1	Findings from the Multi-site SVORI Evaluation Pamela K. Lattimore, Ph.D. Principal Scientist RTI International JRSA National Conference October 11, 2007
2	SVORI Overview Serious and Violent Offender Reentry Initiative funded agencies to develop programs to improve criminal justice, employment, education, health, and housing outcomes for released prisoners 69 agencies received federal funds ($500,000 - $2,000,000 over 3 years) to develop 89 programs that Targeted adult & juvenile populations Incorporated partnerships among state and local agencies to provide comprehensive services to prisoners returning home Were locally designed to meet local needs and organizational capabilities
3	SVORI Program Model
4	Multi-site SVORI Evaluation Implementation: Identify local SVORI program components —what was intended & what was achieved Impact: Determine the extent to which SVORI increased services & programming; improved employment, health, & personal functioning; and reduced criminal activity of returning prisoners Cost-benefit: Determine whether benefits outweigh the costs (subset of impact sites)
5	Program Characteristics: In 2005, PDs Said Focused on the general “serious and violent population” (85%) Focused equally on the pre- and post-release phase (67%) Geographically targeted pre-release (specific institutions) and post-release (specific release communities) Most (90%) programs reported attempting to provide all needed services rather than delivering a specific service or set of services Primary focus of adult programs Employment: 42% Community integration: 27% Substance use: 14% Mental health: 10% Education: 2% Family: 0% Other: 6%
6	Impact Study Respondents Interviewed SVORI participants and comparison subjects 16 programs 14 states More than 300 prisons & jails More than 7,000 completed interviews
7	12 Adult and 4 Juvenile Impact Evaluation Sites Northeast (2 states) Maine Pennsylvania South (5 states) Florida (juvenile only) Maryland Missouri Oklahoma South Carolina (adult & juvenile) Midwest (4 states) Kansas (adult & juvenile) Indiana Iowa Ohio West (3 states) Colorado (juvenile only) Nevada Washington
8	Characteristics of Adult Male Respondents 29 years of age About 40% reported being married or in a steady relationship & about 60% reported having children < 18 years About 60% had a HS diploma or GED About 2/3rds reported having a job 6 months prior prison & 90% “ever had a job” Most had family & friends who were AOD & crime involved About 2/3rds reported drug & alcohol use in the 30 days prior to prison SVORI & Non-SVORI differed on a few measures Race: S more likely to be black, less likely to be white than NS AOD use & treatment: On some indicators, NS more AOD involved than S NS more likely than S to be in prison for parole violation NS had spent less time than S in prison S more likely than NS to be serving time for a drug crime
9	Service Needs & Receipt Bundles Coordination & Supervision (received only): needs assessment, case management, worked with anyone to reintegrate, supervision Transition Services: legal assistance, financial assistance, public benefits assistance, healthcare insurance, mentoring, assistance obtaining documents, finding place to live, transportation, getting drivers license, clothing/food Health Services: tx for physical, MH, AOD, abuse groups, anger management programs Employment/Education/Skills Building Services: assistance finding job, educ. services, money management, gen. lifeskills, assistance w/ personal relationships, changing criminal thinking Family Services: DV support group, batterer intervention Child Services: getting child support, assistance modifying child support debt, custody, parenting class, assistance finding child care (for the ~ 600 cases with children < 18 years)
10	Self-Reported Needs Bundle Scores (30 days pre-release)
11	Self-Reported Service Receipt (30 days pre-release)
12	Self-Reported Service Receipt (3 months post release)
13	Service Need & Receipt
14	So…What Did the Respondents Tell Us About Needs & Services? In general, those not in SVORI have more unmet need than SVORI participants Nonetheless, the overall level of service provision is low according to these self reports Bundle scores will be used in multivariate analyses
15	Impact
16	Impact Evaluation Outcome Measures Self Sufficiency & Quality of Life Employment Housing Family (including contacts, stability & child custody/support) Community Involvement Health Substance Use (sobriety & relapse prevention) Physical Health Mental Health Reduced Criminality Supervision Compliance Reoffending Rearrest Reconviction Reincarceration
17	Taking a Look: Adult Males 15 M Post Release About 3% homeless; about 1/3^rd report living in own place In contrast to baseline findings (re: pre-incarceration), few reported currently living with people using AOD or who have committed illegal acts SVORI subjects more likely than non-SVORI to respond positively to questions about reentry programming Nearly ¾’s have worked since release
18	Taking a Broader View We have more than 100 potential outcome measures at the individual level Community involvement Employment Family Contact/Stability Health/Mental Health Housing Recidivism Substance Use Supervision Compliance
19	SVORI better than Non-SVORI?
20	So?
21	Propensity Score Models: Controlling for Differences Overall, the SVORI and non-SVORI groups are very similar although there are a few differences Propensity score models were estimated in an effort to achieve better balance between the two groups Subjects were assigned to one of five propensity equal-probability bins based on their likelihoods of assignment to SVORI Bins 1 and 5 contained very few subjects, so the subsequent analyses focuses on Bins 2, 3 and 4
22	15-Month Outcomes by Propensity Score Bin
23	So?
24	Treatment Effects
25	Conclusions SVORI: Ambitious effort to improve integrated, individually targeted services through DOC coordination of state & community agencies & organizations Within a year, most programs were able to greatly increase the level of services provided to offenders pre and post release—but levels were far less than 100%, particularly post release From release up through 15 months post release, SVORI participants are doing better—if only moderately so—across a wide range of outcomes
26	What’s Next? Complete acquisition of DOC, DJJ, probation and parole, and arrest data for all study participants (expanded samples in some sites); final data will cover all of 2007 (at least 24 months follow up for all subjects) Continue analysis of interview data, including developing & estimating multivariate outcome models Other analyses: Treatment-needs matching, site differences, cost-benefit/cost-effectiveness Begin to analyze the data for women & boys
27