The purpose of this study is to analyze data from the SVORI multisite evaluation to assess the potential explanations for the mixed effectiveness of reentry programs. The goal is to understand whether or not service-risk/need fit is related to successful reentry outcomes, or whether the needs of returning prisoners are unrelated to their risk of recidivism regardless of how well they are addressed.
The specific objectives identified by the researchers are as follows:
- Understand the nature of need and service delivery among the SVORI evaluation participants;
- Understand the correspondence between risk-need profile and services received, and develop a selection model that identifies which individuals have a high or low correspondence between risk-need profile and services received; and
- Assess whether the degree of service fit with risk-need profiles is related to successful reentry outcomes (i.e. reduced recidivism).
This study utilized data from the 12 SVORI (ICPSR 27101) sites included in the outcome evaluation that involved adult ex-prisoners, and the corresponding comparison group members that did not receive the enhanced services under SVORI funding but still received the general level of reentry programming provided in their state or jurisdiction. To prepare the SVORI data for analysis researchers merged Datasets 1 and 2 (Adult Males and Adult Females) and created seven separate datasets containing Waves 1 through 4 survey data, NCIC crime data, administrative data, and sampling weights.
This study uses panel data and propensity score matching techniques to develop a selection model to examine which of the observed characteristics of offenders are related to need and service receipt, and to investigate how seven key self-reported criminogenic needs in each of the four waves of data collection conducted by the SVORI evaluators corresponded with service receipt and subsequently self-reported and official measures of recidivism in subsequent waves. The researchers used panel data techniques to examine the impact of self-reported need and service receipt on both self-reported and official measures of recidivism across the full year of data collection, rather than examining outcomes within each interview wave. They also examined lagged effects of service receipt compared to reported need; basing their analysis on the fit between needs reported in one wave and services received/outcome observed in the next wave.
Participants in the original SVORI study were adult male and female offenders or juvenile male offenders
convicted of serious and violent offenses. This study focused only on the 2,054 adults in the original SVORI evaluation dataset, and excluded juveniles. Researchers combined the male and female datasets from the original study. In the original evaluation, 863 men and 153 women comprised the SVORI treatment group and 834 men and 204 women comprised the comparison group. For the present study the researchers ignored treatment status, because comparison group participants also received some reentry services and the goal was to examine the relationship between services received and reported needs rather than the impact of SVORI.
Users should consult the Serious and Violent Offender Reentry Initiative (SVORI) Multi-Site Impact Evaluation, 2004-2011 [United States] (ICPSR 27101 - https:/doi.org/10.3886/ICPSR27101.v1) for additional information on the original sampling.
Longitudinal: Cohort / Event-based
Adult male offenders in 12 states and female offenders in 11 states who received Serious and Violent Offender Reentry Initiative (SVORI) programming or were selected as comparison subjects between July 2004 and November 2005.
Arrest Records Data from the FBI's National Crime Information Center (NCIC) Database.
Lattimore, Pamela K., and Christy A. Visher. Serious and Violent Offender Reentry Initiative (SVORI) Multi-site Impact Evaluation, 2004-2011 [United States]. ICPSR27101-v1. Ann Arbor, MI: Inter-university Consortium for Political and Social Research [distributor], 2013-11-07. http://doi.org/10.3886/ICPSR27101.v1
program source code
computer-assisted personal interview (CAPI)
For the present study researchers obtained the SVORI (ICPSR 27101) outcome evaluation datasets from the National Archive of Criminal Justice Data (NACJD). The archive holds four separate datasets from the evaluation: Adult Males Data (Part 1, N=1,697), Adult Females Data (Part 2, N=357), Juvenile Males Data (Part 3, N=337) and official recidivism and reincarceration data (Part 4, N=35,469), which can be linked on a one-to-many basis with the individual-level data in the other three datasets.
Researchers only utilized data from parts 1, 2, and 4 of the original data. The data gathered during the original interviews (in the SVORI study) had been coded into an extensive set of indicator variables and scales by the original evaluators prior to archiving with NACJD. Datasets 1 and 2 in the original study each contain the same 5,698 variables: 5,635 variables from the interview data (Wave 1: 1,018; Wave 2: 1,574; Waves 3 and 4: 1,523), 10 drug test results, 36 recidivism indicators, 14 reincarceration variables, and three weight variables from the original propensity score matching process. Some questions were only asked of juvenile participants, so those variables were empty for the adult sample.
Most of the interview variables were formatted as indicators to show behavioral, and attitudinal constructs. Offender interview variables from the original dataset include demographics, housing, employment, education, military experience, family background, peer relationships, program operations and services, physical and mental health, substance abuse, crime and delinquency, and attitudes. Variables from the offender interviews utilized by the researchers in this secondary analysis included the following:
- Demographics: Age, gender, race, length of incarceration
- Housing: Housing stability, barriers to housing, type of housing
- Employment profile: Employment history, employment status, barriers to employment
- Education: Educational attainment, school attendance
- Family profile: Partner, children, responsibilities, family criminal history, relationships and support, family victimization
- Health: Physical and mental health status
- Substance Use: Types of substances used
- Crime and Delinquency: Criminal history and criminal attitudes, incarceration status, technical violations
- Attitudinal profiles based on validated scales: Self-efficacy, locus of control, readiness for change
- Needs and Services: service needs, types of services received, post-release experiences and offender perceptions of most and least helpful services.
- Crime Outcomes: Self-reported offending at each wave; official indicators of recidivism based on the NCIC dataset
The recidivism and reincarceration indicators describe individuals' arrest or incarceration status at 3, 6, 9, 12, 15, 21, 24, and 36 months (incarceration data were only available up to 24 months).
The response rates for adults were 60 percent at Wave 2, 63 percent at Wave 3, and 68 percent at Wave 4. Approximately 45 percent of respondents had data available at all four waves.
Global Severity Index (GSI); Locus of Control; Readiness to Change; Self-Efficacy; Legal Cynicism scale