Lecture 2: Assessment of Sexual Offenders

What are some of the reasons we might assess sex offenders?

• risk level

• treatment needs

• interests

  • potential victims

• legal reasons

Content of Assessment Interview

• demographic info

• childhood experiences

• experience of abuse

• education history

• employment history

  • protective aspects

• relationship history

• offense history

• description of current offense

Why carry out an assessment with sex offenders?

• decision making about sentencing

• determine risk level

• treatment needs

  • criminogenic needs

  • dynamic risk

• allow access to children?

• level of immediate risk

  • release decisions

  • intensity of supervision in the community

What does functional analysis determine?

the underlying motives & functions for the offending behaviour

What does functional analysis typically involve?

obtaining detailed information about the antecedents, the behaviours, and consequences of offending (the ABC model), to look for patterns and determine the reinforcing and maintaining factors of the behaviour

blank → blank → blank

antecedents; behaviours; consequences

What kind of info might we want to gather in an assessment interview?

• family history

• risk factors

• relationships they have

  • social support

• underlying psychological disorders

• paraphilic interests

  • sexual interests

• access to victims

  • who might be at risk

• info of the offence itself

• triggers

Functional Analysis

• Behaviours carried out along with the accompanying thoughts and emotions

• Not always a straightforward task with offenders due to some level of denial being common

• Cannot assume they’re being completely truthful

Risk Assessment

What does ‘Risky’ mean?

• risk of recidivism

• the type of reoffending

  • predicting sexual/violent offending

• suicide

Risk for what?

• type of reoffending (ex: any general, violent, sexual)?

  • reconviction, new charges, allegations

  • same crime or more serious crimes (ex: noncontact escalating to contact/violent offending)?

• institutional sanctions?

• harm of any kind?

• danger to self?

• over what time period?

  • imminently? next year? next 10 years? ever?

Generations of Risk Assessment

1. Clinical Judgment

2. Actuarial Assessment

3. Dynamic Assessment

1. Clinical Judgment

• Unstructured

• Based on experience and some level of knowledge of literature

2. Actuarial Assessment

• Static, actuarial, based on factors empirically related to recidivism

• Standardized assessment

• Cannot measure change

• Strong evidence

3. Dynamic Assessment

• Standardized assessment, measures change

• Some include Structured clinical judgement

• Fewer validation studies but good evidence

4. (Risk Assessment)

• includes explicit measures of change

• Similar to third but with an integrated case management plan

Static Factors

• Unchangeable, historical factors

• Statistically related to recidivism

• Also called actuarial assessment

• e.g., previous offending history, age, relationship history

Static factor was also called?

actuarial assessment

Dynamic Factors

• Potentially changeable factors; Amenable to change

  • overcome the limitations of purely static actuarial instruments

• Statistically related to recidivism

• Also called criminogenic need factors

• e.g., antisocial cognitions, self regulation problems

• Evidence that they are related to recidivism

Dynamic factors is also called?

criminogenic need factors

Static Assessment: STATIC-99

• 10-item inventory

  • Widely researched and validated

  • Displays good predictive validity

• However:

  • Not validated for use with women

  • Not validated for offenders under the age of 16

  • Not validated for offenders whose only offence involves the possession of child sexual abuse material offenses or statutory rape

  • Evidence of discrepancies between races

What areas does Static-99 assess?

• sexual deviance

• range of potential victims

• persistence (sexual offending behaviour)

• general level of anti-sociality

Static-99: Sexual Deviance

measured by whether the offender has offended against male victims, has never had a long- term relationship, and/ or has committed non-contact sex offences

Static-99: Range of potential victims

measured by whether the offender has offended against unrelated and/or stranger victims

Static-99: Persistence (sexual offending behaviour)

measured by prior number of sexual offences

Issues with Static Risk Assessment

1. provides a probability, not a certainty of future recidivism

2. developed around official recidivism events, thus each risk category inevitably underestimate true re-offense rates

3. may lead clinicians to ignore unusual factors that are relevant to the individual case

4. misleading for individuals not represented or validation samples

5. they take no account of acute risk factors that might indicate imminent re-offending

6. most don’t indicate which factors need to be addressed in treatment for risk to be reduced

Dynamic Risk Factors (Thornton)

• Deviant sexual interest

• Pro-offending (or distorted) attitudes

• Socio-affective (or relational/ intimacy) problems

• Self management/ Emotional regulation difficulties

Evidence-supported Dynamic Risk Factors

• atypical (deviant) sexual interests

• antisocial cognitions/pro-offending attitudes

• relationship problems

• self-regulation problems

atypical (deviant) sexual interests risk factors

• paraphilic sexual interests

• sexual preoccupation

antisocial cognitions/pro-offending attitudes risk factors

• offense supportive attitudes & beliefs

• hostile masculinity

• non-compliance with intervention

relationship problems risk factors

• capacity for adult relationships

• emotional congruence with children

• negative social influences

self-regulation problems risk factors

• impulsivity

• lifestyle instability

• emotional regulation problems

• problem solving deficits

What risk factors do you think would increase someone’s future sex offending?

• managing emotions

• multiple previous offenses

• lack of social support

• ecology/situational context

• substance abuse

• age

  • when the offending started

  • juvenile

  • lower age = increase risk

Dynamic Risk Assessment: STABLE 2007

• 13-item inventory based on sexual and non-sexual dynamic risk factors

• Intended for use with a static risk assessment

• 5 domains:

  • Social influences

  • Intimacy deficits

  • Self-regulation

  • Sexual self-regulation

  • Cooperation with community supervision

• Displays good predictive validity

• Evidence of discrepancies between races

Acute (contextual) Risk/’Panic Now’ Factors

• ‘the concept of risk is inherently contextual, as hazards arise in particular circumstances’

• No matter how good an actuarial assessment is, it can never take into account fluctuations in contextual/proximal factors

Acute Assessment: ACUTE 2007

7-item inventory based on sexual and non-sexual dynamic risk factors

● Victim access

● Hostility

● Sexual preoccupations

● Rejection of supervision

● Emotional collapse, i.e., evidence of severe emotional disturbance/ emotional crisis

● Collapse of social supports

● Substance abuse

• Intended for use with a static risk assessment

• Displays good predictive validity

Components of a 4th Generation Risk Assessment

• Static risk assessment

• Dynamic risk/ risk-needs assessment

• This dynamic assessment can be further divided into the assessment of

  • Stable

  • Acute factors

• Includes a measure of treatment change

• Risk/ Case Management Assessment

Overall Assessment Process

• Interview

  • clinician

• Psychosocial History

• education

• Offending history

• Personality characteristics

• Psychometrics

• test for impulsivity

• test for distorted

• Collateral information

Assessing Risk

• risk principle

• Utilize one of the risk assessment measures that we discussed previously

• Best evidence is for actuarial/ static and combined static/dynamic

How should assessment inform treatment?

• what type of treatment is most effective for the individual

• how intense the treatment should be

Use overall risk level to guide?

the intensity of treatment

Can you think of any issues with static risk assessment?

doesn’t take into consideration of dynamic factors

Risk Principle

Treatment intensity should be proportional to risk level

Assessing Need

• need principle

• Dynamic Risk Assessment measures (e.g., SARN)

• Psychometrics designed to tap into specific Need areas (e.g., distorted attitudes)

• Interview

• Collateral information

Need Principle

Treatment should target criminogenic need

• i.e., those psychological risk factors such as deviant sexual interests, pro-offending attitudes, socio-affective problems and emotional dysregulation

Assessing Responsivity

• responsivity principle

• Psychometrics designed to tap into specific responsivity areas

• Interview

• Collateral information

• E.g., Motivation, Intellectual Disabilities, Psychopathy, Denial, Treatment Readiness

blank characteristics will be what is actually addressed in treatment

need

Assessing responsivity will dictate?

whether any deviations need to be made from the standard format of treatment

• treatment style should be tailored to meet the learning style, and abilities of offenders

What do protective factors decrease?

the risk of recidivism

Examples of Protective Factors

• prosocial involvement

• intelligence

• strong social support

blank traits will be drawn upon & strengthened to address/overcome treatment needs

protective

What are challenges to offender risk assessment?

• Limited resources (e.g., interviews are time consuming and expensive)

• Use of single or multiple assessments?

• Gender-specific or generic instruments?

• Accounting for ethnic and cultural variations

• Accounting for new types of crime and criminals

• Facilitating re-entry and reintegration

Evaluating Evidence through?:

• assessment

• treatment

Assessment

• Examine which items/ measures predict sexual recidivism

• AUC of the ROC

Treatment

Examine which approaches reduce recidivism/ bring about change

How do we compare different risk assessments measures?

ROC (Receiver Operating Characteristics) AUC (Area Under the Curve) Analysis

• Ranges from 0 to 1

• 0.5= chance prediction

• 1= perfect predictive accuracy

• AUC values of .56, .64, and .71 are interpreted as small, moderate, and large predictive accuracy effects

• Example of Interpretation: AUC of .70 indicates that there isa 70% chance that a randomly selected recidivist from the sample would have a higher risk score than a randomly selected nonrecidivist.

Hit Rates & False Alarms

• We plot hit rates against false alarms for each point on a risk assessment measure to give area under the curve analysis

• If we take anybody above the red line as predicted to be dangerous…

• Anybody scoring above the red line who are recidivists are hits, i.e., the correct identification of those who have offended again

• Anybody scoring above the red line who are non-recidivists are false alarms

Hits

Anybody scoring above the red line who are recidivists

• Ex: the correct identification of those who have offended again

False Alarms

Anybody scoring above the red line who are non-recidivists

Score on a Measure (Static-99)

Here if we say anybody above 3 is dangerous, then, hit rate is high (100%), false alarm rate high (70%)

Here if we say anybody above 7 is dangerous, then, hit rate is low (30%), false alarm rate low (0%)

Anchor points: Here if we say anybody above 1 is dangerous, then, hit rate is low (100%), false alarm rate low (100%)

Anchor points: Here if we say anybody above 9 is dangerous, then, hit rate is low (0%), false alarm rate low (0%)

Comparison of Approaches

Individual studies suggest AUC is best for combined Static and Dynamic

• Average AUC for static/actuarial instruments to be 0.68 (averaged over

50 findings)

• Unstructured clinical judgment to be 0.58 (averaged over 12 findings)

More recent meta-analysis AUC predicting sexual recidivism

• total of 31 studies, containing 169 effect sizes from 45,673 risk judgments

• Unstructured risk assessment= 0.625 (5 studies)

• Structured risk assessment= 0.721(5 studies)

Evaluating treatment of effectiveness

• Random assignment

• Descriptive

• Risk Band

• Incidental cohort

• what was the offense rate

• Statistical control of group differences

• Matched controls

• Within treatment change

• how well does groups work in treatment

• Meta-analysis

• pull tg exisiting elements and combine the results

Offender is high risk; what questions might you ask?

• past history

• triggers

• mental health conditions

• substance use

If you were a clinician who needed to assess someone who had committed a sex offense, what would you do?

• early/late onsets

• background characteristics

• risk assessment measure

What does the risk principle support?

varying the intensity of treatment according to the risk level of the individual

Structured Assessment of Risk and Need-Treatment Needs Analysis (SARN)

a clinical framework used to assess sexual offenders’ risk, need, and progress in treatment; treatment planning tool

SARN process

1. actuarial risk is first measured using the RM 2000

2. SARN uses clinical ratings to assess 16 items that measure Thornton’s (2002) four dynamic risk domains

3. uses a scoring protocol that determines the relevance of each factor both as a proximal factor leading up to the offence, and as a factor in the individual’s life in general

What does the ACUTE-2007 assess?

the risk for both sex/violent recidivism and a total score for risk of general recidivism

Motivation for treatment

it can range from acceptance of accountability for offending to a willingness to attend treatment

Random Assignment/Randomized Control Trials

offenders are randomly placed in either a group that will receive treatment, or a no treatment control group

• provide the most conclusive evidence for treatment effects

• controls any pre-existing (known and unknown) differences that exist between the groups

• observed differences can be attributed to treatment. Any differences that are evident between the groups occur by chance

Limitations of Random Assignment/Randomized Control Trials

there is still the risk of differential attrition between the treatment and control group, posing a threat to the internal validity of the study

• expensive to conduct

• practical issues arise that include the likely unwillingness of most large institutional systems to approve of the random allocation of dangerous sexual offenders to a non-treatment group for various reasons related to the real life implications of such a decision

Descriptive

simply followed a group of treated offenders and reported recidivism rates without any comparison to a group of untreated offenders

Risk Band Analysis

compare the observed recidivism rates of the treated group to the rates of reoffence predicted by published risk assessment measures

• does not require a comparison group to be created and allows for examination of risk level by treatment interaction

• Limits: lack of a control group poses a threat to internal validity because it does not allow for the examination of the effect of no treatment

Incidental Cohort Designs

compare treated offenders with a comparison group of offenders who did not receive treatment for reasons that are not likely to be related to risk level

• can confound treatment effects with the time of measurement.

Statistical control of group differences

allows treated and untreated groups to be compared even if they are found to differ on important risk factors

Matched comparison groups

matching the offenders who attended treatment, with those who did not, on factors related to recidivism

Within-treatment Change

Looking at whether or not the group members demonstrate changes in the areas that they are meant to demonstrate change in(i.e., areas targeted in treatment) is a useful way of examining the relationship between treatment effectiveness and reconviction

• avoids problems with cohort, jurisdiction, selection bias, and differences in follow-up because all offenders in the study begin at approximately the same time

• Limits:

  • Studies looking at change within treatment alone (i.e., without relating them to recidivism) cannot provide evidence on their own for the effectiveness of treatment due to lack of an untreated comparison group.

  • They are also often limited by the quality of the psychometrics tests that are used to evaluate the changes.

Comparison of Approaches

designs quantitatively combine the results from a number of studies to determine if there is an overall effect amongst the studies as a whole

• allows for even small effect sizes to be detected given the large sample size that results from amalgamating a number of studies