360 likes | 618 Views
Scientific Basis for Expert Testimony about Violence Risk Assessment. 2. The Scientific Basis for Expert Testimony on Risk Assessment. Progress in the science of violence risk assessmentDefinitions of ?expert" for giving expert testimonyHow do contemporary methods of risk assessment measure up?Co
E N D
1. Scientific Basis for Expert Testimony about Violence Risk Assessment 1 The Scientific Basis for Expert Testimony about Risk Assessment Marnie E. Rice & Grant T. Harris
www.mhcp-research.com
June 2001
2. Scientific Basis for Expert Testimony about Violence Risk Assessment 2 The Scientific Basis for Expert Testimony on Risk Assessment Progress in the science of violence risk assessment
Definitions of “expert” for giving expert testimony
How do contemporary methods of risk assessment measure up?
Conclusions & the future of expert testimony about violence risk
3. Scientific Basis for Expert Testimony about Violence Risk Assessment 3 Progress in the science of violence risk assessment Journal articles on prediction of violence or recidivism
1890- 1969 236
1970- 1979 379
1980- 1989 657
1990- 1999 1232 - Searched PsychInfo Journal Articles- Abstracts for the keywords
(predict* OR recidiv*) AND (violen* OR recidiv*) June 7, 2001- Searched PsychInfo Journal Articles- Abstracts for the keywords
(predict* OR recidiv*) AND (violen* OR recidiv*) June 7, 2001
4. Scientific Basis for Expert Testimony about Violence Risk Assessment 4 Journal articles on prediction of recidivism among sex offenders 1890-1969 8
1970-1979 13
1980-1989 44
1990-1999 117
5. Scientific Basis for Expert Testimony about Violence Risk Assessment 5 Prediction of Violence Before Mid-80’s Examples
Baxstrom (Steadman, 1973)
Quinsey & Ambtman, 1979
Pasewark, Bieber, Bosten, Kiser, & Steadman, 1982
Monahan (1981)
General consensus was (and studies backed that up) that there was no expertise in the prediction of violence
-Baxstrom decision in 1973 was very influential- Showed overprediction of violence- incidence of violence among men all of whom had been kept in because someone thought they were too dangerous to release - yet incidence of future crime and violence was very low-- (But short followup and men were quite old by the time they were released)
-Quinsey & Ambtman psychiatrist- teacher study
- Pasewark et al.- men who eloped were less likely to reoffend than those released (difference was ns.)
General consensus was (and studies backed that up) that there was no expertise in the prediction of violence
-Baxstrom decision in 1973 was very influential- Showed overprediction of violence- incidence of violence among men all of whom had been kept in because someone thought they were too dangerous to release - yet incidence of future crime and violence was very low-- (But short followup and men were quite old by the time they were released)
-Quinsey & Ambtman psychiatrist- teacher study
- Pasewark et al.- men who eloped were less likely to reoffend than those released (difference was ns.)
6. Scientific Basis for Expert Testimony about Violence Risk Assessment 6 Predictors of violent recidivism among mentally disordered offendersHarris, Rice, & Quinsey, 1993 Big predictors
Psychopathy, elementary school maladjustment, age*, personality disorder
Medium predictors
Separated from parents < 16, Failure on prior conditional release,criminal history,never married
Small predictors
schizophrenia*, victim injury, alcohol abuse, male victim N=618N=618
7. Scientific Basis for Expert Testimony about Violence Risk Assessment 7 Predictors of Violent Recidivism Among MDOs (Bonta, Law & Hanson, 1998 meta-analysis) Big predictors
objective risk assessment, antisocial personality, violent history, juvenile delinquency, age*
Medium predictors
nonviolent criminal history, adult criminal history, substance abuse, marital status,
Small predictors
clinical judgement, psychosis*, offense seriousness
8. Scientific Basis for Expert Testimony about Violence Risk Assessment 8 Predictors of Violence Among Psychiatric Patients (MacArthur Risk Study) Big predictors:
Psychopathy (PCL:SV); Adult arrests, Antisocial personality disorder, Major mental disorder without substance abuse*, drug or alcohol abuse, anger (Novaco)
Moderate predictors:
Violent arrests, schizophrenia*, child abuse, threat-control overrride symptoms* delusions at the time of admission* Big= Bivariate correlation >= .16
Moderate= Bivariate correlation > .09 < .16
Small = Bivariate correlation <.09 and significant
Nonpredictor= nonsignificant (approx. < .06)
Note: I’ve listed highest correlation for measures that had many versions (e.g. Novaco Anger Inventory- Behavior, Cognitive, Arousal)
There were many other correlations, I’ve just listed ones most relevant for this talkBig= Bivariate correlation >= .16
Moderate= Bivariate correlation > .09 < .16
Small = Bivariate correlation <.09 and significant
Nonpredictor= nonsignificant (approx. < .06)
Note: I’ve listed highest correlation for measures that had many versions (e.g. Novaco Anger Inventory- Behavior, Cognitive, Arousal)
There were many other correlations, I’ve just listed ones most relevant for this talk
9. Scientific Basis for Expert Testimony about Violence Risk Assessment 9 MacArthur Data (cont) Small predictors
persecutory delusions*, male, BPRS hostility, BPRS thought disturbance, medication nonadherence on admission*
10. Scientific Basis for Expert Testimony about Violence Risk Assessment 10 Nonpredictors of Violent Recidivism From Bonta et al., Hanson et al.
Psychological distress, remorse, treatment history, insight
From MacArthur study:
mania, depression, any delusions, hallucinations, command hallucinations, grandiose delusions, GAF, BPRS Total
11. Scientific Basis for Expert Testimony about Violence Risk Assessment 11 Contemporary methods of combining predictors to assess risk Clinical Opinion (CO)
Blended Clinical (BC)
variables derived from empirical and clinical literature (e.g. HCR-20)
Initial Actuarial (IA)
derived from actual outcome (e.g. MacArthur ICT)
12. Scientific Basis for Expert Testimony about Violence Risk Assessment 12 Contemporary methods of combining predictors to assess risk (cont.)
Established Actuarial (EA)
Derived from actual outcome + cross-validation (e.g. VRAG)
Adjusted Actuarial (AA)
use clinical judgement to adjust ? or ? (e.g., VPS, Webster, Harris, Rice, Cormier, & Quinsey, 1994)
13. Scientific Basis for Expert Testimony about Violence Risk Assessment 13 Blended Clinical Opinion (BC) HCR-20 Historical- Previous violence, Young age at first violence, Marital instability, Employment problems, Substance abuse, Mental illness, Psychopathy, Childhood maladjustment, Personality disorder, Prior failure
Clinical-Lacks insight, Negative attitudes, Active psychosis, Impulsivity, Unresponsive to Tx
Risk Management: Unfeasible plans, Destabilizers, No personal support, Noncompliance with remediation, Stress
Manual: Do Not Total for Clinical Purposes
Two follow-up studies using totals
14. Scientific Basis for Expert Testimony about Violence Risk Assessment 14 Initial Actuarial (IA) (MacArthur ICT) Psychopathy, Prior arrests, Child abuse, Recent violence, Substance abuse, Status, Symptoms*, Employed, Father arrested, suicidal*, Head injury
Violent fantasies, Schizophrenia*, Hostility, Young age, Low functioning, Property arrest, Perceived coercion, Threatening, Negative relationships
Classification Tree and Iterations
Construction but no Replications
15. Scientific Basis for Expert Testimony about Violence Risk Assessment 15 Established Actuarial (EA) VRAG 12 Items: Psychopathy Checklist, ES Maladjustment, DSM-III Personality disorder, Young age, Separation from parents under 16, Prior conditional failure, Criminal Hx score, Never married, DSM-III schizophrenia*, Victim injury, Alcohol abuse score, Male victim
Development 1993; Cross-validation 1997
Over 20 Replications; 6 and 16
Peer Review: 15 journal articles
Reliability, Accuracy, Error Rates -> ROC
16. Scientific Basis for Expert Testimony about Violence Risk Assessment 16 Other Available Instruments None recommended for predicting violence in forensic patients
BC: SARA, SVR-20, EARL-20B, SAC-J
IA: Static-99, Mn-SOST-R
EA: RRASOR
17. Scientific Basis for Expert Testimony about Violence Risk Assessment 17 Definitions of “expert” for giving expert testimony Legal Definitions
U.S.
Frye
Daubert
Canada
Mohan
Scientific Definitions
Einhorn, Faust, Ziskin, & Dawes
18. Scientific Basis for Expert Testimony about Violence Risk Assessment 18 Legal Definitions in theUnited States Frye
Assist trier of fact
Witness qualified by virtue of knowledge, skill, experience, training, or education
General acceptance by respective field
19. Scientific Basis for Expert Testimony about Violence Risk Assessment 19 Legal Definitions in theUnited States Daubert
Has or can be tested
Peer review and publication
Known error rates
Nonjudicial uses
General acceptance in scientific community
20. Scientific Basis for Expert Testimony about Violence Risk Assessment 20 Legal Definitions in Canada Mohan
Relevant to decision before trier
Necessity in assisting trier of fact
No exclusionary rule
Properly qualified expert (specific)
21. Scientific Basis for Expert Testimony about Violence Risk Assessment 21 Scientific DefinitionsEinhorn, Faust, Ziskin, & Dawes Relevance
Different from layperson (and clinical opinion)
Reliable
Valid (more accurate than layperson)
Specialized
Replication and peer review
22. Scientific Basis for Expert Testimony about Violence Risk Assessment 22 Clinical vs. actuarial Grove & Meehl, 1996 “Every day thousands of predictions are made by parole boards..., psychiatric teams, and juries hearing civil and criminal cases.… Freedom of convicted felons or risk to future victims, millions of taxpayer dollars expended by court services, hundreds of millions involved in individual and class action lawsuits...-- these are high stakes indeed. To use the less efficient of two prediction procedures in dealing with such matters is not only unscientific and irrational, it is unethical.” ( p. 320)
23. Scientific Basis for Expert Testimony about Violence Risk Assessment 23 Actuarial vs. clinical prediction -- Monahan 2001 In commenting on the designing of the MacArthur risk assessment study in the late ‘80’s : “More research demonstrating that the outcome of unstructured clinical assessments left a great deal to be desired seemed to be overkill: that horse was already dead.” (Monahan et al. in press)
25. Performance of the VRAG
27. Relative Operating Characteristic
28. Relative Operating Characteristic
29. Replications of VRAG/SORAG (n=24)
30. Scientific Basis for Expert Testimony about Violence Risk Assessment 30 How do contemporary methods of risk assessment measure up? Assumptions
Frye
Witness qualified by virtue of knowledge, skill, experience, training, or education
General acceptance by respective field
31. Scientific Basis for Expert Testimony about Violence Risk Assessment 31 How do contemporary methods of risk assessment measure up? Mohan
Necessity in assisting trier of fact
No exclusionary rule
Properly qualified expert (specific)
32. Scientific Basis for Expert Testimony about Violence Risk Assessment 32 How do contemporary methods of risk assessment measure up? Daubert
Has or can be tested
Peer review publication
Known error rates
Nonjudicial uses
General acceptance in scientific community
33. Scientific Basis for Expert Testimony about Violence Risk Assessment 33 How do contemporary methods of risk assessment measure up? “Scientific” Standard
Different from layperson (CO)
Reliable
More accurate than layperson
Specialized
Replication and peer review
34. Scientific Basis for Expert Testimony about Violence Risk Assessment 34 Clinical decisions using only actuarial instruments Considerable expertise
Different (and more accurate) judgments than other “experts” (and laypersons)
More reliable than laypersons
Use special instruments- diagnosis, psychopathy , VRAG, SORAG, MnSOST-R
35. Scientific Basis for Expert Testimony about Violence Risk Assessment 35 The Scientific Basis for Expert Testimony About Violence Risk Assessment: Conclusions By most standards, CO not expert
By almost any standards, EA expert
EA and the VRAG:
Not perfectly accurate (à la DNA)
“Dynamic” variables
Absolute rates
Unstudied populations
Universal Acceptance Illusive
36. Scientific Basis for Expert Testimony about Violence Risk Assessment 36 The Future Hilton & Simmons, in press;
Rice & Harris, 2001
Legal standards will continue to become more sophisticated wrt scientific evidence
Actuarial instruments to predict violence will continue to improve
Court battles will increasingly be “a battle of actuarial instruments”