Brian Bumbarger, M.Ed. & Stephanie Bradley, Ph.D. Bennett-Pierce Prevention Research Center Penn State University Using BOTH Randomized Trial and Meta-analytic Approaches to Prevent Delinquency and Reduce Recidivism: The Pennsylvania Case Study
Brian Bumbarger, M.Ed. & Stephanie Bradley, Ph.D. Bennett-Pierce Prevention Research Center
Penn State University
Using BOTH Randomized Trial and Meta-analytic Approaches to Prevent Delinquency and Reduce Recidivism: The Pennsylvania Case Study
Collaborative Policy Partners: Linda Rosenberg John Frain
James Anderson Keith Snyder Mike Pennington Clay Yeager
Investigators and Authors:
Brian Bumbarger Stephanie Bradley Brittany Rhoades Cooper Mark Feinberg Mark Greenberg
The EPISCenter and much of the research described here are supported by grants from the Pennsylvania Commission on Crime and Delinquency. Special thanks to the staff of the Office of Juvenile Justice and Delinquency Prevention (OJJDP)
2
§ Strong evidence-base § Demonstrated effects
(confidence) § Sponsored lists
§ E.g., Blueprints
§ Funding requirements
§ Not an EBP for every community need/context
§ Many programs/services
already in widespread use
§ Some of those may be effective
§ Local expertise/fit
A 2-pronged approach…
Promoting Evidence-based Programs
Developing Practice-based Evidence
History of Research-Based Prevention in Pennsylvania
§ 1994: Key state leaders introduce Communities that Care (CTC) § Data-driven strategic prevention planning framework
§ 1994-2002: Initiation of CTC funding by PCCD
§ 16 cycles of CTC model introduced in ~120 communities
§ 1996: PCCD Co-funding of research to develop Blueprints EBP list
§ 1998: Initiation of Evidence-based Program Initiative by PCCD § 12 cycles of EBPs funded over 15 years, resulting in over 200 EBP replications
§ 2001: Targeted Technical Assistance by Prevention Research Center § Focus on implementation quality and fidelity, outcome assessment, & sustainability
§ 2008: Resource Center for Evidence-Based Prevention and Intervention Programs and
Practices created (EPISCenter)
§ Multi-agency Steering Committee Representing Justice, Welfare, Education, and Health § More proactive engaged TA and system coordination
Intermediary and State-‐level Preven1on Support System
Support to Community Preven0on Coali0ons
Improve Quality of Local Innova0ve Programs and
Prac0ces
Support to Evidence-based
Programs
Multi-Agency Steering Committee (Justice, Welfare, Education, Health)
The EPISCenter is a project of the Prevention Research Center, College of Health and Human Development, Penn State University, and is funded by the Pennsylvania Commission on Crime and Delinquency and the Pennsylvania Department of Public Welfare
as a component of the Resource Center for Evidence-‐Based Prevention and Intervention Programs and Practices.
A unique partnership between policymakers, researchers, and communities to bring science to bear on issues of public health and
public safety
EPISCenter’s 3 Key Func0ons: § Build general
prevention capacity
§ Build program-specific capacity
§ Facilitate interaction/ communication between systems
Rhoades, Bumbarger & Moore (2012). The Role of a State-Level Prevention Support System in Promoting High-Quality Implementation and Sustainability of Evidence-based Programs. American Journal of Community Psychology.
Tran
sla'
onal
Research EPISCenter
(Preven0on Support System)
Build general preven'on capacity among prac''oners and policy makers
Technical Assistance
Con'nuous Quality
Improvem
ent
Providers, Prac''oners, & Community Coali'ons (Preven0on Delivery System)
Preven'on Researchers and Program Purveyors
(Preven0on Synthesis & Transla0on System) Policy Makers & Funding Agencies
-‐20
-‐10
0
10
20
30
40
-‐10.8
33.2
-‐10.8
16.4
Delinquency Academic PerformanceNegative Peer Influence School Engagement
5 year Longitudinal Study of PA Youth % Change of CTC/EBP Youth Over
Comparison Group
419 age-grade cohorts over a 5-year period:
youth in CTC communities using
EBPs had significantly lower rates of
delinquency, greater resistance to negative
peer influence, stronger school engagement and
better academic achievement
Feinberg, M.E., Greenberg, M.T., Osgood, W.O., Sartorius, J., Bontempo, D.E. (2010). Can Community Coalitions Have a Population Level Impact on Adolescent Behavior Problems? CTC in Pennsylvania, Prevention Science.
Impact on Juvenile Court Placement Rates: Comparison of Placement Rates for Counties* With and Without an EBI
Bumbarger, B. K., Moore, J., & Rhoades, B. (2010). Impact of evidence-based interventions on delinquency placement rates. Presentation at 2011 Society for Prevention Research annual meeting. Washington, DC.
PA Return-on-Investment Analysis
Jones, D. J., Bumbarger, B. K., Greenberg, M. T., Greenwood, P., & Kyler, S. (2008). The economic return on PCCD’s investment in research-based programs: A cost-benefit assessment of delinquency prevention in Pennsylvania. The Prevention Research Center, The Pennsylvania State University.
Programs/services can be placed along a continuum of confidence based on their evidence or theory
How confident are we that this program or practice is a good use of resources AND improves outcomes for children and families?
Very Confident
ü Evidence-‐based “This program has been rigorously evaluated and shown to work”
ü Promising Approaches “We really think this will work… but we need time to prove it”
• Best Practices • “We’ve done it
and we like it”
EFFECTIVE HARMFUL
" Iatrogenic (Harmful) “This program has been rigorously evaluated and shown to be ineffective or harmful”
Promising Ineffective unknown Very
Confident
*Bumbarger & Rhoades, 2012
What is the Standardized Program Evaluation Protocol (SPEP)?
§ Partnership between probation/juvenile court & service providers
§ Quality improvement process informed by meta-analytic research
§ Specifically aimed at reducing recidivism
• PA has adopted a data-driven, evidence-based approach in improving the juvenile justice system to reduce youth recidivism rates
• PA average rate of recidivism from 2007-2009 was 22%, with a range of 6-33%
• Five medium-sized urban counties self-selected to participate in pilot of SPEP quality improvement process
• Pilot counties’ recidivism rates ranged from 12-24%
• SPEP scores can range from 0-100; scores 50+ shown to be predictive of reduced recidivism at 6 and 12 months
Key Drivers of JJ Services Effectiveness
Service Type
Provider Delivery
SPEP Assessment
Probation Usage
Program Improvement Lifecycle SPEP
Assessment
Understanding SPEP (score)
Improvement Implications
Improvement Plan
Plan Implementation Service
Type
Provider Delivery
SPEP Assessment
Probation Usage
Program Improvement Plan
§ Addresses the areas identified during the SPEP process, as prioritized by stakeholders
§ Identifies the timeframe and method for improvements in accordance with the stakeholder capacities
§ Identifies the needed technical assistance and support which may be necessary to implement improvements.
§ Process for monitoring the progress and outcomes of the Program Improvement Plan.
Most services* score ≥ 50:
* important to consider selection bias of self-selected early adopters
• 23 fully scored services; avg. score was 58, range of 37-100
• Most services scored well on staff training and supervision
• Most services scored poorly on written protocols and response to drift
• VERY poor fidelity to dosage and duration standards
• Self-selection bias; “True” state of the field may be in worse shape than what is depicted here
• Evidence of SPEP predictive validity is very gross measure (+/- 50); limits precision
• Probably not quite ready for scale – better tools and training needed after this pilot phase
• Nearly impossible to convince providers and JPOs NOT to focus on score…potential implications for competition and funding
• Array of services insufficient to match risk assessment
• Further muddies the waters concerning “evidence-based”
• Lower recidivism?? (already quite low – stay tuned)
Limitations and Cautions…
• Qualitative interviews lead JPOs better understanding what programs “really” offer (and for whom services are best suited)
• JPOs are now more routinely sending YLS risk score to providers (better service matching and treatment plans)
• Ongoing education of juvenile court system (Judges) re: relationship of dosage and duration to youth/service outcomes
• Improved relations between probation and providers
Early Benefits from the Pilot Phase
• Service providers are encouraged to more clearly articulate the theory of behavior change of their services
• Providers are developing more specific procedures for coaching, monitoring implementation and addressing drift
• Early discussions of moving to performance-based contracting?
• Evidence that these two models of “evidence” can be complimentary rather than competing!
Early Benefits from the Pilot Phase (cont.)
Thank You!
Brian Bumbarger Preven'on Research Center, Penn State University
www.preven'on.psu.edu www.episcenter.org