Summary Risk Assessment
Lecture 1 - Introduction
Definition and Factors in Risk Assessment
● Definition
○ Older version
■ “Probability calculation that a harmful behavior or event will occur, and involves an assessment about
the frequency of the behavior (or event), its likely impact and who it will affect”
○ More recent version
■ “The attempt to predict the likelihood of future offending in order to identify individuals in need of
intervention”
○ Risk assessment is always talking about future, uses past to do so
○ Risk Assessment must specify
■ Behavior (e.g. violent behavior)
■ Potential damage or harm caused by the behavior (e.g. child sexual abuse)
■ Probability that it will occur, and under what circumstances
● Types of Factors
○ Protective factors
■ Decreasing likelihood of criminal behavior
■ Conceptualization
● Prosocial means or strategies to achieve one or more primary human goods (common life
goals)
○ We all have life goals but use different strategies to achieve those
■ Good Lives Model (GLM) of rehabilitation
● Accomplish a ‘good life’ by fulfilling primary goods
● Offending = maladaptive attempt to attain one or more primary human goods (life goals)
■ Interaction with risk factors, serve as buffer
■ Improve predictive accuracy
■ Potentially more engaging and effective interventions
■ Improvement protective factors -> (non) recidivism
■ Measurement
● SAPROF
○ Structured Assessment of Protective Factors for violence risk
○ Different versions
■ Juveniles and young adults
■ SAPROF-Sexual Offending
■ SAPROF-Intellectual Disabilities
■ SAPROF-Intensive care
○ Risk factors
■ Static
■ Dynamic
○ Static Risk Factors
■ Not changeable / treatable, often historical factors
● E.g. age, gender, ethnicity, first offender, type of offense, first conviction
■ Risk remains the same over years, cannot be used for interventions
● So provide important information but cannot be used in handling of offenders
■ Not used in forensic psychology, more in policy making
■ Assessment
● StatRec
○ Static Risk of Recidivism
○ Gender, age, country of birth, offence type, early convictions, sequence of judicial
contacts
, ○ Provides prediction of recidivism (good predictive validity, AUC = .80)
○ No interview need (all info usually present in file)
● Static99-R
○ Status risk assessment tool for sex offenders
○ 10 items only
○ More specific than StatRex
○ Coding Form
■ E.g. Lower risk with increasing age
○ Translating scores into risk categories
■ Area under the Curve AUC
● Derived from Receiver Operating Characteristic (ROC)
○ Illustrates diagnostic ability
○ Plotting true positive rates (sensitivity) against false positive rates (specificity)
● Extent to which risk assessment instruments accurately differentiate recidivists from
non-recidivists
● Varies between 0 and 1
○ 0: chance performance
○ 1: perfect performance
○ 0.5: uninformative classifier
○ Dynamic Risk Factors
■ Changeable and therefore trestable
■ Interventions aim at change of dynamic risk factors
● Causal status
■ Also called criminogenic dynamic needs
■ Relationship of dynamic risk factors and risk of offending moderated by static risk factors
■ RNR model as theoretical framework
● Dynamic needs
● Risk principle
○ Offenders at high risk for reoffending will benefit most from more intensive
treatment
● Need principle
○ Only those factors associated with reductions in recidivism should be targeted
during treatment
● Responsitivity principle
○ Interventions should be matched to offender characteristics (e.g. level of
motivation, personal circumstances, learning style)
,■ RNR & dynamic rf
● Central eight criminogenic needs (each can be focus for treatment)
○ Antisocial attitudes / orientation
○ Antisocial peers
○ Antisocial personality
○ Antisocial behavior patterns
○ Absence of prosocial leisure / recreation
activities
○ Dysfunctional family
○ Employment issues
○ Substance abuse problems
● Non-criminogenic needs (general recidivism)
○ Personal distress
○ Major mental disorder
○ Low self-esteem
○ Low physical activity
○ Poor physical living conditions
○ Low conventional ambition
○ Insufficient fear of official punishment
■ LS/CMI
● Level of service / case management inventory
● Also based on RNR
● Interview, can be used for treatment
● 8 needs (section)
■ Types of Dynamic Risk Factors
● Stable Dynamic RF
○ Modifiable but unlikely to change
■ Changing maladaptive self-regulation happens slowly
○ Personal skill deficits and learned behaviors
○ Can be changed through process of effortful treatment
○ E.g. impulse control, poor attachment with others
, ● Acute Dynamic RF
○ Modifiable and likely to change
○ Highly transient conditions that only last hours or days
○ Rapidly changing because of environmental triggers or intrapersonal stresses
related to re-offense
○ E.g. intoxication, negative mood
■ Example: Stable-2000/2007
● Stable dynamic risk factors for sexual reoffending represented by following domains
○ Sexual self-regulation
○ General self-regulation
○ Intimacy deficits
○ Compliance and understanding for need of treatment and control
○ Existence supportive significant others
○ Distorted attitudes or attitudes tolerant to sexual violence
■ Example: ACUTE-2007
● Acute changing risk factors related to sexual recidivism
○ E.g. victim access, hostility, sexual preoccupation, rejection of supervision
● Factors for general recidivism
○ E.g. emotional collapse, collapse of social supports, substance abuse
■ Example: DRAOR
● Dynamic Risk Assessment for Offender Re-Entry
● Recidivism risk in community, outpatient
● Inform case planning, risk management
● Stable + acute dynamic rf + protective factors
● Guide for treatment & intervention
● Prediction of
○ Reconvictions
○ Reimprisonment
○ Protective subscale predicted breaches of parole
○ Dynamic risk factors, dual status
■ Predictors of recidivism
■ Potential causes of recidivism from explanatory point of view and as predictions
■ Cause-effect hypothesis
The Cause-Effect Hypothesis
● Assumptions
○ Evidence for association between risk factors and outcome, namely reoffending
○ There must be causal relationship in opposite direction
■ Cause must precede occurrence of effect
○ Cause is different from consequence (distinguish!)
○ Do there exist other inter-correlated factors for an observed association?
■ Can they explain that relation?
○ When there is a cause between Risk Factor A and outcome B, mechanism must be explained
● Dynamic risk factors as causes?
○ They are composite constructs
■ Each factor consists of several sub-factors and / or single factors
■ Example: central eight
● Which underlying factors, different levels, different weights
■ Some risk factors are social factors, mental-state factors, biological factors
○ Lack specificity
■ Risk factor A -> causes -> reoffending
■ Very often: lack of explanatory theories (reason why it causes reoffending)
■ E.g. Ward and Fortune (2016)
● Emotional congruence can be caused by other causes
○ Avoiding adults because of fear and anxiety of adults
○ Consider children as sources of reward and sexual pleasure less in foreground
○ World is dangerous in general, children more innocent
○ Individuals deeply insecure and feel emotionally more safe with children
● -> lack of specificity as multiple explanations for causal relationship / causal alternatives
○ Grain problem
■ Dynamic rf often presented at multiple levels ranging from general domain to more concrete level
■ “How to decide on appropriate level of abstraction when using dynamic rf in explanatory context?
■ Example: three levels
● Items level
● Personality disorder level cluster B
● Domain level
Lecture 1 - Introduction
Definition and Factors in Risk Assessment
● Definition
○ Older version
■ “Probability calculation that a harmful behavior or event will occur, and involves an assessment about
the frequency of the behavior (or event), its likely impact and who it will affect”
○ More recent version
■ “The attempt to predict the likelihood of future offending in order to identify individuals in need of
intervention”
○ Risk assessment is always talking about future, uses past to do so
○ Risk Assessment must specify
■ Behavior (e.g. violent behavior)
■ Potential damage or harm caused by the behavior (e.g. child sexual abuse)
■ Probability that it will occur, and under what circumstances
● Types of Factors
○ Protective factors
■ Decreasing likelihood of criminal behavior
■ Conceptualization
● Prosocial means or strategies to achieve one or more primary human goods (common life
goals)
○ We all have life goals but use different strategies to achieve those
■ Good Lives Model (GLM) of rehabilitation
● Accomplish a ‘good life’ by fulfilling primary goods
● Offending = maladaptive attempt to attain one or more primary human goods (life goals)
■ Interaction with risk factors, serve as buffer
■ Improve predictive accuracy
■ Potentially more engaging and effective interventions
■ Improvement protective factors -> (non) recidivism
■ Measurement
● SAPROF
○ Structured Assessment of Protective Factors for violence risk
○ Different versions
■ Juveniles and young adults
■ SAPROF-Sexual Offending
■ SAPROF-Intellectual Disabilities
■ SAPROF-Intensive care
○ Risk factors
■ Static
■ Dynamic
○ Static Risk Factors
■ Not changeable / treatable, often historical factors
● E.g. age, gender, ethnicity, first offender, type of offense, first conviction
■ Risk remains the same over years, cannot be used for interventions
● So provide important information but cannot be used in handling of offenders
■ Not used in forensic psychology, more in policy making
■ Assessment
● StatRec
○ Static Risk of Recidivism
○ Gender, age, country of birth, offence type, early convictions, sequence of judicial
contacts
, ○ Provides prediction of recidivism (good predictive validity, AUC = .80)
○ No interview need (all info usually present in file)
● Static99-R
○ Status risk assessment tool for sex offenders
○ 10 items only
○ More specific than StatRex
○ Coding Form
■ E.g. Lower risk with increasing age
○ Translating scores into risk categories
■ Area under the Curve AUC
● Derived from Receiver Operating Characteristic (ROC)
○ Illustrates diagnostic ability
○ Plotting true positive rates (sensitivity) against false positive rates (specificity)
● Extent to which risk assessment instruments accurately differentiate recidivists from
non-recidivists
● Varies between 0 and 1
○ 0: chance performance
○ 1: perfect performance
○ 0.5: uninformative classifier
○ Dynamic Risk Factors
■ Changeable and therefore trestable
■ Interventions aim at change of dynamic risk factors
● Causal status
■ Also called criminogenic dynamic needs
■ Relationship of dynamic risk factors and risk of offending moderated by static risk factors
■ RNR model as theoretical framework
● Dynamic needs
● Risk principle
○ Offenders at high risk for reoffending will benefit most from more intensive
treatment
● Need principle
○ Only those factors associated with reductions in recidivism should be targeted
during treatment
● Responsitivity principle
○ Interventions should be matched to offender characteristics (e.g. level of
motivation, personal circumstances, learning style)
,■ RNR & dynamic rf
● Central eight criminogenic needs (each can be focus for treatment)
○ Antisocial attitudes / orientation
○ Antisocial peers
○ Antisocial personality
○ Antisocial behavior patterns
○ Absence of prosocial leisure / recreation
activities
○ Dysfunctional family
○ Employment issues
○ Substance abuse problems
● Non-criminogenic needs (general recidivism)
○ Personal distress
○ Major mental disorder
○ Low self-esteem
○ Low physical activity
○ Poor physical living conditions
○ Low conventional ambition
○ Insufficient fear of official punishment
■ LS/CMI
● Level of service / case management inventory
● Also based on RNR
● Interview, can be used for treatment
● 8 needs (section)
■ Types of Dynamic Risk Factors
● Stable Dynamic RF
○ Modifiable but unlikely to change
■ Changing maladaptive self-regulation happens slowly
○ Personal skill deficits and learned behaviors
○ Can be changed through process of effortful treatment
○ E.g. impulse control, poor attachment with others
, ● Acute Dynamic RF
○ Modifiable and likely to change
○ Highly transient conditions that only last hours or days
○ Rapidly changing because of environmental triggers or intrapersonal stresses
related to re-offense
○ E.g. intoxication, negative mood
■ Example: Stable-2000/2007
● Stable dynamic risk factors for sexual reoffending represented by following domains
○ Sexual self-regulation
○ General self-regulation
○ Intimacy deficits
○ Compliance and understanding for need of treatment and control
○ Existence supportive significant others
○ Distorted attitudes or attitudes tolerant to sexual violence
■ Example: ACUTE-2007
● Acute changing risk factors related to sexual recidivism
○ E.g. victim access, hostility, sexual preoccupation, rejection of supervision
● Factors for general recidivism
○ E.g. emotional collapse, collapse of social supports, substance abuse
■ Example: DRAOR
● Dynamic Risk Assessment for Offender Re-Entry
● Recidivism risk in community, outpatient
● Inform case planning, risk management
● Stable + acute dynamic rf + protective factors
● Guide for treatment & intervention
● Prediction of
○ Reconvictions
○ Reimprisonment
○ Protective subscale predicted breaches of parole
○ Dynamic risk factors, dual status
■ Predictors of recidivism
■ Potential causes of recidivism from explanatory point of view and as predictions
■ Cause-effect hypothesis
The Cause-Effect Hypothesis
● Assumptions
○ Evidence for association between risk factors and outcome, namely reoffending
○ There must be causal relationship in opposite direction
■ Cause must precede occurrence of effect
○ Cause is different from consequence (distinguish!)
○ Do there exist other inter-correlated factors for an observed association?
■ Can they explain that relation?
○ When there is a cause between Risk Factor A and outcome B, mechanism must be explained
● Dynamic risk factors as causes?
○ They are composite constructs
■ Each factor consists of several sub-factors and / or single factors
■ Example: central eight
● Which underlying factors, different levels, different weights
■ Some risk factors are social factors, mental-state factors, biological factors
○ Lack specificity
■ Risk factor A -> causes -> reoffending
■ Very often: lack of explanatory theories (reason why it causes reoffending)
■ E.g. Ward and Fortune (2016)
● Emotional congruence can be caused by other causes
○ Avoiding adults because of fear and anxiety of adults
○ Consider children as sources of reward and sexual pleasure less in foreground
○ World is dangerous in general, children more innocent
○ Individuals deeply insecure and feel emotionally more safe with children
● -> lack of specificity as multiple explanations for causal relationship / causal alternatives
○ Grain problem
■ Dynamic rf often presented at multiple levels ranging from general domain to more concrete level
■ “How to decide on appropriate level of abstraction when using dynamic rf in explanatory context?
■ Example: three levels
● Items level
● Personality disorder level cluster B
● Domain level
