Changes in Risk Assessment
RRASOR
1) Prior sexual charges or convictions
2) Victim gender
3) Age
4) Relationship to victim
Static99
RRASOR plus 6 items
1. Prior sentencing dates
2. Noncontact offenses
3. Index nonsexual violence
4. Prior non-sexual violence
5. Stranger victim
6. Single
Three Generations of Risk
Assessment
•Clinical Judgment
•Actuarial Assessment
•Actuarial Plus Dynamic
Approaches to Risk Assessment
d Subjects Studies
Unstructured clinical .43 1,723 9
Structured professional
judgment .41 844 5
Actuarial (sex) .70 14,160 55
(Hanson & Morton-Bourgon 2007)
What Does Static99 Measure?
Who Gets Caught
Differential Disclosure
Incest
When Did Child Rarely Disclose
Natural Parent
Immediately 17%
Later 30%
Never 55%
Differential Disclosure
Abused and neglected children
Skillful offenders
Decline in Crime 2008
Decline in Rape & Sexual Assault
1999-2008
53%
Base Rates Matter
More
Or
Fewer
Reoffenders for each score
Multiple Norms
Routine Sex Offenders (CSC)
Treatment Samples
Nonroutine Samples
High Risk Samples
Which Norms to Use
Correctional Services of Canada
Routine Cases
N = 2406
No screening procedures
No pre-selection for tx, or civil commitment
Which Norms To Use?
Preselected for Treatment
Referred for sex offender specific
treatment during current or prior
incarceration
Selected but no beds still selected
Non-routine
Treatment sample
High risk samples
Preselected for other reasons, e.g.,
offense severi
Which Norms to Use
High Risk Sample
Preselected for risk
Factors external to Static-99
SVP referral, mentally disordered, not
guilty by reason of insanity, referred for
intensive treatment
Norms
“Most cases will use routine correctional
sample”
Scores Versus Recidivism
Original Norms
Score Risk Years
5 10 15
0,1 Low 6% 9% 10%
2,3 Med/Low 10% 14% 18%
4,5 Med/High 29% 33% 38%
6+ High 39% 45% 52%
New Norms 10 Years
Score Routine5 Routine Treatment High Risk
Observ11ed Adjusted Need
-3 1.4 1.8 3.2
-2 1.8 2.4 4.2
-1 2.3 3.3 5.4 9.8
0 3 4.4 7 12.5
1 3.9 5.7 9 15.7
2 5.1 7.6 11.5 19.7
3 6.6 10 14.5 24.3
4 8.4 13 18.2 29.6
5 10.8 16.9 22.6 35.5
6 13.7 21.7 27.6 41.9
7 17.2 27.8 33.3 48.6
8 21.4 35 39.6 55.3
9 26.3 43.3 46.2 61.9
10 68
High Risk Sample
Bengtson, 2008
Pre-trial psychiatric eval in Denmark
Suspected of retardation or psychosis
Bonta & Yessine, 2005
Dangerous Offenders-
Indeterminate sentences
Potential Dangerous Offenders
Violent Offense after MR
High Risk Sample
Haag, 2005
All Canadian offenders released at MR
Knight and Thornton, 2007
Massachusetts Treatment Center-
Assessed or treated between 1959 -
1984
Sexually dangerous persons
High Risk Samples
Nicholaichuk, 2001
Treated at Clearwater treatment
program – maximum security
forensic mental health facility
Wilson & colleagues
Detained in prison until MR
Relative Risk vs
Absolute Risk
Relative risk consistent across 22
samples
Absolute risk not consistent
Charges vs Convictions
13 samples used charges
15 used convictions
Reconviction Vs. Rearrest
Rearrest = 1.4 x reconviction
10 Years
Rearrest Reconviction
Static 5 24.5% 21%
Static 6 31.5% 27%
Child Molesters vs Rapists
53% child molesters
47% rapists
Rapists & Child Molesters
Tested separately
Both work with new item
Treatment %
Primarily treated samples
Only one untreated sample
Samples
N = 7878
21 samples
Helmus
N = 8412
23 samples
Sample Size
Static99 = 8,893
Logistical regression 10 years
2,528
Samples Sizes
Routine 2,406
Non-Routine 1,642
Treatment 866
High Risk 722
Old Sample Size
Static-99 1086
Aging and Risk of Sexual
Offending
What Difference Does Age Make?
Recidivism & aging -.10
(Hanson & Bussiere, 1998)
Recidivism rates declined steadily with age
Extrafamilial child molesters – maintained
risk longer
N =3751
(Hanson, 2002)
30
Child molesters
25
20 Rapists
Recidivism 15
rate (%)
10 Incest offenders
0
0 10 20 30 40 50 60 70 80
Age-at-release from custody
Age & Recidivism
Reanalysis of Hanson
All groups declined steadily
Own data – linear decrease with age
N = 468
(Barbaree et al., 2003)
Fitted Sexual Recidivism Rates by Age
Graphs
60
50
40
One SA
30 Two SA
Three SA
20
10
0
25-29 30-34 35-39 40-44 45-49 50-54 55-59
Each one unit increase in age was
associated with 98% of the
recidivism rate of the previous
(younger) age
Recidivism rate of 32-year-olds was
98% of recidivism rate of 31 year-
olds
Current age item did not adequately adjust
for age
Could have left old item and “fitted a
complicated curvilinear age adjustment”
Decided instead to create a new age item
With New Age Item
Age of release does not add significant
incremental predictive validity
With old age item –
Age of release does add predictive validity
Meaning: New item adequately controls
for age
Comparing Static-99 to Static-99R
Validation Sample
N = 2,392
ROC ROC
5 Years 10 Years
Static-99 .713 .706
Static-99R .720 .710
New Age Item
Age Score
18 – 34.9 1
35 – 39.9 0
40 – 59.9 -1
60 + -3
Score Range
-3 to 12
Risk Categories
Score Risk Category
-3 through 1 Low
2,3 Low-Moderate
4,5 Moderate-High
6+ High
Do Over-rides Help?
Prediction of Recidivism (ROC)
Recidivism Static-99 Static-99 + Static-99 +
Type Over-ride Stable- 2007
Sexual .77 .75 .81
Any violent .74 .71 .77
Age 61
Attacked 73-year-old woman
Giving him a church tour
Stranger assault
Mr. Johnson
Age Offense History
29 “Has a history of sexual
assault”
46 Attempted molestation of
11-year-old
Mr. Johnson
Age Offense History
48 Violent rape of 17-year-old
boy – beat with chain
58 Molestation 6 yr old boy
59.5 Exposure to officer
No High Risk Aging Sex Offenders?
When to Over-ride?
Recent Offense – within 5 years
History of continuous offending
No evidence of impact of aging
Static, Stable & Acute
Static “Non-changeable life factors that
relate to risk for sexual recidivism,
generally historical in nature”
Static, Stable & Acute
Stable “Personality characteristics, skill
deficits, and learned behaviours that
relate to risk for sexual recidivism that
may be changed through intervention
Acute “Risk factors of short or unstable
temporal duration that can change
rapidly, generally as a result of
environmental or intra-personal
conditions”
Static99R + Stable 2007
Static99R 3 Year Recidivism 5 Year
Score Recidivism
Stable = 5 Stable = 14 Routine High Risk
2 3% 7% 5% 12%
5 7% 18% 11% 25%
7 14% 32% 19% 38%