October 30, 2011

Study: Psychopathy score fails to predict sexual recidivism

Many forensic psychologists believe that psychopathy is a risk factor for sex offender recidivism. Not surprisingly, when forensic psychologists assign a sex offender a high score on a psychopathy test, it increases the risk of extreme legal sanctions such as civil commitment.

But a new study out of Texas found zero correlation between sexual recidivism and psychopathy, as measured by the widely used Psychopathy Checklist (PCL-R). If anything, sex offenders who were arrested for a new sexually violent offense tended to have lower scores on the PCL-R than those who were not rearrested!

Regular blog readers should be familiar with these researchers by now: Dan Murrie, Marcus Boccaccini and crew are the same scholars who informed us of the partisan allegiance phenomenon, in which evaluators working for the government tend to assign far higher PCL-R scores than do those working for the defense.

In their most recent study, they examined PCL-R scores from about 400 sex offenders in Texas who were released from prison and then tracked for anywhere from about two to seven years. They examined not just the total scores on the PCL-R, but also scores on the instrument's two factors, as well as four so-called facet scores. Not one of these seven PCL-R variables was a statistically significant predictor of whether a man would be arrested for a new sex crime.

“Overall, these predictive validity findings were striking because the PCL-R apparently failed to predict the type of outcome (i.e., sexually violent reoffense) for which it was administered in this context,” the authors noted.

Further, in cases in which the PCL-R was administered by more than one evaluator, the study found poor agreement between the two, even though both were working for the government. Consistent with prior research, interrater agreement was higher on Factor 2, which measures antisocial behavior and an impulsive lifestyle, than on Factor 1, which measures the vaguely operationalized personality and emotional dynamics thought to underlie psychopathy.

In an interesting twist, the researchers tried to determine whether some evaluators were more accurate than others at predicting recidivism through PCL-R scores. They identified four highly prolific evaluators; together, these three psychologists and one medical doctor had provided almost two-thirds of the PCL-R scores in the study. Although the PCL-R scores of three of these four evaluators were more likely than other evaluators' scores to correlate with a new arrest for a non-sexual crime, even these evaluators could not produce PCL-R scores that predicted sexual offense recidivism.

Despite the PCL-R’s lack of predictive validity, sex offenders with higher PCL-R scores were more likely than others to be recommended for civil commitment, indicating that the unreliable rating was far from harmless in forensic practice.

The study is: 

Murrie, D. C., Boccaccini, M. T., Caperton, J. and Rufino, K. Field Validity of the Psychopathy Checklist–Revised in Sex Offender Risk Assessment. Psychological Assessment. Click HERE to request a copy from the first author, at the Institute of Law, Psychiatry, and Public Policy.

Of related interest:


4nSick said...

Karen, in your opinion, what are the "best" (most valid) sexual risk assessment instruments out there? I know even the best of them cannot predict sexual recidivism with any amount of certainty. In your opinion, is there a better way to predict recidivism? I sure would like to come up with something better myself. I don't know if it's possible.


Luke said...

Karen, I am wondering why it is important to differentiate between a violent sex crime and violent crimes in general. If the PCL-R can help predict future violence, shouldn't that be taken into consideration in forensic decision-making, even if it isn't sexual violence? Maybe an inmate isn't going to rape anyone else, but if they'll still assault people in say, a robbery, they're not safe in the general public, right? I don't think parole board decisions are limited to considering only the speicific type of risk involved in the conviction offense.


Karen Franklin, Ph.D. said...

4nSick: Good question, but it's a bit too involved for me to get into right now. Perhaps I'll write a blog post on it. The actuarials are certainly limited, but the MATS-1 -- which I've blogged about previously -- shows some promise. Bottom line: As I'm sure you know, predicting the future is always going to be hard.

Luke: You're right about parole board decisions. But the civil commitment laws in the United States are very specifically limited to consideration of future risk of sexual violence. Non-sexual violence risk is not to be considered. It's kind of a crazy set of laws, because -- as you say -- there's a great deal of overlap between different types of criminality.