Security Threat Management in Prison: Revalidation and Revision of the Inmate Risk Assessment for Segregation Placement

Abstract

Prison officials often rely on restrictive housing to promote institutional safety and security. However, a growing body of research indicates this type of confinement has little impact on inmate behavior or institutional order. An alternative approach involves providing the most dangerous and disruptive inmates with increased case management services and other proactive programmatic opportunities. The success of this strategy requires an ability to prospectively and accurately identify the most problematic inmates. The results of this study indicate that Risk Assessment for Segregation Placement (RASP) and its revised Oregon version (RASP-OR) are valid predictors of segregation placement and institutional misconduct. The policy implications of these findings are discussed.

Keywords

prison restrictive housing solitary confinement risk assessment

Introduction

Above all else, prison administrators strive to ensure institutional safety and security (Cullen et al., 1993; DiIulio, 1987; Mears & Castro, 2006).This is no small task considering that more than half (56%) of all state inmates are sentenced for a violent crime (Carson, 2020), and nearly one-fifth (19.1%) are believed to be members of a security threat group (i.e., a gang; Winterdyk & Ruddell, 2010). Findings from a nationally representative survey of prisoners indicate that nearly one of every five respondents (17.4%) reported involvement in a fight or being written up for an assault in the previous year (Beck, 2015).One strategy for increasing safety and promoting order in prison involves segregating dangerous and disruptive inmates into restrictive housing settings (Frost & Monteiro, 2016; Labrecque & Mears, 2019).¹ Despite its promise, several recent reviews of the empirical literature have highlighted that this type of confinement appears to have little to no influence on inmate behavior or institutional order (Gendreau & Goggin, 2019; Labrecque & Smith, 2019a; Morgan et al., 2016; Steiner & Cain, 2016). Gaining a better understanding of the predictors of institutional rule violations and restrictive housing placements is, therefore, vital to assist prison authorities in developing more effective security threat management practices.

If there was a reliable and reasonably accurate mechanism for identifying inmate risk for institutional misconduct and segregation placement upon their entry into the prison system, then this information could be used prospectively to devise appropriate supervision and treatment plans with the aim of reducing the occurrence of these adverse events in the first place. Prison officials, for example, could rely on this information to triage the limited case management services and other programmatic opportunities available toward the highest risk inmates. There are good reasons to suspect that this offender management strategy would help reduce prison misconduct and improve institutional order (see also Labrecque, 2018a; and Smith, 2016). For one, penological scholarship has documented the effectiveness of treatment programs (e.g., education classes, employment training, cognitive-behavioral therapy) in reducing institutional misconduct (French & Gendreau, 2006; Huebner, 2003; Pompoco et al., 2017; Randol & Campbell, 2017; Wooldredge & Steiner, 2015) and has shown promise for reducing placements in restrictive housing as well (Butler et al., 2018; Meyers et al., 2018). More broadly, correctional research has also demonstrated that therapeutic interventions produce the greatest improvements in behavior among higher- versus lower-risk offenders (Andrews et al., 1990; Bonta & Andrews, 2017; Lowenkamp et al., 2006).

Systematic reviews of the literature have uncovered several inmate background variables (e.g., demographics, criminal history), institutional experiences (e.g., prior misconducts), and prison characteristics (e.g., custody level) that are correlated with the perpetration of prison misconduct (Gendreau et al., 1997; Schenk & Fremouw, 2012; Steiner et al., 2014) and placement in restrictive housing (Labrecque, 2018b; Steiner & Cain, 2016). Actuarial risk assessments make use of this type of information by constructing a composite score, where individuals who possess more of these factors are at a greater likelihood for experiencing the outcome than those who have less (Andrews et al., 2006; Bonta, 2002; Bonta & Andrews, 2017). Several risk assessment tools have been found to be strong predictors of institutional misconduct, including the Level of Service Inventory-Revised (LSI-R), the Historical Clinical Risk Management-20 (HCR-20), and the Psychopathy Checklist-Revised (PCL-R; Campbell et al., 2009; Kroner & Mills, 2001; Singh et al., 2011).Research has shown that inmates confined in restrictive housing settings often include the most serious and chronic violators of institutional rules (Labrecque, 2018b; Steiner & Cain, 2016) and that prison staff tend to view these individuals as posing the greatest threat to institutional safety and order (Butler et al., 2012; Labrecque & Mears, 2019). The use of this type of confinement further raises serious ethical and legal considerations (Frost & Monteiro, 2016; Gendreau & Labrecque, 2018; Kapoor & Trestman, 2016), and it is undeniably more expensive to operate than housing in the general prisoner population (Butler et al., 2017; Mears, 2016). From a prison management perspective, therefore, it may be more important to identify risk for segregation placement compared to general prison misconduct.

Recently, two risk assessment scales were developed to identify inmate propensity for restrictive housing placements in the United States (i.e., Risk Assessment for Segregation Placement [RASP]; Labrecque & Smith, 2019b) and Canada (i.e., Risk of Administrative Segregation Tool [RAST]; Helmus et al., 2019). Although both instruments have displayed high levels of predictive accuracy (AUCs>.750), neither has been revalidated with inmates from prison systems other than the ones in which the tools were initially constructed. If it can be shown that these tools are predicatively valid with different samples of inmates, it will increase confidence in the generalizability and utility of these risk measures. Additionally, a growing body of studies discusses the importance of validating and norming risk assessments on local populations (Lovins et al., 2018), further suggesting that weighting the scores of the individual risk items by their predictive strength is one way to maximize the overall performance of the tool (Duwe & Kim, 2016; Georgiou, 2019; Hamilton et al., 2015). Of course, when considering modifications to a risk instrument, it is imperative to evaluate if the revised version is capable of outperforming the original (Cunningham & Sorensen, 2006; Duwe, 2019).

In summary, the current study seeks to advance knowledge on prison risk assessment tools and security threat management practices by addressing two research questions: (1) Does the Risk Assessment for Segregation Placement (RASP) accurately predict segregation placements and institutional misconducts in a sample of prisoners from Oregon?, and (2) Does weighting and norming the scoring of the RASP items on the local Oregon population help improve the predictive validity of the tool on these two outcomes?

Method

Participants and Setting

The participants in this study include the population of inmates who were admitted into the Oregon Department of Corrections (ODOC) custody in calendar year 2016 (N = 4,709). Administrative records on these inmates were tracked for two years following their initial admission date. The selection of this study location is important for several reasons. First, the quality and availability of data collected during the prisoner intake process allowed for the retrospective scoring of the RASP. Second, the ODOC maintains information on inmate segregation placements and institutional rule violations that are necessary for evaluating the predictive accuracy of the risk instrument on these measures. Third, the ODOC is comprised of 14 correctional facilities and is ranked as the 27^thhighest state in terms of its total prisoner population count (Carson, 2020). The RASP was originally developed using a sample of inmates from one of the five most populous prison systems in the United States (Labrecque & Smith, 2019b), and this risk assessment has yet to be validated in another correctional system. The current investigation therefore serves as a test of the generalizability of the RASP in a more average-sized prison system with a more recent sample of offenders.

Measures

The RASP was designed to predict placement in segregation during one's period of incarceration. In its construction, Labrecque and Smith (2019b) used a broad definition of segregation that included placements in restrictive housing for any reason (i.e., discipline, protection, or other administrative purpose) and length of stay (i.e., one day or more). The current study focuses on a narrower and more serious definition of segregation that involved only assignments to the state's Intensive Management Unit (IMU) within specific follow-up periods. The IMU is reserved for inmates who represent a significant threat to institutional safety and security, including those who threaten or inflict bodily injury on another person, pose an immediate threat of escape, or promote or engage in disruptive group behavior. This type of confinement differs from that of disciplinary segregation which separates individuals from the general prisoner population as a specific response to a wide range of rule infractions for a specified period of time. In Oregon, sentences to disciplinary segregation can range from seven to 120 days depending upon the seriousness of the offense and one's misconduct history within the previous two years. Stays in IMU are designed to last seven months with compliant behavior to program expectations. One's time in the unit, however, can be extended for disruptive or non-compliant behavior. This dependent variable is operationalized here as any placement into the IMU within one-year (1 = yes, 0 = no) and two-years of admission (1 = yes, 0 = no).

This investigation also involved an outcome measure of institutional misconduct. The ODOC classifies its rule violations into five levels based on their severity. Level 1 represents the most serious types of infractions: assault, disturbance, distribution, escape, possession of a weapon, and unauthorized organization. This evaluation focused on these severe forms of inmate misconduct because of the heavy influence they have on IMU placement decisions. These are also the types of behavior with which Oregon prison officials reported they are most concerned in preventing (C. Prins, personal communication, July 19, 2019). This variable is defined here as any Level 1 violation within one-year (1 = yes, 0 = no) and two-years of admission (1 = yes, 0 = no).

The RASP consists of six items and was designed to be administered as part of the initial intake process: (1) age, (2) sentence length, (3) violent offense, (4) gang affiliation, (5) mental illness, and (6) custody rating. Age at intake is separated into five categories: aged 46 and older (0 points), aged 36 to 45.99 (1 point), aged 26 to 35.99 (2 points), aged 22 to 25.99 (3 points), and aged 21 or younger (4 points). Current sentence length is divided into four categories: less than 2 years (0 points), 2 to 2.99 years (1 point), 3 to 3.99 years (2 points), and 4 or more years (3 points). Violent offense is defined as having a conviction on one's current sentence for a crime that was violent in nature (1 = yes [1 point], 0 = no [0 points]). Gang affiliation is operationalized as any record of a known affiliation with a gang from a security threat group list (1 = yes [1 point], 0 = no [0 points]).Mental illness is designated as any evidence of a serious mental health disorder (1 = yes 1 point], 0 = no [0 points]). Initial custody rating is divided into three categories: minimum (0 points), medium (1 point), and close or higher. The total RASP score was retrospectively calculated by summing the points of these six items using administrative data that was collected during the admission process. Possible scale scores range from 0 (lower risk) to 12 (higher risk) and were further sub-divided into three risk categories: low-risk (total scores 0 to 4), moderate-risk (total scores of 5 to 8), and high-risk (total scores of 9 to 12). This study also included two demographic variables for gender (1 = male, 0 = female) and race (1 = white, 0 = non-white).

Sample Characteristics

Table 1 presents the descriptive characteristics of study participants. The inmates in this investigation were predominately male and white. The average age of the sample was 35 years old. Although more than 60% of them served less than two years in custody, nearly 20% were sentenced to four or more years. Almost half of the sample was incarcerated for a violent offense. Approximately one-fifth of the inmates were identified as having a gang affiliation, and one-third displayed evidence of a serious mental illness. More than half were classified at intake as minimum custody, roughly 30% as medium custody, and nearly 15% as close or higher custody. The mean RASP score of the sample was 4.1. According to its classification criteria, 60.9% of the prisoners were rated as low-risk, 32.9% as moderate-risk, and 6.2% as high-risk for segregation placement. Among the sample, 1.5% of the inmates were placed in long-term segregation within one year and 2.7% were transferred to this setting within two years. Additionally, 10% had violated a serious institutional rule within one year, and 14% engaged in this type of misbehavior within two years.

Table 1.

Descriptive Statistics of the Study Participants (N = 4,709).

Characteristic	%	n
Male	87.0	4,099
White	77.8	3,664
Age at intake^a	35.3	11.0
Aged 46 or older	18.7	879
Aged 36 to 45.99	23.8	1,122
Aged 26 to 35.99	37.2	1,754
Aged 22 to 25.99	13.3	628
Aged 21 or younger	6.9	326
Sentence length^a	2.5	4.8
Less than 2 years	61.4	2,893
2 to 2.99 years	12.7	598
3 to 3.99 years	6.5	308
4 or more years	19.3	910
Violent offense	47.2	2,222
Gang affiliation	22.1	1,041
Mental illness	35.4	1,666
Custody rating
Minimum	55.2	2,601
Medium	29.9	1,410
Close or higher	14.8	698
Long-term segregation placement (1 year)	1.5	70
Long-term segregation placement (2 years)	2.7	126
Serious institutional misconduct (1 year)	10.0	472
Serious institutional misconduct (2 years)	14.0	660
RASP risk score^a	4.1	2.5
Low-risk	60.9	2,868
Moderate-risk	32.9	1,549
High-risk	6.2	292
RASP-OR risk score^a	−0.6	1.3
Low-risk	65.1	3,067
Moderate-risk	28.5	1,361
High-risk	5.9	281

Note. RASP = Risk Assessment for Segregation Placement. RASP-OR = Risk Assessment for Segregation Placement-Oregon Revised.

Reported values include mean and standard deviation.

Analysis

The formulation of the revised RASP scale—the Risk Assessment for Segregation Placement–Oregon Revised (RASP-OR)—began with a series of binary logistic regression analyses. Using random samples of approximately 50% each, 10 regressions were conducted with the six RASP factors predicting long-term segregation placement within two years of prison admission. The unstandardized beta (b) coefficients for each item were recorded and averaged across the separate models. These values then served as the new item weights in the RASP-OR. Total scores on the revised assessment were calculated by multiplying each original item rating by its new item weight and summing the six values.

This study evaluated instrument utility by first examining the percentage of segregation placements and serious misconducts found within the low-, moderate-, and high-risk groups of the RASP and RASP-OR assessments. It further conducted a series of chi-square analyses to determine if there were any statistically significant differences in outcomes across the risk levels of both versions of the instrument. This investigation further assessed the predictive validity of these risk tools on the segregation and misconduct outcomes through the use of receiver operating characteristic (ROC) curve analysis. This analytic procedure generates an area under the curve (AUC) statistic that estimates how well the assessment identifies the correct outcomes in the sample. Possible AUC values range from 0 to 1, with .5 signifying the assessment performs no better than chance in correctly identifying the outcomes and 1 indicating the assessment predicts outcomes perfectly. For context, the results are interpreted according to the guidelines developed by Rice and Harris (2005) for determining small (AUC = .556), moderate (AUC = .639), and large (AUC = .714) effect sizes. The AUCs are reported along with 95% confidence intervals (CIs) as a mechanism for determining the range of values that are reasonably certain to contain the true effect size (Smithson, 2002). Additionally, if two 95% CIs are found not to overlap, these values are considered to be statistically different from one another (see Cummings, 2012).

Results

Table 2 summarizes the averaged logistic regression results of the six RASP items predicting segregation placement within two years of admission. The values of these unstandardized beta (b) coefficients were used as the item weights in the construction of the RASP-OR. More specifically, the total score on the revised assessment was calculated by multiplying each of the original RASP item ratings by the new weights and summing the six values. Possible scores on the RASP-OR ranged from −6 to 4, and the mean score of the sample was −0.6 (see Table 1). The RASP-OR scores were used to trifurcate the sample by risk level: 65.1% of the prisoners were rated as low-risk (total scores of −6.00 to −0.01), 28.5% as moderate-risk (total scores of 0.00 to 2.38), and 5.9% as high-risk for segregation placement (total scores of 2.39 to 4.00).

Table 2.

Average Logistic Regression Beta (b) Coefficients of RASP Items Predicting Segregation Placement.

RASP item number and description	Average b
1. Age	−0.0322
2. Sentence length	−0.0663
3. Violent offense	0.4766
4. Gang	2.1492
5. Mental illness	0.5217
6. Custody rating	0.6504

Note. RASP = Risk Assessment for Segregation Placement.

Figure 1 illustrates the one- and two-year segregation rates by RASP type and risk category. Across both assessment types, the segregation rates increased in the hypothesized direction. More specifically, 0.4% of the low-risk inmates on the RASP had a segregation placement within one year and 0.9% had a placement within two years, followed by 2.6% and 4.7% of moderate-risk, and 4.8% and 9.6% of high-risk. For the RASP-OR, 0.5% of the low-risk had a segregation placement within one year and 0.7% had a placement within two years, followed by 2.1% and 4.4% of moderate-risk, and 8.9% and 16% of high-risk. Chi-square analyses indicated that all of these relationships were statistically significant (p < .001).

Figure 1.

Percentage of segregation placements by RASP type and risk level at one- and two-year follow-up periods.

Figure 2 depicts the one- and two-year serious misconduct rates by RASP type and risk category. Across both versions, the rates of misconduct increased with each successive risk level. More specifically, 5% of the low-risk inmates on the RASP engaged in a serious institutional rule violation within one year and 6.3% engaged in this type of misbehavior within two years, followed by 15.8% and 22.9% of moderate-risk, and 28.8% and 43.2% of high-risk. For the RASP-OR, 3.6% of the low-risk engaged in a serious rule violation within one year and 5.2% engaged in this type of misbehavior within two years, followed by 18% and 25.4% of moderate-risk, and 41.3% and 54.8% of high-risk. All of these relationships were also found to be statistically significant at the .001 level.

Figure 2.

Percentage of serious misconducts by RASP type and risk level at one- and two-year follow-up periods.

Table 3 provides the results of the ROC analyses that examine the relationship between the total scores from the two variants of the risk assessment and the segregation placement and serious misconduct outcomes at both follow-up periods. All of the AUC values were determined to be statistically significant (p < .001) and large in magnitude. Across both outcomes, the point estimates for the RASP-OR were found to be higher than those of the RASP. The 95% CIs between the two assessment types overlapped with each other in the segregation placement analyses but did not overlap in the serious misconduct analyses. This latter finding indicates that the RASP-OR produced a statistically significant improvement over the RASP in identifying the perpetrators of serious institutional misconduct at both the one- and two-year follow-up periods.

Table 3.

Receiver Operating Characteristic Curve Analyses of RASP and RASP-OR on the Segregation Placement and Serious Misconduct Outcomes, by Follow-up Period.

	Segregation placement		Serious misconduct
	1 year	2 years	1 year	2 years
RASP	.740 [.686,.794]	.770 [.731,.808]	.723 [.699,.746]	.755 [.735,.774]
RASP-OR	.800 [.742,.857]	.835 [.798,.872]	.792 [.769,.814]	.801 [.782,.820]

Note. Reported values are Area Under the Curve (AUC) statistic and 95% confidence intervals.

RASP = Risk Assessment for Segregation Placement. RASP-OR = Risk Assessment for Segregation Placement-Oregon Revised.

Discussion

The use of restrictive housing does not appear to be an effective mechanism for making prisons safer and more secure environments (Gendreau & Goggin, 2019; Labrecque & Smith, 2019a; Morgan et al., 2016; Steiner & Cain, 2016). There are also growing concerns about the potential negative psychological effects that this setting might have on its inhabitants (see e.g., Frost & Monteiro, 2016; Kapoor & Trestman, 2016; Morgan et al., 2016). In response, prison administrators are increasingly searching for better ways to manage their inmate populations. One viable alternative for this task involves providing increased case management services and other proactive programmatic opportunities to the most dangerous and disruptive prisoners in the prison system (see also Labrecque, 2018a; and Smith, 2016). The success of this offender management strategy, however, hinges on the department's ability to prospectively and accurately identify its most problematic inmates. Although a number of risk assessment tools exist that have been shown to be valid predictors of prison misconduct (e.g., LSI-R, HCR-20, PCL-R; see Campbell et al., 2009; Kroner & Mills, 2001; Singh et al., 2011), only two scales have been designed specifically to predict placements in restrictive housing (i.e., the RASP, Labrecque & Smith, 2019a, 2019b; and the RAST, Helmus et al., 2019); and neither has been revalidated on a different sample of prisoners. As inmates held in segregated confinement settings are known to include the most serious and chronic violators of the institutional rules (Labrecque, 2018b; Steiner & Cain, 2016), this group makes the ideal target for the limited services and interventions that are available in prison.

The results of this study revealed that the RASP was a strong predictor of long-term segregation placement and serious institutional misconduct at both the one- and two-year follow-up periods (range of AUCs from .723 to.770). Further, it found that weighting and norming the scoring of the RASP items on the local population was an effective method for increasing the predictive strength of the tool (range of AUCs from .792 to.835). The point estimates generated for the RASP-OR were greater than those for the RASP across all four outcomes examined (range of differences in AUC values between .046 and .069). The RASP-OR was also shown to be a statistically significant better predictor of serious institutional rule violations at both follow-up durations (p < .05).

These findings suggest that the RASP (and the RASP-OR) can be used during the initial intake process to prospectively identify prisoner risk for long-term segregation placement and perpetration of serious institutional misconduct. Armed with this knowledge, prison officials could make more informed inmate supervision and treatment decisions. In both versions of the assessment, a relatively small percentage of prisoners were identified as high-risk (≈6%; see Table 1). Members of this group, however, were much more likely to be placed in long-term segregation and violate a serious institutional rule compared to those in the low- and moderate-risk categories, respectively (see Figures 1 and 2). From a rehabilitative standpoint, it may be advantageous for prison officials to increase supervision practices with high-risk prisoners as well as provide these inmates with greater access to case management services and treatment interventions. There are strong theoretical reasons to suspect that this offender management strategy would help improve institutional safety and security (Andrews et al., 1990; Bonta & Andrews, 2017; Lowenkamp et al., 2006).

The replication of the findings from Labrecque and Smith (2019b) with inmates in a different prison system provides greater confidence in the generalizability of the RASP as a valid predictor of segregation placement and institutional misconduct. As prison administrators seek ways to reduce institutional violence and disorder, the RASP is a practical method for helping authorities make more informed security threat management decisions. These results should encourage prison systems to adopt tools such as the RASP into their classification protocols. Also, it may be beneficial for prison officials to consider norming and weighting these risk instruments on their own local offender populations. The RASP-OR not only displayed better predictive validity than the RASP, but the weighted scoring scheme also has the advantage of providing a wider range of possible scores which allow for greater flexibility in creating the cut-points for the risk categories. For example, if a prison system only has the resources to provide intensive case management and treatment services to 3% of its population, then the operationalization of high-risk can be modified to include prisoners with the top 3% of scores. However, if a prison system is capable of providing more intense services to 10% of its population, then the definition of high-risk can be expanded to include those with the top 10% of scores. To be clear, it is better for prison systems to provide a higher percentage of inmates with more intensive treatment interventions. Given that fiscal constraints limit the number of prisoners who can receive these types of services, however, it is critical for prison officials to prioritize the individuals who pose the greatest threat to institutional safety and order (i.e., those with higher RASP scores).

The security threat management strategy endorsed here suggests that prison officials should provide more intensive case management services and treatment programs to the inmates who possess the greatest probability for being placed in long-term segregation and violating the most serious institutional rules. While the current study provides support for the utility of the RASP and RASP-OR in identifying such risk, it does not assess which types of interventions are most effective in reducing the occurrences of these adverse events. As such, this remains an area in need of further exploration. As a note of caution, engaging high-risk inmates in treatment programs will undoubtedly be more difficult than working with lower risk prisoners. High-risk offenders are known to be less motivated for treatment and to experience greater rates of failure compared to those who are lower risk (Bonta & Andrews, 2017). Prison officials and researchers, nevertheless, should embrace this challenge because the successful implementation of this offender management strategy holds great promise for making prisons safer and more secure environments.

Footnotes

Acknowledgments

Portions of this manuscript were presented at the 2019 Annual Meeting of the American Society of Criminology, San Francisco, CA.

I would like to thank Craig Prins, Rebecca Krueger, and the other members of the Security Threat Management (STM) Re-Alignment Team from the Oregon Department of Corrections (ODOC) for their assistance in making the research possible.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Notes

Author Biography

Ryan Labrecque, PhD, is an assistant professor in the University of Central Florida Department of Criminal Justice. His research interests focus on offender classification and assessment, correctional rehabilitation, restrictive housing, and the transfer of knowledge to practitioners and policymakers.

References

Andrews

D. A.

Bonta

Wormith

J. S

. (2006). The recent past and near future of risk and/or need assessment. Crime and Delinquency, 52(1), 7–27. https://doi.org/10.1177/0011128705281756

Andrews

D. A.

Zinger

Hoge

R. D.

Bonta

Gendreau

Cullen

F. T

. (1990). Does correctional treatment work? A clinically relevant and psychologically informed meta-analysis. Criminology; An Interdisciplinary Journal, 28(3), 369–404. https://doi.org/10.1111/j.1745-9125.1990.tb01330.x

Beck

A. J.

(2015). Use of restrictive housing in U.S. Prisons and jails, 2011–12. Bureau of Justice Statistics.

Bonta

. (2002). Offender risk assessments: Guidelines for selection and use. Criminal Justice and Behavior, 29(4), 355–379. https://doi.org/10.1177/0093854802029004002

Bonta

Andrews

D. A.

(2017). The psychology of criminal conduct (6th ed.). Routledge.

Butler

H. D.

Griffin

O. H.

Johnson

W. W

. (2012). What makes you the worst of the worst?”: An examination of state policies defining supermaximum confinement. Criminal Justice Policy Review, 24(6), 676–694. https://doi.org/10.1177/0887403412465715

Butler

H. D.

Solomon

Spohn

. (2018). Programming in restrictive housing: Considerations for improving outcome evaluations. Criminal Justice and Behavior, 45(8), 1174–1191. https://doi.org/10.1177/0093854818780450

Butler

H. D.

Steiner

Makarios

Travis

. (2017). Assessing the effects of exposure to supermax confinement on offender post-release behaviors. The Prison Journal, 97(3), 275–295. https://doi.org/10.1177/0032885517703925

Campbell

M. A.

French

Gendreau

. (2009). The prediction of violence in adult offenders: A meta-analytic comparison of instruments and methods of assessment. Criminal Justice and Behavior, 36(6), 567–590. https://doi.org/10.1177/0093854809333610

10.

Carson

E. A.

(2020). Prisoners in 2018. Bureau of Justice Statistics.

11.

Cochran

Toman

Mears

Bales

. (2018). Solitary confinement as punishment: Examining in-prison sanctioning disparities. Justice Quarterly, 35(3), 381–411. https://doi.org/10.1080/07418825.2017.1308541

12.

Cullen

F. T.

Latessa

E. J.

Burton

Jr. Lombardo

(1993). The correctional orientation of prison wardens: Is the rehabilitative ideal supported? Criminology; An Interdisciplinary Journal, 31(1), 69–92. https://doi.org/10.1111/j.1745-9125.1993.tb01122.x

13.

Cummings

(2012). Understanding the new statistics: Effect sizes, confidence intervals, and meta-analysis. Routledge.

14.

Cunningham

M. D.

Sorensen

J. R

. (2006). Actuarial models for assessing prison violence risk: Revisions and extensions for the risk assessment scale for prison (RASP). Assessment, 13(3), 253–265. https://doi.org/10.1177/1073191106287791

15.

DiIulio

J. J.

Jr. (1987). Governing prisons: A comparative study of correctional management. The Free Press.

16.

Duwe

. (2019). Better practices in the development and validation of recidivism risk assessments: The minnesota Sex offender screening tool–4. Criminal Justice Policy Review, 30(4), 538–564. https://doi.org/10.1177/0887403417718608

17.

Duwe

Kim

. (2016). Sacrificing accuracy for transparency in recidivism risk assessment: The impact of classification method on predictive performance. Corrections: Policy, Practice and Research, 1(3), 155–176. https://doi.org/10.1080/23774657.2016.1178083

18.

French

S. A.

Gendreau

. (2006). Reducing prison misconducts: What works! Criminal Justice and Behavior, 33(2), 185–218. https://doi.org/10.1177/0093854805284406

19.

Frost

N. A.

Monteiro

C. E.

(2016). Administrative segregation in prison. In Garcia

(Ed.), Restrictive housing in the U.S.: Issues, challenges, and future directions (pp. 1–47). National Institute of Justice.

20.

Gendreau

Goggin

(2019). Solitary confinement and punishment: Effects on misconducts and recidivism. In Polaschek

D. L. L.

Day

Hollin

C. R.

(Eds.), The wiley international handbook of correctional psychology (pp. 131–143). Wiley-Blackwell.

21.

Gendreau

Goggin

E. & Law

& A. (1997). Predicting prison misconducts. Criminal Justice and Behavior, 24(4), 414–431. https://doi.org/10.1177/0093854897024004002

22.

Gendreau

Labrecque

R. M.

(2018). The effects of administrative segregation: A lesson in knowledge cumulation. In Wooldredge

Smith

(Eds.), Oxford Handbook on prisons and imprisonment (pp. 340–366). Oxford University Press.

23.

Georgiou

. (2019). Weights matter: Improving the predictive validity of risk assessments for criminal offenders. Journal of Offender Rehabilitation, 58(2), 92–116. https://doi.org/10.1080/10509674.2018.1562504

24.

Hamilton

Kigerl

Campagna

Barnoski

Lee

van Wormer

Block

. (2015). Designed to fit: The development and validation of the STRONG-R recidivism risk assessment. Criminal Justice and Behavior, 43(2), 230–263. https://doi.org/10.1177/0093854815615633

25.

Helmus

L. M.

Johnson

Harris

A. J. R

. (2019). Developing and validating a tool to predict placements in administrative segregation: Predictive accuracy with inmates, including indigenous and female inmates. Psychology, Public Policy, and Law, 25(4), 284–302. https://doi.org/10.1037/law0000201

26.

Huebner

. (2003). Administrative determinants of inmate violence: A multilevel analysis. Journal of Criminal Justice, 31(2), 107–117. https://doi.org/10.1016/S0047-2352(02)00218-0

27.

Kapoor

Trestman

(2016). Mental health effects of restrictive housing. In Garcia

(Ed.), Restrictive housing in the U.S.: Issues, challenges, and future directions (pp. 199–232). National Institute of Justice.

28.

Kroner

D. G.

Mills

J. F

. (2001). The accuracy of five risk appraisal instruments in predicting institutional misconducts and new convictions. Criminal Justice and Behavior, 28(4), 471–489. https://doi.org/10.1177/009385480102800405

29.

Labrecque

Mears

(2019). Prison system vs. critics’ views on the use of restrictive housing: Objective risk classification or ascriptive assignment? The Prison Journal, 99(2), 194–218. https://doi.org/10.1177/0032885519825492

30.

Labrecque

R. M.

(2018a). Specialized or segregated housing units: Implementing the principles of risk, needs, and responsivity. In Dodson

K. D.

(Ed.), Handbook of offenders with special needs (pp. 69–83). Routledge.

31.

Labrecque

R. M.

Smith

(2019a). The impacts of restrictive housing: A systematic review of the evidence. In Huebner

Frost

(Eds.), Handbook on corrections and sentencing (Vol. 3):The collateral consequences of punishment (pp. 290–310). Routledge.

32.

Labrecque

R. M.

Smith

. (2019b). Reducing institutional disorder: Using the inmate risk assessment for segregation placement (RASP) to triage treatment services at the front-end of prison sentences. Crime and Delinquency, 65(1), 3–25. https://doi.org/10.1177/0011128717748946

33.

Labrecque

R. M

. (2018b). Taking stock: A meta-analysis of the predictors of restrictive housing. Victims and Offenders, 13(5), 675–692. https://doi.org/10.1080/15564886.2018.1468367

34.

Lovins

B. K.

Latessa

E. J.

May

Lux

. (2018). Validating the ohio risk assessment system community supervision tool with a diverse sample from texas. Corrections: Policy, Practice and Research, 3(3), 186–202. https://doi.org/10.1080/23774657.2017.1361798

35.

Lowenkamp

C. T.

Latessa

E. J.

Holsinger

A. M

. (2006). The risk principle in action: What have we learned from 13,676 offenders and 97 correctional programs? Crime and Delinquency, 52(1), 77–93. https://doi.org/10.1177/0011128705281747

36.

Mears

D. P.

(2016). Critical research gaps in understanding the effects of prolongedtime in restrictive housing on inmates and the institutional environment. In Garcia

(Ed.), Restrictive housing in the U.S.: Issues, challenges, and future challenges (pp. 233–295). National Institute of Justice.

37.

Mears

D. P.

Castro

J. L

. (2006). Wardens’ views on the wisdom of supermax prisons. Crime and Delinquency, 52(3), 398–431. https://doi.org/10.1177/0011128705279484

38.

Mears

D. P.

Hughes

Pesta

G. B.

Bales

W. D.

Brown

J. M.

Cochran

J. C.

Wooldredge

. (2019). The new solitary confinement?: A conceptual framework guiding and assessing research and policy on “restrictive housing.” Criminal Justice and Behavior, 46(10), 1427–1444. https://doi.org/10.1177/0093854819852770

39.

Meyers

T. J.

Infante

Wright

K. A

. (2018). Addressing serious violent misconduct in prison: Examining an alternative form of restrictive housing. International Journal of Offender Therapy and Comparative Criminology, 62(14), 4585–4608. https://doi.org/10.1177/0306624X18778451

40.

Morgan

Gendreau

Smith

Gray

Labrecque

MacLean

Van Horn

Bolanos

Batastini

Mills

. (2016). Quantitative synthesis of the effects of administrative segregation on inmates’ well-being. Psychology, Public Policy, and Law, 22(4), 439–461. https://doi.org/10.1037/law0000089

41.

Pompoco

Wooldredge

Lugo

Sullivan

Latessa

E. J

. (2017). Reducing inmate misconduct and prison returns with facility education programs. Criminology and Public Policy, 16(2), 515–547. https://doi.org/10.1111/1745-9133.12290

42.

Randol

B. M.

Campbell

C. M

. (2017). Macro-correlates of inmate violence: The importance of programming in prison order. The Prison Journal, 97(4), 451–474. https://doi.org/10.1177/0032885517711975

43.

Rice

M. E.

Harris

G. T

. (2005). Comparing effect sizes in follow-up studies: ROC area, Cohen’s d, and r. Law and Human Behavior, 29(5), 615–620. https://doi.org/10.1007/s10979-005-6832-7

44.

Schenk

A. M.

Fremouw

W. J

. (2012). Individual characteristics related to prison violence: A critical review of the literature. Aggression and Violent Behavior, 17(5), 430–442. https://doi.org/10.1016/j.avb.2012.05.005

45.

Singh

J. P.

Grann

Fazel

. (2011). A comparative study of violence risk assessment tools: A systematic review and metaregression analysis of 68 studies involving 25,980 participants. Clinical Psychology Review, 31(3), 499–513. https://doi.org/10.1016/j.cpr.2010.11.009

46.

Smith

(2016). Toward an understanding of “what works” in segregation. In Garcia

(Ed.), Restrictive housing in the U.S.: Issues, challenges, and future challenges (pp. 331–366). National Institute of Justice.

47.

Smithson

(2002). Confidence intervals. Sage.

48.

Steiner

Butler

H. D.

Ellison

J. M

. (2014). Causes and correlates of prison inmate misconduct: A systematic review of the evidence. Journal of Criminal Justice, 42(6), 462–470. https://doi.org/10.1016/j.jcrimjus.2014.08.001

49.

Steiner

Cain

M. A.

(2016). The relationship between inmate misconduct, institutional violence, and administrative segregation: A systematic review of the evidence. In Garcia

(Ed.), Restrictive housing in the U.S.: Issues, challenges, and future directions (pp. 165–197). National Institute of Justice.

50.

Winterdyk

Ruddell

. (2010). Managing prison gangs: Results from a survey of U.S. Prison systems. Journal of Criminal Justice, 38(4), 730–736. https://doi.org/10.1016/j.jcrimjus.2010.04.047

51.

Wooldredge

Steiner

. (2015). A macro-level perspective on prison inmate deviance. Punishment and Society, 17(2), 230–257. https://doi.org/10.1177/1462474515577151