Friday, June 23, 2017

Author Q & A with Sebastian Brouillete-Alarie discussing "Three Central Dimensions of Sexual Recidivism Risk: Understanding the Latent Constructs of Static-99R and Static-2002R"

Brouillette-Alarie, S., Proulx, J., & Hanson, K. (2017). Three Central Dimensions of Sexual Recidivism Risk: Understanding the Latent Constructs of Static-99R and Static-2002R. Sexual Abuse: A Journal of Research & Treatment. Online First.

The most commonly used risk assessment tools for predicting sexual violence focus almost exclusively on static, historical factors. Consequently, they are assumed to be unable to directly inform the selection of treatment targets, or evaluate change. However, researchers using latent variable models have identified three dimensions in static actuarial scales for sexual offenders: Sexual Criminality, General Criminality, and a third dimension centered on young age and aggression to strangers. In the current study, we examined the convergent and predictive validity of these dimensions, using psychological features of the offender (e.g., antisocial traits, hypersexuality) and recidivism outcomes. Results indicated that (a) Sexual Criminality was related to dysregulation of sexuality toward atypical objects, without intent to harm; (b) General Criminality was related to antisocial traits; and (c) Youthful Stranger Aggression was related to a clear intent to harm the victim. All three dimensions predicted sexual recidivism, although only General Criminality and Youthful Stranger Aggression predicted nonsexual recidivism. These results indicate that risk tools for sexual violence are multidimensional, and support a shift from an exclusive focus on total scores to consideration of subscales measuring psychologically meaningful constructs.

Could you talk us through where the idea for the research came from?

In French/European countries, professionals tend to be lukewarm towards structured risk assessment. As a native French speaker, I often became involved in debates about the pros and cons of actuarial assessment. By participating in these debates, I became cognizant of the conceptual limitations of this approach. Although many criticisms were warranted (e.g., limited predictive accuracy), one always “struck a nerve” with me: that risk factors are clinically meaningless statistical entities that do not enable a true comprehension of the offender.

Since the dawn of psychology, observable behaviors have been used to infer personality traits (or dynamics of the unconscious mind). In this context, why would risk factors, i.e., measures of criminogenic behaviors, be any different? Although risk factors are first and foremost statistical correlates of recidivism, they are also windows into the psychological and sociological mechanisms that lead individuals to commit crimes. This latent trait approach has been described in the works of Beech and Ward (2004) and Mann, Hanson, and Thornton (2010). Their theoretical frameworks for sexual offender risk assessment illustrated how to integrate static, stable, and acute risk factors in etiological models of risk that have far more clinical resonance than “dry” risk scales. Luckily for me, nobody had (yet) thought to empirically test these models. Thus, it became the overarching goal of my doctoral thesis.

At the start of my Ph.D., I had the luck of being put in touch with R. Karl Hanson and his research team (Kelly M. Babchishin, Maaike Helmus) by my director, Jean Proulx. It turned out that Karl had a project quite similar to mine; he wanted to explore the latent psychological constructs underlying the items in sexual offender risk scales. The goal was to shift practice from the assessment of unidimensional and “atheoretical” risk scores to the assessment of multiple risk-relevant psychological propensities. These constructs could then be combined in specific ways depending on the outcome of interest. In a way, we were exploring the building blocks of risk rather than its finite structure.

This lead to our factor analysis of the Static-99R and Static-2002R items (Brouillette-Alarie, Babchishin, Hanson, & Helmus, 2016). We quickly realized that the literature on this topic was substantive; we found 13+ studies on the factor structure of the Static-99/2002/R. Most studies obtained a solution of 3 factors (ours included): sexual criminality, general criminality, and a third factor related to age and victim characteristics. Unfortunately, none of the studies had conducted any convergent validity analyses. They interpreted the factors by looking at the items constituting each construct. Although this was a good start, we thought that more empirically grounded interpretations were necessary. This led us to the current paper.

What kinds of challenges did you face throughout the process?

Doing convergent validity analyses was, in fact, the easy part. The hard part was coming up with the factor structure in the first place (in Brouillette-Alarie et al., 2016). Jean Proulx and I started the factor analysis project in the spring of 2011, as part of my master’s thesis. Then, we involved Karl’s team, who gave us access to worldwide validation studies of the Static-99. They also (rightfully) told us that our factor analytic procedures were outdated and that we needed to redo everything from scratch (a common occurrence according to Maaike!). We dutifully did so, which led to our 2016 paper.

What do you believe to be the main things that you have learnt about the nature of the risk dimensions of the Static-99R and Static-2002R?

First, we learnt that sexual deviance is not a cohesive whole. Variables concerning sexual criminality clustered in two negatively correlated factors: Persistence/Paraphilia and Youthful Stranger Aggression. These factors were associated with different ends of the agonistic continuum (Knight, Sims-Knight, & Guay, 2013). Persistence/Paraphilia was characterized by modus operandi devoid of physical coercion and intent to harm, while Youthful Stranger Aggression was associated with sexual sadism and hostility. Furthermore, these two dimensions did not predict the same types of recidivism: the former was exclusively related to sexual recidivism, while the latter was predictive of all types of recidivism (like General Criminality). Without surprise, Persistence/Paraphilia was more common in sexual aggressors of children, and Youthful Stranger Aggression was more common in sexual aggressors of women. In sum, our results encourage researchers and evaluators to clearly differentiate between pedophilic and sadistic tendencies, as they refer to substantially different constructs. More often than not, they will not characterize the same offenders. In some rare cases (e.g., sadistic pedophiles), they will nevertheless converge into a very high level of sexual recidivism risk.

Second, we found a strong General Criminality factor that naturally converged with antisocial/psychopathic traits and domains of the Level of Service/Case Management Inventory (LS/CMI; Andrews, Bonta, & Wormith, 2004). This confirms that sexual recidivism risk comprises a general deviance dimension that is common to sexual and nonsexual offenders. The generality of criminal behavior in sexual offenders has already been highlighted by numerous authors (e.g., Lussier, LeBlanc, & Proulx, 2005).

Now that you’ve published the article, what are some implications for practitioners?

Although it is not yet ready to be implemented in forensic practice, we hope that sexual offender risk scales (and those scoring them) will adopt dimensional scores in addition to total scores. Sexual recidivism risk is unanimously considered to be multidimensional, and our current risk tools do not convincingly reflect that. It is more clinically relevant to conceptualize risk as the interaction between psychological constructs and the social environment than the sum of discrete correlates. Our research program tries to bridge the gap between those two perspectives.


  1. I know most subscribers to the list do not seem to appreciate the MSI-II because (most believe)that it implies the explicit pathway model inherently, which is not true, but any way the MSI-II has had these two factors built into the MSI-II for about 25 years (MSI-II versus MSI) in the form of Pedophilic Tendencies (Molester Comparison scale prime ' and the not prime) and the rapist comparison scale. If the alleged offender endorses a number of items on either of these scales they are considered to have attitudes, behaviors and cognitions consistent with what ever factor they endorsed. No one to my knowledge has assessed the predictive characteristic of these scales but there is comparisons as to admittors versus non-admittors. Food for thought related to additional research. Steve Gray Psychologist AZ

  2. I liked Steve Gray's response to the blog... thanks Steve, and concur. I, too, like the MSI-II and find it of practical use. I enjoyed the blog itself, and appreciated Sebastian's deeper conceptualization of the structure and use of actuarial assessment. Nicely done.