July 18, 2013

Most civilly detained sex offenders would not reoffend, study finds

Other new research finds further flaws with actuarial methods in forensic practice

At least three out of every four men being indefinitely detained as Sexually Violent Predators in Minnesota would never commit another sex crime if they were released.

That’s the conclusion of a new study by the chief researcher for the Department of Corrections in Minnesota, the state with the highest per capita rate of preventive detention in the United States.

Using special statistical procedures and a new actuarial instrument called the MnSOST-3 that is better calibrated to current recidivism rates, Grant Duwe estimated that the recidivism rate for civilly committed sex offenders -- if released -- would be between about 5 and 16 percent over four years, and about 18 percent over their lifetimes. Only two of the 600 men detained since Minnesota's law was enacted have been released, making hollow the law's promise of rehabilitation after treatment.

Duwe -- a criminologist and author of a book on the history of mass murder in the United States -- downplays the troubling Constitutional implications of this finding, focusing instead on the SVP law’s exorbitant costs and weak public safety benefits. He notes that "Three Strikes" laws, enacted in some U.S. states during the same time period as SVP laws based on a similar theory of selective incapacitation of the worst of the worst, have also not had a significant impact on crime rates.

The problem for the field of forensic psychology is that forensic risk assessment procedures have astronomical rates of false positives, or over-predictions of danger, and it is difficult to determine which small proportion of those predicted to reoffend would actually do so.

Minnesota has taken the lead in civilly detaining men with sex crime convictions, despite the state's only middling crime rates. Unlike in most U.S. states with SVP laws, sex offenders referred for possible detention are not entitled to a jury trial and, once detained, do not have a right to periodic reviews. Detention also varies greatly by county, so geographic locale can make the difference between a lifetime behind bars and a chance to move on with life after prison.

Ironically, as noted by other researchers, by the time an offender has done enough bad deeds to be flagged for civil commitment, his offending trajectory is often on the decline. Like other criminals, sex offenders tend to age out of criminality by their 40s, making endless incarceration both pointless and wasteful.

The study, To what extent does civil commitment reduce sexual recidivism? Estimating the selective incapacitation effects in Minnesota, is forthcoming from the Journal of Criminal Justice. Contact the author (HERE) to request a copy. 

Other hot-off-the-press articles of related interest:

Risk Assessment in the Law: Legal Admissibility, Scientific Validity, and Some Disparities between Research and Practice 


Daniel A. Krauss and Nicholas Scurich, Behavioral Sciences and the Law

ABSTRACT: Risk assessment expert testimony remains an area of considerable concern within the U.S. legal system. Historically, controversy has surrounded the constitutionality of such testimony, while more recently, following the adoption of new evidentiary standards that focus on scientific validity, the admissibility of expert testimony has received greater scrutiny. Based on examples from recent appellate court cases involving sexual violent predator (SVP) hearings, we highlight difficulties that courts continue to face in evaluating this complex expert testimony. In each instance, we point to specific problems in courts’ reasoning that lead it to admit expert testimony of questionable scientific validity.We conclude by offering suggestions for how courts might more effectively evaluate the scientific validity of risk expert testimony and how mental health professionals might better communicate their expertise to the courts.
Contact Dr. Krauss (HERE) for a copy of this very interesting and relevant article. The following two articles are freely available online:

The utility of assessing "external risk factors" when selecting Static-99R reference groups


Brian Abbott, Open Access Journal of Forensic Psychology

ABSTRACT: The Static-99 has been one of the most widely used sexual recidivism actuarial instruments. It has been nearly four years since the revised instrument, the Static-99R, has been released for use. Peer-reviewed literature has been published regarding the basis for changing the scoring system for the age-at-release item, the utility of relative risk data, and variability of sexual recidivism rate s across samples. Thus far, the peer-reviewed literature about the Static-99R has not adequately addressed the reliability and validity of the system to select among four possible actuarial samples (reference groups) from which to obtain score-wise observed and predicted sexual recidivism rates to apply to the individual being assessed. Rather, users have been relying upon the Static-99R developers to obtain this information through a website and workshops. This article provides a critical analysis of the reliability and validity of using the level of density of risk factors external to the Static-99R to select a single reference group among three options and discusses its implications in clinical and forensic practice. The use of alternate methods to select Static-99R reference groups is explored.

Calibration performance indicators for the Static-99R: 2013 update


Greg DeClue and Terence Campbell, Open Access Journal of Forensic Psychology

ABSTRACT: Providing comprehensive statistical descriptions of tool performance can help give researchers, clinicians, and policymakers a clearer picture of whether structured assessment instruments may be useful in practice. We report positive predictive value (PPV), negative predictive value (NPV), number needed to detain (NND), and number safely discharged (NSD), along with associated confidence intervals (CIs) for each value of the Static-99R, for one data set. Values reported herein apply to detected sexual recidivism during a 5-year fixed follow-up for the samples that the Static-99R developers consider to be roughly representative of all adjudicated sex offenders.

BLOGGER NOTE: I'm posting this research update while stranded at LAX en route to Brisbane, Australia, where I will be giving a series of seminars and trainings at Bond University before flying to Honolulu to give a full-day continuing education training at the American Psychological Association convention. (Registration for that is still open, I am told.) I'll try to blog as time allows, and I hope to see some of you at these venues.

No comments: