Us Doj Internal Prison Classification Systems 2002
Download original document:
This text is machine-read, and may contain errors. Check the original document to verify accuracy.
DEP A A L IN EC ON TIO NS N ATI National Institute of Corrections NT OF J ME U RT ICE ST U.S. Department of Justice S TI TUTE OF C OR Case Studies in Their Development and Implementation R U.S. Department of Justice National Institute of Corrections 320 First Street N.W. Washington, DC 20534 Morris L. Thigpen, Director Larry Solomon, Deputy Director Susan M. Hunter, Chief, Prisons Division Sammie D. Brown, Project Manager National Institute of Corrections World Wide Web Site http://www.nicic.org Internal Prison Classification Systems: Case Studies in Their Development and Implementation Patricia L. Hardyman, Ph.D. James Austin, Ph.D. Jack Alexander, Ph.D. Kelly Dedel Johnson, Ph.D. Owan C. Tulloch, M.C.P. January 2002 The research for this report was funded by National Institute of Corrections under cooperative agreements #93P045HV5, 97P06GIE8, and 98P06GII0 with the National Council on Crime and Delinquency and partially subcontracted to The George Washington University. Points of view or opinions stated in this document are those of the authors and do not necessarily represent the official position or policies of the U.S. Department of Justice or the National Council on Crime and Delinquency. Acknowledgments We would like to sincerely acknowledge the National Institute of Corrections (NIC) for its continuing commitment to the development of the literature on internal classification systems. In particular, we want to acknowledge and thank our program manager, Sammie D. Brown, for her guidance, expertise, involvement, and patience with our efforts. Assisting The Institute on Crime, Justice, and Corrections (ICJC) in this work were Jack Alexander, James Aiken, and Lorraine Fowler, who provided a great deal of expertise in the development and implementation of these internal systems. The internal classification initiatives would have been fruitless without the dedication and endless hours of hard work expended by the steering committees from the state correctional agencies participating in these initiatives. It was their creativity, determination, and tenacity that created, tested, and refined the models. Thank you. The efforts were led by Brian Bemus, Jim Maras, and Brian Belleque in Oregon; Stan Repko and Michelle Ricci in New Jersey; S. Fred Roesel, Chuck Manning, and William Maust, Jr., in Florida; Jean Ann Johnson, Nick Noll, Mike Kemna, and Terry Page in Missouri; Laurie Feiler and Ed Ligtenberg in South Dakota; Major Christine Whidden in Connecticut; James Thatcher in Washington state; and Benjamin Griego, Warden Robert Furlong, and Captain Ron Carter in Colorado. iii Abstract NIC, the National Council on Crime and Delinquency (NCCD), and ICJC worked with eight states over a 7-year period to develop, pilot-test, implement, and evaluate internal prison classification systems. Seven unique models were tested. Florida, Connecticut, and Colorado developed computerized objective, behavior-based models for housing and program assignments. Oregon developed an objective model based on behavioral and compatibility indicators for its female population, although preliminary results suggest that the system works equally well for its male population. New Jersey developed a behavior-based model for identifying the aggression levels of its maximum-custody inmates. Washington state analyzed the utility of the Adult Internal Management System (AIMS) for case management of minimum-custody inmates with long sentences. South Dakota and Missouri developed a personality-based system, the Adult Internal Classification System (AICS), modeled after AIMS. The barriers to the development and implementation of these diverse systems led to the development of a model process and timetable for designing and implementing internal classification systems. The importance of clearly identifying the targeted inmate population and the issues to be addressed by the system was a critical lesson learned by each state. The diversity of the models developed suggests that there is no “best model,” nor should there be one. Instead, the instruments and process must be tailored and validated to the specific populations for which they will be used. A set of standards applicable to all internal classification systems was identified, however. The diversity of the models developed suggests that there is no “best model,” nor should there be one. Instead, the instruments and process must be tailored and validated to the specific populations for which they will be used. v Contents Acknowledgments ..............................................................................................iii Abstract ................................................................................................................v Executive Summary ...........................................................................................xi Chapter 1. Introduction .....................................................................................1 Statement of the Issues ...............................................................................1 The Role of Internal Classification Systems ..............................................1 NIC Internal Prison Classification Initiatives .............................................6 NIC Internal Prison Classification Initiative Tasks.....................................7 Chapter 2. Description of the States’ Internal Classification Initiatives .....11 Connecticut Department of Correction.....................................................11 Colorado Department of Corrections........................................................17 Washington State Department of Corrections ..........................................25 Oregon Department of Corrections...........................................................31 Florida Department of Corrections ...........................................................40 New Jersey Department of Corrections ....................................................53 South Dakota Department of Corrections.................................................63 Missouri Department of Corrections ........................................................65 Missouri and South Dakota Internal Classification Initiative...................66 Chapter 3. Common Problems, Issues, and Solutions ...................................77 Lessons Learned........................................................................................77 Step-by-Step Process for Designing and Implementing an Internal Classification System................................................................................81 Implications and Future Steps...................................................................86 vii Contents Notes ..................................................................................................................89 References...........................................................................................................93 Suggested Readings ...........................................................................................95 Appendix .............................................................................................................99 State of Missouri Department of Corrections Checklist for the Analysis of Life History Records of Adult Offenders (CALH) .............101 State of Missouri Department of Corrections Correctional Adjustment Checklist..............................................................................102 IDOC Internal Classification Form: Maximum Security Facilities .......103 viii Internal Prison Classification Systems List of Exhibits Exhibit 1. Overview of External and Internal Classification Systems ..................3 Exhibit 2. MacDougall Internal Housing Plan ....................................................14 Exhibit 3. Partial List of MacDougall Job and Program Assignments ...............15 Exhibit 4. MacDougall Correctional Institution New Admission Classification Form ..............................................................................................16 Exhibit 5. State of Colorado Department of Corrections Admission Data Summary ..............................................................................................................22 Exhibit 6. Colorado Internal Classification Master Program Scheduling System...............................................................................................23 Exhibit 7. Assaults at Limon Correctional Facility, Fiscal Years 1995–96 to 1998–99 ...........................................................................................................25 Exhibit 8. Washington State AIMS Pilot Test Results: Comparison of Disciplinary Infractions ...................................................................................30 Exhibit 9. Sentence Disciplinary Grid for Oregon Internal Classification System...........................................................................................36 Exhibit 10. Oregon Corrections Information Systems, Recommended Inmate Cell Mates................................................................................................37 Exhibit 11. Florida Risk and Needs Instrument ..................................................45 Exhibit 12. Institutional Disciplinary Index (INDI) by Housing Assignment ....50 Exhibit 13. Housing Placement Results by Gender ............................................51 Exhibit 14. Program Placement According to First Recommended Assignment ..........................................................................................................52 Exhibit 15. Program Placement According to Second or Third Recommended Assignment..................................................................................53 Exhibit 16. Percent Agreement Between Reception Center and Facility Staff on Risk and Needs Items ............................................................................54 Exhibit 17. NJDOC Internal Classification Form: Maximum Security Facilities.................................................................................................60 ix Contents Exhibit 18. AICS History Checklist ....................................................................68 Exhibit 19. AICS Initial Behavioral Checklist ....................................................69 Exhibit 20. AICS Reclassification Behavioral Checklist ....................................70 Exhibit 21. Missouri and South Dakota Pilot Test Reliability Rates Percent Agreement Between Raters One and Two...........................................................72 Exhibit 22. Comparison of Assaults by AICS and AIMS Classification Types for Reclassification Cases: February–March 2000 ...................................73 Exhibit 23. AICS and AIMS Classifications Compared for All Cases Classified: February–March 2000 .......................................................................74 Exhibit 24. Racial Distribution of Inmates Under AICS for Reclassification Cases Classified: February–March 2000 ..................................75 Exhibit 25. Standards and Guidelines for Internal Prison Classification Systems .........................................................................................79 x Executive Summary The past decade has witnessed tremendous growth in the number of adult prison systems that employ objective external classification systems to determine the appropriate custody levels for increasing numbers of adult inmates. In addition, several systems have implemented internal classification systems to guide housing, program, and/or work assignment decisions to promote better inmate management at the facility level. Recognizing the critical need to develop and refine internal classification systems, the National Institute of Corrections (NIC) funded three initiatives to work with state correctional agencies to develop, implement, and assess the impact of objective internal classification systems. The specific goals of these NIC initiatives were: Phase I: Field test internal prison classification systems in three states (1993–96). —Design and pilot-test internal classification systems in three states (Connecticut, Colorado, and Washington state). Several systems have implemented internal classification systems to guide housing, program, and work assignment decisions to promote better inmate management at the facility level. —Conduct a national survey of existing internal classification methods. Phase II: Design, develop, and implement internal prison classification systems (1997–98). —Develop a training curriculum that addresses internal classification issues, including the design, development, implementation, evaluation, and operation of an internal classification system. —Provide training and technical assistance to five state correctional departments that were committed to improving the internal management of their inmates (Oregon, Florida, Missouri, South Dakota, and New Jersey). —Assess the outcome and impact of the training and assistance provided through this initiative. Phase III: Implement and assess internal prison classification systems (1998–2000). —Provide continued technical assistance to help state correctional agencies implement the internal classification systems developed and tested during the Phase II initiative. —Assess the progress to date and impact of the Phase I internal classification systems. —Develop a publication that discusses the state of the art in internal classification and provides guidance to state correctional agencies regarding the design and implementation of internal classification systems. xi Executive Summary NIC sought a diver- Eight states were selected for these initiatives based on the sophistication of their external prison classification systems, the operation of automated inmate tracking systems, a strong commitment from system and facility administrators, the appointment of a working steering committees to move the project forward quickly, and the clarity with which the states could identify problems to be targeted by the internal classification systems. NIC also sought a diversity of correctional systems in terms of the size of the inmate population, geographic location, and the type of system to be developed. The eight states and their initiatives are summarized below. sity of correctional systems in terms of the size of the inmate population, geographic location, and the type of system to be developed. The Connecticut Department of Corrections developed an objective method for making work and program assignments for high-custody male inmates. Because housing assignments within the test facility, MacDougall Correctional Institution, are based on work and program assignments, the primary objective was to provide structure to the housing assignment process. The Washington Department of Corrections sought to enhance the effectiveness of institutional case management and control of inmates by developing programs specific to the inmates’ needs and providing better transition links between correctional facilities and the community. The Colorado Department of Corrections sought to improve the quality and accessibility of information for internal classification decisions (especially housing decisions) by developing an automated master program scheduling system to structure program placement based on the inmates’ risk and needs, program eligibility, and phase of confinement. The Oregon Department of Corrections implemented a behavior-based system to improve the specificity and compatibility of housing assignments in both men’s and women’s correctional facilities. The original plan was to create independent systems, with different criteria and scoring procedures, for its male and female populations. Analysis showed, however, that separate systems for male and female inmates did not appear to be warranted. The system originally designed for the female population has been automated and implemented successfully throughout the system. Continued research is needed, however, to ascertain its validity and reliability. The Florida Department of Corrections embarked on an ambitious process for systemwide implementation of a complex Risk and Needs Model that would identify internal management issues as well as program and job-related needs. The model’s complexity required sophisticated software and automation. Despite these significant challenges, Florida has successfully implemented and automated this system in all of its 141 correctional facilities. The New Jersey Department of Corrections identified maximum-custody inmates for an objective assessment of inmate aggression levels. The original plan called for the transfer of highly aggressive inmates to a centralized housing unit, but New Jersey encountered significant fiscal barriers to this task. Instead, contingent xii Internal Prison Classification Systems on the results of a validity and reliability study, New Jersey plans to establish separate high-aggression units at each of its maximum-security facilities. The Missouri Department of Corrections used the Adult Internal Management System (AIMS) before this initiative to classify its inmates according to behavioral typologies that dictated housing compatibility. To improve the accuracy and utility of this system, Missouri (in partnership with South Dakota) developed the Adult Internal Classification System (AICS), modeled after AIMS. The pilot-test results indicated low inter-rater reliability for the internal classification process. Additional staff training and modifications of the instruments were undertaken to improve the reliability of AICS. The South Dakota Department of Corrections developed AICS, an internal classification system modeled on AIMS, for the internal management of its facilities, in partnership with Missouri. Full implementation of AICS was delayed pending the resolution of validity issues. South Dakota continues to work with Missouri to refine the system. The barriers to implementing these diverse internal classification systems and the strategies for overcoming them have led to several important lessons for this and future initiatives. First, proper implementation requires increased specificity at the beginning of the planning phase and additional opportunities for further refinement toward the end. The major steps for developing and implementing an internal classification system are as follows: 1. Obtain a formal commitment from the central office. 2. Designate a strong project manager and establish a working steering committee. 3. Identify stakeholders and include them in the process. 4. Define problems to be addressed and set realistic goals and measurable objectives. 5. Select the type of internal classification system to be adopted. 6. Select the pilot site. 7. Analyze current housing, work, and program assignment procedures. 8. Conduct a facility program and work assignment inventory. 9. Conduct a facility housing and bed inventory. 10. Develop a prototype instrument and policy manual. 11. Pilot-test policies and instruments. xiii Executive Summary 12. Develop a full implementation plan. 13. Monitor and evaluate the system. Future internal classification system initiatives must respond to the diversity of facilities, populations, factors, and models. This diversity suggests that there is still much to learn about internal classification. xiv In addition to encouraging states to identify the targeted inmate management problems clearly at the outset of the initiative, states are also encouraged to simplify their proposed instruments to reduce staff training needs and make the system more flexible. No distinct set of ideal, generalizable factors was identified to be included in an internal classification system. Instead, the critical risk factors, operational definitions, processes, and timing appeared to be unique to each state and dependent upon its specific goals, resources, and system composition. In other words, there is no “best model,” nor should there be; instruments and processes must be tailored to and validated on the population for which they will be used. Future internal classification system initiatives must respond to the diversity of facilities, populations, factors, and models. This diversity suggests that there is still much to learn about internal classification. At the same time, state and local correctional and detention systems continue to face growing inmate populations and declining resources. The need to develop new management techniques for controlling and servicing the prison population with fewer resources becomes more critical with each new prisoner admitted to the system. Future technical assistance efforts should focus on helping states develop systems that are both practical and feasible, given these harsh realities. one Chapter Introduction Statement of the Issues During the past decade, prison systems have experienced increased pressure to improve their system of classifying inmates according to custody, work, and programming needs. Fueled by litigation and overcrowding, classification systems are viewed as the principal management tool for allocating scarce prison resources efficiently and minimizing the potential for violence or escape. These systems are also expected to provide greater accountability and forecast future prison bed-space needs. In other words, a properly functioning classification system is seen as the “brain” of prison management, which governs many important decisions, including those that heavily influence such fiscal matters as staffing levels, bed space, and programming. Although objective prison classification systems were adopted by many states in the 1980s, the late 1990s witnessed significant improvements in classification practices. The level of overclassification has been reduced, custody decisions are made more consistently, criteria for custody decisions have been validated, inmate program needs are assessed systematically, and institutional safety for both staff and inmates has been enhanced. Despite these improvements, additional issues remain unresolved within prison classification systems. In particular, decisions at the institutional or internal level that guide housing, program, and work assignments need to be as structured and organized as those made at the system or external level. As correctional facilities become more crowded, internal classification decisions play a more significant role. The widespread use of double-celling in high-security units and the expanding use of dormitories for low- and medium-custody inmates have triggered the need for a systematic process for assigning inmates to beds or cells. As inmate populations continue to increase, decisions governing housing and programs, especially for inmates with extremely long sentences, will become increasingly difficult. For correctional officials to make more informed decisions, a second layer of prison classification—internal classification—is now required. A properly functioning classification system is seen as the “brain” of prison management, which governs many important decisions, including those that heavily influence such fiscal matters as staffing levels, bed space, and programming. The Role of Internal Classification Systems To deal effectively with the varying degrees of risk presented by inmates, some prison systems have begun to classify inmates by personality or behavioral typologies 1 Chapter 1 (Megargee et al., 1979; Quay, 1984). Some offender typology systems are designed by psychologists and are reasonably well researched. Others are simply sets of additional criteria applied by individual institutions to augment the external classification system. All of these systems endeavor to address housing, program, and compatibility issues to improve inmate management at the facility level. The task of an internal system is to devise appropriate housing plans and program interventions within a particular facility for inmates who share a common custody level (minimum, medium, close, or maximum). In short, external classification models influence interinstitutional placement, whereas internal management systems focus on intrainstitutional placement and program assignment. Internal classification systems are designed to complement the previously established objective custody-classification (or external) systems. The task of an internal system is to devise appropriate housing plans and program interventions within a particular facility for inmates who share a common custody level (minimum, medium, close, or maximum). In short, external classification models influence interinstitutional placement, whereas internal management systems focus on intrainstitutional placement and program assignment. Exhibit 1 illustrates how an internal classification system should function within a prison classification system. On admission, inmates are first assessed using the external, systemwide classification system to determine their custody level (maximum, close, medium, minimum, or community), program needs, and other needs that may require special housing. This custody classification determines the type of facility or housing unit in which inmates should be housed. For example, maximum-custody inmates will be transferred to a maximum-security facility. On arrival at the appropriate facility, inmates undergo a second formal classification review—the internal classification process. Given their behavior, personality traits, and specific program needs, classification staff identify the appropriate housing unit or cell block, programs, and work assignments for them. As with external classification systems, formal internal classification systems may include structured scoring instruments, staff specialists who have been formally trained to use them, and a reclassification process to update previous classification records. To date, formal internal classification systems have not been widely implemented, although many informal systems do operate. A 1994 survey of state correctional agencies by the National Council on Crime and Delinquency found that only nine states had a formal internal classification system (Alexander et al., 1997). Some kind of internal classification process exists in all prison systems to assign newly arrived inmates to housing units, work assignments, and programs, but these processes are usually informal and rely on subjective criteria. Most respondents to the NCCD survey expressed strong interest in developing a formal internal classification system. Three types of formal internal classification systems that have some level of documentation are listed below. Adult Internal Management System (AIMS) AIMS, developed by Herbert Quay more than 20 years ago, is one of the best known internal management systems.1 Its purpose is to reduce institutional predatory behavior by identifying predators and separating them from vulnerable inmates. AIMS is used by several facilities in the Federal Bureau of Prisons and the 2 Introduction Exhibit 1. Overview of External and Internal Classification Systems Admission to prison Initial classification Custody assessment Program needs assessment Facility designation Transfer to facility Internal classification Housing assignment Program assignment Work assignment Transfer to designated housing area Reclassification External classification – Custody Internal classification – Programs – Facility – Housing – Community programs – Work Ohio Department of Rehabilitation and Correction. The South Dakota and Missouri Departments of Corrections also have fully implemented AIMS for their male inmate population. The South Carolina Department of Corrections implemented AIMS, but has stopped using it as a housing assignment tool. AIMS attempts to identify and separate inmates according to a personality typology. This classification relies on two instruments (copies of the AIMS checklists used by the Missouri Department of Corrections are provided in the appendix): 3 Chapter 1 Life history checklist. This checklist includes 27 items designed to assess inmates on personality dimensions known to relate to their ability to be housed successfully with other types of inmates. Checklist items focus on the inmate’s adjustment and stability during time spent in the community. Correctional adjustment checklist. This checklist includes 41 items that are designed to profile the inmate’s behavior while in a correctional setting. These items focus on the inmate’s record of misconduct, ability to follow staff directions, and level of aggression toward other inmates. AIMS identifies inmates who are likely to be incompatible in On the basis of these scores, inmates are classified into one of five groups. Originally, group labels described personality types (e.g., aggressive-psychopathic, manipulative, situational, inadequate-dependent, and neurotic-anxious), but they have since been revised to the following most common schema: Alpha I, Alpha II, Kappa, Sigma I, and Sigma II. Inmates in these groups are assumed to have different rates and types of institutional misconduct. More specifically, Alpha I and II inmates are characterized as the offenders most likely to be a threat to the safety and security of the facility. Alpha I inmates are more likely to openly exhibit aggressive or assaultive behavior than other types of inmates, whereas Alpha II inmates are more likely to be manipulative. Sigma I and II inmates are unlikely to be assaultive, but they pose other management problems, such as disregarding direct orders and disrupting the orderly operation of the facility. Kappa inmates are least likely to present management problems. Alpha I and II inmates are characterized as predators, whereas Sigma I and II inmates are characterized as inmates at risk of being victimized. Kappa inmates are neither predators nor prey. Thus, AIMS identifies inmates who are likely to be incompatible in terms of housing and inmates who are most likely to pose a risk to the safe and secure operation of a facility. Washington state pilot-tested AIMS as a strategy for housing and program planning for minimum-custody inmates with lengthy sentences. AICS, an internal classification system modeled after AIMS, was pilot-tested by Missouri and South Dakota. terms of housing and inmates who are most likely to pose a risk to the safe and secure operation of a facility. Prisoner Management Classification System (PMC) PMC, developed by Gary Arling and Ken Lerner of the Wisconsin Division of Corrections, was adapted from an offender management system developed originally for probation and parole services (Lerner, Arling, and Baird, 1986). This system is also commonly called Client Management Classification (CMC) or Strategies for Case Supervision (SCS). NIC has widely supported training in the use of this system, and an estimated 8,000 to 10,000 staff nationwide have been trained in its use.2 PMC requires a semistructured interview to assist staff in identifying potential predators and victims and those inmates who require special programs or supervision. The interview (a series of forced-choice responses) and ratings of objective background factors are used to classify each inmate. Once inmates are classified, detailed case handling guidelines provide staff with management techniques for safe and appropriate handling of inmates within their designated housing units. 4 Introduction Equally important, these guidelines highlight programming approaches to prepare inmates for readjustment to the community. On admission, the PMC interview is conducted by a specially trained classification officer and requires approximately 45 minutes. The interview protocol includes 45 attitudinal items that deal sequentially with the inmate’s attitudes regarding the current offense; criminal history (including juvenile justice history); family relationships; relationships with staff, inmates, and peers; current difficulties (psychological, sexual harassment, etc.); and plans after release. In addition to these attitudinal items, 11 factual ratings assess the inmate’s social status and offense history and 8 behavioral ratings assess the inmate’s demeanor during the interview. The assessment concludes with the interviewer’s impressions of the inmate’s most and least urgent problem areas. Inmates are assigned to one of four groups: Limited Setting (LS), Casework Control (CC), Selective Intervention (SI), and Environmental Structure (ES). LS and CC inmates are expected to be more aggressive and harder to control, whereas SI and ES inmates require minimal supervision and should be separated from LS and CC inmates. When necessary, however, SI inmates can be housed with LS and CC inmates. Although experimentally validated and shown to be useful in managing correctional populations, PMC requires significant staff training for inmate assessment, supervision, and interaction. Behavior-Based Systems As an alternative to these systems and their accompanying training requirements, a third type of system has been developed that assesses inmates according to behavioral measures, as reflected by disciplinary records and work performance, in contrast to a personality-based assessment. The strength of behavior-based systems is their high degree of structure, which builds directly on the inmate’s custody score by surveying dynamic measures of in-custody behavior. This type of system has been implemented by the Illinois Department of Corrections (IDOC) in its three maximum-security facilities. Inmates are scored according to their level of aggression, which is largely determined by the severity and frequency of disciplinary conduct and gang-related activities. Several items that are commonly part of external classification systems, such as current offense and age, are also included. A historical assessment of the type of disciplinary incidents and resulting days spent in segregation provides a dynamic measure of the likelihood of aggression. (See the appendix for the internal classification form used in IDOC’s maximum-security facilities.) This ability to assess the change in risk level over time is one of a behavior-based system’s key strengths. Three behavior-based classification systems pilot-tested by three sites in Florida, New Jersey, and Oregon are described in this report. The strength of behavior-based systems is their high degree of structure. 5 Chapter 1 NIC Internal Prison Classification Initiatives Recognizing the critical need to develop and refine internal classification systems, NIC funded three initiatives to work with a total of eight states to develop and implement internal classification systems. The specific goals of these initiatives were as follows: Phase I: field-test internal prison classification systems in three states (1993–96). —Design and pilot-test internal classification systems in three states (Connecticut, Colorado, and Washington state). —Conduct a national survey of existing internal classification methods. Phase II: design, develop, and implement internal prison classification systems (1997–98). —Develop a training curriculum that addresses internal classification issues, including the design, development, implementation, evaluation, and operation of an internal classification system. —Provide training and technical assistance to five state correctional departments (Oregon, Florida, Missouri, South Dakota, and New Jersey) committed to improving the internal management of their inmates. —Assess the outcome and impact of the training and assistance provided through this initiative. Phase III: Implement and assess internal prison classification systems (1998–2000). —Provide continued technical assistance to state correctional agencies to help implement the internal classification systems developed and tested during Phase II. —Assess the progress to date and impact of the Phase I internal classification systems. —Develop a publication that discusses the state of the art in internal classification and provides guidance to state correctional agencies on the design and implementation of internal classification systems. In 1993, for Phase I, NIC developed a cooperative agreement with NCCD to initiate internal classification systems in three states (Connecticut, Colorado, and Washington). The goal of this effort was to design and prepare to implement three different internal classification systems. NIC endeavored to create not a single 6 Introduction model or system to be adopted by all correctional agencies, but, rather, a standard process for designing and implementing internal classification systems (Alexander et al., 1997). NIC endeavored to create not a single The second NIC internal classification initiative was intended to build on the foundation established in Phase I and to further the use of internal classification systems by adult prison systems. The primary goal of Phase II was to assist five state departments of corrections in developing, implementing, and evaluating internal classification policies and procedures to improve their systems’ overall operations. NIC’s third initiative was intended to continue the technical assistance available for full implementation of the systems in the five Phase II states and to assess the status and impact of the internal classification systems implemented by the Phase I states. This report, a product of this third initiative, documents the activities, methods, and outcomes of NIC’s work with the eight states involved in the three internal classification initiatives. The remainder of this section describes the process by which the eight states were selected for participation and outlines the processes undertaken by the states. The next chapter of this report is organized by state and describes— model or system to be adopted by all correctional agencies, but, rather, a standard process process for designing and implementing internal systems. classification systems The problems, issues, and goals of the internal classification initiative and current classification system. The internal classification design process. The internal classification system’s impact on prison operations, its current status, and the next steps to be taken. The third chapter documents the common problems, issues, and solutions encountered across the states, provides a step-by-step process for designing and implementing internal classification systems within state prison systems, and highlights future directions for the development of internal classification systems. NIC Internal Prison Classification Initiative Tasks The processes for selecting states to participate in the respective NIC internal prison classification initiatives were similar for the three phases. The Phase I participants—Connecticut, Colorado, and Washington—were identified through data collected by NCCD during the national survey of the use of internal classification systems among state correctional agencies. The primary criterion for participation in Phase I was that the state had to have a fully automated external classification system. In addition, the sites were selected to reflect a diversity in geographic location and size of the prison system. For Phase II, NIC sent a letter of invitation to the director of each state department of corrections introducing the internal classification demonstration project. States 7 Chapter 1 were offered an opportunity to develop a comprehensive internal classification system tailored to their specific needs. Included with the invitation was an application that asked each state to provide an overview of its current external classification system,3 indicate the level of automation of its external classification system, describe its classification staffing pattern, identify and describe the facility for the pilot test, indicate the composition of the proposed steering committee, and identify the current problems that might be solved or alleviated by an internal classification effort. Applications were received from more than 20 states, representing the entire spectrum of state correctional systems—small to large, all regions of the country, and diverse levels of sophistication with respect to automation and current classification systems. This diversity suggested that internal classification was still a critical issue faced by many correctional agencies. States selected to participate in Phase II had to have— A validated and established external classification system. An automated system for accurately tracking inmate transfers and assignments within and across facilities. High levels of commitment from the warden and key staff at the pilot-test facility. A steering committee composed of representatives from all major operational areas within the department. Adequate resources for staff to design and pilot-test the classification system and to travel to the two national workshops. Clearly identified goals for the internal classification system. The five states chosen were diverse in terms of region of the country; rated capacity and custody level (maximum, close, medium, minimum, special management, and females) of the pilot-test facility; system needs and goals; and type of internal classification system to be evaluated (behavioral, personality traits, etc.). This diversity was sought deliberately; by choosing sites with differing circumstances and questions, NIC could be assured that the initiatives would have broad generalizability. Once the applications were reviewed, the proposed chair of each steering committee was interviewed to clarify the information provided on the application and assess the state’s commitment to the project. States were asked if they were willing to commit staff time and resources for travel to the national seminars and if they were fully prepared for the system changes required to develop and implement an internal classification system. 8 Introduction Based on a review of the applications, additional documentation submitted by the states, and the interview data, Florida, Oregon, New Jersey, South Dakota, and Missouri were selected to serve as the demonstration sites for the Phase II internal classification initiative. For Phase I and II initiatives, the selected states were asked to complete the following tasks: Establish a steering committee and identify a pilot facility for the project. Assess the department’s current external and internal classification systems. Develop preliminary internal classification instruments and procedures. Pilot-test the draft system on a sample of inmates at the selected facility and make any necessary revisions to the system. Develop a plan to implement and monitor the internal classification system. In addition, each Phase II state was asked to participate in two seminars with the four other Phase II states to review internal classification systems, develop implementation plans, and define strategies for monitoring their systems. During Phases I and II, NIC conducted comprehensive onsite assessments of the states’ current external and internal classification systems, practices, policies, and plans. An onsite protocol was developed to ensure comprehensive, consistent, and comparable data. The site assessment protocol included the following activities: During Phases I and II, NIC conducted comprehensive onsite Orientation meeting with the steering committee. assessments of the Interviews with central office and facility administrative staff. states’ current exter- Observations of the external classification process. nal and internal Review of written external classification policies and procedures. classification systems, Observations of the internal classification process. practices, policies, and plans. An onsite Review of written internal classification policies and procedures. Examination of MIS (management information systems) data and analytic capabilities. Exit interview with the steering committee. protocol was developed to ensure comprehensive, consistent, and The next chapter of this report describes each state’s activities and the outcomes resulting from these initiatives. comparable data. 9 two Chapter Description of the States’ Internal Classification Initiatives Connecticut Department of Correction As of April 1, 1995, the Connecticut Department of Correction (CT DOC) housed 13,654 inmates in 16 facilities.4 An examination of the prison population indicated that most of the population was black (45.6 percent) or Hispanic (26.8 percent) and did not have a high school degree (62.5 percent). The modal age group was 20 to 29 years of age with a mean age of 29.8 years. Analyses of the CT DOC population by crime type indicate that 30.9 percent were committed for an offense against persons, 17.5 percent for property offenses, 25.5 percent for drug-related offenses, and 23.5 percent for public order offenses. The average sentence was 6.8 years. Women made up 6.9 percent of the prison population.5 CT DOC identified the MacDougall Correctional Institution near Hartford as the pilot-test facility for the internal classification system initiative. Description of Pilot Site MacDougall Correctional Institution is a modern, 1,017-bed, high-security facility (Level 4 in a five-level security rating system) built in 1993 for male inmates. In 1995, approximately 450 staff were employed at the facility, 210 of which were correctional officers. The average daily population at MacDougall has remained stable. In 1996, for example, the average daily population (ADP) was 959; in 1999, the ADP was 975. The main housing unit consists of five general population units, a 20-cell (doublebunked) restrictive housing unit, and a 22-cell medical unit. The general population housing units are managed as direct supervision or unit management divisions. Each general population housing unit is divided into two housing pods of 48 cells each, most of which are double-bunked. This adds up to a total bed capacity of 955 general population inmates, the primary target of this initiative. The MacDougall facility was designed to hold sentenced, high-custody male inmates with long prison terms. Inmates with long prison terms are likely to be transferred directly from CT DOC’s reception center; they are not viewed as management problems, but must be held in a high-security facility for the initial portion of their term. MacDougall also receives inmates from other institutions who have demonstrated an inability to adjust to lower security facilities and have been transferred with the hope that they will respond positively to the higher security and will 11 Chapter 2 not have to be transferred to the state’s only Level 5 facility, which houses extremely disruptive inmates. The facility was designed to operate large-scale industrial, vocational training, and educational programs. All inmates are assigned to a work detail or are enrolled in vocational training, academic programming, or drug treatment. The primary goal for the internal classification initiative was to develop a mechanism for assigning inmates to these program and work assignments. Comparisons between MacDougall’s population and the overall CT DOC population indicated that in 1995, during the development process, MacDougall had a larger proportion of inmates considered to be a “severe” risk than did the CT DOC system as a whole (95.5 percent versus 29.1 percent, respectively) (Connecticut Department of Corrections, 1999). The MacDougall inmates differed from other CT DOC inmates on two key risk factors: offense severity and history of violent offenses. Three-quarters (74.5 percent) of MacDougall inmates were incarcerated for major offenses, and one-quarter (23 percent) had a history of major or severe violent offenses. In contrast, one-third (33.6 percent) of the overall CT DOC population was incarcerated for a major offense, and less than one-seventh (13.7 percent) had a history of major or severe violent offenses. The MacDougall population had some critical issues to be addressed. Sixty-seven percent of the inmates had significant medical needs and 43 percent had significant mental health needs. In addition, 79 percent of the MacDougall inmates had a history of substance abuse and 22 percent had a known history of sex-related offenses. More than half of the MacDougall inmates (54 percent) were at or below the eighth-grade education level. Their vocational history scores could be attributed to their apparent lack of education. More than half (55 percent) had only a moderate work history and one-third (33 percent) had limited or no work skills. A review of the 1999 statistics indicates that MacDougall’s population has not changed substantially in the past 5 years. The only exception to this observation was that average sentence length has increased from 15.5 years to 18.6 years. The overall risk level has remained the same since 1995: 95 percent of the inmates are classified as a “severe risk” to the safety and security of the institution. Connecticut’s Classification Process CT DOC’s external classification system was implemented in 1992; the most recent revisions were implemented in June 1999. The goal of the external classification system is “to ensure the safety and well being of the community, facility, staff, and the inmate” (Connecticut Department of Correction, 1999: 2). The classification system was designed to assess inmates’ security, custody, and treatment needs using objective risk factors. The system is used by facilities for all inmates, regardless of legal status or sentence length. The classification ratings track individuals throughout the term of confinement. The Office of Offender Classification and Population Management at the Central Office is responsible for the development, implementa12 Description of the States’ Internal Classification Initiatives tion, training, oversight, and management of the external classification function within the department. The objective classification system is based on inmates’ risk and treatment needs. Inmates’ overall risk levels are determined by evaluating their escape history, history of violence, current offense severity, length of sentence, presence of pending charges or detainer, discipline history, and security risk group membership. Each factor is independently assigned a rating between one and five, with one representing the lowest risk and five representing the highest risk. Overall risk level is determined by the highest rating assigned on any one of the seven categories. The risk levels primarily dictate the structure or security required to house the inmates. In addition to the custody risk assessment, the classification process also assesses inmates on seven key areas of need: medical, mental health, education, vocational, substance abuse treatment, sexual treatment, and residence. Housing and work assignments are the responsibility of the classification committee at the receiving facility. Connecticut’s Internal Classification Initiative CT DOC’s internal classification initiative intended to establish an objective method for housing inmates in high-security facilities (Level 4 of 5). CT DOC was not interested in a personality-based system, but preferred a behavior-based system similar to the one developed for Illinois at its maximum-security facilities. Facility administrators were especially concerned about the lack of structure and purpose in the housing and programming decisions for inmates with long sentences. Additionally, they wanted to reduce the possibility of improper work or program assignments for these inmates. At the outset of this project, MacDougall was preparing to expand its work and industrial opportunities. It was hoped that a more structured internal classification process would improve the “fit” between inmates and their work and program assignments. After a 3-month planning process during which the facility and central office staff reviewed current practices and policies at MacDougall, a formal policy for internal classification was issued. This policy declared that “MacDougall Correctional Institution will classify inmates to program and job assignments according to their overall risk score and needs assignments as outlined in Administrative Directive 9.2, Inmate Classification. Inmates shall be assigned to a housing unit based upon their program and/or job assignment.” Connecticut’s internal classification initiative intended to establish an objective method for housing inmates in highsecurity facilities. The last sentence in the policy statement was critical. There was considerable discussion during the development phase as to whether inmates should be housed according to program assignments. From a logistics perspective, housing by program or job assignment greatly facilitates inmate movement, but it may also limit an inmate’s ability to participate in a work assignment or program because of lack of space within the housing unit. Moreover, housing by program or work assignment may intensify or consolidate the influence of gangs over the housing unit. Despite these challenges, CT DOC determined that a match between housing assignment and work or program placement was important to the smooth operation of the facility. Once the decision to house inmates according to work and program 13 Chapter 2 assignment was made, housing units within MacDougall were classified according to the major job and program assignments available. (Exhibit 2 lists the facility’s 12 major housing units and programs.) A detailed description was made of every job and program within the facility, identifying the restrictions, location, pay scale, classification level, number of positions, and number of days per week required. These descriptions are updated periodically as the number and types of positions change. (Exhibit 3 is a sample listing of the job descriptions throughout the facility.) On admission to the MacDougall facility, each inmate is initially assigned to the 95bed H–1 unit for assessment and orientation unless he has already been targeted for some type of special housing (e.g., medical transfer to the hospital unit or L–1). As Exhibit 2. MacDougall Internal Housing Plan Housing Unit Number of Beds Type of Inmates Assigned to Unit1 H–1 95 New inmates admitted to MacDougall from the reception center or interfacility transfer. Inmates admitted for inpatient medical reasons will not go through the orientation process. H–2 96 Unassigned inmates due to program failure or discipline problems. I–1 95 Inmates assigned to vocational programs. I–2 96 Inmates assigned to vocational programs. J–1 95 Inmates assigned to institutional jobs. J–2 96 Inmates assigned to institutional jobs. L–1 96 Inmates assigned to institutional jobs. L–2 95 Special needs inmates unassigned due to a specific medical condition. M–1 96 Inmates assigned to the morning session of an academic school program. M–2 95 Inmates assigned to the afternoon session of an academic school program. Restrictive Housing 40 Inmates placed in accordance with Unit Directive 9.4. Hospital Inpatient Unit 22 Inmates admitted by Health Services staff. Provides services/beds for the entire department. Housing unit janitors are assigned to their respective housing unit. 1 14 Description of the States’ Internal Classification Initiatives Exhibit 3. Partial List of MacDougall Job and Program Assignments Job Assignment Description Location Student Education School Vocational education: Business education Vocational Vocational education education Number Custody Pay Days/ Job of Slots Level Level Week Code 181 4 or below 3 5 1 15 4 or below 3 5 9 Housing janitor Janitor H–1 Unit 10 4 or below 1–4 7 21 Institutional painter Painter Various 4 4 or below 1–4 7 46 part of the orientation process, a counselor reviews with the inmate the facility’s mission and the range and criteria for all programs and work assignments. The orientation process also provides the inmate with an opportunity to indicate any special housing needs. Based on this interview and a review of the inmate’s record, the counselor completes an initial classification form and recommends a housing and program assignment. As shown in Exhibit 4, the internal classification form is a simple list that identifies the most critical programming and job assignments. This form is also used for reclassification, which occurs every 6 months or upon change of need or job status. All inmates who do not have a high school diploma or general equivalency diploma (GED) are enrolled in the academic program. Once the counselor has completed the classification assessment, the unit manager reviews and approves the recommendation. An inmate may request a work or program assignment by submitting a work application to the program or job supervisor. The supervisor reviews the application and recommends approval or denial. All recommendations are reviewed by the classification committee, which consists of a counselor supervisor and staff representing security and inmate programs. The committee makes the final decision and informs the inmate of his work and housing assignment. Inmates are not permitted to refuse any program, work, or education assignment (except substance abuse treatment, which is voluntary). Such a refusal is subject to a major disciplinary action. In such situations, the inmate is placed in an unassigned status with minimal privileges. If the preferred work or program assignment is not available, the inmate’s name is placed on a waiting list and he remains in orientation (or his current assignment) until the desired assignment becomes available. In sum, the Connecticut model is quite straightforward. The overall goal is to place the inmate in the most appropriate work or program assignment. The highest priority is for the inmate to obtain a GED before any institutional or vocational assign15 Exhibit 4. MacDougall Correctional Institution New Admission Classification Form Inmate Name Number: Estimated Release Date: Maximum Release Date: Parole Eligibility: 85% 50% Level Reduction Review Date: (MM DD YY) (Reviewed for accuracy, circle) Yes /No Detainer Score: High Security Status: (Circle) Yes/No (Next Hearing Date): Escape Risk Score: Education Level: School Referral: Yes/No DOB: Next Regular Review Date: (MM DD YY) Data Inmate Orientation Completed: Recommended Unit Placement: M Inmate Requested Placement: L I DNA: (Circle) Yes/No Police Report in File: Yes/No (If No, Data Requested) Comments: Completed by: cc: 16 Master File - (White) Satellite File - (Canary) Inmate's - (Pink) J Date Completed: Description of the States’ Internal Classification Initiatives ment. Housing units are organized around the program and work assignments, and inmates are not classified according to psychological categories or “risk” levels (other than the external classification rating). Because of the simplicity of this system, the amount of staff training required is minimal. A followup visit to the MacDougall Correctional Institution was conducted in February 2000 to assess the current status of the internal classification system. By 1995, the system was fully implemented, and it appears to work well for the facility. A written manual documenting policies and procedures has been developed (MacDougall Correctional Institution, 2000). The system has been modified only minimally since its implementation. The only substantive changes in the system were that vocational programs were added and the substance abuse treatment program no longer functions as a separate housing unit. Participation in the substance abuse program is voluntary and is scheduled around work, educational, or vocational programming. All job and housing assignments are stored on CT DOC’s automated information system. According to staff, the system’s strengths are that: It is simple to understand and apply. Housing units are organized around the program and work assignments, and inmates are not classified according to psychological categories or “risk” levels. Because of the simplicity of this system, the amount of staff Disciplinary incidents are minimized. There is a high rate of successful completion of programs. training required is minimal. Inmates demonstrate high self-esteem and satisfaction with work and program assignments. It enhances security within the facility. Unit and case managers are well aware of the inmates assigned to their areas. Staff also indicated that the system also provides for accountability by both staff and inmates. A review of the number of assaults on staff and inmates indicated that assaults on staff have dropped from 18 in 1996 to 4 in 1999. Assaults among inmates have remained relatively stable: 46 were committed in 1996, 67 in 1997, 46 in 1998, and 67 in 1999. Even though 95 percent of the inmate population is classified as “severe” risk, these data suggest that the internal classification system provides for a safe institutional environment. Colorado Department of Corrections As of September 1, 1995, the Colorado Department of Corrections (CO DOC) housed approximately 11,540 inmates; by December 1999, the population had grown to 15,372 inmates distributed across 24 facilities. An examination of the 1995 prison population indicated that the majority of the population was white (44.7 percent). The largest minority population was Hispanic (26.3 percent), and blacks represented 25 percent of the inmate population. The modal age group was 17 Chapter 2 30 to 39 years, with a mean age of 33 years. Analysis of the CO DOC population by type of offense indicates that 47.2 percent were committed for offenses against persons, 23.2 percent for property offenses, 11.9 percent for drug-related offenses, and 17.6 percent for offenses against public order and other crimes. The average sentence was 12 years. Women make up approximately 7 percent of the prison population (Colorado Department of Corrections, 1999). CO DOC identified the Limon Correctional Facility, located 75 miles east of Denver, as the pilot facility for the internal classification system initiative. Description of Pilot Site Limon Correctional Facility is a modern 953-bed, close-security facility that was constructed in 1991 for male inmates. It is rated as a Level III facility on a scale of one to five. There are positions for 348 employees, 182 of which are corrections officers (Soares, 1999). The average daily population at Limon has remained stable. In 1995, the average daily population (ADP) was 963, and on December 31, 1999, the 1-day count was 939 inmates. The main housing unit consists of 6 living units with 120 cells; each living unit is subdivided into 3 pods. There is also a 28-cell segregation unit. Approximately 25 percent of the cells are double-bunked, resulting in a total capacity of 953. All cells are electronically controlled from an enclosed command center in each living unit. Each pod can be isolated from the other two pods within the living unit, thus providing a great deal of control over inmate movement. The prison is surrounded by a double fence with electronic sensors and razor ribbon. There are three guard towers and a 24-hour perimeter patrol. In sum, Limon is very secure. Like the MacDougall facility in Connecticut, Limon was designed to hold highcustody male inmates sentenced to long prison terms. About one-third of the inmates are transferred directly from the CO DOC diagnostic center; the remainder are transferred from other general confinement prisons. Limon has a wide variety of academic and vocational educational programs, as well as counseling programs, jobs, and athletic facilities. There are program and work slots for approximately 90 percent of the inmates. A comparison between Limon’s population and that of the CO DOC system as a whole indicated that in 1995, during the development process for the internal classification system, Limon had a larger proportion of close-custody inmates than the system as a whole (55.0 percent versus 17.3 percent, respectively) (Colorado Department of Corrections, 1995). The Limon inmates differed from the total CO DOC inmate population on the basis of three key risk factors: offense severity, escape history, and severity of prior convictions. Seventy percent of the Limon population was incarcerated for a highest severity offense, and 15 percent had a history of escape from a Level II or III facility. In contrast, 51.7 percent of the total CO DOC population was incarcerated for a highest severity offense, and only 9 percent had a history of escape from a Level II or III facility. 18 Description of the States’ Internal Classification Initiatives The Limon population had some significant needs to be addressed. About 30 percent of the inmates did not have a high school diploma or GED, 16 percent had significant medical needs (Level 3 or Level 4), 77 percent had mental health needs, 35 percent required sex offender treatment (20 percent had Level 5 sex offender needs), and 95 percent had substance abuse treatment needs (Colorado Department of Corrections, 2000). A review of the most recent criminal history data for the Limon population indicated that the population has changed somewhat during the past 5 years. As of March 2000, a majority (64.4 percent) of the inmates were classified as close custody. Yet the average sentence length increased only slightly, from 26.5 years to 28.0 years. There appeared to be a shift between 1995 and 2000 in the crimes for which the Limon inmates were incarcerated: Offenses against persons dropped from 66.9 to 61.2 percent, and substance abuse related crimes decreased from 4.8 to 2.8 percent. Although the number of property offenders dropped from 18.5 to 13.1 percent, the proportion of offenders convicted of offenses against public order and other crimes increased sharply, from 4.5 to 22.8 percent (Colorado Department of Corrections, 2000: 3). Colorado Classification Process In 1983, CO DOC implemented an objective classification system modeled after the NIC system. Four studies of the Colorado system have been completed, each prompting significant changes. The most recent, completed by Austin et al. (1996), found that although the system is fairly well structured, it was both over- and underclassifying the inmate population. Based upon these recommendations, the system was revised and the modifications were fully implemented by 1998. The classification system was designed to assess objectively an inmate’s security, custody, and treatment needs. The Office of Offender Services within the Central Office is responsible for the development, implementation, training, oversight, and management of the external classification function within the department. All inmates (male and female) are admitted to CO DOC’s main reception center, where they undergo a fairly extensive and detailed initial classification process. The initial classification assessment is based on 13 scoring items that are grouped into 11 categories. The first six items—history of institutional violence, severity of current offense, number of current convictions, offenses resulting in death, severity of prior convictions, and escape history—constitute Part A of the form. These items are tallied to determine whether the inmate should be automatically placed in close custody without considering the Part B items, which include alcohol or drug abuse, current or pending detainer, number of prior felony convictions, stability factors, and time to parole eligibility date. The scores from the Part B items are added to the Part A score to determine the inmate’s final custody level. The system provides for both mandatory and discretionary override considerations. In addition to the custody risk assessment, the classification process evaluates the inmate’s need for medical, mental health, and sex offender treatment. These scores have a direct bearing on the type of facility to which the inmate can be transferred. 19 Chapter 2 The reclassification process parallels the initial classification process, but has only 6 items in both Parts A and B, for a total of 12 risk factors. Part A factors include history of institutional violence, recency of institutional violence, severity of current offense, number of current convictions, offenses resulting in the death of one or more victims, and severity of prior convictions. As with the initial classification, these items are tallied to determine whether the inmate should be automatically placed in close custody. Part B items include detainer or pending charges, escape history, number and type of disciplinary reports, total number of disciplinary reports, and time to parole eligibility date. Colorado’s Internal Classification Initiative Colorado was not interested in developing a risk-based system, but rather CO DOC’s internal classification initiative went a step beyond Connecticut’s internal classification system by automating the housing and program criteria. Colorado’s interests were similar to Connecticut’s in that Colorado was not interested in developing a risk-based system, but rather sought more structured information to ensure that inmates were housed and assigned to programs according to their needs. The specific goals for the Colorado internal classification effort were to— sought more struc- Improve the quality and accessibility of information on which internal classification decisions (especially housing decisions) were based. tured information to Reduce the level of inmate idleness. ensure that inmates were housed and assigned to programs according to their needs. Reduce the number of inmates assigned to inappropriate housing, work, or rehabilitative programs. The primary impetus for the development of the Colorado internal system was the difficulty of accessing and updating information required for making informed housing decisions. Since accessing this information created major inefficiencies and delays in assigning an inmate to a housing unit, job slot, or programs, it was often ignored. Two particularly sensitive areas were separation needs (codefendants, family members, staff conflicts, etc.) and security threat group membership. Relevant information either was not documented or was scattered throughout the CO information system. Information on gang membership, for example, was stored by the automated information system but was accessible only to the gang coordinator. Staff were frequently faced with inmate problems that could have been avoided had the proper information been accessible. Thus, a major goal of this initiative was to improve the reliability and validity of the information on which internal classifications decisions were based. The intent was to create a standardized inmate profile within the information system. CO DOC had previously completed an assessment of the data required for internal classification decisions. Staff had indicated that the primary inmate-related decisions were housing unit, treatment, and job or work assignments. The staff identified the inmate’s medical, psychiatric, and separation needs, as well as security threat group membership, as important considerations for these decisions. 20 Description of the States’ Internal Classification Initiatives Along with improving the validity and reliability of the inmate profile within the information system, the major objectives of the Colorado internal classification initiative included— Encouraging good conduct among inmates; Increasing inmate productivity on work assignments; Establishing a realistic program sequence for long-term inmates; and Increasing the program completion rate. The main strategy for accomplishing these internal classification objectives was to develop a Master Program Scheduling (MPS) system. MPS is an automated system that includes specific criteria for program and housing eligibility at each step of confinement. MPS compiles and tracks information regarding the inmate’s profile; institutional housing, job, and program vacancies; and the facility’s eligibility criteria for internal classification assignments. Admissions Data Summary (ADS). This screen provides general inmate-related information. As part of this project, data fields were added to the screen based on the DOC staff survey regarding information required for internal classification decisions. (Exhibit 5 is a sample ADS.) Access grid. Varying levels of access to the ADS were established so that users could retrieve or enter information within specific data fields. Since access to correctional information is always a sensitive issue, staff information needs and suitability for data entry were identified. Procedures for coordination of information between the central office and facility staff were defined. Data-entry responsibilities. A new policy established lines of responsibility for updating information. Intake and orientation case managers were made responsible for interviewing each new inmate and reviewing and updating the inmate’s ADS. Case managers within the general population housing units were made responsible for updating specified fields. An intelligence coordinator was made responsible for compiling and maintaining data on security threat groups. Data dictionary. A standardized definition was developed for each data element. This definition includes the source and reliability of the information. One major challenge to the system was to regulate the flow of information within MPS. As the system became more accessible and reliable, concerns emerged that the collection and entry of data would become so time consuming that staff would have time for little else. Internal custody information had to be carefully limited and structured or chaos would result. To address this problem, CO DOC developed procedures for structuring MPS custody information. 21 Exhibit 5. State of Colorado Department of Corrections Admission Data Summary Date of Report Date of Interview Name DOC# FBI# AKA Height Weight Sex DOB Age Complexion Disruptive Group Orientation Marks and Scars PHOTO SID # / / / / / / SSN Eyes Birthplace Ethnicity Hair Citizenship SENTENCE DATA Defendant Admission Date County Sentence Date Court / / / / Parole Eligibility Date Effective Date / / / / Sentence Discharge Pre-Sentence Confinement District Attorney Defense Attorney Co-Defendants Offense COURT DATA Class Docket Number Offense Date Sentence (Yr, Mo, Day) Comments Vocational Medical/Dental Psychological Work Leisure Time Alcohol/Drug Pre-Release Warrants Sexual Violence Conduct Pending Charges HISTORICAL CRIMINAL DATA Prior Colorado DOC# Arrests Misdemeanor Arrests Paroles Parole Absconds Parole Revocation Escapes Through Security Walk Away Type Year Prior Probation Probation Revocation Community Placement Last Escape Date Offense Revocation Last Escape Location Disposition PERSONAL DATA Emergency Notification Phone Address Place of Residence Marital Status Military (Branch) Dates Performance Level Claimed Educ. Level Dependents Discharge Apticom Viet.Vet Math Religion Language THIS SENTENCE DATA DOES NOT INCLUDE ANY CHANGES MADE AFTER DATES OF REPORT. CONTACT LOCAL RECORDS OFFICER FOR CURRENT INFORMATION. 22 / / Description of the States’ Internal Classification Initiatives Once MPS was established, internal classification decisions could be tracked through the system. The process begins when inmates first enter the correctional system at the Reception and Diagnostic Center in Denver and continues throughout their tenure at the facility. One priority element for the internal classification process among the intake staff was the availability of information about inmates prior to their arrival at the facility. Exhibit 6 is a flowchart representing the MPS system. At Limon, inmates are temporarily housed in the Intake/Receiving area for orientation, observation, and assessment. To meet prison needs, all newly admitted inmates are assigned to kitchen or general maintenance details. The most desired work assignment at Limon is with the industrial sewing plant because it has the highest pay rate. Yet facility policy dictates that all inmates must spend some time Exhibit 6. Colorado Internal Classification Master Program Scheduling System Master program Offender assessment Job positions Work Education Vocation Mental health Earned time performance Vacancies Inmate attendance Offender assignment Mental health Medical Housing Job Substance abuse Offender Services monitors system via online reports Offender schedule Offender/ program wait list Offender skills and needs Facility schedule Staff schedule 23 Chapter 2 in the less attractive and less lucrative kitchen assignments; inmates are eligible for maintenance assignments before becoming eligible for the sewing assignment. In addition to making basic job assignments, MPS schedules inmates for rehabilitative and treatment programs according to their prioritized need for treatment or other programs, time left to serve, custody rating, special housing flags, the program’s availability, and the program’s eligibility criteria. When inmate-based data are entered into MPS, the optimal time for the inmate to participate in the recommended programs is automatically determined by MPS using the inmate’s profile, program eligibility criteria, and program availability. If an inmate is eligible but cannot be admitted because of limits to program capacity, he is placed on the waiting list. The overall intent is to place inmates in the most appropriate treatment program available. One key issue resolved by the steering committee was the timing for program participation for long-term inmates (i.e., whether participation in academic, vocational training, or substance abuse treatment should occur at the beginning or end of the inmate’s prison term). Staff decided that programs should be required at the beginning of an inmate’s prison term and should be a prerequisite for the desirable industries assignments. These policies were incorporated into MPS programming. Despite its complexity, MPS has undergone few changes since its implementation in 1995. Staff attributed this durability to the thorough planning A followup visit to Limon was conducted in February 2000 to assess the current status of the internal classification system. As of June 1995, MPS had been fully implemented, and it continues to work well for the facility today. Written policies and procedures are available that document the process, responsibilities, and data compiled by MPS (Limon Correctional Facility, 1999). Upon inmate request, the system was incorporated in the Marquiz settlement agreement, and thus has been integrated into the daily decisionmaking processes.6 Despite its complexity, MPS has undergone few changes since its implementation in 1995. Staff attributed this durability to the thorough planning process and the involvement of multiple stakeholders. The steering committee, for example, included representatives of all major operational areas in the facility and central office. Surprisingly, MPS was implemented with only minimal costs. The only startup costs were physical changes to the intake and reception unit and one full-time employee position for the internal classification specialist. Inmate profile, housing, job, program, and special needs data were already stored in the Colorado information system; thus, the primary task of this initiative was to organize the information to make it more accessible and useful. According to staff, the system has led to— process and the A decrease in serious assaults within the facility, both inmate-on-staff and inmate-on-inmate; involvement of mul- Suppression of drugs and contraband; tiple stakeholders. 24 Improvement in the quality and reliability of security group intelligence and separations; Description of the States’ Internal Classification Initiatives Appropriate work and program assignments; and Better control of inmate movement. As shown in Exhibit 7, assaults on staff have remained low despite the influx of higher risk inmates since 1995. This, together with the staff’s confidence in and satisfaction with MPS, suggests that the internal classification system creates a safe environment and provides staff with valid and reliable data. Exhibit 7. Assaults at Limon Correctional Facility, Fiscal Years 1995–96 to 1998–99 Fiscal Year Inmate-on-Inmate Assaults Inmate-on-Staff Assaults 1995–96 10 16 1996–97 7 14 1997–98 17 19 1998–99 14 14 Note: Colorado’s fiscal year begins on July 1st and ends on June 30 of the following calendar year. Staff also identified elements that were essential to the implementation of MPS. Since MPS created more work (e.g., data entry and retrieval), full participation and commitment from central office and facility staff were critical. All staff readily agreed that the internal classification system had to be a high priority for the facility. Another key ingredient for proper implementation was experienced, motivated staff. Attempts to implement a similar internal classification process at another Colorado facility have been slowed because staff lacked the proper experience and training with internal classification systems. Another key ingredient for proper implementation was experienced, motivated staff. Washington State Department of Corrections As of June 30, 1995, the Washington Department of Corrections (WA DOC) housed approximately 11,440 inmates; by July 1999, the population had grown to 14,307 inmates in 13 facilities and 18 work-release and prerelease centers (Washington Department of Corrections, 2000). Examination of the prison population at the onset of the internal classification initiative revealed that a majority of the 1995 population was white (71.6 percent). The largest minority population was black (22.5 percent); Native Americans and Asians were small minorities (2.9 and 2.2 percent, respectively). Nearly three-quarters of the population did not have a high school education. The modal age group was 30 to 39 years, with a mean age of 30.6 25 Chapter 2 years. Analysis of the WA DOC population by type of offense indicated that 61 percent were committed for offenses against persons, 14 percent for property offenses, and 25 percent for drug-related offenses. The average sentence was 10 years. Women made up approximately 7 percent of the prison population (Washington Department of Corrections, 2000). WA DOC identified the Airway Heights Correctional Center (AHCC), 4 miles west of Spokane, as the pilot-test facility for the internal classification system initiative. AHCC was chosen because it was a new institution in which staff were open to suggestions and not committed to “old ways.” The superintendent was new and known to be innovative. The population was considered unique in that the institution was used to house inmates with a minimum 3 (MI 3) custody rating. These offenders scored as minimum custody, but were not eligible for placement within a minimumsecurity facility because they were not within 3 years of release, as required by WA DOC policy; had medical requirements precluding minimum assignment; or were too high a security risk to be assigned to a camp. By WA DOC policy, MI 3 inmates are housed in medium-security facilities with a reduced staffing pattern. Description of Pilot Site AHCC features a 400-bed, minimum-security component that was opened in 1992 and a 1,536-bed medium-security facility that opened in 1994. In 1994 Airway employed 525 staff, 187 of which were corrections officers (Alexander et al., 1997: 40–43). The medium-security facility consists of six 256-bed units and a 20-bed infirmary. Three of the units have “wet cells,” while the other three have “dry cells.” Each unit is shaped like an “X” with four two-tiered wings. A central control station at the intersection of the “X” regulates and monitors inmate movement. The perimeter consists of a double fence with razor ribbon, a tower, and a perimeter patrol. About half of the inmates come directly from the DOC reception center, and the other half are transferred from other confinement facilities. Several academic, vocational, and education programs, as well as a wide range of jobs and athletic facilities, are available to the inmates. A comparison of the AHCC population with the overall WA DOC population indicated that in 1995, during the development process, AHCC had a larger proportion of inmates classified as minimum risk than the state’s total prison population (92 versus 56 percent, respectively) (Washington Department of Corrections, 1995). In 1995, need-factor data were not available for the WA DOC population, but analyses of criminal history data suggested that AHCC inmates had the same average sentence length and time left to serve as the total prison population. Although 92 percent of the AHCC population was classified as minimum custody, 64 percent of the population was incarcerated for crimes against persons; 61 percent of the total WA DOC population was incarcerated for such crimes. As suggested by their MI 3 status, the AHCC inmates who had been incarcerated for serious offenses had adjusted well to institutional life (92 percent of them were classified as low risk), 26 Description of the States’ Internal Classification Initiatives yet had at least 3 years remaining to serve (75 percent). The only need-related data available indicated that about 57 percent of the AHCC inmates did not have a high school diploma or GED. This suggested that the educational attainment of the AHCC population was higher than that of the WA DOC population as a whole. Washington State’s Classification Process WA DOC implemented its current objective (external) classification system in 1989. The classification system was designed as a risk management tool to place offenders in the least restrictive custody designation possible while providing for the safety of the public, community, staff, other offenders, institution guests and visitors, and the orderly operation of the institution (Washington Department of Corrections, 1996). The Office of Correctional Operations, Classification, and Treatment is responsible for the development, implementation, training, oversight, and management of the department’s external classification function. The classification system is designed to encourage offender participation in work, education, treatment, and vocational programs in a manner that results in movement to less restrictive custody levels. Negative behavior is discouraged by ensuring that offenders face consequences for infractions, detainers, escapes, and refusal to participate in programs. The system is designed to help offenders understand how conduct and program efforts affect their custody designation and subsequent placement in correctional facilities. The WA DOC classification has four components: assignment of custody designation, assignment of facility designation, assessment of needs, and review of offender programming needs and case planning. The classification process begins with the reception center’s intake process, which determines the inmate’s custody designation and facility placement. The initial classification assessment is based on five scoring items: severity of current offense, history of violence (institutional and community), detainers, escape history, and age. These items are tallied to determine the inmate’s custody level. The system provides for mandatory overrides based upon the offender’s crime, detainer, and sentence. Discretionary overrides are permitted based on the offender’s behavior, mental health, medical, dental, or program needs. Institutional security concerns may also override the scored custody level. Multiple levels of review are required for all overrides. A facility placement assignment is also made during this process. The placement recommendation addresses custody, security, program, medical, mental health, and dental needs; case management and planning; and other specific offender or institutional needs. Offenders are placed in the least restrictive environment consistent with their initial custody designation. The reclassification process is similar to the initial classification process. Regularly scheduled reviews are conducted according to the time remaining for the inmate to serve. For example, inmates with less than 5 years to serve are reviewed semiannually. Inmates with more than 5 years to serve are reviewed annually. Reclassification is based on the inmate’s current custody designation, institutional infractions during 27 Chapter 2 the past 24 months, program participation, detainers, and escape history. As with the initial classification, these items are tallied to create the custody review score. Again, mandatory and discretionary override factors are considered and reviewed. As previously noted, program needs and case planning must be documented and addressed as part of the initial and custody review processes. Individual goals and steps for addressing the offenders’ needs must be coordinated with their eligibility for placement in a minimum-security facility. Housing, program, and work assignments are based upon individualized case plans. Washington State’s Internal Classification Initiative WA DOC began developing new approaches to internal classification in the late 1980s. The Case Management Classification (CMC) system described earlier was developed and implemented at the Clallam Bay Correction Center. In December 1993, WA DOC expanded the use of CMC as a tool for identifying offenders’ needs and developing individualized case management plans. WA DOC sought an internal classification model that would WA DOC’s major goal for this initiative was to pilot the Adult Internal Management System (AIMS) in AHCC’s K–Unit. As previously described, AHCC’s population is unique in that it is a minimum-custody population that is homogenous with respect to several key criminal history and demographic factors. WA DOC wanted to explore several issues with this population: Use of case management as a classification tool to better manage inmates according to risk of escape and disruptive behaviors; enable it to both classify and manage inmates. The idea Improvement of institutional strategies for managing and controlling offenders; Development of programming responsive to offender needs; was to have an inter- Development of better ties for offender transition to the community; and nal classification Development of indicators for comparing offenders across facilities. system that indicated not only the appropriate housing or programming for the inmate, but strategies for managing the inmate as well. 28 In short, WA DOC sought an internal classification model that would enable it to both classify and manage inmates. The idea was to have an internal classification system that indicated not only the appropriate housing or programming for the inmate, but strategies for managing the inmate as well. To build on its unit management team concept, WA DOC was particularly interested in a model that would require staff involvement at all levels. After reviewing the current literature, AIMS was identified as a unit management strategy that could be implemented in combination with the CMC approach to manage individual behavior. AIMS was selected for the following reasons: It involves both correctional officers and classification staff in the classification and management of offenders. Description of the States’ Internal Classification Initiatives The checklists provide for the collection of criminal history and personality data useful for comparative analyses across facilities. It had been validated as strategy for reducing institutional violence among male inmates. Washington state was particularly interested in case management as a tool for predicting offender behavior and matching risk factors to specific programming to enable the department to better manage offenders in the institution and to better prepare for their transition into the community. WA DOC developed a three-phase planning and implementation strategy for this internal classification initiative. Phase I included pilot-testing AIMS at AHCC to compile a profile of the offender population. This phase required the identification of control and study groups at AHCC and other comparable WA DOC facilities, which were further subdivided into two groups according to the inmates’ time to release. Inmates with less than 3 years to serve were placed in the “short term” group, and inmates with more than 3 years to serve were considered “long term” inmates. Staff were trained on the use of AIMS and data were collected on the characteristics and programming needs of offenders. The groups were compared in terms of rates of infractions and grievances filed by the inmates. Phase II was envisioned as an analysis and planning period to determine the required resources and develop the operational structures to support the case management and community transition model. In Phase III, the case planning and community transition services would be fully implemented. During the 2-month pilot test, data were collected on infractions, classification movement, and complaints or grievances. The major indicator of offender behavior was infractions. In the total AHCC population, the rates of infractions by AIMS category (Alpha, Kappa, and Sigma) conformed to expectations, in that the Alphas had the highest rate of infractions. As shown in Exhibit 8, comparison of the infraction rates for the various groups suggested that the study sample behaved better than expected and the AHCC control groups behaved worse. Although this pilot test was not designed to yield definitive findings, the results were promising. The pilot test was also intended to investigate the usefulness of the AIMS profile data. One key success of this pilot test was the collection of inmate data for program planning. Although a formal study of staff satisfaction with the use of AIMS was not conducted, the issue was discussed at AHCC’s annual inservice training. Virtually all staff agreed that AIMS would greatly enhance the operation of AHCC and would provide key information for case management. Staff also believed that AIMS gave supervisors insight into offender management strategies. Most staff believed that full implementation of AIMS would benefit the facility, but several staff members expressed concerns that housing Alpha offenders together would put undue stress on officers supervising those units. Staff also expressed concern that the Beta units would be called the “rape units” because of the lifestyle and Most staff believed that full implementation of AIMS would benefit the facility, but several staff members expressed concerns that housing Alpha offenders together would put undue stress on officers supervising those units. 29 Chapter 2 Exhibit 8. Washington State AIMS Pilot Test Results: Comparison of Disciplinary Infractions AIMS Test Group AHCC Control Groups AHCC Other Facility LongTerm Group ShortTerm Group LongTerm Group ShortTerm Group LongTerm Group ShortTerm Group Percentage with one or more infractions during preceding year 21 23 37 41 23 29 Percentage with one or more infractions during pilot test 6 4 15 14 5 12 Disciplinary Adjustment behavior of inmates classified as Betas.7 Staff endorsed further exploration of AIMS at AHCC. AIMS was perceived as more accurate than CMC because it provided a profile of the inmate while incarcerated. More important, AIMS does not rely on offender statements, but is based on both the institutional record and the officer’s direct observation of the offender’s behavior. Despite its preliminary success, the internal classification initiative based on AIMS was not implemented because the offender groups identified via AIMS duplicated the offender typology developed under CMC. 30 Despite its preliminary success, the internal classification initiative based on AIMS was not implemented because the offender groups identified via AIMS duplicated the offender typology developed under CMC. Completion of AIMS checklists essentially doubled the staff’s workload without providing additional information. Therefore, a Phase III followup site visit was not conducted. Ultimately, CMC was preferred to AIMS because it provided more direct input into the development of case management plans. AHCC staff also observed that housing offenders according to their “personality” type created greater stress and disruption, particularly within the Alpha units, whereas separating high-risk offenders tended to dilute tensions across all housing units. To date, housing, case management, and program decisions at AHCC are based on CMC. In July 1999, however, WA DOC adopted the LSI (Level of Service Inventory) as a tool for identifying the criminogenic needs or risk factors associated with offenders’ likelihood of recidivism (e.g., substance abuse, peer relationships, education, etc.) and to develop individualized case plans for institutional programs and transition to the community. LSI will also be used to determine community supervision strategies and to allocate resources among offenders. Description of the States’ Internal Classification Initiatives Oregon Department of Corrections As of October 1998, the Oregon Department of Corrections (OR DOC) housed approximately 8,363 inmates in 13 facilities. Approximately 160 inmates were housed in out-of-state facilities. Because population forecasts predict that the prison population will more than double by 2006, the state embarked on an ambitious prison construction program to build seven new prisons between 1997 and 2005. An examination of the prison population by type of offense indicated that 41 percent were convicted of offenses against persons, 29 percent for property offenses, and 31 percent for status offenses. Women comprise 11 percent of the prison population (Oregon Department of Corrections, 1998). OR DOC identified two facilities for its internal classification system initiative. Initially, OR DOC identified Oregon State Correctional Institution (OSCI) as the pilot-test facility. When NIC expressed interest in developing and pilot-testing an internal classification process for a women’s correctional facility, OR DOC readily agreed to expand its initiative to include the Oregon Women’s Correctional Center (OWCC). Both facilities are located in Salem, Oregon, about 3 miles apart. Because of the institutions’ close proximity to each other and to the department’s headquarters, a single project steering committee was organized with subcommittees to address the specific issues for the two facilities. Description of Pilot Sites Oregon State Correctional Institution. The Oregon State Correctional Institution (OSCI) is a medium-security, 840-bed facility that houses males from all counties in Oregon serving sentences for felony convictions. The average daily population during 1997 was 860 inmates. The facility was established by action of the 1955 Oregon state legislature and became fully operational on June 1, 1959. Traditionally, this institution has housed younger inmates, including those remanded to adult prison from juvenile facilities. Since the primary purpose of this facility is long-term confinement, intake and reception activities are limited; most inmates are transferred from other OR DOC facilities. OSCI also serves as the OR DOC facility for inmates with medical or physical problems who do not need to be hospitalized or confined in an infirmary. Additionally, the facility houses OR DOC’s younger inmates, including those remanded to adult prison from juvenile facilities. Although it is a rare occurrence, inmates are sometimes released to the community from OSCI. The types of housing within OSCI include 179 dormitory beds, 246 double cells, 124 segregation cells (88 doubles and 36 singles), and a 64-bed special dormitory for those with medical or physical problems. Oregon Women’s Correctional Center. Oregon Women’s Correctional Center (OWCC), the state’s only medium-security, adult women’s prison and the only fullservice OR DOC prison, opened in January 1965 with a design capacity of 76 beds. Remodeling and double-bunking have expanded the capacity to 190 beds in cells 31 Chapter 2 and dormitories. It maintains a close supervision unit (North Hall), a disciplinary segregation unit, and maximum-custody housing. It also operates as an intake and release center for women of all custody and program levels. Inmate programs include transitional substance abuse treatment, mental health treatment, selfhelp classes, educational and professional technical programs, and an assortment of innovative industrial programs. OWCC has multiple work opportunities for all eligible inmates. Oregon’s Classification Process OR DOC originally designed and validated its objective external classification system in 1989.8 The system was revalidated and amended in June 1993 and again in October 1994. During the fall of 1997, OR DOC undertook another revalidation of the external classification system to update its risk factors. At that time, the department specifically examined the validity of the system for female inmates. Based on this validation effort, minor revisions to adapt the instrument for use with female inmates took effect on May 1, 1998. The purpose of the external classification system is to identify and assign an appropriate custody level to each inmate committed or transferred to the Department of Corrections and housed in an OR DOC-operated or contracted facility.9 The classification and transfer unit of OR DOC’s central office is responsible for the development, implementation, training, oversight, and management of the department’s external classification function. OR DOC’s external classification instrument incorporates numerically weighted public and institutional risk criteria and a scoring matrix to determine the appropriate custody level. The public risk criteria include crime severity (current offense), extent of violence, use of weapons, history of violence, escape history, time left to serve, and felony detainers. The institutional risk criteria include frequency of institutional misconduct, severity of institutional misconduct, primary program compliance, gang affiliation, substance abuse, and age. Based on these criteria, inmates are classified into one of four custody levels: maximum, close, medium, or minimum. Data to score the risk criteria are drawn from OR DOC’s computerized information system, the presentence investigation report, rap sheets, health screening, sentencing or commitment papers, and an interview with the inmate. Results from the battery of tests completed at the inmate’s admission to OR DOC are also available. Overrides of the proposed custody level based on the classification score are approved in approximately 12 percent of the cases. At the OR DOC intake center, overrides require two levels of approvals: from the unit manager and from the classification and transfer unit. Classification reviews must be approved by the program supervisor, the unit manager, the superintendent, and the classification and transfer unit. The most common reasons for increasing the custody level include psychological problems, time remaining to serve, and special management concerns. The most frequently cited reasons for reducing a custody level are to allow inmates access to prerelease programs, to allow them to participate in a work or treatment program, or to allow them to remain in minimum custody because they are not an apparent risk to public or institutional safety. 32 Description of the States’ Internal Classification Initiatives Oregon’s Internal Classification Initiative Oregon decided to participate in NIC’s internal classification initiative for the following reasons: To alleviate problems related to biased housing assignments that had led to a lawsuit resulting in a financial award to inmates. To match inmates more effectively with jobs and programs. To improve institutional population management. To balance the operational needs of special program units with the needs and risks of the inmate population. To develop a prototype for managing inmate populations to minimize intrafacility transfers and interfacility transports. To reduce the severity and frequency of institutional misconduct. OR DOC envisioned an internal classification system that supported the inmates’ program and work priorities and tried to assign them cellmates who were compatible with respect to age, sentence, conduct history, and work or programs. The underlying assumption was that if cellmates were compatible, there would be less tension, fewer transfer requests, and fewer institutional infractions. The goal was to develop a system for collecting, prioritizing, and monitoring inmates’ program and work activities. These data would be used to make housing assignments that supported inmates’ individualized institutional case plans. The plan was to fully automate institutional case plans, work assignments, and program activities so the computer could identify optimal housing, work, and program assignments for each inmate. Although the two facilities’ current internal classification systems differed, their goals were similar. OWCC differed slightly from OSCI in that it wanted to integrate the external and internal classification processes into one sequential decisionmaking process. OR DOC envisioned an internal classification system that supported the inmates’ program and work priorities and tried to assign them cellmates who were compatible with respect to age, Potential barriers to the completion of this project were overcrowding, limited staffing resources, and setbacks in the development and implementation of the incentive level system. Because overcrowding restricted flexibility in housing inmates, the facilities’ ability to pilot-test the new system was limited. Traditionally, housing assignments depended on the first available bed, rather than the optimal slot identified by the new internal classification system. Given the already high demands on the counselors and program staff, OWCC staff were concerned about their ability to complete the project tasks within the period allotted by the NIC grant. This internal classification initiative relied heavily upon the development and implementation of the incentive level system. Thus, delays or barriers encountered by this process would likely hamper the internal classification initiative. sentence, conduct history, and work or programs. 33 Chapter 2 One primary task was to identify the issues and concerns associated with the development of the internal classification system. Their discussions centered on the following questions: Can any internal classification system work effectively in a department suffering from overcrowding? Does a system like AIMS or PMS that separates “aggressors” from “victims” really improve the security and resource utilization of the facilities? Where does housing “compatibility” fit into the internal classification system? What resources and training are necessary to adopt an internal classification system? Could an internal classification system make it harder to fill minimum-custody beds? Where does facility architecture figure into the internal classification system? The steering committee summarized their ideal system as one that was— Dynamic—the system should reflect changes in the inmate’s performance. Consolidated—the system should link existing fragments of information into a concise format. Efficient—the system should improve safety, security, and the utilization of resources. Valid and Defensible—the system should place inmates appropriately and withstand court review. Individualized—the system should provide flexibility for “special management” cases. Simple—the system should not be overly complex and must be user friendly for all staff. Compatible—the system should complement the current external classification instruments. Flexible—the system should operate effectively in a variety of facilities. Useful—the system should provide input to work and program assignments. Improved—the system should incorporate the strengths of the current housing process while correcting its deficiencies. 34 Description of the States’ Internal Classification Initiatives To achieve these objectives, the steering committee concluded that a customized system, rather than a modification of an existing system (e.g., AIMS or PMC), was preferred. Although a single steering committee directed this initiative, the differences in the design of the respective facilities and the differing risks and needs posed by male and female inmate populations led the two facilities to undertake separate design processes. OWCC Internal Classification Design Process. A preliminary step in developing the prototype system at OWCC was a survey of the inmates regarding the factors that they perceived to be important for cellmate “compatibility.” This survey indicated that the most critical factors for women were age, education, participation in treatment, misconduct record, time to serve, number of visitors, criminal history, and hobbies. Building on these data, several prototype instruments were developed and rejected because the large number of variables, categories within variables, and the possible combinations of variables and categories meant that there were too few inmates per profile to provide sufficient latitude in making housing decisions. This issue was resolved by reducing the number of classification criteria. The OWCC prototype instrument considered age, institutional misconduct, time remaining to serve, external custody level, incentive level,10 needs assessment, gang affiliation, relationships, and programs. These factors were cross-tabulated into grids that identified the inmate’s housing “type.” The most critical factor was incentive level. No deviation up or down was permitted; all cellmates were to have identical incentive levels. In the areas of “age/disciplinary reports” and “sentence/custody,” deviations up or down one score were assumed not to affect the roommates’ compatibility. A “treatment/program needs” factor was used to differentiate among inmates with identical scores. The assumption was that inmates with similar needs were more compatible. For the OWCC pilot test, all assignments during October 1998 that resulted in twoperson cell combinations were scored according to the prototype instrument. A total of 157 assignments were considered, 42 of which resulted in two-person combinations. The results indicated that only 10 percent of long-term or initial housing assignments were compatible for all factors on the prototype instrument. Based on these preliminary results, OWCC concluded that the system was feasible in that it was easily scored and compatible cellmates could be identified. Beginning in December 1998, the H Unit of Eastern Oregon Correctional Institution (EOCI) opened as a female housing unit. Because EOCI’s housing configuration and level of services were similar to those of OWCC, a full-scale pilot test of the prototype system was designed. For this pilot test, half of the inmates were assigned according to the prototype instrument and the remainder were assigned randomly to any available cell. The prototype internal classification system was implemented in December 1998 for the female population at EOCI’s H 35 Chapter 2 Although separate systems were originally developed for male and female inmates, with the automation of the internal classification system and its imple- Unit. In February 1999, the system was implemented at OWCC. Between February 1999 and March 2000, the instrument underwent the following modifications: Incentive level (IL) was removed as an element of the model because roommate compatibility and IL did not appear to be correlated. In addition, IL had not been automated and was subject to frequent changes. A grid of six categories for major acts of misconduct during the past year and sentence remaining was built into the instrument to better reflect the misconduct patterns of the female offenders. (See Exhibit 9.) Age by misconduct categories were replaced with more stringent categories for misconduct per time remaining to serve. Age alone appeared to be the most important of all housing criteria; age by misconduct categories produced “false” matches. mentation in March 2000, the same system is now used for males and females alike. Custody level was eliminated, as it was essentially duplicated by the other elements of the model. Using these modifications, the system has been automated so that the computer searches the database to identify the optimal matches among the cells available within the facility. Exhibit 10 shows sample output from the automated internal classification system. Eight potential matches were identified for inmate Hardyman. The optimal match is inmate Suguwara, because both the age and combined SR/DR factors are identical. Prior to making a final assignment, however, relationships and conflicts with other inmates (e.g., gang membership, codefendants, etc.), medical status, previous housing history, inmate incarceration and transition plans, interactions with staff; and public information are also considered. After considering these factors, inmate Kossa is determined to be a potential optional cellmate. Exhibit 9. Sentence Disciplinary Grid for Oregon Internal Classification System 36 Major Acts of Misconduct in Past Year (number of incidents) Sentence Remaining (in years) <1 1–2 2–3 3–5 5–10 10–20 20+ 0 1 1 3 11 12 19 20 1 2 2 3 11 13 19 21 2 2 4 4 12 13 20 21 3 6 6 8 15 16 23 24 4 7 7 8 15 16 23 25 5+ 8 9 9 16 17 24 26 Description of the States’ Internal Classification Initiatives Exhibit 10. Oregon Corrections Information Systems, Recommended Inmate Cell Mates Offender: 11763831 Hardyman, Patricia L. List Openings In: SRCI, Snake River Correction Institution Location: SRCI Cell Number: DSB18A Major RV–1 Year: 1 Max Incarc. Date: 10/30/04 PRD: 10/30/04 Sent. Remaining: 5 nd Gang: Crip Northside 72 DR/SR Score: 11 Birthdate: 12/10/77 Age Score: 2 Position to age score:_____ SRDR score:____ Enter option.....2=Assign bunk OPT ID NBR NAME CELL AGE SR DR COMB GANG ____ 11763831 Hardyman, Patricia DSBA 22 2 5 11 ____ 12310363 Dedel, Kelly 1E25B 21 2 4 11 ____ 12357722 Kossa, Susan 2C37A 22 2 4 11 2 11875757 Suguwara, Stephanie 2J37B 21 2 5 11 ____ 11929741 Ceridini, Ophelia 2J31A 22 2 7 13 ____ 12565117 Capik, Deborah 1A23B 29 2 6 9 ____ 8896693 Tribble, Cassandra 3B24A 30 3 5 9 ____ 12309472 Hale, Jacqueline 1A12B 31 3 4 11 Information Windows: R Relationships/Conflicts X Medical C Counselor chronos Crip Crip Hisp. H Housing History I IIP Inmate Incarceration Plan P Public Information The key findings from this first year Although separate systems were originally developed for male and female inmates, with the automation of the internal classification system and its implementation in March 2000, the same system is now used for males and females alike. Although it is still too early to assess the impact of the internal classification system on the male facilities, initial reports from the facilities have been positive. As of March 2000, the results from the use of the internal classification system at the female facilities were quite promising. Since the pilot tests in February 1999, the internal classification system was manually implemented and tracked at EOCI and OWCC. The key findings from this first year were reductions in the numbers of major acts of misconduct filed, inmates placed in the segregation unit, and housing assignment transfers. were reductions in the numbers of major acts of misconduct filed, inmates placed in the segregation unit, and housing assignment transfers. 37 Chapter 2 Use of the objective internal classification system (ICS) resulted in a 30-percent reduction in the number of institutional acts of misconduct among the female population since its implementation in February 1999. Before ICS, there were an average of 50.42 major misconduct reports issued per month. After ICS was established, the 12-month average was 34.67 per month. Before February 1999, between 10 and 15 percent of the total OWCC population was in the segregation housing unit. Since implementation of ICS, the percentage of OWCC inmates placed in segregation has remained below 7 percent. A longer term analysis of this population is required, however; inmates are transferred to segregation at OWCC from other facilities. The percentage of the EOCI population in segregation has remained low, approximately 5 percent of the population. Unfortunately, there are no baseline data with which to compare this figure. An analysis of the number of noninitial transfers to and from two-person cells at OWCC after implementation of ICS indicated a sharp drop. Between September 1998 and February 1999, there were an average of 138 cell transfers per month. With the implementation of ICS in February 1999, the average number of OWCC noninitial cell transfers per month has dropped to 87, a decrease of 37 percent. With this drop in cell transfers, there has been a 400-percent increase in the length of time OWCC inmates have remained with the current cellmate. EOCI noninitial transfers during the first 6 months after ICS was implemented dropped from 41 to 24 percent of the population per month. This rate is expected to continue to decrease because the initial process of filling H Unit is complete and, therefore, fewer transfers are required for the regular management of the dorm. The system’s built-in reliability is one of its greatest assets. These data suggest that ICS has achieved the goals set forth by the steering committee. The system appears to be flexible and easy to use in that “matches” are readily identified through the automated program, major disciplinary infractions have dropped significantly (by 30 percent), and the number of cell transfers has dropped dramatically. Another important outcome was that the number of OWCC inmates placed in disciplinary segregation dropped by 40 percent. It is important to note that this system is completely void of subjectivity; the data required to “match” the inmates are contained with the department’s information system. The system’s built-in reliability is one of its greatest assets. As of March 2000, automation and full implementation across the department has been accomplished. Research and evaluation of the system will continue and changes will be made as necessary. OSCI Internal Classification Design Process Although OWCC and OSCI took different tracks in designing and testing their respective internal classification systems, the results were very similar. The most critical issue addressed by the OSCI model was the number of inmate-initiated “convenience” moves. Between September 1997 and September 1998, there were 38 Description of the States’ Internal Classification Initiatives 9,400 in-house convenience moves among the OSCI population. OSCI identified five specific goals and objectives for its model: Reduce convenience moves and use assignment office staff more efficiently; Increase cell compatibility among inmates; Offer an incentive for all inmates to work and participate in programs; Enhance program compliance; and Reduce institutional misconduct. The specific factors identified by OSCI to be included in the model were (in order of priority): Inmate incentive level—cellmates must be housed within one incentive level. Age—cellmates must be within 5 years of age. Custody—cellmates must be within one custody level (up or down). Gang affiliation—if an inmate is a recognized gang member, he will not be housed with any other known gang member. The sophistication of OR DOC’s database made it possible to simulate the search for compatible inmates based on the number of matches of these criteria. The simulation database included a random sample of 200 inmates housed in OSCI on October 23, 1998. The simulation query was designed to identify the number of matches, or potential placements, for a given inmate. For example, when a query on the State Identification Number for an inmate was submitted, a list of potential cell locations for housing the inmate was produced. As in OWCC’s internal classification system, incentive level was originally the most critical housing assignment factor, although OSCI’s standard was more flexible (within one incentive level). Thus, the query was designed to select the “best” matches, as well as other acceptable matches, to maximize flexibility and minimize the number of transfers required to house any one inmate. The incentive-level criterion could not be violated, but matches were acceptable if one of the other assignment factors was violated. These simulations indicated that multiple potential cellmates could be readily identified for each inmate. This suggested that the model was sufficiently flexible to meet the identified goals and objectives. Even with this success, several tasks and barriers prohibited full implementation. These included— Full automation of the model into the OR DOC information system; unlike the female population, the OSCI population is too large to manage manually. 39 Chapter 2 Full implementation of the incentive level system, the key internal classification housing factor, had not been completed by OR DOC. Thus, the inmate’s incentive level had to be calculated manually, a time-consuming task for the entire OSCI population. Reduction of the OSCI population; a more manageable number of inmates would permit greater flexibility in the housing assignment process. As previously indicated, although the separate systems were originally developed for male and female inmates, with the automation of ICS and its implementation in March 2000, the same system is used for both males and females. Although it is still quite early to assess its impact on the male population, initial reports from the facilities have been positive. Florida Department of Corrections The Florida Department of Corrections (FL DOC) is one of the largest adult criminal justice systems in the country. As of April 1999, it had approximately 64,700 offenders in custody (5.3 percent of whom are women) in its 141 facilities statewide, which include 60 major institutions, 32 community corrections centers, 5 road prisons, 32 work camps, 3 forestry camps, and 9 drug treatment centers. In addition to these 141 facilities, there are 5 reception centers in which inmates are classified according to their medical needs, security requirements, and work and program eligibility (Florida Department of Corrections, 1999). FL DOC selected the Columbia Correctional Institution (CCI) in Lake City as the pilot facility for the internal classification system initiative for several reasons. First, it houses a variety of inmates (close, medium, and minimum custody inmates; violent and nonviolent inmates; and escape risks), so the sample would be generalizable to the entire FL DOC population. Second, CCI staff were familiar with the internal classification initiative because it had served as the pilot facility in earlier phases of the development of the FL DOC internal classification system. Description of Pilot Site CCI was a “quick build,” or modular, facility built in 1992. At the time of the onsite assessment, the complex included 10 housing units of the following types: General population open-bay units (five units: A, B, C, D, and E): Each unit has two open dormitory-type living units separated by a wall, each of which has separate toilets and showers. Both bays are under the continuous observation of control room staff. Inmates have access to work and program assignments outside of their living area. Secure living units/quads (two butterfly units: F and G): Housing with twoperson cells; four housing units are located in each building. Inmates have access to work and program assignments outside of their living area. 40 Description of the States’ Internal Classification Initiatives Close management unit (one single-cell unit: X): Inmates are housed in single cells. There are three housing units in each building. Inmates remain in their housing unit and do not have outside work assignments. Work-release population open-bay units: (two dormitories: J and K): Units are located within the work camp outside the main facility and fence. Florida’s Current External Classification Process FL DOC has an automated, objective custody classification system implemented in 1991. The system was designed to place inmates into one of four custody levels: minimum, medium, close, or maximum. Although the classification system appears to be adequate, increased pressure for public safety, changes in state sentencing policies, and changes in the characteristics of the inmate population prompted the need to revalidate the external classification system. Currently, the same nine risk factors are considered for both initial classification and reclassification: time remaining to the earliest release date, severity of offense, prior violent offenses, number of recent disciplinary reports, escape history, and stability factors (age, education, and involvement in vocational programs). In addition to these numerically scored risk factors, three mandatory override factors are considered that preclude an inmate from being assigned to minimum custody: active felony detainer, failure to participate in sex offender treatment, or (for inmates who are not U.S. citizens) deportation or a pending decision from the U.S. Immigration and Naturalization Service. Upon an inmate’s admission to a FL DOC Reception Center, an automated case file is built based on the rap sheet, NCIC (National Crime Information Center) report, commitment documents, medical evaluations, county jail records, and a personal interview. The objective classification instrument is scored automatically from these data. A classification team makes the initial classification and program decisions within the first 30 days of admission. This process includes a second interview with the inmate, during which he receives a copy of the classification instrument. The inmate is then transferred to a permanent facility. FL DOC undertook this internal classification initiative to On arrival at the permanent facility, a classification specialist reviews the classification instrument and program plans with the inmate. During an interview with the inmate, internal classification placement decisions regarding work assignments, programs, and institutional objectives are made. Once an inmate has been initially classified at the reception center, the Reclass Custody Questionnaire (DC4–869A) is used for all subsequent assessments. The initial and reclassification processes are identical, but the instruments differ in that the risk factors have different weights. Reclassifications are scheduled according to the time remaining for the inmate to serve. develop an objective, Florida’s Internal Classification Initiative ensured community FL DOC undertook this internal classification initiative to develop an objective, systematic process for housing the inmate population that was cost effective and legal and that ensured community safety. The specific goals were to— safety. systematic process for housing the inmate population that was cost effective and legal and that 41 Chapter 2 Develop an internal classification system that maximizes limited program dollars. Identify those inmates in greatest need and who are most likely to benefit from program participation. Increase the safety of both staff and inmates by developing an objective, internal system to identify those inmates who need long-term confinement, such as close management. In 1996, FL DOC contracted with the Department of Sociology and Criminal Justice at the University of North Florida (UNF) to develop a draft internal classification system called the Risk and Needs Model. The purpose of this system was to establish a framework for identifying offenders’ medical, mental health, emotional, educational, vocational, drug or alcohol abuse, or any other needs in conjunction with the external classification process. To identify effective models for separating violent inmates, the department, with the assistance of UNF, completed a survey of FL DOC classification staff, conducted focus groups with staff and inmates, and reviewed classification literature. Based on the data collected and stakeholders’ expectations, a preliminary Risk and Needs System was developed.11 During March 1997, a manual pilot test using a paper version of the Risk and Needs instrument was conducted at the Columbia, Jefferson, Lancaster, and Marion Correctional Institutions and the North Florida Reception Center. The results suggested that the system was useful, but required further refinement and automation. The software automation design process indicated that, to fully integrate the internal and external classification systems and to resolve security concerns, the current external objective classification system should be revalidated.12 The Florida internal classification system was part of a departmental initiative to automate FL DOC’s reception, case management, and release processes. The department intended to design an automated classification system that served the needs of multiple system stakeholders in the following ways: For the central office and facility-based classification staff, it would assess and group offenders according to appropriate security and custody levels. For Program and Labor Office staff, it would determine programs, services, vocational training, educational programs, and work assignments based on the inmate’s need and the availability of services. For Security Operations and Security Threat Group staff, it would determine the custody level and appropriate housing placement within a facility. 42 Description of the States’ Internal Classification Initiatives For facility-based classification staff, it would schedule reviews of security, custody, and program placement and reassess needs and progress for possible reclassification. For probation and parole staff, it would assess inmates for placement in community transition programs. Essential elements to the initiative’s success included time and MIS staff overload. The deadline for systemwide implementation of the inmate Risk and Needs System was close of business on December 31, 1998. Although information management consultants had multiple, complex projects, the department gave the project top priority. The multiple goals and needs of the stakeholders were translated into 16 risk and need factors:13 Attitude and motivation: The inmate’s attitude and motivation toward participating in primary programs.14 Restructuring potential: The inmate’s potential to benefit from primary program participation. Academic education: Academic education program needs. Vocational education: Vocational education program needs. Substance abuse: Substance abuse program needs (does not include dual diagnosis assessment). Work competency: General work assignment needs. Work release: Potential work release suitability. Outside work: Potential outside work assignment suitability. PRIDE/PIE: Potential PRIDE or PIE assignment suitability.15 Custody: Inmate’s external classification (custody) risk level. Internal management: The inmate’s internal classification risk level, a composite score of the inmate’s prior institutional violence and the classification specialist’s categorization of the inmate into one of the AIMS personality types. Outside influences: Assessment of the influences (positive and negative) that may affect an inmate’s adjustment. Included are family relationships, concern for children’s welfare and safety, associates or friends, intimate relationships, enemies, attorneys or legal representatives, etc. 43 Chapter 2 Housing: Internal housing recommendations by bed type (secure cell, room, or open day dorm), based upon the inmate’s external custody level and recent institutional misconduct. Wellness: Wellness program needs, i.e., the inmate’s need for a course in holistic living habits and positive leisure activities. Go lab/life skills: Go lab, life skills, and other selected self-betterment program needs. Transition program: Transition program participation suitability, based on the inmate’s projected release date. The primary components of the FL DOC internal classification system are the Risk and Needs Instrument and the Inmate Management Plan. The Risk and Needs Instrument (Exhibit 11) collects information about the 16 criteria listed above, as well as information on gang membership via an interview with the inmate. The factors are rated on a scale of one to five (with five representing the greatest need) based on the inmate’s life history, institutional adjustment, and participation in recommended programs and jobs. The system automatically generates work and program placement recommendations and a housing assignment, which The Risk and Needs interview data are integrated with the automated Offender Based Information System (OBIS) data to form the foundation for the Inmate Management Plan. Objective data from OBIS, the classification specialist’s observations, and the inmates’ self-reported information are combined to form a categorical score for each of the 16 risk and need factors. The Inmate Management Plan is the means by which key classification decisions are documented and carried out. The plan tracks the inmate’s primary work or program referrals, goals, short- and long-term objectives, housing recommendations, and current status on each of the risk and need categories. The system automatically generates work and program placement recommendations and a housing assignment, which can be updated by the classification specialist, if necessary. can be updated by the classification specialist, if necessary. 44 The initial Risk and Needs interview is conducted by the classification specialist during the reception process, with subsequent interviews mandated at regularly scheduled intervals and whenever the inmate is transferred to a different facility. Classification staff may also conduct unscheduled interviews at their discretion. The Inmate Management Plan is updated at each reassessment and the inmate is given a written notice of what is expected over the next 6, 12, or 24 months. FL DOC’s internal classification initiative was quite ambitious and complex. In addition to the “paper” pilot test at CCI, the fully automated software that generates the 16 risk and need factor scores and recommendations was tested at 6 other FL DOC facilities between June and September 1998. The primary purpose of the CCI pilot-test process was to determine whether the system was user friendly and useful to the classification specialists during their regular reassessment process. The pilot test included a presentation to CCI staff on the Risk and Needs System; handson training of staff with the relevant computer screens; scoring of inmates on the Exhibit 11. Florida Risk and Needs Instrument DC# STAFF ID # INMATE NAME STAFF SIGNATURE DATE OF ASSESSMENT / TIME ASSESSMENT BEGAN / / : Section I. Internal Management (IM-41) Complete either “A” if it is the inmate’s first plan, or “B” if the inmate has an existing plan. A. Initial Security Threat Group Membership Identification. Initial questions designed to determine if the inmate may have participated in gang activity. These questions are asked during the reception process or if the inmate has not previously had a risk and needs interview. A positive response to any of the questions will result in a referral (CDC generated automatic report) to the institutional inspector and the central office STG office. 1. Are you a member or associate of a neighborhood clique? A. Yes B. No 2. What is the name of the gang or group? 3. What is the place of affiliation? A. Street B. City C. State D. Prison 4. What groups are allies with your group? 5. What groups are enemies with your group? B. Subsequent Security Threat Group Membership Identification. Complete this section if the inmate has received at least one risk and needs assessment interview. Your goal is to discover if the inmate has connections to gang activity or is aware of any gang activity. You do not have to establish such a relationship as a fact. Any positive response to the questions, indicate such, and a report will be automatically forwarded to the STG Coordinator at your facility. 1. Have you joined any groups for protection or any other reason? If yes, what is the group’s name? 2. Since you have been here, were you asked to join any group? If yes, what group? 3. Have you had problems with any group? If yes, what is the group’s name? 4. Do you know of any groups or gangs in this or any other institution? If yes, what is the group(s) name(s)? 45 Exhibit 11. Florida Risk and Needs Instrument (continued) Section II. Outside Influences (IM42) Your goal is to discover if the inmate has any current positive or negative influences that primarily exist beyond the correctional environment that would affect the inmate’s behavior during incarceration. 1. Family relationships (if the inmate has no family, score 3) (3) A. Positive influence (2) B. Neutral influence (1) C. Negative influence (0) D. CRISIS—death or severe illness of immediate family member 2. Concern for children’s welfare and safety: (3) A. Average–normal concerns or regular contact (2) B. N/A–no children; little or no concern (1) C. High—children in foster care or living with relatives other than parent and inmate feels children’s welfare is suspect; children living with parent, but inmate feels children’s welfare is suspect; inmate desires contact with children but has little or no contact. 3. Intimate relationships (husband/wife, girlfriend/boyfriend): (3) A. Positive influence (2) B. Neutral influence (1) C. Negative influence (0) D. None 4. Associates/friends (influence on criminal activity—good or bad?) (3) A. Positive influence (2) B. Neutral influence (1) C. Negative influence 5. Enemies (outside enemies, not within the correctional system): (3) A. No (2) B. Is not sure (1) C. Yes 6. Attorney/Legal Representatives Select only one of the six choices available in A–C below that best describes the inmate’s present relationship status with legal counsel: (3) A. Positive relationship with legal services, private or public, meets one of the following: Has present involvement with outside legal services and has a positive view of assistance Does not have present involvement with outside legal services but has a positive view of assistance received (2) B. Neutral relationship with legal services, private or public, meets one of the following: Has present involvement with outside legal services and has a neutral view of assistance Does not have present involvement with outside legal services but has a neutral view of assistance received (1) C. Negative relationship with legal services, private or public, meets one of the following: Has present involvement with outside legal services and has a negative view of assistance Does not have present involvement with outside legal services but has a negative view of assistance received 46 Exhibit 11. Florida Risk and Needs Instrument (continued) Section III. Attitude and Motivation (IM43) 7. Others (not meeting one of the above categories–other inmates, religious, staff, professional, etc.) If none, mark “B” Neutral Influence. (3) A. Positive influence (2) B. Neutral influence (1) C. Negative influence Your goal is to first determine the inmate’s interest in recommended programs (Question 1). If the inmate has an interest in programs, whether recommended or not, list the inmate’s specific primary program (education, vocation, substance abuse) preferences (Question 2). Take this opportunity to seek the inmate’s preference for work assignments, and list those as well (Question 3). Finally, based on the overall content of the interview so far, rate the inmate’s apparent motivation to participate in programs (Question 4). 1. The inmate is interested in participating in: A. All of the possible recommended primary programs B. Some, but not all, of the recommended primary programs C. None of the recommended primary programs D. N/A, no primary program recommendations (or all primary programs completed) 2. If the answer to #1 is A or B, indicate the inmate’s order of preference of programs (code=course code) 1st Preference: Code 2nd Preference: Code rd 3 Preference: 3. Code Does the inmate have a preference of work opportunities? (code=work assignment code) A. Yes 1st Preference: Code nd 2 Preference: rd 3 Preference: Code Code B. No Preference C. Does not want work assignment 4. The inmate’s interest and motivation to participate in programs, either presently recommended programs or possibly future programs (if none are presently recommended), is rated as: A. Good B. Fair C. Poor 47 Exhibit 11. Florida Risk and Needs Instrument (continued) Section IV. Internal Management–Arrest History and Assessment (IM43) Your goal is to make an overall assessment of the inmate’s internal risk status based on the information available and the impressions developed as a result of the personal interview and a review of the inmate record. It is noted that this subjective assessment is only a part of the evaluation that determines the inmate’s actual IM score. A. Violent Felony Conviction during current incarceration only. Yes or No The data base will suggest a “yes” or “no” entry in this field, but the CPO shall review the inmate record to ensure the data is accurate. The CPO may place an entry in this field if the data base “suggestion” is inaccurate by changing the Y to an N, or vice versa. For instance, the data base may suggest a “yes” because the inmate committed a violent offense (resulting in a conviction) on the same day the inmate was received (or released). The CPO must determine whether the offense occurred while in prison (the “yes” would remain) or on the streets (the CPO would change the suggested “yes” to a “no.”) B. Overall rating based on a R & N interview, your expertise in classifying inmates, review of inmate record, etc. The interviewer shall select the score utilizing the below guidelines that most accurately describe the inmate. Score 5 if: The inmate displayed hostile, aggressive, and/or violent behavior during the interview, or if such behavior is documented and the inmate is resentful of rules and regulations and staff efforts to enforce them. Inmate appears to be easily bored and attempts to create excitement. Responses to questions indicate little concern for the feelings or welfare of others. The potential is present that disciplinary problems in an institutional setting may occur at a higher rate than other population. Is likely to be involved in fights, assaults, threats of bodily harm, extortion, destruction of property and possession of weapons. This assessment does not mean that the inmate will demonstrate this behavior; however, based on this evaluation, this score best describes the inmate’s potential behavior. Score 4 if: The inmate has an increased potential to be hostile to authority figure(s) by demonstrating conning or manipulative behavior. The inmate possesses the skill to organize inmate gangs, or illicit enterprises within the institution. Viewed by staff as being very untrustworthy and unreliable by prison standards. Is not usually directly confrontational but tends to cause problems by operating behind the scenes. Score 3 if: The inmate is worried, anxious, afraid, jittery, easily upset, and unhappy. Appears sad, depressed and tense. Is easily preyed upon by others. High potential to be influenced by others even though is responsible for his/her actions. Usually does not have a high rate of disciplinary reports. Score 2 if: The inmate appears withdrawn, passive, sluggish, and gives the impression of being unconcerned, friendless, indecisive, and submissive. Can be viewed as demanding (verbal), whining, and clinging. Usually does not have a high rate of disciplinary reports. Score 1 if: The inmate is not excessively dependent, although the experience of being in prison may be demoralizing to him/her. Normally does not have an extensive criminal history and does not see himself/herself as a criminal. Is rarely involved in assaults, fights, threats, or extortion within the institutional setting. Usually has a very low rate of disciplinary reports. 48 Description of the States’ Internal Classification Initiatives Risk and Needs Instrument to update the qualitative factors according to their progress, attitudes, and institutional adjustment; and structured feedback sessions, during which staff identified the system’s strengths and weaknesses.16 An important element of the pilot test was the participation of the computer software designers. As problems with the screens were identified, the programmers updated the software and addressed coding errors. The pilot test also provided ample opportunity for staff input into the design and refinement of the instrument. Based on the observations and feedback from CCI staff, FL DOC concluded the following: An important element of the pilot test was the participation of the computer software designers. The Risk and Needs System provided an objective tool for making internal classification decisions—housing, work and program assignments—yet was flexible enough to permit individual decisions according to the characteristics of a particular facility. The automated system was relatively easy to learn and use. It also saved time and effort by providing an immediate electronic record of the Inmate Management Plan. Resistance was inevitable because the system required changes in multiple divisions and disciplines across the department (e.g., programs, security, and classification). The system will provide the platform for other process improvements, such as revision of the external classification system and automated progress reports. The system will provide a method of evaluating inmates’ performance in achieving their restructuring goals as well as a means for evaluating the department’s ability to place inmates according to their needs and risk. The Risk and Needs System was just one component of the FL DOC’s Strategic Plan for 1998–2003. To assess its impact on FL DOC, a series of analyses were conducted based on Risk and Needs assessments completed between October 1998 and October 1999, the first year of use of the Risk and Needs System. Data were assessed according to the four key goals identified by FL DOC at the outset of this initiative. These goals are listed below. In addition, data were compiled to assess the system’s reliability. The results of these analyses suggested that the Risk and Needs System had been implemented successfully. Goal 1: Identify disruptive inmates utilizing objective, inmate behavior based data. Objective 1–1: Use the Risk and Needs to score all inmates. A review of the offender profiles indicated that as of October 31, 1999, approximately 95 percent of the FL DOC incarcerated population had been assessed according to the Risk and Needs System. Most of those not assessed were males within close management or on work release. Because the classification officers were not required to “score” 49 Chapter 2 close-management inmates unless they were to be reassigned, it appeared that the system had been fully implemented. Goal 2: Maintain safe and secure environment for FL DOC staff and inmates. Objective 2–1: Reduce inmate violent and disruptive behavior. To determine whether implementing the Risk and Needs System had reduced institutional violence, the levels of disciplinary infractions among inmates were compared within each housing type before and after inmates were placed according to the Risk and Needs housing recommendation. As Exhibit 12 suggests, the results were somewhat mixed.17 It appeared that for high-risk inmates placed in a secured cell or room, disciplinary infractions decreased under the Risk and Needs System. For those inmates placed in an open bay or special housing, however, infractions increased. Comparisons between inmates who were placed according to the housing recommendation and those who were not showed that disciplinary infractions increased by 11 percent among inmates who were placed according to the housing recommendation. These data suggest that the Risk and Needs System correctly identified the most aggressive inmates and provided for their placement in more secure housing, such as a room or secure cell. The increase in the disciplinary infractions among inmates placed in an open bay, however, suggests the need for special analysis of those inmates whose institutional misconduct increased after assessment to determine whether the system should be refined to better identify inmates suitable for dorms. The data also suggest a need to consider the staffing and supervision patterns for inmates placed in special housing because disciplinary infractions increased dramatically among these inmates. Exhibit 12. Institutional Disciplinary Index (INDI) by Housing Assignment Housing Number Assignment of Inmates Post-Assessment INDI Percent Change Special Housing 1,390 24 31 29 Open Bay 8,444 26 30 15 631 38 35 –8 5,007 52 46 –12 590 28 28 0 16,062 34.95 35.43 1 Room Secured Cell Bed Not Found Total 50 Pre-Assessment INDI Description of the States’ Internal Classification Initiatives Goal 3: Assign inmates to housing, work, and programming based upon their risk, needs, and time to serve. Objective 3–1: Assignment of inmates to housing, work, and program slots according to Risk and Need recommendations. A key goal for the Risk and Needs System was to identify housing, work, and program assignments for inmates based on their risk, needs, and time to serve. The first analysis focused upon the rate at which inmates were housed according to the Risk and Needs housing assignment. During the first year, 63.3 percent of the inmates were placed in the recommended type of housing. Twenty-one percent (20.6 percent) were not placed, even though the recommended bed type was available. Another 12 percent were not placed because the bed type was not available.18 Less than 1 percent (0.5 percent) were not placed in the recommended type of housing due to discretionary management decisions. These data suggested that most inmates were placed in the recommended housing type if the bed space was available. The low rate of discretionary overrides suggested that staff were comfortable with the system’s recommendations. Separate analyses of the data by gender indicated more women (77 percent) than men (63 percent) were placed in the recommended housing type (see Exhibit 13). When inmates were not placed in the recommended housing type, it was usually because the bed type was not available within the system. To determine whether inmates were assigned to work and programming slots according to the recommendations generated by the Risk and Needs System, the actual work and program assignments were crosstabulated with the inmates’ first recommendations. Exhibit 13. Housing Placement Results by Gender Housing Placement Met placement Males Females Total Number Percent Number Percent Number Percent 24,350 62.5 1,722 76.8 26,072 63.3 Did not meet placement 8,267 21.2 212 9.5 8,479 20.6 Community 1,646 4.2 2 0.1 1,648 4.0 No bed type 4,501 11.6 299 13.3 4,800 11.7 187 0.5 8 0.4 195 0.5 38,951 100.0 2,243 100.0 41,194 100.0 Discretionary decision Total Note: Percentages may not add up to 100 because of rounding. 51 Chapter 2 Exhibit 14 illustrates that 93 percent of male inmates whose first recommendation was work were assigned to a job. If the recommendation was for school, alcohol, or vocational programming, however, the recommendation was not likely to be met. In contrast, women who were recommended for programming were somewhat more likely to be assigned to the appropriate program. It also appeared that there were fewer work assignments for the women; only 85 percent of women whose first recommendation was work were assigned to a job. Exhibit 15 illustrates that 74 percent of those inmates not assigned to work or a program according to their first recommendation received their second or third recommendation. The discretionary override rate (14 percent) among males suggested that staff had confidence in the work and programming recommendations generated by the Risk and Needs System, but among female inmates, the override rate was high (28 percent). Further review of the data suggested that most of these were overrides to work assignments. This, again, may indicate a shortage of appropriate programming slots. FL DOC may want to analyze the fit between the programming, vocational, and work slots available and the women’s needs. Goal 4: Achieve consistency in housing and program recommendations. A fundamental question addressed by the assessment of the Risk and Needs System was the reliability of the housing, programming, and work recommendations. Although the Risk and Needs recommendations were based primarily on objective quantitative data, subjective assessments by classification counselors and self-reported data from inmates were also considered. In addition, discretionary overrides were permitted. To assess the inter-rater reliability of the recommendations, the initial assessments completed at the reception center were compared with full assessments completed by the classification staff at the long-term housing facility within 21 days of the inmate’s arrival. This was not an ideal test of the system’s reliability because new or different information may have been available to the classification staff at the receiving facility. Exhibit 14. Program Placement According to First Recommended Assignment Program Placement Substance abuse Females Total Number Percent Number Percent Number Percent 1,487 21.0 64 19.8 1,551 20.9 478 18.1 50 27.0 528 18.7 17 7.2 15 27.3 32 11.0 Work 26,929 92.9 1,420 84.5 28,349 92.5 Total 28,911 74.2 1,549 69.1 30,460 73.9 School Vocational 52 Males Description of the States’ Internal Classification Initiatives Exhibit 15. Program Placement According to Second or Third Recommended Assignment Program Placement Males Females Total Number Percent Number Percent Number Percent Met 2nd recommendation 8,293 61.1 482 51.2 8,775 60.4 Met 3rd recommendation 1,837 13.5 107 11.4 1,944 13.4 Did not meet recommendations 1,603 11.8 93 9.9 1,696 11.7 Discretionary override 1,849 13.6 260 27.6 2,109 14.5 13,582 100.0 942 100.0 14,524 100.0 Total Note: Percentages may not add up to 100 because of rounding. As shown in Exhibit 16, the recommendations generated by the long-term housing facility staff were consistent with those generated at the reception center for most Risk and Needs items. A few key items were problematic; for example, the reliability ratings for the internal management and housing placement recommendations were low. Housing recommendations generated by the reception center staff were consistent with those of the long-term facility staff approximately 80 percent of the time. This suggests that for 20 percent of the inmates, facility staff disagreed with the reception center staff as to whether the inmate should be housed in a dormitory, room, or secure cell. Even more troubling was the low rate of agreement between the reception center staff and long-term housing staff regarding the offenders’ internal management needs. These data suggest that FL DOC should review the scoring for the items on which the rate of agreement was less than 85 percent, provide additional staff training, and develop specific instructions for the items that require subjective ratings. New Jersey Department of Corrections The New Jersey Department of Corrections (NJ DOC) has jurisdiction over approximately 29,000 adult offenders serving sentences of longer than 1 year and supervises approximately 14,000 parolees. From 1980 to 1998, the correctional population increased by more than 400 percent and the parole population doubled. NJ DOC has 14 institutions spread throughout the state: two high maximum security, two maximum/medium security, eight medium/minimum security, one medium/minimum security facility for females. NJ DOC also has one central reception unit for adult and young adult male offenders, and separate reception centers 53 Chapter 2 Exhibit 16. Percent Agreement Between Reception Center and Facility Staff on Risk and Needs Items Risk and Needs Item Percent Agreement Substance abuse 90.96 Vocational 81.13 Education 88.05 Outside work assignment 74.62 PRIDE/PIE 90.01 Restructure priority 82.00 Transit assistance program 89.46 Work competency 88.02 Wellness measure 93.26 Associates and friends 46.57 Child welfare and safety 71.49 Enemies 75.95 Family relationships 75.02 Intimate relationships 57.07 Legal representation 49.34 Other points 52.47 Life skills 97.11 Internal housing placement 79.62 Internal management 54.06 for female offenders and sex offenders, respectively. Most of these facilities are operating at 40 percent above capacity. In 1998, approximately 41 percent of all adult inmates were incarcerated for violent offenses, 21 percent were incarcerated for nonviolent offenses, and 33 percent were incarcerated for drug-related offenses. (For the remaining 5 percent, offenses were not coded.) Ninety percent of NJ DOC inmates had at least one arrest before the current offense, 50 percent had previously been incarcerated in the NJ DOC system, and 50 percent had been on parole. Three-quarters of the inmate population had a history of substance abuse. Almost half of NJ DOC inmates had been sentenced to a term of 5 years or less. Inmates sentenced to terms greater than or 54 Description of the States’ Internal Classification Initiatives equal to 11 years constituted 26 percent of the inmate population. One-quarter of all NJ DOC inmates were classified as maximum or close custody (this includes administrative segregation, gang units, protective custody, and the management control unit). Minimum custody inmates made up 36 percent of the department’s secure facility population. Since 1993, NJ DOC has undertaken several initiatives to improve the practical utility, quality, accuracy, and objectivity of its inmate classification system. In addition, NJ DOC has improved its automated data systems and offender tracking systems to ensure that comprehensive and reliable inmate data are available at the facility level. Before this project, NJ DOC relied on subjective evaluations of offenders’ risk to the safety and security of the operation of the facility to make housing, job, and program assignments. While the individual or group responsible for making work and program assignments varies by institution, in most facilities, an institutional classification committee (ICC) assesses inmate needs and makes job and program assignments. In general, housing decisions are made by the ICC, although housing assignments are made according to job or program assignments in several institutions. Recognizing the shortcomings of a subjective system and the significant benefits of a research-based objective classification system, NJ DOC sought to develop and implement an objective internal classification system. The goals of the NJ DOC internal classification initiative were to— Identify inmates appropriate for placement in therapeutic communities; Develop a mental health treatment program for emotionally disturbed and developmentally disabled inmates; Provide security and safety for both staff and inmates through assigning inmates to appropriate housing units; Establish additional inmate work opportunities so that as many inmates as possible are occupied in meaningful daily job assignments; Effectively integrate the inmate population with housing, education, and work opportunities; and Monitor the inmate population for gang-related or regionally based negative activity. At the onset of the initiative, NJ DOC considered using a modified AIMS system to classify the entire inmate population and identified the Garden State Correctional Center (GSCC) as the pilot facility. GSCC is a treatment-focused, medium-security facility that provides comprehensive substance abuse treatment services to approximately 1,200 inmates. Implementing the modified AIMS system would help identify those inmates with substance abuse problems who were appropriate for housing in the modified therapeutic community setting. As discussed in earlier sections of this report, the AIMS system requires that case managers be familiar with the 55 Chapter 2 behavior and personalities of the inmates on their caseload and be able to translate that knowledge to behavioral and trait checklists. The NJ DOC administration believed that the correctional officers’ union would argue that such requirements were beyond the scope of work identified in their current contracts and would strongly resist the implementation of AIMS. As a result, NJ DOC redefined the scope of the classification project to focus on a smaller segment of the inmate population, those inmates classified as maximum custody. Maximum custody inmates are held in three facilities in New Jersey: Northern State Prison (NSP), East Jersey State Prison (EJSP), and New Jersey State Prison (NJSP). Two samples of inmates were selected from these three facilities to pilot-test an internal classification instrument that would distinguish among levels of aggression. NJ DOC’s intent was to identify inmates with high aggression levels for housing in a self-contained unit that would provide both heightened security and appropriate programming and work opportunities. Description of the Pilot Site The pool of offenders eligible for the maximum custody, high-aggression housing unit was distributed across the three facilities. The original plan included the transfer of highly aggressive inmates to a special housing unit at NSP, a 10-year-old, maximum/high-medium security facility with a rated capacity of 3,000 inmates. Not only does the facility operate general population programs, administrative segregation, and a gang unit, but it also houses minimum custody inmates and operates a regional halfway house outside of the perimeter fence. The Echo and Foxtrot housing units were targeted for housing highly aggressive inmates. The NSP population was distributed across the following custody levels: 13 percent maximum, 19 percent close, 37 percent medium, and 31 percent minimum. New Jersey’s Current Classification Process In 1994, NJ DOC developed an objective external classification system based on the NIC model. This system, which was implemented throughout the department in 1995, had a profound and positive impact on the assessment and assignment of inmates to an appropriate custody level and facility. Staff and inmate acceptance of the external classification system remains high and the objective system has been supported by concrete operational and policy revisions. The successful implementation of this system encouraged the expansion of the concept of objective classification to the facility level. Inmates are received and assessed at the Central Reception and Assignment Facility on external criteria (i.e., involvement with the criminal justice system, history of escape, use of alcohol and drugs, age, education, and employment). Inmates are assigned to maximum, close, medium, gang minimum, minimum, or community custody levels. 56 Description of the States’ Internal Classification Initiatives Empirical research has demonstrated that the classification system has effectively reduced the overclassification of inmates, institutional violence, and walkaways from minimum-security facilities (New Jersey Department of Corrections, 1997). Given its success in implementing the external classification system, NJ DOC was encouraged to pursue an internal classification system to assist in the assignment of inmates within facilities. New Jersey’s Internal Classification Initiative NJ DOC’s internal classification process relies on the subjective decisionmaking efforts of NJ DOC staff to make appropriate assignments for housing, work, and programs. In most institutions, an ICC assesses the needs of each inmate and makes assignments to treatment programs and work details. As previously noted, NJ DOC originally considered implementing a modified AIMS system to classify all of the inmates currently housed in secure settings. Significant resistance was anticipated from the correctional officers’ union to the training and administration of AIMS. As a result, NJ DOC revised its strategy and considered an alternative objective system for housing decisions. Thus, NJ DOC began to develop an internal classification model that would distinguish among groups of maximum custody inmates with differing levels of aggression to make safe and appropriate housing assignments within maximum security facilities. The plan was to house highly aggressive inmates at a special unit at the NSP. Upon refining the direction of the initiative and obtaining the commitment of the newly appointed NJ DOC commissioner, an Internal Classification Review Committee Task Force was formed. The task force was charged with developing a statement of purpose and a strategic plan for implementing the internal classification system departmentwide. A project team was also identified to coordinate the pilot testing and to revise the classification instruments accordingly. The goals of the NJ DOC internal classification model were to— Separate highly aggressive, maximum custody inmates from maximum-custody inmates with moderate or low levels of aggression. Relocate all highly aggressive, maximum custody inmates to one facility, if possible. Provide appropriate housing, work, and treatment programs in a specialized housing unit for this subset of the maximum custody population. Using the instrument shown in Exhibit 17, inmates were identified as having high, moderate, or low levels of aggression. The profiles for the inmates within the respective levels were defined as follows: Highly aggressive inmates are those with a recent institutional history of aggressive, violent, or confrontational behavior. These inmates tend to be younger and generally have been convicted of, or have past histories of, violent offenses. They are characterized as predators and “visible agitators.” 57 Chapter 2 Moderately aggressive inmates have less severe recent histories of violent or aggressive institutional conduct. These inmates do not permit others to take advantage of them. They may have a history or current sentence for a violent offense, but generally behave in a manner that would not threaten their potential return to society. Inmates classified as having low aggression have very limited or nonexistent histories of aggressive institutional misconduct. Usually older inmates, these offenders generally avoid confrontation and serve their sentence as quietly as possible. As shown in Exhibit 17, the NJ DOC Internal Classification Instrument is a behaviorbased instrument that combines scores on current and past criminal involvement, indicators of institutional adjustment, and more specific ratings of aggressive disciplinary infractions. The specific items are current offense involving violence, age, prior assaults, failure to adjust to minimum or medium custody during the past 2 years, number of disciplinary reports in the past 2 years, and administrative segregation placement in the past 2 years. A catalog of specific disciplinary infractions categorizes the reports into highest, high, and moderate severity levels and time since release from management control unit. Scores on all items are tallied to indicate a scored aggression level of low (zero to 6 points), moderate (7 to 11 points) or high (12 or more points). For several reasons, the classification officer may recommend a different aggression level than what is scored. All such recommendations are reviewed by the ICC for a final decision. Acceptable override justifications include the following: A field account of the offense that indicates a higher level of predatory behavior than that indicated on the report. The inmate’s prior criminal record. The inmate’s history of correctional facility adjustment. Reports from professional or custody staff. An override may also be recommended for reasons relating to the best interest of the inmate or to the safe, orderly operation of the department or correctional facility. These justifications must be properly documented to move the offender to a higher aggression level. Conversely, an inmate’s scored aggression level may be reduced upon a review of these same factors, again with proper documentation. Several policies were developed to support the implementation of the system. Once an aggression level has been scored, these policies guide the placement of an inmate in an appropriate housing unit. 58 Description of the States’ Internal Classification Initiatives Medium custody inmates will be housed only with maximum custody inmates with moderate or low aggression levels. Maximum custody inmates with low aggression levels may be housed in maximum, close, or medium security facilities. High-aggression housing units will assist in maintaining the integrity of the current system for processing “Keep Separates.” Inmates in administrative segregation will be classified before their release and appropriately housed according to the scored aggression level. Calculations for the aggressive disciplinary infraction table on the instrument should be based on the date of the inmate’s release from administrative segregation. Before release from the management control unit (MCU) or the final phase of the gang unit (STGMU), inmates will be classified by aggression level. Any inmate who scored in the medium custody level or below on the external classification instrument and scores as highly aggressive will receive an automatic override to maximum custody on the external classification instrument: For non-STGMU inmates, gang-related activity that did not result in a disciplinary infraction will not be scored on the internal classification instrument. Inmates who are classified as maximum custody at the reception unit will be further classified according to their aggression level before any institutional assignment. If inmates return to a maximum security facility within 2 years of release, their prior institutional history will be considered in the internal classification scoring process. This applies to parole, maximum release dates, or any other type of release from custody. Disciplinary records are considered for 2 years from the date of the most recent internal classification. All maximum custody inmates will be scored yearly or on release from a closed unit. Programs will be available for highly aggressive inmates, who will be housed separately in highly structured units and have restrictions on work details. The initial pilot test of this instrument included a sample of 235 cases selected randomly from the NSP, EJSP, and NJSP populations. The mean aggression score was 8.11 overall, with individual facility means ranging from 7.09 at EJSP to 11.77 at NSP, all within the moderate aggression level range. The distribution of inmates by aggression level was as follows: low, 44 percent; moderate, 36 percent; and high, 59 Exhibit 17. NJDOC Internal Classification Form: Maximum Security Facilities Interview Date / / / Current Facility: Inmate’s Name: Number: Current Custody Status: Review Number: Age: Transferred From: Current Internal Classification Status (if applicable): Section I. Criminal and Background Factors a. Current Offense Involving Violence Served less than 3 years = 6 pts. Served 3 to 5 years = 4 pts. Served more than 5 years = 2 pts. None = 0 pts. b. Age Under age 26 = 3 pts. Age 26 to 39 = 1 pts. Over age 39 = 0 pts. c. Prior Assaultive Offenses More than 1 = 4 pts. One = 2 pts. None = 0 pts. Section I Subtotal Section II. Institutional Adjustment 1. Failure to Adjust to Min/Med Custody During Past 2 Years If the failure is the result of a disciplinary report for assault, score 3pts., and score 2 pts. for fighting infractions. All other aggressive infractions score 1 pt. If there is no failure, score 0. 2. 3. Number of Disciplinary Reports Within Past 2 Years 5 or more = 3 pts. 3 to 4 = 2 pts. 2 or less = 0 (if within the past 6 months = 1) Ad Seg Placement Within Past 2 Years Indicate whether Ad Seg placement has occurred for the offenses listed on the Aggressive Infraction Table. Calculate total points using the appropriate weights for each offense and time frame shown in each cell of the table. 60 Exhibit 17. NJDOC Internal Classification Form, Maximum Security Facilities (continued) AGGRESSIVE INFRACTION RATING ADMINISTRATIVE SEGREGATION DAYS 30–90 days Incident Time Frame 90–180 days 181+ days Last 12 mo. 13–24 mo. Last 12 mo. 13–24 mo. Last 12 mo. 13–24 mo. Highest 6 4 8 6 10 8 High 4 2 6 4 8 6 Moderate 2 0 4 2 6 4 4. Released from MCU Within past year = 9 pts. One to two years = 4 pts. Over two years or N/A = 0 pts. Section II Subtotal Section I Subtotal TOTAL SCORE ASSESSED CLASSIFICATION LEVEL A. Scored Aggression Level (circle one) Low 0–6 pts. Moderate 7–11 pts. High 12+ pts. B. Recommended Aggression Level (circle one) Low Moderate High Date Scored By: C. Override D. Assigned Level E. ICC Chairperson F. Next Review / / / Reasons: Low Moderate Date / / High / / / / 61 Chapter 2 20 percent. The mean aggression score was 16.57 for those classified as highly aggressive, 8.75 for those classified as moderately aggressive, and 3.69 for those classified as having a low aggression level. NJ DOC currently houses approximately 4,200 maximum custody inmates. Using the proportions resulting from the pilot test, a projected 840 inmates in the total population would be classified as highly aggressive. Based on the experiences of the coders during the pilot test, the instrument underwent several revisions. The “final” version of the instrument was pilot-tested to ensure its validity and reliability in the maximum-security setting. A random sample of 302 cases was coded and classified by aggression level. These data were analyzed by ICJC to refine the aggression-scale cutoff points and project the distribution of cases by aggression level. NJ DOC needs to further analyze the effect of changes to the aggression-scale cutoff scores. Other variations to be considered are the points allocated to the inmate’s age and institutional adjustment criteria. These analyses will also model the changes in distribution across aggression levels that would result from a change in the length of time that disciplinary reports are tracked. These analyses were scheduled to be completed in the summer of 1999, but were delayed because NJ DOC was converting its management information system to a new, Y2K-compliant client-server management information system (DOCMIS). This new system was necessary to accommodate the new data fields required for the internal classification instrument and to maintain the integrity of existing databases. Although data fields for the internal classification system have been included in the new DOCMIS system, the internal classification system will not be implemented until all critical new MIS applications are fully operational. Originally, NJ DOC planned to house the high-aggression inmates in a centralized location (NSP), but significant funding complications prevented pursuit of this plan. The preliminary cost estimates to establish a high-aggression facility were approximately $375,000, much higher than originally anticipated. This estimate included the enhanced security, classification services, and treatment programs necessary for the appropriate maintenance of inmates on the unit. Serious fiscal constraints and continued growth in the inmate population prevented this expenditure in the fiscal year 2000 budget cycle. In response, NJ DOC developed an alternative plan to implement the internal classification instrument at each of its four maximum/high-medium security institutions, with a special unit in each facility to be designated for housing highly aggressive inmates. To fully implement these units, work and treatment programs that could be provided with existing resources and staff need to be identified. To date, NJ DOC has not completed the design of its internal classification system. Implementation plans are pending the new MIS becoming fully operational. 62 Description of the States’ Internal Classification Initiatives South Dakota Department of Corrections The South Dakota Department of Corrections (SD DOC) includes seven facilities: three main adult facilities, two trusty units, a prison annex, and the prison industries building.19 The adult average daily institutional count during fiscal year 1998 was 2,267 inmates (2,096 males and 171 females) (South Dakota Department of Corrections, 1998). In March 1999, 47 percent of the total inmate population was incarcerated for a violent crime. The South Dakota Penitentiary Complex, a 1,175-bed prison located in Sioux Falls, was selected as the site for the internal classification initiative because it is the state’s primary penal facility and functions as the reception and diagnostic center for all male inmates. These selection criteria were important to the overall goals of the South Dakota initiative: Complete a total review of the SD DOC AIMS policy. Determine the reliability and validity of the AIMS instruments among the SD DOC inmate population. Develop a behavior-based system that allows for differential assignment of the maximum-custody inmates with AIMS codes of A and B. Description of the Pilot Site The Penitentiary Complex consists of the penitentiary, built in 1889, and the Jameson Annex, built in 1993. Although the complex is centrally administrated, the facilities function independently. The penitentiary and Jameson’s A/B Unit, for example, have separate exercise yards and industries buildings. The complex is considered a maximum security facility, although it also contains medium and minimum security housing units. The penitentiary consists of linear blocks, most of which are five tiers high (one block is four tiers high). Jameson Annex has three units, each with three wings. Two of these units have double cells; the third unit is composed of dormitories. South Dakota’s Classification System As previously indicated, the Penitentiary Complex houses SD DOC’s 56-bed admission and orientation (A & O) unit. The intake process entails assessment of the inmate’s educational, chemical dependency, medical, and mental health needs. Inmates committed for sex-related offenses are also evaluated by the mental health unit for the Sex Offender Treatment Program (STOP). The initial classification custody form includes five numerically scaled factors: violence in the current offense, pattern of past violence, escape history, time remaining to serve, and institutional misbehavior. Four mandatory override factors are not expressed numerically: need for sexoffender treatment, incomplete criminal history information, serving escape sen63 Chapter 2 tence, and pending charges or holds. An inmate who needs STOP treatment, for example, cannot be placed in minimum custody housing. Discretionary overrides are permitted with adequate documentation but are reported to be extremely rare (less than 1 percent of all cases). Inmates are reclassified at least annually or on the receipt of new information regarding any custody factor. The reclassification custody form is the same as the initial classification custody form except for three modifications: severity of current offense is weighted less and two additional items are considered—poor and exceptional institutional behavior. The rate of discretionary overrides at reclassification is not tracked, but staff estimate that overrides are used in less than 1 percent of the cases. As to the validity of the system, a 1996 study conducted by the DOC indicated that there was “a statistically significant, though weak, relationship between initial assessed risk and conviction for class 4 or 5 disciplinary infractions.” (Mande, 1996) Since this study did not include the factors that are distinctive to the reclassification form, actual risk behavior and exceptional behavior, the reclassification instrument has not been empirically validated. As previously indicated, at the outset of this initiative, SD DOC used the standard AIMS instruments and scoring process. AIMS was completed during the admission and orientation process and was not reviewed as part of the reclassification process because the system is designed to measure personality traits assumed to remain constant. Before the AIMS profile is completed, inmates with convictions for violent offenses are housed separately from nonviolent offenders. Before this initiative, the reliability and validity of SD DOC’s use of AIMS had not been evaluated. Staff who observed the complex before and after the implementation of AIMS report that it has reduced the exploitation of vulnerable inmates. Furthermore, staff believe that AIMS accurately identifies unique personality types with distinct behaviors, thus testifying to its face validity. SD DOC issued a policy requiring that all inmates be classified under AIMS (SD DOC Policy 48.6). When AIMS was first implemented, staff were formally trained and written instructions were distributed. Yet SD DOC has not provided formal training or written instructions to new staff, nor does the department provide inservice training on AIMS. Inmates’ AIMS classification does not affect their job or program assignments. These decisions are made through a program and work schedule that indicates the start and stop dates and the number of vacant slots. The case manager maps out inmates’ program and work schedules for their entire term based on their external classification custody level, needs, skills, sentence length, and available slots in the required programs. This schedule is known as the Inmate Program Directive. South Dakota’s Internal Classification Initiative SD DOC undertook an internal classification initiative in concert with the Missouri Department of Corrections. Therefore, the initiatives and pilot-test results are reported together, following the description of Missouri’s current systems. 64 Description of the States’ Internal Classification Initiatives Missouri Department of Corrections As of March 1999, the Missouri Department of Corrections (MO DOC) housed approximately 24,900 inmates in 22 facilities (Missouri Department of Corrections, 1999). These 22 facilities included 16 correctional centers, 1 camp, 4 reception centers, and 1 treatment center. Most of the correction centers offer academic instruction, GED testing, and substance abuse assessment and education. All MO DOC inmates are required to work, attend school, or participate in treatment programs full-time during their term of incarceration. MO DOC implemented AIMS in 1987. While the initial implementation was reported to reduce inmate-on-inmate victimization, the department sought to improve and refine the system through this initiative. Three key problems with the existing system had been identified: AIMS had never been evaluated for the MO DOC system. There was racial disparity in that urban, black inmates were more likely to be classified as Alphas (predatory inmates), and rural, white inmates were more likely to be classified as Sigmas (victims). Housing options were limited, as separation of Alphas from Kappas and Sigmas precluded the creation of a honor block, which was a goal of the MO DOC. Crossroads Correctional Center was selected as the pilot-test facility because both the administrative and classification staff were committed to reworking AIMS. Crossroads was a new maximum security prison (opened in March 1997) with a mixed inmate population. Many inmates were serving long sentences for violent offenses, whereas others had been placed in maximum custody due to poor institutional adjustment at other facilities. The population had a higher percentage of young, minority, urban inmates than the MO DOC population in general, but most of the staff were transferred from medium security prisons or were new recruits from the local rural communities and thus had limited experience with such inmates. Improving the reliability of the internal classification system was particularly critical, as many of the MO DOC staff were new and inexperienced. The internal classification system is the primary mechanism for guiding the staff in the assessment and housing assignment process. Description of the Pilot Site Crossroads Correctional Center is a 1,510-bed, maximum-security prison located on the outskirts of Cameron, Missouri, about 45 miles north of Kansas City. The prison consists of five housing buildings, an activity building, and an administration building. Each housing building has four wings and a central control bubble. The wings have two floors with 18 double, wet cells per floor. Therefore, each building 65 Chapter 2 houses 288 inmates. In addition, there are 60 single cells and 10 infirmary beds in the activity building. Missouri’s Current Classification Process MO DOC uses a version of the Correctional Classification Profile (CCP) external classification system developed by the Correctional Services Group. Each inmate is assessed using a scale of one to five on seven domains: medical, mental health, public risk, institutional risk, education, vocational training, and work skills. The public risk score is based on six factors: time to earliest release, detainers, escape history, violence in the criminal history, violence in the current offense, and sex offense (past or current conviction). The institutional risk score is based on two factors for first-time offenders: escape history and adjustment on probation and parole. For repeat offenders, two additional factors are considered: inmate assault history and institutional risk score at time of release. Although the reclassification schedule varies according to the inmate’s custody level, inmates are reclassified at least annually. Significant changes in any of the custody factors will generate a classification review as well. The inmate’s case manager is responsible for conducting the custody reclassification. The inmate’s custody reclassification is discussed with him, audited by an independent staff person, and then reviewed by the warden’s designee. All overrides for “aggravating or mitigating risk information,” “exceptional disciplinary adjustment,” or “other factors” must be approved by the warden and director of classification. At reclassification, the same seven domains are assessed. The only change in the calculation of the public risk score is a point reduction for sex offenders who have completed the sex offender treatment program. The major difference between the initial and reclassification instruments is that at reclassification, the institutional risk score is based on multiple measures of the severity and persistence of disciplinary problems and the custody level at which the problems occurred. Staff report no problems with the external reclassification process. Missouri and South Dakota Internal Classification Initiative As previously reported, both South Dakota and Missouri were currently using AIMS when NIC introduced this internal classification initiative. Both states sought this opportunity as a mechanism for refining and validating AIMS for their institutional populations. Although both states attributed their relatively low rates of institutional predatory incidents and protective custody admissions to AIMS, concerns remained. The key issues and questions follow: Are the current inmate populations comparable to the populations on which AIMS was developed? If not, is AIMS still adequate? 66 Description of the States’ Internal Classification Initiatives Is AIMS subject to racial bias? Can the racial disparity be reduced? AIMS reduces flexibility for inmate housing. Alphas and Sigmas cannot be housed together, though either can be housed with Kappas. Kappas are few in number, so they have to be moved frequently as Alpha and Sigma counts fluctuate. Is AIMS compatible with the idea of establishing an honor dorm or block? How reliable is AIMS? How valid is AIMS? After reviewing the literature on the development and analyses of AIMS,20 Missouri and South Dakota joined forces to develop an internal classification system modeled after AIMS. The primary modifications identified by joint steering committees were to—21 Develop “situational” items to be put on the behavioral checklists. Revise the history checklist to focus on behavior before incarceration and revise the behavioral checklists to focus on prison adjustment. Develop a reclassification instrument and process. While the original AIMS assumed a static personality profile, both states sought a mechanism for updating the classification based on the inmate’s adjustment to institutional life. The behavioral checklists were to be updated annually and the weight of the history checklist is to decrease progressively over time. Collapse the two predator and situational subtypes in AIMS into one predator and one victim type. Revise the checklists to eliminate redundant items, clarify vague and outdated language, and add new items. Clarify the definition of items. The internal classification system developed by South Dakota and Missouri was named the Adult Internal Classification System (AICS). AICS assigns inmates to one of three offender types (Type 1 inmates are aggressive, Type 2 inmates are situational, and Type 3 inmates are vulnerable) based upon checklists composed of yes-no questions. The first instrument is similar to the AIMS life history; the AICS history checklist assesses inmates’ social and criminal history before incarceration (see Exhibit 18). The second AICS checklist, the initial behavior checklist, parallels the AIMS correctional adjustment checklist. The initial behavior checklist considers inmates’ adjustment to institutional life and their interactions with other inmates and staff upon admission to the correctional system (see Exhibit 19). A major difference between AICS and AIMS is that AICS does not assume that the “personality type” remains constant throughout an inmate’s incarceration. A 67 Exhibit 18. AICS History Checklist DOC Number: Received Date: Offender Name: Wing: Facility: Rater Last Name: Cell: Rater Position: Enter a “1” for each behavior trait which best describes the inmate’s history. If the trait is not descriptive of inmate’s life history, enter a “0.” On item #1, enter “0” if the offender has no prior incarcerations. 1. Prior incarceration demonstrated predatory behavior 2. Uncontrollable as a child 3. Has expressed guilt over offense 4. Expresses need for self-improvement 5. Socially withdrawn 6. Passive, indecisive, easily led 7. Defied authority 8. Requires assistance from others in daily living 9. Accepts responsibility for family obligations 10. Stable work history 11. Ties with criminal elements 12. Felony first arrest 13. Depressed, suicidal, prior psychological hospitalizations 14. Stable relationship for at least 5 years 15. Disciplinary difficulty in school 16. Commission of offense related to financial reverses suffered prior to offense 17. Accepted responsibility for his/her situation 18. Apprehended due to poorly planned, poorly executed crimes that display very inept, incompetent behavior 19. Self-centered 20. Hostile to criminal justice supervision 21. Physically aggressive 22. Thrill-seeking 68 Exhibit 19. AICS Initial Behavioral Checklist DOC Number: Received Date: Offender Name: Wing: Facility: Cell: Initial Classification Date: Final Initial Classification Level: Rater Last Name: Rater Position: Enter a “1” for each behavior trait which best describes the inmate’s history. If the trait is not descriptive of inmate’s life history, enter a “0.” 1. Demands immediate satisfaction 2. Tries, but cannot seem to follow directions 3. Cooperative with staff 4. Afraid of other inmates 5. Brags 6. Passive, easily led 7. Difficulties with free time 8. Avoids fights unless directly challenged 9. Seems to be unhappy, complains often 10. Argues with or questions staff 11. Moody, temperamental 12. Respectful of others 13. Worried, anxious 14. Quick-tempered 15. Continually seeks help from staff 16. Acts before thinking 17. Seems dull and unintelligent 18. Physically aggressive 19. Dependable in assignments 20. Sluggish, drowsy 21. Openly defies rules and supervision 22. Jittery, jumpy, unable to relax, seems afraid 69 Exhibit 20. AICS Reclassification Behavioral Checklist DOC Number: Received Date: Offender Name: Wing: Facility: Cell: Initial Classification Date: Final Initial Classification Level: Rater Last Name: Rater Position: Enter a “1” for each behavior trait which best describes the inmate’s history. If the trait is not descriptive of inmate’s life history, enter a “0.” 1. Victimizes inmates 16. Difficulties with free time 2. Fights within the last year, assault on staff within the last 5 years 17. Dependable in assignments 18. Seems dull and unintelligent 3. Avoids fights unless directly challenged 19. Demands immediate satisfaction 4. Thrill seeking 20. Brags 5. Sluggish, drowsy 21. Continually seeks help from staff 6. Acts before thinking 22. Openly defies rules and supervision 7. Socially withdrawn 23. Moody, temperamental 8. Accepts responsibility 24. Passive, easily led 9. Cooperative with staff 25. Respectful of others 10. Argues with or questions staff 26. Seems unhappy, complains often 11. Physically aggressive 27. Worried, anxious 12. Quick-tempered 28. Stirs up trouble among inmates 13. Jittery, jumpy, unable to relax, seems afraid 29. Prefers to be with physically aggressive inmates 14. Tries to control group for own interest 30. Afraid of other inmates 15. Tries, but cannot seem to follow directions 70 Description of the States’ Internal Classification Initiatives reclassification behavioral checklist that relies heavily upon recent institutional adjustment must be scored annually to monitor changes in the behavior and adjustment of the inmate (see Exhibit 20). At initial classification, the history and initial behavior checklists are weighted equally. With each annual reclassification, however, the weight of the history checklist decreases by 10 percent, while the weight of the reclassification behavioral checklist increases by 10 percent. The AICS checklists are composed of a series of subjective yes/no ratings that describe the three offender types. All responses are weighted and tallied. The offender type with the highest score determines the inmate’s internal classification type. Type 1 responses are weighted slightly higher than Type 2 or 3 responses to ensure that aggressive, predatory offenders are identified. If, for example, the case worker indicates that an offender has an equal number of Type 1 and Type 3 characteristics, he or she will be classified as a Type 1 offender. The preliminary checklists were pilot-tested twice to assess the reliability of AICS, assess its impact on housing assignments, and identify any problems with the revised items, instructions, or staff training. Because the most critical aspect of the first pilot test was the reliability of the instruments, both states selected random samples of cases, trained staff members to use the revised instruments, and required that each case be classified independently by at least two persons. A detailed training manual was prepared, and all coders participated in training sessions during which the instruments were explained and instructions for coding were provided. The first pilot-test samples consisted 149 cases from Missouri and 199 from South Dakota. The samples were stratified by custody level (maximum, medium, and minimum), race (minority versus white), and classification type (initial versus reclassification). The first question addressed by the pilot tests was the reliability of the instruments for each of the two states. For the purposes of this analysis, reliability was defined as agreement between “rater one” and “rater two” on the respective checklist items and the final classification, which was based on the combination of the two checklists’ categorizations. Reliability was a critical issue, because inconsistent scoring would cast doubt on the results of any further analyses. Exhibit 21 shows that inter-rater reliability was unacceptable for both states. The first pilot test results indicated the following: During the first pilot test, the reclassification behavioral checklist was less reliable than the history checklist. This was probably, at least in part, because staff had less experience with the reclassification process and were required to score the checklist based upon behaviors with low base rates. Reliability rates from the first pilot test ranged between 74 and 64 percent. The highest rate of agreement was 74 percent for the initial classification decisions in Missouri. These rates of inter-rater agreement were too low for the other analyses to be considered significant. 71 Chapter 2 Reliability rates were remarkably similar in the two states, which suggests that the problems were inherent in the instruments, training, or the scoring and observation processes. Exhibit 21. Missouri and South Dakota Pilot Test Reliability Rates Percent Agreement Between Raters One and Two Assessment Missouri Percent Agreement South Dakota Percent Agreement Pilot Test 1 Pilot Test 2 Pilot Test 1 Pilot Test 2 Initial Classification 74 64 73 78 Reclassification 64 84 69 88 Based on the first pilot test, several key problems were identified: Operational definitions needed to be further clarified and refined. Redundant items needed to be eliminated. Staff needed to understand the benefits and purpose of revising the checklists and the importance of their efforts. The staff training process had to be reviewed and updated. Quality control procedures for monitoring the AICS had to be enhanced. The instruments did not differentiate strongly enough among inmates; 90 percent of the inmates scored the same on 42 of the 87 items on the checklists. The small number of Type 2 items weakened the instruments because undue weight was placed on a small number of items. The states conducted a second pilot test with 25 randomly selected initial classification and 50 reclassification cases from each state, for a total of 150 cases. The issues highlighted during the first pilot test were addressed through further refinement of the instruments, staff training exercises, and the development of procedures. The results from this second pilot test are shown in Exhibit 21. The data suggested that the reliability rates improved, except for initial classifications in Missouri.22 Although the modest reliability rates rendered the subsequent analyses somewhat suspect, it was worth noting that for both states, AICS produced a sharp increase in the number of inmates classified as Type 2. Corresponding decreases in the number of inmates classified as Type 1 and 3 were also observed. 72 The states met for a fourth joint workshop in September 1999 to review the results of the pilot tests and plan for the implementation and monitoring of AICS. The final Description of the States’ Internal Classification Initiatives versions of the checklists were agreed upon and monitoring reports were developed to assess each state’s original internal classification goals. (See Exhibits 18–20.) An automated program for data entry of the checklists, calculation of final scores, determination of offender types, and production of monitoring reports was developed through a separate NIC short-term technical assistance program.23 Data collection to assess the impact of AICS was initiated in January 2000 at the Penitentiary Complex in South Dakota and the Crossroads Correctional Center in Missouri. The first question considered by the impact study was the ability of AICS to differentiate between assaultive and nonassaultive inmates. The preliminary reclassification data from South Dakota that compare the number of assaults during the past 12 months across the respective AICS and AIMS classification types are provided in Exhibit 22. AICS predicts that Type 1 inmates are more likely to be assaultive than Type 2 and 3 inmates, and AIMS predicts that Alpha inmates are more likely to be Exhibit 22. Comparison of Assaults by AICS and AIMS Classification Types for Reclassification Cases: February–March 2000 AICS Types Inmates Committing South Dakota Missouri Type 1 Type 2 Type 3 Total Type 1 Type 2 Type 3 Total No assaults 33 87 23 143 94 257 32 383 1 assault 8 2 1 11 5 0 0 5 2 or more assaults 0 1 0 1 1 0 0 1 41 90 24 155 100 257 32 389 Total AIMS Types Inmates Committing South Dakota Alpha Kappa Missouri Sigma Total Alpha Kappa Sigma Total No assaults 95 17 29 141 295 25 56 376 1 assault 10 1 1 12 4 0 0 4 2 or more assaults 0 0 1 1 0 0 0 0 105 18 31 154 299 25 56 380 Total 73 Chapter 2 assaultive than Kappa or Sigma offenders. SD DOC data suggested that 96.5 percent (110 out of 114 inmates) were accurately identified as nonpredatory (87 Type 2 and 23 Type 3 inmates were correctly identified as nonassaultive), while eight Type 1 inmates were correctly identified as predatory (8 out of 41 cases, or 19.5 percent). Overall, 76 percent of the AICS predictions were correct compared to 26 percent of the AIMS predictions. Among the Missouri inmates, AIMS correctly identified all of the predatory inmates as Alpha. Yet the false positive rate was also quite high; 79 percent of the inmates (299 of 380) were identified as Alpha, 99 percent of whom were not involved in assaultive behavior. These data suggest that AIMS did not differentiate among Missouri inmates. In contrast, all of Missouri’s predatory inmates were classified by AICS as Type 1 offenders, whereas none of those classified as Type 2 or 3 offenders were involved in predatory behavior. These analyses are preliminary and should not be interpreted as a validation of AICS. In addition, the low base rate of assaults among Missouri inmates precludes more indepth analyses of the predictive power of AICS versus AIMS. On an individual level, however, it is disappointing to find that among South Dakota inmates, whereas AIMS correctly predicted 10 of the 13 assaulters (76 percent), AICS correctly predicted only 8 of the 12 assaulters (66 percent). Definitive statements are difficult, given that inmates were not housed as AICS would dictate. Thus, the system’s actual impact cannot be determined. A second important objective of the South Dakota and Missouri initiative was to increase the flexibility of the housing assignment process by identifying Type 2 inmates, who, it is assumed, can be safely housed with either Type 1 or Type 3 inmates. Exhibit 23 shows that AICS classified a greater number of inmates as Type 2 than AIMS in both South Dakota and Missouri. Of 222 South Dakota inmates classified in February and March 2000, AIMS classified 37 as Kappa, whereas AICS classified 145 as Type 2. Similar results were observed for Missouri, where 65 percent of the inmates assessed were classified as Type 2. Thus, AICS would give facilities greater flexibility in making housing assignments. Exhibit 23. AICS and AIMS Classifications Compared for All Cases Classified: February–March 2000 South Dakota Missouri AIMS Type 1 Type 2 Type 3 Total Type 1 Type 2 Type 3 74 Total Alpha 36 82 12 130 87 204 23 314 Kappa 3 31 3 37 2 23 2 27 Sigma 1 32 22 55 7 37 19 63 Total 40 145 37 222 96 264 44 404 Description of the States’ Internal Classification Initiatives Exhibit 24. Racial Distribution of Inmates Under AICS for Reclassification Cases Classified: February–March 2000 South Dakota Missouri Race Type 1 Type 2 Type 3 Total Type 1 Type 2 Type 3 Native American Total 14 26 3 43 0 2 0 2 6 6 0 12 76 121 5 202 White 20 58 21 99 25 133 27 185 Total 40 90 24 154 101 256 32 389 Black/Other The third critical objective for the Missouri and South Dakota internal classification initiative was to reduce the racial disparity that was observed under AIMS. Chisquare analyses of the South Dakota and Missouri data suggest that there was a significant relationship between race and AICS classification. Exhibit 24 illustrates that the number of white inmates identified as Type 3 and the number of black inmates identified as Type 1 were higher than expected if the classification system was free of disparity. The preliminary data shown in Exhibit 24 suggest that racial bias may have been reduced, but not eliminated, under AICS. The racial disparity was greater among the Missouri assessments than in those in South Dakota. Missouri and South Dakota Internal Classification: Next Steps Given the low reliability of AICS, analysis of the impact of the modifications to the checklists and AICS classification process are inconclusive at best. The lack of certainty as to the “correct” classification of any one inmate seriously impedes the ability to determine the accuracy or reliability of the distribution of Type 1, 2, and 3 inmates across the populations. Moreover, this lack of certainty limits any findings about the flexibility in the housing process afforded by the larger number of Type 2 inmates. Yet these data are promising. Although reliability issues need continued attention, AICS will need to be implemented, as least in a pilot facility, to fully assess the system’s utility and validity. Missouri has undertaken a validation study of AICS; the results, however, will not be available until the summer of 2002. 75 three Chapter Common Problems, Issues, and Solutions Lessons Learned The complex and varied experiences of the eight states involved in this initiative provided substantial insight into the process, core elements, and critical barriers to implementing internal classification systems. All parties, including NIC, learned new lessons and added depth and dimension to their understanding of the key issues. Some of the lessons learned had been noted in previous NIC initiatives, while others lent new growth and complexity to the field of internal classification. This chapter will include a summary of the most poignant lessons and provide illustrations from the states’ experiences. Lesson 1: Plan to plan, and then double your resources Proper implementation of any classification system requires a strong commitment, time, and very careful planning. This was a key lesson learned by NCCD in the first internal classification initiative in 1993, and its importance was reinforced throughout subsequent initiatives. Although NIC embarked on this project with what appeared to be ample time, staff, and resources and then carefully selected sites based upon their commitment, resources, and goals and objectives, even this wellfunded and multidimensional initiative met significant challenges. For example, the need for time onsite to assist, observe, prompt, and encourage the states to design, pilot-test, and complete the required research tasks exceeded the evaluators’ expectations and budget. This is not to imply that the state staff were incompetent or uncommitted. On the contrary, across all of the states, the staff (central office, institutional, research, etc.) were highly competent, committed, and knowledgeable about classification. The challenge was to ensure that each department’s resources (staff time, funding, and knowledge) continued to focus on this initiative. Proper implementation of any classification system requires a strong commitment, time, and very careful planning. More specifically, at the onset, Oregon struggled to develop and test this model within OWCC. The level of overcrowding limited bed space flexibility, and OWCC staff had insufficient time to develop the model. Not until additional bed space was found at EOCI and a member of the steering committee was shifted from OWCC program services to the classification and transfer unit did the OWCC initiative progress. With the adequate resources—staff time and bed space—OWCC staff were able to meet all deadlines for the development, testing, and implementation of the internal classification system for the female population. The situation in Florida was totally different. From the outset, the Risk and Needs System was a top prior77 Chapter 3 ity not only for the Central Office Classification Bureau, but for the entire department. Extensive computer, staff, and travel resources were tapped to design, automate, and implement the system. Only with this significant and enduring commitment was Florida able to carry out its ambitious initiative to design, pilottest, and then implement the system statewide by December 1998. Lesson 2: Keep it simple Because the behavioral systems are quantitative and objective, high levels of reliability and validity were easier to achieve. A second lesson learned during this initiative was that simplicity is paramount. The models developed and tested by New Jersey and Oregon were behavior-based and relatively simple to score and implement. Although New Jersey is still refining its model, full implementation is clearly feasible. Oregon started with a relatively simple instrument that was further simplified with additional testing and implementation. Because the behavioral systems are quantitative and objective, high levels of reliability and validity were easier to achieve. In contrast, Florida’s complex Risk and Needs System required a significant expenditure of staff time and fiscal resources to develop and implement. Few states have the level of MIS sophistication, availability of computer program staff, or the fiscal resources to design and automate a similar model. Despite its impressive complexity and automation, the system depends on classification specialists’ subjective assessments of key risk and needs factors, so reliability and validity may be difficult to maintain. Over time, the automated systems developed and implemented by Connecticut and Colorado have proven to be both reliable and valid. These systems were implemented with relatively few resources and appeared to meet the internal classifications goals of the facilities for which they were designed. Their strength lies in their simplicity and objectivity. Unlike external classification systems, there is no distinct set of internal classification factors that is generalizable across systems. 78 Missouri and South Dakota were using AIMS when they joined this initiative. Both states expended tremendous time and resources to develop the AICS model, but they faced significant reliability problems because the system relied on multiple checklists composed of numerous subjective assessments of the inmate’s personality. Although the specificity and clarity of AICS appear to have improved and further efforts toward implementation are strongly encouraged, the experiences of these two states suggest that states may struggle to meet the staff training, monitoring, and fiscal resources demanded by personality-based systems. Lesson 3: Internal classification is unique to each system The diversity of models considered and tested by these eight states clearly indicated that, unlike external classification systems, there is no distinct set of internal classification factors that is generalizable across systems. The operational definitions, processes, timing, and other critical factors are unique to each state and depend on its specific goals, resources, and system composition. Even behaviorbased systems, which are clearly the most objective systems available, appear to require significant tailoring and system specificity. For example, the model developed by New Jersey is quite different from that developed by the Illinois Common Problems, Issues, and Solutions Department of Corrections for close-custody inmates. Colorado, on the other hand, developed an automated, relatively simple, behavior-based system for the Limon facility, which still required modification before it could be implemented in a similar Colorado maximum-security facility. In other words, there is no “best model,” nor should there be. As recommended in the literature on external classification systems, the instruments and process must be tailored to and validated on the population for which they will be used. Although internal classification systems should be responsive to individual departments of corrections, Alexander and colleagues identified a set of broad standards that are applicable to all internal classification systems (Alexander et al., 1997: 70–71). These standards are provided in Exhibit 25. There is no “best model,” nor should there be. As recommended in the literature on Exhibit 25. Standards and Guidelines for Internal Prison Classification Systems external classification systems, the instru- 1. The primary objective of the internal prison classification system is to match inmates with institutional level housing, program, and work assignments; these provisions should be consistent with the inmate’s external classification designation. ments and process 2. The internal prison classification system should encourage the classification of inmates in the least restrictive housing and custody levels within a facility. population for which must be tailored to and validated on the they will be used. 3. Internal classification systems should complement existing external classification systems. Consequently, they should not be developed until the department has a well functioning external prison classification system. 4. The internal classification system must have a clearly defined set of measurable goals and objectives prior to implementation. 5. There must be a formal set of written procedures and policies governing the internal classification process. In particular, these policies should include a very structured administrative process and clear operational definitions of how the system will be used to make housing and program/work assignments. 6. The classification process must provide for the collection, automation, and distribution of complete, high-quality, verified data to ensure inmates are housed and programmed according to the internal system. 7. Instruments used in the internal classification system must be objective, reliable, and valid. They should also be designed to be fully automated and easy for staff to use. 79 Chapter 3 Exhibit 25. Standards and Guidelines for Internal Prison Classification Systems (continued) 8. Internal classification systems must allow for the use of overrides (both discretionary and nondiscretionary). As a rule, the discretionary override rate should be in the 5 to 15 percentage range among all internal classification decisions that occur in a given year. 9. Internal classification system must include policies and procedures for screening and evaluating prisoners who are management problems and those who have special needs. In particular, criteria and procedures for assigning inmates to administrative segregation, disciplinary segregation, protective custody, mental health, and medical health care units must be part of the system. 10. There must be provisions to involve the inmate in the internal classification process; the inmate should have input into the classification process (without compromising security) and should receive a copy of his/her internal and external classification assessments. 11. There must be provisions for a systematic review of the internal classification decisions by supervisory staff. 12. The internal classification system should be reviewed annually and formally evaluated every 3 years. 13. Internal classification procedures must be consistent with constitutional requisites. 14. Administration and line staff must be involved in the design and implementation of a classification system and a pilot test of the preliminary system must be conducted prior to implementation. A fundamental Source: Alexander et al. (1997). lesson learned from this initiative was the importance of the clarity of the purpose, goals, and anticipated impacts of the internal classification system. 80 A fundamental lesson learned from this initiative was the importance of the clarity of the purpose, goals, and anticipated impacts of the internal classification system. The expectations and needs of the department (central office), facility administrators, and line staff will determine the specific “make and model” of the system. As illustrated by the steps listed below for designing and implementing an internal classification system, each department of corrections and, in some circumstances, individual facilities, will need to design and refine an internal classification system that will respond to its own unique problems, issues, goals, and resources. Moreover, it is important for each system to implement its model completely to reach definitive conclusions about its effectiveness. In South Dakota and Missouri, for example, housing assignments need to be made according to inmates’ AICS classification. Only then can the true relationship between the classification and occurrence of institutional violence be assessed empirically. Common Problems, Issues, and Solutions Lesson 4: Automation is critical Automation of the internal classification system was a critical step for each state. The Florida, Colorado, and Connecticut systems were designed as automated systems that combined risk, programming, and institutional job assignment data to provide staff with housing and programming recommendations. Information systems and customized computer programs played central roles in the design and testing of these systems, even for small subpopulations within individual facilities. On completion of the design and testing process, Oregon, South Dakota, and Missouri discovered that automation was critical to fully implement the systems. Oregon was able to access local resources to automate its internal classification system; South Dakota and Missouri, however, required additional technical assistance from NIC to fully automate the scoring of AICS. New Jersey’s internal classification system was affected by MIS-related issues in a slightly different way. Full development and implementation were interrupted by the need to update the department’s entire MIS. Fortunately, the variables and modules for implementing its High Aggression Model have been built into NJ DOC’s new information system. Step-by-Step Process for Designing and Implementing an Internal Classification System As previously indicated, the importance of careful planning was not a surprise. One product of the 1993 initiative was a set of strong recommendations and a step-bystep process for designing and implementing an internal classification system (Alexander et al., 1997). These steps were updated based on the experience of the eight states and new insights. Time parameters are provided for completion of the design and implementation process within 1 year. Although the proposed timelines may need to be adjusted to respond to the resources and needs of the state, the project should be completed in a timely manner to ensure that the system will address the goals and systems trends while they are still current. Step 1: Obtain formal commitment from the DOC Central Office (month 1). Step 2: Designate a strong project manager and establish a working steering committee (month 1). The current initiative reinforced the notion that a strong project manager with designated time to conduct the required tasks is critical. Given that most staff are busy with their current workload, the project manager’s normal responsibilities and priorities must be shifted to ensure that the project stays on track. This same prioritization of duties applies to the steering committee and the line staff responsible for data collection. This experience was best illustrated by Oregon; the design and testing of that state’s internal classification system was stalled until an OWCC staff person was reassigned to the classification and transfer unit, where he could devote A strong project manager with designated time to conduct the required tasks is critical. 81 Chapter 3 all of his time to developing and implementing the system. The Colorado staff attributed the success of their initiative to the project manager’s commitment and the representation of all major operational components of the facility and central office on the steering committee. Step 3. Identify stakeholders and organize their inclusion in the process (months 1 and 2). It is important to recognize the potential impact that the The internal classification system, like the external system, affects all operations within a given facility. It is important to recognize the potential impact that the new system will have on the various stakeholders throughout the facility and department. Florida, for example, clearly identified the stakeholders of the Risk and Needs Model and developed the system to respond to their respective goals. Efforts toward inclusion were instrumental in reducing resistance to the implementation of the system. new system will have on the various stakeholders throughout the facility and department. Step 4. Define the problems to be addressed and set realistic goals and measurable objectives (month 2). This is probably the most critical, and most difficult, step to complete. Identification of the precise issues prompting the design or revision of the internal classification system is essential to a successful effort. The members of the steering committee should agree on the relevant issues, the desired outcomes, and how they will be achieved. Although each stakeholder may assign a different priority to identified problems and goals, all members of the steering committee should understand the importance of their contributions and be clear on what is to be accomplished. The importance of this dynamic was highlighted in New Jersey, where the effort to create an internal classification system stalled for several months because of a failure to include all stakeholders in the identification of the problem and goals. These challenges were exacerbated by unforeseen shifts in organizational priorities. The project manager plays an essential role through a thorough knowledge of and commitment to the problems, goals, and objectives identified by the steering committee. Although they did not develop measurable objectives at the outset of their planning process, both Connecticut and Colorado were specific in their desire for an objective, behavior-based system for assigning inmates to work and programming. As their new systems were implemented, these states were able to provide quantitative data to show that they were efficient and effective. Oregon achieved similar clarity in expressing its desire for a system that was responsive to both genders. Data were collected throughout the pilot-test, implementation, and automation process to monitor the system’s impact on key outcome indicators, such as noninitial transfers, major disciplinary reports, and segregation populations. Florida was the only state that outlined specific quantitative, measurable objectives as part of its initial statement of goals and objectives. In contrast, Washington’s process shows how clarity of intended goals for the internal classification system is even more important than quantitative, measurable, objective results. Focusing on both internal classification and inmate management, the internal classification sys- 82 Common Problems, Issues, and Solutions tem was intended to indicate the appropriate housing or programs for an inmate, as well as strategies for behavior management. Step 5. Select the type of internal classification system for adoption (month 2). The shift in the timing for this step is the most significant modification to the earlier prescriptions for designing and implementing an internal classification system. Alexander et al. (1997) suggested that this step should be undertaken only after baseline data collection has been completed. The Phase II initiative revealed, however, that the type of internal classification system to be developed should direct the selection of the pilot site and the collection of baseline data. A thorough review of the literature or technical assistance from a provider with expertise in classification should dictate the type of system to be implemented. As discussed in the previous chapter, seven different models were explored. Six states developed unique models for statewide implementation and one state (New Jersey) developed populationspecific instruments. Clearly, the type of model to be implemented has a tremendous impact on the design of the pilot test and other implementation steps. Moreover, the interplay between goals and research is most relevant here. In Washington, pilot-test results suggested that AIMS accurately identified predatory and nonpredatory inmates but contributed little to the development of individualized case-management strategies. Because AIMS did not further the primary goal of the system, Washington decided to stop using it. Step 6: Select a pilot site (months 2 and 3). The selection of a pilot site depends on the specific problem, goals, and type of system envisioned. For example, New Jersey originally identified the Garden State Correctional Center, a medium-security, treatment-oriented facility, as the pilot site. When the focus of its project shifted to identifying and managing maximumcustody inmates, it recognized that the Northern State Prison was a more appropriate site. Regardless of the unique characteristics or issues to be addressed by the internal system, the pilot site staff must demonstrate the commitment and offer the necessary resources to carry out the project. Once the site is identified, the steering committee and project manager must solicit a solid commitment from the facilitylevel stakeholders, who also must be included on the steering committee. The selection of a pilot site depends on the specific problem, goals, and type of system envisioned. The pilot-test facility for the Washington initiative was selected primarily because of the number of its minimum-custody prisoners with long sentences, the target population for the case-management and transition strategies. Similarly, as seen in Colorado and Connecticut, new facilities can offer an opportunity to develop and implement innovative internal classification procedures. Step 7. Analyze current housing, work, and program assignment procedures (months 3 and 4). A baseline description of the facility’s current housing units, work and program assignments, and external and internal classification processes is the foundation of 83 Chapter 3 Baseline data should be collected on each outcome indicator. sound planning. These data permit an understanding of the scope of the new internal classification system. Moreover, baseline data should be collected on each outcome indicator identified by the steering committee and stakeholders (such as rates of institutional violence and number of housing transfers). These baseline indicators are used to highlight the impact of the new or revised system on the facility. If both software and expertise are available, a “map” of the various cell blocks, housing units, and program assignments will illuminate any questionable or inappropriate patterns to be addressed by the new system. Suspect patterns include disproportionate numbers of gang members assigned to the same cell blocks, special-management inmates housed in general population, maximum- and minimum-custody inmates housed together in the same cell, and inmates placed in programs they do not need. Extensive baseline data illustrating work and program assignments as well as offender needs were required for the Florida, Connecticut, and Colorado initiatives to incorporate the eligibility requirements into the automated computer program. The steering committees focused on the operation of the new system and ensured that policies and eligibility criteria were updated to support its goals and objectives. Step 8. Conduct facility program and work assignment inventory (months 4 and 5). A system may also benefit from an inventory of the facility’s program and work assignments. The utility of such an inventory depends on the type and purpose of the internal classification system to be developed. For systems designed to assign inmates to work and program slots (e.g., Colorado and Connecticut), this step is essential. Florida’s initiative also required a detailed database that identified the admission criteria, number of slots, location, pay schedule, and other factors for all programs across all FL DOC facilities. In contrast, because Oregon’s internal classification systems did not affect work or program assignments, a detailed inventory was not required for implementation. Step 9: Conduct ongoing inventory of available bed space (months 4 and 5). The need for an inventory of the facility’s available beds was echoed across states, although at different points and for different reasons. Oregon, for example, struggled with bed space restrictions throughout the project. Not only was bed space identified as a consideration for the type of system to be developed, it was a significant barrier during the pilot tests at OWCC and OSCI and remains a barrier for full implementation at OSCI. Similarly, New Jersey was forced to modify its implementation plan because of the unmanageable costs associated with shifting populations, upgrading security, and training staff to implement the high-aggression model solely within the Northern State facility. Instead, New Jersey was forced to decentralize its high-aggression unit across four facilities. 84 Common Problems, Issues, and Solutions Step 10. Develop prototype instruments, policies, and procedure manuals (months 6 and 7). Drafting new policies and procedures was a significant challenge for seven states, requiring multiple revisions and multiple pilot tests. (Washington was an exception only because it opted not to implement AIMS.) States should be prepared for this step to exceed original deadlines. With the exception of Florida, Phase II states still consider their current systems to be a “work in progress” and expect further refinement and specification. States undertaking internal classification system initiatives should not be discouraged if multiple prototypes are discarded; modifications are an important part of the process. If the states were polled, they would advise, “Keep it simple.” Complex instruments or those with many variables do not provide adequate flexibility for making housing, work, or program decisions within overcrowded, understaffed facilities. In addition, they require significant supporting documentation and staff training for administration and scoring, which can prove to be both cumbersome and expensive. Modifications are an important part of the process. Step 11. Pilot-test procedures and instruments (months 7 through 9). Once the prototype instruments, policies, and procedures were created, pilot tests using a random sample of inmates highlighted the elements that needed further modification and provided a preliminary assessment of the instruments’ impact on the larger correctional system. A systematic random-sampling scheme should be used, unless the facility population is less than 250 inmates. This methodology entails selecting every nth case from a current list of the facility’s population to reach a total of 250 cases. For example, if the facility population is 750, every third case should be assigned to the pilot-test sample. The entire inmate population should be sampled, including the special-management populations. Proper sampling permits an estimate of the distribution of classification levels across the entire population and highlights housing and program needs. Conclusions about the accuracy and impact of the system cannot be made without confidence in the reliability of the instruments. As Missouri and South Dakota learned, the low rates of inter-rater reliability called the overall feasibility of the system into question. In Oregon and Florida, ongoing pilot tests triggered further refinement of the instruments. Empirical testing can also identify ineffective and inefficient processes, triggering more revisions to the overall strategy. In Washington, the pilot test of AIMS demonstrated that AIMS duplicated current case-management strategies and thus, its implementation was not warranted. Step 12. Develop a full implementation plan (month 10). Once the system has been refined to achieve appropriate levels of validity, reliability, and feasibility, full implementation can proceed. A detailed time-task chart that accounts for the complexity of implementation across facilities and staffing patterns must be developed. The importance of staff training and automation cannot be 85 Chapter 3 Regardless of how simple the instruments are or how few criteria are built into the model, ultimately, automation is essential. overemphasized. Training is key to achieving reliability, especially as new staff are required to use the system. Regardless of how simple the instruments are or how few criteria are built into the model, ultimately, automation is essential. For example, in Oregon, upon implementation of even a relatively simple model (with only four factors), managing and tracking the system for just the female population has proven to be a full-time job. Full implementation will bring additional challenges and obstacles and may even require additional modifications to the instruments. Such challenges should be expected, and time and resources should be built into the plan so that obstacles can be appropriately addressed. One barrier encountered by South Dakota and Missouri was that the initial plan for the development and implementation of AICS did not fully account for the reliability problems encountered during the pilot tests. This delayed the development of the data, policies, and automation that would have enabled the departments to shift from AIMS to AICS. In contrast, Colorado staff attributed their success to the careful planning and detailed implementation processes undertaken by the steering committee. The implementation of the model provided the opportunity “to see the model in action” and to test its real-world application to the correctional facility. Step 13. Monitor and evaluate (ongoing). Ongoing monitoring and feedback are critical to implementing a new system in a complex and dynamic correctional setting. A monthly reporting format should be developed that provides easy access to management information. Regular reports will alert classification staff to problems with the system and will also signal an appropriate time for revalidation of the system. Florida’s next step, for example, is to develop a portfolio of user-friendly management reports and statistics for monitoring the system. Validity and reliability studies have been recommended to seven states to ensure that each system operates properly once fully implemented. The details available in these management reports may also improve corrections departments’ ability to secure additional resources for population management. From the outset of the pilot test, Oregon was the most diligent at collecting and tracking quantitative data for monitoring the impact of the internal classification system at EOCI and OWCC. Florida was also able to generate comprehensive data to assess the success with which each goal and objective was achieved. Implications and Future Steps Researchers are always calling for more research, more planning, and more data. Thus, suggestions for planning, technical assistance to states, and additional analyses sound redundant and trite. Yet the continued challenges faced by the eight states involved in this initiative and the need for specific onsite expertise and complex data analyses testify to the practical utility of this suggestion. 86 Common Problems, Issues, and Solutions The diversity of facilities, populations, factors, and models suggests that much remains to be learned about internal classification. At the same time, state and local correctional and detention systems continue to face growing inmate populations and declining resources. Development of new management techniques for controlling and servicing the prison population with fewer resources becomes more critical with each new admission. Future technical assistance efforts should focus on assisting states in developing systems that are both practical and feasible, given these harsh realities. Just as researchers dependably call for more research, correctional administrators are consistently asked to “do more with less.” Scarce resources should provide maximum returns, so future initiatives should concentrate on models that require reasonable efforts in terms of staff training, validation, and implementation. The diversity of facilities, populations, factors, and models suggests that much remains to be learned about internal classification. In response to this call, NIC has indicated a commitment to continue supporting the development of comprehensive classification systems. NIC continues to provide long- and short-term assistance to state correctional agencies to— Update and improve objective classification systems. Enhance the capacity of correctional agencies to enable continued improvements. Validate and revalidate classification instruments. Expand the knowledge of research and strategies for appropriate and effective correctional classification practices that respond to emerging issues and management concerns. Encourage the development of comprehensive objective classification systems. With these future initiatives, it is anticipated that new lessons will be learned and previous lessons will be further refined. As shown in previous initiatives, although internal classification is critical for the efficient, safe, and orderly management of the system, the design and implementation of a valid and reliable model is both difficult and time consuming. Although most systems attempt to answer the seemingly simple question, “Where should I put this inmate?” the complex interplay of goals, factors and criteria, resources, and outcomes challenges even the most sophisticated systems. 87 Notes 1. AIMS has been subject to only limited evaluation (Quay, 1984). Preliminary results show that facilities using AIMS experienced significantly lower rates of staff and inmate assaults than facilities not using an internal classification system. In South Carolina, serious disciplinary incidents were reduced by 18 percent after the implementation of AIMS. 2. PMC has been rigorously evaluated and has been shown to be helpful in identifying and separating potential victims from potential predators. Outcome research indicates that separating inmates by type and the subsequent differential supervision of inmate types resulted in decreased institutional misconduct and increased staff and inmate safety. 3. States were required to submit with their application copies of their current external classification forms (initial, reclassification, and needs assessment instruments) and manual as well as written documentation of their current internal classification process, if available. 4. As of July 1999, Connecticut had 20 correctional facilities. 5. Connecticut Department of Corrections. (April, 1995). Statistical Reports. Hartford, CT: Research Division. 6. The Marquiz settlement was the result of a civil action suit by Limon Correctional Facility inmates following a series of deaths at the facility in 1992. 7. Sex offenders are housed in all units, but their lifestyles and behavior tend to place them in the Beta units. 8. This description of the OR DOC External Classification System was based, in part, upon OR DOC Administrative Rules 291–104–005 through 291–104–035, effective October 1, 1994, and interviews with OR DOC Classification and Transfer Unit Staff. 9. OR DOC Administrative Rule 291–104, p.1. 10. At the same time, OR DOC was in the process of developing an incarceration plan system that would identify and track inmates’ participation in treatment and vocational and institutional work. Upon admission to OR DOC, the inmates were placed in level one. Based upon program compliance and institutional record, they could progress to level six. Inmates’ privileges and assignment to preferred facilities, housing units, and work assignments were linked to their incentive level. 11. The internal classification initiative stakeholders were identified as the staff from institutional programs, inmate labor office, security, security threat group, probation and parole, field, and classification staff. 89 Notes 12. To date, the FL DOC external classification system has not been revalidated. 13. For a full description of the definition and scoring for each of the 16 risk and need factors, see Florida Department of Corrections (1998). 14. For purposes of the Risk and Needs System, FL DOC programs were categorized as “primary” or “other” based on their potential to aid offenders in avoiding substance abuse and reintegrating successfully into the community. Substance abuse treatment, academic, special education, and vocational education were identified as “primary” programs. “Other” programs included, but were not limited to, wellness, life skills, and transition. 15. PRIDE stands for the FL DOC Prison Rehabilitative Industries and Diversified Enterprises or PIE (Prison Industries Enhancement) Program. The criteria assesses the inmate’s need for new or enhanced training skills. 16. Both CCI classification and security staff participated in the hands-on pilot test. The debriefing sessions included the facility classification staff, central office classification staff, facility executive staff, facility security staff, and computer technical staff. 17. Institutional Disruption Index (INDI) represents the inmate’s involvement in institutional violence, supervision risk, and major and minor infractions. 18. The difference between the category “did not meet bed type” and “no bed type” was that the second category, “no bed type,” included placements that did not meet the recommendation because the recommended bed type was not available within the FL DOC system. 19. The seven facilities and their respective locations are South Dakota State Penitentiary, Sioux Falls, SD; Jameson Annex, Sioux Falls, SD; Springfield State Prison, Springfield, SD; Salem Public Safety Center, Pierre, SD; Yankton Trusty Unit, Yankton, SD; Redfield Trusty Unit, Redfield, SD; and Pheasantland Industries, Sioux Falls, SD. 20. Herbert Quay (1984) described how AIMS was created, how to implement it, and what effect it had on the Federal Bureau of Prisons, and Patricia Van Voorhis (1994) compared the reliability and validity of five classification systems, including AIMS. 21. The Missouri Steering Committee included Jean Ann Johnson, associate superintendent of inmate management, Crossroads; Terry Page, assistant superintendent of inmate management, Crossroads; Mike Kemna, superintendent, Crossroads; Nick Noll, chief psychologist, Crossroads, Correctional Medical Services; and Amy Gertz, caseworker 2, Crossroads. The South Dakota Steering Committee was composed of Ed Ligtenberg, director of classification, transfer, and scheduling; Laurie Feiler, director of noninstitutional programs/planning; Rick Leslie, case 90 Internal Prison Classification Systems manager, Penitentiary Complex; Dennis Block, unit manager, Penitentiary Complex; and Jorgene Williams, transfer coordinator. 22. Further analyses of the Missouri data indicated that where there were disagreements in the final classification type between the two raters, the inconsistencies were never Type 1 versus Type 3, the types that must be housed separately. There were two such inconsistencies among the SD DOC cases, however. 23. ACCESS software, with the automated instruments and monitoring reports, is available from the National Institute of Corrections. 91 References Alexander, J., J. Austin, S. Brown, L. Chan, S. He, and P. Stokes. 1997. Internal Prison Classification Systems: A Field Test of Three Approaches. San Francisco, CA: National Council on Crime and Delinquency. Austin, J., J. Alexander, S. Anuskiewicz, and L. Chin. 1996. Evaluation of Colorado Department of Corrections Classification System. San Francisco, CA: National Council on Crime and Delinquency. Colorado Department of Corrections. 1995. Electronic statistical file, August 28, 1995. Colorado Springs, CO: Colorado Department of Corrections, Research Division. ———. 1999. Web site: www.doc.state.co.us/statistics.htm. Accessed in May 1999. ———. 2000. “Profile Data—Limon Correctional Facility.” March 24, 2000. Colorado Springs, CO: Colorado Department of Corrections, Office of Planning and Analysis. Connecticut Department of Correction. 1999. Objective Classification Manual 1999. Suffield, CT: Connecticut Department of Correction, Classification and Population Management Unit. Florida Department of Corrections. 1998. Risk and Needs System Guide. Tallahassee, FL: Office of Security and Institutional Management, Bureau of Inmate Classification and Management. ———. 1999. Web site www.dc.state.fl.us. Accessed in April 1999. Lerner, K., G. Arling, and S.C. Baird. 1986. “Client Management Classification Strategies for Case Supervision.” Crime and Delinquency 32 (3): 254–71. Limon Correctional Facility. 1999. “Internal Classification.” Operational memorandum, August 15, 1999. Limon, CO. MacDougall Correctional Institution (MacDougall). 2000. Internal Classification Manual. Suffield, CT: Connecticut Department of Correction, Classification and Population Management Unit. Mande, Mary J. 1996. Inmate Classification Study. Boulder, CO: MJM Consulting Services. Funded by the National Institute of Corrections. Megargee, E.I., and M.J. Bohn, Jr., with J. Meyer, Jr., and F. Sink.1979. Classifying Criminal Offenders: A New System Based on the MMPI. Beverly Hills, CA: Sage Publications. Missouri Department of Corrections. 1999. Web site: www.corrections.state.mo.us. Accessed in April 1999. 93 References New Jersey Department of Corrections. 1997. Report on the Implementation of the Objective Classification System in the New Jersey Department of Corrections. Trenton, NJ: New Jersey Department of Corrections, Policy and Planning Division. Oregon Department of Corrections. 1998. Quick Facts. Salem, OR: Oregon Department of Corrections, Office of Public Affairs. Quay, H. 1984. Managing Adult Inmates: Classification for Housing and Program Assignments. College Park, MD: American Correctional Association. Soares, R.A. 1999. “Marquiz Settlement 9–K–1470, May 20, 1997–May 20, 1999, Reporting Period 12/98B–02/99.” Limon, CO: Limon Correctional Facility. South Dakota Department of Corrections. 1998. “Department of Corrections: Monthly Statistics Report—November 1998.” Online at www.state.sd.us/corrections/doc.htm. Accessed on November 30, 1998. Van Voorhis, P. 1994. Psychological Classification of the Adult Male Prison Inmate. Albany, NY: State University of New York Press. Washington Department of Corrections. 1995. Electronic statistical file. Spokane, WA: Washington Department of Corrections, Research Division. ———. 1996. “Classification Policy No. 300.380.” Olympia, WA: Washington Department of Corrections, Secretary of the Department. ———. 2000. Web site www.wa.gov/doc/content/secstatistics.htm. Accessed in May 2000. 94 Suggested Readings Alexander, J. 1986. “Classification Objectives and Practices.” Crime and Delinquency 32(3): 323–38. Alexander, J., and J. Austin. 1992. Handbook for Evaluating Objective Prison Classification Systems. Washington, DC: U.S. Department of Justice, National Institute of Corrections. NIC 010675. Austin, J. 1986. “Evaluating How Well Your Classification System Is Operating: A Practical Approach.” Crime and Delinquency 32 (3): 302–22. Austin, J., D. Holien, L. Chan, and C. Baird. 1989. Reducing Prison Violence by More Effective Inmate Management: An Experimental Field Test of the Prisoner Management Classification (PMC) System. San Francisco, CA: National Council on Crime and Delinquency. Buchanan, R., K. Whitlow, and J. Austin. 1986. “National Evaluation of Objective Prison Classification Systems: The Current State of the Art.” Crime and Delinquency 32 (3): 272–90. Chaiken, J., M. Chaiken, and W. Rhodes. 1993. “Predicting Violent Behavior and Classifying Violent Offenders.” In Understanding and Preventing Violence, vol. 4. Consequences and Control. Ed. A. Reiss and J. Roth. Washington, DC: National Academy Press: 217–95. Clements, C. 1982. “The Relationship of Offender Classification to the Problems of Prison Overcrowding.” Crime and Delinquency 28 (1): 72–81. Cohen, R.L. 1991. Prisoners in 1990. Washington, DC: U.S. Department of Justice, Bureau of Justice Statistics. May. NCJ 129198. Cowden, J. and Durkin, G. (Summary, 1984). “Development and validation of differential risk prediction scales: an enhancement of the original Wisconsin risk/needs classification system.” Cowden, J. (Summary, 1984). “Development and validation of risk prediction and needs assessment scales for juvenile offenders in Wisconsin.” Deemer, W., and Conine, A. 1984. “An Institutional Programming Model for the Adult Developmentally Disabled Offender.” Corrections Today June: 30–31. Evatt, P., S. Brown, and L. Fowler. 1989. “Offender Classification: Don’t Overlook This Important Security Strategy.” Corrections Today July: 34–37. Hardyman, P., J. Austin, K. Dedel, J. Alexander, and J. Aiken. 1999. “Where Do I Put This Inmate?” Designing and Pilot Testing Internal Classification Systems in Five States. Washington, DC: The Institute on Crime, Justice and Corrections at The George Washington University. 95 Suggested Readings Hare, R. 1996. “Psychopathy: A Clinical Construct Whose Time Has Come.” Criminal Justice and Behavior 23 (1): 25–54. Hart, S., R. Hare, and A. Forth. 1996. “Psychopathy as a Risk Marker for Violence: Development and Validation of a Screening Version of the Revised Psychopathy Checklist.” In Violence and Mental Disorder: Developments in Risk Assessment, ed. J. Monahan and H. Steadman. Chicago: University of Chicago Press: 81–98. Hepburn, J., and L. Goodstein. 1986. “Organizational Imperatives and Sentencing Reform Implementation: The Impact of Prison Practices and Priorities on the Attainment of the Objective of Determinate Sentencing.” Crime and Delinquency 32(3): 339–65. Herrick, E. 1989. “The Surprising Direction of Violence in Prison.” Corrections Compendium 14 (6):1, 4–17. Innes, C.A. 1988. Profile of State Prison Inmates, 1986. Special Report. Washington, DC: U.S. Department of Justice, Bureau of Justice Statistics. January. NCJ 109926. Innes, C.A., and L.A. Greenfield. 1990. Violent State Prisoners and Their Victims. Special Report. Washington, DC: U.S. Department of Justice, Bureau of Justice Statistics. July. NCJ 124133 Jacobsen, V. 1983. “Industrial Unit Management.” Corrections Today August: 88–91, 108. Kane, T. 1986. “The Validity of Prison Classification: An Introduction to Practical Consideration and Research Issues.” Crime and Delinquency 32 (3): 367–90. Lansing, D., J. Bogan, and L. Karacki. 1977. “Unit Management: Implementing a Different Correctional Approach.” Federal Probation 41 (1): 43–49. Leeke, W., and H. Mohn. 1986. “Violent Offenders.” Corrections Today May: 22–24. Levinson, R. and J.D. Williams. 1980. “Inmate Classification: Security/Custody Considerations.” Federal Probation 43 (1): 37–43. Levinson, R., and R. Gerard. 1986. “Classifying Institutions.” Crime and Delinquency 32 (3): 291–301. Levinson, R. 1988. “Developments in the Classification Process: Quay’s AIMS Approach.” Criminal Justice and Behavior 15 (1): 24–38. ———. 1999. “Internal Classification and Unit Management,” in R. Levinson, Unit Management in Prisons and Jails. Lanham, MD: American Correctional Association: 75–95. 96 Internal Prison Classification Systems McCorkle, R.C. 1993. “Fear of Victimization and Symptoms of Psychopathology Among Prison Inmates.” Journal of Offender Rehabilitation 19 (1–2): 27–41. Maitland, A., and R.D. Sluder. 1998. “Victimization and Youthful Prison Inmates: An Empirical Analysis.” The Prison Journal 78 (1): 55–73. Morris, N. and D. Rothman, eds. 1995. Oxford History of the Prison: The Practice of Punishment in Western Society. New York: Oxford University Press. National Council on Crime and Delinquency. 1986. Crime and Delinquency. Special Edition: Prison Classification Systems 32 (3). ———. 1992. Criminal Justice Sentencing Policy Statement. San Francisco, CA: National Council on Crime and Delinquency. O’Donnell, I. and K. Edgar. 1998. “Assault in Prison: The ‘Victim’s’ Contribution.” The British Journal of Criminology 38 (4): 635–50. Quay, H.C. 1973. An Empirical Approach to the Differential Behavioral Assessment of the Adult Offender. Washington, DC: U.S. Department of Justice, Federal Bureau of Prisons. ———. 1983. Behavioral Classification System of Adult Offenders: Technical Manual and Administrator’s and User’s Manual. August. Washington, DC: U.S. Department of Justice, National Institute of Corrections. NIC 000151 Quay, H.C., and C. Love. 1989. Behavioral Classification for Female Offenders. Research Report. July. Washington, DC: U.S. Department of Justice, National Institute of Corrections. NIC 007804 Smith, W.A. and C.E. Fenton. 1978. “Unit Management in a Penitentiary: A Practical Experience.” Federal Probation 42 (3): 40–46. Solomon, L. 1980. “Developing An Empirically Based Model for Classification Decision-Making.” Prison Law Monitor, 2 (9): 234–37. Toch, H. 1992. Living in Prison: The Ecology of Survival, rev. ed. Washington, DC: American Psychological Association. Woodredge, J.D. 1998. “Inmate Lifestyles and Opportunities for Victimization.” Journal of Research in Crime and Delinquency 35 (4): 480–501. Wright, K. 1991. “A Study of Individual, Environmental, and Interactive Effects in Explaining Adjustment to Prison.” Justice Quarterly 8: 217–42. ———.1991b. “The Violent and Victimized in the Male Prison.” Journal of Offender Rehabilitation 16 (3–4): 1–25. 97 Appendix State of Missouri Department of Corrections Checklist for the Analysis of Life History Records of Adult Offenders (CALH) . . . . . . . . . . .101 State of Missouri Department of Corrections Correctional Adjustment Checklist . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .102 IDOC Internal Classification Form: Maximum Security Facilities . . . . . . . . .103 State of Missouri Department of Corrections Checklist for the Analysis of Life History Records of Adult Offenders (CALH) INMATE NAME DOC NUMBER CORRECTIONAL OFFICER DATE COMPLETED INSTRUCTIONS: Place a checkmark before each behavior trait that describes this inmate’s life history. 1. Has few, if any friends. 2. Thrill-seeking. 3. Preoccupied; “dreamy.” 4. Uncontrollable as a child. 5. Has expressed guilt over offense. 6. Expresses need for self improvement. 7. Socially withdrawn. 8. Weak, indecisive, easily led. 9. Previous local, state or federal incarceration. 10. Tough, defiant. 11. Irregular work history (if not a student). 12. Noted to be not responsive to counseling. 13. Gives impression of ineptness, incompetence in managing everyday problems in living. 14. Supported wife and children. 15. Claims offense was motivated by family problems. 16. Close ties with criminal elements. 17. Depressed, morose. 18. Physically aggressive (strong arm, assault, reckless homicide, attempted murder, mugging, etc.). 19. Apprehension likely due to “stupid” behavior on the part of the offender. 20. Single marriage. 21. Expresses feelings of inadequacy, worthlessness. 22. Difficulties in public schools. 23. Suffered financial reverses prior to commission of offenses for which incarcerated. 24. Passive, submissive. 25. Bravado, braggart. 26. Guiltless, blames others. 27. Expresses lack of concern for others. 101 State of Missouri Department of Corrections Correctional Adjustment Checklist INMATE NAME DOC NUMBER CORRECTIONAL OFFICER DATE COMPLETED Please indicate which of the following behaviors this inmate exhibits. If the behavior describes the inmate, circle “1.” If it does not, circle “0.” PLEASE COMPLETE EVERY ITEM. 102 0 1 1. Worried, anxious. 0 1 25. Passive, easily led. 0 1 2. Tries, but cannot seem to follow directions. 0 1 26. Talks aggressively to other inmates. 0 1 3. Tense, unable to relax. 0 1 27. Accepts no blame for any of his trouble. 0 1 4. Socially withdrawn. 0 1 28. Continually complains; accuses staff of unfairness. 0 1 5. Continually asks for help from staff. 0 1 29. Daydreams; seems to be mentally off in space. 0 1 6. Gets along with the hoods. 0 1 30. Talks aggressively to staff. 0 1 7. Seems to take no pleasure in anything. 0 1 31. Has a quick temper. 0 1 8. Jittery, jumpy; seems afraid. 0 1 32. Obviously holds grudges; seeks to “get even.” 0 1 9. Uses leisure time to cause trouble. 0 1 33. Inattentive; seems preoccupied. 0 1 10. Continually uses profane language; Curses and swears. 0 1 34. Attempts to play staff against one another. 0 1 11. Easily upset. 0 1 35. Passively resistant; has to be forced to participate. 0 1 12. Sluggish and drowsy. 0 1 36. Tries to form a clique. 0 1 13. Cannot be trusted at all. 0 1 37. Openly defies regulations and rules. 0 1 14. Moody, brooding. 0 1 38. Often sad and depressed. 0 1 15. Needs constant supervision. 0 1 39. Stirs up trouble among inmates. 0 1 16. Victimizes weaker inmates. 0 1 40. Aiding or abetting in breaking rules. 0 1 17. Seems dull and unintelligent. 0 1 41. Considers himself unjustly confined. 0 1 18. Is an agitator about race. 0 1 19. Continually tries to con staff. 0 1 20. Impulsive; unpredictable. 0 1 21. Afraid of other inmates. 0 1 22. Seems to seek excitement. 0 1 23. Never seems happy. 0 1 24. Doesn’t trust staff. IDOC Internal Classification Form Maximum Security Facilities I BACKGROUND DATA / / / Current Facility Interview Date Inmate’s Name IDOC # Primary Offense Class Sentence Length IDOC Admit Date / / / MSR/MAX Date Transferred From Transfer Date Current Age Transfer Reason Disciplinary Inmate Request Medical Keep Separate Initial Class PC to General Pop Other Type of Classification Review Initial Reclass Current Classification/Supervision and Disciplinary Grade Information (Circle the current designation for each item below.) Escape Level L M H EXH Security MN MD MX Grade A B C Interview OTS Other Enemies checked by Number of known enemies at correctional center If there are known enemies at this correctional center, please list below. 1. 4. Name IDOC# Name 5. 2. Name IDOC# Name 6. 3. Name IDOC# Name II / / / / / / IDOC# IDOC# IDOC# CRIMINAL AND BACKGROUND FACTORS 1. Current Security Level Max = 3 pts Med = 1pt Min = 0 pts 2. Current Offense Involved Violence Against Others (If yes, and time served is less than 1 year, score 6pts. If yes and time served is 1 year or more score 2pts. If no, score 0 pts.) 3. Current Detainer/Warrant for Violence Against Others (Do not score convictions scored in item 2 above. Yes = 1 pt No = 0 pts 4. Current Age Under 23 = 3 pts; 23 to 25 = 2 pts; 26 to 30 = 1pt; 31+ = 0 pts SECTION II SUBTOTAL 103 III INSTITUTIONAL ADJUSTMENT FACTORS A. Failure to Adjust to Min/Med Facility During Past Two Years (If involved in assault, score 3 pts. If involved in fighting, score 2pts. Otherwise, score 1 pt. If no failure, score 0pts.) B. Segregation to Segregation Transfers During Past Two Years (If yes, score 5pts. If no, score 0 pts.) C. Maximum Security Segregation Placements Past Two Years (Indicate whether a segregation placement has occurred for the offenses listed in the table below. Calculate total points using the appropriate weights for each offense and time frame shown in each cell of the table below.) ADMINISTRATIVE SEGREGATION DAYS 30 to 90 days 90 to 180 days 181+ days Incident Time Frame Last 12 mo. 13-24 mo. Last 12 mo. 13-24 mo. Last 12 mo. 13-24 mo. Staff Assault 6 4 8 6 10 8 Inmate Assault 6 4 8 6 10 8 Dangerous Disturbance 6 4 8 6 10 8 Fighting 6 4 NA NA NA NA Threat or Intimidation 6 4 8 6 NA NA Dangerous Contraband 6 4 8 6 10 8 Violent State/Fed Violations 6 4 8 6 10 8 Arson 6 4 8 6 10 8 Gang Activity 6 4 8 6 NA NA TOTAL POINTS SECTION III SUBTOTAL TOTAL SCORE (Sections II, III) IV 104 ASSESSED CLASSIFICATION LEVEL A. Scored Aggression Level (Circle one) B. Recommended Aggression Level (Circle one) C. If different from scored level, reasons if any: Low 0-5 pts Low 0-5 pts D. E. F. G. Screened by Reviewed by Concur Nonconcur If different from recommendations, reasons if any: H. I. Final Aggression Level Designation Next Scheduled Review Date / LOW / / Moderate 6-11 pts Moderate 6-11 pts High 12+ pts High 12+ pts Date Date MODERATE HIGH / / / / / / User Feedback Form Please complete and return this form to assist the National Institute of Corrections in assessing the value and utility of its publications. Detach from the document and mail to: Publications Feedback National Institute of Corrections 320 First Street N.W. Washington, DC 20534 1. What is your general reaction to this document? ______Excellent ______Good ______Average ______Poor ______Useless 2. To what extent do you see the document as being useful in terms of: Useful Of some use Not useful Providing new or important information Developing or implementing new programs Modifying existing programs Administering ongoing programs Providing appropriate liaisons 3. Do you believe that more should be done in this subject area? If so, please specify the types of assistance needed.____________________________________________________________________ 4. In what ways could this document be improved? ________________________________________________ 5. How did this document come to your attention? ____________________________________________ 6. How are you planning to use the information contained in this document? __________________________ 7. Please check one item that best describes your affiliation with corrections or criminal justice. If a governmental program, please also indicate the level of government. _____ Citizen group _____ College/University _____ Community corrections _____ Court _____ Department of corrections or prison _____ Jail _____ Juvenile justice _____ Legislative body _____ Parole _____ Police _____ Probation _____ Professional organization _____ Other government agency _____ Other (please specify) 8. Optional: Name: ____________________________________________________________________________ Agency: ____________________________________________________________________________ Address: __________________________________________________________________________ Telephone:__________________________________________________________________________ Internal Prison Classification Systems: Case Studies in Their Development and Implementation