Abstract
A signature feature of the get-tough era in American corrections has been the proliferation of restrictive housing (RH). Although sometimes equated with solitary confinement, this housing encompasses a variety of distinct forms of incarceration. They are unified by an emphasis on restricted movement and privileges—yet vary in their design and uses. Despite that fact, little is known about the prevalence of different forms of housing. To address this research gap, we use a case study of Florida policy and administrative records data to illuminate the variety of RH types and the varying prevalence of each. We then discuss the implications of the findings for the study of RH uses and impacts and for policy.
Recent decades have been witness to a dramatic turn toward more severe punishment. That can be seen in the rise of what has been termed “mass incarceration” (Pratt, 2018; Travis et al., 2014), but it can also be seen in the proliferation of restrictive housing (RH; Beck, 2015; Garcia, 2016). Definitions and precise prison system terms for different forms of RH vary. By most accounts, this housing goes beyond traditional maximum-security housing in having a variety of goals, such as punishment for infractions or management of individuals deemed to be especially violent or disruptive. It can also impose a range of restrictions, including the potential for single-cell confinement, being confined to a cell for up to 24 hours per day, and having few privileges (Labrecque et al., 2021; Mears et al., 2019; Shalev, 2009). The United States has always relied on different types of what might be called RH (N. Morris & Rothman, 1995; Pizarro & Stenius, 2004; Riveland, 1999). However, the seeming ubiquity of this phenomenon in contemporary times has led to increased concerns about its use, including whether it is morally or legally appropriate, its possible harms to the mental and physical well-being of those incarcerated in this housing and its impacts on successful re-entry (Collins, 2004; Frost & Monteiro, 2016; Labrecque & Mears, 2019; U.S. Department of Justice, 2016; Wildeman & Andersen, 2020).
One question, though, centers on the extent to which RH in fact is used. Beginning in the 1980s, increased attention focused on estimating the prevalence of super-maximum, or “supermax,” incarceration, which can be viewed as one type of RH (Mears, 2016; National Institute of Corrections, 1997; Riveland, 1999). Such work focused primarily on individuals who might serve time in isolation for extended periods for management purposes. In the last decade, increased attention has turned to a wider range of specialized housing that may or may not include solitary confinement—but that relies on a myriad of restrictions. This has resulted in attention to prison cells that are described with inconsistent terminology, but that regardless of terminology, are generally used for protection, punishment, or administrative purposes (Shalev, 2009). It has led to a focus on factors that may predict placement in this housing as well, though studies primarily focus on long-term segregation (e.g., Mears et al., 2021) or short-term disciplinary confinement (DC; for example, R. G. Morris, 2016). Not least, it has led to varying estimates of the extent of RH (e.g., Beck, 2015; Correctional Leaders Association and the Arthur Liman Center for Public Interest Law [CLA-Liman], 2020; Mears, 2016). These estimates, however, generally do not distinguish between types of RH or, by extension, provide details about each type.
Accordingly, this study seeks to respond to scholars’ calls to examine the contours of the get-tough era in contemporary corrections and, in particular, to illuminate the range of RH and to understand its nature and prevalence (see, generally, Frost & Monteiro, 2016; Labrecque et al., 2021; Mears et al., 2019; Naday et al., 2008). Shedding light on the “dark figure” of an extreme housing that has garnered considerable critique can help to highlight prison system reliance on diverse types of RH, the risks of generalizing from studies of one housing type to another, and policy considerations that may attend to the housing. Critical to such an undertaking is a need to directly engage with characterizing unique types of RH and their function in prison systems. To this end, we discuss prior work and the risks of failing to distinguish types of RH and the extent of their use. We then provide a case study of Florida’s prison system, drawing on a discussion of the State’s administrative policies and on analyses of administrative records data, to highlight variation in the types of RH that exist and in the extent to which they are used. We then conclude by discussing implications of the findings for evaluating scholarship on RH and continuing policy debates about its use.
Background
Rise of RH
The “get tough” era in criminal justice policy led to increasingly punitive responses to criminal offending, with an attendant emphasis on punitive, control-oriented prison management practices (Pratt, 2018; Travis et al., 2014). Concerns about potential harms of this shift are evident in research on long-term solitary confinement, which has highlighted a range of concerns centering on the notion that the housing—typically justified as necessary to maintain order and safety in prison systems—is inhumane, adversely affects mental health, and may worsen behavior as well as re-entry outcomes (Collins, 2004; King, 1999; Kurki & Morris, 2001; Labrecque & Smith, 2019b; Mears, 2016; Richards, 2015; Riveland, 1999; Smith, 2006). Early research focused on the lockdown of the federal penitentiary in Marion, Illinois in 1983, and the subsequent construction of super-maximum prisons across the country (see, for example, King, 1999; Toch, 2003; Ward & Werlich, 2003). These facilities symbolized a return to total isolation of individuals from the community and fellow incarcerated persons in a manner akin to the model of confinement of early American prisons such as Pennsylvania’s Eastern State Penitentiary (N. Morris & Rothman, 1995; Pizarro & Stenius, 2004).
More recently, scholars and critics of long-term solitary confinement have expanded their focus to include any of a range of types of incarceration—collectively termed “restrictive housing”—that impose restrictions beyond what typically would occur with traditional maximum-security custody units (Garcia, 2016). This attention has had the benefit of drawing attention to more than long-term solitary “supermax” housing. It has, for example, led to research on short-term disciplinary segregation, including its effects (e.g., R. G. Morris, 2016) and factors associated with placements into it (e.g., Cochran et al., 2018). At the same time, the terminology risks obscuring important distinctions among types of housing, each of which may vary in their purposes, durations, and levels of isolation and restrictions (Labrecque et al., 2021; Mears et al., 2019). Shalev (2009) and others (CLA-Liman, 2020), for example, have highlighted that there is RH used for protecting incarcerated persons at risk of victimization, other RH used for punishing individuals who violate prison rules, and still other housing used for administrative purposes, such as separating an individual pending investigation of a potential need for protective housing or of involvement in a rule violation.
This situation presents several problems, such as the risk of misgeneralizing about a study of one type of RH to another (Labrecque et al., 2021; Mears et al., 2019). If, for example, a study finds that DC occurs frequently, it would not mean that long-term solitary confinement for management purposes occurs as often. Debates about solitary confinement in particular risk equating housing that may be dissimilar across a range of dimensions, such as whether individuals have a cellmate or are housed alone and whether they spend years or days in a given type of housing. In addition, there is the likelihood that the appropriate outcomes for gauging effectiveness and impacts may vary depending on the housing type. For example, relative to long-term stays in single-cell confinement, short-term stays in double-bunk protective confinement might have different effects, or effects of different magnitudes, on outcomes such as mental health, victimization, and infractions. As we argue below, while there is assuredly a need for increased research on the diverse range of impacts of different types of RH, there is also a need for greater clarity about the existence and nature of these types of housing and their prevalence.
How Frequently Is RH Used?
Over the past two decades, researchers have generated estimates of RH prevalence using varying research designs and conceptualizations of what it entails. One of the earliest sources of information on RH comes from a 1997 report by the National Institute of Corrections (NIC). The report’s national survey of correctional departments was the first to document the use of supermax housing or long-term segregation for management purposes. According to the report, however, this terminology led to varying interpretations and reporting across departments because some did not have “supermaxes” while others had multiple units that fit the description. These issues aside, 34 states reported having some form of housing that resembled supermax confinement. King (1999) later extended the findings of the NIC study with new information and site visits. The updated results revealed that nearly 20,000 supermax beds existed across the country, which amounted to housing capacity for 1.8% of the prison population. A 2004 survey of supermax wardens indicated that the use of such housing had increased since King’s study to approximately 25,000 individuals in supermaxes across 44 states (Mears, 2006).
More recent studies have focused on additional forms of segregated housing. A prominent example in this area is an ongoing research collaborative between correctional administrators and the Yale Law School. Initially, this group focused on documenting the use of administrative segregation, which they defined, “as the placement of inmates in a cell (either alone or with a cellmate) for approximately 23 hours a day, and which can last for thirty days or more” (Metcalf et al., 2013, p. 2). Their current surveys aim to measure the broader prevalence of RH, which they equate with solitary confinement and define as “the placement of an individual in a cell for an average of at least twenty-two hours per day for fifteen or more continuous days” (CLA-Liman, 2020, p. 2). Based on this definition, the percentage held in RH was estimated to range from 4.9% in 2015 to 3.8% in 2019 (Association of State Correctional Administrators and the Arthur Liman Public Interest Program [ASCA-Liman], 2016; CLA-Liman, 2020). As with the earlier supermax surveys, this body of research asks correctional systems to report prevalence using the survey’s definition of RH, which may invoke different interpretations and counts. Separately, scholars have expressed concerns about such an approach due to the possibility that correctional departments may purposely underreport their use of segregation (Naday et al., 2008).
Another prevalence estimate is Beck’s (2015) study of RH using self-reported data from the National Inmate Survey. The survey includes questions about current housing status as well as “time spent in disciplinary or administrative segregation or solitary confinement” during the past 12 months (p. 16). The findings indicate that 4.4% of imprisoned persons were held in either administrative segregation or solitary confinement at the time of the survey, and more than 20% had spent time in such housing or DC in the past year. In so doing, the study conflated types of housing that could be used for a variety of purposes—such as long-term management, short-term punishment, or protection—across prison systems. The study thus provides a baseline estimate of RH use but does not reveal the prevalence of specific types of segregation.
Juxtaposed against such work are several studies that provide prevalence estimates based on administrative records data from select correctional systems or facilities. For example, a 2016 report from the U.S. Department of Justice on RH in the Federal Bureau of Prisons found that about 5% of federal prisoners were housed in special housing units for disciplinary punishment or administrative purposes in 2015, while less than 1% were in special management units or the federal supermax prison. A 2018 report by Tasca and Turanovic found that nearly 10% of RH placements in Arizona involved disciplinary segregation, whereas administrative segregation accounted for less than 3% of placements. Other studies that are state-specific, such as Lovell et al.’s (2020) Washington study or Mears et al.’s (2021) Florida study, examine the prevalence of a specific type of RH, such as long-term solitary confinement for managing individuals deemed to be unmanageable in other ways. International studies also provide insight into RH use outside the United States, including administrative segregation lasting 6 days or longer in Canada (Helmus et al., 2019) and short-term DC in Denmark (Wildeman & Andersen, 2020).
Limitations of Prior Research
Scholars have drawn increased attention to RH and, in so doing, made strides in identifying potential impacts of this housing. At the same time, several limitations and research gaps stand out. One limitation is that many prior prevalence estimates rely on surveys of reported RH use rather than administrative records detailing actual housing practices. A second is that few studies distinguish between types of RH (Labrecque et al., 2021). For example, many accounts mention protective custody and DC when discussing RH (e.g., Garcia, 2016; Shalev, 2009) but blend the two when describing the nature of this housing, its use, and its impacts. Other accounts focus on long-term solitary confinement without discussing or examining other types of RH (e.g., Mears & Bales, 2010), while still others focus only on DC but not other types (e.g., R. G. Morris, 2016; Wildeman & Andersen, 2020). There is, at the same time, the problem of researchers and correctional systems employing similar terms, such as “administrative segregation,” to refer to different types of housing and of using different names to refer to the same type of housing (Mears et al., 2019). For studies to be more readily interpretable and that allow for appropriate comparisons, attention to distinguishing unique types of RH is needed.
A third and related limitation of many accounts is the absence of information about the precise nature of particular types of RH, the policies specific to them, and the conditions and restrictions attending to their use. Exceptions exist, such as Labrecque et al.’s (2021) recent account of “administrative segregation” across states, but these focus on select types, such as long-term supermax housing (e.g., Butler et al., 2013; Naday et al., 2008). Recognition and description of all types of RH, including their purposes and designs, is necessary for creating a more comprehensive and accurate account of this housing, for providing, in turn, appropriate comparisons in studies and reviews and for making accurate generalizations, and, not least, for opening lines of additional inquiry.
This situation gives rise to a central gap in research on RH, namely, a lack of information about the extent to which different types of this housing are used. It includes information about duration. It is not safe to assume, for example, that supermax confinement necessarily entails months and even years in prison (Mears & Bales, 2010) or that DC and other such RH necessarily entail shorter stays (see, generally, Garcia, 2016). What actually is the typical exposure to the different types of housing? How frequently do individuals experience any given type of RH? To what extent do they experience stays in multiple types of the housing? In addition, given the significant concerns about re-entry (e.g., Travis et al., 2014) to what extent do individuals go straight from a given type of RH back into society, with no transition period in general population facilities? One reason that the latter question is relevant is because of the concern that prisoners in general need preparation for re-entry and that RH may further impede their readiness for a return to society. Few studies, however, have examined RH timing relative to release (see, however, Lovell et al., 2007; Mears & Bales, 2010). In short, what are needed are studies that distinguish different types of RH, examine their unique nature, and identify the prevalence and use of each type.
Current Study
This study seeks to shed light on the nature and heterogeneity of RH, including variation in the use of different types of RH. To this end, we present a case study of Florida, the third-largest state prison system in the country, which houses more prisoners than 19 states combined (Carson, 2020). Although Florida differs in some respects from other prison systems, prior literature shows that all prison systems employ multiple forms of RH (Garcia, 2016; Labrecque et al., 2021).
Conceptual clarity regarding the types and nature of RH is critical to understanding its use. We begin first by providing a policy analysis of Florida Administrative Code, which defines each type of RH, including the purpose and conditions of the housing and the criteria for admission. This analysis serves to illuminate the distinct nature of types of RH. Another aspect of the housing that remains elusive is how many people experience each type. To address this research gap, we analyze administrative records data to illuminate the prevalence of RH in Florida and how each type can vary across a range of dimensions, including frequency and duration.
Method
Administrative Policy Analysis
Our study begins with a review of definitional and operational differences across distinct RH types in Florida’s rules. The Florida Administrative Code contains the official administrative policies and practices of state agencies, including the Florida Department of Corrections (FDC). These rules outline specific types of RH, including when and how they should be used. We examine policies for all specialized nongeneral population housing used to address safety and security concerns, focusing on those that were in effect during the period of our administrative records data. The analysis does not distinguish between security custody levels because these classifications apply to all incarcerated persons. Rather, we follow prior accounts (CLA-Liman, 2020; Garcia, 2016; Mears et al., 2019; Shalev, 2009) that distinguish between administrative, disciplinary, and protective confinement categories.
For each housing type, we identify and then discuss the definitions, placement and review procedures, time limits, and conditions of confinement, according to stated policy. In all cases, RH placement decisions involve the Institutional Classification Team, or ICT, which includes a facility’s warden or assistant warden, classification supervisor, and correctional officer chief. The State Classification Office, or SCO, oversees classification decisions for the entire state prison system, and also plays a central role in decision-making. We discuss their roles in relation to the placement decisions for RH.
Administrative Records Analysis
Separately, we then examine administrative data from the FDC to estimate the use of different types of RH in Florida, with a focus on the prevalence of exposure and typical durations of time spent in each type. The analyses use administrative records of Florida prison terms that began on or after July 1, 2007, and ended on or before December 31, 2015 (n = 216,875 prison terms for 184,183 unique persons). Administrative records on facility movements and status assignments provide the opportunity to examine exposure to each RH type. Specifically, we constructed a data file with one row for every day of the prison term, identified stays in each type of housing and then created the following measures.
To examine overall prevalence, we determined both the frequency and percent of prison terms with a stay of 1 or more days in each housing type. We also counted the number of stays in each type, which we operationalized based on the number of times the prison term involved a unique stay as determined by entry and exit dates. We next calculated the average stay duration in days. Here, we divided the total number of days spent in each type by the total number of stays to generate an average number of days per RH stay rounded to a whole day. We calculated the cumulative duration spent in separate forms of housing by totaling the number of days in each type over the entire prison term. Both duration measures exclude days spent away from the FDC for events like court hearings or interstate transfers because the RH status does not continue while individuals are at outside facilities. These measures, however, do include days where individuals are housed elsewhere in the FDC, such as a mental health or medical unit, because their “restrictive” status still is in effect at these times. We also identified whether the prison term involved exposure to other RH stays. Specifically, this measure indicates whether an individual spent time in the applicable housing and at least 1 day in another type of RH during their prison term. Finally, we identified whether the prison term ended with a direct release from each housing type to the community, which captured whether there was an RH assignment on the release date.
Results
Administrative Policy Analysis
Types of RH
An analysis of the Florida Administrative Code shows that there are seven types of RH available in Florida prisons. Each constitutes “restrictive housing” in relying on segregation of individuals from the general prison population, isolation in a cell alone or with another person, limited out-of-cell time for recreation or programming, and other restrictions. These similarities mask, however, the qualitatively different purposes, conditions, and operations of each housing type.
As review of Table 1 shows, Florida uses terminology that distinguishes types of RH based on the goals and the duration of confinement. It relies on several subtypes of housing that serve a management function. The first is Maximum Management (MM), which is a temporary housing status for individuals who pose an extreme security risk beyond the level of control available in other RH units or on death row (Fla. Admin. Code R. 33-601.820). Close Management (CM) consists of three levels of extended segregated housing to manage individuals who cannot be managed in other ways and do not meet the requirements for MM placements (Fla. Admin. Code R. 33-601.800). Close Management 1 (CM1) is the most restrictive and isolating, and, like MM, entails extended solitary confinement (i.e., with no cellmate). The two next levels—Close Management 2 and 3 (CM2 and CM3)—do not require isolation or as many restrictions. A fifth type of management-focused housing is Protective Management (PM), which is used to separate vulnerable individuals from the general population for their own safety, not for discipline (Fla. Admin. Code R. 33-602.221).
Restrictive Housing Variation: An Illustration Based on Florida
Note. MM = Maximum Management; CM = Close Management; PM = Protective Management; DC = Disciplinary Confinement; AC = Administrative Confinement.
The two remaining types of RH have goals that do not center on management and that involve shorter-term stays. The first, DC, is a form of punishment for individuals found guilty of violating facility rules (Fla. Admin. Code R. 33-602.222). These units essentially operate as a jail within the prison system. In addition, Florida uses Administrative Confinement (AC) to temporarily segregate individuals from general population for a variety of reasons (e.g., investigations and pending disciplinary hearings) while facility personnel determine a more appropriate housing plan (Fla. Admin. Code R. 33-602.220). These units operate as a form of pretrial detention or ad hoc auxiliary housing.
Nature of Each Type of RH
Definitional differences provide initial insight into the diversity of RH. We next turn our attention to describing qualitative differences in the nature of these types of segregation. We begin with the most restrictive housing and then discuss the other types.
MM involves temporary placement in solitary confinement for individuals who qualify for CM or death row housing and have recently committed a serious act (e.g., staff assault, life-threatening inmate assault, escape or escape attempt; Fla. Admin. Code R. 33-601.820). Placements into MM involve a referral process and hearing by the ICT. State policy does not specify time limits on placement. However, the ICT reviews an individual’s MM status every week for the first 60 days of placement. MM imposes the most restrictions upon placement, including severely limited outdoor recreation time (e.g., 2 hours once every 30 days during the first 60 days of confinement), visitation (i.e., legal visits only), reading materials, and canteen purchases. These restrictions may be reduced over time depending on the individual’s behavior.
MM is architecturally distinct from other RH types with a cell-within-a-cell design. A solid metal outer door completely blocks out views of the wing and natural light, while a recessed inner cell imposes another layer of separation. It resembles accounts of supermax units and the original solitary cells at Pennsylvania’s Eastern State Penitentiary as well as the modern-day extreme isolation in California’s Pelican Bay State Prison. This design imposes the most isolation of any housing in the state but is restricted to a single wing in one prison facility.
CM consists of three levels of long-term RH (Fla. Admin. Code R. 33-601.800). The highest level—CM1—is also similar to what is traditionally viewed as supermax housing (Mears & Bales, 2010; Riveland, 1999). Specifically, CM1 entails solitary confinement for 22 hours or more per day for a policy-stated minimum of 6 months. Individuals can be referred to CM1 for serious acts that pose a risk to system safety and security (e.g., an incident causing death or injury, staff assaults, sexual assault, inciting a riot, successful or attempted escapes, possession of weapons, drug trafficking). During CM1 placements, individuals have limited outdoor recreation time (i.e., 2 hours up to 3 days per week), no day room access, no out-of-cell work assignments, and in-cell only educational programming. CM1 also limits access to the outside world—individuals are allowed one personal phone call and one 2-hour noncontact visit every 30 days after completing 30 days in CM with no major rule violations.
CM 2 and 3 also involve long-term segregation for management reasons but with fewer restrictions. In addition, CM2 and CM3 cells are designed to house two individuals at a time, whereas CM1 cells only provide for solitary confinement. However, it is possible that individuals in CM2 and CM3 may be housed alone at certain periods or facilities, such as when placements to them are low. Currently, the three CM levels together form a step-down process where individuals start in CM1 and may later progress to less-restrictive levels. (According to FDC personnel, however, historically individuals could be and were referred to any of the three levels depending on a given safety or security concern.) Referrals to CM2 involve similar but less-serious safety and security concerns as CM1 (e.g., pattern of predatory actions or threatening behavior, participation in a riot or contraband trafficking). And direct CM3 referrals may involve actions such as aiding escapes, possession of contraband, validated gang membership, or adjustment problems, behaviors that are serious but may be less so than those that may result in CM1 or CM2 placements.
Regardless of the level, all CM placements require a three-part decision process—initial referral, ICT hearing, and SCO decision. After a classification officer submits a CM referral, a hearing is held by the ICT to interview the incarcerated individual and to review the evidence in support of the placement. The ICT then sends their recommendation to the SCO, who reviews all documentation before making a final placement decision. If the SCO agrees to a CM placement, the current facility houses the individual in a confinement cell until they can be transferred to a CM facility. A similar review process occurs every 6 months, and can result in continued placement or a transfer to another CM level or back to general population.
As occurs with CM1, individuals placed in CM2 and CM3 have limited outdoor recreation time. However, both levels allow for day room access—4 hours per day up to 2 days a week for CM2 and 4 hours per day up to 5 days a week for CM3. Individuals in CM2 are eligible for a limited number of work assignments, whereas those in CM3 can be assigned to jobs throughout the facility. CM2 and CM3 also allow for in-cell and out-of-cell educational program participation. Not least, both levels provide greater access to the community. Placed individuals can make one phone every 2 weeks in CM2 and every 7 days in CM3. After a period without major rule violations, persons in CM2 can have one 3-hour non-contact visit every 2 weeks while those in CM3 can have one 4-hour contact visit.
Unlike other prison systems, Florida does not have a supermaximum prison like California’s Pelican Bay State Prison or the Federal Bureau of Prison’s USP Administrative Maximum (ADX). Rather, CM units are concentrated at a small number (typically, five to seven) of facilities throughout the state (Office of Program Policy Analysis and Government Accountability, 2019); the overall number has fluctuated over time. In addition, whereas many states built technologically advanced segregation units with centrally controlled cell locks and in-cell surveillance, Florida retrofitted existing facilities to accommodate CM housing.
PM differs substantially from these types of management-focused RH. PM housing segregates individuals at-risk of victimization from the general population; placement is for their own safety, not because their behavior poses a threat (Fla. Admin. Code R. 33-602.221). In addition, incarcerated persons can request placement into PM by submitting a statement that details their safety concerns (Fla. Admin. Code R. 33-602.220). The facility then places the person into AC pending an investigation and ICT review. The ICT interviews the individual and considers evidence of victimization, threats, or harassment, the individual’s reputation as an informant, their conviction history, and their physical characteristics (e.g., size, build, age). After this review, the ICT forwards their recommendation to the SCO, who makes the final status determination. Individuals who are denied protective custody can appeal the decision.
Another distinguishing feature of PM is that there are no stated minimum or maximum time limits. Rather, Florida policy emphasizes returning individuals to general population as soon as it is safe to do so; this requires that an individual’s protective status be re-evaluated regularly. Moreover, unlike other RH, protected persons can request a release from PM that the ICT and SCO then review during ongoing placement evaluations. Still another distinguishing feature is that Florida policy explicitly states that PM should mimic general population housing to the extent possible, albeit with separate accommodation. Thus, placed individuals do not in theory face the same set of stated restrictions as other types of RH. For example, PM allows free movement and interaction with other persons within the unit. However, the need for separate facility or in-unit spaces for routine privileges (e.g., visitation, outdoor exercise, programming, work assignments) may result in more limited access than what is available to their counterparts in general population units.
In addition to the above types of management-related housing, Florida also uses two types of RH—DC and AC—that, in general, are for short-term stays. Incarcerated persons can be placed in DC if they are found guilty of violating system rules during a disciplinary hearing process (Fla. Admin. Code R. 33-602.222). Placement into DC begins when prison personnel file a formal disciplinary report describing the alleged rule violation. The accused individual is notified of the report and an internal investigation begins. Once the investigation is complete, a hearing is scheduled with the facility’s disciplinary team (for major rule violations) or hearing officer (for minor violations; Fla. Admin. Code R. 33-601.306). The hearing resembles a traditional court process whereby a hearing member reads and reviews the charges and asks for a plea. If the accused individual pleads not guilty, then they are given the opportunity to present exculpatory evidence and witness statements and provide a closing statement (Fla. Admin. Code R. 33-601.307). If the disciplinary team or hearing officer decides the individual is guilty of the alleged rule violation, then they can sentence the person to various sanctions, including placement in DC. The amount of time in DC varies by offense—more serious violations have a maximum duration of 60 days while minor violations are typically 30 days or less (Fla. Admin. Code R. 33-601.314). Stays may exceed these durations if the disciplinary team orders consecutive placements in DC for multiple violations.
AC, on the contrary, is temporary segregation for a variety of administrative or decision-making purposes (Fla. Admin. Code R. 33-602.220). Specifically, senior correctional officers can place individuals in AC pending internal classification decisions or facility transfers, disciplinary investigations or hearings, security concern assessments stemming from unresolved outside charges, requests for protection, or CM evaluations. The ICT reviews AC placements within 72 hours to decide whether the individual should return to general population or remain in AC pending final resolution of a case. AC placements are generally limited to 15 days or less, with the possibility of limited time extensions for certain reasons upon system approval.
Both DC and AC restrict telephone privileges and out-of-cell exercise. However, DC imposes additional restrictions beyond AC, including limits on personal items, canteen purchases, reading materials, and visitation. All major prison facilities in the state have DC and AC units that are physically separated from general population. Cells in both types of units are designed for shared confinement but, depending on referrals, may entail solitary confinement.
Our policy analysis highlights that RH can involve a range of conditions—different reasons for placement, durations, restrictions, and levels of isolation—even though each type clearly differs from general population housing (Mears et al., 2019). The terminology used to describe each type can easily cause confusion without additional contextual clarity. Numerous prison systems, for example, call their long-term solitary confinement units “administrative segregation” and thus, prior research has assumed that “administrative confinement” in Florida operates in this way (see, e.g., Labrecque et al., 2021). Moreover, Florida does not use the terms “solitary confinement” or “supermax” to describe any of their RH units. However, the conditions of both CM1 and MM reflect the traditional conceptualizations of extended isolation. Research that combines RH types, such as DC and long-term segregation for management purposes, obscures how each may result in different outcomes (see, e.g., Labrecque & Smith, 2019a; Zgoba et al., 2020).
Administrative Records Analysis
Overall Prevalence
The Florida Administrative Code provides initial insight into the policy-based differences between types of RH in Florida. We now turn to examining whether these distinctions exist in practice. To this end, analyses of administrative records data are shown in Table 2. Inspection of the table reveals that there is substantial variation in the prevalence of different types of RH. In general, long-term housing for management-related purposes is rare, while short-term confinement is quite common across prison terms. Specifically, housing that requires solitary confinement—MM and CM1—affected one-hundredth of 1% and 1% of prison terms, respectively. Transitional long-term management housing—CM2 and CM3—occurred among less than 2% of prison terms. After MM, PM was the second-least common form of RH. On the contrary, both forms of short-term confinement were normative experiences during the study period. Nearly 35% and 48% of prison terms involved a stay in DC or AC, respectively.
Profile of Different Types of Restrictive Housing Experiences Across Prison Terms, 2007–2015
Note. N = 216,875 prison terms for 184,183 unique persons; RH = restrictive housing; all terms = across all admissions-and-releases July 2007 to December 2015; if placed = among applicable RH placements; MM = Maximum Management; CM1 = Close Management 1; CM2 = Close Management 2; CM3 = Close Management 3; PM = Protective Management; DC = Disciplinary Confinement; AC = Administrative Confinement.
The substantial differences in the use of different types of RH can readily be seen in Figure 1. As Panel A shows, the type of housing that often evokes the most concern—long-term solitary confinement (i.e., MM and CM1) that may last 6 months or more—is infrequent, with less than 1% of prison terms entailing this type of incarceration. By contrast, short-term confinement for disciplinary or administrative reasons occurs far more often, collectively accounting for 82% of stays in RH.
Restrictive Housing in Florida, 2007–2015
Number of Stays
When examining the number of stays in each RH type, we find that the vast majority of prison terms that involved some form of long-term management housing were based on a single stay. Short-term confinement, however, shows a cycling process whereby individuals experience repeated exposure to DC and AC. As Panel B in Figure 1 displays, nearly 50% of prison terms with any DC involved two or more separate stays in this housing. A similar pattern was observed for AC. The upper range for the number of stays in Table 1 underscores how these types of housing differ. Some prison terms involve up to five separate stays in CM1. For DC and AC, we see as many as 23 and 35 separate placements, respectively, during the course of a single incarceration. It warrants mention that “snapshot,” or static prevalence estimates that measure housing status on a single day do not capture this type of cycling into and out of certain types of RH and thus miss that for some incarcerated persons, frequent exposure to certain types of RH, especially DC and AC, is common.
Average Stay Duration
Substantial variation in RH experiences by the average duration of housing stays can also be seen in the analyses. Panel C in Figure 1 shows that the median number of days per stay in any of the five types of management-related housing is longer than the typical amount of time in short-term confinement. Specifically, stays in PM have the longest duration followed by CM1. The average stay in CM2 and CM3 is typically around 175 days. DC stays typically last a month whereas the average stay in AC is approximately a week.
When examining the ranges for each housing type in Table 2, we can see that some prison terms include confinement placements lasting more than a year while others involve a single day in MM, all three levels of CM and PM. Shorter stays may occur if a placement is overturned upon review, or if individuals are placed in RH close to their release date. The ranges highlight the lengthy placement durations that can exist for all types of RH, especially for the CM types of confinement (CM1, CM2, and CM3) and PM. At the same time, the median and average durations, which typically hew closer to the lower end of the ranges, reinforce that such extremes are not typical.
Cumulative Duration
The findings for cumulative duration mirror the above patterns. Over the course of a prison stay, people spend less time in short-term confinement overall despite a higher incidence of repeated exposure. Approximately 95% of prison terms with AC and 80% with DC involved cumulative durations of 90 days or less. The reverse is true of long-term confinement—the normative experience is cumulatively longer durations. More than one-third of prison terms with PM exposure and roughly one-fourth with CM1 and MM placements involved a total duration of 1 year or longer. The majority of prison terms with any CM2 or CM3 involved cumulative durations of less than 1 year. These durations contradict a common assumption that all RH placements continue for years on end. Given the fact that 77% of state prisoners serve less than 3 years (Kaeble, 2021), multiyear housing placements in fact can only occur for a small subset of the prison population with especially lengthy sentences.
Other RH Stays
Inspection of the Table 2 indicates that prison terms involving one type of RH typically include other types as well. That is, the universal experience among people ever exposed to RH during an incarceration term is exposure to multiple forms of RH. These findings make sense, given the purposes of each. AC, for instance, is used for temporarily confinement until a disciplinary hearing can be completed. Thus, individuals who experience DC will likely have experienced AC first.
Direct Release
There are different patterns for direct release to society, as shown by Panel D in Figure 1. More than half of prison terms involving CM1 and more than two-thirds of terms with PM involve direct release from this housing to the community. That is notable in part because of the lengthy stays in this housing. Nearly one-third of terms involve direct release from MM, CM2, or CM3. Direct release from either form of short-term confinement is substantially less common—less than 10% of prison terms involve a direct release from DC or AC.
Discussion
To illuminate the “dark figure” of RH, this study examined seven different types of RH in Florida and the relative prevalence of each. We found that types of RH vary greatly, with each differing in its design and purpose. Some are for protection, others for disciplinary punishment, others for administrative operational goals, and still others for varying degrees of management. Some allow for double-bunking, some require solitary confinement, and some impose more restrictions than others. Although we only examined Florida, similar observations can be made about RH in other prison systems. For example, California, Ohio, and Arizona all use multiple forms of confinement (Labrecque et al., 2021). California uses at least three types of RH: detention units for short-term disciplinary purposes, security housing units for longer-term management purposes, and protective housing units for the safety of incarcerated individuals (California Department of Corrections and Rehabilitation, 2021). Ohio uses a hierarchy of segregated housing, including multiple levels of extended RH as well as short-term “limited privilege” housing (Ohio Department of Rehabilitation and Correction, 2020). Arizona has three types of RH—maximum custody units, close management, and protective custody—that vary in duration and use of solitary confinement (Arizona Department of Corrections, Rehabilitation & Reentry, 2021).
In addition, we identified substantial variation in the extent to which each type is used. Long-term RH, including solitary confinement, occurs rarely, amounting to less than 2% of all prison terms during the study period. By contrast, DC housing occurs among approximately one-third of all incarcerations. AC was even more common, with almost half of prison terms experiencing such placements.
The variation in frequency and duration across these types of housing was substantial. For some, such as long-term solitary confinement, nearly all who experienced the form had one stay. By contrast, repeated stays were common in DC and AC. The typical duration of segregation was over 6 months for those in long-term solitary confinement, other types of management confinement, or protective housing. DC typically lasted 1 month, and AC stays 1 week. Notably, across all types of housing, outlier instances of individuals serving years in the housing existed, and most individuals who experienced a given type of RH experienced others as well.
Not least, we found substantial variation in direct releases to society. Over half of prison terms involving the most common type of long-term solitary confinement were released directly from this housing to society, and over two-thirds of those in protective custody experienced direct releases. Direct release from DC or AC was rare.
These findings should be viewed in light of several limitations of the study. Our analysis of administrative policy and administrative records data constitutes a case study of a single state during the study period. Both the specific policies and the use of RH assuredly will vary across states as well as within states over time, and prison systems may add to or vary the nature of specific types of housing for operational or security reasons and as prison populations change (see, e.g., Sakoda & Simes, 2021). Another limitation is that the study’s findings are based on an admission-and-release cohort, which may provide different estimates than those from “stock” population, or “snapshot” samples of prison populations. We also are not able to assess how the conditions of confinement outlined in state policy translate into practice, including what restrictions incarcerated individuals actually experience or how much time they are able to spend out of their segregation cells.
We turn now to implications of these findings for research and policy. First, similar studies are needed in other states to create an empirical foundation for comparing the precise nature of the different types of RH in each state and their use. That substantial variation in RH exists has been suggested in prior work and accounts of RH (CLA-Liman, 2020; Mears et al., 2019; Naday et al., 2008; Shalev, 2009), but this study highlights in more detail the nature of and reason for each type. And it draws attention to the use of a broader spectrum of RH, including types of long-term management housing, which may be used by prison systems. Labrecque et al. (2021) recently analyzed state prison policy descriptions of “administrative segregation.” Their study advanced understanding of the nuances and variation around various “segregation” policies, yet, like other accounts (e.g., CLA-Liman, 2020) risked conflating or missing a variety of RH that does not necessarily entail solitary confinement or a focus on long-term management. This study and theirs suggest the need for caution when generalizing studies of one type of housing to other types, and, as Labrecque et al. (2021) and others (e.g., Mears et al., 2019) have emphasized, the importance of describing the conditions of the type of housing under investigation.
We echo this recommendation and would argue that the next generation of research needs to center on state-by-state descriptions of each distinct type of RH, each type’s unique goals and conditions, and the extent to which each is used. Given the confusion that circles around the use of particular terms, studies ideally will use descriptive language that captures these different dimensions. The very term “administrative segregation,” for example, can signify different types of housing across states, as is the case in Florida, where it is used for a variety of jail-like purposes, while in Texas it consists of solitary confinement for long-term management purposes. To the extent that states move toward transitioning individuals into and out of longer-term solitary confinement and, more generally, seek alternatives to excessive restrictions or sanctions (Digard et al., 2018), new variations on existing housing, as well as new terms for the housing, may arise. Keeping abreast of such changes and ensuring accuracy in how results are presented will become even more important in coming years.
Second, research is needed on the extent to which each type of housing is used and on its impacts. Comparing the demographic and behavioral characteristics of those placed into RH is a critical next step in identifying if, and to what extent, each type is used to segregate certain types of incarcerated individuals. Impact evaluations are needed that identify the extent to which the goals associated with each type of RH are achieved. Research is needed as well that can distinguish between levels of housing that have different stated purposes but are used to address similar behavior. For example, what specific factors related to misconduct (e.g., severity, frequency, and victim position) result in a referral to long-term management housing rather than short-term DC?
Assessing the impacts of each housing type will be challenging. The relevant outcome for protective custody, for example, may be the extent to which a vulnerable individual experiences less victimization. For DC, the relevant outcome may be reduced misconduct. For long-term transitional management housing, it may be a greater ability to navigate social interactions. In all instances, behavioral outcomes, such as misconduct and recidivism, as well as physical and mental health outcomes would be relevant to assess and compare across housing types. For example, short-term AC may not have the same adverse effect on mental health as long-term solitary segregation. Individuals may be more likely to comply with institutional rules in protective custody to maintain their housing status than individuals in DC. To our knowledge, no state and no studies provide this type of comprehensive comparative assessment of their RH. Yet, to ensure appropriate use and to gauge effectiveness, such assessments will be needed. They will require including duration measures, such as total continuous or cumulative exposure to particular types of housing. Not least, states will want to monitor and assess individual and organizational factors, such as staffing levels and training and rehabilitative programming, which contribute to the use and impacts of each type of RH (Lovell et al., 2020).
Third, the significant durations of time that individuals spend in any of the different types of housing, with some spending years in it—and the large percentage returning to society directly from the housing—underscores the risks to these individuals and to society when they leave prison. Research is needed that sheds light on their re-entry experiences. At the same time, prison systems will want to minimize time in the various types of housing, given that providing rehabilitative programming and services in them is challenging. The limited programming RH individuals have almost assuredly reduces their likelihood of successfully navigating their way in prisons and in having healthy, prosocial re-entries into society.
Finally, the study’s findings suggest additional warrant for prison systems to reconsider their reliance on extreme forms of housing. There is little basis for viewing extended stays in any type of RH as the most cost-efficient way to manage prison systems (Digard et al., 2018; Garcia, 2016). At the same time, there is considerable basis for viewing such housing as potentially harmful (Frost & Monteiro, 2016; Kapoor & Trestman, 2016; King, 2018; Labrecque & Smith, 2019b; Mears, 2016; Richards, 2015; Shalev, 2009). The warrant stems both from the plethora of types of RH, the long durations that some individuals face in this housing, and the large numbers of individuals directly released into society. Of course, prison systems operate with fiscal constraints. Prison systems will need more funding and resources to implement alternatives, such as greater reliance on rehabilitative programming and well-trained staff. Investments, too, will be needed to reduce incarceration of individuals who may be more cost-efficiently punished and assisted in communities (Travis et al., 2014).
Footnotes
Authors’ Note:
The authors thank the editor and anonymous reviewers for their thoughtful suggestions in improving this article. This project was supported by Grant No. 2016-IJ-CX-0014 awarded by the National Institute of Justice, Office of Justice Programs, U.S. Department of Justice. Points of view 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 Florida Department of Corrections.
