Sunday, December 11, 2011

A Jerry Sandusky flight of fancy...

Caveat: In reading this post, I would ask that you allow me a “flight of fancy”, as it were. Under normal conditions, risk assessments are made of persons before the court and are conducted using validated tools and procedures in a standard fashion. Attention to detail and precision are important goals in risk assessment. The following pays attention to those prescriptions in spirit; however, I’m not suggesting that my process below would be acceptable if we were actually attempting to assess risk posed for true judicial or case management purposes.

So, here goes…

Best as I can figure, the score Jerry Sandusky would get on the Static-99R is -1. Based on what has been reported in the media (admittedly, this is not a formal risk assessment, but you’ll see where I’m going with this), that’s what you get. Negative one.

According to the reports: Mr. Sandusky has been married for a long time and has no other criminal history; the alleged victims are all persons he appears to have known for a while; and there is apparently no violence other than that directly associated with sexual offending. So, no points for “ever lived with a lover…”, “index non-sexual violence”, “prior non-sexual violence”, “prior sexual offenses”, “non-contact sexual offenses”, “prior sentencing dates”, or “any stranger victims”.

That leaves one point for “any unrelated victims” and one point for “any male victims”. Then, of course, you have to consider his age. Sandusky is in his sixties, which means that his score on the age item would be -3. Total it up and you get -1. If you follow the nominal risk category labels suggested by the Static-99R’s developers (Karl Hanson and David Thornton), this score would put Sandusky in the low risk group according to static (life history, or unchangeable) factors.

Then, there’s the issue of the appropriate normative sample against which to compare him. Since 2008, Hanson, Thornton, and others associated with Static-99R have suggested that it would be inappropriate to simply use a one-size-fits-all normative group against which to compare individual offenders. Options for comparative samples include “routine”, “non-routine”, “preselected for treatment need”, and “preselected for risk and need”. The intent is to ensure that we really do compare apples with apples and oranges with oranges. Meta-analytic data suggest that using a normative group that includes all offenders amounts to a veritable “fruit salad” in which the heterogeneity of the sample leads to possible over- or under-estimation of risk for some individual offenders.

In a short paper originally published in the CCOSO Newsletter (republished in The Forum of the Association for the Treatment of Sexual Abusers ATSA), Thornton, Hanson, and Helmus suggested that the we might be able to use scores on measures of dynamic risk to help us choose an appropriate normative group for better interpreting an offender’s risk as scored on Static-99R. Stable dynamic risk factors are those that assist us in measuring risk according to factors related to personality or other longstanding aspects of someone’s make-up—but which are theoretically amenable to change with treatment or other interventions. Using the Stable-2007, Thornton et al. suggested that, among persons with the same Static-99R score, higher scores on the Stable-2007 would indicate higher levels of criminogenic need and an attendant increase in overall risk to reoffend.

Lacking any clinical interview with Mr. Sandusky, it is difficult to speculate what his exact score might be on any risk measure but, again, if we simply infer from information presented in the media, we can get a rough idea of what stable dynamic risk factors might be pertinent.

Of course, we have to emphasize that the reported offenses are all still allegations. For this exercise, one must assume that what’s been reported in the media is correct. So, for sake of argument, it appears reasonable to assume at least some degree of marital strife. It also seems reasonable to strongly suspect deviant sexual preferences and to presume the presence of emotional identification with children. Last, it seems likely to me that sexual pre-occupation and some degree of poor cognitive problem-solving and impulsivity might also be reasonably assumed.

Putting all this together, Mr. Sandusky would very plausibly be assessed as being of at least moderate risk to reoffend based on Stable-2007. Thornton et al. would suggest that this is indicative of a need for participation in a formal treatment program aimed at addressing risk for sexual reoffending and other aspects of lifestyle instability and dysfunction. This is to be contrasted with psychoeducational or short-term counseling programs usually offered to lower risk offenders (in keeping with the tenets of the Risk-Need-Responsivity Model).

At this point, it is likely worthwhile to briefly revisit the punishment vs. treatment issue. I’ve expounded on this much more precisely in previous posts, but it bears reiterating that there is little if any research showing that punishment alone reduces recidivism. The meta-analytic data suggests that this relation actually may work in reverse to what many might assume, in that it appears that punishment without programming leads to increased reoffending. Whatever we might think of individual studies supporting (or not) the benefits of providing treatment and other human service interventions, some really large meta-analytic studies show a strong effect of lower risk as a consequence of cognitive-behavioral programs, including treatment for sexual offenders. Bottom line, however much we might want to punish the Sanduskys of the world, punishment will not get us nearly as much bang for our risk reduction buck as will treatment. It’s fine to want to sanction someone for their actions, but we shouldn’t kid ourselves that that will eliminate risk to reoffend.

Back to our flight of fancy risk assessment exercise…

Referring to the Static-99R Evaluators’ Workbook, persons in the “Preselected for Treatment Need” normative sample have a roughly 3.1% chance of sexually reoffending within five years, or a 5.4% chance of doing so within 10 years of release. Pretty low, eh? I bet that the majority of the public would be shocked to learn that Sandusky’s putative risk for reoffense is so low; especially, given the degree of media attention and public outcry this case has garnered to date.

However, this is an excellent example of how the process of risk assessment doesn’t always line up with public perception. We sometimes let our emotional impulses determine how we think about the risk posed by the offender. If he actually committed the offenses, it is very reasonable to assume that Mr. Sandusky did an awful lot of harm to a large number of boys and young men. Such offenses naturally evoke some degree of anger, disgust, and denunciation. However, none of those responses necessarily makes for good risk assessment or, ultimately, good risk management.

Should he be found guilty of the offenses, Mr. Sandusky will be subject to punishment for his actions. It is also possible that he will attend a treatment program for persons who have sexually offended. Depending on the length of the sentence, his health, and the release practices of the state in which he is incarcerated, it is also possible that he may someday be released back to the community. That brings us back to his Static-99R score and the consideration of Stable-2007 variables.

Should he be released at some point in the future, it would be scientifically defensible to consider the risk Sandusky poses for reoffense to be low. One of the salient risk factors to be considered here is age-related desistance. Of course, Sandusky is alleged to have been offending as recently as only a few years ago, when he was already of advanced age. Some might be inclined to say that this negates age-related desistance as a modifier of risk potential—I disagree. The data I’ve seen so far on the age-related desistance issue do not suggest that age is a dichotomous factor. Rather, it is incremental or continuous. What that indicates is that all persons are likely to “slow down” in most areas of life as they age, but not necessarily in the same exact fashion. All in all, advancing age WOULD BE a risk-reducing factor for Sandusky, just like it is for every sexual offender—it might just not be as much of a risk-reducing factor for him as it might be for his peers.

In closing, the intent of this blog post is not to garner sympathy for offenders nor somehow to excuse their actions. Rather, my intent was to highlight the apparent disconnect between our emotions about offending and our knowledge about scientifically defensible risk assessment processes. Concluding that Jerry Sandusky would be in the low risk range for sexual reoffending is not the same thing as concluding that his actions are acceptable or mitigating the harm done to the putative victims. Neither does that rating suggest that Mr. Sandusky should be exempted from criminal prosecution, criminal sanction, or eventual evidence-based risk management processes should he ever be released.

Hopefully, as this case goes forward, Sandusky will be properly assessed by a trained professional and a scientifically defensible plan of action will be developed that assists him in addressing whatever risk factors are actually present. All of this would be intended to reduce the level of risk he might pose to others in the community. Ultimately, that’s the dilemma of people who sexually abuse. We can reduce risk among the most reviled of our citizens, but only through compassionate rehabilitation and sensible community supervision.



  1. I just read your blog for the first time today, and I appreciated your willingness to pursue your "flight of fancy". For the sake of on-going discussion I think it served it's purpose (at least for me it did). Your comments related to punishment vs. treatment, and thoughtful response vs. emotional response were also appreciated and again (at least for me) right on the money. The place that I experienced discomfort while reading your blog was the place I always experience discomfort when discussing Risk Assessment tools, and I am also left feeling caught between a professional rock and a hard place. I respect scientific process and data-driven outcomes, hence my use of Risk Assessment models. On the other hand, there is also another "tool" that is always operating for me at the same time....I tend to think of it as "common sense informed by knowledge and experience". When I engage this particualr tool, which by the way is not emotion-driven at all, I come to a very different conclusion based on the same media derived asumptions that you have made. In short, it is hard for me to imagine that given the time, relationship, access and opportunity, that Mr. Sandusky would not abuse again. Utilizing this tool and the assumptions made about his previous behavior with young boys, and given these circumstances, Mr. Sandusky would appear to me to actually be a very high risk to reoffend. I imagine that many experienced and non-emotionally-reactive professionals in the field would come to the same conclusion and would not trust the Risk Assessment as presented by you. I understand how you got there, I understand the tool, but as in other circumstances I've faced throughout my career, it doesn't quite fit with the reality that I know. As a thoughtful professional this presents quite a conundrum, but there it is. Thanks again for your thoughtful piece.....

  2. "Bottom line, however much we might want to punish the Sanduskys of the world, punishment will not get us nearly as much bang for our risk reduction buck as will treatment. It’s fine to want to sanction someone for their actions, but we shouldn’t kid ourselves that that will eliminate risk to reoffend."

    How did you draw that conclusion? Incarcerated sex offenders do not molest children; therefore, the risk of reoffending is 0. Take away the opportunity and their desire is a moot point. Also, you are correct that the public would be shocked to find out that Mr. Sandusky's risk for reoffending would be considered low or moderate based on a specified assessment - which is why logically there's little faith in anything that treatment or assessments have to offer towards the goal of public safety.

  3. The best judge of future behavior is past behavior. Although Sandusky has no known prior convictions he does have MANY alleged victims. Assuming he is convicted in some of the cases it's simply idiotic to assume that he will turn over a new leaf and quit raping boys - as if a guilty verdict somehow lowers his desire to sexually abuse male children. The mere thought of the possibility of a judge weighing a risk assessment heavily that "proved" that this type of defendant is a low risk to the community is over the top, don't you think? Evidenced based practices are only as accurate as the scientific measurements used.