The difficulties of predicting future violence Edward P. Mulvey, Ph.D. Western Psychiatric Institute...

40
The difficulties The difficulties of predicting of predicting future violence future violence Edward P. Mulvey, Ph.D. Edward P. Mulvey, Ph.D. Western Psychiatric Institute and Western Psychiatric Institute and Clinic Clinic University of Pittsburgh School of University of Pittsburgh School of Medicine Medicine [email protected] [email protected] Conference on Campus Violence Conference on Campus Violence Columbia Law School Columbia Law School April 4, 2008 April 4, 2008
  • date post

    19-Dec-2015
  • Category

    Documents

  • view

    218
  • download

    1

Transcript of The difficulties of predicting future violence Edward P. Mulvey, Ph.D. Western Psychiatric Institute...

The difficulties of The difficulties of predicting future predicting future

violenceviolence

Edward P. Mulvey, Ph.D.Edward P. Mulvey, Ph.D.Western Psychiatric Institute and Western Psychiatric Institute and

ClinicClinicUniversity of Pittsburgh School of University of Pittsburgh School of

[email protected]@upmc.edu

Conference on Campus ViolenceConference on Campus ViolenceColumbia Law SchoolColumbia Law School

April 4, 2008April 4, 2008

Goals Goals

Provide some background about Provide some background about general methods for making general methods for making predictions of violencepredictions of violence

Identify the inherent challenges to Identify the inherent challenges to predicting campus violencepredicting campus violence

Make some general recommendations Make some general recommendations about strategies for addressing this about strategies for addressing this problemproblem

Approaches to Predicting Approaches to Predicting Incidents of ViolenceIncidents of Violence

Actuarial ApproachActuarial Approach

Same information considered for Same information considered for every personevery person

Consistent method for combining Consistent method for combining informationinformation

Factors considered not necessarily Factors considered not necessarily “causal” for that individual“causal” for that individual

Clinical ApproachClinical Approach

Individualized judgmentIndividualized judgment

Range of relevant factors considered Range of relevant factors considered is very broad; ideally, an integrative is very broad; ideally, an integrative viewview

Operates from a theory of how Operates from a theory of how violence might occur or unfold in the violence might occur or unfold in the individual’s lifeindividual’s life

Problems with the Actuarial Problems with the Actuarial ApproachApproach

All the information has to be All the information has to be availableavailable

Generally assumes that all the Generally assumes that all the factors apply the same to everyonefactors apply the same to everyone

The risk estimate is devoid of theoryThe risk estimate is devoid of theory

Problems with the Clinical Problems with the Clinical ApproachApproach

Depends on the person doing itDepends on the person doing it Variability in:Variability in:

Information gatheredInformation gathered Ways information is combinedWays information is combined

Human biases occurHuman biases occur RecencyRecency VividnessVividness

Affected by organizational demands Affected by organizational demands Optimization is assumedOptimization is assumed ““Satisficing” is more commonSatisficing” is more common

Actuarial vs. Clinical Actuarial vs. Clinical PredictionPrediction

Old debate Old debate SuicideSuicide Job/academic successJob/academic success

General findingsGeneral findings Actuarial Approach generally more Actuarial Approach generally more

accurateaccurate Clinical Approach more versatile Clinical Approach more versatile

Issue of reliability and validity Issue of reliability and validity

Reliable, but not valid

Valid, but not reliable

Actuarial Risk Assessment ToolsActuarial Risk Assessment Tools General violence/recidivism (mainly in mentally ill General violence/recidivism (mainly in mentally ill

individuals)individuals) Historical-Clinical-Risk Management-20 (HCR-20) Historical-Clinical-Risk Management-20 (HCR-20) Violence Risk Appraisal Guide (VRAG) Violence Risk Appraisal Guide (VRAG) Classification of Violence Risk (COVR)Classification of Violence Risk (COVR) Violent Offender Risk Assessment Scale (VORAS)Violent Offender Risk Assessment Scale (VORAS)

Special purpose instrumentsSpecial purpose instruments Domestic violence Domestic violence

Spousal Assault Risk Assessment (SARA)Spousal Assault Risk Assessment (SARA) Risk of sex offenseRisk of sex offense

Static-99 Static-99 Risk of violence among juvenilesRisk of violence among juveniles

Early Assessment Risk List for Boys (EARL-20B) Early Assessment Risk List for Boys (EARL-20B) Manual for the Structured Assessment of Violence Risk in Youth (SAVRY)Manual for the Structured Assessment of Violence Risk in Youth (SAVRY)

Issues regarding actuarial Issues regarding actuarial instrumentsinstruments

Increasingly popular because of technologyIncreasingly popular because of technology Optimization on chanceOptimization on chance

Shrinkage inevitableShrinkage inevitable Application on particular sample might not Application on particular sample might not

be appropriatebe appropriate Effect of contextEffect of context

Information availabilityInformation availability Outcome of decision Outcome of decision

Not a replacement for clinical judgment. Not a replacement for clinical judgment. Integration of actuarial and clinical Integration of actuarial and clinical information is the goalinformation is the goal

Inherent Challenges to Inherent Challenges to Predicting Campus Predicting Campus

ViolenceViolence

Major General John Sedgwick

Problem #1: Low Base Problem #1: Low Base RatesRates

Actually violent Actually violent

YesYes NoNo

PredictedPredicted Yes 80Yes 80180180 260 260

Violent Violent NoNo 2020 720720 740740

100100 9009001,0001,000

Assume that one in ten individuals is actually violent over a given time period

Assume that the instrument correctly identifies 80% of the violent individuals80% of the nonviolent individuals

true positives

true negativesfalse

negatives

false positives

ImplicationsImplications

No technology will predict rare eventsNo technology will predict rare events Secret Service study of school shootings: Secret Service study of school shootings:

“There is no accurate or useful profile of the “There is no accurate or useful profile of the school shooter”. school shooter”.

““profiles” may be valuable, but not because profiles” may be valuable, but not because they are predictivethey are predictive

The utility of screening and assessment The utility of screening and assessment is not to predict for an individual, but is not to predict for an individual, but to identify groups with higher base ratesto identify groups with higher base rates to focus prevention resources to focus prevention resources

Problem #1: Low Base Problem #1: Low Base RatesRates

Problem #2: Context Problem #2: Context matters and situations matters and situations

changechange

Problem #2: Context Problem #2: Context matters and situations matters and situations

changechange Violence is usually Violence is usually

dependent on proximal situational dependent on proximal situational factors factors

transactional transactional

Opportunities for violence Opportunities for violence may differ substantially across may differ substantially across

individualsindividuals can be altered by lifestyle changescan be altered by lifestyle changes

presence of alcohol presence of alcohol living arrangementsliving arrangements

ImplicationsImplications

Risk statusRisk status may be important, may be important, but so are fluctuations in but so are fluctuations in risk risk statestate Move toward management of high Move toward management of high

risk individuals and situations risk individuals and situations Conditional prediction modelConditional prediction model

““if….then” formulation of riskif….then” formulation of risk Monitoring and management of Monitoring and management of

“dynamic predictors”“dynamic predictors”

A research example:A research example: Substance use and Substance use and

violenceviolence

Study DesignStudy Design Select group of individuals who were highly likely to Select group of individuals who were highly likely to

have frequent involvement in violencehave frequent involvement in violence

Weekly interviews providing daily reportsWeekly interviews providing daily reports Violent incidents Violent incidents Substance use reports at daily levelSubstance use reports at daily level

Alcohol (number of drinks)Alcohol (number of drinks) Marijuana useMarijuana use Other drugs (mostly cocaine)Other drugs (mostly cocaine)

Analyses of the concurrent and lagged relationships Analyses of the concurrent and lagged relationships of substance use and violenceof substance use and violence

Conditional Probability SERIOUS VIOLENCE

Conditional Probability ANY VIOLENCE

P (violence| alcohol only) 2.3 % 7.9%

P (violence| marijuana only ) 1.6 % 4.2%

P (violence| other drug only) 2.5 % 7.4%

P (violence| alcohol and mj) 4.4 % 10.5%

P (violence| alcohol and otherdrugs) 4.5 % 10.2%

P (violence| alcohol, mj, otherdrugs) 9.0 % 19.4%

P (alcohol | violence) 21.7 % 20.2%

P (marijuana | violence) 32.4 % 28.0%

P (other drugs| violence) 6.2 % 5.5%

P (alcohol & any drug use) 9.3 % 9.3%

P (alcohol & any drug use| violence) 23.1 % 21.0%

SeriouSerious s

violencviolencee

Any Any violenceviolence

Alcohol Alcohol (> 3 (> 3

drinks)drinks)

MarijuaMarijuanana

Other Other drugsdrugs

Percent of Percent of daysdays

1.4%1.4% 3.9%3.9% 8.7%8.7% 20.6%20.6% 2.5%2.5%

Day Before

Day AfterSerious Violenc

eAlcohol

Marijuana

Other Drugs

Serious Violence 5.4 1.9 1.5 2.1

Alcohol 2.4 9.5 2.1 2.8

Marijuana 1.6 2.3 31.5 1.5

Other Drug 1.5 2.2 1.5 48.1

Odds ratios for substance use and Odds ratios for substance use and violence violence

one day apart for serious violenceone day apart for serious violence

ExamplesExamples

--------|||||||-|----------------------------------------------------|---|-----

Case 8

Case 2080

-|------|------------|------|---------------|-|------|--------------|---------|--|--------|----|---|------|------------|------|---------------|-|------|--------------|---------|--|--------|----|--

Testing the relation of Testing the relation of violence and alcohol use over violence and alcohol use over

time time

FindingsFindings Evidence for a lagged effect for alcohol Evidence for a lagged effect for alcohol

use (greater than three drinks) on use (greater than three drinks) on violence, but not the other way aroundviolence, but not the other way around

No significant lagged relationships No significant lagged relationships either way for marijuana use or other either way for marijuana use or other drugsdrugs

Even controlling for different types of Even controlling for different types of substance use, violence on one day substance use, violence on one day predicts for the next daypredicts for the next day

Use of multiple substances on prior day Use of multiple substances on prior day also increases likelihood of violencealso increases likelihood of violence

Problem #1: Low Base Problem #1: Low Base RatesRates

Problem #2: Context Problem #2: Context matters and situations matters and situations

changechange

Problem #3: Late Problem #3: Late adolescence is all about adolescence is all about

change change

Problem #3: Late Problem #3: Late adolescence is all about adolescence is all about

changechange Late adolescence brings: Late adolescence brings:

Independent social roles Independent social roles ““trying on” lifestylestrying on” lifestyles ongoing brain developmentongoing brain development Different patterns of substance useDifferent patterns of substance use

Onset period for many mental disordersOnset period for many mental disorders Substance abuse: 20 years oldSubstance abuse: 20 years old Mood disorders: 30 years oldMood disorders: 30 years old Schizophrenia: 20 years old (males); 30 Schizophrenia: 20 years old (males); 30

years old (females)years old (females) Involvement in violence drops off, even Involvement in violence drops off, even

in serious offendersin serious offenders

Self Reported Offending Self Reported Offending Serious Adolescent Offenders - Serious Adolescent Offenders -

males only – average age 16 at first males only – average age 16 at first interviewinterview

0

2

4

6

8

0 6 12 18 24 30 36

Months after Initial Interview

Se

lf R

ep

ort

Va

rie

ty

Sc

ore Group Group

4 4 (15.1%(15.1%

))

Group Group 5 5

(8.5%)(8.5%)

Group Group 2 2

(33.8%(33.8%))

Group Group 3 3

(18.3%(18.3%))

Group Group 1 1

(24.2%(24.2%))

ImplicationsImplications

Lack of solid history for use in making Lack of solid history for use in making judgments about future violence judgments about future violence

Mix of developmental features and valid Mix of developmental features and valid symptomssymptoms Impulsiveness, moodiness, feelings of being Impulsiveness, moodiness, feelings of being

picked on, feelings of rejection, tendency to picked on, feelings of rejection, tendency to blame othersblame others

Diagnostic labels are less validDiagnostic labels are less valid

Likely to confuse Likely to confuse risk markersrisk markers with with risk risk factorsfactors

Assessments have a limited shelf lifeAssessments have a limited shelf life

Guidelines for structuring Guidelines for structuring risk assessmentsrisk assessments

1. Get the right 1. Get the right information on the right information on the right individualsindividuals

Screen Screen using structured measures, using structured measures, and and assessassess for risk when warranted for risk when warranted

Use all available information Use all available information consistentlyconsistently

Use relevant assessment toolsUse relevant assessment tools Group characteristicsGroup characteristics Target behavior of interestTarget behavior of interest Use actuarial instrument as an “anchor”Use actuarial instrument as an “anchor”

Best Bets for individual Best Bets for individual assessmentassessment

history of violencehistory of violence impulsivity (process from ideation to action)impulsivity (process from ideation to action) active ideation (mainly hostility and anger)active ideation (mainly hostility and anger) drug and alcohol usedrug and alcohol use psychopathypsychopathy perceived threatperceived threat plan/access to meansplan/access to means opportunities for violent encountersopportunities for violent encounters coping strategiescoping strategies

2. Consider history in detail2. Consider history in detail

When there is past violence, assess:When there is past violence, assess: what happenedwhat happened what factors explain the incidentwhat factors explain the incident what factors protect against violencewhat factors protect against violence which risk and protective factors are which risk and protective factors are

currently in effect or likely to be in effectcurrently in effect or likely to be in effect When there is no past violence, assess:When there is no past violence, assess:

““close calls”close calls” recent changes in life that may exceed recent changes in life that may exceed

coping capacitycoping capacity risk factors for violence, based on risk factors for violence, based on

appropriate assessment protocolsappropriate assessment protocols

3. For prevention efforts, 3. For prevention efforts, distinguish between risk distinguish between risk markersmarkers and risk and risk factorsfactors Key is to focus on causal, dynamic risk Key is to focus on causal, dynamic risk

factorsfactors

QuestionsQuestions Is this a risk factor for a particular type of Is this a risk factor for a particular type of

violence?violence? Are there conditions that influence the Are there conditions that influence the

relationship between the risk factor and relationship between the risk factor and violence?violence?

Does the risk factor play a Does the risk factor play a causalcausal role in role in violence? violence?

If so, is the risk factor capable of being If so, is the risk factor capable of being modifiedmodified??

4. Take context and risk 4. Take context and risk state seriouslystate seriously

A single risk assessment is useless A single risk assessment is useless without a management planwithout a management plan

Assess individual factors and Assess individual factors and conditions periodically in high risk conditions periodically in high risk casescases

Create an environment where students Create an environment where students trust authorities enough to share trust authorities enough to share information information