. The number of adult and adolescent risk instrument validation publications would exceed 1,000. However, validation samples predominantly comprise White participants. Of samples that include diverse populations, few compare within-sample validation estimates across participant cultural backgrounds, particularly for violent outcomes. The disproportionate focus on White offenders and patients contrasts with actual correctional populations where some minority groups (i.e., Indigenous; African American, Latino/as) are heavily overrepresented” (p. 428).

“Findings from the comparatively smaller number of investigations analyzing within-sample cross-cultural generalizability of risk instruments are inconclusive” (p. 428). Several research studies have found that, “Indigenous offenders regularly receive higher risk scores across several major adult and youth instruments [Static 99; LS/CMI; LSI-R; SAVRY; VRS-SO; YLS/CMI] compared with White offenders” and that risk assessment tools, “demonstrate reduced predictive validity for this group compared with White offenders” (p. 428). “Much less scientific attention has been afforded to African American and Latino/a offending populations in this space. The structural inequities and subsequent criminal justice contact experienced by both groups is well documented and a continual subject of forensic inquiry, yet this has not translated to comparable participant representation in risk validation investigations” (p. 428).

“This literature warrants several observations. First, evidence for the equivalent cross-cultural predictive validity of risk assessment instruments is unclear. The restricted and varied nature of the available findings precludes an assertion that instruments commensurately extend to minority groups. Moreover, such an assertion is perhaps premature given that minority groups do not receive the levels of representation in validation studies typically afforded to White populations. Second, while some studies have indicated that specific instruments demonstrate predictive accuracy for particular minority samples, this accuracy is recurrently poorer compared with White populations and necessitates interrogation. Third, the information derived from risk instruments often influences client liberties and treatment decisions—if instruments are consistently less accurate for particular cultural groups, then individuals from such groups may be legally and/or medically disadvantaged by their use. As such, cross-cultural biases that might unintendedly permeate the risk assessment process require deliberation” (p. 429).

Translating Research into Practice

“A client’s culture plays a pivotal role in the manifestation of risk. Culture determines behavioral norms and expectations, acceptable responses to threat, emotional presentation, modes of communication, goals, and motivations as well as explanations or remedies for illness, dysfunction, and delinquency. Moreover, culture helps to define what is a disease or harmful behavior. Risk assessment instruments are often crafted within the precepts of one cultural group, ignoring the broad experiential variability across diverse populations. Prediction estimates are often less accurate for ethnic minority groups, yet the notion of “close enough is good enough” enables the field to maintain current standards of risk assessment practice. This is unfortunate for some overrepresented, disadvantaged ethnic minority groups with histories of injustice and mistreatment by both criminal justice and mental health systems. Risk assessment instruments need to be reflective of the population they are measuring given the influence assessment information has on decision making impacting both public safety, offender treatment, and civil liberties. For these reasons all scientific efforts must be exhausted to ensure instruments are culturally fair, relevant, and nondiscriminatory. This article has outlined suggestions to facilitate such a process. The Ewert v. Canada decision should not be viewed as unreasonable by forensic clinicians and researchers, but rather as an opportunity to review and perhaps modify measures that may be unjustly impacting a portion of their clientele” (p. 434).

“Several actions can be undertaken to ensure that risk assessment is a fair and balanced exercise for all cultural groups concerned. The ultimate questions we must first examine are if, how and where culture impacts risk assessment. Put simply, is risk assessment compromised by cultural differences in (a) the underlying properties of instruments, (b) instrument item content, and (c) evaluator bias, or a combination of the three? These inquiries must be negotiated before contemplations of cultural renorming transpire” (p. 433).

“First, greater methodological sophistication is required to investigate factorial invariance in risk instruments. Rarely have risk instruments been subject to such testing. Without analysis of measurement equivalence, we cannot be sure if differences in mean scores across cultural groups are entirely attributed to differences between groups on the latent construct. To date, this connection has been largely assumed minus the necessary testing” (p. 433).

“Second, item content may need to be modified to mirror culturally specific explanations and manifestations of behavioral phenomena. This approach requires substantial input from multicultural health professionals and community members to ensure that item descriptions are in line with their experiential reality. This may involve the broadening of certain categories to encompass typical minority practices and understandings (i.e., family/kinship systems, cultural specific bereavement coping strategies), and in some cases demarcating ambiguous terminology (i.e., perceiving\hostile cues)… It is important that minority communities and stakeholders are engaged in the instrument development and validation processes. The specialized knowledge gained from cultural inclusion provides this exercise with a level of authenticity and credibility that existing tools have failed to secure” (p. 433).

“Third, AUC estimates must be complemented with additional predictive validity procedures. The comparison of regression slopes allows for the detection of significant differences in the outcome variable (offending) by equivalent increases in risk by cultural group. Positive and Negative predictive values also enable researchers to compare the prospective accuracy of high-risk and low-risk participants across culture” (p. 433).

“Our final suggestion pertains to a practitioner’s capacity to work effectively cross-culturally. Clinicians must be regularly educated in cultural competency or safety to ensure that risk assessment interviews gather relevant and meaningful information without unintentionally alienating, offending, or demeaning the client. A clinician’s cross-cultural aptitude can shape the risk assessment experience (i.e., patient candor or rapport, symptom interpretation, and risk precision) and should not be undervalued. Efforts to familiarize oneself with the practices, perspectives, values and belief systems of multicultural clientele can improve cultural awareness. An understanding of the historical injustices, structural inequalities and ongoing discrimination faced by minority clients, and the impact these incidents have on contemporary agency and wellbeing allows for a multilayered, contextualized assessment. A self-awareness of one’s own biases, privilege and assessor/patient power imbalance can also help minimize harmful heuristics and strengthen therapeutic relationships. The extent to which assessor biases play a role in cross-cultural risk assessment is an empirical question and warrants immediate academic scrutiny. This is particularly crucial given the capacity for assessor bias already identified in forensic assessment” (p. p. 434).

Other Interesting Tidbits for Researchers and Clinicians

“These concerns affect both actuarial and SPJ modes of risk assessment in different ways. The rigidity of actuarial instruments (the norms of which are often sample-specific) could impede the identification of cultural idiosyncrasies. For SPJ instruments, the clinical liberty afforded to risk evaluation enables raters to potentially incorporate culture-specific information. However, without adequate training on culture-specific behavioral norms, biases (stereotypical or otherwise) could easily pervade the process. SPJ instruments also draw items from a body of empirical literature based on decades of research, though much of it framed by Western methodologies, belief systems, and behavioral expectations. Arguably, both forms of risk assessment may contain structural biases in their constitution. Furthermore, the discretionary element of SPJ instruments similarly enhances or prejudices an assessment. Either way, it is important to acknowledge that an ethnocentric “starting point” for instrument construction” (p. 429).

“One difficulty with this approach is the potential essentialization of culture. There is often greater heterogeneity within a culture than between cultures. However, ignoring culture entirely in risk assessment (a colorblind approach) could mean a deferral to dominant culture contexts and a diminishing of cultural experiences. In light of these concerns, a culturally themed instrument should still take a “person-centered” approach and avoid social categorization. If an individual identifies with a particular cultural group, we must examine his or her own interpretations of behavior, deviance and illness, rather than associating the person’s cultural identity as a proxy for a suite of traits and characteristics. It may be useful here to consider the inclusion of cultural interview guidelines as an addendum to risk instruments to bridge cultural distance. For example, the DSM–5 Cultural Formulation Interview is a supplementary set of questions that help clinicians understand a patient’s cultural experience of illness as part of a diagnostic assessment” (p. 433).

Join the Discussion

As always, please join the discussion below if you have thoughts or comments to add!

Authored by Amanda Reed

Amanda L. Reed is a first year student in John Jay College of Criminal Justice’s clinical psychology doctoral program. She is the Lab Coordinator for the Forensic Training Academy. Amanda received her Bachelor’s degree in psychology from Wellesley College and a Master’s degree in Forensic Psychology from John Jay College of Criminal Justice. Her research interests include evaluator bias and training in forensic evaluation.

Gender Differences Apparent in the Use and Interpretation of the SAVRY for Juvenile Offenders

Forensic-Training-AcademyGender plays a role in the association between the risk/need and protective factors on the SAVRY and risk of violence. This is the bottom line of a recently published article in Law and Human Behavior. Below is a summary of the research and findings as well as a translation of this research into practice.


Featured Article| Law and Human Behavior | 2015, 1-15lhb


Identifying Gender Specific Risk/Need Areas for Male and Female Juvenile Offenders: Factor Analyses with the Structured Assessment of Violence Risk in Youth (SAVRY)


Ed. L. B. Hilterman, Tigburg University and Justa Mesura, Consultancy & Applied Research, Barcelona, Spain
Ilja Bongers, GGzE Center for Child and Adolescent Psychiatry, Netherlands, and Tigburg University
Tonia. L. Nicholls, University of British Columbia and Simon Frasier University
Chijs van Nieuwenhuizen, Tilburg University and GGzE Center tor Child & Adolescent Psychiatry, Eindhoven, The Netherlands


By constructing risk assessment tools in which the individual items are organized in the same way for male and female juvenile offenders it is assumed that these items and subscales have similar relevance across males and females. The identification of criminogenic needs that vary in relevance for 1 of the genders, could contribute to more meaningful risk assessments, especially for female juvenile offenders. In this study, exploratory factor analyses (EFA) on a construction sample of male (n = 3,130) and female (n = 466) juvenile offenders were used to aggregate the 30 items of the Structured Assessment of Violence Risk in Youth (SAVRY) into empirically based risk/need factors and explore differences between genders. The factor models were cross-validated through confirmatory factor analyses (CFA) on a validation sample of male (n = 2,076) and female (n = 357) juvenile offenders. In both the construction sample and the validation sample, 5 factors were identified: (a) Antisocial behavior; (b) Family functioning; (c) Personality traits; (d) Social support; and (e) Treatability. The male and female models were significantly different and the internal consistency of the factors was good, both in the construction sample and the validation sample. Clustering risk/need items for male and female juvenile offenders into meaningful factors may guide clinicians in the identification of gender-specific treatment interventions.


juvenile offenders, gender differences, factor analysis, risk management, violence risk assessment

Summary of the Research

“In recent decades research has demonstrated differences in risk factors between male and female juvenile offenders. For instance, risk factors related to family and social relationships have been found to be more important for female adolescents than for male adolescents (Cauffman, 2008; Fields & Abrams, 2010; Zahn et al., 2008). Compared with male juvenile offenders, female juvenile offenders also have a higher likelihood of exhibiting more mental health problems (Grande et al., 2012; Marston, Russell, Obsuth, & Watson, 2012). Further, according to Funk (1999), the presence of antisocial peers is more important for male juvenile offenders than for female adolescents. Factor analytic studies on risk assessment tools can provide information on differences in relevance of risk and protective factors between groups, for example genders. However, research on the psychometric structure of risk assessment tools for adolescents is scarce, especially regarding gender differences” (p.2).

Professionals in the Catalonian Justice Department in Catalonia, Spain were trained to administer the SAVRY to juveniles from 2006 until 2008. Interrater reliability for accuracy was excellent. Each assessment was given twice throughout the duration of the offenders stay in the juvenile sector. The following 5-factor model was assessed: antisocial behavior, family functioning, personality, social support, and treatability. Construction and validation samples were also used to support the 5-factor model.

“In the first step, we examined which factors could be found in the risk and protective items of the SAVRY that refer to gender specific risk/need areas. In the second step, the fit of the factor models for both male and female juvenile offenders was evaluated in validation samples and subsequently the difference between the models for both genders was measured. In the third step, we explored whether the found solutions could be accounted for by one underlying second-order construct. Finally, we analyzed the internal consistency of the different factors” (p.2).


“The inclusion of Poor School Achievement and Community Disorganization in the Antisocial Behavior factor for female, but not for male offenders, suggests that female offenders have more acting out behavior in familiar environments like school or their neighborhood. This could be related to findings that female adolescents are more likely to be involved in relational or social aggression (Herrman & Silverstein, 2012; Miller, Winn, Taylor, & Wiki, 2012; Zahn et al., 2008; that may take place in familiar settings like school, the neighborhood, and the family) than in direct aggression, while males tend to be more involved in direct aggression toward strangers (Card, Stucky, Sawalani, & Little, 2008; Herrman & Silverstein, 2012; Skara et al., 2008; Zahn et al., 2008). The inclusion of the Community Disorganization item in the Antisocial Behavior factor could also indicate that violent and delinquent behavior among female adolescents was, compared with male juveniles, closer to their living environment (Kling, Ludwig, & Katz, 2005). Our results are consistent with research that indicates that female juvenile offenders benefit more, compared with male adolescents, from programs like “Move to Opportunity” in which their families are moved to safer neighborhoods (Clampet-Lundquist, Edin, Kling, & Duncan, 2011; Kling et al., 2005)” (p.10).

“For both males and females, Factor 2, Family Functioning, refers to a history of dysfunctional family relations. The inclusion, for female offenders, of current poor parental management in the Family Functioning factor also suggests that the association between a dysfunctional family history and poor supervision and harsh or inconsistent discipline by caregivers could be stronger for females than for males” (p.11).

The authors also found that female juvenile offenders with behavioral problems were associated with substance abuse and self-harm. Treatability and Social Support also differed based on gender wherein female mental health needs were mostly related to treatment compliance, but male treatment compliance was associated with the absence of protective factors.

Translating Research into Practice

“The clinical interpretation of the gender specific factors found in this study may also suggest different strategies to reduce recidivism risk for male and female juvenile offenders. This is a potentially promising development since there are indications that traditional treatment interventions designed on the basis of risk management tools for male juvenile offenders are less effective for female juvenile offenders (Miller, Leve, & Kerig, 2012; Vitopoulos et al., 2012). Based on these findings, it is recommended that risk assessment tools like the SAVRY provide specific information regarding gender specific effects of risk and protective factors. By doing so, it could be easier for assessors to decide which items are more relevant for female or male offenders, interpret gender specific differences, and adjust risk reduction strategies accordingly” (p.11)

“For a future version of SAVRY it would be important to consider the inclusion of additional protective factors (e.g., positive future orientation, positive relationships with parents and peers, and prosocial romantic relationships, which could be especially relevant for female offenders; Oudekerk & Reppucci, 2010) as a means of improving balanced assessment and treatment planning (see also de Vries Robbé et al., 2015; Viljoen et al., 2014). In addition, coding the protective items on a three-point scale, instead of dichotomously, would offer the advantage of more flexibility in scoring, which would make the tool more dynamic and more prone to measure change over time” (p.12).

“The separate risk/need assessment models for male and female juvenile offenders, and/or the inclusion of empirically based information on specific gender effects of risk and protective items may guide clinicians in the identification of gender-specific treatment needs; this is essential information to inform the debate regarding the necessity of gender-specific treatment interventions” (p.12).

As the authors have suggested, identifying gender differences on the SAVRY can aid professionals in understanding both the motivations behind juvenile criminal behavior and the resulting risk associated with violence. It can also be helpful in developing programs and treatments necessary to target gender specific risk needs and reduce recidivism.

Other Interesting Tidbits for Researchers and Clinicians

“Future research that evaluates field implementations by comparing and contrasting research-based SAVRY assessments and clinical SAVRY assessments (e.g., looking at agreement by different categories of assessors; testing predictive accuracy) would be able to address some of these gaps in knowledge. Second, the generalizability of the results to non-Spanish populations of juvenile offenders is limited and this study needs to be replicated in other cultural settings in, for instance, other European countries, North America and Latin America. Finally, although the sample contained a large proportion of immigrants (male: 43.2%; female: 20.8%), replication in ethnic and cultural minorities would be important to gain more insight into risk and protective factors important to diverse populations” (p.12)

Join the Discussion

As always, please join the discussion below if you have thoughts or comments to add!

Authored by: Sarah Hartigan

6Sara Hartigan is a second year Forensic Psychology Master’s student at John Jay and hope to obtain a Ph.D. in Clinical Forensic Psychology in the future. My main areas of interest include clinical evaluations and developing treatment interventions within the forensic population.

YLS/CMI cross-cultural validity uncertain for low-risk Indigenous youth

Forensic-Training-AcademyThe YLS/CMI accurately predicts recidivism in high-risk Australian offenders, yet has difficulty in low and moderate risk groups, particularly for non-English speaking youth. This is the bottom line of a recently published article in International Journal of Forensic Mental Health. Below is a summary of the research and findings as well as a translation of this research into practice.

ijfmhFeatured Article| International Journal of Forensic Mental Health | 2015, Vol. 14, No. 3, 193-204

Does the Youth Level of Service/Case Management Inventory Generalize Across Ethnicity?


Stephane M. Shepard, Forensic Behavioral Science, University of Technology, Melbourne, Australia
Jay P. Singh, Global Institute of Forensic Research, Reston, Virginia; Faculty of Health Sciences, Molde University College, Molde, Norway, Department of Psychology, Universitat Konstanz, Konstanz, Germany
Rachael Fullam, Centre for Forensic Behavioural Science; Victorian Institute of Forensic Mental Health (Forensicare), Melbourne, Australia


Few studies have explored the cross-cultural utility of youth violence risk instruments. Moreover both the discrimination and calibration performance indicators of such instruments are rarely investigated. This study aimed to address both gaps in the literature by exploring the predictive validity of the YLS/CMI instrument for an Australian multi-ethnic cohort of young offenders in custody. The YLS/CMI total score was able to discriminate between reoffenders and non-reoffenders for the overall sample; however, a breakdown across ethnicity showed only strong effects for Australian English Speaking Background youth. Despite an inability to distinguish re-offenders from non-reoffenders for both Culturally and Linguistically Diverse and Indigenous youth, the instrument was able to accurately predict recidivism for high-risk youth for all ethnic groups. Findings highlight the importance of adopting both calibration and discrimination indicators when assessing predictive validity.


YLS/CMI, re-offending, violence risk assessment, forensic psychology, risk assessment

Summary of the Research

“The risk-need-responsivity (RNR) model is the dominant offender assessment and rehabilitation model in correctional systems around the globe. A key to this model is identifying higher risk individuals such that additional resources can be allocated to them. Criminogenic needs which are dynamic factors linked to offending behaviours are targeted for treatment so that an individual’s level of risk is reduced. The success of the RNR model spawned the development of a number of risk instruments which help practitioners design informed treatment strategies after identifying a client’s needs and level of risk. Only one of these instruments was developed for use with young offenders, namely the Youth Level of Service/Case Management Inventory” (p. 193).

“The YLS/CMI is a general risk/needs instrument for young offenders aged 12–17 years. It comprises static and dynamic risk factors encompassing previous offending behaviours, environmental influences, substance use and antisocial attitudes and behaviours. The YLS/CMI total score has demonstrated moderate associations with young offender recidivism across a variety of international correctional settings. A meta-analysis of 22 YLS/CMI related studies found that the instrument predicted general recidivism and to a lesser extent violent recidivism… Despite these results, a dearth of research has examined instrument’s cross-cultural generalizability… Risk assessment instruments must be able to demonstrate utility across cultural groups to avoid misclassification and subsequent correctional mismanagement. This is of critical importance for Indigenous Australians who have historical systematic discrimination, particularly within Australian government institutions including health and justice… Failure to validate risk instruments for minority groups could have deleterious repercussions given contemporary research pointing to the growing use and influence of the YLS/CMI during legal decision-making at various levels. Regional validity is of further value given that YLS/CMI validation studies conducted outside of North America have been shown to obtain lower predictive accuracy estimates compared to North American studies” (pp.193-195).

“This study aims to address these significant gaps in the literature by exploring the comparative predictive and discriminative ability of the YLS/CMI for three ethnic subgroups in a representative young Australian custodial population. Although this is, to our knowledge, the first Australian study to ascertain the cross-cultural validity of the YLS/CMI in custodial settings, we anticipate potential differences in item and total scores across ethnicity…The final sample comprised 207 participants remanded or serving a custodial order in two Youth Justice Centres in the Australian state of Victoria. The study capture rate over 12 months was exceptionally high ensuring a representative sample” (p.195). The sample was comprised of English Speaking Background (ESB; white and/or Caucasian participants of European descent.), Culturally and linguistically diverse (CALD; minority groups from non- English-speaking backgrounds) and Indigenous (IND; Aboriginal Australian or Torres Strait Islander heritages) youth. Participants were assessed using a semi-structured interview to complete the YLS/CMI and six month follow-up for recidivism.

“As anticipated, group differences across instrument total and item scores were observed. The Indigenous group obtained the highest aggregate mean total score. This score was significantly higher than the aggregate mean total score for the CALD group… Regarding specific risk items, the Indigenous group received significantly higher scores on two domains (i.e., substance abuse, leisure/recreation) compared to the CALD group” (pp.199-200).

“Using the suggested YLS/CMI high-risk cut-off point, CALD offenders who scored above the threshold were almost three times more likely to reoffend than those who scored below the cut-off. In comparison, ESB and IND youth who scored above the threshold were approximately two times more likely to have reoffended. High-risk youth from all three ethnic subgroups were more likely to violently reoffend than to not reoffend. These findings suggest that the discriminative utility of the YLS/CMI total score is clearly greater for clients deemed as a higher risk, regardless of ethnicity. Low to moderate scores on the YLS/CMI total are unable to distinguish reoffenders from non-reoffenders for CALD youth in particular” (pp. 200-201).

“High proportions of high-risk participants across ethnic groups re-offended. Fewer participants who were adjudged to be high risk violently reoffended, although the instrument still correctly predicted this outcome for the majority of high-risk offenders. These findings indicate that the YLS/CMI demonstrates strong predictive accuracy for general recidivism for high-risk youth regardless of ethnic background… Conversely, the instrument demonstrated low predictive ability for participants who scored below the threshold. The majority of participants with scores below the threshold regardless of ethnicity still reoffended” (p. 201).

Translating Research into Practice

“The YLS/CMI instrument was found to be able to identify with reasonable accuracy who will reoffend across different ethnic groups. However, in the study, this facility appears to transpire for ESB youth and only for CALD and Indigenous youth deemed as a high risk by the instrument. The instrument exhibits a noticeable weakness in identifying Indigenous and CALD recidivists from non-recidivists who are a lower risk for future offending. In clearer terms, the instrument has difficulties discriminating between high- and low-risk ethnic minority young offenders” (p. 201). “We cannot rely on the AUC [area under the curve; a statistical analysis commonly used to measure discriminant validity] estimate alone when ascertaining the predictive utility of a risk instrument. This study clearly demonstrates the ostensible predictive accuracy of the YLS/CMI for CALD and Indigenous youth despite those same groups receiving low AUC indices. As such, it is necessary to adopt both discrimination and calibration analyses in future investigations or else important clinical information is potentially lost. This study indicates that assessors can be confident that high-risk young offenders regardless of ethnicity, who are judged as a high risk on the YLS/CMI, are a strong likelihood of reoffending. At the same time, assessors need to be cognizant that the ability of the YLS/CMI to predict recidivism for lower-risk CALD and Indigenous offenders is uncertain” (p. 201).

“Clinicians need to be acutely aware of the prejudicial impact of risk instruments if they are misused, particularly given the supposed inaccuracy of the moderate/low-risk classification as a predictor of minority desistance. Mirroring earlier research, the Indigenous group in this study collectively received a comparatively higher mean instrument total score despite the instrument demonstrating lower discriminative value. A heavy focus on risk level (which for the YLS/CMI is associated with a cumulative total score) may need to be obviated when assessing clients from specific cultural backgrounds. Total scores may be artificially inflated because of unalterable static factors associated with minority disadvantage. Moreover, the unique influence of both static and dynamic factors on recidivism for some minority groups is without the requisite supporting empirical data. In addition, assessing CALD and Indigenous youth while the knowledge base on cultural specific protective factors is limited, presents challenges for balanced risk assessment with these groups. While the risk categorization of a client is undeniably an important part of the RNR model, replicative work needs to be conducted before we are confident that the instrument satisfactorily adjudicates cross-cultural risk. It may be that risk thresholds have distinct cut-off points on the instrument for different ethnic groups” (p. 202).

“It is important to remember that the focus of risk assessment, especially with under-researched minority groups, should be on specifically targeting problematic individual risk factors for risk management purposes. Reviewing the criterion coding for risk items to ensure cultural applicability is recommended. Notably, subsequent treatment schemes should be facilitated in ways that incorporate cultural learning styles. Therefore it is incumbent on forensic professionals to first, be aware of cultural differences and then translate such awareness into culturally service delivery. As the cross-cultural risk instrument validation and culturally specific risk factor literature are both still in their infancy, it is important that clinicians embrace and attain cultural proficiency in the interim” (p.202).

Other Interesting Tidbits for Researchers and Clinicians

“The higher instrument scores are a likely reflection of the well-documented disadvantaged and marginalized status of many Indigenous Australians who are exposed to deleterious circumstances including family disruption, unemployment, community violence, substance abuse, and stress… These factors undoubtedly impact the likelihood of engagement in criminal activity and contribute to Indigenous overrepresentation in the justice system. As such, the tool appears to accurately account for this greater exposure to risk items for Indigenous youth. However, it is important to note that widespread socioeconomic disadvantage may engender higher scores on unchangeable historical items for many Aboriginal forensic clients potentially inflating risk scores” (p. 199-200).

“Interestingly, participants who were not released from custody or did not permit access to their criminal records obtained significantly lower YLS/CMI total scores compared to the consenting cohort who were released into the community. While this finding may have accounted for the unusually high recidivism rate of the released group, it is difficult to speculate why young offenders who are potentially serving longer sentences or refuse access to their offending history are more likely to present as a lower risk for re-offense” (p.200).

Join the Discussion

As always, please join the discussion below if you have thoughts or comments to add!

Authored by Megan Banford

Megan BanfordMegan Banford is a master’s student in the Forensic Psychology program at John Jay College. She graduated in 2013 from Simon Fraser University with a B.A. (Honors) and hopes complete a PhD in clinical forensic psychology. Her main research interests include violence risk assessment and management, juvenile offenders and public policy.

Changes in Dynamic Factors During Treatment Predicts Postdischarge Recidivism

lhbChangeability of dynamic risk and protective factors during treatment is useful for measuring development in forensic patients related to reduced violent recidivism. This is the bottom line of a recently published article in Law and Human Behavior. Below is a summary of the research and findings as well as a translation of this research into practice.

Featured Article | Law and Human Behavior | 2014, OnlineFirst

Changes in Dynamic Risk and Protective Factors for Violence During Inpatient Forensic Psychiatric Treatment: Predicting Reductions in Postdischarge Community Recidivism


Michiel De Vries Robbé, Van der Hoeven Clinic, Utrecht, the Netherlands
Vivienne de Vogel, Van der Hoeven Clinic, Utrecht, the Netherlands
Kevin S. Douglas, Simon Fraser University and Mid-Sweden University
Henk L. I. Nijman, Radboud University and Altrecht, Den Dolder, the Netherlands


Empirical studies have rarely investigated the association between improvements on dynamic risk and protective factors for violence during forensic psychiatric treatment and reduced recidivism after discharge. The present study aimed to evaluate the effects of treatment progress in risk and protective factors on violent recidivism. For a sample of 108 discharged forensic psychiatric patients pre- and post-treatment assessments of risk (HCR-20) and protective factors (SAPROF) were compared. Changes were related to violent recidivism at different follow-up times after discharge. Improvements on risk and protective factors during treatment showed good predictive validity for abstention from violence for short- (1 year) as well as long-term (11 years) follow-up. This study demonstrates the sensitivity of the HCR-20 and the SAPROF to change and shows improvements on dynamic risk and protective factors are associated with lower violent recidivism long after treatment.


protective factors, risk assessment, treatment progress, violence, desistance

Summary of the Research

“Studies on the dynamic factors of the HCR-20 have shown good predictive validities for violence at short- as well as long-term follow-up and have demonstrated their usefulness for treatment guidance and evaluation of violence risk. However, few studies have investigated the relationship between changes in dynamic risk factors and treatment progress or reductions in violence risk. The present study investigated the usefulness of the joint assessment of the HCR-20 and the SAPROF for measuring changes in dynamic risk and protective factors during treatment. The aim was to evaluate the predictive validity of treatment progress as measured by the tools (i.e., reductions in risk factors and improvements in protective factors) for treatment success. It was hypothesized that dynamic risk factors and protective factors would change over time during treatment and that improvements on risk factors and protective factors would be negatively related to violent recidivism after treatment” (p. xx).

Participants were 108 male patients with a history of either general violent or sexually violent offending. Treatment length averaged 5.65 years and short-term recidivism was defined as any violence committed during a 12-month follow-up period, while long-term recidivism during the maximum average follow-up period of 11 years.

The HCR-20 and SAPROF were coded at the start and end of treatment, and comparisons were made between violent recidivists and those who did not violent reoffend.

Pretreatment risk and protective factors did not significantly differ between future recidivists and nonrecidivists. “However, at discharge future recidivists had higher scores on both dynamic HCR-20 subscales and lower scores on all three SAPROF subscales compared to those participants who turned out to be long-term desisters. These differences were significant for the clinical risk scale and for the internal and motivational protection scale, as well as for the total SAPROF and HCR-SAPROF index posttreatment scores” (p. xx).

“On average patients with final judgments that were rated one risk category higher than other patients (moderate vs. low, or high vs. moderate) were 11 times more likely to recidivate violently within one year after treatment and three times more likely to recidivate in a violent offense in the long run” (p. xx)

Translating Research into Practice

“At the end of treatment, protective factor total scores were significantly greater for the nonrecidivists and dynamic risk factor total scores significantly lower, indicating that the patients who changed the most while in treatment (i.e., those who showed the greatest improvements in risk and protective factors) were the most resilient to violent offending” (p. xx).

Positive changes during treatment appear to reduce recidivism in the short and long term. The likelihood of short term recidivism may be 2 times less likely with a 5 point reduction on the HCR-20 over treatment, and 2 times less likely with a 5 point increase on the SAPROF. Additionally, a 5-point increase over treatment on the SAPROF may indicate over 3 times reduced probability of short term recidivism. “Thus, those who changed the most during treatment still showed significantly lower violent recidivism rates long after treatment had ended… the present findings exemplify that treatment changes can have fairly stable long-term positive effects on abstention from violence.”

“The findings in this study suggest that the HCR-20 dynamic risk factors and the SAPROF protective factors could be useful to measure meaningful change in risk and protection and potentially provide for attainable treatment targets in clinical practice. This implies the HCR-20 and the SAPROF could be useful in guiding effective violence reduction efforts” (p. xx).

Professionals evaluating treatment effectiveness and efficacy may benefit from analyzing the treatment changes measured by the HCR-20 and SAPROF. “This connection between dynamic factor changes and treatment success is promising for the value of dynamic risk assessment tools for evaluating treatment progress, predicting successful treatment outcome and guiding treatment interventions in psychiatric practice. Furthermore, these findings are supportive of the apparent effectiveness of forensic psychiatric treatment in reducing the likelihood of violent reoffending in a high-risk offender population” (p. xx).

Other Interesting Tidbits for Researchers and Clinicians

“The finding that change scores have relatively stable predictive validities across follow-up times suggests that, at least for the current specific subgroup of patients, improvements in risk and protection levels during forensic psychiatric treatment of offenders at high risk of violent recidivism may have long-term effects in making society safer. Due to individual cases of recidivism getting much attention in the media, the general public opinion on the usefulness of treatment of offenders at high risk has become quite skeptical. The positive results on the changeability of risk and protective factors for violent offending observed in the present study may therefore present a valuable finding in support of the potential effectiveness of forensic psychiatric treatment.”

Join the Discussion

As always, please join the discussion below if you have thoughts or comments to add!

Contributing Author

BanfordMegan-picThis post was authored by Megan Banford.

Megan is a graduate student in the Forensic Psychology program at John Jay College. She graduated in 2013 from Simon Fraser University with a B.A. (Honors) and hopes to attain her PhD in clinical forensic psychology. Her main research interests include violence risk assessment and management, juvenile offenders and public policy.

Use of Violence Risk Assessments Tools a Growing Global Phenomenon

ijfmhThe use of psychological assessment instruments for evaluations of risk is a growing phenomenon around the world. This is the bottom line of a recently published article in International Journal of Forensic Mental Health. Below is a summary of the research and findings as well as a translation of this research into practice.

Featured Article | International Journal of Forensic Mental Health | 2014, Vol. 13, No. 3, 193-206

International Perspectives on the Practical Application of Violence Risk Assessment: A Global Survey of 44 Countries


Jay P. Singh, Sarah L. Desmarais, Cristina Hurducas, Karin Arbach-Lucioni, Carolina Condemarin, Kimberlie Dean, Michael Doyle, Jorge O. Folino, Verónica Godoy-Cervera, Martin Grann, Robyn Mei Yee Ho, Matthew M. Large, Louise Hjort Nielsen, Thierry H. Pham, Maria Francisca Rebocho, Kim A. Reeves, Martin Rettenberger, Corine de Ruiter, Katharina Seewald, & Randy K. Otto


Mental health professionals are routinely called upon to assess the risk of violence presented by their patients. Prior surveys of risk assessment methods have been largely circumscribed to individual countries and have not compared the practices of different professional disciplines. Therefore, a Web-based survey was developed to examine methods of violence risk assessment across six continents, and to compare the perceived utility of these methods by psychologists, psychiatrists, and nurses. The survey was translated into nine languages and distributed to members of 59 national and international organizations. Surveys were completed by 2135 respondents from 44 countries. Respondents in all six continents reported using instruments to assess, manage, and monitor violence risk, with over half of risk assessments in the past 12 months conducted using such an instrument. Respondents in Asia and South America reported conducting fewer structured assessments, and psychologists reported using instruments more than psychiatrists or nurses. Feedback regarding outcomes was not common: respondents who conducted structured risk assessments reported receiving feedback on accuracy in under 40% of cases, and those who used instruments to develop management plans reported feedback on whether plans were implemented in under 50% of cases. When information on the latter was obtained, risk management plans were not implemented in over a third of cases. Results suggest that violence risk assessment is a global phenomenon, as is the use of instruments to assist in this task. Improved feedback following risk assessments and the development of risk management plans could improve the efficacy of health services.


Violence, risk assessment, survey, international, mental health

Summary of the Research

A total of 2135 mental health professionals around the world were surveyed with respect to their risk assessment practices. The majority of the respondents were women (60.3%) and the average age of respondents was 43.9 years (SD = 11.0), with an average of 15.9 years (SD = 10.7) spent in practice. “Approximately half of their time in the past 12 months was spent on clinical activities, most often in forensic psychiatric hospitals followed by private practice and correctional institutions. Additional professional responsibilities over the past 12 months included administrative duties and teaching, with comparatively less time spent on research activities” (p. 198). Mental health professionals were contacted through web-based surveying, in a variety of languages, to examine the similarities and differences in risk assessment practices. The majority of participants were psychologists given that one of the inclusion criteria was that participants have previous training and practice in assessment.

Assessment Instruments Being Used

Two primary approaches to risk assessment have been identified: the actuarial approach, which combines protective and risk factors for violence in a mechanical, statistically-based way within a particular population of interest, and the structured professional judgment (SJP) approach, which provides an evidence-based structure within which the evaluator may use clinical discretion. The three most commonly used instruments for violence risk assessments identified in this study were the Historical, Clinical, Risk Management-20 (HCR-20), the Psychopathy Checklist-Revised (PCL-R), and the Psychopathy Checklist: Screening Version (PCL:SV) over both a lifelong and twelve-month time period. More than any other instrument, the HCR-20 was used to conduct, develop, and monitor risk for violence assessments and risk management plans. Personality instruments, such as the PCL-R, do not accurately predict violence as well as tools specifically designed for violence risk assessment but may still be helpful in the overall evaluation of an individual;’s risk for violence. In Europe, the HKT-30 (an instrument that is based on the HCR-20 but adds an additional 10 factors to consider) and the FOTRES were used almost exclusively.

Overall, both actuarial methods and SJP approaches had success and popularity among professionals. This research underscores that the decision regarding which of the two approaches to use should be based on the particular characteristics of the individual being evaluated as well asthe specific environment in which the professional works. Participants using SJP methods rated them as very helpful for their assessments. SJP tools like the HCR-20 or the HKT-30 appear to be more useful in international settings as actuarial approaches are impacted more by local factors than are SJP instruments, causing greater variability in probabilities of violence risk.

Implementation of Assessment Tools

“In terms of professional discipline, psychologists reported using instruments to structure their violence risk assessments more often than did psychiatrists or nurses both over their lifetime and in the past 12 months. Nurses reported more often obtaining feedback on whether their risk management plans had been implemented and that their risk management plans were implemented more often than psychologists or psychiatrists. Finally, psychologists reported taking significantly longer to conduct both unstructured violence risk assessments and structured violence risk assessments” (p. 199)

These authors “found that respondents who used instruments to develop management plans frequently did not know whether their plans had been implemented (44.6%) and, amongst those who did, proposed plans were not implemented in over a third of cases (34.6%)” (p. 202-203). Execution of risk assessment and management plans is necessary to manage risks.

Global Trends and Practices in Risk Assessment

“Compared to North America, Europe, and Australasia, respondents in Asia and South America reported completing a smaller proportion of risk assessments with the aid of an instrument” (p. 199). This highlights the differences in professional practices and differences across different countries. In addition, the lack of representation of these countries in the development of risk assessment research and practice may contribute to the slower adoption of these clinical practices in these countries. Other cultural restrictions included unauthorized translation of foreign materials or lack of awareness for instruments that are commercially available.

As the need for quality violence risk assessment grows on a global scale, it will be important to identify appropriate tools for risk assessments in South America and Asia.

Translating Research into Practice

This survey has two primary implications for practice: (1) we need to Improve communication regarding the quality and follow up of violence risk assessment and management plans, and (2) we need more information regarding how cultural factors impact the usage and quality of violence risk assessments throughout different countries.

Risk assessments provide a primary means of communication for mental health professionals. As such, there must be a continuum of channels for communication to ensure the clinical assessment is used to its fullest potential. If a mental health professional is required to perform assessments with no direct follow up or no follow through, then assessments lose their utility. Only one-third of participants reported receiving feedback on the accuracy of their assessments. “Social psychology research demonstrates that judgment accuracy increases when decision-makers receive feedback about their performance” (p. 202). It is important to receive feedback on the accuracy of assessments to determine whether additional training, education, or experience is needed. On the other hand, identifying professionals with strong assessment skills could help train other professionals to improve their skills. The main goal is to strengthen assessment skills and become experienced using these tools which will help improve the predictive validity of risk assessments.

“Moving forward, clinical training programs in these areas may wish to incorporate modules on violence risk assessment tools, funding agencies may wish to issue grants to encourage the development of novel instruments in native languages or the authorized translations of available tools, and there needs to be increased discussion in the field about the strengths and limitations of the contemporary literature and best-practice recommendations in international settings” (p. 203). Cultural differences across countries can be controlled to an extent as more collective research is transformed into international, unified models of clinical practice.

Other Interesting Tidbits for Researchers and Clinicians

Participants in this survey reported an average of 435.5 violence risk assessments in their lifetime with the use of a structured instrument for over half of those assessments. On average, about 35 violence risk assessments were performed within a twelve-month time frame. The time to conduct a structured assessment took an average of 7.8 hours while an unstructured assessment took an average of 2.8 hours. This 5-hours difference is significant but structured assessments offer more valid and reliable conclusions.

Younger participants and participants who had shorter careers were more likely to use structured violence risk assessments than older participants or participants with longer careers. This could be evidence towards a shifting philosophy in clinical practice that strives for the use of Structured Professional Judgment approaches.

Join the Discussion

As always, please join the discussion below if you have thoughts or comments to add!

Special Contributor

PatrickAndreaContributions to this post were made by Andrea Patrick.

Andrea Patrick is a first year masters student studying Forensic Psychology at John Jay College of Criminal Justice. In the future, she hopes to be directly working with forensic populations providing risk assessments and clinical evaluations.

Psychiatric Symptoms Do Not Precede Criminal Behavior for the Majority of Mentally Ill Offenders

lhbPsychiatric symptoms do not precede criminal behavior for the majority of mentally ill offenders. This is the bottom line of a recently published article in Law and Human Behavior. Below is a summary of the research and findings as well as a translation of this research into practice.

Law and Human Behavior 2014, Vol. 38, No. 5, 439-449

How often and How Consistently do Symptoms Directly Precede Criminal Behavior Among Offenders with Mental Illness?


Jillian K. Peterson University of California, Irvine
Patrick Kennealy University of South Florida
Jennifer Skeem University of California, Irvine
Beth Bray University of North Dakota
Andrea Zvonkovic Columbia University


Although offenders with mental illness are overrepresented in the criminal justice system, psychiatric symptoms relate weakly to criminal behavior at the group level. In this study of 143 offenders with mental illness, we use data from intensive interviews and record reviews to examine how often and how consistently symptoms lead directly to criminal behavior. First, crimes rarely were directly motivated by symptoms, particularly when the definition of symptoms excluded externalizing features that are not unique to Axis I illness. Specifically, of the 429 crimes coded, 4% related directly to psychosis, 3% related directly to depression, and 10% related directly to bipolar disorder (including impulsivity). Second, within offenders, crimes varied in the degree to which they were directly motivated by symptoms. These findings suggest that programs will be most effective in reducing recidivism if they expand beyond psychiatric symptoms to address strong variable risk factors for crime like antisocial traits.


offenders, mental illness, mental health symptoms, crime, recidivism

Summary of the Research

Approximately 1 million people in the criminal justice system have a major mental disorder such as schizophrenia, bipolar disorder, or major depressive disorder. This constitutes about 14-16% of the 7.3 million people who are under correctional supervision. A direct link between criminal behavior and psychiatric symptoms is the assumption of most policy initiatives, and the majority of research has focused on the role of these symptoms in causing crime.

Identifying how often offenders commit crimes that are motivated by symptoms of mental illness will vary based on how symptoms of mental illness are defined. Including only hallucinations and delusions provides a narrower definition than including more normative traits encompassing anger, impulsivity, irritability or aggression and will result in a lower estimate of mental health symptoms motivating criminal behavior.

As the definition of symptoms of mental illness broadens beyond psychosis, it becomes increasingly difficult to distinguish between symptoms and normative risk factors for crimes. For example, anger is correlated with symptoms of psychosis (delusions and command hallucinations), personality disorders (emotional instability), mood disorders (irritability and “anger attacks”), and post-traumatic stress disorder; however, anger is also a fundamental and functional human emotion that is a robust dynamic risk factor for violence among both general offenders and psychiatric inpatients. It is difficult to distinguish between anger as a human emotion and anger as a symptom of mental illness. To make this distinction, anger must be examined in relation to an individual’s typical behavior. This difficulty in distinguishing between symptoms and normative traits also remains true for impulsivity.

This research examined the relation of symptoms of mental illness and criminal behavior in a sample of 143 offenders with major depression, bipolar disorder, or schizophrenia spectrum disorders. Participants completed a 2-hour interview that focused on past criminal behavior, mental health symptoms, and the connection between the two. Among these 143 offenders, a total of 429 crimes were coded with respect to the degree to which they were directly a result of symptoms of mental illness.

Question 1: How Often Do Offenders Commit Crimes Motivated by Mental Health Symptoms?

The first aim of this study was to examine how often psychiatric symptoms are related to criminal behavior. Symptoms of psychotic disorder, bipolar disorder, and depression were considered and analyzed with respect to whether the symptoms immediately preceded the crime and increased its likelihood of occurrence.

Proportion of direct crimes by diagnosis

Of the 429 crimes coded, almost two thirds (64.7%) were coded as completely independent and less than one in 10 (7.5%) were coded as completely direct. About one third (27.9%) of crimes fell in the middle of the continuum, indicating mixed or moderate symptom involvement.

Schizophrenia spectrum distribution

Of crimes committed by participants with schizophrenia spectrum disorders, 23% were completely or mostly related directly to symptoms. Of crimes related to schizophrenia spectrum disorders, 42% were crimes against another person, 42% were property crimes, and 16% were minor crimes such as trespassing.

Bipolar disorder

Of crimes committed by participants with bipolar disorder, 62% were completely or mostly related directly to symptoms. Of crimes related to symptoms of bipolar disorder, 39% were crimes against another person, 42% were property crimes, and 19% were minor crimes.


Some 15% of crimes committed by participants with depression were completely or mostly related directly to symptoms. Of crimes related to symptoms of depression, 39% were crimes against another person, 15% were property crimes, and 46% were minor crimes” (p. 444).

Question 2: How Consistently Are Symptoms of Mental Illness Linked to Criminal Behavior Over Time, Across Incidents?

Additionally, this study examined whether the relationship between symptoms and crime is consistent with the offender’s criminal history. That is, the extent of consistency between symptoms and crimes was examined to determine whether there is a subgroup of mentally ill offenders that consistently commit crimes in response to their symptoms.

“The 18% of crimes coded as mostly or completely related directly to symptoms were scattered among 38% of offenders. Of the 38% of offenders with at least one direct crimes, most (66.7%) also committed at least one crime that was coded ‘mostly or completely’ independent. This suggests that the relationship between symptoms and criminal behavior varies over time within an offender” (p. 445).

The “results indicate that little or no variance in direct continuum scores can be attributed to offenders—that crimes are inconsistently related to symptoms within a given offender, over time. The majority of offenders who committed a “mostly or completely” direct crime committed at least one crime independent of symptoms as well” (pp. 445-446).

Translating Research into Practice

These “findings question the accuracy of past distinctions between offenders with mental illness whose criminal behavior is or is not directly caused by symptoms. These findings also underscore the fact that symptoms other than psychosis can lead directly to criminal behavior. As noted earlier, however, distinguishing between symptoms that are specific to major mental disorder and features that may be found among offenders without mental illness can be difficult. Further investigation of specific symptoms of Axis I disorders in causing crime is needed.

[These] findings also question the current policy focus on controlling symptoms as a means toward recidivism reduction. As shown in prior literature reviews, system solutions like diversion programs that focus predominantly on symptom control tend to have little effect on recidivism. The findings in this study indicate that effective mental health treatment may prevent a minority of crimes from occurring (about 18%, according to our findings), but would likely not improve criminal justice outcomes for the vast majority of offenders with mental illness. In keeping with past, [these] results suggest that psychiatric symptoms are not robust, independent risk factors for criminal recidivism.

Instead, most offenders with mental illness—whether they occasionally commit a crime that is directly motivated by symptoms or not—may benefit from interventions that reduce recidivism for offenders without mental illness. For example, cognitive– behavioral treatment focused on criminal cognition or services that target variable risk factors for high-risk offenders have been shown to reduce criminal recidivism for general offenders. Developing a better understanding of causal factors for recidivism among offenders with mental illness can inform better correctional interventions, both in institutions and probation and parole” (p. 447).

Join the Discussion

As always, please join the discussion below if you have thoughts or comments to add!

Special Contributor

BeltraniAmandaContributions to this post were made by Amanda Beltrani.

Amanda M. Beltrani is a first-year graduate student in the Forensic Psychology masters program at John Jay College of Criminal Justice in New York. Her professional interests include forensic assessments and criminal matter evaluations. Amanda plans to continue her studies in a doctoral program after completion of her Masters degree.

Please Join Us for a 3-hr Live Webinar on the HCR-20 Version 3

HCR20We are pleased to announce a 3-hour live webinar on the newly revised Version 3 of the HCR-20 to be led by threat assessment expert and lead author of the measure, Dr. Kevin Douglas. The webinar will be held on 3 different dates and offered at 2 different times on each date. We are limiting space in each session so participants will have the opportunity to interact with Dr. Douglas and ask questions regarding the development, use, and implementation of this risk assessment instrument. If you are interested in learning more about this important risk assessment instrument or are considering adopting it for use in your clinical practice, please join us for this webinar.

About the HCR-20 Version 3

The HCR-20 is the most commonly used violence risk assessment measure across 44 different countries. It helps professionals in correctional, mental health, and forensic settings make decisions about who poses higher versus lower risk for violence, either within institutions or in the community, and to devise and monitor violence risk management plans. The HCR-20 (Version 2) has been evaluated in more than 100 studies and implemented or evaluated in at least 32 countries.

Version 3 of the HCR-20 was recently released. Version 3 maintains the basic features of Version 2, but has additional features that will help decision makers to determine which risk factors are most relevant at the individual level, how to produce a meaningful case formulation, how to develop helpful risk management plans, and how to make decisions about level of violence risk. In addition, some of the items have changed from Version 2 to Version 3.

About Dr. Kevin Douglas

Douglas, Kevin.jpgKevin S. Douglas is Associate Professor and Associate Chair, Department of Psychology, Simon Fraser University. He is also a Guest Professor of Applied Criminology at Mid-Sweden University, and a Senior Research Advisor at the University of Oslo. Dr. Douglas received his law degree in 2000 from the University of British Columbia, and his Ph.D. in clinical (forensic) psychology from Simon Fraser University in 2002. He received a Michael Smith Foundation for Health Research Career Scholar Award (2005-2010), and was the recipient of the Saleem Shah Award for Early Career Excellence in Psychology and Law (2005), awarded jointly by the American Psychology-Law Society and the American Academy of Forensic Psychology. His research has been funded by the National Science Foundation (USA), Canadian Institutes of Health Research, Social Sciences and Humanities Research Council of Canada, and the Michael Smith Foundation for Health Research. His research addresses violence risk assessment and management, the association between various mental and personality disorders (i.e., psychosis; psychopathy) and violence, and dynamic (changeable, treatment-relevant) risk factors. He is co-author of the Historical-Clinical-Risk Management-20 (HCR-20) violence risk assessment measure, which has been translated into 18 languages and is used broadly around the world (roughly 35 countries) in correctional, forensic, and psychiatric settings to help guide decisions about violence potential and how to reduce it. Dr. Douglas is lead author on the latest (third) revision of the HCR-20, called the HCR:V3. More recently, he has been conducting research on other violence-related adverse experiences, including suicide-related behavior, and being victimized by violence. On these topics, Dr. Douglas has authored over 100 journal articles, books, or book chapters.

About the Webinar

This is a 3-hour live webinar that allows for interaction with Dr. Douglas. Participants will receive a certificate of completion as well as 3 Continuing Education credits for attending.

Dates and Times

November 5th, 2014 | 9:00am (PST)/Noon (EST)     OR     1:00pm (PST)/4:00pm (EST)

November 10th, 2014 | 9:00am (PST)/Noon (EST)     OR     1:00pm (PST)/4:00pm (EST)

November 12th, 2014 | 9:00am (PST)/Noon (EST)     OR     1:00pm (PST)/4:00pm (EST)

Learning Objectives

This 3-hour live webinar will:

• Describe the revision process for the development of Version 3 and outline the changes that differentiate Version 2 from Version 3

• Provide an overview of the items and a discussion regarding how the measure can be used in practice (implementation issues)

• Provide an opportunity to interact with Dr. Douglas and ask questions


The cost for this webinar is $149


Registration for this 3-hour live webinar is now open. Interested participants can register by clicking the “Register Now!” button below.


Register Now

Please use the buttons below to share this post and we look forward to seeing you there!!

Communication Format Influences Perceptions of Offender Risk

lhbPerceptions of risk vary by the communication format used and laypersons appear more receptive to communication messages that provide an interpretive label than those that provide statistical results. This is the bottom line of a recently published article in Law and Human Behavior. Below is a summary of the research and findings as well as a translation of this research into practice.

Featured Article | Law and Human Behavior | 2014, Vol. 38, No. 5, 418-427

Same Score, Different Message: Perceptions of Offender Risk Depend on Static-99R Risk Communication Format


Jorge G. Varela, Sam Houston State University
Marcus T. Boccaccini, Sam Houston State University
Veronica A. Cuervo, Sam Houston State University
Daniel C. Murrie, University of Virginia
John W. Clark University of Texas at Tyler


The popular Static-99R allows evaluators to convey results in terms of risk category (e.g., low, moderate, high), relative risk (compared with other sexual offenders), or normative sample recidivism rate formats (e.g., 30% reoffended in 5 years). But we do not know whether judges and jurors draw similar conclusions about the same Static-99R score when findings are communicated using different formats. Community members reporting for jury duty (N = 211) read a tutorial on the Static-99R and a description of a sexual offender and his crimes. We varied his Static-99R score (1 or 6) and risk communication format (categorical, relative risk, or recidivism rate). Participants rated the high-scoring offender as higher risk than the low-scoring offender in the categorical communication condition, but not in the relative risk or recidivism rate conditions. Moreover, risk ratings of the high-scoring offender were notably higher in the categorical communication condition than the relative risk and recidivism rate conditions. Participants who read about a low Static-99R score tended to report that Static-99R results were unimportant and difficult to understand, especially when risk was communicated using categorical or relative risk formats. Overall, results suggest that laypersons are more receptive to risk results indicating high risk than low risk and more receptive to risk communication messages that provide an interpretative label (e.g., high risk) than those that provide statistical results.


Static-99R, communication, risk assessment, sexual offender

Summary of the Research

“Research suggests that between 75% and 80% of laypersons believe that sexual offenders will reoffend. In contrast, meta-analytic research has found a sexual recidivism rate of approximately 11.5%. This large discrepancy between public perceptions and the research that forms the basis of risk estimates highlights the importance of examining how risk messages are understood and used in legal decision making…Evaluators who use risk assessment instruments must effectively communicate their findings to those who actually make decisions about offenders, including judges and jurors. Risk assessment results are of little value if experts cannot effectively communicate their findings to legal decision makers. Thus, the primary goal of the current study was to examine whether different risk communication formats for the same Static-99R score lead venirepersons (i.e., community members who presented for jury duty) to reach the same or different conclusions about recidivism risk. A secondary goal was to examine whether venirepersons view some risk communication message formats as more useful or understandable than others” (pp. 418-419).

“The current study extends risk communication research into the domain of sexual offender proceedings, in which risk communication, particularly based on the Static-99R, is ubiquitous. [The authors] compared the influence of three Static-99R risk communication formats—categorical, risk estimate, and relative risk—on venire- persons’ perceptions of sexual offenders (e.g., dangerousness and likelihood of reoffense). Existing research suggests that venirepersons view categorical messages as more useful than those with statistical information and, as a result, we should expect the greatest differences in perceptions of dangerousness and likelihood of reoffending between high- (Static-99R = 6) and low- (Static-99R = 1) scoring offenders when information is presented categorically. Existing research also suggests that venirepersons may devalue risk assessment results when they indicate low risk, suggesting that we should expect our participants to report that Static-99R findings are less useful when they indicate low risk. We also should expect venirepersons to be especially likely to report being confused by relative risk messages, because they combine technical statistical information with a comparison to the undefined ‘typical’ sex offender” (p. 420).

“Participants read a two-page document that included one of six versions of a sex offender risk assessment case…The case descriptions varied across two dimensions—the offender’s Static-99R score (i.e., risk level) and risk communication format. The low-scoring offender was assigned a Static-99R score of 1 and the high-scoring offender was assigned a Static-99R score of 6.

The three risk communication formats were categorical, risk estimate, and relative risk. In the categorical conditions, the offender’s risk was communicated in the following manner: ‘According to the Static-99R developers, Mr. Donaldson’s score of 1 (or 6) places him in the Low (or High) risk category for being charged with another sexual offense.’ In the relative risk conditions, the case description reported the Static-99R score and the offender’s risk was described as ‘three fourths the recidivism rate of the typical sex offender’ for the low-score condition and ‘2.91 times the recidivism rate of the typical sex offender’ for the high-score condition…In the risk estimate conditions, the case description read ‘in the Static-99R research sample, 9.4% (or 31.2%) of men who scored 1 (or 6) on the Static-99R (like Mr. Donaldson) were rearrested for a sexual offense within five years’” (p. 420).

Participants were 211 adult community members called for jury duty in an urban jurisdiction. The average age was 44.12 years (SD = 14.03) and the racial/ethnic breakdown was: 53.1% White, 44.1% Black, and 2.9% other ethnicity. Participants were asked to make several ratings regarding the hypothetical offender, including their perceptions of likelihood of committing a new sexual offense in the next 5 years, level of risk, and importance of the Static-99R results.

Perceptions of the Offender

“Overall, 95% of participants indicated that the offender would commit a new sex offense in the next 5 years. Given this lack of variance, [the authors] did not compare responses across study conditions” (p. 421).

With respect to the ratings of risk, “[o]verall, the findings indicate that participants viewed the high- and low-scoring offenders as having significantly different levels of risk when Static-99R results were communicated using a categorical format, but not when results were communicated using relative risk or recidivism rate formats” (p. 421).

“There was also some evidence that participants who read about the same Static-99R score viewed the offender differently depending on risk communication format. Among participants who read about a Static-99R score of 6, those who read a categorical message assigned higher risk composite ratings than those who read a recidivism rate message and those who read a relative risk message. Among participants who read a Static-99R score of 1, those who read a categorical message assigned the lowest composite ratings, although they were not significantly lower than those from participants who read recidivism rate and relative risk messages” (pp. 421-422).

Perceptions of the Static-99R Results

In terms of perceptions regarding the importance of the Static-99R results, participants rated the Static-99R results as being more important in the high-score condition. With respect to the type of communication used, in the high-score condition, the categorical message was rated as the most important whereas in the low-risk condition, it was rated as the least important.

Translating Research into Practice

“Every day, the justice system makes decisions about sexual offenders after considering risk communication, particularly based on the Static-99R. Yet, so far, no studies have examined how decision makers understand or use this risk communication. [These authors] found that different risk communication formats for the same Static-99R score might lead venirepersons to different conclusions about recidivism risk. When the offender had a high Static-99R score, participants rated him as more dangerous and likely to reoffend when the risk communication included a categorical message than a numerical message (i.e., relative risk or risk estimate). When the offender had a low Static-99R score, risk ratings were generally high but similar across the three risk communication conditions. Indeed, participants viewed the high- and low-scoring offender as similarly likely to reoffend when risk was communicated numerically.

Participants’ responses to questions about the importance and understandability of Static-99R results may help explain the relatively high-risk ratings in the low-score conditions. Participants presented with a low Static-99R score were more likely than those presented with a high score to report that the Static-99R results were difficult to understand, and they also rated the Static-99R results as relatively unimportant. These patterns applied more clearly to participants presented with categorical and relative risk communication messages than those presented with recidivism risk messages. Indeed, participants who were presented with recidivism rate messages responded similarly to each of [the] measures, regardless of whether they read a risk communication message that corresponded with a high or low Static-99R score” (p. 424).

“Confirmation bias may help explain the varied pattern of participants’ responses to risk communication messages. Confirmation bias is the tendency to selectively seek and interpret information in a manner consistent with one’s beliefs and expectations. In [this] study, nearly all participants, across all experimental conditions, reported that the offender would likely reoffend within the next 5 years. It is reasonable to assume that many participants in [this] study assumed that most sexual offenders reoffend, as have participants in other research…One implication of these findings is that experts and attorneys (most likely defense/respondent attorneys) should consider directly addressing jurors’ a priori beliefs about sexual offenders…But confirmation bias cannot completely explain [these] findings, as the risk communication format also seemed to influence perceptions of offender risk. For example, [the] findings suggest that participants perceived only some of the Static-99R high-score messages as actually conveying higher risk than the corresponding low-score messages. Participants clearly assigned higher risk ratings to the high- scoring than low-scoring offender when risk was communicated using categorical messages. In contrast, there was only a small difference in risk ratings between the high- and low-score (risk) conditions when participants were presented a relative risk message and almost no difference when participants were presented a recidivism rate message.

The finding that community members were most responsive to risk communication formats that provide interpretive guidance (i.e., categorical labels) is consistent with clinicians’ preferences for communicating risk using nonnumerical messages. But one danger in using categorical labels is that clinicians become less descriptive and more prescriptive, implicitly recommending a course of action to the court rather than simply providing factual data. Thus, even using categorical messages requires conscientious clinicians to communicate with caution and clarity” (p. 424).

“[These] findings also suggest that venirepersons either neglect risk ratios or do not understand them, especially when the risk ratio suggests low risk. Participants presented with a low score and a relative risk message should have reported that the offender was less likely than other offenders to reoffend. Yet, almost 80% of participants in the low-score/relative risk condition reported that the offender was more likely than other offenders to reoffend, despite having just read that his recidivism rate was ‘approximately three fourths’ that of the typical offender.

There are several possible explanations for this finding. The first is that jurors simply view all offenders as at a high risk for reoffending, regardless of risk communication messages. Again, a second possible explanation is that innumeracy may have led to a misunderstanding of the risk message. Regardless of the explanation, the implication for practice is that experts need to spend time explaining the recidivism risk of the typical offender. In other words, relative risk communication information may need to be accompanied by a thorough explanation of its basis and application. Testimony explaining the meaning of risk ratios and how they apply to a specific case may be necessary to ensure that decision makers understand the risk information when formulating risk” (p. 425).

“Overall, the findings of the current study suggest that venirepersons approach decision making with the expectation that sexual offenders are dangerous and quite likely to sexually reoffend. Their expectations appear resistant to influence by risk assessment messages, especially when they are informed an offender’s risk is low. Researchers and clinicians in forensic psychology are understandably focused on developing and properly using instruments, including actuarial instruments. But until the field can communicate to decision makers the results of these measures—in understandable and constructive ways—the practical value of rigorous assessment methods will be greatly constrained “ (p. 425).

Other Interesting Tidbits for Researchers and Clinicians

“A consistent finding across risk, understandability, and importance measures was that participants who read about recidivism rates responded similarly, regardless of whether they read about the rate for a low or high Static-99R score. One possible explanation for this pattern of findings is that laypersons so consistently and so greatly overestimate recidivism risk that actual estimated rates have little impact on their opinions. Another possible explanation is that the difference in recidivism rates (9.4% vs. 31.2%) was too small to be salient or meaningful, despite having used rates from the Static-99R normative group with the highest recidivism rates (high risk/needs). A final possibility is simply that innumeracy (i.e., lack of understanding and facility with numbers and mathematical concepts) left [the] participants unable to make use of recidivism rate data. Scholars have demonstrated that innumeracy is a significant problem among legal decision makers and has hindered mock jurors’ perceptions of violence risk.

One implication of these findings is that it is unlikely that the difference between high and low Static-99R score recidivism rates from other Static-99R normative groups will be salient to jurors. The Static-99R & Static-2002R Evaluator’s Workbook suggests that rates from the routine sample norms, which are lower than the high-risk/need norms used in the current study, are appropriate in most sex offender evaluations. Using the routine sample norms, the estimated 5-year sexual recidivism rates for scores of 1 and 6 are 3.8% and 14.7%, respectively. Because only 4% of offenders score higher than 6 on the Static-99R, most of the 5-year recidivism rates reported by evaluators will be 14.7% or below” (p. 425)

Join the Discussion

As always, please join the discussion below if you have thoughts or comments to add!

Causal Attributions Influence Scores on the HCR-20

ijfmhCasual attributions—judgments about the internal or external cause of an individual’s actions—influence scores on the HCR-20. This is the bottom line of a recently published article in International Journal of Forensic Mental Health. Below is a summary of the research and findings as well as a translation of this research into practice.

Featured Article | International Journal of Forensic Mental Health | 2014, Vol. 13, No. 1, 8-17

Investigating the Influence of Causal Attributions on Both the Worksheet and Checklist Versions of the HCR-20


Jennifer Murray, School of Life, Sport and Social Sciences, Edinburgh Napier University, Sighthill Campus, Sighthill Court, Edinburgh, Scotland
Kathy E. Charles, School of Life, Sport and Social Sciences, Edinburgh Napier University, Sighthill Campus, Sighthill Court, Edinburgh, Scotland
David J. Cooke, Department of Psychology, Glasgow Caledonian University, 70 Cowcaddens Road, Glasgow, Scotland
Mary E. Thomson, Newcastle Business School, Northumbria University, City Campus East, Newcastle Upon Tyne, England


Attribution theories suggest that when assessing an individual’s actions, judgments are made about the cause of these behaviours and often these judgments focus on internal or external causal explanations. The current research investigated the effects of internal and external attribution on the scoring of the HCR-20 and the possibility of differences in scoring between two ways of using the HCR-20 (using the HCR-20 as a worksheet versus checklist). No differences were present in the scoring between the checklist versus worksheet. Attribution effects were present within the Historical Scale, Clinical Scale, and overall scoring of the HCR-20. Ratings were higher within the internal attribution condition than the external one, indicating that judgments made using the HCR-20 are subject to attribution effects in a similar manner as unaided violence risk assessments.


attribution, decision-making, clinical judgment, HCR-20, violence risk assessment

Summary of the Research

“Attribution theories posit that in order to understand the cause of our own or others’ behaviors, we will apply either an internal or an external explanation for the behaviour. When an internal cause is attributed to an individual’s behavior, such as personality or temperament, the individual is often considered to be in control of or responsible for their behavior. Conversely, when the cause of an individual’s behavior is attributed externally, for example, as a result of the environment, the behavior is considered to be less in the individual’s control and their level of responsibility for the outcome is therefore perceived to be low.

The current research aimed to investigate whether the HCR-20 would be affected by attribution effects [and] whether the way in which it is completed will affect the outcome of the risk assessment” (p. 8).

“The HCR-20 is one of the most widely researched, commonly used, and leading structured professional judgment (SPJ) violence risk assessment tools…SPJ tools, such as the HCR-20, do not aim to predict future violence, but instead aim to guide the clinician through the risk assessment using risk factors derived from both empirical and practice-based evidence and inform suitable risk management plans and interventions” (p. 9).

“The HCR-20 should be completed in conjunction with the HCR-20 manual and a ‘worksheet’ is often used to facilitate this. The worksheet guides the user in gathering information, to consider the presence and relevance of risk factors associated with the case, consider different hypothetical scenarios of violence (best-, most realistic-, worst-case scenarios), develop risk management strategies suitable to the case, and finally to document their summary judgments. Spaces are provided underneath each risk item where the clinician can provide evidence from the case sup- porting his or her assertions of whether the item is present, to what extent and whether its presence (or lack of) is of relevance to the risk assessment. In presenting and combining the evidence in this way, it is thought that decision making will potentially be less biased. However, not all clinicians use the worksheet in practice, or indeed the manual (this is not best practice, however, as the manual should be used to guide all assessments); the HCR-20 is at times used purely as a checklist, where the 20 factors are numerically coded on a 0–2 scale. In both the checklist and worksheet formats, the assessor should also formulate a final risk judgment indicating whether the person poses a low, moderate, or high risk” (p. 10).

“Attribution bias occurs when judgments are made about a person or situation but greater or lesser weight is placed on internal or external factors, skewing the viewpoint formed…When information relating to an individual’s internal state, such as mood or personality, is focused upon, internal causal theories about their behavior tend to be formed (internal attribution) and their actions are attributed on something internal to that person. If, on the other hand, external or situational factors are focused on, the individual’s behavior is attributed to something external to them and outside of their control (external attribution)” (p. 10).

Studies on attribution have “found that individuals whose behavior was described to imply more control over their actions (i.e., internally attributed) were considered more dangerous and to have committed a more serious crime than when the individual was described to emphasize the role of external factors on their behavior (i.e., externally attributed), despite the description of the crimes and key features of the individual remaining constant across the internal/external conditions. In order to investigate these attribution effects in relation to more modern risk assessment practice, it is crucial to identify whether attributionally manipulated in- formation influences the scoring of SPJ tools, such as the HCR-20, or whether this effect is confined to unaided clinical judgments” (p. 11).

“The current research [aimed] to investigate the effects of internally and externally manipulated attributional information on the scoring of the HCR-20, and whether this differs across completing the HCR-20 as a checklist or using the worksheet format… A 2×2 independent groups design was used, with the independent variables being the two modes of completion for the HCR-20 (checklist versus worksheet) and the two attributionally manipulated crime-based scenarios (internal versus external). The dependent variable was the numeric responses to the items measured on the HCR-20” (p. 11).

Participants were 40 master’s level psychology students who had an average of 17.5 years of education.

“A significant main effect of attributional manipulation was found for the Historical scale, the Clinical scale, and for the overall scoring of the HCR-20…Participants scored the offender significantly higher in the internal manipulation than in the external one for the Historical and Clinical scales and for the overall HCR-20 score. No main effect of attribution was found in the scoring of the Risk Management scale” (p. 13).

“No main effect of the ‘version’ of the HCR-20 used (checklist versus worksheet) was apparent in any of the three scales or the overall HCR-20 score, indicating no differences” (p. 13).

Differences across attributional manipulations appeared to be most prominent for the following items: Previous Violence, Substance Use Problems, Psychopathy, and Personality Disorder on the Historical scale; Negative Attitudes and Unresponsiveness to Treatment on the Clinical scale; and Exposure to Destabilizers on the Risk Management scale.

Translating Research into Practice

“The finding that individuals completing the HCR-20 as a checklist rated the offender to be a ‘high risk’ more often in the internal manipulation than in the external one provides further evidence that attribution influences assessments of violence. These findings are of paramount importance as they highlight that even when using an SPJ tool, judgments can still be affected by attribution bias.

The current findings identified a number of areas where attribution effects may be more present within the HCR-20; both in terms of scale level and item level. As the HCR-20 is divided into three scales (Historical, Clinical, and Risk Management) the scales most affected by attribution can be identified. Attribution effects were present in the Historical and Clinical scales, but not in the Risk Management scale. In both of the affected scales, numerical judgments of risk were higher in the internal manipulation than in the external one” (p. 15).

In terms of item-level differences in attribution-related scoring, it is important for clinicians to be aware of those items in which attribution effects are most prominent (i.e., Previous Violence, Substance Use Problems, Psychopathy, Personality Disorder, Negative Attitudes, Unresponsiveness to Treatment, and Exposure to Destabilizers) so that they might be mindful of these as a means of reducing attribution effects on these items and the overall judgment of risk that is made. Bearing these attribution effects in mind will help clinicians improve their decision making and risk formulation using the HCR-20.

“The current findings further indicate that the scoring of the HCR-20 does not differ as a function of the checklist versus worksheet modes of completion. This finding is encouraging in terms of applying the findings of existing research, which utilized the HCR-20 as a checklist, to clinical practice…In addition, the present findings support and strengthen those of earlier research findings in the area, in that internally manipulated scenarios are rated higher in terms of risk than externally manipulated ones. Thus, attribution effects are not confined to only unaided judgments of violence risk assessment, but apply also to those made using SPJ tools” (p. 16).

Other Interesting Tidbits for Researchers and Clinicians

“These findings, however, should be considered with caution as it is clear that the effect size, and therefore practical significance/magnitude, of the findings are small. The statistical significance should therefore be considered with some tentative appraisal when discussing the magnitude of this finding, and the overall strength of meaning behind it. Indeed, future research should investigate various levels of clinical expertise and use various vignettes in order to assess the robustness of these findings; ideally with larger sample sizes. However, the current findings could be viewed as a first step towards systematically investigating the effect of known decision-making biases and errors on the completion of the HCR-20” (p. 15).

“Concerning the scoring of the HCR-20 when used as a check-list compared to the worksheet, no significant differences were found. This would appear to be a positive indication for the applicability of existing research (which largely utilizes the HCR-20 as a checklist) to practice, where it is the fuller worksheet adjunct to the HCR-20 manual that should be used. It can therefore be proposed that existing research investigating the HCR-20 is indeed applicable and generalizable to practice” (p. 16).

Join the Discussion

As always, please join the discussion below if you have thoughts or comments to add!