Sexual Offending

definition of sexual aggression/assault/violence

• Any actual, attempted, or threatened sexual contact with a person who is non-consenting, or unable to give consent (Boer et al., 1998).

UCR Survey Data for Sexual Offences Reported to Police in Canada in 2020

- Sexual assault make sup the bulk of it

- 0.4 aggravated sexual addault – smaller

- Child pornography – 10 = possession

- Child porn – making – 19

DSM Criteria for Paraphilia

• Recurrent, intense, sexually arousing fantasies, sexual urges, or behaviors generally involving:

• Nonhuman objects (e.g., fetishism) – classical condition component

• Suffering or humiliation of oneself or sexual partner (e.g., sadism, masochism) – does not count if it’s consensual and mutual

• Children or nonconsenting persons (e.g., pedophilia, voyeurism, frotteurism – rubbing up against someone in a closed room)

common and uncommon paraphilias

• Pedophilia

• Exhibitionism

• Sadism and masochism

• Fetishism

• Zoophilia and bestiality

• Necrophilia

• Klismaphilia (enemas)

• Ampotemnophilia (amputations)

• Coprophilia (feces); coprophagia (ingestion)

• Urophilia (urine)

• Telephone scatologia (obscene telephone calls)

• Etc., etc., etc.

Seto’s motivation facilitation model of child sexual abuse

- Need motivating factor – paraphilia, high sex drive, intense mating effort

o Without this – you would take the offence away

- Add or Subtract facilitation factors –

o 1. Antisocial personality factors

 Antisocial personality

 Central 8 factors

o 2. State factors

 Alcohol use

 Passed out at a part

o Without these factors – someone might have iiegal thoguhts or inhibition but wont act on it

- A study with this mode at usask followed this model and was irrelevant of gender

what are the base rates of sexual recidivism

- Look at who already committed sex offences and followed up

- After 5 years = 9%

- After 10 = 13%

- After 15 = 16%

- After 20 = 18%

- After 25 = 18.5%

- Average estimate when usask was asked = 60% with treatment = 40%3

- Looks at official charges – might be lower due to unreporting

sexual recidivism in a sample of rapists over 15 years

recidivism in a boy victim offenders is double of that girls

-        Extend incest recidivism is least

why might recidivism with boy victims be higher than those of girls

More entrenched paraphilia is dragging this behaviour and other synamics of hypersexuality

o Crimes of opportunity

sexual recidivism based on past offenses

- Those with a previous sex offence have double the recidivism rates than those without a previous sex offender

sexual recidivism based on their age

- Those who are younger than 50 at release have almost more than double the recidivism than those at 50 at release

- Sexual violence follows age crime curve

what factors influence base rates of sexual recidivism

• Many factors influence base rates of sexual recidivism such as sexual offense victim profile, a history of prior sex offenses, and age

joseph fritzl case motivation

- Joseph Fritzl – had a lot of static variables that would not put him at high risk, but he has a lot of dynamic factors that caused his behaviour

broad domains of sexual violence risk assessment

sexual deviance

antisociality

other domains

sexual deviance - broad domain

• Atypical or illegal sexual interests and compulsive or hyperactive sexual behaviour/arousal

antisociality - domain

• Broad propensity for rule violation – essentially the Central Eight

- General criminal attitudes

- Bad peers

sexual violence risk assessment measures

empirical actuarial - static - second generation

empirical actuarial - dynamic - 3rd/4th generation

structured professional judgement - third generation

empirical actuarial static - second gen measures

• Rapid Risk Assessment for Sexual Offense Recidivism (RRASOR) – replaced by below

• Static-99/99R

• Static-2002/2002R – was supposed to replace above

• Minnesota Sex Offender Screening Tool (Mn-SOST-R)

empirical actuarial - dynamic - third and fourth gen

• Stable 2000/2007 (3rd gen)

• Violence Risk Scale-Sexual Offense version (VRS-SO) (4th gen)

strctured professional judgment - third gen measures

• Sexual Violence Risk-20 (SVR-20) – like the HCR-20

• Estimate of Risk for Adolescent Sexual Offense Recidivism (ERASOR)

• Risk for Sexual Violence Protocol (RSVP) – variation of SPJ tool

static 99r

- Higher score high risk

- Only using static variables

- Made in 1999

- Risk goes down as they are older – age crime curve

- Prior sex ofences – sexual and non sexual

- Ever lived with lover

- Type of victim

- Rating for each variable -3 to 0

- Score between -3 and 12

- Scores ebove 6 = high

- Area under the curve – 0.7 – large effect size

Stable - 2007 items

- Stable dynamic variables

- Developed in 2007

- Developed on community supervision sample

- AUC = 0.67

- N = 615

- 13 items – 3 point scale

- Some sexual variance specific items

- Relationship problems

- Loneliness

- Emotional identification children – those who have harmed children

Violence Risk Scale - Sexual Offense Version Items

- Like the VRS – but more specific to sexual offending populations

- From clear water sample

- Static and dynamic items – 4 point to scale

- 2-3 point – criminogenic items

- Use static 99 to estimate risk levels instead of static variables

- Used with stages of change

- Median AUC static – 0.69

- Median AUC dynamic – 0.67

- Total median AUC dynamic – 0.67

- Medium effect size

Predictive Accuracy of Sexual Violence Risk Tools

- Large scale metanalysis for sexual offending assessment tools

- Mid 0.60 to low 0.61

- LSI is lowest – doesn’t have sexual specific content

- 0.68 – all dynamic measures

- All predict well but have different purposes

Sexual Violence Risk Tools Prediction of Sexual Recidivism - Summary of Prediction Meta Analyses

- Medium to large effect size for sexually specific tools

- VRS-SO is highest of the dynamic tools with large effect size

- HCR-20 is small effect size since it’s a general tool

- Unstructured clinical judgement – lower than 50 effect size

Which instruments are better? Static instruments or dynamic ones?

• Both do a pretty good job

• Static actuarial measures (e.g., Static 99)

• Mean d = .67, k = 81, N = 24,089

- Cannot inform treatment

• Dynamic (e.g., Stable 2007; VRS-SO)

• Mean d = .66, k = 29, N = 5,838

What’s the difference between static and dynamic?

• Purely static measures

• Are limited in their ability to inform treatment

• May overlook important domains of risk (i.e., not comprehensive enough)

• Are unable to capture changes in risk (e.g., through treatment)

Concentual Quandary - Case and Risk

- Case A pre tx score of 50 with 7 pts change = post tx score of 43

- High risk

- Case B – pre tax score of 25 – 0 points of change = 25

- Low risk and low change

- Case A is higher risk since numerical score is linked to recidivism -

- Case c – PRE TAX SCORE OF 50 – 0 PTS OF CHANGE = 50 post tx score

- High risk and low change

- Case C is highest risk now

- Case D starts with 25 – 5 pts of change = 20

- HAS LOWEST RISK

- High risk and high change

Four Risk Change Groups based on Static 99R

- Low risk low change

- Low risk high change

- High risk low change

- High risk high change

Trajectories of Sexual Recidivism as Function of Baseline Risk Level and Change

- Deeper the line = more recidivism

- Red line = high risk and low change = case C

- Low risk and high change = green

- Blue – low risk low change

- Case a – purple – high risk high change

If you only used the Static 99R – you wouldn’t see any change so you would unnecessarily be detaining people

Trajectories of Sexual Recidivism as a Function of Base Risk Level and Change

• When assessing risk-relevant change it is critical to account for baseline risk level.

• The dynamism of risk underscores the necessity of using dynamic instruments with populations that may have been exposed to credible change agents

- Need to use static to assess base line for risk

How to use change information in sexual violence risk assessment

Look at 10 year follow up for individualized rates of recidivism for each score

Recidivism rate of 0 at 13 – not many people have that score

- Extreme scores are infrequent

- Base rate is lower

- U can predict what the score should be considering the other scores – logistic regression

There is a linear association but not perfect

Rationale of using change information in sexual violence risk assessment

• Low, medium, high not very informative

• Best if you can use exact scores

• But recidivism rates unstable at extreme scores

• Use logistic regression to estimate rates of recidivism

• More clinical utility and practicality for decision-making purposes, while reducing bias

What the rates should should be given baseline scores and rates of change

• Four samples of treated sex offenders with VRS-SO and Static-99R ratings (N = 913)

• Minimum 10-years of offending opportunity

• 5 and 10-yr LR generated estimates for sexual and violent recidivism

• Risk scores entered into LR equation

• Excel spreadsheet computes risk estimate

•  VRS-SO Calculator

- 5 year estimated sexual recidivism as a function of mean baseline risk score and change

conclusions on sexyal violence risk assessment

• Several risk factors, both static and dynamic identified

• Several measures developed to provide a structured and systematic appraisal of risk

• Base rates tend to be low so prediction is difficult – rely on logistic regression

• Research suggests measures, on average, capable of predicting AUC .65 to .75 (Hanson & Morton-Bourgon, 2009)

• No evidence that any given measure has superior accuracy in predicting sexual recidivism

• Goal to treat and manage risk to prevent recidivism

most common methodology in sex offending treatment

single treatment outcome study

single treatment outcome study

• Treated sex offenders compared to a comparable control group of untreated offenders

• Reduced recidivism in treated offenders supports treatment efficacy

majority of sex offender treatment programs currently operating in north america are…

behavioural in nature

characteristics of sex offender treatment programs

psychoeducationa

grouo

homework assignemnts

multimodal

oppurtunitiies for individual therapy

psychoeducational characteristics of sex offender treatment group

- Teaching group

- Teaching CBT strategies for concerns issues and problem areas

- Thinking errors

- Inappropriate arousal

benefit of group sexual treatment

- They are all similar – can sense BS

- Universality – common empathy and knowledge for each other’s concerns

current sex offender treatment approaches

CBT

multimodal

• modifying deviant fantasis andinterest

• replace with prosocial

inhibiting medicine

challending attitudes

understanding cycle and addiction

bad relationships

anger management assertiveness substance abuse concerns

increasing social supports

community supports

issues with single treatment outcome study

• Need to ensure groups are comparable on variables that could influence the outcome (i.e., dependent variable)

• Random assignment if feasible

• Matching groups on key variables

• Statistical controls of key variables

conclusion of singe treatment outcome program effectiveness

• 1. No sexual offense treatment program (SOTP)

• 2. Treatment as usual (standard prison based SOTP; group CBT)

• 3. Rockwood CBT strength-based group program

- Found that 3 groups weren’t equivalent

o Rockwood has lower recidivism

o The three groups difference significantly on risk variables that could influence the DV

o Summed the items to create a brief actuarial risk scale (bars) and groups diffs lardely eliminated when examined by risk level

o How do we control for risk

 Risk stratified – combine low medium and high within the group and then compare them between each other

- Rockwood and TAU effects over no treatment

- Rockwood effects over TAU – but primarily for high risk group[

- Most significant difference for high risk people – risk principle

how to interpret odds ratio

• Treatment effect  OR < 1.0 (95% CI below 1.0)

• No treatment effect  OR ≥ 1.0 (95% CI includes 1.0)

• Negative treatment effect  OR > 1.0 (95% CI above 1.0)

overall analyses for SOTP effectivesness

- Odds ratio for sexual recidivism – 0.64

- Outlier was that odds ratio was for sexual recidivism after treatment was 2% higher than 1

conclusions on golbal findings of SOTP

• Significant ES, with or without outlier, for all three outcomes

• Remarkable continuity in ES magnitude and observed rates of sexual recidivism for treatment-control group comparisons across studies

• Substantial ES heterogeneity

• Underscores need for moderator analyses

- Moderator analyses – variables that impact effect size

o Some programs are better some programs are worse – why some might conlude that treatment is effective versus not

moderatory analyses

variables that impact effect sizes

example - using clinicnas vs not

when supervision is provided in SOTP

significant effect - larger - 0.56 odds ration

no supervision provided - 0.74 odds ratio

service quality for SOTP services

- RNR principles

- Weaker – 1 or no principles

- Promising – 2 principles

- Most promising – all principles

- Successively smaller odds ratios and more reductions in recidivism

arousal conditions and SOTP

- Not a requirement

- No significant effect when arousal conditions is not used

- Significant effect when arousal conditioning is used

o Learning how to regulate drive

o Regulate fantasy content

o Textbook – what are these strategies

program moderatory conclusions

• Promising and Most Promising programs associated with largest reductions in sexual recidivism

• Larger ES associated with supervision of service delivery

• Program interventions?

• Arousal conditioning (most programs)  larger ES

- Light detector test – in Canada not admissible as evidence

o Those who didn’t use it – large effect size