||Assessing Sex Offenders: Problems and Pitfalls
||Terence W. Campbell
||Charles C Thomas, ©2005
Charles C Thomas, Publisher, LTD
2600 South First Street
Springfield, Illinois, 2004
Price: $61.95 hard
Following a number of high profile sexually violent crimes in the
1990s, several states passed sexually violent predator (SVP) statutes,
targeting high-risk sex offenders who had served their sentences and
were about to be released from prison. Currently there are sexual
predator statutes in 16 states.
Although there are variations in the different states' SVP statutes, in
general, to commit an individual, the state must prove that (a) the person
has been convicted of a relevant sexual offense in the past; (b) the
person currently suffers from a "mental disorder" or "mental abnormality"
predisposing him to engage in acts of sexual violence; and (c) the person
is likely to commit a sex offense in the future. The civil commitment
follows a jury trial.
After the state determines that a sex offender should be considered for
possible civil commitment, he is evaluated by a state psychologist. If
probable cause is found, he will be evaluated by another state evaluator
and by a defense evaluator. These professionals are generally
psychologists, although some psychiatrists also do evaluations. The
evaluation consists of a lengthy and detailed file review, an interview,
sometimes psychological testing, often a rating on the Hare Psychopathy
Checklist (PCL-R), and ratings on one or more actuarial assessments. The
amount of face-to-face contact ranges from several hours to none at all
when the sex offender refuses to speak to the evaluator.
Since actuarial methods are believed to be the most accurate way of
assessing long-term recidivism risk, a variety of actuarial instruments
have been developed over the past several years for this purpose. Evaluators generally use one or more of them in combination with their
clinical judgment. The problem is that these procedures are far from
perfect and evaluators seldom include adequate qualifications as to their
This is the focus of Dr. Campbell's book. Dr. Campbell agrees the
research is clear that actuarial assessment is superior to clinical
judgment. But he argues quite persuasively that none of the actuarial
methods for assessing sexual offense recidivism have advanced enough to
improve over making predictions based entirely on the base rate. He notes
that unless a prediction scheme has exceptional accuracy, attempting to
predict events in a population with a low antecedent probability will
inevitably lead to an unacceptable level of false positives. Unless the
base rate for recidivism for sex offenders exceeds 30 percent, barring
unusual circumstances, the most accurate prediction will always be that a
person will not engage in sexual violence. But none of the actuarial
instrument developed to date have more than modest accuracy.
Dr. Campbell notes the increasing number of SVP
commitments and the high costs of these programs and asks two "unavoidable
questions" concerning these statutes: "(1) Do these statutes rely on
assessments of acceptable accuracy to identify sex offender recidivism
risk? (2) Are the treatment programs available to committed SVPs
sufficiently effective?" The remainder of the book addresses these two
questions. Recidivism risk can be assessed in four basic ways: clinical
judgment, guided clinical judgment, actuarial assessments, or adjusted
actuarial assessments. Dr. Campbell examines all of these in detail by
asking a series of questions proposed by Thomas Grisso1.
- Do SVP assessment procedures now satisfy the heightened scientific
standards to which they should be subjected?
- Are SVP assessment procedures now sufficiently standardized?
- Are there manuals now available for SVP assessment procedures that
satisfy applicable standards?
- Are the levels of inter-rater reliability for SVP assessment
procedures now at an acceptable level?
- Are SVP assessment procedures being rushed prematurely into legal
- In legal proceedings, can SVP assessment procedures survive full
disclosure of their limitations?
- Do SVP assessment procedures create an appearance of precision that
substantially exceeds their actual accuracy?
Dr. Campbell observes that there are unresolved problems undermining
the actuarial methods used in sex offender risk assessment. All of the
actuarial schemes stress risk factors and ignore protective factors. Most
of the research, and hence the items, on the actuarial instruments are
static factors, that is, factors that will never change, although dynamic
factors are extremely important. None of the actuarial instruments
adequately consider age-for the RRASOR and the Static-99 the cut- off age
is 25; for the MnSOST -R it is 30. But the base rate for recidivism is
much lower for older sex offenders.
There are other problems with the commonly-used actuarial techniques,
which Dr. Campbell addresses in detail, including the disagreements
between evaluators about whether to use both the Static-99 and the RRASOR
or just the Static-99, the changes in the Static-99 coding rules, and
field reliability problems. He notes the lack of peer review in the MnSOST-R, problems that appeared when it was independently analyzed, and
concludes it should not be used.
He explains the receiver operating characteristic (ROC) statistic that
is used when determining the classification accuracy of an actuarial
procedure. The ROC methods yield what are called area under the curve (AUC)
values, that correspond to the probability of a recidivist's score
exceeding a non-recidivist's score. But the AUC value does not identify
the different levels of sensitivity and specificity at various cutoff
scores and for samples with recidivism base rates less than 30 percent,
actuarial instruments are no more accurate than relying on the base rate
alone. Using them, however, can create a misleading appearance of
Dr. Campbell next addresses what he terms "supplementary" assessment
procedures, including the Psychopathy Checklist-Revised (PCL-R), Violence
Risk Appraisal Guide (VRAG), Sex Offender Risk Appraisal Guide (SORAG) the
Minnesota Multiphasic Personality Inventory (MMPI-2) the (Millon Clinical
Multiaxial Inventory (MCMI-llI), and the Sexual Violence Risk-20 (SVR-20). Dr. Campbell concludes that there are no supplementary assessment
procedures that elevate the basic actuarial instruments above the
threshold of general acceptance.
Most evaluators, however, do not rely exclusively upon actuarial
assessments; instead they typically adjust their assessment upward or
downward using factors such as completing a treatment program or comments
made by treatment staff. But according to Dr. Campbell this procedure,
called "adjusted actuarial assessment" (AAA), amounts to a retreat into
clinical judgment and unknown errors of judgment. It does not represent an
Dr. Campbell next discusses the requirement that the sex offender meet
the criteria for a diagnosable mental disorder, or mental abnormality. He
notes the shortcomings with the DSM-IV and observes that its reliance on
clinical judgment creates significant problems for SVP evaluators who fail
to interview offenders but still give diagnoses. Completing an actuarial
assessment without an interview may be defensible, but making a DSM-IV
diagnosis without one is not. Dr. Campbell points out that the DSM-IV
contains no inter-rater reliability data and he discusses problems with
the personality disorder and paraphilia diagnoses commonly found in SVP
evaluation reports. He concludes that, like the actuarial techniques, the
DSM- IV, when subjected to Grisso's seven questions, falls far short.
Finally Dr. Campbell addresses issues in the the treatment of sex
offenders, including informed consent, treatment planning, the relapse
prevention model, and other problems with SVP therapy. Although brief,
this chapter provides an excellent overview of the controversies and
problems in the treatment provided to the civilly committed sex offenders. He notes that on average the recidivism risk of offenders undergoing
treatment is reduced by only approximately 7½ percent. This assumes the
best, most current, state-of -the art treatment.
The final chapter, "ethical obligations," applies the American
Psychological Association's ethical standards and testing standards to SVP
evaluations. Dr. Campbell believes that these standards mean that
psychologists are ethically required to acknowledge the limitations of the
actuarial instruments and other techniques when they write reports and
testify concerning their SVP evaluations.
Dr. Campbell's book should be studied by anyone who does risk
assessments, not only for SVP evaluations but for sentencing evaluations
and conditional discharge considerations. Whether or not evaluators elect
to use actuarial methods, clinical assessments, or some combination, it is
incumbent upon them to be familiar with the arguments Dr. Campbell
presents in this book. Psychologists must be able to explain clearly and
knowledgeably just what methods they used, why they used them, the
research supporting these methods, and the limitations of these methods.
1 Thomas Grisso: Invited address: Ethical
issues in evaluations for sex offenders re-offending. Presented at "Sex
offender re-offense risk prediction." Madison, WI, Sinclair Seminars,
Reviewed by Hollida Wakefield, Institute for Psychological Therapies.