PRESENTATION - Risk Assessment
What is Risk Assessment?
Defining Risk Factors
Risk Assessments in Civil Settings
civil commitment
child protection
risk assessment whena chuld is abused; risk if you put the child back in the family?
immigration laws
school and labour regulations
if smn wants to work, what’s the risk they’ll offend
duty to warn
it is your repsnibility to warn smn that an individual is high risk… you must notify the relevant ppl (like the schoolboard if theyre trynna be a teacher) and the police
limits of confidentiality
when ur doing risk assessment, there is no confidentiality; which has to be explaiend to the person on which your doing the risk asessment… the reason for this is because it’s not therapeutic; you’re trying to give your report of their risk to someone like a judge
Risk Assessments in Criminal Settings
Predicting Risk
Predictions: Decisions Versus Outcomes
Base Rates
critical for creating a valid instrument… no instrument is valid without taking into account base rates.
base rates = how often does the crime occur generally?
if everyone commits that crime, then you can’t do the risk assessment because the base rate is way too high…
if no one ever does it, the probability that that person will ever do it is zero.
risk assessments work best for crimes where the base rates are in the middle range
base rates for men are pretty good… most violent behaviour amongst men has a moderate base rate: so we can make good predictions
for women, violent criminal behaviour is exceptionally low; so risk assessments for women are almost useless
Judgement Error and Biases
heuristics
we don,t know why it works, but we beleive it works;
freud after seeing a bunch of ppl with sexual problems developed a theory for it… it doesn’t work anymore because most problems now don,t revolve around sexuality
illusory correlation
just because ppl in gangs often get involved in crimes, you assume they’re more likely to offend…
but some gangs are pro-social,
ignoring base rates
reliance on salient or unique cues
detectives sometimes get in their own head; ‘because he turned his eyes away, he must be guilty’
overconfidence in judgements
risk assessment is based on judgement; so if you always think you’re right, you ignore the possibility you’re wrong… Hirsch would send his risk assessment to a team to get their opinions
Making Judgements and Doing the Profiling
Unstructured Clinical Judgement
this is the worst form of risk assessments (no rules/guidelines… this is the kind of profiling that was done before scientific profiling was developed… like the profile of hitler)
decisions characterized by professional discretion and lack of guidelines
subjective
no specific risk factors
no rules about how risk decisions should be made
Actuarial Prediction
Gold standard
VRAG is preferred here
Decisions based on risk factors that are selected
and combined based on empirical or statistical
evidence
Most actuarial risk instruments include only static
risk factors
Evidence favours actuarial assessments over
unstructured clinical judgment
only measures static assessment (what the person is like now and what their risk of offending is based on their current profile)
Structured Professional Judgement
usually done by a team
HCR20 is preferred here
Decisions guided by predetermined list of risk
factors derived from research literature
Judgment of risk level is based on professional
judgement
Diverse group of professionals (psychologists and/or criminologists)
this has a tendency to measure future things a lot better (what is the risk this person will feel frustrated and lash out?)
usually, you’d do actuarial prediction and structured professional judgement… and then you look at how well they coincide
Types of Predictors
Static Risk Factors
Historical
Factors that cannot be changed
where they grew up, childhood,
Dynamic Risk Factors
Fluctuate over time
Factors that can be changed
who their friends are, level of education, where they live
Acute vs. stable dynamic risk factors
Important risk factors
Dispositional Risk Factors
These consist of demographic characteristics:
Age
Gender
And personality traits:
Impulsivity
Psychopathy
Historical Risk Factors
Clinical Risk Factors
Contextual Risk Factors
Female Offenders and Risk
More similarities in risk factors for men and women
than differences
• Gender-specific risk factors
– History of self-injury
– Poor self-esteem
Gender differences in criminality
Women engage in less crime
Women reoffend at lower rates
Childhood victimization more prevalent
Mental disorders more prevalent
usually crime committed by women is relational (i.e., they’re doing the crime on behalf of someone else, usually a partner)
example: fella in toronto who raped and killed 2 young women, and his partner helped him kill them and she did it because she didn’t want to lose him
when they get arrested, its usually prostitution and drug use, not violent crime.
Fundamental Definitions and the Nature of Risk
weather forecasting within 7-day period is 80% accurate; it’s similar to risk assessment (e.g., determining what the risk is that someone will be violent in the future)
VRAG: very good accuracy over 10 years… it’s still a guessing game involving your judgement…
Risk Factor: This is defined as a measurable feature of an individual that serves to predict the behavior of interest, such as criminal behavior or violence.
Risk as a Continuous Range: According to Steadman (), risk should not be viewed as a binary (present/absent) but rather as a range.
probabilities change across time
interaction among offender characteristics and situation
Temporal Fluidity: Risk levels and probabilities are not static; they change across time.
Interactionist Perspective: Risk is the result of a complex interaction between the specific characteristics of the offender and the situational context.
risk management:
risk in low-security prison vs halfway house vs high-security prison
how swriously do we have to minotr risk? depends on the invidiual
The Dual Framework of Risk Assessment
Risk assessment is comprised of distinct components:
Prediction: The process of determining the likelihood that an individual will engage in a specific behavior (e.g., violence) in the future.
Management: The development and implementation of strategies to mitigate identified risks and prevent the behavior from occurring.
Risk Assessment in Civil and Criminal Settings
Civil Settings:
Civil commitment (involuntary hospitalization).
Child protection (assessing risk of maltreatment or neglect).
Immigration laws (determining admissibility based on safety risks).
School and labor regulations.
Duty to Warn: Legal obligations of professionals to notify third parties of potential threats.
Limits of Confidentiality: Identifying when the need for public safety supersedes client privacy.
Criminal Settings:
Assessments are conducted at critical decision points including Pretrial, Sentencing, and Release (parole/probation).
Legal Priority: In these contexts, public safety interests outweigh attorney-client privilege.
The Logic of Prediction: Outcomes and Errors
Predictions are evaluated against actual outcomes using a four-quadrant matrix:
True Positive (Correct): The individual was predicted to reoffend and they did reoffend.
False Positive (Incorrect): The individual was predicted to reoffend but they did not reoffend.
False Negative (Incorrect): The individual was predicted to not reoffend but they did reoffend.
True Negative (Correct): The individual was predicted to not reoffend and they did not reoffend.
The Role of Base Rates in Prediction
Base Rate Definition: Represents the percentage () of people within a specific population who commit a criminal or violent act.
Statistical Difficulty: Prediction becomes significantly more difficult when base rates are extremely high or extremely low.
Low Base Rate Errors: When an event is rare (low base rate), there is a higher tendency for assessments to produce False Positives.
Frequency Correlation: It is fundamentally easier to predict frequent events than infrequent ones.
Cognitive Biases and Judgment Errors
Several psychological factors can impair the accuracy of risk assessments:
Heuristics: Cognitive shortcuts that can lead to systematic errors.
Illusory Correlation: The false belief that two variables are related when they are not.
Ignoring Base Rates: Failing to account for the natural frequency of an event in the population.
Salience Bias: Over-reliance on unique or highly memorable cues rather than statistically significant ones.
Overconfidence: Excessive certainty in the accuracy of one's professional judgments.
Methodological Approaches to Risk Assessment
Unstructured Clinical Judgment:
Characterized by professional discretion and a lack of specific guidelines.
The process is subjective.
Lacks specific risk factors or standardized rules for making decisions.
Actuarial Prediction:
Decisions are based on risk factors selected and combined through empirical or statistical evidence.
Most actuarial instruments focus primarily on static risk factors.
Empirical Superiority: Current evidence favors actuarial assessments over unstructured clinical judgment for accuracy.
Structured Professional Judgment (SPJ):
Decisions are guided by a predetermined list of risk factors derived from research literature.
The final judgment of the risk level is based on the professional's clinical judgment within that structured framework.
Typically involves a diverse group of professionals.
Taxonomy of Risk Predictors
Static Risk Factors:
Historical in nature.
Factors that are fixed and cannot be changed (e.g., past criminal history).
Dynamic Risk Factors:
Factors that fluctuate over time and are amenable to change/intervention.
Stable Dynamic Factors: Change slowly over months or years (e.g., attitudes, long-term substance use).
Acute Dynamic Factors: Change rapidly over days or hours (e.g., intoxication, sudden mood shifts).
Core Domains of Risk Factors
Dispositional Risk Factors:
Demographics: Age and Gender.
Personality characteristics: Impulsivity and Psychopathy.
Historical Risk Factors:
Records of past antisocial behavior.
Age of onset of antisocial behavior (earlier onset often predicts higher risk).
Personal history of childhood maltreatment.
History of past supervision failure, escape attempt, or maladjustment within an institution.
Clinical Risk Factors:
Substance use disorders.
Major mental disorders, specifically diagnoses of schizophrenia or affective disorders.
Threat/Control Override (TCO) Symptoms: Specific psychotic symptoms where an individual feels threatened or believes their internal controls are being overridden by outside forces.
Contextual Risk Factors:
Lack of social support for daily living.
Easy access to weapons.
Easy access to potential victims.
Gender Considerations in Risk Assessment
Similarities vs. Differences: There are generally more similarities in risk factors between men and women than there are differences.
Gender-Specific Risk Factors:
History of self-injury.
Poor self-esteem.
Gender Differences in Criminality:
Women generally engage in less crime overall.
Women reoffend at lower rates than men.
Childhood victimization is more prevalent among female offenders.
Mental disorders are more prevalent among female offenders.
Practical Clinical Exercise: Edmund Kemper
Instruments: The practice involves using the following tools:
PCL-R: Psychopathy Checklist-Revised.
VRAG: Violence Risk Appraisal Guide.
Context: Profiles and risk assessments are conducted on convicted felon Edmund Kemper to prepare students for Assignment .