Lec 6/ Ch 10

Question 1

What is Risk Assessment?

• risk assessment
• 2 parts of the assessment
• 4 settings they are conducted
  
  • define civil setting
    
    • Civil commitment
    • child protection
    • MH professional duty to warn
  • 3 types of criminal settings
• Solicitor-client privilege

4 Types of Prediction Outcomes

• true +ve
• true -ve
• false +ve
• false -ve
• how are errors related
• Consequences of false +/-
• Fatal vs non-fatal false +
• • Base rate
• Base rate problem
• low base rates = ?

Answer 1

What is Risk Assessment?/violence risk assessment

• assess probability of whether the person will commit violence
• 2 processes in risk assessment (focus = prevention)
  
  • 1. Prediction: identify risk factors for future violence
  • 2. Management: dev interventions to reduce future violence
       * Focus: treatments to reduce the risk or manage the risk
       *

Risk Assessments: When are They Conducted?

Civil contexts: one’s private rights and the related legal proceedings

• Civil commitment: hospitalize those w/ MD if they pose PS threat
• Child protection: protect child from abuse/neglect, gov agency can remove child from parents/guardian
• Immigration: prevent terrorism
• MH professionals
  
  • Duty to warn: need to intervene if patient will act violently

Criminal Settings

• pretrial, sentencing, release (ex. parole application)
• Disclose info about potential risk
  
  • Solicitor-client privilege: lawyers and experts hired by lawyers can freely discuss case w/ clients
  • Public safety outweighs doctor patient confidentiality and Solicitor-client privilege

Types of Prediction Outcomes

• True +: correct prediction (predicted violent person -> reoffend)
• True -: correct prediction (predicted not violent -> didn’t reoffend)
• False +: incorrect (predicted violent -> didn’t reoffend)
• False -: incorrect (predicted not violent -> reoffend)
• The errors are related: false positive errors decrease -> false -ve increase
• Consequences of false +/-
  
  • False + -> deny freedom; false -ve -> harm public safety (ex. another child harmed by sex offender)
• False + can be tolerable if cons are not super severe (ex. more supervision on offender when released)
• False + can be fatal if cons are high (ex. death penalty)

The Base Rate Problem

• Base rate: % of ppl in pop who commit crime or violence
• If the base rate is too high/low -> wrong predictions
• Ex. low base rate -> many false +ve
  
  • High profile school shootings have lots of media attention, but occur infrequently -> wrongly classify many teens as potential shooters

Question 2

Baxstrom and Dixon studies

• Baxstrom v. Herald (1966)
  
  • ruling on Baxstrom
  • subsequent event
  • study result after following those released against Dr’s advice
• Dixon v. Attorney, 1971
  
  • what happened after the case
  • result
• 2 main findings from the 2 studies
• Canadian and US courts ruling on risk assessments

Methodological Issues

• ideal way to eval risk
• our reality when eval risk
• Monahan and Steadman 1994
  
  • 3 weaknesses of rs on prediction of violence
  • MacArthur Violence Risk Assessment Study (base rate for agency records vs collateral/patient records)

Answer 2

Baxstrom and Dixon studies

• Baxstrom v. Herald (1966)
  
  • US SCC ruled Baxstrom is detained beyond his sentence and released him
  • After this case, 300 MD offenders from psychiatric hospitals were released or transferred to less secure institutions
    
    • Study: Followed 100 of these patients released in the community but MH professionals were against
      
      • Only 20 were arrested after 4 years; 7 committed violent offence
      • -> low accuracy of MH professionals predicting violence
• Dixon v. Attorney, 1971
  
  • after the case, 400 forensic patients were released
  • only 60 (15%) were arrested/rehospitalized due to violence 3 yrs after
• Point: MH professionals suck at making accurate predictions of violence -> many patients were locked up for no reason
• 2 key findings
  
  • 1. Base rate for violence is low
  • 2. False +ve rate is very high (86% and 85%)

Methodological Issues

• ideal way to eval risk
  
  • assess a large # of offenders -> release them to community regardless of risk level -> track offenders to see if they reoffend
• Issue: unethical to release high risk indiv
• Reality: can only evaluate risk assessment instrument using a low-risk reoffending pop -> conclusions are limited
• Monahan and Steadman 1994
  
  • 3 weaknesses of rs on prediction of violence
  • 1. Limited # of risk factors studied; there’s many
  • 2. Rs use official crime records as DV (criterion issue)
       * Issue: many crimes are unreported
       * Violent crimes go undiscovered and underestimated (false + → true +)
       * MacArthur Violence Risk Assessment Study
       
       • Official agency records, base rate of violence =5%
       • patient and collateral reports, base rate = 30% → 6 x higher
       • Collateral reports: info from ppl/agencies who knows patient/offender
  • 3. Need to measure more dimensions of violence (how Dv is defined)
       * Severity of violence (threat vs severe)
       * Types of violence (spousal vs sexual)
       * Targets of violence (fam vs stranger)
       * Location (institution vs community)
       * Motivation (reactive; unplanned violent response due to provocation) vs instrumental (violence used for a goal)
       * Some risk factors are associated w/ certain forms of violence
       * Ex. history of sexual offences predict future sexual offences, not bank robberies

Question 3

Judgment Error and Biases

• illusionary correlation
• Dr confidence and accuracy

Approaches to The Assessment of Risk

• 3 methods of risk assessment
  
  • Unstructured
  • Mechanical
    
    • Actuarial prediction
  • SPJ
• Skeem & Monahan 2011
  
  • Violence risk assessment approaches has 4 components
  • unstructured vs actuarial vs SPJ

Types of Risk Factors

• Risk factor
• 2 main types: static, dynamic
• RF as a cont construct
  
  • 2 ends
  • middle

Answer 3

Judgment Error and Biases

• Heuristics -> inaccurate decisions
  
  • Assume some traits are associated w/ risk but are not
• Illusionary correlation: belief that a correlation exists b/w 2 events that in reality are not or less correlated than believed
  
  • Ex. clinician thinks there is high correlation b/w MD and violence
    
    • Inconsistent findings
  • Ex. clinicians ignore base rates of violence
    
    • Clinicians are may not be aware of how often those w/ characteristics act violent
      
      • Ex. bizarre delusions may be used as a sign of high risk
  • Clinicians are overconfident in risk assessments -> more likely to recommend interventions
  • They are not accurate (ex. Dr. Grigson)

Approaches to The Assessment of Risk

• 3 methods of risk assessment
• 1. Unstructured clinical judgement: discretion decisions and lack guidelines
     * No predefined rules; Risk factors vary across clinicians and cases
• 2. Mechanical prediction (opp): has predefined rules on what factors to consider, how info is collected, how info is combined to make risk decisions
     * Risk factors do not vary across clinicians and cases
     * Actuarial prediction: decisions are based on risk factors selected and combined based on epical or stat association w/ a specific outcome
     
     • IOW: measured many risk factors -> followed offenders -> only those risk factors related to reoffending in the sample are selected
     • actuarial prediction was equal or better than unstructured clinical judgement
     • Criticism: static risk factors; can’t measure change in risk over time; no info on interventions
• 3. Structured professional judgement (SPJ) – middle ground
     * “professional” = assess by diverse professionals (ex. officers, social workers)
     * Decisions are guided by a predetermined list of risk factors that are selected from rs and professional lit
     * Judgement of risk lv is based on evaluator’s professional judgement
     * Rs still ongoing on actuarial vs SPJ which is more accurate
• Skeem & Monahan 2011
  
  • Violence risk assessment approaches has 4 components
  • Not all assessment hv them all
  • A. Identify empirically valid risk factors
  • B. Determine a method for measuring these risk factors
  • C. Establish a procedure for coming scores on the risk factors
  • D. produce an estimate of violence risk
• Unstructured clinical judgement = none
• Actuarial = all
• Structure professional judgement
  
  • Yes: identify and measure risk factors
  • Vary: combine and produce risk factors

Types of Risk Factors

• Risk factor: measurable feature of a person that predict future violence
• 2 main types: static, dynamic
  
  • Static RF (historical risk factors) do not fluctuate over time, don’t change by treatment (ex. age)
  • Dynamic RF (criminogenic needs): fluctuate over time, will change (ex. antisocial attitude – treatment can change this)
• Rs now conceptualize risk factors as a cont construct
  
  • One end = static risk factors
  • other end = acute dynamic risk factors
    
    • Acute dynamic RF: change rapidly w/in days, hrs, min, prior to offence (Ex. -ve mood, lv of intoxication)
    • Middle = stable dynamic RF
      
      • change only over mo, yr
      • target for intervention
      • Ex. criminal attitudes, coping ability, impulse control

Question 4

Important Risk Factors

• 4 risk factors
  
  • 1. Historical/static RF
       * Past b
       * Age of onset
       
       • Exception F
         * Childhood history of maltreatment
       • physical abuse/neglect vs sexual abuse
  • 2. Dispositional RF
       * personality
       
       • impulsive & recidivism
       • psychopathy & recidivism
       • psychopathy + deviant sexual arousal predicts ?
  • 3. Clinical RF
       * substance use
       
       • drug-violence link: direct and indirect effects
       • drug-crime link
         * MD
       • public perception vs reality
       • psychosis influence
  • 4. Contextual/ Situational RF
       * Lack of social support
       
       • Henggler et al: 4 kinds of support
       • family relationship
         * Access to weapons or victims
• Terrorist RF

Answer 4

Important Risk Factors

• 4 risk factors
• 1. Historical/static RF: events experienced in the past (ex. criminal history v, history of violence)
• 2. Dispositional RF: traits, attitudes, style (ex. demographic v; age gender, psychopathy)
• 3. Clinical RF: MD symptoms and type
• 4. Contextual/ Situational RF: aspects of indiv’s env that (ex, access to victim and weapons, lack of social supports)

Terrorist RF

• There are no unique risk factors among violent terrorist
• Also, they do not have the same risk factors as general offenders

Historical Factors

• Past b: most accurate b of future b (ex. violence)
• Age of onset:
  
  • Those w/ antisocial b young are more chronic and serious offenders
    
    • 50% of male youth convicted of violence offence in early adult
    • Age onset is not as strong of predictor for F offenders
• Childhood history of maltreatment
  
  • Having a history of physical abuse/neglect is associated w/ increased risk of violence
  • Widom 1989
    
    • Those who were victims of sexual abuse were no more likely than those who were not sexually abused to commit offences
    • Those who were victim of physical abuse/neglect were more likely to commit crime compared to those who were not

Dispositional Factors

• Personality characteristics
  
  • impulsiveness & psychopathy are associated w/ more risk of crime and violence
    
    • High impulsivity (low SC) → recidivistic
    • Psychopathy: callous (insensitive), grandiose, manipulative, lack remorse, impulsive, irresponsible
      
      • Psychopathy predicts reoffending in many countries, for both sex,
        
        • Weakly or unrelated to violent reoffending in female teens
• psychopathy + deviant sexual arousal predicts sexual recidivism (recidivism rate = 70%; vs others = 40%)
  
  • Deviant sexual arousal = preference for inappropriate stimuli (ex. kids, violent nonconsensual sex)

Clinical Factors

• Substance use
  
  • Drug and OH use are associated w/ crime and violence
  • Drug-violence link
    
    • direct effects (from drug on brain) and indirect effects (use violence to get drugs)
  • Drug-crime link
    
    • Heroin & crack is most associated w/ crime
    • some drugs: more taken = more crime
    • OH and drug use mod related to recidivism
    • Those w/ substance abuse diagnosis, rate of violence for men (22%) and women (17%)
    • Sexual risk -> associated w/ drug use
    • Point: MD predict minor drug use; high risk b were predicted by polysubstance abuse
• MD
  
  • Public thinks MD and violence are related
  • Most ppl w/ MD are not violent
  • Those w/ affective disorders and schizo are linked w/ higher rates of violence
  • Those w/ suicide attempts and self-harm b were more likely to engage in verbal and physical aggression to others
  • Psychosis (knowing what’s real or not) → 50-70% increase in the odds of violence

Contextual Factors

• Lack of social support
  
  • Henggler et al 1998
    
    • 4 kinds of support
    • 1. Instrumental – provide necessities of life
    • 2. Emotional – give strength
    • 3. Appraisal – give aid or courage
    • 4. Info – give new facts
  • Study: relationship b/w offender w/ MD and family is related to violence
• Access to weapons or victims
  
  • increases chance for violent recidivism (actuarial observation)

Question 5

2 myths Risk Assessment

• do they provide objective info
• is there clear consensus on which assessment is used

Risk-Assessment Instruments

• Hanson & Thornton static 99
  
  • what it predicts
  • type of RF (SPJ/actuarial/unstructured)
• Webster et al HCR-20
  
  • what it predicts
  • type of RF (SPJ/actuarial/unstructured)
  • 3 main scales
• Campbell et al 2009
  
  • Results: instruments’ predictive effectiveness: effect ?
  • HCR-20 & institutional violence: effect?

Theory -Coping relapse model of criminal recidivism

• 3 parts

Answer 5

2 myths

• Risk statement may not provide objective info on an offender’s risk to offend
  
  • ranking of low, mod, high labels can mean diff things in diff contexts
• There is no clear consensus among rs as to which type of risk assessment should be used (ex. actuarial vs SPJ)

Risk-Assessment Instruments

• Hanson & Thornton 1999 - static 99
  
  • predict sexual recidivism,
  • actuarial scale; static items
  • Score = 0-12; 10 item, actuarial scale
  • Scores has 4 risk categories: low, mod-low, mod-high, and high
  • Items
    
    • Young age at time of release
    • Ever lived w/ intimate partner
    • Prior nonsexual violent convictions
    • # of prior sex offences; sentences
    • Any male victims; unrelated victims; stranger victims, noncontact sex offences; index nonsexual violent convictions
• Webster et al 1997 - HCR-20
  
  • predict violent b
  • SPJ approach
  • list of 20 items organized into 3 main scales that align RF into
    
    • Past (historical)
    • Present (clinical)
    • Future (risk management)
    • 1 Historical (static nature)
      
      • ex. Past violence; age of first violent offence; PD
    • 2 Clinical (current, dynamic RF)
      
      • ex. Lack of insight; Impulsivity
    • 3 Risk management (future community or institutional adjustment)
      
      • ex. Exposure to destabilizers; Lv of personal support
• Campbell et al 2009
  
  • Meta-analysis – compared predictive effectiveness of several risk assessment
  • Results: effect size: .25 to .3 (strong)
  • HCR-20 (r = .31) , LSI-R were most predictive of institutional violence
• Yang et al 2010
  
  • Similar results, each risk assessment tool = good predictors of violence, no tool is superior

Where Is the Theory?

• Zamble and Quinsey 2007
• Coping relapse model of criminal recidivism
• 1^st event: env trigger
• 2^nd event: emo and cog appraisal of the event
  
  • If stressed/-ve emo → cope (good vs bad)
• 3^rd - 2 factors that affect one’s response: indiv and response mech
  
  • Indiv influences (stable)
  • Available response mech (dynamic)

Question 6

Unique Sub-populations of Offenders

• criminality
• can we apply risk assessment tools for Caucasian male offenders to them

Answer 6

• Subpop hv differences in criminality
• some risk assessment tools for Caucasian male offenders can be used w/ F and abor; but not always the case

Question 7

• Protective factors
• 3 limitations of risk assessment
• desistance
• 4 reasons for desistance

Answer 7

What about Protective Factors?

• Protective factors: factors that mitigate or reduce the likelihood of antisocial acts or violence of high risk offenders
• Protective factors
  
  • Prosocial involvement
  • Strong social supports
  • Positive social orientation (ex. school, work)
  • Strong attachments (not to antisocial other)
  • Intelligence

Risk assessment has limitations

• No method can determine specific risk lv for indiv
• inconsistent agreement on SPJ’s lv of risk (low, mod high)
• Measures in one country/pop may not generalize to others
  
  • risk assessment measures need to be validated in the community they will be used

Why Do Some Individuals Stop Committing Crimes?

• Desistance: process of ceasing to engage in criminal b
• Potential reasons
  
  • may be related to “good” work or “good” marriages
    
    • Those w/ longer criminal histories are less likely to marry, and tend to marry criminal others
  • Age is related to criminal b: older -> less offending b/c of maturation
  • Insight triggered by -ve events connected to their criminal lifestyle
    
    • Ex. stupid to chop someone due to smth trivial
  • Social avoidance: avoid situations that make you violent

Question 8

• Parole
  
  • does it reduce length of sentence?
  • is it automatically granted?
  • Purpose

Answer 8

Parole in Canada

• Definition: release of offenders from prison into the community before their max prison sentence is complete.
  
  • Parole does not reduce the length of a sentence; it affects the way the sentence is served.
  • Allows offenders to serve their sentence in the community under supervision of a parole officer. (help reintegrate in society in some sense)
  • Parole is not automatically granted (need to apply when it is available for you) → Parole Board of Canada makes parole decisions
• The majority of offenders are serving fixed-length sentences; they will be released once their sentence ends.
• Parole contributes to public safety by helping offenders re-integrate into society as law-abiding citizens through a gradual, controlled, and supported release with conditions.