Risk Assessment of Sexual Offenders

1
Risk Assessment of Sexual Offenders

• Mr Steven M Wright
• University of South Australia
• ? 2001

2
Why the current emphasis ?

• Media reporting
• Political pressures
• community notification Megans Law (USA)
  Sarahs Law (UK)
• Guide Intervention who and what to target
• Legal obligations / Ethical concerns
• Sexual predator legislation (United States)
• Duty to warn/protect

3
Risk Assessment Objectives(Hart, 2001)

• We never know an individuals risk for violence
  we merely estimate it assuming various
  conditions.
• Evaluations of individuals to
• (a) Characterise the risk that they will commit
  violence in the future, and
• (b) Develop interventions to manage or reduce
  that risk
• The task is to understand the factors associated
  with how and why individuals chose to offend in
  the past, and to determine whether these or other
  factors might lead the individual to make similar
  choices in the future.

4
Recidivism risk factors (Hanson, 2000)

• Static historical and unchangeable
• age, criminal history, demographic
  characteristics
• Dynamic predictors
• Stable dynamic (sexual preferences, cognitive
  distortions)
• Acute dynamic (intoxication, emotional states)

5
Sexual offense recidivism

• Although there is a rich clinical literature on
  sexual offenders, there has been relatively
  little work on assessing sexual violence risk
  among sexual offenders, particulary with regard
  to sexual-reoffending.
• Hanson Bussiere (1998) meta-analysis of the
  scientific literature (28,972 offenders)
  highlighted the importance of historical or
  static factors in sexual violence recidivism
  risk.
• sexual deviance (phallometric assessment) Age
  (young)
• prior sexual offences Never married
• early onset of sexual offending Personality
  disorders
• victim choices (family membersltacquaintancesltstran
  gers)
• failed to attend/dropped out of treatment
• Concluded that recidivism rate for sexual
  violence low contrary to popular opinion. 13.4
  percent of offenders committed a new sexual
  offense within the 4-5 year follow up period.

6
Among sexual offenders, non sexual recidivism was
best predicted by the same variables that predict
recidivism among nonsexual criminals (Andrews
Bonta, 1994). Often these offenders tended to be
young, single and have antisocial/psychopathic
personality disorders, and have a history of
prior violent and nonviolent offenses. Factors
not related to sexual offense recidivism included
having a history of sexual abuse as a child,
substance abuse and general psychological
problems (anxiety, depression, low self-esteem
etc.)It is suggested that whilst the extent to
which sexual offenders are distressed does not
predict recidivism, such offenders may react
deviantly when distressed.
7
Hanson Harris (2000)Dynamic risk factors in
sexual offendingThe purpose of this study was
to identify factors that could be useful for
officers supervising sexual offenders in the
community. Overall, substantial differences were
observed between the 208 sexual offenders who
sexually recidivated while on community
supervision and a comparison group of 201
non-recidivists. In comparison to the
non-recidivists, the recidivists had a greater
history of sexual deviance, such as diverse types
of victims, stranger victims, juvenile offenses
and paraphilias (e.g., exhibitionism,
cross-dressing). As well, the recidivists showed
more signs of an antisocial lifestyle than did
the non-recidivists. The recidivists were more
likely to meet criteria for antisocial
personality, psychopathy (PCL-R), and had higher
scores on objective risk scales (SIR and VRAG).
8
Officer interviews indicated that the recidivists
displayed more problems while on supervision than
did the non-recidivists. In particular, the
recidivists were generally considered to have
poor social supports, attitudes tolerant of
sexual assault, antisocial behaviour, poor
self-management strategies and difficulties
cooperating with supervision as indicated by
being disengaged, manipulative or absent. The
overall mood of the recidivists and
non-recidivists was similar, and each had
equivalent levels of life stress and negative
affect, but the recidivists tended to show an
increase in anger and subjective distress just
prior to re-offending. In other words,
psychological symptoms appeared as acute, but not
stable, risk factors. With rare exceptions, the
same risk factors applied to both rapists and
child molesters.
9
Statistical Methods of describing and quantifying
the accuracy of risk predictions

• Correlations
• ROC AUCs (receiver operating characteristic
  analysis)

10
Sex Offender Risk Assessment Measures (Campbell,
2000)

• Professional judgement
• Unstructured or clinical
• Structured
• Actuarial decision making
• Clinically Adjusted Actuarial Prediction
• Multifactorial approaches and classification
  trees (to come)

11
Professional judgementMost commonly used
method for violence risk assessmentFlexible,
requires limited training and resources

• Un-structured based on idiosyncratic
  impressions
• Poor predictive validity,
  unreliable and false positive bias
• Predictive accuracy only slightly better than
  chance
• (r.10, Hanson Bussiere, 1998)
• Structured Imposes structure on evaluation
• Must refer to at minimum a fixed and explicit
  set of risk factors. Combine ratings on such
  to guide
• assessment of risk.
• Sexual Violence Risk 20
• (SVR-20 Boer, Hart, Kropp
  Webster, 1997)
• Structured Risk Assessment 99 (SRA-99
  Thornton, 1999)
• Matrix 2000 (Thornton, 2000)

12
SVR-20 (Boer, Hart, Kropp Webster, 1997)

• 20 standard risk factors
• Three main areas Psychosocial adjustment
• Sexual offending
• Future plans
• Rate as present, possibly present or not
  present
• Translate into low, moderate, or high risk
  categories

13
Sample Conclusion

• Based on a comprehensive risk assessment, it is
  my opinion that should he be released into the
  community Mr Smith poses a high risk for sexual
  violence relative to other sex offenders
  incarcerated in the Correctional Service.
• According to the available information, all of Mr
  Smiths sexual offences have been paedophilic in
  nature, involving the non-coercive sexual contact
  of young boys with whom he was acquainted through
  casual contact. There is no information to lead
  me to believe that his offences will change in
  nature or escalate in severity in the near
  future.
• Based on his past offences, if Mr Smith
  recidivates his victims are most likely to be
  boys between the ages of 6-12 years who live
  within a few miles of his residence. Given the
  long standing nature of Mr Smiths paraphilia,
  its resistance to treatment, and his extensive
  history of sexual offending, the most effective
  way to manage his risk of sexual violence is
  through incapacitation, that is, by denying his
  request for parole. Should Mr Smith by released
  into the community, risk management strategies
  should focus on intensive supervision. Electronic
  monitoring, frequent meetings with a parole
  officer might be effective supervision strategies.

14
Correlations/ ROC AUCs Hart (2000)

• Any violence
  Sexual violence
• r AUC r AUC
• PCL-R .45 .76 .20 .69
• VRAG .56 .83 .26
  .71
• SORAG .64 .88 .36 .77
• RRASOR .40 .73 .48 .77
• SVR-20 .52 .81 .31 .74

15
The SVR-20 (1998) can be purchased from
Psychological Assessment Resources

• www.parinc.com/

16
Actuarial devices

• Commonly-used adjunctive method for violence risk
  assessment
• Utilise statistical techniques to generate risk
  predictors
• Generally equal or superior to clinical judgement
  with respect to consistency (reliability) and
  accuracy (validity)
• Highly structured/systematic
• Objective limited role of discretion,
  empirically based and scientific
• Rapid Risk Assessment for Sexual Offense
  Recidivism
• (RRASOR Hanson, 1997)
• Sexual Offence Risk Appraisal Guide
• (SORAG Quinsey, Harris, Rice Cormier, 1998)
• Minnesota Sex Offender Screening Tool (Revised)
• (MnSOST-R Epperson, Kaul Huot, 1995)
• Static 99 (Hanson Thornton, 1999)

17
Rapid Risk Assessment of Sexual Recidivism
(RRASOR Hanson, 1997)

• 4 item actuarial instrument rated from official
  records
• Intended to be relatively brief screening
  instrument for predicting sexual offense
  recidivism
• Based on meta-analytic research and re-analysis
  of existing data sets.
• Items weighted according to ability to predict
  likelihood of recidivism over periods of 5-10
  years. Total scores range from 0 6 with a 10
  year estimated likelihood of recidivism ranging
  from 6.5 73.1 percent. Most offenders have
  scores which range between 1 and 4.
• Items Prior sex offenses (not
  including index offenses)
• Age at release (current age)
• Victim gender
• Relationship to victim

18
RRASOR Summary

• No manual
• Minimal peer reviewed studies
• Doesnt consider deviant sexual preferences,
  personality, treatment compliance or other
  dynamic variables.
• Insensitive to context, change
• Utility in assessing post-treatment changes in
  risk status limited.
• Potentially useful psychological instrument for
  establishing elevated risk of sexual violence
• Good predictive accuracy in development and
  validation samples (Hanson Thornton, 2000)
• r 0.27 AUC 0.71 (Hanson, 1997) sexual
  recidivism
• r 0.22 AUC 0.72 (Sjostedt Langstron,
  2000)
• sexual recidivism
  4 year follow up

19
The RRASOR (1997) is available to download from

• http//www.sgc.gc.ca/epub/corr/e199704/e199704.htm

20
Sex Offender Risk Appraisal Guide (SORAG
Quinsey et al, 1998)

• Authors strongly pro-actuarial based on
  Penetanguishene studies
• Modification of the VRAG (Quinsey et al, 1998)
• Do the findings generalise ?
• the universe is homogenous with respect to
  forensic institutions (Quinsey et al, 1998)
• 14 item actuarial instrument, with range of
  scores from 1 9.
• Includes both static and dynamic factors
• At least four of the factors included in the
  items have received little empirical support (ie
  history of alcohol abuse history of non-violent
  offenses marital status diagnosis of
  schizophrenia) (Campbell, 2000)

21
SORAG (1998) Items

• Living with biological parents until age 16
• Elementary school maladjustment
• History of alcohol problems
• Marital status
• Nonviolent offense history
• Violent offense history
• Sexual offense history
• Sex and age of index victim
• Failure on prior conditional release
• Age at index offense
• DSM-III criteria for any personality disorder
• DSM-III criteria for schizophrenia
• Phallometrically measured deviant sexual
  interests
• PCL-R score

22
SORAG (1998) Summary Likelihood of recidivism
is estimated for only general violence Restricted
in clinical usage due to inclusion of PCL-R
(training requirements) No manual. Lack of peer
reviewed support. ROC-AUCs 0.82 (Belanger
Earls, 1996) parole failure or recidivism of any
kind. 0.63 (Firestone, Bradford, Greenberg, Nunes
Broom, 2001) violent (including
sexual) recidivism follow up 7 years
23
The SORAG (1998) is available inQuinsey, V.L.,
Harris, G.T., Rice, M.E. Cormier, C.A. (1998).
Violent Offenders Appraising and Managing Risk.
The American Psychological Associationhttp/
/www.apa.org/books/431604A.html
24
Minnesota Sex Offender Screening Tool -
Revised(MnSOST-R Epperson, Kaul Hesselton,
(1998)

• 16 item actuarial instrument, constructed
  applying retrospective methods
• Incorporates both historical and institutional
  information (ie treatment participation).
• Designed specifically to predict sexual
  recidivism (unlike the VRAG and SORAG)
• Scores divided into 4 categories, with estimated
  recidivism rates from 16 88 percent over 6
  years.

25

• Correlations/ ROC AUCs
• More accurate at discriminating between sexual
  recidivists and non-recidivists than the RRASOR.
• r 0.45 0.77 AUC (Epperson et al, 1998)
• sexual recidivism follow up 6 years
• r 0.35 0.73 AUC (Epperson et al, 2000)
• sexual recidivism follow up 6 years

26
The MnSOST-R (1998) can be downloaded from

• http//psych-server.iastate.edu/faculty/epperson/m
  nsost_download.htm

27
Static-99(Hanson Thornton, 1999)

• Actuarial instrument consisting of 10 items
• Combination of items from 2 scales (RRASOR
  Hanson, 1997) and Thorntons Structured Anchored
  Clinical Judgement Scale (SAJC Grubin, 1998)
• Sample N 1,301 (Canada UK)
• Moderate predictive accuracy for sexual
  recidivism (r.33, AUC .71) and violent
  (including sexual) recidivism (r.32, AUC .69).
• Only small incremental improvements over the
  original two scales.
• Reliance on static factors.

28
Static-99 (Hanson Thornton, 1999) Items

• Prior sexual offences (same rules as in RRASOR)
• Prior sentencing dates (number of distinct
  occasions on which the offender has been
  sentenced for criminal offences of any kind)
• Any conviction for non-contact offences
• Index non-sexual violence
• Prior non-sexual violence
• Any unrelated victims
• Any stranger victims
• Any male victims
• Young
• Single

29
The STATIC-99 can be downloaded from

• http//www.sgc.gc.ca/epub/corr/e199902/e199902.htm

30
Clinically Adjusted Actuarial Prediction

• Adjusting actuarial predictions either up or down
  depending on professional judgement
• Structured Risk Assessment 99 (SRA-99
  Thornton, 1999)
• Stepwise process including
• Initial classification of risk Static-99
• Consider offenders functioning on dynamic risk
  factors to revise the original risk
  classification
• Consider offenders response to treatment
• Consider offenders typical offence pattern in
  association with situational factors
• Reflects diversity of assessment domains.
• Yet to be subject to systematic empirical
  evaluation.

31
Latest researchPredictive accuracy
(Correlations/ROC - AUCs)Barbaree, Seto,
Langton Peacock (2001)
VRAG, SORAG, RRASOR Static-99 predicted general
recidivism, serious (violent and sexual)
recidivism, and sexual recidivism. MnSOST-R
predicted general recidivism but not serious or
sexual recidivism, PCL-R predicted general and
serious recidivism but not sexual recidivism.
32
Conclusions

• It is possible to predict sexually violent
  recidivism in sex offenders with moderate
  accuracy
• Validity of structured professional judgements
  may equal that of actuarial instruments

33
References

• Hanson, R.K. Harris, A.J.R. (2001). A
  structured approach to evaluating change among
  sexual offenders. Sexual Abuse Journal of
  Research and Treatment, 13(2) 105-122.
•  
• McCarthy, J. (2001). Risk assessment of sexual
  offenders. Psychiatry, Psychology Law, 8(1)
  56-64.
•  
• Hanson, R.K. Thornton, D. (2000). Improving
  risk assessments for se offenders A comparison
  of three actuarial scales. Law and Human
  Behavior, 24(1) 119-136.
•  

34
Hanson, R.K. Harris, A.J.R. (2000). Where
should we intervene ? Dynamic predictors of
sexual assault recidivism. Criminal Justice
Behavior, 27(1) 6-35. Hanson, R.K. Bussiere,
M.T. (1998). Predicting relapse a meta-analysis
of sexual offender recidivism studies. Journal of
Consulting and Clinical Psychology, 66(2)
348. Barbaree, H.E., Seto, M.C., Langton, C.M.
Peacock, E.J. (2001). Evaluating the predictive
accuracy of six risk assessment instruments for
adult sex offenders. Criminal Justice Behavior,
28(4) 490-521.
35
The Sex Offender Need Assessment Rating (SONAR)
A Method for Measuring Change in Risk
Levels2000-1By R. Karl Hanson Andrew
HarrisCorrections ResearchDepartment of the
Solicitor General of Canadahttp//www.sgc.gc.ca/e
pub/corr/e200001a/e200001b/e200001b.htmThe
Development of a Brief Actuarial Risk Scale for
Sexual Offense Recidivism1997-04By R. Karl
Hanson, Ph.D.Department of the Solicitor General
of Canadahttp//www.sgc.gc.ca/epub/corr/e199704/e
199704.htm  
36
Static 99 Improving Actuarial Risk Assessments
for Sex Offenders1999-02By R. Karl
HansonDepartment of the Solicitor General of
Canada, OttawaDavid ThorntonHer Majestys
Prison Service, Londonhttp//www.sgc.gc.ca/epub/c
orr/e199902/e199902.htm Dynamic Predictors Of
Sexual Recidivism1998-1by R. Karl Hanson
Andrew HarrisCorrections ResearchDepartment of
the Solicitor General Canadahttp//www.sgc.gc.ca/
epub/corr/e199801b/e199801b.htm
37
Predictors of sexual offender recidivism a
meta-analysis1996-04By R. Karl Hanson Monique
T. BussièreCorrections ResearchDepartment of
the Solicitor General Canadahttp//www.sgc.gc.ca/
epub/corr/e199604/e199604.htm