Chapter 25: Violence Risk Assessment - Research and Practice

Violence Risk Assessment

• how psychologists assess the risk of future violence

  • to determine amount of supervision needed in prison and after release

    • considering rights and liberties of public and offender

The Evolution of Violence Risk Assessment

• unstructured clinical judgement → actuarial risk assessment → structured professional judgement

• can use any of them but most use structured professional judgement (this works the best)

Unstructured Clincal Judgement

• clinicians rely on

  • professional experience

  • intuition

  • case-specific information

• strengths

  • flexible

  • individualized

  • inexpensive

  • able to consider things that may not be included in a more strucutred assessment

• limitations

  • subjective

  • vulnerable to cognitive bias

  • lower inter-evaluator reliability and predictive validity

Actuarial Risk Assessment

• assessment

  • use empirically supported risk factors

  • factors are scored using a fixed formula

  • produces a probability or risk category

• strengths

  • standardized

  • objective

  • transparent

  • high inter rater reliability

• limitations

  • may rely heavily on statistic factors (factors that don’t change over time)

    • age at first offence or previous offences

    • vs. dynamic factors that change like mental states, medication, treatment adherence

  • may not generalize across populations

  • less individualized

  • limited usefulness for treatment planning

• adjusted actuarial assessment — starts with score but can adjust with factors not included in actuarial tool (clinical override)

  • may reintroduce bias

Structured Professional Judgement

• includes…

  • static factors

  • dynamic factors

  • sometimes protective factors

• combines…

  • research-based risk factors

  • professional judgement

  • case-specific information

• goal:

  • not just to predict violence, but to help prevent and manage it

• risk can change depending on circumstances — not a permanent assessment of the person

  • focuses on overall wellbeing of individual

    • why they did it

    • how can we help

    • how can we make sure it doesn’t happen again

Risk Factors that Increase Likelihood of Violence

• previous violence

• substance use problems

• relationship instability

• treatment noncompliance

• impulsivity

Protective Factors that can Reduce or Buffer Risk

• social support

• treatment engagement stability

• personal strengths

Base Rate

• how common a behaviour is in a specific population over a specific period

Predictive Accuracy

• how well a tool distinguishes between those who reoffend and those who do not

• key accuracy concepts:

  • sensitivity

  • specificity

  • positive predictive value

  • negative predictive value

  • area under the curve

Predictive Accuracy — Sensitivity

• correctly identifying those who reoffend

Predictive Accuracy — Specificity

• correctly identifying those who do not

Predictive Accuracy — Positive Predictive Value (PPV)

• probability that someone assessed as high risk actually reoffends

Predictive Accuracy — Negative Predictive Value (NPV)

• probability that someone assessed as low risk does not reoffend

Predictive Accuracy — Area Under the Curve (AUC)

• how well a tool distinguishes recidivists from non-recidivists

Limitations of Violence Risk Assessment

• structured assessment is better than unstructured judgement, but violence risk can never be predicted with certainty

  • no tool predicts violence with certainty

  • results may not generalize across populations

  • cognitive bias can still influence assessments

  • risk estimates can be misunderstood or framed differently

  • group-level findings do not perfectly predict individual behaviour