Download Analysis of Arrest Patterns Following SVORI Release: Factors Influencing Recidivism and more Schemes and Mind Maps Law in PDF only on Docsity!
The author(s) shown below used Federal funding provided by the U.S.
Department of Justice to prepare the following resource:
Document Title: Desistance from Crime over the Life Course
Author(s): Pamela K. Lattimore, Debbie Dawes, Kelle
Barrick
Document Number: 252080
Date Received: September 2018
Award Number: 2012-R2-CX-
This resource has not been published by the U.S. Department of
Justice. This resource is being made publically available through the
Office of Justice Programs’ National Criminal Justice Reference
Service.
Opinions or points of view expressed are those of the author(s) and
do not necessarily reflect the official position or policies of the U.S.
Department of Justice.
NIJ Grant Number 2012-R2-CX-
RTI Project Number 0213489
Desistance from Crime over the Life Course
Final Summary Report
September 2018
Prepared for
National Institute of Justice
810 7th^ Street NW
Washington, DC 20531
Prepared by
Pamela K. Lattimore
Debbie Dawes
Kelle Barrick
RTI International
3040 Cornwallis Road
Research Triangle Park, NC 27709
This resource was prepared by the author(s) using Federal funds provided by the U.S. Department of Justice. Opinions or points of view expressed are those of the author(s) and do not
Contents
Acknowledgements....................................................................................................................................... ii
Exhibits......................................................................................................................................................... iii
Abstract........................................................................................................................................................ iv
Problem and Purpose.................................................................................................................................... 1
Research Design............................................................................................................................................ 3
Findings ......................................................................................................................................................... 5
Differences between Recidivists and Desisters ........................................................................................ 8
Recidivism Gap Analyses........................................................................................................................... 9
Negative Binomial Recidivism Model Results ......................................................................................... 10
Patterns of Rearrest ................................................................................................................................ 11
Self-Reported Employment, Drug Use, Violence, and Crime.................................................................. 13
Qualitative Findings ................................................................................................................................ 14
Conclusions and Implications for Policy and Practice................................................................................. 16
References .................................................................................................................................................. 18
Appendix A: Additional Exhibits................................................................................................................ A-
Appendix B: Qualitative Analysis and Codes..............................................................................................B-
Exhibits
Exhibit 1. Subject characteristics at Wave 1 SVORI interview (approx. 30 days prior to release, 2004 -
Exhibit 2. Distributions of lifetime and post-SVORI arrests for the SC sample through December 2015. ... 7
Exhibit 3. Arrest history (prior to SVORI incarceration) for the SC sample (N = 479)................................... 7
Exhibit 4. Post-SVORI arrests for the SC sample (N = 479) ........................................................................... 8
Exhibit 5. Comparison of baseline characteristics of those who desisted and those who recidivated........ 9
Exhibit 6. Lognormal survival results for time to first, second, and third post-SVORI arrests ................... 10
Exhibit 7. Negative binomial results for number of post-SVORI arrests..................................................... 10
Exhibit 8. Distribution of post-SVORI arrests over 10 years’ follow-up following SVORI release ............... 12
Exhibit 9. Logistic model results for any arrest in years 1, 2 and 3 following SVORI release ..................... 13
Exhibit 10. Self-reported housing, employment, drug use, and violence at Waves 2, 3, and 4 SVORI
interviews and 2015-2016 interview, desistance sample only................................................................... 14
This resource was prepared by the author(s) using Federal funds provided by the U.S. Department of Justice. Opinions or points of view expressed are those of the author(s) and do not
iii
Abstract
P URPOSE : The Desistance from Crime over the Life Course study focused on 479 men and women
from South Carolina who were enrolled as participants in a multi-site reentry program evaluation shortly
before prison release in 2004-2005. The study goals were to (1) update information on current status
across multiple domains, including housing, employment, and substance use; (2) collect additional
recidivism data to examine long-term offending patterns; and (3) acquire information about factors
individuals associated with decisions to desist from further criminal activity.
R ESEARCH S UBJECTS : Between 2004 and 2005, 345 men, 79 boys, and 55 women incarcerated in
South Carolina were interviewed about 30 days before release from prison/juvenile detention for
inclusion in the multi-site evaluation of the Serious and Violent Offender Reentry Initiative (SVORI). The
male subjects (men and boys) were about 27 years and the female subjects were about 30 years of age at
the time of release. The men were most likely to be black (63%) while the women were equally likely to
be black or white (46%). The women were less likely than the men to report having been employed
during the 6 months prior to their original incarceration but were more likely to report having a high
school diploma or equivalent. Overall, the participants had extensive criminal histories and about 33%
reported having received drug treatment prior to the interview.
M ETHODS : Arrest records through December 31, 2015 were obtained from the South Carolina Law
Enforcement Division for all 479 subjects. The interview instrument included modules from the original
SVORI interviews, new modules focused on factors linked to emerging theories of desisting behavior,
and a life event calendar captured details of their experiences between their original release from prison
and the interview. Interviews were conducted between 9/10/2016 and 3/3/2017 with 208 individuals (
men and 34 women). Quantitative analyses included (1) descriptive analyses of interview data to identify
the current status of individuals; (2) lognormal survival, negative binomial, and logistic recidivism
analyses; and (3) qualitative analyses of responses related to the factors associated with criminal
persistence and desistance.
R ESULTS : Recidivism, measured as at least one new arrest, was experienced by 90% of the sample.
On average, individuals had about 7 arrests (with an average of 11 charges) after their SVORI
incarceration. Majorities of the sample had at least one post-SVORI charge for a person, property, and
public order/order offense, while 49% had at least one new drug charge. Results from sequential
lognormal survival models predicting time to first arrest, second arrest, and so forth following the original
release show those older at release had, on average, longer times to rearrest following the first arrest,
while property priors were associated with shorter times. Consistent with previous findings, those with
more prior arrests had shorter times to subsequent arrests and those who were older at first arrest had
longer times to subsequent arrests. Negative binomial model results show that having completed high
This resource was prepared by the author(s) using Federal funds provided by the U.S. Department of Justice. Opinions or points of view expressed are those of the author(s) and do not
iv
Problem and Purpose
Given that most prisoners return to society and soon re-engage in criminal activity, research on the
dynamic process of desistance from criminal activity among released prisoners is important to policy
makers and criminal justice practitioners. A better understanding of these processes could help to identify
more effective strategies to reintegrate offenders into the community while reducing the likelihood that
they return to criminal activity. Yet, the understanding of the process of desistance from crime is
underdeveloped.
Much of the focus of criminal interventions has been to address needs and deficits correlated with
criminal activity (e.g., Lattimore and Visher, 2014, MacKenzie, 2006; Visher, Lattimore, Barrick, &
Tueller, 2017). These deficits include drug use, mental health issues, and limited education and job skills.
Wraparound services, including transportation and housing, and help obtaining identification and licenses,
have also been included in the bundles of services provided by reentry programs hoping to prevent
returning prisoners from re-engaging in criminal conduct. Research suggests cognitive-behavioral
approaches and programs that target criminogenic factors and individual needs and focus on individual-
level change may be most effective at reducing recidivism among adults and juveniles (Andrews &
Bonta, 2003; Andrews et al., 1990; Aos et al., 2006; Fonagy & Kurtz, 2002; Lipsey, 1995; Lipsey &
Cullen, 2007; MacKenzie, 2006). In line with this research, Visher et al. (2017) show services associated
with individual needs (mental health and substance use treatment, assistance working on personal
relationships, training on changing criminal attitudes, anger management, and education) were more
likely to be associated with reduced recidivism than more practical services (case management, needs
assessment, reentry planning and programming, life skills, and employment services) (also see Lattimore,
Barrick, Cowell, Dawes, Steffey & Tueller, 2012).
Criminologists have proposed that the factors associated with desistance from criminal activity may
differ from factors associated with ongoing engagement. For example, identity transformation (from the
criminal to the noncriminal) has been suggested as a necessary first step away from criminality (e.g.,
Giordano, Cernkovich, & Rudolph, 2002; Paternoster & Bushway, 2009). Absent this transformation,
programming that focuses on structural or instrumental factors like employment skills and assistance
finding housing or transportation may be of limited use in reducing recidivism.
For example, Giordano and colleagues place an emphasis on human agency and stress the
importance of individual identity and cognition in explaining desistance, proposing a theory of cognitive
transformation grounded in symbolic interactionism (Giordano, et al., 2002, 2007). Paternoster and
Bushway (2009) build on this and other early work with a theory of multifaceted identity that includes a
“working self” and “possible self.” Individuals commit to the working (criminal) self until its costs
outweigh its benefits—gradual change in identity occurs as failure and dissatisfaction are linked to the
This resource was prepared by the author(s) using Federal funds provided by the U.S. Department of Justice. Opinions or points of view expressed are those of the author(s) and do not
working self. Once the criminal identity is weakened, identity change is possible. As described by
Paternoster and Bushway (2009, p. 1105), the “perceived sense of a future or possible self as a non-
offender coupled with the fear that without change one faces a bleak and highly undesirable future
provides the initial motivation to break from crime.” After the criminal identity is weakened and a non-
offender possible self is considered, then the individual might move toward conventional institutions such
as employment and marriage.
Some evidence in support of the identity theory of desistance (ITD) emerged from a study examining
the long-term recidivism of drug-using offenders (Bachman, Kerrison, Paternoster, O’Connell & Smith,
2016 ; also see Na, Paternoster, & Bachman, 2015; Paternoster, Bachman, Kerrison, O’Connell, & Smith,
2016). Interviews were conducted with 304 individuals who participated in drug treatment demonstration
projects in 1989 and 1990 approximately 20 years after the original studies. The interview data were
qualitatively analyzed and the results suggest that offenders who desisted from crime and substance use
underwent an identity transformation that was motivated by the realization that they needed to change to
avoid an undesirable future, such as dying in prison.
Na, Paternoster, and Bachman (2015) used longitudinal data from the same cohorts of individuals
followed by Bachman and colleagues to examine the role of self image and efforts to improve self (as
measured by treatment seeking). They found support in the results of their growth-curve models for these
factors to be associated with long-term desistance from substance use and crime (measured by arrest).
Using different analytic approaches to examine ITD with the same data, Paternoster and colleagues
(2016) subsequently found that postitive identity and treatment-seeking was positively and significantly
related to longer time to rearrest.
Additional evidence is suggested in interview data collected for the Multi-site Evaluation of the
Serious and Violent Offender Initiative (SVORI)^1 (see, e.g., Lattimore and Visher, 2009). During follow-
up interviews (3, 9, and 15 months post release from incarceration associated with study enrollment),
respondents were asked to identify reasons why they were newly incarcerated—if they were
incarcerated—and why they were no longer engaging in criminal behavior—if they were not incarcerated.
The responses suggested clear differences between the two groups with many of the incarcerated
respondents ‘blaming’ others for their continued engagement in crime and subsequent incarceration and
those who were not incarcerated pointing to changes in themselves. For example, the most common
(^1) The Multi-site Evaluation of the Serious and Violent Offender Reentry Initiative was supported by NIJ
grants 2003-RE-CX-K101 and 2004-RE-CX-002; see, e.g., The Multi-site Evaluation of SVORI: Summary and Synthesis (December 2009) by P.K. Lattimore and C.A. Visher available from the National Criminal Justice Reference Service https://www.ncjrs.gov/pdffiles1/nij/grants/230421.pdf. Prisoner Reentry Services: What Worked for SVORI Evaluation Participants was funded by NIJ grant 2009-IJ-CX-0010; see, e.g., Lattimore, P.K., Barrick, K., Cowell, A., Dawes, D., Steffey, D., & Tueller, S. (April 2012). Prisoner Reentry Services: What Worked for SVORI Evaluation Participants (https://www.ncjrs.gov/pdffiles1/nij/grants/238214.pdf).
This resource was prepared by the author(s) using Federal funds provided by the U.S. Department of Justice. Opinions or points of view expressed are those of the author(s) and do not
For the current study, a new interview instrument was developed that included modules from the
original evaluation, new modules that focused on factors linked to emerging theories of desisting
behavior, and a life event calendar to capture details of their experiences between their original release
from prison (2004 or 2005) and the date of the interview. The instrument covered a number of domains
over the life-event calendar period, including basic demographic information, education, attitudes (e.g.,
legal cynicism), receipt of programs and services, family (marriage, children, and intimate relationships),
physical and mental health, criminal identity, peers, social support, location and living arrangements,
employment and income, leisure activities, stressful life events, substance use, and avoided and
committed criminal behavior. Although most items had closed responses, several were open-ended.
Extensive tracing efforts were used to locate individuals for interviews that were conducted between
September 10, 2016 and March 3, 2017. Of the original 479 study participants, 29 were confirmed to have
died^4 ; 24 had moved from South Carolina^5 ; and 34 were unavailable during the interview period^6. We
were unable to contact (locate) 149 individuals. There were 30 refusals by respondents and 3 refusals by
others for potential respondents. Two interviews were terminated by the respondent before completion
(consent withdrawn) and were lost. Thus, 208 individuals were successfully located and completed the
interview (171 in the community and 37 in prison)^7. Of the 208, 174 were male (24 of whom were
juvenile males at the time of their original release) and 34 were female. Interviews were conducted by
experienced and trained field interviewers using laptop computers. The field interviewer read a series of
questions from the laptop screen and entered the respondent’s answers. For some questions, individual
responses were audio recorded for subsequent transcription.
Administrative arrest data through December 31, 2015 were obtained from the South Carolina Law
Enforcement Division (SLED). Analyses comparing characteristics at the time of the original study
release showed no significant differences between those who responded to the new interview and those
who did not respond.
Quantitative analyses included (1) descriptive analyses of interview data to identify the current status
of individuals; (2) recidivism analyses; and (3) qualitative analyses of responses related to the factors
associated with criminal persistence and desistance. Recidivism analyses included negative binomial
(^4) Deaths were confirmed for 2 boys (at initial SVORI release), 24 men, and 3 women. (^5) The structure of the interview required that it be administered in person so individuals out of state were
not contacted for a potential telephone interview. (^6) Of those unavailable to be interviewed, eight were incarcerated out of state (or in one case, in Federal
custody), one was in segregation in a SC prison, and two were in jail. (^7) The overall response rate was 46.2% (208 interviews of 450 eligible—nondeceased—respondents). There
was variability in response rates across the demographic groups—the response rate was 43.7% for the male participants (150 of 321 or 46.7% for men at SVORI release; 24 of 77 or 31.1% for juvenile males at SVORI release) and 65.4% for females (34 of 52 female subjects).
This resource was prepared by the author(s) using Federal funds provided by the U.S. Department of Justice. Opinions or points of view expressed are those of the author(s) and do not
models examining the number of arrests following the original release (with offset for prison
incarceration and time at risk), lognormal survival models examining the factors associated with the time
to rearrest for each new arrest following the original release up to 10 new arrests (“gap analyses”); and
graphical analyses that plotted number of arrests each year following the original release and
accompanying logistic regression models that examined factors associated with rearrest in each of 10
years following the original release.
The qualitative analyses were based on the short-answer items that were either manually entered into
the computer by the interviewer or audio-recorded and then transcribed. The open-ended questions
covered: (1) reasons for avoiding criminal behavior in at least the past 6 months; (2) reasons for avoiding
criminal behavior in periods in which they were tempted to commit crime; (3) life events that occurred
during periods when they were less likely to engage in criminal behavior; and (4) life events that occurred
during periods when they were more likely to engage in criminal behavior. Audio recordings of
interviews were transcribed and merged with those that were manually typed. Qualitative content analysis
was used to examine themes in the short response items. An initial set of codes was developed and
additional codes were added after iteratively reading the transcripts. (See Appendix B.)
Findings
Characteristics of the subjects at the time of their original SVORI interview are shown in Exhibit 1.
The original sample of 479 included 424 men and boys and 55 women. For this study, the two male
samples were combined so the men were somewhat younger then the women at release because of the
inclusion of the boys with the adult men. The men were most likely to be black (63%) while the women
were equally likely to be black or white (46%). The women were less likely than the men to have reported
being employed during the 6 months prior to their original incarceration (53% v. 65%), but were more
likely to have reported having a high school diploma or equivalent (66% v. 39%). Overall, the sample had
an extensive criminal history and 36% were classified as high risk; 33% reported having had drug
treatment prior to the original interview.
This resource was prepared by the author(s) using Federal funds provided by the U.S. Department of Justice. Opinions or points of view expressed are those of the author(s) and do not
Exhibit 2. Distributions of lifetime and post-SVORI arrests for the SC sample through December 2015.
0
2
4
6
8
10
12
0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 45 60
% of Sample
Number of Arrests
Lifetime Arrests Post-SVORI Arrests
Note: Two juvenile males had zero arrests, reflecting original confinement on a juvenile detainer and no subsequent recidivism arrests.
Exhibit 3 shows additional information about arrest history prior to the SVORI incarceration. On
average, individuals had experienced nearly 8 arrests (14 charges) prior to their SVORI incarceration.
Most had at least one prior charge for each of a person, property, drug, and public order/order offense.
Exhibit 3. Arrest history (prior to SVORI incarceration) for the SC sample (N = 479)
Variable Mean SD Number arrest events 7.570 6. Has a prior person charge=1 0.653 0. Has a prior property charge=1 (^) 0.666 0. Has a prior drug charge=1 (^) 0.557 0. Has a prior public order/other charge=1 0.704 0. Number prior charges 13.818 12. Number prior person charges (^) 2.144 2. Number prior property charges (^) 5.557 8. Number prior drug charges 2.159 3. Number prior public order/other charges 3.958 4.
Exhibit 4 shows information about arrests following the SVORI incarceration. Ninety percent had
been arrested at least once. On average, individuals had nearly 7 arrests (with an average of 11 charges)
This resource was prepared by the author(s) using Federal funds provided by the U.S. Department of Justice. Opinions or points of view expressed are those of the author(s) and do not
after their SVORI incarceration. Most had at least one post-SVORI charge for each of a person, property,
and public order/order offense, while 49% had at least one new drug charge.
Exhibit 4. Post-SVORI arrests for the SC sample (N = 479)
Variable Mean SD Number arrest events 6.900 6. Has an arrest=1 0.900 0. Days to first arrest (^) 531.622 596. Has a person charge=1 (^) 0.520 0. Has a property charge=1 0.601 0. Has a drug charge=1 0.491 0. Has a public order/other charge=1 (^) 0.810 0. Number of charges (^) 10.923 9. Number of person charges 1.382 2. Number of property charges 3.271 5. Number of drug charges (^) 1.390 2. Number of public order/other charges 4.879 5.06 4
Differences between Recidivists and Desisters
Ninety percent of our sample experienced at least one arrest following their SVORI release in 2004-
2005. There are substantial differences between those with no new arrests (desisters) and those with one
or more (recidivists). Exhibit 5 presents results on several key measures. The desisters were older at
release, more likely to be white (less likely to be black), and more likely to have completed high school
(or equivalent). About 88% of both the desisters and the recidivists were male.
A criminal past was definitely a prologue for these subjects. Desisters were considerably older at
first arrest (24 on average versus 16) and had fewer prior arrests (5 versus 8), arrest charges (9 versus 14)
and convictions (2.7 versus 4.7)^8. Desisters had fewer juvenile detentions and were less likely to have
been serving time for a probation/parole violation. There were also significant differences in conviction
offenses—desisters were much more likely to have been serving time for a person/violent offense and
much less likely to have been serving time for a property offense. Fewer desister were classified as high
risk on the LSI-R (12.5% versus 39%). The desisters were also less likely to have reported receiving
substance abuse treatment (21% versus 34%).
(^8) Desisters were also less likely to have any prior charges for property, drug, and public order/other
offenses and to have fewer of these charges on average. While they were more likely to have a prior person/violent offense, the mean number of person/violent charges was less for the desisters than the recidivists. (Data not shown.)
This resource was prepared by the author(s) using Federal funds provided by the U.S. Department of Justice. Opinions or points of view expressed are those of the author(s) and do not
Exhibit 6. Lognormal survival results for time to first, second, and third post-SVORI arrests
Variable First Arrest Second Arrest Third Arrest Estimate St. Err Estimate St. Err Estimate St. Err (Intercept) (^) **4.84***** 0.484 **4.854***** 0.708 **4.401***** 0. Race=white (^) 0.222 0.16 - 0.03 0.214 0.044 0. Employed 6M prior to incarceration (^) 0.104 0.154 - 0.093 0.203 0.247 0. Completed 12th^ grade or equivalent 0.235 0.154 0.033 0.205 0.138 0. Age at release (^) 0.008 0.014 **0.059**** 0.019 0.027 0. Conviction offense=Person/Violent 0.334 0.2 - 0.059 0.263 - 0.009 0. Conviction offense=Property (^) **- 0.575**** 0.195 - 0.077 0.255 - 0.209 0. Conviction offense=Drug (^) 0.094 0.207 0.094 0.272 0.037 0. Conviction offense=Public Order/Other (^) - 0.1 0.19 - 0.15 0.248 - 0.209 0. In for probation/ parole violation (^) - 0.033 0.159 - 0.02 0.21 0.02 0. Times juvenile lockup (^) - 0.086 0.053 - 0.06 0.069 0.019 0. Total prior arrests (^) **- 0.026***** 0.007 **- 0.033***** 0.009 **- 0.04***** 0. Male=1 - 0.206 0.235 - 0.449 0.318 - 0.207 0. Age at first arrest (^) **0.077***** 0.018 0.012 0.028 0.056 0. **Log(scale) 0.379***** 0.035 **0.612***** 0.038 **0.57***** 0. N 460 417 373 Note: *** p < .001, ** p < .01, * p <.
Negative Binomial Recidivism Model Results
Results from the negative binomial model are shown in Exhibit 7. The dependent variable is the
number of arrests (0 to 33) following the SVORI-related release and predictor values are from the SVORI
pre-release interview. Completing high school (or the equivalent) predicts fewer post-SVORI arrests,
while being younger at first arrest and having more priors predict more post-SVORI arrests.
Exhibit 7. Negative binomial results for number of post-SVORI arrests
**Variable Incident Rate Estimate Std. Error z value (Intercept) 0.004***** - 5.49 7 0.31 8 - 17. Race=white 1.026 0.02 6 0.10 3 0.25 10 Male=1 0.932 - 0.070 0.15 1 - 0. Age at release 0.986 - 0.01 4 0.00 9 - 1. Completed 12th^ **grade or equivalent 0.698***** - 0.35 9 0. 100 - 3.601 1 Employed 6M prior to incarceration 1.132 0.124 0.098 1. **Age at first arrest 0.945***** - 0.05 7 0.012 - 4.546 4 Times juvenile lockup 1.033 0.032 0.03 3 0. **Total prior arrests 1.027***** 0.02 7 0.004 6.516 4 Conviction offense(s)=Person/Violent 0.837 - 0.178 0.12 7 - 1. Conviction offense(s)=Property 1.147 0.137 0.12 4 1.105 9 Conviction offense(s)=Drug 0.96 - 0.041 0.13 2 - 0. Conviction offense(s)=Public Order/Other 1.137 0.12 9 0.1 20 1.074 3 In for probation/ parole violation 1.037 0.03 7 0.10 1 0.36 40 Note: *** p < .001, ** p < .01, * p <.
This resource was prepared by the author(s) using Federal funds provided by the U.S. Department of Justice. Opinions or points of view expressed are those of the author(s) and do not
Patterns of Rearrest
The patterns of rearrest were also examined using an innovative method for visualizing event
patterns and more traditional logistic regression. Exhibit 8 shows the number of arrests each year
following the SVORI release for each individual in the sample. (The graph is composed of stacked
horizontal lines with each line corresponding to one individual. Each line shows the number of arrests
each year following release with a gradient of color—darker segments representing more arrests during
the year; see Tueller, Van Dorn, & Bobashev, 2016.) Death events are indicated by an “o” and the lines
for the individuals who died during follow-up are at the bottom of the chart.
The dark orange lines at the top of the chart for the initial 1-year period indicate that some
individuals had many arrests (up to 9) in the year following release. About half of the sample had no
arrests during that first year as is indicated by the large number of lines that are light yellow during the
initial 1-year period. The 10% who experienced no new arrests for the full 10-year period following their
SVORI release are the group of lines at the bottom of the chart (above the 29 who died during the follow-
up period). The overall graphic suggests that the pattern of arrest events varies considerably across these
individuals—e.g., ranging from the desisters (no arrests in any year) to those with many arrests early on to
those with high frequencies of arrests throughout the 10-year period.
To examine arrest patterns more closely, a series of logistic models were estimated to examine the
factors predicting any arrest each year. Exhibit 9 shows the first three years’ results. (Results for all 10
years are in Exhibit A-3 .) Age and criminal history were most often predictive of an arrest, consistent
with expectations.
This resource was prepared by the author(s) using Federal funds provided by the U.S. Department of Justice. Opinions or points of view expressed are those of the author(s) and do not
Exhibit 9. Logistic model results for any arrest in years 1, 2 and 3 following SVORI release
Variable Year 1 Year 2 Year 3 Odds Ratio
Estimate St. Err Odds Ratio
Estimate St. Err Odds Ratio
Estimate St. Err
(Intercept) (^) 0***** - 7.989***** 1.451 0.002 - 6.256***** 1.279 0.001 - 7.558***** 1. Race=white (^) 1.42 0.351 0.522 2.539 0.932 0.406 0.893 - 0.113 0. Employed 6M prior to incarceration 0.585 - 0.535 0.53 1.416 0.348 0.454 0.488 - 0.718 0. Completed 12th grade or equivalent 0.837 - 0.178 0.596 0.805 - 0.216 0.429 3.093 1.129**** 0. Age at release (^) 0.934 - 0.068 0.053 1.079 0.076** 0.034 1.054 0.053 0. Conviction offense= Person/Violent 2.392 0.872 0.618 2.068 0.727 0.536 1.349 0.299 0. Conviction offense= Property 0.966 - 0.035 0.569 1.203 0.185 0.467 1.15 0.14 0. Conviction offense= Drug 1.327 0.283 0.564 3.171 1.154** 0.539 1.838 0.609 0. Conviction offense= Public Order/Other 1.11 0.104 0.553 0.521 - 0.652 0.646 0.699 - 0.359 0. In for probation/ parole violation 1.292 0.256 0.421 1.146 0.136 0.44 0.801 - 0.221 0. Times juvenile lockup 0.86 - 0.151 0.14 0.981 - 0.019 0.129 0.8 - 0.223 0. Total prior arrests (^) 1.017 0.017 0.019 0.957 - 0.044** 0.022 0.973 - 0.027 0. Male=1 (^) 0.837 - 0.178 0.621 0.103* - 2.278***** 0.605 0.643 - 0.441 0. Age at first arrest (^) 1.197 0.18**** 0.064 0.965 - 0.036 0.049 1.054 0.052 0. N with no arrests (^) 234 227 257 N with any arrest (^) 198 205 175 Note: *** p < .001, ** p < .01, * p <.
Self-Reported Employment, Drug Use, Violence, and Crime
The in-person interviews conducted in 2015-2016 (N = 208) covered a variety of domains. Results
are reported here for key domains—employment, self-reported drug use, and self-reported violence and
crime ( Exhibit 10 ). Results from this interview are compared to results for the same sample from the
original SVORI follow-up interviews. (Results comparing the full SC sample to the 2016 sample are in
Exhibit A-4 .)
Individuals were somewhat less likely to be working in 2015-2016 than they were immediately
following their SVORI release—about two-thirds reported currently working in the 2015-2016 interview
compared to about 80% who reported working in 2005-2007. In reports on their behavior since their
SVORI release in 2004-2005, respondents were more likely to report drug use and having engaged in
criminal behavior over this much more extensive period than during the 3-15 months since their SVORI
release. Only 38% reported no illegal drug use post-SVORI, although nearly two-thirds (66%) said they
This resource was prepared by the author(s) using Federal funds provided by the U.S. Department of Justice. Opinions or points of view expressed are those of the author(s) and do not
had used no illegal drugs other than marijuana. In contrast, most respondents in the earlier interviews
reported no illegal drug use of any kind. Although fully 90% had experienced at least one post-SVORI
arrest, most (59%) said that they had not engaged in any criminal behavior since that release. A similar
percentage (18%) was incarcerated at the time of the 2016 interview as the Wave 3 and 4 interviews.
Exhibit 10. Self-reported housing, employment, drug use, and violence at Waves 2, 3, and 4 SVORI
interviews and 2015-2016 interview, desistance sample only
Wave 2 (2004 2006) (N = 151)
Wave 3 Wave 4 (2005 2006) (2005 2007) (N = 151) (N = 174)
2016 Interview ( 2015 2016) (N = 208) Mean (Standard Deviation) Currently employed in legal job (^) 0.70 2 (0.45 9 ) 0.8 10 (0.394) 0.80 3 (0. 400 ) 0.678 (0.46 9 ) No self-reported drug use 0.768 (0.423) 0.589 (0.49 4 ) 0.57 9 (0.495) 0.3 80 (0.48 7 ) No self-reported drug use other than marijuana or steroids 0.907 (0.29 1 ) 0.79 5 (0.405) 0.789 (0.409) 0.663 (0.47 4 ) No perpetration of violence 0.735 (0.44 3 ) 0.576 (0.49 6 ) 0.64 4 (0.480) 0.745 (0.43 7 ) No criminal behavior 0.90 7 (0.29 2 ) 0.850 (0.358) 0.738 (0.44 2 ) 0.58 7 (0.49 4 ) No violent or weapons crimes 0.834 (0.37 3 ) 0.675 (0.4 70 ) 0.68 4 (0.466) 0.716 (0.45 2 ) Not reincarcerated at follow-up 0.91 4 (0.281) 0.821 (0.384) 0.82 2 (0.38 4 ) 0.822 (0.383)
Qualitative Findings
Embedded within the interview were opportunities to elicit open-ended responses to provide insight
into respondents’ perceptions concerning factors associated with engaging in or desisting from criminal
behavior. Qualitative content analysis was used to evaluate these responses in an effort to understand
motivations related to desistance and continued criminal activities. Of particular interest were responses
related to identity and identity transformation.
To probe why individuals avoided criminal behavior, interviewees were asked whether they had
engaged in robbery, property crimes, prostitution, drug dealing, intimate partner violence, and other
violence during the approximately 10 years following their SVORI incarceration release. When
individuals responded ‘yes,’ they were asked a series of follow-up questions, including the timing and
frequency of involvement in these activities.
The 112 interviewees who indicated that they had not engaged in any of these criminal behaviors
during at least the past 6 months were asked to tell the interviewer why they had not committed a crime
recently. One of the most common reasons cited involved incarceration having a deterrent effect. About
one-third of the respondents indicated that they had been deterred from committing more crimes. For
example, one respondent indicated that, “I learned my lesson.” while another said, “… if you commit
crimes you will go to jail, you will pay the price, it is wrong.”
This resource was prepared by the author(s) using Federal funds provided by the U.S. Department of Justice. Opinions or points of view expressed are those of the author(s) and do not
