The Pathways to Desistance Study:
Implications for Intervention and Policy
Edward P. Mulvey, Ph.D.
Law and Psychiatry Program
Department of Psychiatry
University of Pittsburgh School of Medicine
Presentation at Pennsylvania State University, State College, PA
November 7, 2012
Project Background
Supported by









Office of Juvenile Justice & Delinquency Prevention
National Institute of Justice
John D. & Catherine T. MacArthur Foundation
National Institute on Drug Abuse
Pennsylvania Commission on Crime & Delinquency
Arizona Governor’s Justice Commission
Robert Wood Johnson Foundation
William Penn Foundation
William T. Grant Foundation
Working Group Members
 Edward Mulvey
 Laurence Steinberg
 Elizabeth Cauffman
 Laurie Chassin
 George Knight
 Carol Schubert
 Sandra Losoya
 Robert Brame
 Jeffrey Fagan
 Alex Piquero
Reasons for the study

Richer information about serious adolescent
offenders

Picture of the desistance process
• Individual maturation
• Life changes
• Systems involvement

Improved practice and policy in juvenile justice
• Risk assessment
• Targeted interventions and sanctions
Study design



Two sites: Philadelphia and Phoenix
Enroll serious adolescent offenders
• 1,354 felony offenders, aged 14 -18
• Females and adult transfer cases
Regular interviews over seven years
• Initial interviews
• Time point interviews (background characteristics,
psychological mediators, family context, relationships, community
context, life changes)
Release interviews
Other sources of information
• Collateral interviews
• Official records
•

What we look at
Background Characteristics
•
•
•
•
•
•
•
•
•
•
Personal characteristics (e.g. family,
Academic achievement and commitment
Routine activities
Offense history
Alcohol and drug use/abuse
Exposure to violence
Psychopathy
Emotional reactivity
Acculturation
Personality
•
•
•
•
•
•
•
•
•
•
Psychological development
Mental health and threat-control symptoms
Head injury
Use of social services
Perceptions of opportunity
Perceptions of procedural justice
Perceived thrill of doing crime
Moral disengagement
Religious orientation
Costs and rewards of offending
Psychological Mediators
Family Context
• Parental Monitoring
• Parental Relationships
• Parent orientation
Personal Relationships
• Relationships with romantic partner & friends
• Peer delinquency and gang involvement
• Contact with caring adult
Community Context
• Neighborhood conditions
• Community involvement
• Personal capital and social ties
Life Changes
Monthly data available regarding:
Living arrangements
School involvement
Legal involvement
Work
Romantic relationships
Social service
involvement/sanctions
Living situation calendar
Month 1
Month 2
Month 3
Month 4
Month 5
Month 6
Subject 1
900 West
Huntington
St Gabe’s
Hall
900 West
Huntington
St Gabe’s
Hall
Vision
Quest
Youth
Forestry
Camp
Subject 2
2429 W.
Augusta
Madison
Street Jail
1808 S.
Wilmot
1808 S.
Wilmot
1808 S.
Wilmot
Tucson
Prison
5050 Master
4th and
Norris
4th and
Norris
4th and
Norris
House of
Corrections
House of
Corrections
Subject 3
Who are these adolescents?
 At Enrollment
• 16 years old on average
• 86% male
• Average of two prior court appearances
32% had no prior petitions to court
Most of priors were for a person crime
 Ethnically diverse
2%
25%
29%
44%
Caucasian
African American
Latino
Other
Most Serious Adjudicated Charge –
Study Index Petition
30
25
20
15
10
5
0
Philadelphia
Phoenix
Progress so far
 Average of 90% of interviews completed at each time point
 Over 21,000 interviews completed
 Archiving data at the Interuniversity Consortium for Political and
Social Research (ICPSR). Initial data will be released for public
access in June, 2012.
 Dissemination efforts
• 7 OJJDP bulletins published or under review
• Over 60 academic papers in print or under review; book in
progress
• Over 100 presentations to academics, public officials, service
providers
Examples of topics being investigated
•
•
•
•
•
•
•
•
•
Procedural justice
Perceptions of risk/benefit of crime
Psychosocial maturity and criminal offending
Effects of substance use treatment
Acculturation/enculturation
Family functioning
Perceptions of opportunities
Neighborhood effects
Service Provision/Institutional Care
Patterns of Offending
Monahan, K. C., Steinberg, L., Cauffman, E., Mulvey, E. P. (2009). Trajectories of Antisocial Behavior and Psychosocial
Maturity from Adolescence to Young Adulthood. Developmental Psychology, 45(6), 1654-1668
Mulvey, E. P., Steinberg, L., Piquero, A. R., Besana, M., Fagan, J., Schubert, C., Cauffman, E.. (2010). Trajectories of
desistance and continuity in antisocial behavior following court adjudication among serious adolescent offenders.
Development and Psychopathology, 22, 453-475.
Monahan, K., Piquero, A., Glasheen, C., Schubert, Mulvey, E.P. (in press). Does time matter? Comparing trajectory
concordance and covariate association using time-based and age-based assessment. Crime and Delinquency.
Re-arrest
• 74% of the sample were arrested again within
the 7-year follow-up period
• Average of 4 arrests (among those rearrested)
80
68
% of sample
70
60
55
55
Person
50
40
30
Property
Drug
31
Weapon
Sex
20
10
0
4
Prevalence of Arrest by Year
(% of sample with an arrest)
35
30
Percent
25
20
15
10
5
0
y1
y2
y3
y4
y5
y6
y7
Mean rate of re-arrests in each
wave (for those with an arrest)
0.012
0.01
0.008
0.006
0.004
0.002
0
y1
y2
y3
y4
y5
y6
y7
Number of arrests per days in the community. Ex: 1 arrest in 121 days in community = .008,
1 arrest in 65 days in the community = .015, 3 arrests in 183 days in community = .016
Median Severity Ranking for Arrests
across time
(within month)
7
6
5
4
Series1
Log. (Series1)
3
2
1
0
1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 41 43 45 47 49 51 53 55 57 59 61 63 65 67 69 71 73 75 77 79 81 83 85
1 = status offense, 2=misdemeanor, 3 = possession of narcotics (excluding glue and marijuana), 4 = felony, not part 1,
5=major property felonies, 6=burglary, 7=drug felony, 2nd degree sex offense, 8 =felonious assault, felony w/ weapon
9 =murder, rape, arson
Self-reported offending
7 year follow-up period – only males – controlling for time on street
High stable
10%
Drop-off
21%
Low
rising
12%
Lowest
26%
Low
stable
31%
Average number of re-arrests
per year in the community
3
2.6
2.5
2.2
2
1.8
1.5
1.5
1.1
1
0.5
0
High stable
Drop-off
Low-rise
Low-stable
Lowest
Proportion of each offending pattern
group in each crime group
100%
90%
80%
70%
60%
Persisters
50%
Late Onset
40%
Desisters
30%
Mid stable
20%
Low stable
10%
0%
Violent
Crime
Property
Crime
Weapons Drug Charge
Charge
Other
Average number of risk markers above
the sample mean by offending pattern group
5
4.5
4
3.5
3
2.5
2
1.5
1
0.5
0
High stable
Drop-off
Low rising
Offending Group
Range 1- 7
Low stable
Lowest
Policy statement #1
Serious juvenile offenders vary considerably in
patterns of offending, risk factors, and life
situations. Policies about placement or program
eligibility which are based on criteria related to
the presenting offense are a poor predictor of
future recidivism or positive development.
To increase the impact of investments in justice
interventions, it is important to promote decision
frameworks or statutes that
a) consider cumulative risk and addressable needs, and
b) target services to the highest risk offenders.
Substance Abuse Disorders/
Substance Use
Schubert, CA, Mulvey, EP, Glasheen, C. (2011). The influence of mental health and substance use problems and
criminogenic risk on outcomes in serious juvenile offenders. The Journal of the American Academy of Child and
Adolescent Psychiatry, 50(9), 925-937
Mulvey, E.P., Schubert, C.A., Chassin, L. (2010). Substance use and offending in serious adolescent offenders.
OJJDDP Bulletin. Washington, DC: United States Department of Justice, Office of Juvenile Justice and
Delinquency Prevention.
Chassin, L., Knight, G., Vargas-Chanes, D., Losoya, S., Naranjo, D. (2009). Substance Use Treatment Outcomes in
a Sample of Serious Juvenile Offenders, Journal of Substance Abuse Treatment, 36(2), 183-194
% of sample arrested at each wave by
frequency of alcohol use*
35
Percent Arrested
30
25
20
Alc-no use
Alc -some use
15
Alc-frequent use
10
Frequent use = those
In top 1/6 of reported use
5
0
6m
12m
18m
24m
30m
36m
*The same pattern holds for frequency of marijuana use and
frequency of “other drug” use. The pattern holds for all seven years for all substances.
Prevalence of disorders by gender
60
50
53
43
45
40
35
Male
30
Female
20
10
0
Substance Use
Mood/anxiety
Does having a substance use or mood/anxiety
disorder matter for long term outcomes?
Approach: Moderation Analyses
Risk
(no disorder group)
Risk
(disorder group)
Outcome
a
b
Does a equal b?
Outcome
The role of substance use and
mood/anxiety disorders on outcomes
Findings
 The mental health disorders alone do not affect the
relationship between risk markers and outcomes.
Given a certain risk, adolescents with disorders have
no greater or lesser chance of either positive or
negative outcomes (holds for males and females).
 A substance use disorder does change the
relationship between risk markers and both gainful
activity and re-arrest over 6 years. It makes things
significantly worse.
Do changes over time in levels of
substance use affect offending?
 Approach:
• Look at within individual changes at each time and see if
they raise level of offending
• See whether these effects are different, depending on the
trajectory groups
• Consider substance use, having an antisocial partner, being
a parent, and level of gainful activity (school and work) all
at the same time - also consider time out of community.
 Findings:
• Across all groups, level of substance use and having an
antisocial partner increase offending significantly and
independently
 1.2 to 12 times more offending for substance use
 1.2 to 4 time more offending for antisocial partner
Are these adolescents getting
substance use services?
Looking at those adolescents with a diagnosed
substance use problem
% with service
Average
intensity of
sessions
Adult
Setting
Juvenile
Setting
Community
45%
67%
14%
1 every 10 days
1 every 5 days
1 every 50 days
Does substance use treatment work if
given to these adolescents?
 Approach:
• Within individual, does treatment involvement in one time
period affect substance use and offending in the next time
period?
• Test of treatment as actually provided
• Controlling for “street time”, drug testing, and characteristics
predicting who gets treatment
 Findings:
• Family involvement and duration of treatment necessary
• Significant short term (6-month) effects of treatment on
• Marijuana use
• Offending
• Alcohol reductions (not dependent on family involvement)
• Not a one-shot “inoculation”
Policy statement #2
Increase the provision of substance abuse
services to serious adolescent offenders in
both institutions and in the community,
making sure that they are of adequate
intensity (e.g., meet NIDA standards for
enrollment time) and preferably involve family
members. Youth who are supported by family
members have better outcomes.
Institutional Placement
Loughran, T., Mulvey, E., Schubert, C., Fagan, J., Losoya, S., Piquero, A. (2009). Estimating a dose-response
relationship between length of stay and future recidivism in serious juvenile offenders. Criminology, 47,
699-740.
Mulvey, E., Schubert, C.A., Chung, H. L. (2007). Service use after court involvement in a sample of serious
adolescent offenders. Children and Youth Services Review, 29(4), 518-544. NIHMS154499,
PMCID2777728.
Schubert, CA; Mulvey, EP, Loughran, TA, Losoya, S. (2012) Perceptions of institutional experience and
community outcomes for serious adolescent offenders. Criminal Justice & Behavior, 39(1), 71-93.
Institutional placements over 84 months
Subject 52928
Adult
Setting
3
Juvenile
Setting
2
Treatment
Facility 1
Community
0
0
Age 14
10
20
30
40
Month Enrolled
50
60
70
80
Institutional placements over 84 months
Subject 4716
Adult
Setting
3
Juvenile
Setting
2
Treatment
Facility 1
Community 0
0
Age 17
10
20
30
40
Month Enrolled
50
60
70
80
Institutional placements over 84 months
Subject 691
Adult
Setting
3
Juvenile
Setting
2
Treatment
Facility
1
Community
0
0
Age 15
10
20
30
40
Month Enrolled
50
60
70
80
Probation vs. placement
Unadjusted comparison of re-arrest rate
Mean Yearly Rate of Re-Arrest,
by Placement Status
rate
1.4
1.2
1.0
0.8
0.6
0.4
1.20
0.63
0.2
0.0
probation
placement
Propensity score matching
 Two step process:
• A propensity score is calculated for each case. It is the
predicted probability that you get placed given all of
the background characteristics considered
• Take each placed case and match it to one or more
probation case with similar propensity score
 We then can look to see if the placed group looks
similar to the matched probation group on a variety
of characteristics that might affect the outcome
 If the groups look alike, we can attribute any
difference in the outcomes to the fact that they
were placed
Treatment effect of placement
Matched groups
Mean Yearly Rate of Re-Arrest,
by Placement Status After Matching
1.4
1.2
rate
1.0
0.8
1.06
1.20
0.6
0.4
0.2
0.0
probation
placement
No significant differences between groups in rate of re-arrest
Do longer institutional stays
reduce re-arrest?
Approach
 Length of stay is broken up into discrete “doses”
 Methods to get similar cases across different
levels of the “dose”
• 65 of 66 variables show no difference among
the groups, meaning we can rule them out as
causes of differences in outcomes
 Response Curve is estimated
Dose-response curve
3 month intervals as doses
Expected Rate of Re-Arrest,
by 3 mo. Dose Category
3.0
2.55
2.5
rate
2.0
1.11
1.5
1.35
1.08
1.04
1.0
0.5
0.0
0-3
3-6
6-9
9-12
> 12
Findings
 Overall, no effect of placement on rate of rearrest (if anything, it may increase re-arrest)
 For intermediate lengths of stay (i.e., 3-13
months), there appears to be little or no
marginal benefit for longer lengths of stay
Policy statement #3
Reduce the rate of placement of serious
adolescent offenders in institutions as well as
the duration of these placements. Increase
the level of community-based services to
these adolescents.
Points to remember
 This is a sample of serious adolescent
offenders. Results aren’t applicable across the
whole juvenile justice system.
 Outcome measures rely on self reports, but
arrests and official records support the results
so far.
 The study was conducted in only two cities.
While there is considerable variability across
cities, the outcomes regarding development
over time appear consistent in each site.
“Be of use”
Contact Information
Edward P. Mulvey, Ph.D.
mulveyep@upmc.edu
Carol A. Schubert, MPH
schubertca@upmc.edu
www.pathwaysstudy.pitt.edu