Key Takeaways
In 2005, the California Department of Corrections and Rehabilitation (CDCR) was given a rehabilitative mission, which included providing programs to reduce recidivism by addressing the needs of imprisoned people. This report examines how effectively the agency has delivered on this mission by describing the education, employment, and rehabilitative programs available to people released from prison between 2015 and 2019, assessing program participation, and mapping underlying trends in recidivism outcomes.
- About 1 in 1000 people living in California in any year were released from prison during that year; 64 percent of over 37,000 people released in 2019 had participated in at least one prison program.→
- People entered prison with significant needs that reflect their pathways into prison. On average, people read at the eighth-grade level and had sixth-grade math skills. Over two-thirds needed substance use disorder treatment. More than four in ten needed cognitive behavior interventions and employment programs.→
- Program capacity expanded dramatically between 2015 and 2019. Participation rates increased across all program areas—doubling or tripling for some—but remained below an ambitious goal of 70 percent.→
- Programs faced challenges with accurate targeting. At most, 43 percent of people who initially tested into an education program and 21 percent of people with an assessed substance use need participated.→
- Recidivism rates are lower than before these reforms but remain high. Over half of those with prior prison histories and 43 percent of first-timers were rearrested for felonies. Likewise, 28 percent and 21 percent, respectively, were reconvicted for felonies within two years. Though we do not attempt to connect recidivism to program participation, stubbornly high recidivism rates indicate the need to enhance prison programs to improve reentry pathways.→
Based on information that has not been available before, we offer context for the role of prison programs in people’s lives by describing pathways into prison, investigating factors that affect program participation, and setting the stage to trace post-prison pathways. Each of these elements highlights aspects of people’s lives before, during, and after prison that determine which programs are available to and appropriate for them.
Introduction
After the California Department of Corrections was rebranded the California Department of Corrections and Rehabilitation (CDCR) in 2005, a panel of experts was convened in 2006 to help CDCR realize its new rehabilitative mission. The panel recommended expanding program offerings and targeting programs to those with assessed need, with the aim of reducing recidivism by building social, job, and educational skills. Ten years later, two oversight agencies questioned whether CDCR had achieved these aims.
The Legislative Analyst’s Office (LAO) and the state auditor found that people who needed in-prison programs often did not receive them. Even when they did, the programs did not achieve their aims, which include building life skills, changing during- and post-prison behavior, and reducing reoffending (Taylor 2017; Howle 2019). Both agencies recommended an independent evaluation of prison programs, which CDCR enlisted PPIC to conduct.
The problems LAO and the auditor identified are not unique to California. A 2016 national survey of imprisoned people showed that less than half had participated in education programs and only about one third had participated in job training programs (PPI 2022). Other state and federal prison systems also struggle to assess people’s needs and provide programs that address them (Latessa et al. 2015; Byrne 2020; Russo et al. 2020). Federal prisons did not have processes in place to assess need until recently. Even in states like Ohio, which assesses needs, the programs offered do not target them with evidence-based interventions (Latessa et al. 2015).
Prison programs have not been rigorously evaluated to determine whether they address the problems that landed people in prison and promote successful reentry pathways. Most evaluations are at least a decade old, employ weak evaluation strategies, and examine small samples for short follow-up periods (Byrne 2020; Duwe 2017; Nur and Nguyen 2022). We illustrate how we will overcome these limitations and lay the groundwork for evaluating prison programs in California by providing previously unavailable information about people released from prison, the prison programs that were available to them, their participation, and recidivism trends.
We describe all in-prison programs administered by CDCR between 2014 and 2019. During this time, California prisons offered programs in three areas: education, employment, and rehabilitative programs. Education programs span adult basic education through college. Employment programs include career technical education (CTE) and job readiness programs. Rehabilitative programs include substance use disorder treatment and cognitive behavior interventions that address assessed needs of imprisoned people, such as managing anger, improving relationships, and understanding victims.
California prisons—and prison programs—touch a substantial share of Californians
About 1 in 1000 people living in California each year were released from prison in that year. Between 2015 and 2019, California prisons released 168,482 people—between 35,000 and 40,000 individuals each year, as shown in Figure 1. Ten percent were released two or more times, for a total of 185,310 releases. Most people participated in at least one program in prison. In 2015, 46 percent of those released had participated in at least one education, employment, or rehabilitation program. By 2019, 64 percent had participated.
Each year between 35,000 and 40,000 people return to California communities from prison
Number of people released from California prisons by year
SOURCE: Author calculation from CDCR administrative data.
NOTES: This figure shows the number of people released from prison in California each year from 2015 to 2019. If a person was released twice within a year, we only count their first release within that year, but people can be counted twice over the five years. These “first-in-year” releases number 185,007. See Technical Appendix Table B1 for more information.
People released from prison are more likely to be male, middle-aged, and Black or Native American
Average characteristics of released prisoners diverge from those of California’s population (Technical Appendix Figures B1 and B2). Released people are far more likely to be male. While just over half of the state population was female, just 7 percent of people released from prison were.
Most released prisoners are in young adulthood or early middle-age. Between 2015 and 2019, people left prison at an average age of 37 years. By 2019, the share between the ages of 25 and 44 had increased to nearly two-thirds of those released. By contrast, just one-third of Californians are similarly aged.
Black and Native American people are starkly overrepresented. Though Black people were 6 percent of California’s population from 2015 to 2019, they were 25 percent of released prisoners—a fourfold differential. Similarly, the share of Native American people released from prison held at triple that of the state’s population.
This report lays the groundwork necessary to evaluate California prison programs by providing previously unavailable information about formerly imprisoned people, including their prior justice system involvement, the programs available to them in prison, and trends in recidivism. The next section discusses some factors that affect program participation: pathways into prison, program availability, and time imprisoned. We then describe participation in the main education, rehabilitative, and employment programs. The last two sections chart recidivism trends and discuss our findings.
Factors That Shape Program Participation
Whether people can participate in programs depends on many factors—and we can touch on only a few in this report. Between 2015 and 2019, the composition of the released population shifted toward a greater share of those imprisoned for the first time—most of whom had committed violent crimes—and people served increasingly longer sentences, which may have given them more time to engage with programs. Starting in 2014, programs became increasingly available across all prisons in the system as the state worked to provide rehabilitative opportunities to 70 percent of eligible people.
Pathways into Prison
We characterize the criminal histories of formerly imprisoned people to investigate whether CDCR offered programs to the people who needed them and whether other types of programs may be needed. In general, we can think about people entering prison along two broad pathways: those imprisoned for the first time and those returning to prison. By assessing similarities and differences in the prior circumstances of these two groups, we can suggest potential areas for program intervention. For instance, higher prevalence of certain types of crimes, such as domestic violence, can highlight a need for prison-based interventions to mitigate them (Hasisi et al. 2016; Stansfield et al. 2022).
Half of those released had been imprisoned previously
About half of released prisoners (53%) had spent time in prison earlier in their lives; about 10 percent had been sent to prison three or more times. However, the early years of the cohort were more likely to include prior prisoners: in 2015, 57 percent of those released had been imprisoned before; by 2019, 51 percent had been (Figure 2).
Most people recently committed violent crimes—and domestic violence was prevalent
In the three years before prison, the cohort members were arrested an average of five times. Three of those arrests included felony charges. Cohort members were convicted twice, on average, and once for a felony. By contrast, 8 percent of the sample (about 13,200 people) had not been convicted and 1 percent (about 2,200 people) had zero arrests in the three years before imprisonment.
Most people released from California prisons have violence in their recent pre-prison past. Forty-seven percent had been convicted of a violent felony in the three years before prison. Assaults were by far the most common type of violent crime; 57 percent of violent convictions were assaults. Domestic violence is the most common type of assault; at least 7 percent of the 2015–19 cohort was imprisoned for domestic assault.
Violence was more likely for first-timers than for those previously imprisoned
People serving their first prison term were more likely to have felony arrests and convictions for violent crimes (Figure 3). Nearly three-quarters of first-timers, but just under half of those with prior prison experience, had been arrested for a violent crime. Similarly, 59 percent of first-timers, but 35 percent of those previously imprisoned, had been convicted of a violent felony in the three years before being imprisoned.
Most released prisoners were arrested for and convicted of violent felonies in the three years before prison
Violence within three years before prison, by prison history
SOURCE: Author calculation from CDCR and DOJ administrative data.
NOTES: N= 167,861 people with DOJ data at first release. Controlling offenses are those that carry the longest sentence. Our definition of a violent felony differs from the Penal Code. See the “Description of the data” text box for more information.
By contrast, rates of prior domestic violence are similar for both groups: less than 1 percentage point separates their felony domestic violence conviction rates. Therefore, rates of domestic violence hold steady as criminal histories lengthen, whereas other forms of violence abate.
Time Spent in Prison
How long people spend in prison also influences participation in programs. People who serve shorter sentences may have fewer opportunities to participate in programs due to timing and scheduling constraints. By contrast, people who serve longer sentences may have greater opportunities to participate in programs, many of which aim to help people overcome the deeper job, family, and health-related challenges related to long-term imprisonment, such as extended absence from the labor market, dwindling social support, and inevitable aging processes.
Two-thirds of people were imprisoned for less than two years
On average, the release cohort members spent three years behind bars. However, the length of time people spent in prison varied widely, as shown in Figure 4. Fourteen percent of those released were imprisoned for six months, whereas 15 percent were imprisoned for five or more years.
About half of those released from prison spent between six months and two years behind bars
Time imprisoned, 2015–2019
SOURCE: Author calculation from CDCR administrative data.
NOTES: N=185,292 releases. 18 people sentenced to life or life without parole excluded. Percentages in bars indicate the share of releases involving people who had served the specified length of time.
People typically serve only a portion of their prison sentence
Most people who leave California prisons have served about 60 percent of their sentence (Figure 5). Section 2933 of the Penal Code, which was enacted in response to Expert Panel recommendations to provide people with incentives to work toward rehabilitation, allows prisoners to earn “good conduct credits” that can cut their sentences in half (Technical Appendix A).
Sentences have lengthened—which means people are spending more time in prison
California’s criminal justice reforms have changed why people go to prison. Returning to prison after a parole violation became less likely after realignment. Some felony property and drug crimes were downgraded to misdemeanors when Proposition 47 passed in 2014 (Technical Appendix A). As a result, people increasingly serve more time in prison after serious and violent felony convictions that carry longer sentences, as prescribed by the Penal Code of California. People released in 2015 spent an average of 35 months in prison, compared to 39 months for those released in 2019. Likewise, people released in 2015 were sentenced to 57 months, whereas those released in 2019 had been sentenced to 64 months (Figure 5).
Term lengths have lengthened because sentences have lengthened
Time sentenced and served by release year
SOURCE: Author calculation from CDCR administrative data.
NOTES: N=185,292 releases. 18 people sentenced to life or life without parole excluded. Zero sentences (n=137) were changed to one year.
People imprisoned for longer periods have higher program participation rates
People who spend less than six months in prison are least likely to participate in programs. Even those imprisoned for up to one year have lower participation rates than those imprisoned longer, especially in rehabilitative and employment programs (Figure 6). As described below, people typically receive education programs first, which may help explain higher participation rates in the education area. Options to work independently (rather than in a classroom setting) also contribute to higher participation rates in education programs, as shown in Technical Appendix Figure C6.
People who spend less time in prison are less likely to participate in programs
Program participation by time served and program area
SOURCE: Author calculation from CDCR administrative data.
NOTES: N=185,310. Education programs presented in this figure include primary and secondary general population and voluntary education programs, but do not include college. Rehabilitative programs include all core and specialized programs. Employment programs include CTE and Transitions.
CDCR policy may have constrained program access for those serving shorter sentences
CDCR guidelines may have inadvertently limited access to programs for people serving shorter sentences. During our study period, the Division of Rehabilitative Programs (DRP) developed a “lifecycle” that illustrated ideal program timing for a person serving five years, which is the average sentence length. The lifecycle prioritized education programs, which people should enter soon after intake. Thereafter, estimated time to release matters: people who are within one year of release receive priority to enroll in employment and rehabilitative programs. In this respect, the lifecycle seems to have worked as designed: 75 percent of people entered programs in the rehabilitative area within one year of their release date (Technical Appendix Figure C5).
Program Availability
Whether people can participate in programs depends on program availability: where programs are offered across the prison system, during which years, and their capacity during those years shapes how many and which people have access to them. By 2019, all CDCR-administered programs became available in all prisons, and program capacity had increased in all program areas. However, most programs had lengthy waitlists.
Program availability varied by program, by prison, and across years
Program availability varied by prison between 2014 and 2019. For instance, primary and secondary education was available everywhere; community colleges increased the availability of college courses over time. Similarly, rehabilitative programs expanded from the 13 reentry hubs throughout the system. In addition, the capacity of programs across program areas expanded as the legislature increased funding for programs and CDCR worked to meet staffing and participation goals set in the Blueprint (Technical Appendix A).
Between 2014 and 2019, capacity for primary and secondary education programs increased 9 percent—from enough slots to educate 30 percent of the 2015 prison population at once to enough slots for 35 percent of the 2019 population. Budgeted capacity more than tripled for core rehabilitative programs and employment programs—from enough slots for 8 percent of the prison population to participate in 2015 and 30 percent to participate in 2019 (Figure 7).
CDCR dramatically expanded the capacity of programs between 2014 and 2019
Program capacity by area
SOURCE: 2014–2019 Annual C-ROB reports.
NOTES: Slots for education programs include adult basic education and GED courses and exclude college courses. Employment includes CTE and Transitions. Rehabilitation programs include substance abuse treatment, criminal thinking, anger management, and family relationships programs. Slots for education programs include adult basic education and GED courses and exclude college courses. Employment includes CTE and Transitions. Rehabilitation programs include substance abuse treatment, criminal thinking, anger management, and family relationships programs. Rehabilitation and Transitions programs have finite schedules; capacity for these programs reflects the maximum number of people who participate each year. Education programs and CTE programs are more open-ended; capacity for these programs is the maximum number of people who can enroll at once.
Education Programs
Within DRP, the Office of Correctional Education (OCE) administers education programs, which range from adult basic education (i.e., elementary education for adults) through college courses and include supportive programs for those with special needs.
A recent study that synthesized research on correctional education programs in the United States indicates that they can reduce recidivism and increase employment. Incarcerated people who received education programs were 32 percent less likely to recidivate and 12 percent more likely to find employment after release, according to a synthesis of the results of higher-quality research studies (Bozick et al. 2018). Yet some of the included studies were conducted in jails, and all were published before 2012. Thus, the state has much to learn about the educational programs available to prisoners in the last decade and whether they also reduce recidivism.
Primary and Secondary Education
Elementary and high school courses are provided in traditional classroom formats, referred to as general population (GP), and via the voluntary education program (VEP), which allows people to learn on their own. In both formats, people follow individualized pathways toward demonstrating competency in areas guided by the College and Career Readiness Standards, a subset of the Common Core State Standards. The individualized structure allows new students to enter GP classrooms as others exit.
Adult Basic Education
ABE programs map onto primary school grade levels. ABE I spans kindergarten through third grade; ABE II fourth through sixth grade; and ABE III covers seventh and eighth grades. Intake assessments signal appropriate ABE levels. Participation in ABE is mandatory for individuals who read below the ninth-grade level and do not have a high school credential (i.e., a high school diploma or GED certificate). Nonparticipation can carry negative consequences, as described in the “Measuring program participation and targeting accuracy” text box.
GED and High School Equivalency
Individuals without a high school credential must participate in secondary education, which includes courses in language arts, mathematical reasoning, science, and social studies. Students earn a GED, high school equivalent certificate, or a high school diploma. Most take the computer-based GED test, which is preferred by OCE. People living in secure units and fire camps can take the paper-based High School Equivalency Test (HiSET). People within 40 semester credits of earning their diploma have the opportunity to satisfy state graduation requirements.
College
Individuals who have a high school credential can take postsecondary courses offered primarily by community colleges. College courses are offered through classroom-based instruction in prisons and correspondence programs. Participation in postsecondary education is completely voluntary. However, CDCR requires those who enroll to take transferable courses that lead to an associate or bachelor’s degree.
Availability of Education Programs
Education programs were available in all prisons between 2014 and 2019. In executing the Blueprint, CDCR hired over 200 new academic and vocational instructors, which expanded program capacity, as shown in Figure 8. Daily budgeted capacity for classroom-based (GP) programs held roughly steady between 16,000 and 17,000 slots. By contrast, VEP program slots grew from 24,000 to more than 28,000.
CDCR expanded VEP capacity while GP slots held relatively steady between 2014 and 2019
Daily education program capacity by fiscal year
SOURCE: 2014–2019 Annual C-ROB reports.
NOTES: This figure includes primary and secondary education courses and excludes college courses. Since education programs are open-ended, capacity indicates the maximum number of people who can participate on any day.
The legislature incentivized community colleges to provide courses in prisons
Before 2014, imprisoned students participated in college courses mainly by correspondence. Senate Bill 1391 (2014) incentivized community colleges to work with CDCR to offer in-person courses by directly compensating them. As a result, college courses became widely available in most prisons between 2014 and 2019.
In fall 2017, over 4,400 prisoners enrolled in face-to-face classes (Mukamal and Silbert 2018). By the end of the 2018–19 fiscal year, 24 colleges (23 community colleges and one CSU campus) offered classes at 34 prisons and four camps outside of prisons where inmate firefighters are trained (C-ROB 2019).
Assessed Educational Achievement
Though information on educational attainment is largely incomplete in the CDCR data, the department assesses people’s reading, language, and math skills upon intake. Initial Test of Adult Basic Education (TABE) reading scores place people without high school credentials into education programs. TABE scores range from 0 to 12.9 and map roughly onto grade levels.
Released prisoners had reading and math skills at the middle-school level
Eighty-five percent of those released had at least one TABE assessment. On average, released people read at the eighth-grade level and scored at the sixth- to seventh-grade level in math, as shown in Figure 9. Across each release year, the reading scores first assessed for each cohort remained stable, whereas math and total battery scores—which include reading, math, and language—increased for each cohort.
Average TABE scores indicate middle-school level reading and math abilities
Average TABE score at first assessment, by release year
SOURCE: Author calculation from CDCR administrative data.
NOTES: This figure displays a person’s first non-zero, non-missing reading, math, and total battery score. We use non-zero scores as there are unusually high proportions of zero scores that may reflect unadministered tests, not actual zero scores. In the case that a person was released more than once in a year, we use scores from their first release that year. The TABE complete battery includes reading, math, and language.
Participation in Primary and Secondary Education Programs
We present participation in primary and secondary education programs relative to 155,000 released people who have not earned a high school credential, and based on the first time they were released in each year. Four in ten participated in at least one primary or secondary education course. One in three people released in 2015 participated in primary or secondary education courses. By contrast, half of those released in 2019 did. Between 2016 and 2019, total participation in these programs increased from 43 percent to 49 percent, as shown in Figure 10.
People released in 2019 were more likely than those released earlier to participate in education courses
Participation rates in education courses by release year
SOURCE: Author calculation from CDCR administrative data.
NOTES: N=154,718 released people without a high school credential or missing prior education data. Participation includes involvement in any education course—general, voluntary, or both. If a person was released more than once in a year, we use information from the first release. People can participate in more than one course during a prison stay. Program participation information was less complete in 2015.
Non-classroom instruction accounted for increased participation rates
Increasing participation in voluntary education explains participation gains. General population participation held fairly steady between 2016 and 2019, whereas those participating only via voluntary education or a combination of general population and voluntary education increased 3 and 4 percentage points, respectively (Technical Appendix Figure D2).
People who test into ABE are more likely to participate than those who test into GED
We assess whether people initially assessed at a level then ever participated in an education program at that level (Figure 11). We likely overstate the shares of those in need who participated (left column) and those who participated who were not in need (right column) because we do not examine reassessments.
Those who tested into primary education had greater participation rates than those who tested into secondary education, perhaps because people who test into the lowest levels receive priority for educational programs. Fewer than 3 in 10 people who initially tested at the high school level ever took a GED course. However, nearly 6 in 10 GED course participants first tested into this level. About 4 in 10 people who first tested into adult basic education ever participated in the corresponding course.
Within adult basic education, targeting is least accurate for ABE III. Just 35 percent of those assessed at ABE III participated, compared to more than 40 percent of those assessed at ABE I or ABE II. At 31 percent, ABE III courses also had the lowest share of participants assessed to need that level of education program. ABE III courses also had higher rates of participation among those initially assessed at a different level: 13 percent of people who tested into another level instead took an ABE III class, relative to 9 percent for ABE II and 5 percent for ABE I.
People assessed into primary and secondary education courses participated at low rates
Targeting accuracy of primary and secondary education programs
SOURCE: Author calculation from CDCR administrative data.
NOTES: N=154,718 released people without a high school credential or missing prior education data. Participation includes involvement in any education course—general, voluntary, or both. If a person was released more than once in a year, we use information from the first release. People can participate in more than one course during a prison stay. Each cell in the figure has a different underlying N as described in Technical Appendix Tables D3 and D4.
Regression to lower education levels is more common than progression to higher levels
In the case of ABE II and ABE III, we can ask whether participation rates among those who first tested into another level suggest progression—people are advancing through levels—or regression—people are entering or returning to a lower level. We find more evidence of regression than progression.
At the ABE II level, more than 6 in 10 may have regressed because they were initially assessed at a higher level—30 percent had tested into ABE III and 32 percent had tested into GED. Similar results hold for ABE III—and we do not yet know why (Technical Appendix Table D4). On the other hand, 38 percent of people who were assessed as ABE I participated in ABE II, which suggests that they progressed into ABE II. Likewise, 35 percent of those assessed at the ABE II level participated in ABE III.
Participation in college courses more than doubled between 2015 and 2019
Approximately 8 percent of people released from California prisons between 2015 and 2019 participated in at least one college course. Participation in college courses more than doubled from fewer than 2,000 participants among those released in 2015 to more than 4,000 among those released in 2019, as shown in Figure 12.
Double the number of people in the 2019 cohort had participated in college courses compared to the 2015 cohort
Participation in college courses by release year
SOURCE: Author calculation from CDCR administrative data.
NOTES: N=14,829 people released from 2015 to 2019 participated in college courses. If a person was released more than once in a year, we use information from the first release.
Rehabilitative Programs
At the recommendation of the Expert Panel, CDCR began offering rehabilitative programs in all prisons between 2015 and 2019. The Office of Program Operations (OPO) within DRP administers these programs, which include substance use disorder treatment and cognitive behavioral interventions that address assessed anger management, criminal thinking, and family relationships needs. CDCR also offers smaller-scale programs that target specialized needs, which we discuss in Technical Appendix E.
Prior research suggests that substance use disorder treatment and cognitive behavior interventions can reduce recidivism (Duwe 2017; MacKenzie 2006). For instance, meta-analysis showed that people who received prison-based residential substance use treatment were 37 percent less likely to recidivate than those who did not (Mitchell et al. 2012). However, the included studies are more than a decade old—and research on cognitive behavioral interventions is even older, with most conducted in the 1980s and 1990s (Byrne 2020; Duwe 2017). As we shall see, most of California’s prisoners need these programs—and the state needs to know whether the programs improve in-prison behavior and reduce recidivism.
Substance use disorder treatment
The SUDT program is a cognitive behavior intervention that aims to minimize reoffending by reducing drug and alcohol use. DRP has endorsed Cognitive Behavioral Interventions for Substance Abuse (CBI-SA), which the University of Cincinnati Corrections Institute developed. The group-based program includes eight modules that build cognitive, social, emotional, and coping skills to reduce reliance on drugs and alcohol. Students learn about the linkages between thoughts and behaviors, how to recognize the risks that behaviors and settings can pose, and strategies to manage emotions. They then develop post-prison plans that include problem solving (UCCI 2021).
Anger management
Anger management programs help participants identify and process emotions and reduce maladaptive behaviors, such as aggression, impulsivity, hostility, anger, and violence. Through anger management programs, DRP aims to “help displace out-of-control, destructive behaviors with constructive pro-social behavior.” Though DRP did not recommend evidence-based programs, two common programs offered during our study period were Controlling Anger—Learning to Manage It (CALM) or Aggression Replacement Training (C-ROB 2015).
Criminal thinking
Criminal thinking programs aim to reduce criminal behavior by changing how people think. DRP recommends Thinking for a Change 4.0, which National Institute of Corrections created. Thinking for a Change focuses on cognitive restructuring, social development, and problem solving. Cognitive restructuring helps people pay closer attention to their thoughts and the risks they may pose. Social development cultivates social skills, including active listening, posing questions, giving feedback, and negotiating with others. Problem solving instruction focuses on strategies to change negative thinking to work towards productive outcomes in social interactions.
Family and relationships
Family and relationships programs aim to improve participants’ relationships, including with family members and romantic partners. For example, the Partners in Parenting program focuses on building communication skills between parents and between parents and children. Parents learn how to discipline children and modify their behavior. Through such programs, CDCR aims to promote healthy family values and parenting skills. Family reunification can also occur for those who have had little to no contact with family for an extended time.
Availability of Rehabilitative Programs
Rehabilitative programs became available in all prisons between 2014 and 2019 (Technical Appendix Table E2). As shown in Figure 13, the budgeted capacity for rehabilitative programs also increased dramatically. Substance use disorder treatment slots increased 124 percent; capacity for anger management program and criminal thinking programs grew more than 150 percent; and family and relationships program capacity more than doubled (C-ROB 2015, 2019).
CDCR dramatically expanded the capacity of rehabilitative programs between 2014 and 2019
Annual budgeted capacity for rehabilitative programs, 2014–19
SOURCE: 2014–2019 Annual C-ROB reports.
NOTES: Only slots for core rehabilitative programs are shown. Capacity for other programs is described in Technical Appendix E. Since rehabilitative programs have finite duration, capacity indicates the maximum number of students who could complete programs in a year.
Assessed Need for Rehabilitative Programs
During our study period, CDCR assessed rehabilitative needs using the Correctional Offender Management Profiling for Alternative Sanctions (COMPAS). The COMPAS assessment generates a score for level of need; this level of need is the primary (but not sole) eligibility indicator for rehabilitative programs. People whose scores reflect medium or high need can qualify for programs, whereas those with low need usually volunteer.
Ninety-two percent of those released had a COMPAS assessment (Figure 14). On average, 68 percent of people released between 2015 and 2019 had been assessed to need substance use disorder treatment; 47 percent had an assessed anger management need; 40 percent had an assessed criminal thinking need; and 25 percent had an assessed need for family and relationships programs.
COMPAS assessments revealed highest need for substance use disorder treatment and lowest need for family and relationships programs
Assessed needs for core rehabilitative programs at first assessment
SOURCE: Author calculation from CDCR administrative data.
NOTES: For those with COMPAS data, we use scores aligned with their earliest test date for each particular subject. For substance use, N=159,670. For anger management, N=157,264. For criminal thinking, N=159,670. For family and relationships, N=157,134. For people released more than once between 2015–2019, we use their earliest COMPAS scores from each stay. If a person was released twice in a year, we report scores from their first release in that year. We use scores from core COMPAS assessments only and do not include scores from re-entry COMPAS assessments.
Participation in Rehabilitative Programs
We report participation rates for rehabilitation programs based on the first time each prisoner in the cohort was released in each year. Participation in all rehabilitative programs more than doubled between 2015 and 2019 but peaked at just 19 percent of those released—and only for substance use disorder treatment.
Across the five years, substance use disorder treatment had the highest participation rates and family and relationships programs had the lowest (Figure 15). On average, 15 percent of people had participated in substance use disorder treatment; 10 percent had participated in an anger management program; 10 percent had participated in a criminal thinking program; and 6 percent had participated in family and relationships programs.
Participation rates more than doubled for most rehabilitative programs
Participation rates in core rehabilitative programs by release year
SOURCE: Author calculation from CDCR administrative data.
NOTES: N=185,007. This figure shows the percentage of people in each release year who participated in each type of rehabilitative program, regardless of variation. Program varietals include reentry, Long-Term Offender Pilot Program, and general. In the case that a person is released from prison twice in the same year, we use that person’s information from their first release that year.
Most people assessed to need rehabilitative programs did not participate
Participation is mandatory in anger management, criminal thinking, family and relationships, and substance use disorder treatment programs for those with medium or high needs, and failure to participate can lead to citations for rules violations and other consequences; however, whether people can participate is conditional on availability—whether the prisons had programs and how many slots were staffed—and accessibility, or how many other people need programs and how long people will be imprisoned. People without assessed needs can also volunteer to participate. However, those with assessed needs should receive programs before those who volunteer.
Substance use disorder treatment was the best targeted of the core rehabilitative programs, as shown in Figure 16. Eighty-two percent of those who participated in substance use disorder treatment had medium-to-high needs. Even so, most people who needed the program did not participate. Just one in five people with medium-to-high need participated in substance use disorder treatment.
Substance use disorder treatment was targeted most accurately, while family and relationship programs were targeted least accurately
Targeting accuracy for rehabilitative programs
SOURCE: Author calculation from CDCR administrative data.
NOTES: N=185,007. This figure shows the percentage of people in each release year who participated in each type of rehabilitative program, regardless of variation. Program varietals include reentry, Long-Term Offender Pilot Program, and general. In the case that a person is released from prison twice in the same year, we use that person’s information from their first release that year.
Most people assessed to need anger management and criminal thinking programs did not receive them, even as some people assessed at low need participated. About 74,000 people were assessed with medium-to-high anger management needs, and about 63,500 people had an assessed criminal thinking need.
Just 17 percent of people assessed with medium-to-high anger management needs participated in an anger management program while 34 percent of participants had no assessed need. Likewise, 18 percent of those assessed at medium or high need participated in criminal thinking programs while 42 percent of participants had low or no needs.
Family and relationships programs were the most inaccurately targeted rehabilitative programs. Just 11 percent of people assessed to need family and relationships programs received them. Moreover, most people who received the program did not need it: 60 percent of participants were assessed as low need.
Employment Programs
Employment programs and prison work show promise for boosting employment after release and lowering recidivism (Hess and Turner 2021; Nur and Nguyen 2022). In their recent evaluation of Prison Industry Authority (PIA) programs in California, James Hess and Susan Turner (2021) found that individuals within the program who received career technical education, as compared to those who did not, were 15 percentage points less likely to be rearrested (35% vs. 47%) and 6 percentage points less likely to be reconvicted (15% vs. 21%) within three years.
CDCR’s other employment programs have not been evaluated. The Office of Correctional Education (OCE) oversees career technical education (CTE) programs, which are independent of PIA, and a job readiness program called Transitions. The Cal-ID program, which the Office of Program Operations (OPO) oversees, also falls in this area. Additionally, in future work we will examine prison jobs or work assignments, which are administered by the Division of Adult Institutions (DAI), and described in Technical Appendix F.
Transitions
Transitions is a bespoke job readiness program designed to prepare prisoners for community reentry by teaching them how to find jobs, hold jobs, and manage their money. In modules related to finding jobs, students learn to understand their work preferences, identify employment opportunities, and effectively present themselves and their skills to potential employers. Modules related to money management focus on budgeting, saving, and understanding current credit and banking services. Participation in Transitions is limited to those close to their release dates and who have a medium-to-high employment need.
Career Technical Education
CTE programs aim to provide prisoners with training that increases their chances of securing skilled jobs that pay a living wage after prison. The department offers programs that provide imprisoned students with opportunities to earn industry-recognized certificates, which can lead to apprenticeships and entry-level positions within their field upon release. CDCR offers programs in six career sectors (building and construction trades, energy and utilities, business and finance, public service, manufacturing and product development, and transportation). Through 2019, there were no eligibility requirements for CTE enrollment.
CAL-ID
Employment and housing “are nearly impossible to obtain” without an official form of identification (NRRC 2016: 2). Assembly Bill 2308 (2014) which created the Cal-ID program, tasked CDCR and the Department of Motor Vehicles (DMV) with providing valid identification cards to “all eligible” released people. Original eligibility criteria included having had a California driver’s license or ID card in the past; having a picture on file with the DMV that was taken within 10 years of release; having no outstanding fees; and being able to provide a verifiable full name, date of birth, Social Security number, and evidence of presence in the United States. Several subsequently enacted bills incrementally expanded eligibility and reduced fees associated with ID provision.
Availability of Employment Programs
Between 2015 and 2019, opportunities for prisoners to participate in employment programs expanded. However, employment programs expanded in different ways due to the nature of the programs: some became more widely available, budgeted capacity increased for some, and some expanded in both ways.
Transitions is a standalone program, which expanded when an administrative shift affected who operated the program and where it was offered. Before 2016, the Division of Rehabilitative Programs (DRP) contracted with multiple external organizations to provide the Transitions program, which was available only in select prisons. Thereafter, OCE assumed responsibility for providing the program, updated the curriculum, and began expanding to all prisons. By 2019, Transitions programs were available in all prisons. Accordingly, the budgeted capacity for Transitions programs exploded from 2,400 slots in 2014 to more than 20,000 in 2019.
The Cal-ID program is also a standalone program. Like Transitions, it was rolled out across prisons during this period. However, the Cal-ID program is not constrained by capacity because “participation” is an application process, not an instructional program.
Career technical education is an umbrella term that captures 18 instructional programs. Though some courses are widely available, others are limited to prisons with facilities to accommodate them (see Technical Appendix Figure E1). Between 2014 and 2019, the budgeted capacity for CTE programs increased 30 percent from roughly 8,500 to nearly 11,000 slots, as shown in Figure 17 (C-ROB 2015, 2019).
CDCR expanded the capacity of CTE and Transitions programs between 2014 and 2019
Budgeted capacity for employment programs, 2014–19
SOURCE: 2014–2019 Annual C-ROB reports.
NOTES: Career technical education includes all available programs. CTE capacity reflects the maximum number of daily participants because some courses exceed one year. By contrast, Transitions programs have finite duration; capacity is the maximum number of participants per year.
Participation in Employment Programs
We discuss participation in employment programs based on the first time each prisoner in the cohort was released in each year. Participation rates for all employment programs at least tripled between 2015 and 2019. However, what participation means varies across employment programs. For the Cal-ID program, participation is an application process. By contrast, Transitions and CTE courses are instructional programs.
Forty-one percent of released prisoners had an assessed employment need
Unlike rehabilitation and education programs, employment programs did not have formal need-based eligibility criteria during our study period. Yet Transitions and CTE programs aimed to help people with employment needs, as described above. Therefore, we assess how accurately they were targeted to people with assessed employment needs, as determined via the COMPAS (Technical Appendix Figures F5 and F6). We do not assess targeting accuracy for the Cal-ID program, which could facilitate employment but has other aims.
Participation in the Cal-ID program quadrupled
Between 2015 and 2019, 15 percent of those released—nearly 27,000 people—received a Cal-ID. During that time, participation in the program quadrupled (Figure 18). In 2015, 8 percent of those released had applied for a Cal-ID and 5 percent received one. By 2019, 29 percent of those released had applied for a Cal-ID and 19 percent received one.
Participation in employment programs increased dramatically
Participation rates by release year for Transitions, CTE, and Cal-ID
SOURCE: Author calculation from CDCR administrative data.
NOTES: N=185,007. If a person was released twice in a year, we report participation associated with their first release in that year. We calculate the share of people in CTE programs by identifying the first release of all unique individuals released each year and determining whether they participated in any CTE program during that prison term. Within a year, no person (or their participation) is counted twice, but a person can be double counted across years. Just 303 people are released twice in one calendar year.
Participation in Transitions accelerated after OCE began providing it
While the department was making the switch from contract-based to OCE program provision, just 8 percent of those released in 2015 participated in Transitions. Participation increased dramatically under OCE (Technical Appendix Figure F2). In 2017, 22 percent of those released had participated. By 2019, 28 percent had, as shown in Figure 18.
Participation in CTE was low and concentrated in computer technology courses
Participation in career technical education nearly tripled over time but remained low across release years. Even in 2019, just 12 percent of prisoners had participated in CTE programs (Figure 18). Participation was especially low relative to the variety of programs available.
Participation is higher in programs that are more widespread. In 2018, computer technology and electronics were the most widely available (Technical Appendix Figure F8). With nearly 3,800 students, participation in computer technology classes outpaced participation in other CTE courses. The next most popular course, electronics, had 1,700 participants (Figure 19).
Computer technology was the most common CTE course among released prisoners
Number of people participating in CTE programs by type, 2015–19
SOURCE: Author calculation from CDCR administrative data.
NOTES: N=15,786 participants released from 2015–19. We do not show roofing or computer coding in this figure, as both programs have less than 30 participants.
Neither Transitions nor CTE programs were targeted to those with assessed employment needs
Transitions provides general skills required to succeed in the job market and CTE programs teach trade-specific skills. Although people with assessed employment needs could benefit from Transitions and CTE programs, just 24 percent of people assessed with medium-to-high employment need participated in Transitions and most participants (57%) were assessed at low need (Technical Appendix Figure F5). Similarly, only 9 percent of those with medium-to-high assessed employment needs participated in any CTE program while 73 percent of CTE participants had no assessed needs (Technical Appendix Figure F6).
Recidivism Trends
Prison programs aim to promote successful reentry pathways by addressing prisoners’ needs and building on their competencies. Along with other life-course outcomes such as employment, recidivism indicates whether people are following successful reentry pathways. As CDCR formulated its Blueprint, the state’s comparatively high reconviction rates were concerning. Nationally, 37 percent of prisoners released in calendar year 2012 were reconvicted in two years, compared to 44 percent of those released in California in fiscal year 2012 (Durose and Antenangeli 2021; CDCR 2017).
California’s reconviction rate has since decreased. An average of 36 percent of people released between fiscal years 2015–16 and 2018–19 were reconvicted in two years (CDCR 2022, 2023a, 2023b, 2024). Greater program participation in the wake of the Blueprint could account for some of the decrease—but so could other factors. For example, some criminal justice reforms, including Proposition 47, lowered recidivism rates around certain crimes (Bird et al. 2018). Additionally, changes in arrest and conviction rates amid the pandemic likely impacted recidivism rates (Harris 2023; Premkumar et al. 2023).
In this section, we describe rearrest and reconviction rates to establish a baseline to guide our ongoing research. We do not associate recidivism outcomes with program participation. At this stage of our evaluation, we are still learning about the processes that lead to program participation and completion (see “Why are program participation rates so low?” text box). When we have that information, we can rigorously assess the relationship between prison programs and recidivism. Without it, we might misrepresent that relationship. In future work, we will estimate the impact of program participation on recidivism and other reentry indicators, including employment and health outcomes.
Nearly half of rearrests involve supervision violations and drug possession
On average, 62 percent of those released between 2015 and 2019 were rearrested for any crime. However, less than half of released prisoners were rearrested on felony charges and 16 percent were rearrested on felony charges we categorize as violent, as shown in Figure 20.
Within two years of release, 62 percent were rearrested and 37 percent were reconvicted
Survival curves indicate the percentage of people who had not experienced the event
SOURCE: Author calculation from CDCR and DOJ data.
NOTES: N=167,861 people with DOJ data at first release. Recidivism is relative to first release with 621 missing DOJ data. Our definition of a violent felony differs from the Penal Code. See the “Description of the data” text box for more information.
Nationally and in California, most people who are eventually rearrested are rearrested within one year (Alper et al. 2018). Half of those rearrested for any crime were rearrested within five months. Half of those rearrested for violent felonies were arrested within about 10 months.
People released from California prisons were rearrested 382,000 times within two years. Most of these rearrests were for less serious crimes. At least 47 percent were for supervision violations (31 percent) or drug possession (at least 17 percent). Similarly, of the released people, 15 percent were rearrested on a drug possession-related charge and more than 30 percent were rearrested for a supervision violation (Technical Appendix Figure G2).
A substantial share of rearrests do not lead to reconviction
Forty percent of those rearrested for any crime were not reconvicted of any crime. About half of people who were rearrested on a felony charge were reconvicted of a felony within two years. The same holds true for those rearrested for drug possession: half were reconvicted. Comparatively, 64 percent of those rearrested for a violent felony within two years were not reconvicted within that timeframe.
The lower probability of reconviction—especially for more serious crimes—is partly attributable to prosecution times. It takes more time to reconvict people than to rearrest them. Of reconvictions that took place within one year of rearrest, the average time between arrest and conviction is about three months.
People with prior prison histories are more likely to reoffend than first-timers
People released for the first time were less likely to recidivate than people who had previously been imprisoned, as shown in Figure 21. Fifty-five percent of first-timers were rearrested and 32 percent were reconvicted. By contrast, 69 percent of people with prior prison terms were rearrested and 42 percent were reconvicted.
Across offense levels and types, people with prior prison histories have higher rearrest and reconviction rates than first-timers, which aligns with prior research (Rosenfeld et al. 2005). Twenty-eight percent of those with prior prison histories were reconvicted for felonies, as were 21 percent of first-timers. Previously imprisoned people were more likely than first-timers to be reconvicted of all crime types—conduct, drug, property, and violent.
Recidivism rates are higher for the previously imprisoned
SOURCE: Author calculation from CDCR and DOJ data.
NOTES: N=167,861 people with DOJ data at first release. Rearrest and reconviction are within two years of release date. See the “Description of the data” text box for more information about crime classifications.
Previously imprisoned people were also less likely than first-timers to participate in most prison programs (Technical Appendix Figure G3). Their participation lagged at least 4 and up to 10 percentage points behind first-timers in all education, Transitions, and rehabilitative programs, with the notable exception of substance use disorder treatment. Fifty-seven percent of those with prior prison histories participated in substance use disorder treatment, compared to 43 percent of first-timers.
Rearrest and reconviction rates plummeted with the onset of the pandemic
Two-year rearrest and reconviction rates held reasonably steady in the three years leading to the pandemic. Of people released from prison between 2015 and 2018, about 63 percent were rearrested, half were rearrested for felonies, 39 percent were reconvicted, and one-quarter were reconvicted for felonies.
Beginning with prisoners released in the first quarter of 2018, recidivism rates steadily decreased as the pandemic—during which Californians saw dramatic decreases in arrest and conviction rates—encompassed more of the two-year follow-up (Harris 2023; Premkumar et al. 2023). Rearrest rates for those released in the fourth quarter of 2019 fell 10 percentage points relative to the first quarter of 2018, and felony rearrest rates dropped 8 percentage points. Reconviction rates plummeted 13 percentage points, and felony reconviction rates fell 7 percentage points (Figure 22). As our evaluation unfolds, we will need to account for these underlying downward trends to ensure we do not attribute underlying decreases in recidivism rate to program participation.
Reconviction rates fell 8 percentage points amid the pandemic
Two-year rearrest and reconviction rates by release quarter
SOURCE: Author calculation from CDCR and DOJ administrative data.
NOTE: N=167,861 people with DOJ data at first release.
Conclusion and Policy Recommendations
In 2012, CDCR set a goal to meet the rehabilitative needs of 70 percent of eligible people in prison. Between 2014 and 2019, budgeted program capacity rose rapidly across program areas; and program participation rates increased sharply. Participation in primary education increased more than 60 percent and participation in secondary and postsecondary programs doubled. Participation rates for most rehabilitative programs also doubled, while rates tripled for Transitions and career technical education. Compared to people released from prison in 2015, nearly four times as many people released in 2019 applied for a state ID.
Despite significant investments in prison programs that led to steep increases in participation rates, achievement of the 70 percent goal remained distant as of 2019. Most people entered prison with needs that ranged from low educational achievement to high risk of substance use disorder. Yet just 9 percent of the five-year cohort had participated in any CTE program and just one in five participated in any core rehabilitative program. Participation rates for education programs were higher—averaging 40 percent across all programs and years.
Participation rates could be increased by targeting programs more accurately. Most people with assessed need for programs did not participate and some people without assessed needs did. At most, 20 percent of people assessed to need rehabilitative programs participated in them. Similarly, between 28 and 42 percent of people who tested into education programs participated in them.
Programs should also address the changing needs of imprisoned people, their families, and California communities. Nearly two decades of reforms have reshaped California’s formerly imprisoned population. Realignment had a profound impact on the state’s justice system that was “still unfolding” in 2015, as the effects of AB 109 comingled with those of Proposition 47 (Raphael and Lofstrom 2015). Realignment reduced the likelihood of being sentenced (and resentenced) to prison, thereby altering pathways into prison in ways that have implications for programs and how they can support people when they return to their communities.
Though people in prison for the first time became an increasing share of the released prisoner population, people with prior prison histories were still the majority of those released in 2019—and they were 14 percentage points more likely to be rearrested and 10 percentage points more likely to be reconvicted within two years than those released for the first time. Yet the previously imprisoned participate in most programs—with the marked exception of substance use disorder treatment—at lower rates than first-timers.
Most previously imprisoned people are age 35 to 54, and their reentry needs differ from those of younger people (Burke et al. 2022; Technical Appendix Figure C1). A key difference lies in “instrumental” family support or how families encourage reentry by providing housing and financial assistance; this support tends to keep people stably housed, even as they may struggle to maintain employment and desist from crime (Mowen et al. 2019). Older people tend to receive less support and they are more likely to reoffend when they do not have it (Western 2018). To reduce reoffending among older people who have deeper ties to the justice system (Technical Appendix Figure C3), prisons may need programs that help people secure housing and establish financial security, whether through employment or connecting people to social services.
Released people who do return home may bring violence with them. Domestic violence is prevalent among those released from prison and the reasons why are poorly understood (Mowen and Fisher 2021; Stansfield et al. 2022). In California’s formerly imprisoned population, assaults were the most common type of violent crime and domestic assaults were the most prevalent type of assault, even among older people. The high prevalence of domestic assault and its persistence as people age suggests that many families could benefit from programs specifically designed to mitigate it, though we appreciate the challenges of identifying and developing domestic violence reduction programs that work (but see Hasisi et al. 2016).
Yet substance use and abuse may pose the greatest impediment to people making the most of available education and employment opportunities in prison and after they are released. Among the core rehabilitative programs, participation rates were highest in substance use disorder treatment. Even still, only one in five people who needed treatment participated. Moreover, a substantial share of recidivism was related to drug and drug paraphernalia possession. Just six possession-related crimes accounted for 17 percent of the rearrests within two years of release.
Realignment provided people opportunities to reduce their sentences by maintaining good behavior in prison. As a result, people released between 2015 and 2019 served about 60 percent of their sentences. Forty percent of released people had spent less than one year imprisoned—and they were least likely to participate in programs. Similarly, California’s jail inmates typically serve sentences of less than one year and lack access to programs (Tafoya et al. 2016; May et al. 2014). Therefore, identifying feasible ways to help people serving shorter projected sentences enroll in and complete prison programs could also benefit California’s jail inmates—and we aim to identify such strategies in future research.
The preceding discussion leads to five main recommendations.
- The state may need to make stronger investments in programs to meet the needs of all eligible people.
- CDCR may need to adapt its programs to meet the needs of a changing population.
- Identifying and providing violence interventions, especially for those who will return to their families, should be a high priority for CDCR and the state.
- The state and CDCR should prioritize ensuring broader access to substance use disorder treatment.
- Efforts to expand and evaluate prison programs, especially for those released after serving short sentences, can inform efforts to expand jail programs and parole programs.
Since 2018, PPIC has worked with CDCR’s Division of Rehabilitative Programs (DRP) and the Office of Research (OR) to design and implement a robust, multi-year prison program evaluation. With this report, we laid the groundwork for understanding California’s vast prison program environment and the people who exit prison after experiencing it. Many questions remain unanswered. As our research progresses, we aim to improve program offerings, operations, and outcomes to help CDCR fulfill its mission to rehabilitate imprisoned people and promote the well-being of the communities to which they return.