COLORADO DEPARTMENT OF CORRECTIONS
OFFICE OF PLANNING AND ANALYSIS
2862 SOUTH CIRCLE DRIVE
COLORADO SPRINGS, CO 80906
HTTP://WWW.DOC.STATE.CO.US

JOE ORTIZ
Executive Director
NOLIN RENFROW
Director of Prisons
C. SCOTT HROMAS
Director of Research

ACKNOWLEDGEMENTS
Many people were instrumental in seeing this project through to its completion, particularly Larry Reid
and Nolin Renfrow who were the instigators for the research. Thank you to the following committee
members for your input into the study: Nolin Renfrow, Larry Reid, Gary Golder, Ron Leyba, Joanie
Shoemaker, John Stoner, Jim Michaud, Charlie Smith, Scott Hromas, Kristi Rosten, Daryl Vigil, Leonard
Vigil, and Paul Naudeau. Leonard and Paul went the extra mile in collecting important data for this project.
Thanks also to Alysha Yarbrough for her research on the legal cases and Kelli Klebe for her research input
and review along the way.

ACKNOWLEDGEMENTS

i

EXECUTIVE SUMMARY
Born out of the necessity to provide safety to the public, staff, and inmates, prison systems have rapidly
expanded the use of administrative segregation or supermax facilities. In 1983, the federal system opened the
first modern-day supermax; by 1999, 32 states were operating or planned to open supermax facilities
(National Institute of Corrections, 1997). Supermax facilities, based on the consolidation model of
concentrating the most dangerous criminals together (Hershberger, 1998), provide tight controls to ensure
reductions in violence and serious prison disruptions.
The nature of administrative segregation, primarily solitary confinement, has alarmed many individuals
and groups, including the Human Rights Watch (1997, 2000). Still, greater concern is devoted to the belief
that solitary confinement is psychologically damaging to mentally ill inmates. Further speculation holds that
solitary confinement may even produce symptoms of mental illness in inmates who had no mental health
symptoms when they were placed in administrative segregation.
The constitutionality of administrative segregation has been challenged many times over in courts.
Although judges have vocalized their contempt for long-term solitary confinement, none determined that
supermax facilities violate the “cruel and unusual punishment” clause of the Eighth Amendment. In brief,
courts have generally deemed administrative segregation unsuitable for many types of inmates, such as
mentally ill, developmentally disabled, or mere nuisance inmates.
There exists no empirical evidence that supermax confinement either is or is not psychologically
harmful. Despite the preponderance of supermax literature, it is largely theoretically driven rather than
empirically based. The extreme polarity among researchers reveals their inherent biases, making it nearly
impossible to draw meaningful conclusions from the literature. There exists a void of even basic statistics on
supermax inmates; even less is known about the psychological profiles of inmates confined to supermax.
The present study seeks to inform the design for a subsequent prospective research project while
providing basic statistics about Colorado administrative segregation. This study first compared Colorado
Department of Corrections’ (CDOC) administrative segregation population to the overall prison population,
across demographic characteristics, criminal history, institutional behavior, needs levels, and standardized
psychological assessments. The purpose was to understand who are the inmates serving their sentences under
administrative segregation conditions. Secondly, this study included a longitudinal design to examine length of
stay, return to administrative segregation, recidivism rates, and institutional behavior. Finally, an analysis of
administrative segregation hearings was conducted.

Study 1: Population Comparisons
The present study compared 981 administrative segregation inmates to the overall prison population (N
= 16,171), excluding inmates under community supervision. Administrative segregation inmates were
identified through classification levels on June 30, 2003. Demographic, criminal history, and needs assessment
data were obtained from administrative information in the CDOC database.
• Demographics. Compared to the general prison population, inmates in administrative segregation
were more likely to be male, Hispanic, younger, and not married.
• Criminal History. Inmates in administrative segregation were more likely to be violent, have longer
incarcerations, and have higher Level of Service Inventory – Revised (LSI-R) scores.
• Pre-segregation Custody Levels. Prior classification levels were 70% close custody, 17% medium,
8% maximum (new cases or parole violators), 4% minimum-restrictive, and 1% minimum.
• Institutional Behavior. Administrative segregation inmates had more than triple the rate of
disciplinary infractions and punitive segregations than general population inmates. Administrative
segregation inmates accounted for 24% of the disciplinary violation even though they represent only
6% of the population.
• Needs. Inmates in administrative segregation presented with greater needs in psychological, mentally
retarded / developmentally disabled, assaultive, and self-destruction areas.
• Security Threat Group (STG). The incidence of STG involved inmates in administrative
segregation was 66% versus 26% in the general population. There were fewer inmates involved with

EXECUTIVE SUMMARY

ii

•

•

Crips, White Supremacists, and Bloods in administrative segregation than in population, but more
Surenos-13, Gallant Knight, and Prison Gang. STG involved inmates have equal or less needs than
other inmates in administrative segregation.
Offenders with Mental Illness (OMI). OMIs not only appeared at a greater rate in administrative
segregation, but they were found to have more serious psychiatric symptoms than the mentally ill in
the general population. In addition to their mental health needs, this group presented with greater
academic, vocational and medical needs.
Primary Determinants of Placement. A logistic regression was conducted to determine offender
characteristics that might predict administrative segregation placements. STG involvement was the
strongest predictor; the odds ratio indicated that inmates with STG involvement were 4.5 times more
likely to be placed in administrative segregation than those with no involvement. Violent, Hispanic,
single, and OMI inmates had significantly greater odds of such a placement than inmates without
those traits.

Study 2: Longitudinal Analysis
The longitudinal sample consisted of 3,003 inmates who were placed in administrative segregation
between January 1995 and December 2003. All data were downloaded from the CDOC administrative
database.
• Segregation Durations. The median length of stay in administrative segregation was 18.83 months.
OMIs had shorter durations than inmates with no diagnosis, while STG members had longer
durations than those with no affiliation.
• Release Types. The majority (59%) of inmates released from administrative segregation back into to
the general prison population. The remainder release directly to the community: 20% mandatory
parole, 13% sentence discharge, and 8% discretionary parole.
• Return to Segregation. 12% of inmates returned to administrative segregation within 1 year of their
release (either to prison population or community) and 20% return within 2 years.
• Recidivism. Recidivism rates for offenders in administrative segregation were higher than the overall
CDOC 3-year rate of 50%. Inmates releasing directly to the community averaged a 66% recidivism
rate and inmates releasing to prison prior to the community averaged a 60% rate.
• Role of Institutional Behavior. Disciplinary violations not only decreased as a result of
administrative segregation, they were also related to offenders’ release from segregation. Clearly,
disciplinary violations did not represent the only factor dictating release decisions; the other factors
remain yet unmeasured.
• Trend of Segregating OMIs. Five year prison and administrative segregation populations showed
that OMIs have been over-represented in administrative segregation. The rate increased substantially
as of July 2003, coinciding with a dramatic decline in mental health professionals and rehabilitation
programs that were a casualty of budget cuts.

Study 3: Administrative Segregation Hearings
Inmates who are determined to be a threat to the facility are removed from population and placed in a
segregation cell. They attend an administrative segregation hearing where it is determined whether such a
placement is warranted. The case is next reviewed by the warden who has the opportunity to affirm, reverse
or modify the decision. Lastly, each case is reviewed by CDOC’s Offender Services staff prior to
reclassification and transfer. Data from 842 hearings from August 2003 to July 2004 were used for this study.
• Rate of Placement: 90% of hearings resulted in an administrative segregation placement. On
average, 69 inmates were placed in administrative segregation per month.
• Reasons for Placement. The most common reasons for placement included multiple disciplinary
violations, advocating facility disruption, STG activity, and assaults.

EXECUTIVE SUMMARY

iii

TABLE OF CONTENTS
INTRODUCTION

1

HISTORY OF ADMINISTRATIVE SEGREGATION
ROLE OF ADMINISTRATIVE SEGREGATION IN TODAY’S PRISONS
OBJECTIONS TO ADMINISTRATIVE SEGREGATION
CONSTITUTIONALITY OF SUPERMAX
EFFECTS OF SUPERMAX ON PSYCHOLOGICAL FUNCTIONING
ADMINISTRATIVE SEGREGATION IN COLORADO
PRESENT STUDY

1
2
3
4
5
7
7

STUDY 1: POPULATION COMPARISONS

9

METHOD
PARTICIPANTS
MATERIALS
PROCEDURE
RESULTS AND CONCLUSIONS
POPULATION COMPARISONS
STG INVOLVEMENT
MENTALLY ILL
LOGISTIC REGRESSION

9
9
9
10
10
10
15
17
20

STUDY 2. LONGITUDINAL ANALYSIS

22

METHOD
PARTICIPANTS
PROCEDURE
RESULTS AND CONCLUSIONS
LENGTH OF STAY
RETURN TO ADMINISTRATIVE SEGREGATION
RECIDIVISM
INSTITUTIONAL BEHAVIOR
TREND OF SEGREGATING MENTALLY ILL OFFENDERS

22
22
22
22
22
24
25
26
29

STUDY 3: ADMINISTRATIVE SEGREGATION HEARINGS

30

METHOD
PARTICIPANTS
PROCEDURE
RESULTS AND CONCLUSIONS

30
30
30
30

DISCUSSION

33

FUTURE RESEARCH

34

REFERENCES

35

APPENDIX A

37

APPENDIX B

37

TABLE OF CONTENTS

iv

Just as prison is society’s solution to errant citizens, supermax is prisons’ answer to contentious
inmates. Born out of the necessity to provide safety to the public, staff and inmates, prison systems have
expanded the use of administrative segregation, or supermax, facilities. By the year 1998, nearly 2% of the
entire prison population was housed in administrative segregation (Human Rights Watch, 2000; King, 2000).
Administrative segregation facilities provide tight controls to ensure reductions in violence and serious prison
disruptions.
The nature of administrative segregation, primarily solitary confinement, has alarmed many individuals
and groups including the Human Rights Watch (1997, 2000). Still, greater concern is devoted to the belief
that solitary confinement is psychologically damaging to mentally ill inmates. Further speculation holds that
solitary confinement may even produce symptoms of mental illness in inmates who had no mental health
symptoms when they were placed in isolation. The following literature review highlights the controversial
issues and reviews empirical studies on administrative segregation.

History of Administrative Segregation
Toch (2003) described early antecedents to today’s supermax facilities, in which experiments of social
isolation and stimulus deprivation were conducted, emanating in the early 1800’s. These concepts were
reportedly abandoned because of their negative effect on inmates’ psychological functioning, which subsided
with the lessening of those conditions. Administrative segregation for the “worst of the worst” would appear
to have been forsaken until circa 1930.
Alcatraz Island Prison emerged in 1934, offshore from San Francisco, as the first modern-day
supermax facility, operated by the Department of Justice. The island, surrounded by icy, dangerous waters,
provided an ideal setting to impose solitary confinement on its tenants, the most ruthless predators, escape
artists, and racketeers following the post-Prohibition, post-Depression American era. Alcatraz served as the
first large-scale model of concentrating difficult prisoners in one facility. The demise of Alcatraz Island was
attributed to decaying conditions and a shift towards a rehabilitation philosophy rather than a failed concept
of super-maximum confinement (Hershberger, 1998).
The Federal Bureau of Prisons (BOP) next commissioned the prison in Marion, Illinois in 1963 to
replace Alcatraz. In actuality, all but 10 Alcatraz prisoners were transferred to various facilities other than
Marion, which consequently was not run as a supermax facility until 20 years later (King, 2000). In 1983, the
prison went into permanent “lockdown” or solitary confinement status, following a stream of serious
incidents culminating in the inmate murder of two correctional officers. The BOP followed up the Marion
Prison concept with the Administrative Maximum Prison in Florence, Colorado in 1994. Both the Marion
and Florence prisons operate today as administrative segregation facilities.
At the same time the Florence facility was materializing, state systems produced a proliferation of
supermax facilities. The National Institute of Corrections (NIC, 1997) indicated that 34 states were either
operating supermax housing or planned to open supermax facilities by 1999. While some prisons were
retrofitted to meet administrative segregation standards, many supermax facilities were newly constructed.
Despite the seemingly virtual explosion of supermax prisons, administrative segregation and control units
have operated on a smaller scale for decades (Zinger, Wichman, & Andrews, 2001).
NIC (1999) noted the heterogeneity across different jurisdictions’ definition of supermax. Supermax is
considered a generic term; many facilities became known under a variety of titles, including control units, security
housing units or SHUs, intensive management unit, and security controls unit (Haney, 2003; NIC). Additionally, a
supermax facility may refer to an entire facility or a distinct unit within a facility. However, administrative
segregation is differentiated from punitive segregation, disciplinary segregation or segregation; these are a time-limited
response to a disciplinary infraction after due process hearings resulting in a finding of guilt.
Recognizing the difficulty NIC experienced in its 1997 survey of supermax facilities, King (2000, p.
171) defined three essential elements of supermax housing:
(i)
accommodation which is physically separate, or at least separable, from other units or
facilities, in which
(ii) a controlled environment emphasizing safety and security, via separation from staff and
other prisoners and restricted movement, is provided for

INTRODUCTION

1

(iii)

prisoners who have been identified through an administrative rather than a disciplinary
process as needing such control on grounds of their violent or seriously disruptive
behavior in other high security facilities.

Role of Administrative Segregation in Today’s Prisons
Despite accusations that supermax prisons are a needlessly harsh deterrent tactic and an answer to
politicians’ “tough on crime” platforms, prison officials find administrative segregation a necessary
component of effective prison management. Their increasing popularity is a direct result of their success in
maintaining order within the larger prison environment.
Traditionally, prisons have employed the dispersion model (Hershberger, 1998; NIC, 1999). As such,
problem inmates are scattered throughout the correctional system, distributing the burden across facilities and
staff. Often these inmates are moved around the prison system to disrupt their alliances and provide
temporary staff relief (Hershberger). This model is obviously limited by the number of prisons available at the
proper custody level (NIC). In contrast, the concentration or consolidation model centralizes high risk,
dangerous inmates in a tightly controlled facility (Hershberger; NIC). The supermax concept is the archetype
of this model.
Circumstances under which inmates are placed in administrative segregation may be either involuntary
or voluntary. Inmates who exhibit violent or disruptive behavior while incarcerated among the general prison
population are placed in administrative segregation involuntarily. Inmates who are at risk of being harmed by
other inmates are also placed in administrative segregation. The second type is called protective custody, and
they are housed under the same rules and privileges as involuntary cases. One study found that in 90% of the
cases, protective custody inmates themselves requested to be placed into administrative segregation (Carriere,
1989).
Involuntary administrative segregation placements are usually made based on observable behavior. NIC
(1999, p. 6) indicated that “inmates who have demonstrated that they are chronically violent or assaultive,
who present a serious escape risk, or who have demonstrated a capacity to incite disturbances or otherwise
are threatening the orderly operation of the general population institution may become target populations.”
The greater challenge to prison administrators is to identify inmates as a risk prior to a serious incident, with
the understanding that prediction strategies are unreliable and subjective. Therefore, threats are viewed as
objective behavior that may provide the foundation for an administrative segregation decision.
Supermax is characterized as a “minimum privilege, maximum control facility” (Hershberger, 1998, p.
56) in which solitary confinement is the primary security measure. Most supermax facilities confine inmates to
their cells for 23 hours per day, allowing just 1 hour for personal hygiene and exercise. Inmate movement is
severely restricted, with multiple restraints placed on inmates before leaving their cell. Personal contact is kept
to a minimum. Even contact with staff is limited; therapy and worship services may be provided through
videoconferencing or secure barrier. Visitations are allowed on a limited basis, but are generally no-contact,
meaning there is a physical barrier between inmates and visitors.
Even though staff contact and inmate movement is severely restricted, administrative segregation
facilities are equipped with more staff and security devices than typical prison facilities. Many of the supermax
facilities boast state-of-the-art prison technology (Berge, Geiger, & Whitney, 2001; Hershberger, 1998),
thereby reducing personal contact even further. Nonetheless, more staff are needed due to multiple-officer
escorts, increased supervision and searches, and individualized services (e.g., providing meals) at cell doors
(NIC, 1999).
Newer supermax facilities tend to be self-contained in order to circumvent transportation issues. Many
prisons deliver a range of programs and services at the inmates’ cell or within the facility. The usual prison
capacity for medical and dental services may be expanded for emergency medicine and minor surgeries to
keep inmates within the confines of the facility (Berge et al., 2001; Kurki & Morris, 2001; Shepperd & Geiger,
1996).
Release procedures may vary by supermax facility, but behavioral compliance with institutional rules
generally dictates the conditions under which an inmate may be released (NIC, 1999). In order to move to
lower custody situations, inmates are informed of the conditions under which they may be released. NIC

INTRODUCTION

2

recommends regular assessments to provide adequate rationale for retaining inmates within an extended
control facility.
Supermax facilities are more costly to operate, but those costs are offset by simplified staff training as
well as staff and inmate perceptions of increased safety (Hershberger, 1998). A lessening of rigid controls
within the general prison population affords those individuals greater freedom and access to programs. It is
believed that administrative segregation is also a deterrent to other inmates, promoting increased safety
among the remaining population (Ward, 1999).

Objections to Administrative Segregation
Administrative segregation is inarguably the most restrictive environment used to incarcerate inmates,
giving way to concerns about prisoner’s psychological adaptation to solitary confinement. Prison officials are
criticized for resorting to this tactic to maintain safety. The decline of the rehabilitation movement, coupled
with inadequate staff training, is believed to be strongly correlated with the increase in administrative
segregation (King, 2000; Toch, 2001). Critics also suggest that the need for super-maximum facilities was
overestimated, resulting in pressures to fill expensive high custody beds with lower risk inmates (King; Kurki
& Morris, 2001).
The Human Rights Watch (2000) suggests that four factors related to supermax confinement
potentially violate a person’s constitutional rights. Human Rights Watch does not outwardly oppose solitary
confinement of all inmates. Instead, their argument focuses on loose eligibility criteria, extreme harshness of
conditions, duration of confinement, and abuse by staff.
Correctional authorities have been accused of using vague or broad placement criteria. It is not
required that corrections afford inmates the same due process as punitive segregation because it is an
administrative decision rather than a disciplinary action, bringing into question the appropriateness of
placements. Administrative segregation is suitable for violent, dangerous inmates who present imminent risk.
Mentally ill inmates may find themselves placed in administrative segregation, because of a lack of other
suitable placements, protective custody reasons, or disruptive behavior related to their mental illness.
However, less restrictive environments are advised for nuisance inmates, inmates with an accumulation of
minor, nonviolent disciplinary infractions or those in need of protective custody (Human Rights Watch, 2000;
Kurki & Morris, 2001; NIC, 1999; Toch, 2001). Some even contend that placement based on gang member
status is inappropriate (Haney, 2003; Toch).
Many find the conditions of solitary confinement to be excessively harsh and inhumane (Haney, 2003;
Human Rights Watch, 1997, 1999, 2000; King, 2000; Kurki & Morris, 2001; Toch, 2001). The list of the
unacceptable conditions, although not exclusive, includes: lack of windows, 24-hour lighting, lack of exercise
in general and outdoor recreation in particular, limited personal contact, denial of reading materials or other
meaningful activity, and limited therapeutic services. It is believed that many of these conditions are in place
for punishment rather than actual security reasons.
The duration of administrative segregation typically outlasts disciplinary segregation, extending years
rather than months. The related concerns are broadly outlined as extreme solitary confinement and limited
sensory stimulation that leave the inmate ill-equipped to reintegrate into the prison culture or society at large
(Human Rights Watch, 2000). Furthermore, lengths of stay may be extended due to minor infractions of the
rules, resulting in a punishment disproportionate to the seriousness of the behavior. Ultimately, corrections
staff have complete discretion over an inmate’s release from administrative segregation.
Supermax facilities are characterized by the complete control exerted over inmates by correctional staff
(Hershberger, 1998; Human Rights Watch, 2000). The typical “we-they” prison dynamic between inmates and
staff is exacerbated in supermax settings where inmates have little control over their environment (Human
Rights Watch; Kurki & Morris, 2001; NIC, 1999), thereby introducing staff abuse into the realm of
possibilities. Prisoner abuses have been discovered and punished in administrative segregation settings (Kurki
& Morris), but in other situations Human Rights Watch (p. 4) found that “management has tacitly condoned
the abuse by failing to investigate and hold accountable those who engage in it.”

INTRODUCTION

3

Constitutionality of Supermax
Most of the legal challenges surrounding administrative segregation have been on the grounds of a
Fourteenth or Eighth Amendment violation. The Fourteenth Amendment of the U.S. Constitution
guarantees the rights of its citizens, stating “…nor shall any State deprive any person of life, liberty, or
property, without due process of law; nor deny to any person within its jurisdiction the equal protection of
the laws.” Therefore, the state must adhere to certain procedures in deciding to deprive inmates of their
liberty interest (Collins, 2004). The greater the potential loss of liberty (e.g., involuntary medication), the more
stringent are the due process procedures required.
The need for procedural due process rests on the answer to whether administrative segregation creates a
liberty interest. Pursuant to a 1983 Supreme Court decision, due process was necessitated if the institutional
rules were written such that they created a liberty interest (Hewitt v. Helms). Unfortunately, this rule
discouraged institutions from delineating the conditions under which an inmate would be placed in
segregation (Collins, 2004). Subsequently the Supreme Court created a new rule in 1995 that a liberty interest
is created only when there is an “atypical and significant deprivation in relation to the normal incidents of
prison life” (Sandin v. Conner). Hence, segregation that does not pose an atypical and significant hardship is not
subject to due process, including such confinement that may occur during a period of investigation into
inmates’ misconduct (Jones v. Baker, 1998).
At what point is administrative segregation considered an atypical and significant hardship? The most
compelling challenge of this phrase occurred with a class action suit against the Ohio State Penitentiary
(Austin v. Wilkinson, 2002) which produced a ruling contrary to previous cases. The judge ruled in favor of the
plaintiffs stating that their due process and liberty interest had been violated. Segregation was deemed atypical
and a significant hardship because the combination of conditions were significantly more restrictive than
other Ohio state correctional facilities (e.g., isolation, lack of control over heating and lighting, no outside
recreation) and because of the length of confinement. The court upheld the Hewitt v. Helms decision (1983)
that these inmates were entitled to minimal procedural requirements, specifically timely notice of an
administrative segregation evidentiary hearing, reason for confinement, and sufficient opportunity to respond.
The bounds of the Fourteenth Amendment have been questioned in other circumscribed lawsuits. A
New York court found that periodic review of inmates’ confinement is required (McClary v. Kelly, 1998);
extended confinement in segregation without a review hearing is a constitutional violation. Furthermore, in
Wright v. Smith (1994), a New York court ruled that inmates who refuse protective custody or are denied their
request for it will be provided a hearing within 14 days.
Despite these rulings, there is awareness at the judicial level that prison officials should be afforded
enough flexibility and latitude to swiftly manage a volatile environment. The reluctance to unilaterally impose
procedural due process for administrative segregation placements represents a legal balancing act between the
needs of the institution and individuals’ interests. The courts must simultaneously ensure fair treatment of all
inmates.
The Eighth Amendment ensures prisoners protection from cruel and unusual punishment. Because this
concept is subjective, the Supreme Court has established the following standards as the yardstick: (a) shocks
the conscience of the Court, (b) violates the evolving standards of decency of a civilized society, (c)
punishment that is disproportionate to the offense, and (d) involves the wanton and unnecessary infliction of
pain (Collins, 2004). Furthermore, prison officials must be shown to demonstrate “deliberate indifference” to
a prisoner’s basic human need. Cases concerning conditions of confinement, which fall under the Eighth
Amendment, must consider the totality of circumstances where each individual condition might not be a
violation but the combination of conditions might constitute one.
Madrid v. Gomez (1995) was a landmark case that found California state officials in violation of the
Eighth Amendment by housing mentally ill inmates in the SHU at Pelican Bay State Prison. Although the
SHU was not considered a violation for all inmates, the totality of conditions for certain subgroups in this
particular setting over extended periods of time did. The unit in question was considered a modern
forerunner of supermax prisons; unsanitary conditions or antiquated buildings were not in question. The
court reasoned however: “For [mentally ill] inmates, placing them in the SHU is the mental equivalent of
putting an asthmatic in a place with little air to breathe…Such inmates are not required to endure the horrific
suffering of a serious mental illness or major exacerbation of an existing mental illness before obtaining

INTRODUCTION

4

relief.” Not only was it ruled cruel and unusual punishment to place mentally ill inmates in the SHU, those at
reasonably high risk of suffering mental illness as a result of SHU conditions were also restricted.
In another benchmark case regarding conditions of confinement (Ruiz v. Johnson, 1999), a Texas judge
ruled that the “extreme deprivations and repressive conditions of confinement of Texas’ administrative
segregation units…violate the Constitution of the Unites States’ prohibition against cruel and unusual
punishment, both as to the plaintiff class generally and to the subclass of mentally ill inmates housed in such
confinement.” Although administrative segregation itself was not deemed unconstitutional, the deprivation
of “even the most basic psychological needs” such as human contact, psychological stimulation, and human
dignity was. Mentally healthy individuals would decompensate under such conditions; the symptoms and
responses are exacerbated for mentally ill inmates. Upon investigation into the plaintiffs’ claims, experts
testified on “a world in which smeared feces, self-mutilation, and incessant babbling and shrieking are almost
everyday occurrences.” Warehousing mentally ill offenders in administrative segregation was found an
obvious violation of the U.S. Constitution.
Even more despairing was the evidence from this trial that Texas officials acted with sufficient
culpability in regards to housing mentally ill inmates in administrative segregation. Judge Justice wrote in his
ruling: “Whether because of a lack of resources, a misconception of the reality of psychological pain, the
inherent callousness of the bureaucracy, or officials’ blind faith in their own policies, [Texas Department of
Correctional Justice] has knowingly turned its back on this most needy segment of its population.”
A class action suit brought against the Wisconsin prisons was settled out of court in 2001 (Jones ‘El v.
Berge). Similar to the Ruiz v. Johnson (1999) case, individual conditions did not constitute cruel and unusual
punishment, but the totality of the circumstances did (e.g., cell temperatures, nocturnal lighting, lack of
outdoor recreation). State officials agreed to a number of conditions (e.g., regulated cell temperatures,
reducing nocturnal lighting by 60%, inmate choice of indoor or outdoor recreation, minimum of 5 hours per
week out of cell activity), which included not placing severely mentally ill inmates in supermax as a routine
procedure and providing adequate mental health services to ameliorate the effect of the setting on an inmate’s
illness.
Finally, another lawsuit – Rasho v. Snyder – has been filed against the Tamms Prison in Illinois, stating
violations of mentally ill offenders’ rights under the Eighth and Fourteenth Amendments: exposing mentally
ill prisoners to excessively harsh conditions amounts to cruel and unusual punishment and unlawful
discrimination. This case is currently pending trial in the Illinois Southern District court.

Effects of Supermax on Psychological Functioning
Certainly, this discussion evokes a strong emotional response from proponents and opponents alike.
Despite the preponderance of supermax literature, it is more theoretically-driven than empirically-based.
There exists a void of even basic statistics on supermax inmates. Still less is known about the psychological
consequences of solitary confinement. Furthermore, the limited empirical studies available are wrought with
methodological flaws.
The extreme polarity of researchers makes it nearly impossible to interpret research findings; weak
designs or findings are prone to subjective interpretations that support researchers’ beliefs. External
researchers often tend to be advocates for inmates; internal researchers typically operate from an equally
biased, albeit opposing, perspective (Ward, 1995). University researchers perhaps offer the greatest
opportunity for objective research, but they may be perceived as naïve and easily swayed by either inmates or
staff (Ward).
A significant body of literature devoted to administrative segregation is qualitative in nature (Benjamin
& Lux, 1975; Human Rights Watch, 1997, 1999; King, 2000; Kurki & Morris, 2001), using case study designs
and personal accounts. These studies usually involve prolonged periods of on-site observation and interviews
with staff and inmates. They provide the reader with minute details of every day life in solitary confinement
and descriptions of psychological anguish that force the reader to consider issues and perspectives they might
not otherwise obtain.
Case study and similar designs are perhaps most useful for developing theories that can be tested
empirically. Nonetheless, there are serious limitations to be considered. Small sample sizes, as are the norm in
case studies, mean findings may not generalize to all, or even most, segregated offenders. Particularly

INTRODUCTION

5

concerning is that sampling procedures are often not discussed, suggesting that special care was not taken to
select a representative sample. Additionally, these approaches do not provide a relative comparison of the
subjects’ behavior in other settings. Inmates who report serious psychological difficulties in segregation may
experience those same problems in other prison settings or in the community at large.
Researchers may reference older studies that used inadequate study groups, environments, or
experimental stimulus to support their claims that segregation is or is not psychologically harmful (see
Benjamin & Lux, 1975; Pizarro & Steinus, 2004; Suedfeld, Ramirez, Deaton, & Baker-Brown, 1982; Zinger et
al., 2001). Inmates in supermax settings have been compared to prisoners of war (POWs), yet the situation of
isolated and tortured POWs held indefinitely without trial with barred communication does not adequately
portray modern-day prison inmates or their experiences. College students or inmates who volunteer to timelimited segregation stays neither represent the typical inmates placed in segregation against their will nor are
they subjected to indeterminate or lengthy periods of segregation (Benjamin & Lux). Furthermore, studies
conducted in field or laboratory settings that try to emulate a segregation unit, particularly sensory deprivation
and isolation studies, have more severe conditions than those found in today’s supermax settings. When
research participants, study environments, and experimental stimulus differ so radically from the true
environment in question, it is difficult to generalize the findings to the population in question.
Direct studies of solitary confinement, conducted with inmates in administrative segregation, are more
valuable for understanding the population, although they are not without their limitations.
Haney (2003) randomly selected 100 SHU prisoners at Pelican Bay Prison for assessments by interview.
He found elevated symptoms of psychological trauma (e.g., anxiety, headaches, impending nervous
breakdown, lethargy) within the sample as compared to a national probability sample. This study also
demonstrated a greater prevalence rate of psychopathological features (e.g., ruminations, social withdrawal,
irrational anger) than the population at large. However, this study involved only one study group, measured at
a single interval. It is not surprising that prisoners as a whole, not just those in segregation, would differ
dramatically from a non-clinical, non-incarcerated sample. Furthermore, because change over time was not
assessed in this design, it can not be determined whether segregation produced an increase in symptoms.
Comparison groups are necessary to determine whether the differences are attributable to the
experimental stimulus (supermax confinement). Studies conducted on administrative segregation which
include an inmate control group provide a better comparison. In one study, segregated prisoners were not
found to differ in significant ways from a control group, although prisoners in general were different from
standardized samples across multiple measures (Suedfeld et al., 1982). The researchers reported a moderate
correlation of length of stay in segregation to depression (r = .35) and hostility (r = .47), however they did not
specifically examine differential change in groups over time. In another study, Canadian offenders in
segregation were compared to randomly selected non-segregated offenders (Motiuk & Blanchette, 2001). This
study found that the segregated group had more criminal justice system involvement, poor education, skills
deficits, family dysfunction, antisocial attachments, chemical dependencies, thinking problems, and antisocial
attitudes than non-segregated offenders. Furthermore, segregated offenders had a higher recidivism rate than
the non-segregated offenders. A third study found that severe mental disorders were higher among
segregation populations than the general population, particularly schizophrenia and bipolar disorder (Hodgins
& Cote, 1991). Major depression was lower in segregation than the general population, and suicide attempts
were of equal proportion between samples.
Establishing differences between groups is important for understanding the population and the types
of services needed in administrative segregation, but it does little to explain whether the supermax
environment is making them worse or if those with the greatest psychological disturbances are sent to
supermax. If the goal is to determine the harm that supermax prisons inflict, then it is necessary to evaluate
their psychological functioning over the course of their confinement. Longitudinal studies, therefore, are
essential to assess change over time.
Danish inmates in solitary confinement were compared to non-segregated offenders (Andersen,
Sestoft, Lillebaek, Gabrielsen, Hemmingsen, & Kramp, 2000), and the results indicated that psychiatric
disorders were higher among offenders in solitary confinement than those not segregated. However, those
disorders included primarily adjustment and depressive disorders rather than psychotic disorders. Because of
releases and transfers from solitary confinement, the 228 subjects at the beginning of the study declined to 14

INTRODUCTION

6

within 3 months. In another longitudinal study, Zinger and his colleagues (2001) found that mental health and
psychological functioning did not deteriorate over time, although segregated offenders had psychological
indices that were often elevated over non-segregated offenders. This study was limited to a 60-day period, and
it suffered a 40-44% refusal rate (depending on group) and a 56% attrition rate. Although longitudinal design
has many advantages over others, extremely high refusal and attrition rates bring into question the
generalizeability of its findings to administrative segregation overall. Additionally, short durations may not
fully represent the experiences of most offenders in long-term solitary confinement.
Taken together, these findings suggest that inmates in administrative segregation are different from
their peers in the general prison population, particularly in regards to their psychological functioning. What is
still not known is if these differences are attributable to harmful effects of solitary confinement or if inmates
with more serious psychological problems are being placed into administrative segregation at a
disproportionately high rate.

Administrative Segregation in Colorado
Colorado began large-scale use of administrative segregation with the genesis of Colorado State
Penitentiary (CSP) in 1993. In 1998, Colorado reported 5.6% of its population was housed in administrative
segregation as compared to the national average of 1.8% (King, 2000). Only three other states reported a
higher concentration of inmates in supermax beds; even the BOP had less than 1% in administrative
segregation. A major contributing factor to Colorado’s high rates include its zero-tolerance policy towards
prison gang activity, as established by Colorado Revised Statute 17-1-109. This statute empowers wardens to
take reasonable measures to confine persons with known gang associations and to prevent recruitment of
new members.
Despite Colorado exceeding national averages, another larger prison modeled after CSP is scheduled
for construction. Yet, it remains unknown how many potential violent acts have been diverted with
administrative segregation as a management tool.
At a national level, corrections professionals are not making as strong a case for administrative
segregation as opponents are making against it. The constitutionality of it remains in question, with NIC
(1999) issuing strong cautions to corrections regarding its use and the judicial system voicing acrimonious
reactions against its implementation. NIC (p. 22) reports “Typically, ‘new’ programs in the field of corrections
are not based on extensive research. Some are born of emerging needs; some are created in reaction to a crisis
or emergency; others are the result of political agendas.”
Colorado’s liberal use of administrative segregation coupled with national outcries against supermax
underscore the critical need for empirically-based research. None of the debates are grounded in rigorous
scientific theory because virtually no empirical data exists. It is in the best interest of corrections to conduct
internal research to understand the nature of administrative segregation; doing so can reduce the likelihood of
the court system making decisions about prison operations as well as ensure that best practices are
implemented.
The corrections community needs to carry a stronger voice at a national and local level, by initiating
unbiased empirical research that is not tainted by the political underpinnings inherent in previous studies.
Research needs to explore who is assigned to administrative segregation and for what reasons, the impact of
solitary confinement for long durations, effects on special populations, and the adherence of operating
procedures to nationally recognized standards.

Present Study
A prospective research design is recommended to specifically examine changes in psychological
functioning of mentally ill and non-mentally ill inmates as a result of administrative segregation placements.
The nature of that study will require extensive planning, resources, and collaboration; a repeated measures
design with a carefully selected comparison group(s) is recommended. It is further recommended that a
process evaluation be conducted for the two larger administrative segregation facilities in Colorado.
The present study seeks to inform the design for a subsequent prospective research project while
providing basic statistics about Colorado administrative segregation. This study first compared Colorado

INTRODUCTION

7

Department of Corrections’ (CDOC) administrative segregation population to the overall prison population,
across demographic characteristics, criminal history, institutional behavior, needs levels, and standardized
psychological testing. The purpose was to understand who are the inmates serving their sentence under
administrative segregation conditions. Secondly, this study included a longitudinal design to examine length of
stay, return to administrative segregation, recidivism rates, and institutional behavior. Finally, an analysis of
administrative segregation hearings was conducted.

INTRODUCTION

8

STUDY 1: POPULATION COMPARISONS
Method
Participants. The present study compared 981 administrative segregation inmates to the overall prison
population (N = 16,171), excluding inmates under community supervision. Administrative segregation
inmates were identified through classification levels on June 30, 2003. Of these, 11 were on the waitlist and
not yet placed in administrative segregation. The remaining administrative segregation inmates were in one of
four facilities: CSP, Sterling Correctional Facility (SCF), San Carlos Correctional Facility (SCCF), and Denver
Women’s Correctional Facility (DWCF).
Materials. The Level of Service Inventory – Revised (LSI-R; Andrews & Bonta, 1995) is a semistructured interview that assesses criminal risk. Information obtained in the interview is verified through
official offender records and other sources. Each item is scored using a coding system of either 0 or 1, with a
1 indicating that an item is true. The resulting overall LSI-R score can range from 0 to 54 and is used to assign
the level of supervision for community-based offenders and to determine allocation of services (Motiuk,
Motiuk, & Bonta, 1992). The LSI-R showed moderately strong predictive validity (r = .31) for 1-year
recidivism rates with Colorado parolees (O’Keefe, Klebe, & Hromas, 1998).
The MCMI-III (Millon, Davis, & Millon, 1997) consists of 175 true/false items. The inventory
provides diagnostic information in the areas of personality disorders and clinical syndromes. Internal
consistency for the clinical scales ranges from .66 to .90 with 20 of the 26 scales having alpha coefficients in
excess of .80. Test-retest reliability coefficients for the subscales ranged from .82 to .96 (Millon et al.). The
MCMI-III was only administered to inmates incarcerated after 1995.
The Culture Fair Intelligence Test (CFIT; Cattell & Cattell, 1973) is a non-verbal measure designed to
assess general mental capacity in terms of fluid ability, meaning the ability to perceive relationships, to
analyze, and to reason in abstract or novel situations. The goal of this measure is to use items which are free
of cultural bias usually associated with language, cultural background, and educational level. The CFIT is a
multiple-choice paper and pencil test administered in group settings. It consists of four subtests, each
measuring a different area; these areas require the examinee to make classifications, complete series, solve
matrices, and evaluate conditions, all of which are perceptual tasks. CDOC uses a version of the CFIT which
is designed for adults in the average range of intelligence. Conversion tables are used to change the raw CFIT
scores into normalized standard IQ scores. Internal consistency reliability estimates vary between high .70s to
.90s depending on the scale. Test-retest reliabilities run in the low .80s and equivalent-forms reliabilities range
from .58 to .72 (Koch, 1992; Tannenbaum, 1965). The CFIT’s convergent validity with other intelligence tests
has an average correlation of .70 (Koch).
The Brief Psychiatric Rating Scale (BPRS; Overall & Gorman, 1962) is a 24-item scale most commonly
used to assess patients with psychiatric disorders. It is designed to allow for the rapid review of changing
symptoms (Lukoff, Nuechterlein, & Ventura, 1986; Ventura, Lukoff, Nuechterlein, Liberman, Green, &
Shaner, 1993). It measures positive symptoms, general psychopathology and affective symptoms. Some items
can be rated after observation of the patient; others require clinical interview to obtain the patient’s self report
information. Each of the 24 symptom constructs are rated on a 7 point scale of severity ranging from 1 (not
present) to 7 (extremely severe). If a specific symptom is not rated, ‘NA’ is marked indicating it was not assessed.
The research conducted, on both the older 18 item BPRS and the newer 24 item expanded version, has
indicated that there are five factors (or scales) to which the individual items are associated; these are thinking
disorder, withdrawal, anxiety/depression, hostility/suspicion, and activity (Burger, Calsyn, Morse,
Klinkenberg, & Trusty, 1997; Hedlund & Vieweg, 1980). A study measuring the psychological disability of
clients attending self help agencies found the 24-item scale had internal consistencies of .79 at baseline and
.74 after a 6 month follow up (Segal & Silverman, 2002).
The Tests of Adult Basic Education (TABE) – Forms 7 and 8 are designed to measure adult
proficiency in reading, mathematics, language and spelling. Assessment using the TABE gives the information
needed to place learners in the appropriate lessons for their particular skill deficiencies (CTB/McGraw-Hill,
1994). The tests have five levels of difficulty ranging from limited literacy to advanced. Each level consists of 263
items which yield seven subscale scores. Examinees first take the “locator test” which is made up of 25

STUDY 1: POPULATION COMPARISONS

9

vocabulary items and 25 mathematics items; the raw scores from this initial test are then used in determining
what level of the test should be administered. Final scoring of the tests can be done using different systems
which generate grade equivalent (GE) scores and percentile-rank scores. Although the percentile-rank scores
yield more accurate results of the examinees abilities, the GE scores are most easily interpreted. The GE
scores were developed equating TABE standard scores to GE scores on the California Achievement Tests.
Technical aspects of the TABE — Forms 5 and 6 (an older version of the tests) shows Kuder-Richardson
Formula 20 (KR20) reliabilities ranging from .71 to .94 (Bauernfeind, 1992). A technical report examining the
TABE 7 and the 2002 Tests of General Educational Development (GED) shows correlations between the
scores on corresponding content areas on the TABE and GED as ranging from .52 to .57. The correlation
between the TABE total battery score and the GED average score was .63 (CTB/McGraw-Hill, 2004).
Procedure. Demographic, criminal history, and needs assessment data were obtained from administrative
information in the CDOC database. Inmates are routinely processed through the diagnostic unit, and data is
gathered through various sources including official records, diagnostic interview, and pencil-and-paper tests.
Resulting from the diagnostic assessment process are ratings across different needs levels, including academic,
vocational, sex offender, substance abuse, medical, psychological, assaultiveness, self-destruction, and mental
retardation or developmental disabilities (MR/DD). Each level is rated on a 5-point scale, where scores of
three and higher are indicative of problem areas. Levels may be reevaluated during offenders’ incarceration.
The psychological needs level has two parts – the 5-point rating scale as well as a qualifier. Similar to
the other scales, a rating of three or greater indicates the need for mental health services. Offenders who have
an elevated psychological rating need a diagnostic assessment, have been identified as mentally ill, or are
experiencing serious emotional distress not related to a pervasive mental illness (e.g., divorce, grief over loss
of child). The qualifier on the psychological needs rating clarifies whether a qualifying CDOC mental illness
exists (‘C’ for chronic or ‘O’ for organic), further assessment is warranted (‘T’ for temporary or rule-out
diagnosis), or a non-qualifying disorder is present (‘N’ for non-qualifying).
Within CDOC, a C or O qualifier denotes an offender with mental illness (OMI) and includes the
following disorders: bipolar mood disorders, major depressive disorder, depressive disorder not otherwise
specified, dysthymia, paranoid/delusional disorders, schizophrenic disorders, schizophreniform disorder,
schizo-affective disorder, psychotic disorder not otherwise specified, induced psychotic disorder, brief
reactive psychosis, dissociative identity disorder, post-traumatic stress disorder (PTSD), and cluster A
personality disorders (schizoid, schizotypal, and paranoid). These disorders, selected from the Diagnostic and
Statistical Manual – IV (American Psychiatric Association, 1994), characterize individuals who experience the
greatest perceptual distortions or mood disorders, which require more frequent monitoring and treatment
than other psychiatric diagnoses.
Several standardized tests are administered in the diagnostic unit to obtain a baseline measure: the LSIR , MCMI-III, CFIT, and TABE. Violent offenders are defined as serving their current prison term for a
violent crime as defined in CDOC’s annual statistical report (Rosten, 2004, p.70).
Institutional behavior, such as violations of the code of penal discipline (COPD) and involvement in
security threat groups (STG), are recorded electronically over the course of an offender’s incarceration. There
are three levels of STG involvement: member, associate, and suspect. Levels are ascertained by field
intelligence officers who rate offenders’ involvement across 11 items (e.g., self admission, moniker, gang
tattoos, identification by law enforcement). Each item carries a weight ranging from 5 to 20 points, and
summative scores determine STG involvement.
Psychological data, such as the BPRS and psychological needs level, are updated periodically during an
offender’s incarceration. The BPRS is administered only to diagnosed mentally ill offenders, ideally at 6month intervals or more often as needed. The psychological needs level may or may not change, depending
on the degree of psychological disruption an inmate displays.

Results and Conclusions
Population Comparisons. Descriptive statistics were generated for the entire prison population, for the
administrative segregation sample, and for each administrative segregation facility sample (see Table 1). Oneway chi-square and t tests were conducted between each sample and the general population, using an alpha
level of .01. Significant differences are highlighted.

STUDY 1: POPULATION COMPARISONS

10

Table 1. Population Comparisons across Demographic, Criminal History & Institutional Behavior Factors
Ad Seg
CSP
SCF
SCCF
Population
(N = 16,171)

DWCF

(n = 734)

(n = 180)

(n = 26)

(n = 30)

100%
0%

100%
0%

100%
0%

0%
100%

37%
17%
42%
4%
31.6 (8.4)

36%
17%
43%
4%
31.8 (8.5)

41%
15%
40%
4%
30.2 (7.8)

39%
15%
42%
4%
33.6 (9.2)

33%
37%
27%
3%
32.0 (8.4)

41%
36%
23%

59%
26%
15%

58%
27%
15%

58%
30%
12%

85%
4%
11%

52%
24%
24%

21%
48%
31%

9%
59%
32%

9%
62%
29%

9%
55%
36%

9%
35%
56%

25%
21%
54%

8%
23%
59%
10%

9%
23%
62%
6%

8%
21%
65%
6%

13%
23%
57%
7%

14%
41%
45%
0%

18%
36%
32%
14%

73%
19%
8%

71%
22%
7%

71%
22%
7%

74%
22%
4%

65%
27%
8%

70%
23%
7%

85%
14%
1%

89%
9%
2%

91%
7%
2%

88%
11%
1%

88%
12%
0%

74%
23%
3%

11%
5%
10%
74%
48%
10%
4.4 (7.8)

40%
5%
21%
34%
65%
32%
16.9 (16.0)

43%
5%
22%
30%
69%
33%
17.6 (16.5)

37%
7%
23%
33%
51%
22%
14.2 (12.4)

4%
8%
8%
80%
69%
69%
20.5 (24.6)

3%
0%
17%
80%
40%
17%
13.7 (11.5)

2.4 (4.6)

8.5 (7.6)

8.2 (7.5)

9.4 (8.0)

9.8 (7.1)

7.9 (8.1)

4.2 (4.6)

5.8 (4.4)

6.2 (4.3)

4.6 (3.8)

7.4 (8.1)

3.5 (2.9)

33.6 (7.4) 33.6 (7.2) 32.5 (7.4)
35.2 (7.7)
Note. Highlighted values indicate the sample is statistically different (α = .01) from the CDOC inmate population.

37.7 (8.9)

Gender
Male
Female
Ethnicity
Caucasian
African American
Hispanic
Other
Mean age (SD)
Marital status
Single
Married/common law
Div/sep/wid
High school
Diploma
Equivalency test
Neither
Highest grade completed
Grade school
Less than 12th grade
12th grade
Beyond high school
Prior incarcerations
None
One
Two or more
Status type
New commitment
Revocation
Other
STG member status
Member
Associate
Suspect
None
Violent offender
% with previous ad seg
Mean # COPDs this
incarceration (SD)
Mean # punitive seg this
incarceration (SD)
Mean years served this
incarceration (SD)
Mean LSI-R (SD)

93%
7%
45%
22%
30%
3%
35.4 (10.4)

30.0 (8.2)

(n = 981)

97%
3%

In comparison with the population, administrative segregation inmates tended to be male, Hispanic,
younger, not married, violent, possess a GED as opposed to high school diploma, have STG involvement,
and spent a longer period under CDOC supervision for the current incarceration. They were three times

STUDY 1: POPULATION COMPARISONS

11

more likely to have had a prior placement in administrative segregation than were inmates in the general
population. Their frequency of COPD infractions and punitive segregations over this incarceration was more
than triple of the general population. Furthermore, administrative segregation inmates had higher LSI-R
scores. Other significant differences were found; however, the differences were slight and appeared to be of
limited practical significance.
The ways that inmates in individual administrative segregation facilities differed from the general
population was also explored. Gender was an obvious difference because prisons are gender-specific. The
SCCF sample was dramatically different on marital status from other groups, being highly unlikely to be in a
married relationship. CSP had a significantly higher rate of violent offenders and an overall longer time under
CDOC supervision. It should be noted that SCCF and DWCF may not vary statistically from the general
population as a function of their low sample size.
Prior classification levels were examined to determine where administrative segregation decisions were
most likely to occur. It was found that 70% were close custody, 17% medium, 8% maximum (new cases or
parole violators), 4% minimum-restrictive, and 1% minimum. These figures indicate that the bulk of
placements were already housed in high security units.
The 981 administrative segregation inmates accumulated a total of 16,598 COPD convictions over their
current incarceration, accounting for 24% of COPDs even though they represent only 6% of the population.
Disciplinary actions were examined by type, with class I disciplinary violations representing the most serious
and class III the least (see Table 2). They had a greater frequency of class I disciplinary convictions compared
to the overall population. This finding held true for inmates in three of the four segregation facilities.
The most frequent COPD convictions are listed in Table 3, which represent an accumulation over the
current incarceration through June 30, 2003. A distinction is not made here whether the violations occurred
prior to or during administrative segregation placement. This relationship is explored more fully in study 2.
Table 2. COPD Convictions by Class Type
Population
Ad Seg
CSP
SCF
SCCF
DWCF
N
%
n
%
%
%
%
%
Class I
7,281
10%
2,519
15%
15%
13%
27%
13%
Class II
57,419
81%
13,089
79%
79%
79%
71%
82%
Class III
5,983
9%
990
6%
6%
8%
2%
6%
TOTAL
70,683
100%
16,598
100%
100%
100%
100%
100%
Note. Highlighted values indicate the sample is statistically different (α = .01) from the CDOC inmate population.
Table 3. Most Frequent COPD Convictions
Description
Disobeying a lawful order
Unauthorized possession
Violating posted operational rule
Verbal abuse
Advocating facility disruption
Possession or use of dangerous drugs
Unauthorized absence
Fighting
Threats
Tattooing and/or possession
Assault
Refusal to work
Bartering/selling goods
Damage to property
Other

STUDY 1: POPULATION COMPARISONS

Class
IIB
IIA
III
IIB/III
IIA
II
IIB/III
IIA
IIA
IIA
I
II
IIB
IIA
I - III

Population (N = 70,683)
# Counts
Percent
11,798
16.7%
6,373
9.0%
4,592
6.5%
4,318
6.1%
3,719
5.3%
3,176
4.5%
3,120
4.4%
3,035
4.3%
2,899
4.1%
2,859
4.0%
2,752
3.9%
2,534
3.6%
2,408
3.4%
1,730
2.4%
15,370
21.8%

Ad Seg (n = 16,598)
# Counts
Percent
2,856
17.2%
1,324
8.0%
705
4.2%
1,307
7.9%
1,426
8.6%
357
2.2%
464
2.8%
500
3.0%
1,056
6.4%
502
3.0%
1,149
6.9%
316
1.9%
359
2.2%
770
4.6%
3,507
21.1%

12

Inmates in administrative segregation had more class I violations, which represent a serious threat or
disruption to prison security, than the general population; 83% had at least one class I violation. For the
remainder (n = 167), total number of violations ranged from 0 to 46 with a median of 4. A file review was
conducted for 82 cases that had fewer than the median number of violations. The results indicated that 46
inmates were immediately classified to administrative segregation upon incarceration for the following
reasons: high profile trial, murder conviction, assault on peace officer, prison escape and/or attempts,
behavior in county jail, interstate transfer, or death penalty. Nine cases had previously discharged their
sentence from administrative segregation and review of their cases upon reincarceration warranted placement
at the same custody level. Finally, 27 offenders were placed in administrative segregation for their institutional
behavior on the current incarceration. Reasons for these placements primarily included gang activity or fewer
than four class II disciplinary violations (i.e., fighting, advocating facility disruption, threats).
Summary statistics on standardized tests administered at the Diagnostic Unit are presented in Table 4.
Administrative segregation inmates scored similarly on the Culture Fair IQ test as the overall inmate
population. Although their math and total TABE scores were statistically lower, the difference was slight.
Table 4. Population Comparisons across Standardized Testing
Population
(N = 13,347)

Culture Fair Intelligence Test
TABE Reading
TABE Math
TABE Language
TABE Total

M
SD
101.3
13.4
8.8
3.7
7.2
3.2
7.5
4.0
7.9
3.6
Population
(N = 11,681)

Ad Seg

(N = 666)

M
SD
100.4
13.5
8.6
3.6
6.9
3.1
7.1
4.0
7.6
3.6
Ad Seg
(N = 567)

MCMI–III Personality Patterns (% Base Rate Scores > 75)
1- Schizoid
13%
2a- Avoidant
25%
2b- Depressive
19%
3- Dependent
15%
4- Histrionic
9%
5- Narcissistic
17%
6a- Antisocial
22%
6b- Aggressive
15%
7- Compulsive
6%
8a- Passive-aggressive
23%
8b- Self-defeating
12%
s- Schizotypal
6%
c- Borderline
8%
p- Paranoid
7%
MCMI–III Clinical Syndromes (% Base Rate Scores > 75)
a- Anxiety
35%
h- Somatoform
2%
n- Bi-polar: manic disorder
4%
d- Dysthymic disorder
16%
b- Alcohol dependence
23%
t- Drug dependence
16%
r- Post traumatic stress
8%
ss- Thought disorder
3%
cc- Major depression
5%
pp- Delusional disorder
4%

STUDY 1: POPULATION COMPARISONS

p
n.s.
n.s.
<.01
n.s.
<.01
p

19%
28%
19%
15%
7%
19%
29%
21%
3%
33%
14%
10%
13%
12%

<.001
n.s.
n.s.
n.s.
n.s.
n.s.
<.001
<.001
<.01
<.001
n.s.
<.001
<.001
<.001

41%
2%
5%
18%
27%
19%
10%
5%
7%
9%

<.01
n.s.
n.s.
n.s.
n.s.
n.s.
n.s.
<.01
n.s.
<.001

13

Inmates in administrative segregation demonstrated more traits associated with personality-related (Axis
II) disorders than the inmate population. They had higher MCMI-III scores on schizoid, antisocial,
aggressive, passive-aggressive, schizotypal, borderline and paranoid personality disorders and lower scores on
compulsive personality disorder. There were also differences across three clinical syndromes, specifically
anxiety, thought disorder, and delusional disorder. Interestingly, all of the cluster A personality disorders
(paranoid, schizoid, and schizotypal) were elevated in the administrative segregation sample; cluster A
characterizes individuals who evidence clear signs of oddness or eccentricity, prefer social isolation, and have
difficulty relating to others.
Figure 1 shows the occurrence of CDOC inmates who present with specific needs across 10 domains.
Many of the diagnostic levels may change throughout an offender’s incarceration to reflect current needs,
such as the psychological level, while others (e.g., substance abuse) remain unchanged because they are used
merely as a screening tool. For the levels that are assessed in a dynamic way, the current level as of June 30,
2003 was used.
Administrative segregation inmates did not vary from the prison population on academic, vocational,
substance abuse, sex offender, or medical needs areas. However, they presented with significantly greater
mental health needs, including broad psychological or emotional distress, mental illness as defined by CDOC,
MR/DD, assaultive behavior, and self-destruction or suicide tendencies. Because these particular levels
fluctuate over time, it can not be determined from these comparisons whether their needs were elevated prior
to administrative segregation placement.
Figure 1. Comparison of Administrative Segregation Inmates to General Population Needs
Percent Rated 3 - 5
0%

10%

20%

30%

40%

50%

60%

70%

80%

90% 100%

Academic
Vocational
Sex Offender

Needs Level

Substance Abuse
M edical
Psychological **
OM I **
M R/DD **
Assaultive **
Self-Destruction **
**p < .01

GP

Ad Seg

Note. OMI status is denoted by a psychological code qualifier of C or O.

STUDY 1: POPULATION COMPARISONS

14

Needs levels are presented by segregation facility in Table 5. Percentages indicate inmates scoring three
or higher on the needs level, except the OMI category which is denoted by a ‘C’ or ‘O’ qualifier on the
psychological level. Again, some differences may not be statistically significant because of small SCCF and
DWCF samples, but they may have practical importance.
This data highlights some interesting differences between the populations housed within each facility.
Although the entire administrative segregation sample did not vary from the general population across five
needs areas, females presented substantially greater academic and medical needs and encompassed fewer sex
offenders. The SCCF sample also had greater academic needs than typical inmates. The population at SCCF
was not composed entirely of OMIs because some offenders may have a non-qualifying mental illness, mental
retardation, or developmental disorder, or they may be undergoing diagnostic assessments to rule-out
diagnoses and malingering. Overall, administrative segregation inmates had elevated psychological needs
levels, but the prevalence of OMIs was not elevated at CSP. However, CSP had a higher rate of MR/DD and
self-destructive inmates, as did SCCF, than was the norm. Taken together, these findings may have important
implications for allocation of mental health resources.
Table 5. Needs Levels by Administrative Segregation Facility
CSP
Population
(N = 16,171)

Academic
Vocational
Sex Offender
Substance Abuse
Medical
Psychological
OMI
MR/DD
Assaultive
Self-Destruction

SCF

(n = 734)

(n = 180)

33%
87%
23%
78%
17%
31%
17%
8%
77%
22%

38%
91%
22%
81%
13%
37%
23%
6%
57%
18%

33%
89%
25%
77%
19%
22%
15%
5%
45%
16%

SCCF

(n = 26)

62%
100%
35%
70%
23%
100%
89%
27%
77%
50%

DWCF
(n = 30)

57%
97%
3%
86%
47%
80%
67%
7%
55%
25%

Note. Highlighted values indicate the sample is statistically different (α = .01) from the CDOC inmate population.

STG Involvement. Inmates involved in a STG constituted a disproportionately large number of the
administrative segregation sample. Table 6 presents the most common STG groups in CDOC. A higher rate
of Surenos-13 and Gallant Knights were found in administrative segregation; fewer Crips, Bloods, and White
Supremacists reside in segregation than the general population.
Table 6. STG Affiliation
Name

Population
n
%
Crips
768
18%
White Supremacists
657
16%
Surenos-13
573
14%
Bloods
429
10%
Gallant Knights
262
6%
Security Threat *
233
6%
Prison Gang
141
3%
Folk
141
3%
Other
1,045
25%
Total
4,249 100%
* Non-specific STG threat to facility.

STUDY 1: POPULATION COMPARISONS

Ad Seg
n
%
66
10%
79
12%
153
24%
38
6%
64
10%
20
3%
46
7%
15
2%
170
26%
651
100%

CSP
(n = 511)
10%
12%
25%
5%
10%
3%
7%
3%
25%
100%

SCF
(n = 121)
9%
17%
19%
8%
11%
3%
5%
2%
26%
100%

SCCF
(n = 5)
0%
0%
20%
20%
0%
0%
0%
0%
60%
100%

DWCF
(n = 6)
33%
0%
17%
0%
17%
0%
33%
0%
0%
100%

15

Descriptive data were examined by STG status (see Table 7). Because the number of associates (n = 51)
was so small, they were grouped with the suspect category for analytical purposes. Differences among STG
groups centered primarily on demographics rather than criminal history or institutional behavior, although
inmates with STG involvement have been under CDOC supervision for less time than those with no
involvement. Members were exclusively male and tended to be Hispanic, younger, and single. Inmates with
any STG involvement were more likely to attain a GED than a high school diploma.
Table 7. Demographic, Criminal History & Institutional Behavior by STG Member Status
Associate/
Suspect
None
Member
p
(n = 391)
(n = 260)
(n = 330)
Gender
<.001
Male
100%
98%
93%
Female
0%
2%
7%
Ethnicity
<.001
Caucasian
26%
41%
47%
African American
13%
15%
24%
Hispanic
57%
39%
26%
Other
4%
5%
3%
Mean age (SD)
28.3 (5.7)
30.4 (7.8)
36.4 (9.3)
<.001
Marital status
<.01
Single
63%
55%
57%
Married/common law
26%
31%
22%
Div/sep/wid
11%
14%
21%
High school
<.01
Diploma
6%
9%
15%
Equivalency test
62%
60%
52%
Neither
32%
31%
33%
Highest grade completed
n.s.
Grade school
10%
9%
9%
23%
21%
24%
Less than 12th grade
64%
64%
57%
12th grade
Beyond high school
3%
6%
10%
Prior incarcerations
n.s.
None
74%
67%
71%
One
20%
24%
22%
Two or more
6%
9%
7%
Status type
n.s.
New commitment
90%
88%
90%
Revocation
9%
12%
7%
Other
1%
0%
3%
Violent offender
68%
60%
64%
n.s.
% with previous ad seg
34%
30%
31%
n.s.
Mean # COPDs this
16.5 (13.3)
16.1 (14.2)
18.1 (19.9)
n.s.
incarceration (SD)
Mean # punitive seg this
9.0 (7.7)
8.3 (7.3)
8.0 (7.7)
n.s.
incarceration (SD)
Mean years served this
5.3 (3.2)
5.4 (3.9)
6.6 (5.7)
<.001
incarceration (SD)
Mean LSI-R (SD)
34.2 (6.9)
33.6 (6.8)
32.7 (8.3)
n.s.

STUDY 1: POPULATION COMPARISONS

16

Figure 2 presents comparisons by group across the 10 need levels. Inmates with STG involvement have
similar or lesser needs than those with no involvement. Specifically, they are less likely to be a sex offender,
have mental health or self-destructive concerns, or have medical issues.
Figure 2. Needs Levels by STG Member Status

Percent Rated 3 - 5
0%

20%

40%

60%

80%

100%

Academic
Vocational
Sex Offender **

Needs Level

Substance Abuse
Medical **
Psychological **
OMI **
MR/DD
Assaultive
Self-Destruction **
**p < .001

Member

Associate/ Suspect

None

Mentally Ill. Due to concerns regarding mentally ill inmates in administrative segregation, characteristics
of this population were examined in depth. Administrative segregation inmates were grouped into three
categories based on the existence of a qualifying diagnosis. The first category included OMIs, the second had
an other or non-qualifying diagnosis, and the third had no diagnosis. Inmates categorized with no diagnosis
may indeed have an Axis I diagnosis, but without accompanying mental health needs that warrant increased
attention.
Table 8 presents descriptive data for the administrative segregation sample, grouped by mental health
status. Chi-square analyses showed that OMIs had a greater rate of female offenders than the other two
categories. OMIs and inmates with non-qualifying disorders tended to be Caucasian, had a slightly lower
education level and were significantly less likely to have STG involvement. Inmates with no qualifying
diagnosis had fewer COPDs than those with a qualifying diagnosis. Furthermore, they had lesser needs in
academic, vocational, and medical areas as well as lower self-destruction levels (see Figure 3).

STUDY 1: POPULATION COMPARISONS

17

Table 8. Demographic, Criminal History & Institutional Behavior by Mental Health Status
Other
None
OMI
(n = 210)

Gender
Male
Female
Ethnicity
Caucasian
African American
Hispanic
Other
Mean age (SD)
Marital status
Single
Married/common law
Div/sep/wid
High school
Diploma
Equivalency test
Neither
Highest grade completed
Grade school
Less than 12th grade
12th grade
Beyond high school
Prior incarcerations
None
One
Two or more
Status type
New commitment
Revocation
Other
STG member status
Member
Associate
Suspect
None
Violent offender
% with previous ad seg
Mean # COPDs this
incarceration (SD)
Mean # punitive seg this
incarceration (SD)
Mean years served this
incarceration (SD)
Mean LSI-R (SD)

(n = 137)

(n = 634)

90%
10%

98%
2%

47%
18%
31%
4%
33.0 (8.8)

52%
15%
30%
3%
31.6 (8.7)

55%
23%
22%

53%
31%
16%

61%
26%
13%

13%
46%
41%

8%
57%
35%

9%
63%
28%

14%
27%
50%
9%

10%
26%
60%
4%

8%
20%
66%
6%

p
<.001

99%
1%
<.001
30%
18%
48%
4%
31.1 (8.1)

n.s.
n.s.

<.01

<.01

n.s.
71%
20%
9%

69%
24%
7%

72%
22%
6%

89%
10%
1%

86%
11%
3%

91%
8%
1%

22%
9%
15%
54%
63%
70%
19.2 (18.4)

31%
7%
21%
41%
58%
72%
18.9 (18.2)

48%
4%
23%
25%
66%
67%
15.8 (14.5)

n.s.
n.s.
<.01

9.2 (6.7)

9.2 (9.8)

8.1 (7.3)

n.s.

5.7 (4.8)

5.2 (4.0)

5.9 (4.3)

n.s.

35.0 (7.6)

34.0 (7.8)

33.0 (7.1)

n.s.

n.s.

<.001

STUDY 1: POPULATION COMPARISONS

18

Figure 3. Needs Level by Qualifying Mental Illness Diagnosis
Percent Rated 3 - 5

0%

20%

40%

60%

80%

100%

Academic *
Vocational *

Needs Level

Sex Offender
Substance Abuse
Medical *
MR/DD
Assaultive
Self-Destruction **

* p < .01, **p < .001

OMI

Other

None

Axis I diagnoses, as assigned by mental health clinicians, were examined for the entire CDOC
population; the most frequent diagnoses are listed in Table 9 for the population as well as the administrative
segregation sample. Axis I disorders of administrative segregation inmates ranked similarly to those occurring
within the prison population, except for lower prevalence rates of major depression and other depressive
disorders. Although drug use and dependence are ranked as the top Axis I diagnoses in CDOC, this is an
underestimation of actual substance abuse problems. Because CDOC has a separate assessment process to
identify substance abusers, clinicians oftentimes do not assign a substance abuse diagnosis where one is
warranted.
Mentally ill inmates are routinely assessed on the BPRS, typically at 6-month intervals or more
frequently. Table 10 lists mean BPRS scale and total scores for mentally ill inmates system-wide by custody
level. Overall, it should be noted that scores were very low given the possible range of scores from 24 to 168.
OMIs at close and administrative segregation custody levels had higher BPRS total, thinking disorder,
hostility suspicion, and activity scale scores than the remainder of inmates. Administrative segregation inmates
scored significantly higher than close custody inmates on thinking disorder and hostility suspicion scales.
Mean BPRS scores were examined by facility: 33.99 (SD = 8.18) for CSP, 34.38 (SD = 6.74) for SCF,
43.58 (SD = 11.26) for SCCF and 39.44 (SD = 10.77) for DWCF. A one-way analysis of variance (ANOVA)
found that SCCF inmates had significantly higher scores than the other facilities and DWCF had higher
scores than CSP and SCF, F(4, 417) = 13.09, p < .001.

STUDY 1: POPULATION COMPARISONS

19

These findings indicate inmates in administrative segregation present more psychiatric symptoms than
inmates at lower custody levels, particularly in areas suggestive of psychotic behavior (i.e., grandiosity,
hallucinations, unusual thoughts), hostile and suspicious behavior, and greater activity levels (i.e., tension,
excitement, motor hyperactivity). These findings, compounded by those of the MCMI-III, indicate a relatively
higher prevalence of psychosis, as well as inmates who perhaps prefer solitude, among the administrative
segregation sample.
Table 9. Axis I Diagnoses for Mentally Ill Inmates (N = 4,317)*
Population
Diagnostic and Statistical Manual – IV Categories
N
%
Drug Use/ Dependence
1,067
25%
Major Depression/ Depressive Disorders
760
17%
Bipolar Disorders
684
16%
Dysthymic Disorders
469
11%
Schizophrenia/ Psychotic Disorders
394
9%
Anxiety Disorders/ PTSD/ Phobias
336
8%
Alcohol Use/ Dependence
269
6%
Other Disorders
172
4%
Sexual and Gender Identity Disorders
118
3%
Disorders Usually Diagnosed in Childhood
48
1%
* Sample size reflects diagnoses rather than inmates; inmates may have multiple diagnoses.
Table 10. BPRS Scores by Custody Level (N = 2,498)
Ad Seg
Close
BPRS Scale (score range)
(n = 313)
(n = 612)
M (SD)
M (SD)
Thinking Disorder (5 - 35)
6.66 (2.68) 6.26 (2.68)
Withdrawal (6 - 42)
7.45 (2.09) 7.43 (2.20)
Anxiety-Depression (5 - 35)
8.83 (3.13) 8.94 (3.66)
Hostility Suspicion (3 - 21)
5.27 (2.59) 4.88 (2.66)
Activity (5 - 35)
6.70 (2.51) 6.80 (2.50)
BPRS Total (24 - 168)
36.35 (8.81) 35.22 (9.47)

Medium
(n = 584)
M (SD)
5.82 (1.80)
7.28 (1.90)
9.39 (3.75)
4.21 (1.92)
6.36 (2.04)
33.97 (7.85)

Ad Seg
n
%
93
25%
43
11%
58
15%
50
13%
41
11%
22
6%
15
4%
33
9%
14
4%
7
2%

Min-Restr
(n = 647)
M (SD)
5.87 (2.14)
7.35 (1.81)
9.38 (3.59)
3.98 (1.64)
6.18 (1.96)
33.83 (8.01)

Minimum
(n = 342)
M (SD)
p
5.80 (1.97) <.001
7.35 (2.07)
n.s.
9.23 (3.70)
n.s.
3.83 (1.40) <.001
6.15 (2.07) <.001
33.54 (7.88) <.01

Logistic Regression. A logistic regression was conducted to determine offender characteristics that might
predict administrative segregation placements. The dependent variable was administrative segregation (yes,
no); inmates in the general population who had previously been in segregation were excluded to maximize
group differences. The following variables, which differentiated administrative segregation inmates from the
general population in the univariate comparisons, were added to the logistic regression equation in three
blocks: (1) gender (male, female), age, Hispanic (yes, no), high school diploma (yes, no), single (yes, no), and
OMI (yes, no); (2) LSI-R score, STG involvement (yes, no), number of punitive segregations, number of
disciplinary violations, and violent offender (yes, no); and (3) base rates scores on MCMI-III schizoid,
antisocial, aggressive, passive-aggressive, schizotypal, borderline, paranoid, and delusional disorder scales.
Table 11 gives the results of the analysis; only variables found to be a significant predictor are presented
here. The equation correctly classified inmates at a rate of 80%. STG involvement was the strongest
predictor; the odds ratio indicated that inmates with STG involvement were 4.5 times more likely to be placed
in administrative segregation than those with no involvement. Violent, Hispanic, single, and OMI inmates
have significantly greater odds of such a placement than inmates without those traits.
Interestingly, Hispanic ethnicity remained a significant predictor even when accounting for both
ethnicity and STG involvement. It appears that Hispanic gang members are segregated at a greater rate than
gang members of other cultures. This finding is consistent with the earlier finding of increased placements for
predominantly Hispanic gangs (Surenos-13, Gallant Knights) and fewer placements of African American

STUDY 1: POPULATION COMPARISONS

20

(Bloods, Crips) and Caucasian (White Supremacists) gangs. This may reflect greater violence among Hispanic
gangs, more visible activities, or a greater perceived threat.
Table 11. Logistic Regression Results
Variable
B
STG involvement
1.51
Violent offender
.88
Hispanic
.65
OMI
.42
Single
.38
# punitive segregations
.18
LSI-R
.04
Schizotypal
-.01

p
<.001
<.001
<.001
<.05
<.01
<.001
<.001
<.05

STUDY 1: POPULATION COMPARISONS

Odds Ratio
4.50
2.43
1.91
1.53
1.47
1.19
1.04
.99

21

STUDY 2. LONGITUDINAL ANALYSIS
Method
Participants. Participants included 3,003 inmates who were placed in administrative segregation between
January 1995 and December 2003. Offenders classified as administrative segregation but never moved to
such a facility (e.g., discharged sentence before transfer) were not included. Also excluded from the study
were inmates already in administrative segregation as of January 1995, unless they released and later returned
to segregation in which case only latter episodes were examined.
Death row inmates were excluded from this study.
Participants were mostly male (95%) and were on average 29 years old upon their initial placement.
Caucasian and Hispanic ethnic backgrounds were equally distributed at 37% each, with African Americans
composing 23% of the sample and other ethnic representation totaling 3%.
Procedure. All data were downloaded from the CDOC administrative database. Dates of administrative
segregation, both placements and releases, were based on classifications rather than time served in an
administrative segregation facility. Moves to and from administrative segregation facilities are made according
to available bed space at the proper custody facility; all inmates are held in a punitive segregation environment
prior to transfer into administrative segregation. Their release upon reclassification is also made by bed
availability.
Other data obtained from the database download included segregation facility, release types, COPD
infractions, mental health needs levels, STG involvement, and recidivism dates.

Results and Conclusions
Eighteen percent of participants had more than one administrative segregation episode. Except where
noted otherwise, only the first episode was analyzed to meet the assumption of independent observations.
Length of stay. Time in administrative segregation was examined by episode (see Table 12), including
cases still active as of December 2003. It should be noted that true durations are actually longer than those
represented here. Inmates placed in segregation prior to 1995 (excluded from this study) would tend to
increase the lengths of stay slightly. Because these numbers are small, however, they would have only a small
impact on median months, which is perhaps the most reliable measure of duration. Mean central tendencies
are unduly influenced by a small number of inmates who have quite long lengths of stay.
Length of stay was analyzed by the facility where administrative segregation was served (see Table 13).
Oftentimes, inmates are in multiple facilities as their needs, or the needs of CDOC, change. When a sufficient
sample size was present, data were examined by specific placement patterns (e.g., CSP then SCCF). Inmates
who serve their placement in CSP, whether in that facility alone or in conjunction with another one, have the
longest durations. The longer stays at CSP may be attributable to pro-unit participation. Inmates at the prounit are not confined to the degree of administrative segregation, but re-classification does not occur until
their fifth month in the pro-unit.
Table 12. Length of Stay for 1995 – 2003 Admission Cohort
Months in Administrative Segregation
Episode
N
Median
Mean
Maximum
1
3,003
16.53
21.17
107
2
552
12.45
15.77
73
3
65
9.17
11.85
41
4
8
8.49
8.62
17
5
2
3.79
3.79
5
Total
3,003
18.83
24.35
107

STUDY 2: LONGITUDINAL ANALYSIS

22

Table 13. Length of Stay by Segregation Facility
Months in Administrative Segregation
Facility
N
Median
Mean
Maximum
CSP
1,994
20.16
24.05
107
SCF
479
11.63
10.86
40
SCCF
64
11.25
14.22
79
Females
160
8.87
9.58
37
90
24.95
33.42
102
CSP Æ SCCF
133
20.07
21.53
49
SCF Æ CSP
83
17.91
25.36
87
Other
Months in administrative segregation, accumulated across episodes, was examined by mental health
status and STG involvement. OMIs had shorter durations (Median = 16.03, SD = 16.22) than inmates with no
diagnosis (Median = 19.86, SD = 19.86), but longer stays than inmates coded as other diagnosis (Median =
14.00, SD = 13.95), F(2, 2982) = 24.22, p < .001. STG members had the longest duration (Median = 21.88,
SD = 19.58), while associates/suspects (Median = 16.76, SD = 19.22) had the same length of stay as those
with no affiliation (Median = 17.41, SD = 18.68), F(2, 3002) = 15.47, p < .001.
Release from administrative segregation was categorized into four types: (1) return to general population
(GP) or release for (2) sentence discharge, (3) mandatory parole, or (4) discretionary parole. An additional 5
inmates died in administrative segregation from 1995 to 2003, of which four were from natural causes and
one was a suicide. Release types are presented in Table 14. Inmates predominantly return to GP, but a
substantial percentage (41%) released directly to the community. Inmates granted discretionary parole may
reflect the Parole Board’s desire to have those offenders under community supervision prior to them
discharging their sentence. Table 15 provides release type by administrative segregation facility.
Table 14. Release Type for Administrative Segregation Release Cohort
Sentence
Mandatory Discretionary
Episode
GP
Discharge
Parole
Parole
1
1,455
235
411
173
2
25
93
74
37
3
3
10
10
4
4
2
1
1
0
Totals
1,485
339
496
214
Percent
59%
13%
20%
8%
Table 15. Release Type by Segregation Facility
Sentence
Facility
GP
Discharge
CSP
64%
10%
SCF
60%
13%
SCCF
65%
12%
Females
71%
12%
CSP/SCCF
78%
6%
SCF/CSP
65%
7%
Other
39%
17%

Mandatory
Parole
17%
23%
20%
13%
10%
28%
39%

Totals
2,274
229
27
4
2,534
100%

Discretionary
Parole
9%
4%
3%
4%
6%
0%
5%

Length of stay was found to vary by release type, F(3, 2273) = 57.33, p < .001. The median length of
stay was 20.50 months for release to GP, 17.19 months for mandatory parole, 16.73 months for discretionary
parole, and 13.06 months for sentence discharge. Posthoc analyses indicated that time in segregation was
similar for discretionary and mandatory parole releases, but sentence discharges were correlated with
significantly shorter terms and release to general population meant significantly longer segregation episodes.

STUDY 2: LONGITUDINAL ANALYSIS

23

Release type varied by mental health status and STG involvement, which in part accounts for different
lengths of stay by group. For mentally ill offenders, those categorized as having ‘other’ diagnoses were more
likely to discharge their sentences than release back to population, which corresponds to this group’s shorter
stays, χ2 (6, 2258) = 17.26, p < .01. Associates/suspects were more likely to discharge their sentences than
they were to return to GP, χ2 (6, 2274) = 26.58, p < .001. While this explains why associates/suspects have
shorter segregation stays than members, release type does not account for the different durations between
members and those with no involvement.
Return to Administrative Segregation. In the present study, 18% of offenders released from administrative
segregation returned at some point during the study period. Standardized return rates by release year are
presented in Table 16. Two-year return rates appear more stable over time than one-year rates. However,
across both measures, the lowest rates occurred for inmates released from segregation in 1999.
Figure 4 presents the return rates by release type. The lowest return rates were for inmates who
discharged their sentences without community supervision and subsequently returned to prison for
committing a new crime. Inmates who release direct from segregation to the community are closely
scrutinized for administrative segregation placement in the event they fail in the community.
Figure 5 examines return rates for the various administrative segregation facilities. As a cautionary
statement, comparisons should not be made across facilities because many factors, including release type,
were not taken into account. Return rates are presented for descriptive purposes; if the intent was to make
inferences regarding the effectiveness of one facility over another, offender characteristics and release types
would need to be controlled for across groups.
Table 16. 1 and 2 Year Returns to Administrative Segregation
Return to Ad Seg
Release Cohort
N
1 Year
2 Year
1996
156
18%
22%
1997
222
12%
20%
1998
207
11%
18%
1999
267
8%
18%
2000
250
10%
18%
2001
345
13%
21%
2002
400
13%
-Figure 4. Return to Administrative Segregation by Release Type

% Returned to Ad Seg

35%

1 year
2 year

30%

30%
25%

15%

17%
12%

12%

10%
4%

5%

20%

20%

19%

20%

8%

6%

0%
GP

Sentence
Discharge

Mandatory Parole

Discretionary
Parole

Overall

Administrative Segregation Release Type

STUDY 2: LONGITUDINAL ANALYSIS

24

Figure 5. Return to Administrative Segregation Rates by Facility

% Returned to Ad Seg

25%
20%
15%

1 year
2 year

20%
18%

18%

18%

16%

14%
12%

11%

12%

10%

11%

12%

7%

5%
0%
CSP

SCF

SCCF

Females

CSP/SCCF

Other

a

Administrative Segregation Release Type
a

Includes placement in multiple administrative segregation facilities.

Return to segregation was analyzed according to mental health status and STG involvement. There was
no difference across mental health groups for 1- and 2- year return to prison rates. Similarly, there was no
difference between STG groups at 1 year. However, STG members were significantly more likely to return to
segregation within 2 years (26%) than associates (21%). Members were twice as likely to return than
individuals with no STG affiliation (13%), F(2, 1487) = 16.64, p < .001.
Recidivism. Recidivism rates were examined for up to three years following release from prison, for those
who released (see Table 17). A distinction is made for whether they released directly to the community or first
to GP. There appears to be a slight trend towards lower recidivism rates if inmates transitioned through GP
prior to release, however, the difference is not statistically significant (see Appendix A). It is possible that this
trend is attributable to improved recidivism rates for those who completed CSP’s pro-unit, which should be
investigated through further research.
The recidivism rates of administrative segregation inmates are considerably higher than CDOC’s overall
recidivism rates. CDOC’s 3-year recidivism rates are as follows: 27% for sentence discharge, 64% for
mandatory parole, 54% for discretionary parole, and 50% overall (Rosten, 2004). It is not at all inconceivable
that the highest custody inmates have exceedingly poor recidivism rates.
Table 17. Recidivism Rates by Release Type
Release Type
1 Year
Released from Ad Seg via:
(n = 698)
Sentence Discharge
10%
Mandatory Parole
56%
Discretionary Parole
48%
Total
41%
Released to GP prior to:
Sentence Discharge
Mandatory Parole
Discretionary Parole
Total

(n = 396)
11%
53%
47%
42%

STUDY 2: LONGITUDINAL ANALYSIS

2 Year
(n = 549)
27%
78%
65%
60%

3 Year
(n = 418)
38%
84%
69%
66%

(n = 296)
21%
70%
60%
55%

(n = 221)
27%
74%
70%
60%

25

Recidivism rates were examined by facility (see Table 18). Because the earlier analyses demonstrated that
transition to GP did not significantly impact recidivism rates, the categories were collapsed. Each successive
year is cumulative, meaning that inmates who recidivated in their first year of release would be included in the
return rates for second and third years. However, the overall numbers decline with each successive year
because fewer inmates had 2- or 3-year at-risk periods. Again, given that each facility houses inmates with
different risk factors and release types, inferences can not be made regarding one facility’s effectiveness over
another.
Table 18. Recidivism Rates by Administrative Segregation Facility
1 Year at Risk (1995–2000) a 2 Years at Risk (1995–2001) a 3 Year at Risk (1995–2002) a
Facility
# returns % returns
Total
# returns % returns
Total
# returns % returns
Total
CSP
310
40%
785
373
57%
655
320
63%
510
SCF
52
42%
125
32
63%
51
-----SCCF
22
60%
37
24
75%
32
23
77%
30
Females
34
44%
78
39
57%
69
38
67%
57
Other b
34
49%
69
23
61%
38
18
62%
29
Total
452
41%
1,094
491
58%
845
399
64%
626
a Years in parentheses define release years. b Includes placements in multiple administrative segregation facilities.
Table 19 examines recidivism rates by mental health status and STG involvement. Additional analyses,
not reported here, were conducted to determine if holding release type constant across groups impacted the
results. The pattern of results was similar and did not change interpretation of findings. Actual recidivism
rates are reported below because they are easier to interpret than adjusted rates.
The results indicated that recidivism rates did not differ by severity of mental health diagnosis. On the
other hand, STG involvement was related to recidivism. STG members have the highest recidivism rates
followed by associates/suspects. Those with no STG involvement have the lowest rates.
Table 19. Comparison of Recidivism Rates Across Mentally Ill and STG Involvement Groups
Qualifying Mental Illness Diagnosis
OMI
Other
None
1 Year Recidivism
2 Year Recidivism
3 Year Recidivism

# return

% return

Total

61
60
50

47%
61%
65%

129
98
77

# return

1 Year Recidivism
2 Year Recidivism
3 Year Recidivism

186
205
158

Member

# return

15
14
7

% return

42%
64%
58%

Total

36
22
12

STG Involvement
Associate/Suspect

# return

375
414
350

% return

Total

# return

% return

Total

# return

48%
72%
77%

385
283
205

119
127
109

40%
55%
61%

295
233
179

147
159
142

% return

Total

41%
57%
64%

924
722
548

None

% return

Total

36%
48%
56%

414
329
255

p
n.s.
n.s.
n.s.

p
<.01
<.001
<.001

Institutional Behavior. A repeated measures design with one within-subjects factor (time) and one
between-subjects factor (OMI) was proposed to analyze the rate of COPDs over time, using 6 month
intervals beginning at 6 months pre-segregation and ending with 6 months post-segregation. However,
because participants may have short durations before or after segregation, as well as varying segregation stays,
the longitudinal sample was partitioned into 13 groups. Doing so enabled maximum flexibility to examine
different paths that offenders might experience.
Inmates with less than 3 months prior to segregation (n = 254) or with less than 6 months in
segregation (n = 276) were excluded. Additionally, inmates still in segregation (n = 599) at the end of the
study (December 2003) were excluded, unless their episode was longer than 3 years. Some offenders may

STUDY 2: LONGITUDINAL ANALYSIS

26

have been excluded for more than one of these reasons. Hypothetically, the 599 still in segregation would
naturally fall into one of the 13 groups given enough time. However, these findings can not be applied to
inmates who are quickly placed in segregation (e.g., high-profile criminals, parole returns) or to those who
serve a brief time in segregation.
Table 20 describes how the 13 groups were categorized. The first six groups included inmates who
returned to GP for a minimum of 6 months while the second six groups included inmates who released either
directly to the community or within 6 months of return to GP. For those with fewer than 6 months in prison
prior to segregation, a 6-month rate of COPDs was computed. Similarly, a 6-month rate of COPDs was
calculated for participants in groups 6, 12, and 13 who were housed in excess of 36 months in segregation (all
COPDs from 36 months to release were factored in). Where there was an insufficient number of OMIs, a
within-subjects only design was conducted.
Mean plots for every group are available in Appendix B.
There was a significant time main effect for all groups. In all cases except group 13, the rate of COPDs
dropped substantially upon entry into administrative segregation. Upon the last measurement period prior to
release, the rate approached zero. For those who returned to GP for at least 6 months, COPDs increased
following release but not to pre-segregation levels. This pattern suggests that placement in administrative
segregation reduces the rate of COPDs, although they may not be the best measure of compliant behavior as
the setting precludes certain violations (i.e., assault, fighting, bartering). Even so, COPDs is not a wholly
inadequate measure as evidenced by group 13 which had longer than average segregation stays presumably
related to elevated rates of disciplinary infractions.
Groups 1 through 6 are particularly interesting when considering whether behavior dictates release. At
first glance, it appears as though it does. However, groups 3 through 5 demonstrate very low levels of
violations commencing around 6 or 12 months, but they are maintained in administrative segregation for
longer periods. Considering that the average number of COPDs was less than .5 for these periods, most
inmates would be accumulating no disciplinary violations for months or years without release. It should be
noted that COPDs is not the only measure of behavior, albeit it an important one. This is another area for
further research to understand why there is a need to segregate these inmates for such long durations.
There were significant OMI main effects for 7 of the 8 comparisons, indicating that OMIs have more
disciplinary violations than non-OMIs. This finding suggests that mentally ill inmates were more disruptive,
perhaps because adapting to the prison environment poses a greater challenge to them. Nonetheless, they
represent a population that is more difficult to manage.
Three interaction effects were significant. For groups 1, 2, and 8, it can be interpreted that the rate of
change in COPDs over time was different for OMIs than non-OMIs. This appears primarily due to presegregation levels of COPDs. Over time, COPDs for OMIs drop to levels found with non-OMIs, but they
have a greater distance to travel.
Group 13 was different from the other groups. This was a disruptive group of individuals who
continued to have write-ups and violations while in segregation, particularly the OMIs, and consequently were
retained within segregation. This population may be of particular interest for further study because they
clearly are unmanageable even in segregation. It may be that segregation is particularly incompatible for these
individuals and that another approach is needed. Or it may indicate that these individuals are so unruly and
combative that the safety of the prison environment depends on their extreme confinement.

STUDY 2: LONGITUDINAL ANALYSIS

27

Table 20. Factorial Repeated Measures ANOVA
Months in
Sample Size
Repeated Measures ANOVA Results
Group
Ad Seg
Total OMI OMI
Time
Time x OMI
Released to GP with at least 6 months incarcerated following release:
1
6-12
124
20
F(2, 257)= 78.44**, η2=.39
F(2, 257) = 5.08*, η2=.04
F(1, 122)= 6.77*, η2=.05
2
2
2
13-18
281
35
F(1, 279)= 12.70**, η =.04 F(2, 600) = 160.70**, η =.37
F(2, 600) = 5.73*, η2=.02
3
19-24
235
31
F(1, 233)= 8.37*, η2=.04
F(3, 716) = 77.23**, η2=.25
F(3, 716) = 2.70, η2=.01
2
-4
25-30
143
9
-F(2, 248) = 87.75** , η =.38
2
5
31-36
90
8
--F(3, 283) = 48.85**, η =.35
6
>36
217
18
F(5, 1050) = 21.39**, η2=.09
F(5, 1050) = 0.44, η2 = .00
F(1, 215)= 0.14, η2=.00
Released to community or < 6 months incarcerated following release:
7
6-12
257
36
F(1, 255)= 13.26**, η2=.05
8
13-18
207
28
F(1, 205)= 10.78*, η2=.05
9
19-24
128
23
F(1, 126)= 7.59*, η2=.06
10
25-30
74
6
-11
31-36
59
3
-12
>36
59
8
-Still in administrative segregation after more than 36 months:
13
>36
112
21
F(1, 110)= 20.85**, η2=.16
** p < .001, * p < .01.

Within-Ss
MSE
2.75
3.07
2.54
8.08
3.51
4.85

F(1, 362) = 146.95**, η2=.37
F(2, 353) = 82.45**, η2=.29
F(3, 374) = 57.53**, η2=.31
F(2, 171) = 48.50**, η2=.40
F(3, 194) = 12.16**, η2=.17
F(5, 273) = 15.11**, η2=.21

F(1, 362) = 5.03, η2 = .02
F(2, 353) = 6.11*, η2 = .03
F(3, 374) = 1.98, η2 = .02
----

8.27
13.63
7.69
15.96
11.56
8.90

F(4, 468) = 14.06**, η2 = .11

F(4, 468) = 2.18, η2 = .02

12.36

Trend of Segregating Mentally Ill Offenders. Five year prison and administrative segregation populations were
examined in 6-month increments. Figure 6 displays the rate of OMIs in the population and segregation.
Historically, OMIs have been over-represented in administrative segregation. However, the rate increased
substantially, particularly in the last year. This increase coincides with a dramatic decline in mental health
professionals and rehabilitation programs that were a casualty of budget cuts.
Figure 6. Five-Year Prevalence Rates of OMI in Administrative Segregation

30%
CDOC Population

Ad Seg Only

25%

Percent OMI

25%
19%

20%

20%

25%

21%

17%

15%

15%

14%

15%

15%

15%

10%
5%
0%
6/99 12/99 6/00 12/00 6/01 12/01 6/02 12/02 6/03 12/03 6/04
Month/Year

STUDY 2: LONGITUDINAL ANALYSIS

29

STUDY 3: ADMINISTRATIVE SEGREGATION HEARINGS
Method
Participants. Data from 842 hearings from August 2003 to July 2004 was used for this study. Participants
included 820 inmates who had administrative segregation hearings, with 22 of these offenders having two
hearings during this timeframe. Inmates were 95% male (n = 779) ranging in age from 18 to 62 (M = 29.8, SD
= 7.8). The ethnic composition of inmates was as follows: 41% Hispanic, 37% Caucasian, 19% African
American, 2% Native American, and 1% Asian.
Procedure. Inmates who were determined to be a threat to the facility were removed from population and
placed in a segregation cell. They attended an administrative segregation hearing where it was determined
whether such a placement was warranted. The case was next reviewed by the warden who had the
opportunity to affirm, reverse or modify the decision. Lastly, each case was reviewed by CDOC’s Offender
Services staff prior to reclassification and transfer. This division, in conjunction with the Director of Prisons,
also had the authority to affirm, reverse or modify the decision.
The Offender Services personnel who reviewed administrative segregation hearings maintained a
database designed for research purposes. The database application was put into use in mid-July 2003 and
variables included hearing dates, decisions, reasons for placement decision, and a narrative of the facts
surrounding the case.

Results and Conclusions
Of the 842 hearings, 90% resulted in an initial decision to place the inmate in administrative segregation
(n = 759). Nine of those decisions were reversed and one was modified. Ultimately, 762 of the 842 hearings
resulted in an administrative segregation placement.
The number of hearings conducted by month is shown in Figure 7 and the number of hearings by
facility is presented in Figure 8. Facilities not shown in Figure 8 had no hearings during the year. Hearings
were on the decline for the first few months and peaked in March. During the spring, there were major
facility disturbances at LCF, FCF, AVCF, and CCF that accounted for a large number of hearings.
Reasons for administrative segregation placements are shown in Figure 9. Typically, offenders were
placed in administrative segregation for multiple reasons, averaging 4.02 reasons (SD = 1.56). The most
common of which was serious management problem, indicative of offenders who pose a threat by remaining
within the general population. Only one inmate had this listed as the sole placement reason. Other common
reasons included multiple disciplinary violations, advocating facility disruption, STG activity, and assaults,
many of which were STG-related. The most frequent ‘other’ reasons included failure in the CSP’s pro-unit
program, parole failure following release from administrative segregation, and dealing drugs.
Figure 7. Administrative Hearings by Month

Number of Hearings

120
100
80
60
40
20
0
Aug

Sep

Oct

Nov

Dec

Jan

Feb

Mar

Apr

May

Jun

Jul

Month (2003 - 2004)

STUDY 3: ADMINISTRATIVE SEGREGATION HEARINGS

30

Figure 8. Hearings by Facility

Number of Hearings
0
ACC (1%)

20

40

60

80

100

105

BCCF (1%)

7

BVCF (9%)

74
1

Facility (% of Operational Capacity)

CCCF (2%)

25
74

CCF (22%)
CSP (0%)

3
41

CTCF (5%)
15

CWCF (7%)
DCC (0%)

2
79

DRDC (16%)
DWCF (3%)

26

FCF (6%)

81

FLCF (3%)
FMCC (1%)

13
5

HCCF (4%)

30

KCCF (1%)

9
102

LCF (11%)
SCC (1%)

3
129

SCF (5%)
SCCF (2%)
TCF (2%)

140

3

AVCF (10%)

CCC (1%)

120

4
11

STUDY 3: ADMINISTRATIVE SEGREGATION HEARINGS

31

Figure 9. Reasons for Administrative Segregation Placements
0%

10% 20% 30% 40% 50% 60% 70% 80% 90% 100%

ASSAULT ON INMATES…
20%

Weapon/Serious Injury
5%

No Weapon

11%

No Injury
ASSAULT ON STAFF…
Weapon/Serious Injury
No Weapon

1%
4%
8%

No Injury
STG INVOLVEMENT…
Related Assaults

28%
8%

Recruitment

41%

Activity
INTRODUCTION OF…
10%

Drugs
Tobacco

2%

POSSESSION OF…
11%

Weapon

7%

Drug Paraphenelia
ESCAPE…
Serious Threat
With Force
Without Force

3%
0%
2%
14%

Physical altercation/ fight
Compromising staff/ intimidation

3%

High profile

2%
12%

Threats

29%

Advocating facility disruption
15%

Enciting a riot

56%

Multiple COPDs

89%

Serious management problem
Refusing assignment
Other

3%
24%

STUDY 3: ADMINISTRATIVE SEGREGATION HEARINGS

.

32

DISCUSSION
This study revealed an administrative segregation profile that was very discernable from that of the
general population. As a whole, the administrative segregation population has a worse criminal history and
more disruptive institutional behavior than the general prison population, which would confirm that
Colorado’s administrative segregation is used for the worst of the worst. Further support was provided in
study 3 where the bulk of placement reasons occurred for serious offenses such as violent assaults on other
inmates, STG involvement, multiple disciplinary violations, and facility disruptions.
While the administrative segregation population clearly delineates itself from the general population as a
more disruptive, difficult to manage group of offenders, the question of whether Colorado sets its bar too
low remains. In the absence of a national standard or guideline, it is unclear how disruptive is enough to
warrant segregation. In study 1, 17% of administrative segregation cases had never been convicted of a class I
disciplinary violation, a significant indicator of a seriously disruptive inmate. A file review of these cases
indicated that many offenders established disruptive, violent patterns through their crime, prior Colorado
incarcerations, or their incarcerations in other jails or prisons. The remainder had less overt behavior patterns,
having committed some class II violations or demonstrating gang involvement.
Research is subject to its own limitations. It is important to remember that statistics examine groups,
averages, and sample distributions. A court of law, on the other hand, looks at individual cases. For this
reason, it is not enough to say that, on average, administrative segregation placements are justified. Even one
inappropriate placement is too many.
Administrative segregation durations are of equal interest to placements. In Colorado, these stays
average one and a half to two years. In the absence of more statistics, it is unknown how long this is in
comparison to national averages. However, it is important to note that OMIs had shorter durations than their
mentally healthy counterparts. As much as possible, OMI placements are made in the “kinder, gentler
settings” provided at DWCF and SCCF where there is a greater emphasis on intensive mental health services
and quick transitions back to population. Unfortunately slots are limited by number, and staff are forced to
make difficult decisions of which inmates deserve prioritization for these beds.
Disciplinary violations not only decreased as a result of administrative segregation, they were also
related to offenders’ release. Clearly, COPDs did not represent the only factor dictating release decisions; the
other factors remain yet unmeasured. Given that 41% of inmates release directly to the community, coupled
with a correspondingly high recidivism rate, the need exists to transition offenders to lower custody prisons
as early as possible.
Perhaps the most informative findings from this study described the types of inmates in segregation.
Two unique profiles emerged with very little overlap between the two: mentally ill and STG-involved inmates.
Interestingly, there still remained an undefined administrative segregation group that fit into neither group.
Segregating STG-involved inmates represents a departmental management decision to reduce the negative
impact of gang related activities in the general population. The same is not true of the OMI group, although
high segregation rates appear to be an unintended effect, likely attributable to recent budget deficits.
This study was specifically concerned about OMIs in administrative segregation. Not only was a
disproportionately high prevalence of Axis I and II disorders discovered, but administrative segregation
inmates exhibited more severe psychiatric symptoms. These represent serious implications for operating such
a facility. It is the responsibility of staff to monitor OMIs and ensure that they do not decompensate,
particularly in regards to suicidal behavior. Solely by definition, this population requires more rigorous
programming and treatment services than most. In addition to their mental health needs, this group presents
with greater academic, vocational and medical needs which can not be neglected on the basis of their
administrative segregation placement.
This trend of incarcerating more and more OMIs in solitary confinement has even greater implications
for non-administrative segregation facilities. With the decline of mental health resources, rates of
administrative segregation placements for Colorado OMIs have dramatically increased. Whether the
prevailing perspective that prison mental health services are ancillary changes to one where they are accepted
as integral or mental health training is brought to the front lines (i.e., security staff), prisons need to learn how
to better manage their mentally ill inmates and protect them. Optimally both scenarios would be in effect,

DISCUSSION

33

where mental health clinicians work closely with line staff to employ the best management strategies. A
system which condones “check-ins” (inmates who, fearing for their safety, engage in the minimal behaviors
necessary to land them in solitary confinement), by merely allowing them to happen would benefit from
procedural review. Colorado should take heed from Judge Justice’s ruling in Ruiz v. Johnson (1999): “Despite
its institutional awareness of these conditions, [Texas Department of Correctional Justice] has failed to take
reasonable measures to protect vulnerable inmates from other, predatory prisoners and overzealous,
physically aggressive state employees.”

Future Research
First and foremost, the need persists for researchers to tackle the empirical question of whether
administrative segregation inflicts psychological harm on inmates, both for mentally ill and mentally healthy
subgroups. A well-designed study would include a control group and a repeated measures design. The
repeated measures design can ascertain whether offenders decompensate over long periods of solitary
confinement while the use of a control group can discern whether any significant psychological changes are
attributable to the administrative segregation environment specifically or are simply associated with the
overall prison environment. In keeping with the broad range of criticisms aimed at solitary confinement,
measures across a variety of psychological dimensions (e.g., suicide ideation, hopelessness, psychosis) should
be included. Assessments gained through multiple sources (e.g., inmate self-report, mental health clinician,
researcher) enable a researcher to analyze convergence of measures, thereby strengthening the study findings.
Furthermore, research should try to understand if some types of individuals, such as those with Cluster A
personality disorders, do better in a segregation environment than others.
Process evaluations that take into account actual prison operations in Colorado’s administrative
segregation facilities should be undertaken. As demonstrated in Ruiz v. Johnson (1999), clearly written policies
and procedures are not nearly enough; a court of law will examine actual practices. In fact, even ACA
accreditation did little to assist the corrections officials’ position in the Texas lawsuit because a paper review
was deemed an inadequate substitute for a thorough on-site evaluation. Process evaluations conducted by an
agency or unit external to the prison program, where there are not repercussions for either the prison or the
researchers, is essential to obtain an unbiased perspective that might reveal program features not obvious to
those working there everyday and promote positive change.
Finally, it is recommended that an outcome study of CSP’s Pro-Unit be conducted. There is some
evidence to suggest that inmates who complete this transitional program from administrative segregation
have better outcomes than those who release directly to the community. If this is a transitional model that
provides successful reintegration back into the general prison population and/or society at large, then it
should be endorsed as an ideal model within Colorado and nationally.

DISCUSSION

34

REFERENCES
American Psychiatric Association (1994). Diagnostic and statistical manual of mental disorders (4th ed.).
Washington, DC: Author.
Andersen, H. S., Sestoft, D., Lillebaek, T., Gabrielsen, G., Hemmingsen, R., & Kramp, P. (2000). A
longitudinal study of prisoners on remand: Psychiatric prevalence, incidence, and psychopathology in
solitary vs. non-solitary confinement. Acta Psychiatr Scand, 102, 19-25.
Austin v. Wilkinson, 204 F. Supp. 2d 1024 (N.D. Ohio 2002).
Bauernfeind, R. H. (1992). Tests of Adult Basic Education: Forms 5 and 6. In D. J. Keyser, & R. C.
Sweetland (Eds.), Test Critiques Vol. IX. (pp. 591-596). Austin, TX: Pro-ed.
Burger, G. K., Calsyn, R. J., Morse, G. A., Klinkenberg, W. D., & Trusty, M. L. (1997). Factor structure of the
Expanded Brief Psychiatric Rating Scale. Journal of Clinical Psychology, 53, 451-454.
Benjamin, T. B., & Lux, K. (1975). Constitutional and psychological implications of the use of solitary
confinement: Experience at the Maine state prison. Clearinghouse Review, 9, 83-84.
Berge, G., Geiger, J., & Whitney, S. (2001). Technology is the key security in Wisconsin supermax. Corrections
Today, 63, 105-113.
Carriere, K. D. (1989). Protective custody in Canada: A review of research and policy responses. Canadian
Criminology Forum, 10, 17-25.
Cattell, R. B., & Cattell, H. E. P. (1973). Measuring intelligence with the Culture Fair Tests. Champaign, IL: Institute
for Personality and Ability Testing.
Collins, W. C. (2004). Correctional law for the correctional officer (4th ed.). Lanham, MA: American Correctional
Association.
CTB/McGraw-Hill. (1994). Tests of Adult Basic Education (TABE): Version 7-8. Monterey, CA: CTB/McGrawHill.
CTB/McGraw-Hill. (2004). Tests of Adult Basic Education (TABE): TABE Form 7 to the 2002 GED Linking
Report.
Retrieved
July
6,
2004,
from
the
CTB
McGraw-Hill
Web
site:
http://www.ctb.com/media/articles/pdfs/AdultEducation/TABE_Form7_2002_GED_Linking_Rep
ort.pdf?FOLDER<>folder_id=59997&bmUID=1089131925485.
Haney, C. (2003). Mental health issues in long-term solitary and “Supermax” confinement. Crime and
Delinquency, 49, 124-156.
Hedlund, J.L., & Vieweg, B.W. (1980). The Brief Psychiatric Rating Scale (BPRS): A comprehensive review.
Journal of Operational Psychiatry, 11, 48-62.
Hershberger, G. L. (1998). To the max: Super-max facilities provide prison administrators with more security
options. Corrections Today, 60, 54-57.
Hewitt v. Helms, 459 U.S. 460 (1983).
Hodgins, S., & Côté, G. (1991). The mental health of penitentiary inmates in isolation. Canadian Journal of
Criminology, 33, 175-182.
Human Rights Watch (1997). Cold Storage: Super-Maximum Security Confinement in Indiana. New York: Human
Rights Watch.
Human Rights Watch (1999). Red Onion State Prison: Super-maximum security confinement in Virginia.
Human Rights Watch, 11, 1-24.
Human Rights Watch (2000). Out of sight: Super-maximum security confinement in the United States.
Human Rights Watch, 12, 1-9.
Jackson, M. (2001). The psychological effects of administrative segregation. Canadian Journal of Criminology, 43,
7-9.
Jones v. Baker, 155 F.3d. 810 (6th Cir. 1998).
Jones ‘El v. Berge, 164 F. Supp. 2d 1096 (W.D. Wisc. 2001).
King, R. (2000). The rise and rise of super-max: An American solution in search of a problem? Punishment and
Society, 1, 163-186.
Koch, W. R. (1992). Culture Fair Intelligence Test. In D. J. Keyser, & R. C. Sweetland (Eds.), Test Critiques
Vol. I. (pp. 233-238). Austin, TX: Pro-ed.

REFERENCES

35

Kurki, L., & Morris, N. (2001). The purposes, practices, and problems of supermax prisons. Crime and Justice,
28, 1-21.
Lukoff, D., Nuechterlein, K., & Ventura, A. (1986). Manual for the expanded brief psychiatric rating scale.
Schizophrenia Bulletin, 13, 261-276.
Madrid v. Gomez, 889 F. Supp. 1146 (N.D. Cal. 1995).
McClary v. Kelly, 4 F. Supp. 2d 195 (W.D.N.Y. 1998).
Miller, H. A., & Young, G. R. (1997). Prison segregation: Administrative detention remedy or mental health
problem? Criminal Behaviour and Mental Health, 7, 85-94.
Millon, T., Davis, R.D., & Millon, C. (1997). Manual for the Millon Clinical Multiaxial Inventory-III (MCMI-III) (2nd
ed.). Minneapolis: National Computer Systems.
Motiuk, L. L., & Blanchette, K. (2001). Characteristics of administratively segregated offenders in federal
corrections. Canadian Journal of Criminology, 43, 131-144.
NIC (1997). Supermax housing: A survey of current practice, special issues in corrections. Longmont, CO: US
Department of Justice, National Institute of Corrections.
NIC (1999). Supermax prisons: Overview and general considerations. Longmont, CO: US Department of Justice,
National Institute of Corrections.
O’Keefe, M. L., Klebe, K. J., & Hromas, C. S. (1998). Validation of the Level of Supervision Inventory (LSI) for
community based offenders in Colorado. [Technical Report]. Colorado Springs, CO: Department of
Corrections.
Overall, J. E. & Gorham, D. R. (1962). The brief psychiatric rating scale. Psychological Reports, 10, 799-812.
Pizzaro, J. & Stenius, V. M. K. (2004). Supermax prisons: Their rise, current practices and effect on inmates.
The Prison Journal, 84, 248-264.
Rosten, K. L. (2004). Statistical report: Fiscal year 2003. [Technical Report]. Colorado Springs, CO: Department
of Corrections.
Ruiz v. Johnson, 37 F. Supp. 855 (S.D. Tex. 1999).
Sandin v. Conner, 515 U.S. 472 (1995).
Segal, S. P., & Silverman, C. (2002). Determinants of client outcomes in self-help agencies. Psychiatric Services,
53, 304-309.
Shepperd, R. A., & Geiger, J. R. (1996). Closed maximum security: The Illinois supermax. Corrections Today, 58,
84-88.
Suedfeld, P., Ramirez, J., Deaton, J., & Baker-Brown, G. (1982). Reactions and attributes of prisoners in
solitary confinement. Criminal Justice and Behavior, 9, 303-340.
Tannenbaum, A. J. (1965). Culture Fair Intelligence Test. In O. K. Buros (Ed.), The sixth mental measurements
yearbook (pp. 721-723). Lincoln, NE: University of Nebraska Press.
Toch, H. (2001). The future of supermax confinement. The Prison Journal, 81, 376-388.
Toch, H. (2003). The contemporary relevance of early experiments with supermax reform. The Prison Journal,
83, 221-228.
Ventura, J., Lukoff, D., Nuechterlein, K., H., Liberman, R. P., Green, M. F., & Shaner, A. (1993). Training and
Quality Assurance with the BPRS. Los Angeles, CA: UCLA Department of Psychiatry and Biobehavioral
Sciences.
Ward, D. A. (1995). A corrections dilemma: How to evaluate super-max regimes. Corrections Today, 57, 104109.
Ward, D. A. (1999). Supermaximum facilities. In Prison and jail administration: Practice and theory, edited by P. M.
Carlson & J. S. Garrett. Gaithersburg, MD: Aspen.
Wright v. Smith, 21 F. 3d 496, 510 (2nd Cir. 1994).
Zinger, I., Wichman, C., & Andrews, D. A. (2001). The psychological effects of 60 days in administrative
segregation. Canadian Journal of Criminology, 43, 47-88.

REFERENCES

36

APPENDIX A
Comparison of Recidivism Rates via Release to GP
Recidivism Rates
2-Year
t(221) = 0.87, p = .39
t(332) = 1.57, p = .12
t(286) = 0.85, p = .40

1-Year
t(270) = -0.25, p = .80
t(484) = 0.56, p = .57
t(334) = 0.19, p = .85

Sentence Discharge
Mandatory Parole
Discretionary Parole

3-Year
t(174) = 1.40, p = .16
t(233) = 1.81, p = .07
t(226) = -0.20, p = .84

APPENDIX B
Group 1

8

COPDs

6

OMI
Non-OMI

4
2
0
pre

0

6

post

Measurement Intervals

Group 2

8

COPDs

6
OMI
Non-OMI

4
2
0
pre

0

6

12

post

Measurement Intervals
Group 3

8

COPDs

6
OMI
Non-OMI

4
2
0
pre

0

6

12

18

post

Measurement Intervals

APPENDICES

37

Group 4

8

COPDs

6
4
2
0
pre

0

6

12

18

24

post

Measurement Intervals

Group 5

8

COPDs

6
4
2
0
pre

0

6

12

18

24

30

post

Measurement Intervals

Group 6

8

COPDs

6
OMI
Non-OMI

4
2
0
pre

0

6

12

18

24

30

36

post

Measurement Intervals

APPENDICES

38

Group 7

8

COPDs

6
OMI
Non-OMI

4
2
0
pre

0

6

Measurement Intervals

Group 8

8

COPDs

6
OMI
Non-OMI

4
2
0
pre

0

6

12

Measurement Intervals

Group 9

8

COPDs

6
OMI
Non-OMI

4
2
0
pre

0

6

12

18

Measurement Intervals

APPENDICES

39

Group 10

8

COPDs

6
4
2
0
pre

0

6

12

18

24

Measurement Intervals

Group 11

8

COPDs

6
4
2
0
pre

0

6

12

18

24

30

Measurement Intervals

Group 12

8

COPDs

6
4
2
0
pre

0

6

12

18

24

30

36

Measurement Intervals

APPENDICES

40

Group 13

8

COPDs

6
OMI
Non-OMI

4
2
0
pre

0

6

12

18

24

30

36

Measurement Intervals

APPENDICES

41