Sexual offenders do great societal damage and cause high levels of justifiable public concern. Over the past 10 years psychology has developed the ability to reliably sort male sexual offenders into low, moderate, and high risk for sexual recidivism (MnSOST-R, Risk Matrix-2000, RRASOR, STATIC-99) based upon historical, static, non-changeable risk factors. However, the “static” structure of these tests effectively precludes their ability to measure changes in risk. In this brief review, we outline a timeline and provide supporting data regarding the development of specialized tools—the STABLE-2007 and the ACUTE-2007—designed to assess and track changes in risk status over time by assessing changeable “dynamic” risk factors.
“Stable” dynamic risk factors are personal skill deficits, predilections, and learned behaviours that correlate with sexual recidivism but that can be changed through a process of “effortful intervention”. It is hypothesized that should “effortful intervention” (read: treatment or supervision) take place in such a way as to reduce these risk-relevant factors there would be a concomitant reduction in the likelihood of sexual recidivism. “Acute” dynamic risk factors are highly transient conditions that only last hours or days. These factors are rapidly changing environmental and intrapersonal stresses, conditions, or events that have been shown by previous research to be related to imminent sexual reoffence. Measures of dynamic risk potential, like STABLE-2007 and ACUTE-2007, should be used to inform correctional managers as to how much risk they are managing, to inform decisions on levels of community treatment and supervision, and estimate changes in risk status (pre-post testing) after treatment or other interventions.
In the late 1990s, we began to investigate the relationship between sexual recidivism and dynamic, changeable, risk factors that correlated with sexual recidivism. This work produced the SONAR assessment, which demonstrated adequate internal consistency and a moderate ability to differentiate sexual recidivists from non-recidivists. Extending this work we broke the SONAR into two parts creating a stable measure of dynamic risk, the STABLE-2000 (16 items) and an acute measure of dynamic risk, the ACUTE-2000 (8 items).
The Dynamic Supervision Project
To test these new instruments we instituted a prospective study, the Dynamic Supervision Project, involving every Canadian Province and Territory and the States of Alaska and Iowa. Over a five-year period, 156 parole and probation officers completed Static, Stable, and Acute risk assessments on 997 sexual offenders across 16 jurisdictions. All of the probation and parole officers scoring risk of reoffence for these community-based sexual offenders were trained in sexual offender risk assessment by attending a two-day training that focussed on scoring actual case examples. Inter-rater reliability was high, with intraclass correlations (ICCs) of .89 found for STABLE-2000 total scores (k = 87) and a median ICC of .90 observed for the ACUTE-2000 (k = 75). Sexual, Violent and “other” recidivism information was gathered from official criminal histories after a median of 41 months of follow-up.
Results showed that both the STABLE-2000 and the ACUTE-2000 added predictive validity above that demonstrated by the STATIC-99 alone. The sexual recidivism rate for this widely disparate group of community-based sexual offenders was 7.6% after three years (n = 790). The results also suggested that the scoring procedures could be improved. Based upon empirical results, changes in scoring of the STABLE-2000 and the ACUTE-2000 were recommended and this resulted in the development of two improved dynamic risk measures, the STABLE-2007 and ACUTE-2007.
The STABLE-2007 assesses 13 stable risk factors that have been shown to correlate with sexual recidivism: significant social influences, capacity for relationship stability, emotional identification with children, hostility toward women, general social rejection, lack of concern for others, impulsivity, poor problem solving skills, negative emotionality, sex drive and pre-occupations, sex as coping, deviant sexual preference, and co-operation with supervision. Each of these 13 items is scored on a three-point scale (0 = no problem evident, 1 = some problem evident, 2 = significant problem evident), for a total of 26 possible points. Emotional identification with children is not scored for those offenders who do not have a child victim and the scale is subsequently scored out of 24 points for that group. The offender’s STATIC-99 score is then combined with his STABLE-2007 score to produce percentage estimates of sexual recidivism, sexual recidivism plus sexual breaches, violent recidivism, any criminal recidivism (breaches excluded), and any criminal recidivism including breaches at one, two, three, and four years.
Changes from the STABLE-2000 to the STABLE-2007
Several changes were made in the evolution of the STABLE-2000 to its current incarnation—the STABLE-2007:
1. In the original STABLE-2000 there were three attitude measures. The STABLE-2000 assessed attitudes of Sexual Entitlement, Attitudes Supportive of Sexual Assault, and Child Molester Attitudes. None of these attitude measures was found to be significantly related to sexual recidivism; however, each was significantly related to both violent and general recidivism. As a result, these three attitude measures were not included in the STABLE-2007.
2. The STABLE-2000 item “Lovers/Intimate Partners” has been strengthened by adding a second subsection making note of whether the offender has ever had an intimate relationship with an adult partner of two years duration. This reinforced item has been re-titled as “Capacity for Relationship Stability” in the STABLE-2007.
3. Emotional Identification with Children (STABLE-2000) showed a significant but non-linear relationship with sexual recidivism in child molesters, but did not show a significant relationship to any type of recidivism in assaulters of adults. This item, while retained in the STABLE-2007 is now only scored for offenders who have at least one victim aged 13 or less at the time of the assault.
4. The STABLE-2000 item “Deviant Sexual Interests” was scored by self-report, specialized testing (phallometric testing), or by evident persistence of one type of victim choice in the criminal history. As scored in this sample, this item was only weakly related to sexual recidivism. The revised item for the STABLE-2007 takes into greater account the number of sexual victims and the number of sexual offences that demonstrate a deviant sexual preference in the offender (e.g., male child victims).
5. The STABLE-2000 was scored by summing the highest score in each of six sub-domains for a total score out of a possible 12 points. In the STABLE-2007, all items are weighted equally and individual item scores are simply summed for a total score out of a possible 26 points.
The ACUTE-2007 assesses seven acute, rapidly-changing risk factors that correlate with sexual recidivism. This scale contains two factors. The first factor predicts sexual and violent reoffending and uses the following four risk factors: victim access, hostility, sexual pre-occupation, and rejection of supervision. The second factor predicts general criminal recidivism using the four factors above plus emotional collapse, collapse of social supports, and substance abuse for a total of seven items. Each of these seven items is scored on a four-point scale (0 = no problem evident, 1 = some problem evident, 2 = significant problem evident, and IN = “Intervene Now”) for a total of 14 possible points. An “Intervene Now” score calls for immediate intervention to prevent imminent reoffence or supervision catastrophes, such as suicide. Once the ACUTE-2007 has been scored, this outcome is combined with the offender’s STATIC-99/STABLE-2007 score to estimate an overall risk priority. The offender is nominally classified as a low, moderate, or high risk for sexual and violent recidivism and as a low, moderate, or high risk for general criminal recidivism. Appropriate, empirically-based risk ratios can then be applied to determine intervention priority.
Changes from the ACUTE-2000 to the ACUTE-2007
In the original ACUTE-2000 there were eight items to score, one of them optional. Officers in the field were able to nominate a “unique” risk factor—unique to that offender. Data collected from rating these proposed “unique” risk factors did not predict and, hence, the “unique” factor question was dropped from the STABLE-2007, for a total item count of seven factors.
The STABLE-2007 and the ACUTE-2007 are easier to score than their predecessors and combinations of the STATIC-99 and STABLE-2007 produced Receiver Operating Characteristic (ROC) values for sexual reoffence commonly in the .76 range. When used by “conscientious” officers, the STATIC-99/STABLE-2007 combined scores in this sample produced ROCs of .84 for sexual reoffence and .80 for violent recidivism. For all types of recidivism, STABLE-2007 and ACUTE-2007 assessments were found to add predictive power above and beyond that available to assessments of static risk alone.
Although it is wonderful to see that “conscientious” officers and the conscientious application of these tests can improve the overall accuracy of prediction, the variability between the “common ROC” of about .76 and the “conscientious officer” ROC of .84 is worrisome. Variability of this nature points to the necessity of training the officers who carry out these assessments. Furthermore, officers must have the opportunity to do sufficient numbers of assessments each year to remain familiar and comfortable with these assessment instruments as a whole. In addition, this points to several systemic issues in risk assessment. Correctional systems using risk assessment measures of any type must assure not only adequate initial training for their staff but that staff demonstrate their ability to score cases against established criteria. Staff need to be given the opportunity and the time to engage in peer-review exercises and inter-rater reliability exercises to ensure reliable application of these instruments. An important factor in the “calibration” of scoring any risk assessment within any jurisdiction is the development and maintenance of a “decision log” for the jurisdiction. Here, when officers gather to discuss differences in scoring, the resulting decisions must be recorded and circulated to all officers scoring the instrument to ensure that officers are all “singing from the same hymnal”.
The findings of the Dynamic Supervision Project provide further evidence that trained community supervision officers can reliably score valid and useful sex offender risk assessments. Results of this nature, even taking into account the need for replication and cross-validation, suggest significant policy and practice implications for the community supervision of sexual offenders. The STABLE-2007 and the ACUTE-2007 have demonstrated predictive validity beyond that of the SONAR and the STABLE-2000/ACUTE-2000 packages. The authors no longer support or recommend the use of the SONAR, the STABLE-2000 or the ACUTE-2000, but recommend the STABLE-2007 and ACUTE-2007 for assessing dynamic changes in risk for sexual offenders.
For more information, readers are directed to:
Hanson, R. K., Harris, A. J. R., Scott, T-L, & Helmus, L. (2007). Assessing the risk of sexual offenders on community supervision: The Dynamic Supervision Project. User Report, Corrections Research, Ottawa: Public Safety Canada. Available at
[Back to Top]