Assessing Psychopathy_ An Overview of the Hare Scales

Document Sample
Assessing Psychopathy_ An Overview of the Hare Scales Powered By Docstoc
					Risk Assessment
Possibilities and Impossibilities

Stephen D. Hart
Simon Fraser University

Violence Risk Assessment
Evaluations of people to:
Characterize the risk they will commit violence in the future Develop interventions to manage risk

The clinical task is to:
Understand how and why people chose to act violently in the past Determine whether these or other factors may lead them to make similar choices again

Goals of Risk Assessment
Prevent violence

More specifically...
Guide intervention Improve consistency of decisions Improve transparency of decisions
Protect clients’ rights Liability management

Which Risk Factors Should We Assess?

Identifying Risk Factors
Three primary criteria
Empirical (predictive accuracy) Professional (practical utility) Legal (fairness and reasonableness)

Problems: Empirical Criterion
Not everything that is important has been proven or validated scientifically
Can lead to exclusion of “good” but rare or difficult-to-assess risk factors

Prediction  cause, explanation, or intervention
Can lead to inclusion of “bad” but common or easy-to-assess factors

Example: The SIEVE
Age Sex Facial hair Foot size

Young is bad Male is bad Dense is bad Big is bad

Problems: Professional Criterion
Focus on dynamic factors may bias risk assessments
Can lead to exclusion of “good” but static or easy-to-ignore factors

Conventional wisdom of professionals may be plain wrong
Can lead to inclusion of “bad” but vivid or dramatic factors

Example: Clinical Intuition
Depression Anxiety Intelligence Rorschach

Present is good Present is good High is good Seeing viscera is bad

Problems: Legal Criterion
Useful for excluding risk factors, but not for including them It can be argued that almost any risk factor is unfair or unreasonable in some respect

How Should We Conduct Risk Assessments?

Methods of Risk Assessment
Unstructured professional judgement Anamnestic assessment Structure professional judgement

Psychological tests Actuarial tests

Unstructured: Features
No constraints on evaluation
Any information can be considered Information can be gathered in any manner

No constraints on decisions
Information can be weighted and combined in any manner Results can be communicated in any manner

Unstructured: Limitations
No systematic empirical support
Low agreement (unreliable) Low accuracy (unvalidated) Foundation is unclear (unimpeachable)

Relies on charismatic authority Decisions are too general Focus is not on action

Anamnestic: Features
Imposes some structure on evaluation
Must consider, at a minimum, nature and context of past violence

Logically related to development of risk management strategies Consistent with “relapse prevention” or “harm reduction” approaches

Anamnestic: Limitations
Unknown reliability Unknown validity Assumes that history will repeat itself
Violent careers are static Violent people are specialists

Structured Judgement
Designed to prevent an outcome

Imposes major structure on evaluation
Must consider, at a minimum, a fixed and explicit set of risk factors Specifies process for information-gathering

Imposes minor structure on decision
Specifies language for communicating findings

HCR-20, version 2
Webster, Douglas, Eaves, & Hart (1997) Designed to assess risk for violence in those with mental or personality disorders 10 Historical, 5 Clinical, and 5 Risk Management factors

Previous violence Young age at first violence Relationship instability Employment problems Substance use problems Major mental illness Psychopathy Early maladjustment Personality disorder Prior supervision failure

Clinical/Risk Management
Lack of insight Negative attitudes Active symptoms of major mental illness Impulsivity Unresponsive to treatment Plans lack feasibility Exposure to destabilizers Lack of personal support Noncompliance with remediation attempts Stress

Structured: Limitations
Requires “retooling” of evaluation process
Systematized information-gathering New training and technology

Justification for use requires induction
What works elsewhere and with other people will here and with this person

Assumes professionals can exercise discretion appropriately

Psychological Tests: Features
Measure some disposition that predicts violence, according to past research Reliability and validity of test-based decisions has been evaluated Imposes major structure
On some part of the evaluation process On some part of the decision-making process

Psychological Tests: Limitations
Require professional judgment
Which tests to use How to interpret scores

Justification of use requires induction
What works elsewhere and with other people will here and with this person

Symptom construct rating scale
Requires clinical/expert judgment Based on “all data”

Data obtained from two primary sources:
Review of case history (required) Interview/observation (recommended)

Part 1 Superficial Grandiose Deceitful Lacks remorse Lacks empathy Doesn’t accept responsibility Part 2 Impulsive Poor behavioral controls Lacks goals Irresponsible Adolescent antisocial behavior Adult antisocial behavior

Risk Scales: Features
Designed to predict an outcome High-fidelity
Optimized for specific outcome, time period, population, and context

Impose rigid structure
On all of the evaluation process On all of the decision-making process

Risk Scales: Limitations
Still require professional judgment
Which scales to use How to interpret scores

Justification of use requires induction
What works elsewhere and with other people will here and with this person

Results may be easily misinterpreted
Pseudo-objective, pseudo-precise

Violence Risk Appraisal Guide
Quinsey et al. (1998) Constructed in adult male patients assessed or treated at a maximum security hospital 12 items weighted according to ability to postdict violence over 7 year follow-up Total scores divided into 9 bins, with estimated p(violence) from 0% to 100%

VRAG Items
PCL-R score Elem. school problems Personality disorder Age (—) Separated from parents under age 16 Failure on prior conditional release Nonviolent offense history Never married Schizophrenia (—) Victim injury (—) Alcohol abuse Female victim (—)

How Accurately Do We Assess Risk?

A Complex Phenomenon
Studying the accuracy of risk characterizations is difficult due to the complexity of…
The characterizations The violence The follow-up Indexes of accuracy

The Violence
Nature: Victims (stranger vs. acquaintances); motivations (instrumental vs. reactive), context (co-factors) Severity: Physical or psychological harm (threats vs. battery vs. homicide) Imminence: Timing (sooner vs. later) Frequency: Number of events (single vs. multiple)

The Characterizations
Evaluator: Professional vs. researcher; novice versus expert Process: Clinical vs. actuarial, contextual vs. context-free Timing: Admission vs. discharge, static vs. dynamic Metric: Uni- vs. multi-dimensional, categorical vs. continuous

The Follow-Up
Data source: Patient vs. collaterals vs. records Time at risk: Weeks vs. months vs. years Interventions: Dynamic factors, life events Monitoring: Continuous vs. endpoint

Indexes of Accuracy
Comparison group: Other patients vs. normals Statistic: Uni- vs. multi-variate, time Weighting of errors: Equal vs. differential Interpretation: Chance vs. status quo vs. perfection

Science Responds to Complexity
Prediction Low Risk High Risk Outcome Not Violent Violent

 

 

Example: Psychopathy
Prediction Low Risk

Outcome Not Violent Violent 90 12 24 40

High Risk

Harris, Rice, & Cormier (1991)



In this study...
Accuracy of positive predictions is 77% Accuracy of negative predictions is 79% Overall accuracy is 78% Chance-corrected agreement is 53% Correlation is .53 Odds ratio is 12.5



So, how did we do?
Relative to chance: Great! Relative to perfection: Awful! Relative to the status quo: ???

What is the status quo?
Predictions of violence using the PCL-R typically have an effect size (r) of about .25-.35; the average effect size for psychosis is about .20-.30 An effect size of .40 may be the “forensic sound barrier” But what is the status quo in other human endeavors?

Lipsey & Wilson (1993) reviewed 302 meta-analyses Determined typical effect sizes for psychological, educational, and medical interventions

Psychological Interventions
CBT —  depression Psychotherapy — any  Correctional programs (youths) — any  Diversion (youths) —  recidivism Correctional treatment (adults) — any  .44 .39 .23 .20 .12

Educational Interventions
Small classes —  class climate Tutoring —  grades Small classes —  grades Media campaigns —  seatbelt use .26 .20 .10 .06

Medical Interventions
Speech therapy —  stuttering Bypass surgery —  angina pain Cyclosporine —  organ rejection Bypass surgery —  mortality ASA —  heart attack .54 .37 .15 .07 .04

Violence Predictions in Context
Speech therapy —  stuttering .54 CBT —  depression .44 Bypass surgery —  angina pain .37 Psychopathy —  violence .25-.35 Psychosis —  violence .20-.30 Small classes —  class climate .26

It is possible to assess violence risk in a reliable and valid manner
Risk factors assessed should reflect scientific, professional, and legal considerations Risk can — and should — be assessed in different ways Violence risk assessments are, on average, as good as most other prognostications

It is impossible to make specific predictions of future violence for a given individual with a high degree of scientific precision or certainty
We never know risk; we merely estimate it assuming certain contexts God doesn’t play dice

Contact Information
Stephen D. Hart, Ph.D. Department of Psychology Simon Fraser University Burnaby, British Columbia Canada V5A 1S6
Tel: 604.291.5485 / Fax: 604.291.3427 E-mail: URL:

Shared By: