The Michigan Department of Corrections Special Alternative Incarceration Program

Transcription

1 The Michigan Department of Corrections Special Alternative Incarceration Program First Year Process Evaluation: An Independent Review of Program Improvements Submitted by James Austin Gabrielle Chapman The JFA Institute 5 Walter Houp Court, NE Washington DC September, 2009

2 KEY POINTS The Special Alternative to Incarceration (SAI) program was modified significantly in 2008 to become an integral part of the Michigan Prisoner ReEntry Initiative. As such it should no longer be characterized as a traditional boot camp program but an intensive prison diversion and/or release program. The planned SAI program modifications include a comprehensive risk and needs assessment of each person admitted to the program and the development and delivery of an individualized program plan. There is also a well coordinated release plan with the offender s probation and parole officer that facilitates seamless continuation of services and progress being made by the offender. These program modifications have positioned the SAI program to continue to produce cost-effective reductions in the Michigan prison population without jeopardizing public safety. The modified SAI program is clearly saving significant prison bed space and should be able to demonstrate lower recidivism rates in the future. It is therefore recommended that the SAI program be continued until a Phase II Impact Evaluation focused on specific recidivism reduction is completed. It is also recommended that the current restrictions (both legislative and administrative) on admission to the SAI be reviewed and amended to maximize the number of offenders that can be safely placed in SAI, thus maximizing bed savings and improving public safety. i

3 EXECUTIVE SUMMARY FINDINGS 1. Based on recommendations made by the Office of the Auditor General and JFA, the Michigan Department of Corrections (MDOC) has fundamentally altered the Special Alternative to Incarceration (SAI) program screening and intervention/programmatic structure. These changes have significantly enhanced the SAI s potential to reduce both recidivism and MDOC agency costs. 2. The annual cost of the SAI is approximately $13.7 million with bed capacity of approximately 500 persons. 3. When the projected MDOC length of stay is computed for the SAI participants who have been sentenced to prison, the overall costs savings for each participant is $36, The cost savings for the probation cases is less clear since it is not certain that the SAI sanction is diverting them from prison. That said and using conservative estimates, the estimated cost savings for that population is $2,543 per referral. 5. The SAI is currently scheduled to sunset on September 30, Should this occur, the current prison population will increase by approximately 2,000 inmates as early as next year (2010). This unmet bed demand translates into a need for approximately two additional full prison facilities, at a cost of $40-$50 million in annual operational costs. If the program is authorized to continue the 2,000 bed demand will be averted. 6. The SAI is now accepting a greater number of prisoners (as opposed to probationers) which results in increased operating capacity as well as enhanced net savings due to the higher costs of prison incarceration. The larger numbers of prison offenders also lowered per offender costs. 7. The SAI continues to enjoy an extremely high program completion rate of approximately 90%. This level of success is associated with decreased recidivism and the savings associated with lower levels of incarceration. 8. Programmatic changes have been made that reduce in-efficient and redundant programmatic services. 9. The vast majority of offenders have individualized program plans that are informed by SAI s standardized intake risk/needs assessment. These services appear to be delivered in a professional manner by staff that is well trained in such methods. ii

4 10. Interviews with SAI participants show a high level of satisfaction with the staff and services being provided to them. Representative comments by the SAI participants are as follows: The Corporals helped so much and make you feel that you are not just another number.... they care and don t want you to come back. They [the SAI staff] give you incentives not to come back and the tools to help you in your future These changes have significantly enhanced the probability that SAI will have a greater impact on recidivism rates. 12. The SAI has made significant efforts to improve their relations with the courts and within the DOC (community supervision and the Parole Board). These changes should result in a greater number of referrals by the court and a smoother transition from SAI to community supervision. RECOMMENDATIONS 1. This process evaluation has found that the recommended changes in program screening and intervention/programmatic structure have been successfully made in the SAI Program. These changes have significantly enhanced the SAI s potential to reduce recidivism, the MDOC prison population and agency costs. 2. Due to this finding, it is recommended that the SAI Program should be continued until a scientifically robust recidivism test is completed in the Impact Evaluation study (Phase II). Any final decision regarding the future of the program should be deferred until the results of Phase II are made available. 3. The SAI should expand its daily population to meet its current funding level and further enhance its cost effectiveness. The program is clearly saving prison bed space and is at least as effective as of other release methods for DOC inmates in terms of recidivism. 4. To continue the positive growth of SAI, it is also suggested that a pilot test be conducted to expand the SAI to include prisoners who have denied parole at their initial hearing. Since there will be an excess number of these persons compared to the number of beds available, the assigned should be based on random assignment thus allowing a rigorous test of the suitability of these persons for the program. 5. The MDOC needs to ensure that the probation referrals are likely to be incarcerated or that they pose a high risk to recidivate had the SAI program. To that end, probation referrals should be assessed more carefully to ensure they are either prison or straddle sentencing guideline cases or are assessed as moderate to high risk to recidivate based on the COMPAS instrument. iii

5 TABLE OF CONTENTS Page KEY POINTS... EXECUTIVE SUMMARY..... RECOMMENDATIONS..... i ii iii INTRODUCTION.. 1 HISTORY OF CORRECTIONAL BOOT CAMP-BASED PROGRAMS 3 Impact on Recidivism and Costs... 5 Evaluation Results for the SAI. 7 DESCRIPTION OF THE MICHIGAN SAI PROGRAM. 10 Key SAI Program Changes.. 13 EVALUATION METHODOLOGY AND DESIGN 16 Design 16 Data. 16 RESULTS 19 SAI Screening and Admissions Attributes 19 SAI Services and Interventions program Outcomes 36 SAI Participant Interviews. 39 Organizational Impact of Modified Program 43 Up-Dated Cost Benefit Assessment.. 45 CONCLUSIONS AND RECOMMENDATIONS. 52 APPENDICES Appendix A: Summary of 2009 Changes to SAI Program Structure Appendix B: SAI Program Briefs iv

6 LIST OF TABLES Table 1. SAI Admissions, Primary Offenses for SAI Admissions, Prison Admissions, Probation Admissions, COMPAS Scales, Prison Referrals to SAI 6. COMPAS Scales, Probation Referrals to SAI 7. SAI Program Outcomes by Referral Source, 2007 and SAI Program Outcomes, All Cases, 2007 and SAI Program Graduates v. Non-Completers SAI Operating Budget 11. Estimated Cost Savings for SAI Prison Referrals 12. Estimated Cost Savings for SAI Probation Referrals LIST OF FIGURES Figure 1. Historic SAI population ( Historic SAI Population by Type ( ) 3. Source of SAI Population 4. Current SAI Offender Programs and Services 5-7. Examples of Offender s Daily Schedules 8. Flow of Programmatic Information v

7 INTRODUCTION This report summarizes a preliminary process evaluation of the Special Alternative Incarceration (SAI). The evaluation is referred to as a process evaluation as it seeks to document how the SAI is now functioning on a number of key programmatic areas. This evaluation was needed as the SAI program underwent a fundamental transformation in 2008, which makes it no longer a traditional Boot Camp but a re-entry program that embodies some of the key principles for a successful program. The SAI program made these significant modifications as part of the overall restructuring of the Michigan Department of Corrections (MDOC). The revised SAI program was modified to be fully integrated under the Michigan Prisoner ReEntry Initiative (MPRI) model. Planned and completed program modifications include individualized risk/needs assessments that utilize standardized, validated instruments, offender targeted programming protocols, and strong community collaborations. For this reason, the SAI is now a key part of the larger MPRI model, which is rapidly expanding throughout MDOC. By way of background the MPRI was initiated by the MDOC in response to the current interest in enhancing prisoner re-entry procedures and lowering prisoner recidivism rates. As reported by the US Department of Justice and the MDOC, there has been little progress over the past three decades on lowering state and federal prisoner recidivism rates. The MPRI seeks to reverse this trend by enhancing the prisoner s ability to succeed in their communities once released. This is accomplished by assessing the prisoner s risk and needs at admission to prison, providing those programs that will assist in meeting the identified needs, and then continuing such services during the transition period from prison to community supervision (parole). Other aspects of the MPRI are to promote offender accountability and victim participation in the process. 1

8 The newly designed SAI represents one method for achieving these MPRI objectives. More narrowly it fits within the Phase I or Getting Ready component of the MPRI where the DOC works with incarcerated people to help them prepare for their imminent release from a DOC facility. This is achieved by making a comprehensive of the persons risk to recidivate and the underlying factors that influence that risk level. Once the assessment is complete, the MDOC attempts to ensure the prisoner receives services that address as best they can the identified needs and thus lower the risk to public safety. The first SAI was implemented in 1988 Camp Sauble, a minimum security prison camp for males located in the northwestern part of Michigan s Lower Peninsula. At that time it was a very traditional boot camp program that did not provide for a comprehensive risk and needs assessment or programs that matched the inmate s identified risk and needs. Rather, it relied upon military discipline, hard work, and physical challenges. Over the next 20 years the MDOC boot camp program expanded and contracted as several sites were converted back to minimum security camps. But until recently it relied upon a traditional boot camp structure. Today the SAI program exists at a single site located near Ann Arbor, which currently houses approximately 425 program participants on any given day. The remainder of this report provides an updated assessment of the SAI and documents the extent to which the SAI has altered its screening and intervention methods. A preliminary cost-benefit analysis is also completed. Finally, we discuss prior recidivism analysis and outline how we intend to update that work over the next four months. 2

9 HISTORY OF CORRECTIONAL BOOT CAMP-BASED PROGRAMS Although the SAI is no longer a traditional boot camp program, it is useful to summarize what research has been conducted on such programs and their known effects on recidivism and costs. While correctional boot camps became popular in the U.S. in the 1980s and 1990s, the combination of military-like discipline, physical challenge, and strict adherence to hierarchical authority and offender detention dates back to the 1800s. Louis Dwight, founder of the Prison Discipline Society, first conceptualized the Auburn Model in the early 1820s. The correctional model got its name from the facility that adopted Dwight s vision of sharp, swift punishment and rigid discipline, the Auburn State Prison. The Auburn facility s Principal Keeper and his Deputy further developed the Auburn Model. The two men s significant military backgrounds influenced their plan for facility management and control. The military-based model they created emphasized work, strict discipline, marching, drills, consistent daily routines, and corporal punishment for rule violations (Colvin, 1997). Inspired by a tough on crime ethos, the modern version of military style boot camps reemerged as a correctional model in the early 1980s. The boot camp program appealed to the conservative political culture as it emphasized personal accountability and retribution. The first modern boot camps were established in Georgia and Oklahoma in The 1980s and 1990s saw the height of boot camp proliferation in the U.S. According to Parent (2003), by 1995 there were 75 state correctional adult boot camps, 30 state and locally run juvenile boot camps, and 18 boot camps run by local jails. The modern boot camp model married the military discipline, hard work, and physical challenge of the early Auburn Model. However, some of the new programs began placing a greater emphasis on rehabilitation components such as social skills training, education and 3

10 substance abuse treatment and some programs incorporate an aftercare component as well (Cowels and Castellano, 1995; Kurlychek, 2006). For offenders, the change factors targeted in a boot camp program include positively affecting an individual s values, beliefs, and attachments through structure, discipline, teamwork, skills training, and consistent exposure to prosocial interaction and attitudes. According to MacKenzie and Armstrong (2004), most of today s correctional boot camps, operate under a constructive punishment philosophy... if a person experiences or is placed within an environment of radical change, this environment will create a reasonable amount of stress. As a result of this stress, people will be particularly susceptible to external influences (9-10). Like its military counterpart, the primary intent of the boot camp process is resocialization. It is assumed that within a context of stress, individuals are more amenable to psychological and behavioral change. Despite their initial popularity, today s boot camps housed only a small percentage of U.S. correctional populations at any given time and are declining in their number. For example, at the peak of their popularity in 1990, most states had less than 250 participants and operated only one or two programs statewide (MacKenzie and Armstrong, 2004). Over the past decade, the number of boot camps actually declined due to moderate to negative evaluation results, media attention to the negative impact of shock incarceration programs, and increased government funding for rehabilitation focused sanctions. According to Parent (2003) by 2000, over 30% of state run boot camps in the U.S. had closed. The correctional boot camps that remain in operation today tend to have a primary emphasis on the rehabilitation component and only a secondary focus on military discipline. They also strive to incorporate a strong community aftercare component. 4

11 Impact on Recidivism and Costs The popularity of boot camps in the 1980s and 1990s was based on a promise of significantly reduced recidivism, declining prison populations, and lower correctional operating costs. Unfortunately, many studies have not confirmed these outcomes. One problem has been the lack of rigorous studies. For example, lack of robust comparison groups, negligible implementation measures, limited information on subject characteristics, high rates of attrition, and poorly constructed comparison groups that exclude dropouts from the analysis (Austin, 2000) can all contribute to invalid results. But based on what adequate studies do exist the results generally have been negative. According to Parent s 2003 review of research, boot camps failed to produce significant reductions in correctional populations and recidivism due to their unstandardized programming components, absence of community reentry planning, and early release mandates that made volunteering for boot camps unnecessary as a means of shortening sentences (1). Most researchers have found that boot camp graduates do not do not fare worse in terms of re-arrest, new conviction, or return to prison. In some cases, boot camp participants are more likely to have their parole revoked due to technical violations (Austin, 2000; Flowers et. al., 1991; MacKenzie and Parent, 1991; MacKenzie and Shaw, 1993; MacKenzie et. al., 1990, Parent, 2003). The calculation of boot camp cost savings is also problematic. Most studies directly compare boot camp per diem costs with prison per diem costs to calculate cost savings. This comparison is misleading and can often overstate actual cost savings. For example, one of the largest budget items in corrections is staffing. Correctional staffing patterns are not reduced unless there are significant decreases in prison population. Small or moderate population decreases will only decrease marginal costs. 5

12 Additionally, many boot camp programs admit persons who may not be true prison diversion candidates. Costs and corrections populations can only decrease if persons who would otherwise serve time in prison were diverted to boot camps for a shorter length of stay. This issue is exacerbated by the restrictive eligibility criteria associated with boot camp participants (e.g., only non-violent offenders and/or persons with no prior felony convictions). There is also a related danger of net widening with any new or nontraditional criminal justice sanction. If boot camps are used as a sanction option for persons who would otherwise be released with restitution and/or a fine, both correctional costs and populations can actually increase. Finally, correctional systems (e.g., state, local, federal) vary greatly in how per diems are calculated making both inter and intra system comparisons difficult at best. As mentioned earlier in this discussion, the programmatic components of boot camps vary greatly. Wilson and MacKenzie s 2006 meta-analysis, found boot camps that had a primary focus on treatment, therapeutic and aftercare components showed larger positive effects than boot camps with a primary focus on military training and discipline. This concurs with Parent s (2003) earlier finding that boot camps improved participants attitudes and behavior. Parent (2003) also found that due to the structure and continued activity of boot camps, the programs produced a safer correctional environment for both participants and staff. While boot camp participants have not shown significantly reduced recidivism rates relative to comparison groups, studies have shown a positive impact on other prosocial characteristics such as attitudes, self-esteem, problem solving techniques, community attachment, and impulsivity (Austin, 2000; Bottcher and Isorena, 1994; Clark et. al., 1994; MacKenzie and Armstrong, 2004; MacKenzie et. al., 2001; MacKenzie and Shaw, 1990; MacKenzie and Souryal, 1995; Parent, 2003; Peters et. al., 1997). Overall, evidence suggests that boot camps are no worse then the alternatives.... (e.g., probation or jail time) (Wilson and Mackenzie, 2006:83). However, if used as a prison 6

13 alternative, as the SAI program is currently designed, correctional boot camps can result in small to moderate relative reductions in prison bed demand (MacKenzie and Piquero, 1994; MacKenzie and Parent, 1991; Parent, 2003) as well as modest relative reductions in correctional operating costs (Parent, 2003). Evaluation Results for the SAI In terms of recidivism, the MDOC 2009 Annual SAI Report found that SAI graduates who were originally sentenced to prison and later admitted to the SAI program had a 9.4% lower three year return to prison rate (38.9%) as compared to all other prisoners released on parole (48.3%). In terms of the type of recidivism, the SAI lower rate is solely attributable to a lower technical violation and absconding from parole rate. The two groups have similar rates of being retuned to prison for a new crime. The problem with this analysis is that there is no effort to control for differences between the SAI parolees and the non-sai parolees. Furthermore, it only includes SAI graduates and excludes SAI non-graduates who may have a higher recidivism rate. Finally, the analysis excludes SAI participants who are sentenced by the courts to SAI as part of a probation term. As noted later in this report, the SAI probationers make up a large proportion of the SAI population so any analysis must include them as well. The Michigan Auditor General conducted their own analysis that attempted to improve upon the MDOC assessment methods. Specifically, the AG study said it created a comparison group that met the criteria of the SAI program but were not admitted although it was not clear how this was done. In fact, the AG s comparison group was matched to SAI participants only on the criminal offense and not the other SAI program eligibility criteria. The sample sizes were often small which can create unstable results. As with the MDOC evaluation the SAI program failures were not included in the AG recidivism results and the follow-up time (2 or 3 years) was not controlled for. However, they did conduct a recidivism study for the SAI probationers and a 7

14 SAI eligible probation population. Using this design the AG reported that the SAI prison graduates had a higher return to prison rate. The differences between the SAI probationer SAI graduates and the probationer SAI eligible group was even larger. Relative to costs, the MDOC compared the SAI operational costs to that of a Level I facility in This analysis found that the SAI costs were higher largely due to the greater level of intense supervision. But these higher operational costs were mitigated by the shorter length of stay experienced by the SAI graduates (90 days versus up to 36 months). Using these assumptions the MDOC reported that the SAI averts $35 million in DOC costs per year. In terms of cost savings calculations, it should be noted that the broadened eligibility criteria used by the SAI Program may provide additional cost savings. As mentioned earlier in this report, the SAI Program has broadened its intake to include medically compromised offenders, including those in need of outpatient mental health. Because of their medical and/or mental health issues these offenders have higher than average per diems when in prison facilities. The marginal cost savings for these inmates in the SAI Program would therefore be even higher than that of the average incarcerated offender. The AG s 2007 report also concluded that the program is more expensive per day than a Level I facility but that the program is cost-effective based on the averted prison term (up to 36 months). Unlike the MDOC, the annual cost savings for the SAI was only $2.5 million per year and only if the SAI operated at full capacity. However, one should add that the cost analysis by both the DOC and AG should have used marginal and not fully loaded Level I security costs. In other words, unless one can demonstrate that absent the closure of a facility that otherwise would have been needed, the actual cost savings are much less since personnel savings (which are the bulk of a daily correctional operational cost) are not being realized. In summary, the two studies disagreed on the recidivism analysis but concurred on the cost savings findings. Namely, the SAI is more expensive to operate than a Level I facility or 8

15 camp but that the projected savings from reduced periods of incarcerations or diverted prison admissions resulted in an overall cost savings to the state. 9

16 DESCRIPTION OF THE MICHIGAN SAI PROGRAM While the current SAI Program is an integral part of the overall MDOC mission of evidence-based and best-practice adult corrections and a vital prong of the MPRI model of reentry, the original SAI Program included a significant focus on militaristic training. As a result of the emphasis on the physical training admissions to the program did not include persons with even minor medical or mental health issues thus limiting the pool of potential diversions from a more expensive prison option. While all inmates were well screened upon admission, the program in its original form did not utilize individual needs and/or risk assessments and the limited programming offered did not emphasize evidence-based programs and services. Essentially, good efforts made by staff and DOC administration toward the development and usefulness of the SAI Program; but due to the admissions, assessment, and programming standards the SAI was not getting the biggest bang for the Michigan tax payer buck. The JFA Institute was asked to review the SAI program in 2008, which resulted in a number of suggestions to alter the program. These recommendations along with the MDOC s MPRI philosophy, structure and objectives became the basis of significant change in the form and function of the boot camp. A new operational emphasis was designed to optimize the SAI program. A revised organizational culture also developed from new administration staff that brought a high level of experience to the table and even further enhanced the initial recommendations for program modification. Over the last five years the SAI Program population has varied from a low of 242 in July 2008 to a high of 480 in April 2006 (see Figure 1). The average monthly population for the first 5 months of 2009 was 367 which is higher than the population for same period in

17 Figure 1. Historical SAI Population Total SAI Population by Month January May Jan 03 Apr 03 Jul 03 Oct 03 Jan 04 Apr 04 Jul 04 Oct 04 Jan 05 Apr 05 Jul 05 Oct 05 Jan 06 Apr 06 Jul 06 Oct 06 Jan 07 Apr 07 Jul 07 Oct 07 Jan 08 Apr 08 Jul 08 Oct 08 Jan 09 Apr 09 While the total population numbers at the SAI facility have fluctuated over the past few years, the make up the offender population has changed significantly. Figure 2 illustrates the breakdown of SAI population by type: offenders who were sentenced to prison and persons who were sentenced by the courts to complete SAI as part of their probation term. The former (prisoners) are clearly people who otherwise would have had a longer prison term had the SAI not existed. As will be reported later on in the report the estimated period of imprisonment for this group is 2-3 years. The latter (probationers) are people whose offense suggests a prison term based on the state s sentencing guidelines. The extent to which the SAI is filled by prisoners, the greater the likelihood the SAI is impacting prison beds. 11

18 In 2003, probationers made up the bulk of the population at a monthly average of 60% while prison-bound trainees made up only 40% of the population on average. The pattern began to reverse in 2005 and reached a 50/50 prisoner/probationer ratio in By 2009 this ratio has changed to 64% prisoners and 34% probationers. As noted earlier, this is especially significant as optimal use of the SAI program, in terms of cost savings and population management, requires that SAI beds are used as a true alternative to incarceration. Figure 2. Historic SAI Population by Type SAI Population , Probationers and Prisoners Probationers Prisoners *June population figures used ; May population used for The current program has a daily population of approximately (397 males and 43 females). Of this population, almost two-thirds (64%) are prisoners (253 male and 33 female) (see Figure 3). Probationers make up only about 30% of the trainees (147 male and 10 female). 1 Population information based on SAI population on 5/11/09. 12

19 This is quite a difference from the population make up in September 2008 when the SAI held only 278 inmates (263 males and 215 female) and only a little more than half the trainees were state prisoners. It should also be noted that the MDOC reports a high completion rate of 90% or higher which is far better than most correctional treatment programs. For example, the well publicized Prop 36 drug treatment program in California reports that only 25% of all program referrals actually complete the program. A program completion rate is essential for a program to demonstrate cost savings and meaningful recidivism reductions. Figure 3. Source of SAI Population Population 2008 Population 2009 Prison 53% Prison 64% Probatn 47% Probatn 36% Key SAI Programmatic Changes In an effort to fully align itself with the evidence-based, MPRI-driven mission of the MDOC, the current SAI Program model made a number of strategic changes based on recommendations from JFA and best practice models (see Appendix A for a detailed, point-bypoint comparison between the former and current SAI Program). During 2008, the Special Alternative Incarceration Program (SAI) was established as a Michigan Prisoner ReEntry 13

20 Initiative (MPRI) In-Reach Facility. Early on in this transition it was evident that SAI needed to demonstrate significant changes in its structure and operation to meet the standards of Evidence Based Practice established in the MPRI Model. SAI is unique in the sense that they are a Phase I (Getting Ready) and a Phase II (Going Home) facility requiring the approval of the offender s judge to participate in the 90 day program. Changes are illustrated in this report using the MPRI three faze, seven decision point paradigm. As you can see in Appendix A, many significant changes in program content and practice have been made. The most significant changes were in the area of individual assessment, programming, and operations. In terms of assessment, the SAI Program now utilizes the Correctional Offender Management Profiling for Alternative Sanctions (COMPAS) instrument as a risk and need assessment for each offender within 7 days of admission. Through the use of this instrument as well as accompanying organizational modifications pertaining to the admissions process, the SAI Program has made the transition from a group oriented assessment and/or program assignment process in 2007 to individualized assessment and assignment in The SAI Program continues to strengthen and streamline this process in The individualized assessment process allows for the optimization of offender programming. The program maintains the overarching military milieu that serves to successfully retain consistency, order, and security for staff and inmates but now also includes a significantly enhanced service and/or treatment array designed to positively impact offenders and discourage reoffending upon release. The current SAI Program model engages offenders by offering thoughtful, proven, evidence based programming within a structured quasi-boot camp environment. The SAI Program has identified and incorporated well researched evidence-based programming appropriate for the operational format and length of stay. Program content is not the only modification that has been made in the current quasi-boot camp model. The camp has 14

21 improved program implementation through staff training, the use of documented service protocols, increased use of pre/post tests, and improved tracking mechanisms. The SAI Program curriculum is specifically designed to prepare a person for re-entry. The current program incorporates a strong community service based approach, which supports reentry efforts and can contribute positively to post release adjustment of offenders. Additionally, there is an emphasis on the continuum of offender care with a high level of communication among reentry players (e.g., probation officers, out patient post release counseling services, housing, family reunification, etc.). SAI s use of COMPAS and community Transition Teams also help to tie the SAI program into the MPRI. The directed program modifications noted above have accompanied a broadened admissions eligibility that includes persons with minor, treatable medical conditions and certain levels of mental illness. This initiative this will continue to expand during Finally, it should be noted that the myriad of positive changes in the SAI Program model have not only had a positive affect on offenders. Facility staff have benefitted significantly with increased morale. This outcome will be discussed later on in the report. 15

22 EVALUATION METHODOLOGY AND DESIGN Design A process evaluation design was used for this stage of the SAI Program. A process study measures how a particular program is operating and then evaluates whether the program operations are consistent with the stated program goals and objectives. One assumes that the program goals and objectives have some theoretical or evidentiary basis. If the implementation process is as intended, then there is an expectation that the program s desired goals and objectives will be realized. If not, then one would expect no measurable impact or possibly even negative consequences. In the case of SAI, the program is based in a theory that a person s propensity for deviant activity can be more significantly affected over a shorter period of time than would otherwise occur under traditional probation and prison sanctions. These changes will occur by exposing the person to treatment and rehabilitative services that target the person s individual criminogenic attributes. To maximize the impact of a program like SAI, judges must be mindful of the unintended consequence of net widening, or increasing the total number of persons admitted into the criminal justice system by introducing a new or revised sanction rather than decreasing or stabilizing the number of persons exposed to the criminal justice system. Furthermore, in addition to lowering the expected recidivism rate over a shorter period of time, the program costs are lower than the costs associated with traditional probation, prison, and parole interventions. Data A process evaluation requires a variety of quantitative and qualitative data. With respect to the quantitative data, there were two primary sources: MDOC and the facility s own records and 16

23 databases. The first is the existing MDOC data system which maintains detailed information on all persons admitted to the MDOC s correctional facilities and community supervision (probation and parole). The DOC provided data files of all persons admitted to the newly designed program that included information from 2007 through A data file of all prison and probation admissions for the same period was used to compare the attributes of the other probation and prison admissions. These data files allowed the researchers to determine how many trainees enter the program, their attributes and how they differ from other prisoners and probationers, the length of time in program, their release dates from prison and the post release supervision period. Aside from offender data, cost data was also obtained. The primary source for this data was the Annual Report published in mid-march Each year the SAI Program works with MDOC s central office to develop cost comparison figures. In addition to these data we reviewed a sample of COMPAS and Transition Accountability Plan (TAP) instruments for 20 inmates (10 male and 10 female). Additionally, we reviewed the programs these offenders started/completed, as well as their daily schedules. This information was used to determine the level of consistency from initial risk/needs assessment through completion of target programs. These data were supplemented by a data download from a recently developed program tracking database. The latter data listed all programs completed by SAI participants. In terms of qualitative data, site visits were planned for the purposes of 1) observing program operations, 2) conducting interviews with program staff, and 3) conducting interviews with program participants. The staff and program participant interviews were structured face-toface interviews and were designed to determine the level of program implementation from staff and program participant perspectives. Suggestions for immediate changes were also solicited during the interviews. 17

24 Twenty-seven interviews were completed, 20 with trainees and 7 with staff. The offender interviews consisted of 10 male and 10 female SAI trainees, all of whom had been at the SAI Program for 30 days or more. The twenty offenders were selected at random from a daily trainee roster and included persons aged 19 to 56 years who came to SAI as probationers and as prisoners. Most of the offenders interviewed had been incarcerated before (70%) either in jail or prison. All selected trainees were available and consented to participate in an interview with the researchers. At the start of the interview, the trainees were assured of the confidentiality of their responses and were advised that their participation was voluntary. All interviews took place in a temporarily vacant office in the administration building at the facility and lasted an average of approximately 20 minutes each. In addition to the offender interviews, several staff interviews were conducted to briefly assess the current organizational culture of the SAI work environment and the impact of the SAI program modifications. A sample of staff (6 males, 1 female) ranged in rank from Corporal to Acting Sergeant and included several persons employed at SAI before and after the programmatic changes. Much like the offender interviews, the staff was assured that their responses would be confidential and they were told their participation was purely voluntary. All interviews took place in a temporarily vacant office in the administration building at the facility and lasted an average of approximately 20 minutes each. 18

25 RESULTS This process evaluation was guided by research questions that fell into three basic programmatic categories: screening and admission attributes, services and interventions, and costs. The results of this study are presented in these three categories: 1) Screening and Admission Attributes, 2) Services and Interventions, including Program Outcomes, 3) Participant Interviews. SAI Screening and Admission Attributes The SAI admissions over the past three years have increased, going from a little under 950 admissions in 2007 to over 1,500 admissions in 2008, an increase of approximately 58% in just one year. To date (7/10/09), 710 offenders have been admitted to SAI in Based on this number, it appears that admissions for 2009 will be even higher at the end of As can be seen in Table 1, the demographic characteristics of SAI admissions have remained similar over the past 3 years. Admissions have been primarily male with some indication of increase in female trainees in 2009 and the offenders have been approximately 50% White and 50% Black consistently since The age of SAI admissions has remained constant as well, with persons in the 18 to 22 year age range making up almost 50% of the admissions. The offense categories of those persons admitted to the SAI program have remained somewhat stable between 2007 and 2009 with non-assaultive offenses making up approximately 50% of the primary offenses for all SAI admissions all three years examined. Drug and other assaultive offenses made up approximately 25% of primary offenses for the remaining trainees during this same period. The offenses for the prison and probation referrals were examined (see Table 2) and the offense make up of the two groups of SAI admissions do 19

26 not differ significantly. Other non-assaultive offenses make up about 50% of the primary offenses for both prison and probation Table 1. SAI Admissions, Sex Race* Age Referral Source # % # % # % Male Female Black White Asian Indian Mexican Other Unknown < = Probation Prison *When comparing across tables it should be noted: Race information for offenders comes from two different sources: Offender Management Network Information (OMNI) and CMIS. The race labels used in CMIS and OMNI are different. While both systems use the labels ASIAN, BLACK, and WHITE. The labels MEXICAN, INDIAN, and OTHR-RACE are only used in CMIS; the labels AMERICAN INDIAN OR ALASKAN NATIVE, NATIVE HAWAIIAN OR OTHER PACIFIC ISLANDER, and UNKNOWN are only used in OMNI. 20

27 admissions into SAI for all three years examined however, drug offenses make up a slightly higher percentage of primary offenses for prisoners compared to probationers. The reverse is true for probationers, with other assaultive offenses making up a slightly higher percentage of offenses compared to prisoners. None of the prisoner offenses at admission were criminal sexual conduct offenses and these make up and less than 0.5% of the probationer offenses. Table 2. Primary Offenses for SAI Admissions, Referral Type Year of Admission Offense Category # % # % # % Prisoners Probationers Other Nonassaultive % % % Drug % % % Other Assaultive % % % Criminal Sexual Conduct 0 0.0% 0 0.0% 0 0.0% TOTAL % % % Other Nonassaultive % % % Drug % % % Other Assaultive % % % Criminal Sexual Conduct 1 0.3% 3 0.4% 0 0.0% TOTAL % % % Total prison and probation admissions during the same period were compared to see how the SAI admissions differ from system-wide prison and probation admissions. As illustrated 21

28 in Table 3, the demographic characteristics of the prison admissions remained relatively stable between 2007 and Like the SAI admissions, the prison admissions remained over 90% male but unlike the SAI population, the percentage of female admissions did not increase over the past three years, rather, it decreased slightly from 7.1% of prison admissions in 2007 to 6.2% of admissions for this group in Like the SAI group, the race make-up of this admissions group is about evenly divided between Black and White offenders for all 3 years. Not surprisingly, the total prison admissions group has a different age structure with only 18% of admissions aged years compared to 43% of SAI admissions aged years. The majority of offenders who were admitted to SAI via prison were for other non- assaultive offenses in 2007, 2008, and A little over 30% of prison admissions were convicted of other assaultive offenses and only 14% of these offenders were admitted with drug offenses between 2007 and Statewide probation admissions are examined in Table 4 below. Like the SAI and prison admissions, males are the vast majority of admissions. However, females make up a higher percentage of the statewide probation admissions -- 20% versus less than 10% for SAI admissions and prison admissions. Once again, the race breakdown is about 50% White and 50% Black for all three years. However, the age of probation admissions differs from both the SAI and prison groups. The SAI admissions were the youngest with almost 50% of the offenders falling into the year age range. The age of prison admissions was distributed between 18 and 47 years with only about 17% of them falling into the 18 to 22 year age range. The age of admission in the probation group falls between the two previously examined groups, with about 30% of the admitted offenders in the year age group and most of the remaining offenders (approximately 40%) aged years. 22

29 Table 3. Total Prison Admissions, Offense Age Race* Sex # % # % # % Male 11, , , Female Black White Asian Indian Mexican Other < = Criminal Sexual Conduct Other Assaultive Drug Other Non Assaultive *When comparing across tables it should be noted: Race information for offenders comes from two different sources: Offender Management Network Information (OMNI) and CMIS. The race labels used in CMIS and OMNI are different. While both systems use the labels ASIAN, BLACK, and WHITE. The labels MEXICAN, INDIAN, and OTHR-RACE are only used in CMIS; the labels AMERICAN INDIAN OR ALASKAN NATIVE, NATIVE HAWAIIAN OR OTHER PACIFIC ISLANDER, and UNKNOWN are only used in OMNI. 23

30 Table 4. Probation Admissions, Male Female Missing Race* Age American Indian Or Alaskan Native Asian Black Native Hawaiian Or Other Pacific Islander Unknown White < = Missing *When comparing across tables it should be noted: Race information for offenders comes from two different sources: Offender Management Network Information (OMNI) and CMIS. The race labels used in CMIS and OMNI are different. While both systems use the labels ASIAN, BLACK, and WHITE. The labels MEXICAN, INDIAN, and OTHR-RACE are only used in CMIS; the labels AMERICAN INDIAN OR ALASKAN NATIVE, NATIVE HAWAIIAN OR OTHER PACIFIC ISLANDER, and UNKNOWN are only used in OMNI. 24

31 Aside from demographic characteristics, offender risk and program needs were examined to further examine the scope of the SAI population. As was discussed earlier, the SAI Program now utilizes the COMPAS instrument as a risk and need assessment for each offender within 7 days of admission. COMPAS scales for a snapshot of the population taken on May 19, 2009 (see Tables 5 and 6). Fifteen key scales that measure risk and programmatic needs were examined for this population. There was a great deal of consistency in terms of risk and program need for the prison referrals to SAI. The majority of folks fell into the low risk/need dimension on all fifteen scales with 3 exceptions (Table 5). For substance abuse, 50% of trainees were rated identified as having a high level of risk/need with 20% and 27% in the medium and low risk/need groups respectively. On the recidivism and violence risk dimensions approximately 30% of trainees were rated as high risk on these scales and about 30% were rated as medium risk, leaving approximately 40% the trainees identified in the low category. Table 5. COMPAS Scales, Prison Referrals to SAI PRISON REFERRALS High Medium Low Scales # % # % # % Violence Risk % % % Recidivism Risk % % % Criminal Involvement % % % Non-compliance History % % % Violence History % % % Current Violence % % % Social Isolation % % % Substance Abuse % % % Criminal Personality % % % Cognitive Behavioral % % % Family Criminality % % % Financial Problems % % % Vocation/Education % % % Social Environment % % % Residential Instability % % % 25

32 Table 6. COMPAS Scales, Probation Referrals to SAI PROBATION REFERRALS High Medium Low Scales # % # % # % Violence Risk % % % Recidivism Risk % % % Criminal Involvement 2 1.4% % % Non-compliance History % % % Violence History % % % Current Violence % 5 3.4% % Social Isolation % % % Substance Abuse % % % Criminal Personality % % % Cognitive Behavioral % % % Family Criminality % % % Financial Problems % % % Vocation/Education % % % Social Environment % % % Residential Instability % % % Table 6 displays the ratings for offenders who entered the SAI Program as probation referrals. These persons tend to be younger and they enter the SAI Program earlier in their criminal careers, hence they had a probation sentence as opposed to a prison sentence. Like the prison referrals, over 50% of the probation SAI trainees were rated as having a low risk/need on the following dimensions: criminal involvement, violent history, current violence, social isolation, family criminality, financial problems, and residential instability. While about 40% of the prison referrals were rated as having a low violence and recidivism risk, only 16% of the probation referral fell in the low risk group with 47% and 51% rated as high risk on the violence and recidivism risk dimensions. Based on these data it appears that the higher risk probationers, or those considered most likely prison bound are being admitted to the SAI Program. For the non-compliance history dimension, 41% of the probation referrals fell into the high risk/need category while another 42% fell into the low risk/need category. The probation 26