Residential Positive Achievement Change Tool (R-PACT) Validation Study June 2013 Carter Hay, PhD, Principal Investigator Alex Widdowson, MA, Research Assistant STATE OF FLORIDA Department of Juvenile Justice
Residential Positive Achievement Change Tool (R-PACT) Validation StudyJune 2013
Carter Hay, PhD, Principal InvestigatorAlex Widdowson, MA, Research Assistant
STATE OF FLORIDADepartment of Juvenile Justice
Introduction
This presentation describes key conclusions from a validation study of the Residential Positive Achievement Change Tool (R-PACT).
This study was initiated by DJJ and we have consulted throughout with Meg Bates, Mike Baglivio, and Kathy Jackowski.
STATE OF FLORIDADepartment of Juvenile Justice
2
About the R-PACT
The R-PACT is a survey instrument used by Florida DJJ to:
o identify residential youth’s criminogenic needs/risks
o guide the development of intervention strategies—directing the right services to the right youths
o assist in determining progress.
STATE OF FLORIDADepartment of Juvenile Justice
3
About the R-PACT (continued)
Other details:
o 52 domains: 26 risk and 26 protective scores
o Some domains relate to history, others to current circumstances
o Administered by residential staff at multiple points
STATE OF FLORIDADepartment of Juvenile Justice
4
R-PACT domains
STATE OF FLORIDADepartment of Juvenile Justice
5
I
R-PACT Domains 1: Record of referrals 3A: School history 3C: current academic status 3D: Current vocational training 4A: Historic use of free time 4B: CWl·ent use of free time 5A: Employment history 5B: Employability 5D: Program supervised tasks 6A: History of relationships 6B: Cunent relationships 7A: Family history 7B: Parent/ caregiver relationships
••
8A: 8B: 9A: 9B: 10: 11: 12: 12A: 12B: 12C: 12D: 12E:
AlcohoVdrug history CWl·ent alcohoVdrug use Mental health history Current mental health Current attitudes Cunent aggression Current skills Dealing "ith others Dealing with situations Dealing with emotions Controlling impulses Controlling aggression
THE FLORIDA STATE U IVERSTTY COli Eta:. Of CRI/\I INOI OLY & CR IM INAl JUS rlc..l:.
Illustrative items from key domains
Domain 3C: Current academic status
o “Believes getting an education is of value;” “believes school is encouraging;” is “close to” educational staff; behaves in the school setting.
Domains 12, 12A-12E: Social skills
o For Domain 12C (dealing with emotions): “expressing feelings;” “understanding the feelings of others;” “dealing with someone else’s anger.”
STATE OF FLORIDADepartment of Juvenile Justice
6
The study focused on two questions:
Do R-PACT domain scores predict who will become a reoffender?
What do R-PACT scores reveal about changes that youths experience during the residential stay?
The study’s focus
STATE OF FLORIDADepartment of Juvenile Justice
7
Our approach
Data come from a sample of 4,735 residential completions in FY 2010-2011.
o 85% male, 15% female
o 53% black, 35% non-Hispanic white, 11% Hispanic
o Average age = 16.5 years
STATE OF FLORIDADepartment of Juvenile Justice
8
Our approach (continued)
Analyses of youth change focused on changes between the Initial and Exit R-PACTs.
Reoffending is defined in terms of adjudications, adjudications withheld, and convictions for new law violations occurring within twelve months.
o Reoffending rate in this sample: 40%
STATE OF FLORIDADepartment of Juvenile Justice
9
1. Roughly 60% of the R-PACT risk and protective scales affect reoffending.
o Scoring “well” translated into a reoffending rate of about 34-35%.
o Scoring “poor” translated into a reoffending rate of about 45-46%.
o This held across males, females, blacks, Hispanics, whites, and those above or below age 16 at intake.
Findings
STATE OF FLORIDADepartment of Juvenile Justice
10
2. Domain 1 Record of referrals (the extent and seriousness of youths’ prior offending) is the most important individual domain.
o Scoring “high”: A reoffending rate of about 53%.
o Scoring “low”: A reoffending rate of about 28%.
Findings (continued)
STATE OF FLORIDADepartment of Juvenile Justice
11
3. Other domains had consistent effects that were at least moderate in strength.
o 3C Current academic status (values education)o 5B Employability (knowledge, aspirations, and plans)o 6B Current relationships (antisocial/social peers)o 10 Current attitudes and behaviorso 11 Current aggression (attitudes/involvement)o 12/12A-12E Current social skills (dealing with emotions, others, difficult situations, controlling impulses)
Findings (continued)
STATE OF FLORIDADepartment of Juvenile Justice
12
Findings (continued)
4. A number of these important domains worked together to affect reoffending.
For our most predictive multivariate risk factor model, reoffending was:
o 55% for those high on all risk variables
o 40% for those at the mean on all risk variables
o 26% for those scoring low on all risk variables
STATE OF FLORIDADepartment of Juvenile Justice
13
Findings (continued)
5. Some domains are unimportant for predicting reoffending:
o 9A Mental health historyo 9B Current mental health o 5A Employment historyo 8A Alcohol/drug historyo 7A Family history
STATE OF FLORIDADepartment of Juvenile Justice
14
Findings (continued)
6. Youths showed considerable improvement from the Initial to the Exit R-PACT:
o 14% reduction in risk factor scores
o 28% increase in protective factor scores
o The greatest improvements were in the social skills domains—30%-50% of youths were classified as major improvers.
STATE OF FLORIDADepartment of Juvenile Justice
15
Findings (continued)
Aggregate change in protective scores0
200
400
600
Freq
uenc
y
-.4 -.2 0 .2 .4 .6 .8Total R-PACT Percent Maximum Change: Protective
STATE OF FLORIDADepartment of Juvenile Justice
16
7. All demographic subgroups showed Initial-to-Exit improvements.
Findings (continued)
STATE OF FLORIDADepartment of Juvenile Justice
17
Initial‐to‐Exit changes across different groups
Findings (continued)
STATE OF FLORIDADepartment of Juvenile Justice
18
8. Youths who experienced the greatest Initial-to-Exit improvements had lower reoffending; however:
o This was primarily the case among those with a lower record of prior offending.
o One exception: For high prior offending youth, improved social skills for dealing with emotions and controlling impulses reduced reoffending.
Findings (continued)
STATE OF FLORIDADepartment of Juvenile Justice
19
9. Improvements in social skills were especially likely to reduce reoffending.
o Other notable areas:- 3C Current academic status (committed to, excels in
education) - 5D Supervised tasks (performance of tasks in the
facility)- 6B Current relationships (involvement with antisocial
peers)- 10 Current attitudes and behaviors (antisocial
attitudes/behaviors)
Findings (continued)
STATE OF FLORIDADepartment of Juvenile Justice
20
1. The R-PACT should be used to make assessments about which youths are most likely to reoffend.
o This can be done for all demographic subgroups.
o The focus should be on the most predictive domains.
o These assessments should guide transition, release, and after-care decisions.
Recommendations
STATE OF FLORIDADepartment of Juvenile Justice
21
2. The R-PACT should be used to assess youth changes during the residential stay and to guide performance plans.
o Special attention should be devoted to improvements in social skills (with existing programs like Thinking for a Change).
Recommendations (continued)
STATE OF FLORIDADepartment of Juvenile Justice
22
3. Programs should continue to prioritize and where possible intensify attention to youths’ commitment to doing well in school.
Recommendations (continued)
STATE OF FLORIDADepartment of Juvenile Justice
23
4. DJJ should closely track and where necessary increase intervention and treatment dosage/frequency for moderate and high prior offending youths.
o This follows from the conclusion that Initial-to-Exit improvements reduced reoffending most notably for low prior offending youths.
Recommendations (continued)
STATE OF FLORIDADepartment of Juvenile Justice
24
5. DJJ should closely track and where necessary intensify post-release supervision and intervention, especially for moderate and high prior offending youth.
o Youths with more extensive prior offending face greater obstacles to successful reintegration.
Recommendations (continued)
STATE OF FLORIDADepartment of Juvenile Justice
25
1. Because of differences in how youths responded to the residential stay, research should consider sources of this variation:o Differences across programso Differences across interventions (e.g., Thinking for a
Change or Aggression Replacement Training)o Differences across program types (e.g., secure vs. non-
secure)o Differences across programs’ behavior management
systems and practices (e.g., use of Protective Action Response [PAR] vs. positive reinforcement)
Priorities for future research
STATE OF FLORIDADepartment of Juvenile Justice
26
2. New research should examine differences across youth in the frequency, duration, consistency, and fidelity of social skills training.
Priorities for future research (continued)
STATE OF FLORIDADepartment of Juvenile Justice
27
3. New research should examine the transition from residential facilities to post-release supervision/treatment.
o A key question: What changes do youths experience in risk and protective factors during the period after release?
o This can involve linking the R-PACT data collected during residence with the C-PACT data collected after release
Priorities for future research (continued)
STATE OF FLORIDADepartment of Juvenile Justice
28
4. New research should examine the R-PACT’s scoring system for creating the domain risk and protective scales.
o Does the current scoring system maximize the predictive accuracy of the R-PACT?
Priorities for future research (continued)
STATE OF FLORIDADepartment of Juvenile Justice
29
For more information:
STATE OF FLORIDADepartment of Juvenile Justice
Florida Department of Juvenile Justice2737 Centerview DriveTallahassee, FL 32399-3100www.djj.state.fl.us
Office of Residential ServicesMeg W. Bates, MSPrograms and Policy Coordinator850/[email protected]
Bureau of Research and PlanningMichael Baglivio, PhD850/[email protected]
Kathy Jackowski, MS850/[email protected]
The Florida State UniversityCollege of Criminology & Criminal JusticeCarter Hay, PhD, Principal [email protected]
Alex Widdowson, MA, Research Assistant