An Examination of Recidivism Among Inmates Released From a Private Reentry Center and Public Institutions in Colorado

Abstract

In this study, we examine comparative rates of recidivism for Colorado inmates released from a private reentry center and public facilities. Using a sample of 6,102 inmates released from 2008 to 2012, we measure overall recidivism and recidivism for a new crime. Applying a quasi-experimental methodology, we find that overall rates of recidivism are comparable. Prior to matching, 46.8% of the treatment and 61.3% of the comparison group recidivated, and the difference is statistically significant. However, after matching on relevant covariates, the difference was reduced to nonsignificance. In contrast, those released from the private facility are more likely to return to prison for a new offense. Approximately 14% of those in the private facility committed a new crime compared with 9% prematching and 8% postmatching. Despite these modest differences in the rate of recidivism, the overall time to return to prison is comparable between the groups.

Keywords

recidivism private prison quasi-experimental design propensity score analysis

Introduction

One in 37 individuals is currently under some form of correctional supervision in the United States (U.S.), straining federal, state, and local government budgets (Kaeble & Glaze, 2016). On average, it costs US$31,286 to house an individual in prison for 1 year (Henrichson & Delaney, 2012). Furthermore, the correctional system places a financial burden on the public, costing taxpayers US$39 billion over the last decade (Henrichson & Delaney, 2012). Since 1999, there has been a steady growth in private sector correctional facilities and programming as well as a political movement to increase privatization in the government sector (Chang & Thompkins, 2002; Kleis, 2010). Privatization of the correctional system can come in the form of private for-profit or nonprofit facilities such as prisons, jails, reentry centers, and various outpatient community corrections services (Bayer & Pozen, 2005). Since the U.S. correctional system began the shift toward privatization in the early 1980s (Kleis, 2010), the use of private prisons has grown rapidly (Lee, 2012). For example, since 1999, the private prison population has grown by 83%, with approximately 126,000 individuals housed in private state and federal prisons in 2015 (Geiger, 2017). Private correctional facilities now account for 8% of all state and federal prisons (Geiger, 2017).

The privatization of the correctional system relies on the rationale that privately run facilities will alleviate some of the financial burden that the government assumes, as well as decrease recidivism through the use of rehabilitative programming. Any decrease in recidivism further contributes to the economic benefit of these programs. For example, in an examination of a prerelease substance abuse treatment program run by Community Education Centers, Inc. (CEC), French, Fang, and Fretz (2010) found that the treatment program yielded significant economic savings through a lower recidivism rate and operating costs. However, the wider body of research in the area is mixed with regard to whether private facilities and programs are as cost-effective (Lundahl, Kunz, Brownell, Harris, & Van Vleet, 2009). In a meta-analysis of the cost-effectiveness of private corrections, it was found that private correctional facilities were more cost-effective than public facilities in half of the studies (Lundahl et al., 2009).

Although a great deal of research has assessed various facets of privately run correctional facilities, including the quality of care (e.g., Greene, 2000), safety (e.g., Bayer & Pozen, 2005), and issues related to rehabilitation (e.g., Kleis, 2010), the assessment of recidivism rates of participants has received less of an empirical focus. Many privately run correctional facilities, including reentry centers, have programming that is aimed explicitly at decreasing recidivism (Bales, Bedard, Quinn, Ensley, & Holley, 2005). Again, results from previous investigations have been mixed. The inconclusiveness of the findings is likely due, in large part, to methodological differences across studies. To that end, the current study seeks to address the issue of effectiveness, comparing private and public institutions in reducing recidivism.

Previous Studies of Recidivism Between Private and Public Facilities

Previous research has found support for the effectiveness of treatment to reduce recidivism in private facilities and programs. For example, Fretz, Heilbrun, and Brown (2005) and Heilbrun and colleagues (2008) examined the differences in recidivism between inmates housed at a CEC in New Jersey; both point to lower recidivism rates among the private prisoners. In comparison with Department of Corrections’ inmates released from work camps, the private facilities fared better with regard to rearrests, reconviction, and reincarceration (Fretz et al., 2005), and these differences persisted at all follow-up time measures (up to 12 months). In an analysis of female offenders in a CEC facility with gender-specific programming compared with women released from a prison in New Jersey, Heilbrun and colleagues (2008) found that women with CEC programming had lower rates of rearrest. Taken together, these results suggest that innovative treatment programs offered outside of the traditional correctional system and tailored to specific populations (e.g., gender-specific programming) may facilitate successful reentry to the community. Although in both studies, the CEC (treatment) group was higher risk, neither was able to account explicitly for selection bias. That is to say, characteristics associated with the likelihood of recidivism may be differentially distributed among these groups. Therefore, these investigations were not able to systematically isolate the impact of the program on recidivism.

Results have been mixed in comparisons between private and public facility recidivism rates. Several quasi-experimental studies find results favorable to private prisons. Using data on male and female inmates in Florida, the inmates of two private prison facilities were matched to public prison inmates, based on demographic and offense characteristics (Lanza-Kaduce, Parker, & Thomas, 1999). Except for technical violations, offenders released from private prisons had a lower overall recidivism rate comparatively (measured within 12 months of release). In addition, the seriousness of the reoffense was lower than those released from the public facilities and the time to recidivate was comparable between the two groups (Lanza-Kaduce et al., 1999). In a follow-up study, the authors were able to replicate the results while extending the postrelease analysis period to 4 years (Lanza-Kaduce & Maggard, 2001). However, these studies suffered from methodological problems, including not taking into consideration the amount of exposure to the private facility; inmates may be transferred between facilities and analyses must isolate the impact of the level of exposure.

Farabee and Knight (2002) took a slightly different approach to the analysis of Florida inmate recidivism, specifying their treatment group as those that were in the private facility 6 months before release. The authors’ choice to impose the restriction of at least a 6-month period of incarceration was aimed at isolating the impact of the institution, as shorter stays may not produce a measurable impact. They found that up to 3 years after release, both facilities had the same recidivism risk for male inmates (both adult and youthful), but female inmates from the private prison (both adult and youthful) had a lower risk of recidivism. In addition, Farabee and Knight (2002) conducted qualitative analyses with personnel at the institutions. Although, the employees perceived the outcomes to be comparable between public and private prisons, they did note that private prisons allowed for greater flexibility with regard to programming.

Other studies that employ rigorous methodological analyses and build into the analyses the relative exposure to the private facility do not find reduced recidivism at private prisons. In 2005, a statewide study compared the documented recidivism rates of privately and publicly run correctional facilities in the state of Florida for up to 60 months post release (Bales et al., 2005). As the first analysts to take into consideration the proportion of time spent in the private facility, Bales and colleagues (2005), using survival analysis, report that the facilities did not differ significantly in rates of recidivism. Furthermore, these results held when the population of interest was adult males, adult females, and youthful male offenders. Likewise, in a comparison of Florida for-profit and nonprofit private facilities compared with state-run facilities, juveniles released from for-profit private facilities had a higher risk of being readjudicated or rearrested at the 12-month follow-up, compared with nonprofit and state-run facilities (Bayer & Pozen, 2005). Although for-profit private facilities also demonstrated less operating costs, the authors suggest that the added economic burden of the increase in recidivism negates these savings.

Replicating and extending the methodology in Bales and colleagues (2005) to Oklahoma, Spivak and Sharp (2008) found that adult men in privately run prisons were 5% more likely (35%-30%) to be reincarcerated than public prison inmates at the 4-year follow-up period. In contrast, female offenders were less likely to recidivate, but this difference was not statistically significant, likely due to the sample size. Finally, in a recent study of recidivism using a sample from Minnesota, privately run prison inmates were matched with prison inmates in public facilities (Duwe & Clark, 2013). Compared with public releases, privately run prison inmates had a modestly higher risk of rearrest and reconviction after release (Duwe & Clark, 2013).

Prison Privatization in Colorado

In 1995, Colorado House Bill 1352 provided the opportunity for the privatization of corrections via a proposal process. This led to the creation of the 1999 Private Prisons Monitoring Unit, charged with the overseeing of private prison contracts and operations. The Private Prisons Monitoring Unit works with the Colorado Department of Corrections’ Central Classification Unit to ensure that all inmates housed within private institutions are appropriately classified and assigned to a facility with an appropriate custody level (with the maximum security level for a private prison inmate being medium security as outlined in Colorado Revised Statutes 17-1-104.9, Custody levels for state inmates at private prison). Additional provisions include Colorado Department of Corrections’ training and the provision of medical, food, and educational services. In addition, the Private Prisons Monitoring Unit assigns individual monitors to each private facility who are required to spend a minimum of 20 hours per week at their assigned facility to oversee operations. There are also additional medical, mental health, and food service monitors that oversee all of these services at all Colorado Department of Corrections’ private facilities. There are currently five private prison contracts in Colorado. In 2009, during the peak of Colorado’s rate of incarceration, Colorado’s private prisons accounted for approximately 27% of all incarcerated inmates (Wells, O’Keefe, & Allen, 2013).

Current Study

This study contributes to the current body of knowledge regarding the effectiveness of privately run institutions that offer unique programming to reduce recidivism in several ways. First, this research examines the impact of privatization on recidivism in a state that has not been explored before. Second, this study isolates a facility that is mandated to provide prerelease services. As Wright (2010) suggested, private facilities may be more advantageous because they have the possibility to provide more innovative programming to facilitate successful reentry into the community. Third, the investigation employs a quasi-experimental design to isolate the impact of the facility type in a causal design. This is an important research strategy because it is logical that individuals assigned to a privately run facility may differ on key factors that affect their likelihood of recidivism, and, therefore, any observed differences may be spurious. Following the more methodologically rigorous studies (Bales et al., 2005; Duwe & Clark, 2013; Spivak & Sharp, 2008), we apply a rigorous statistical design to control for variables that affect recidivism rates between these facilities. That is, propensity score matching allows for the analysis of the counterfactual, the recidivism rate given assignment to the comparison (i.e., public) facility.

The Research Setting

The focus of the investigation is the evaluation of a private facility operated by CEC. This particular Colorado facility is a 700-bed male facility that opened in 2005 and has been designated as a medium-security prison by the Colorado Department of Corrections, working with male offenders who are either on their way back to prison due to a parole violation or on their way out of prison to community release programs. It is important to note that this facility is not an alternative to prison, but is part of the continuum of programming for inmates transitioning to the community. The center uses a Positive Peer Culture, providing a number of programs that include an assessment of the individual’s current risk/needs. The individual-based curriculum has a 6-month time frame, with allowances, for instance, where an individual remains at the facility for a briefer period. At the end of the individual’s stay, an assessment report is generated that outlines the person’s risk/needs factors, programs completed, strengths and protective factors, and further treatment recommendations. The assessment reports are sent to the community boards where the individual resides, providing up-to-date information for making an informed decision. The assessment reports are also used by the Colorado Board of Parole in making release decisions (CEC, 2015).

This reentry center was designed using the model of Talbot Hall in New Jersey as an interim step prior to inmates having full access to the community. The programming includes the assessment of inmates’ treatment needs in addition to their preparedness for release into the community. As Wojtowicz and Liu (2006) noted at the conclusion of the assessment and treatment plan, recommendations are made for an inmate’s continued participation within the community. The offenders are also further classified according to these assessments. Using data from the New Jersey Contract Administration Tracking System (the database currently used to place and track New Jersey State offenders in community corrections), Wojtowicz and Liu (2006) found that there was a decreasing trend of “inmate walkaway rates” during a 7-year observation period; the establishment of the assessment and treatment center approach had a significant and positive impact on the likelihood of walkaway behaviors.

Method

Data

Data came from Colorado’s Department of Corrections and included all inmates released from January 1, 2008, to December 31, 2012, resulting in a sample of 20,087 total inmates. Of those, 6,825 had been a private facility client at some point during their incarceration. Given that inmates travel between facilities, it is important to isolate the proportion of time spent in a public or private facility. Following the methodology of prior studies (Bales et al., 2005; Duwe & Clark, 2013; Spivak & Sharp, 2008), we incorporate the proportion of time spent at the facility in designing the research groups. To obtain the “cleanest” estimates while still maintaining sample size, we restricted our treatment group to inmates who spent 75% or more at the private facility compared with inmates who spent 75% or more in public facilities. This results in a final sample of 6,102 inmates, of whom 2,024 were considered private (i.e., treatment group) and 4,078 public (i.e., comparison group).

Recidivism

This study operationalizes recidivism in two ways. First, a general reincarceration measure is used to capture cases in which the inmate returned to prison. Second, a new crime measure was constructed to differentiate between those who recidivated due to a technical violation compared with a new offense. Overall, approximately 61% of those released from the private facility recidivated compared with 45% of those released from the public facility. When the type of recidivism is taken into account, approximately 9% of the inmates released from the public facilities returned to prison for a new crime compared with 13% of the private inmate sample.

Matching Criteria

At a minimum, matching criteria should include demographic (i.e., sex, race/ethnicity, and age) and offense-related information (i.e., type of offense and prior record; Nagin, Cullen, & Jonson, 2009). Given that this sample was all men, sex was not considered in the matching criteria, but the other demographic information was included. Furthermore, education and gang status were also included as individual-level predictors. As shown in Table 1, the majority of respondents were White, with the largest minority being Hispanic. Approximately 20% of inmates had completed high school. For offense type, inmates were matched on offense degree, which ranged from 0 to 6. Prior record was accounted for with a count of total incarcerations; prior incarcerations ranged from 1 (indicating first-time offenders) to 7. Approximately 73% of the total sample were first-time offenders. Finally, the level of supervision assigned to the offender, their average sentence length, and information regarding their needs were also included.

Table 1.

Descriptive Statistics on Matching Criteria.

Variable (% missing)	State(n = 4,078)	Private(n = 2,024)
Variable (% missing)	ƒ(%)	ƒ(%)
Demographics
Race/ethnicity (0.00%)
White	1,906 (46.74%)	1,011 (49.95%)
African American	655 (16.06%)	421 (20.80%)
Latino/Hispanic	1,377 (33.77%)	529 (26.14%)
Other	140 (3.43%)	63 (3.11%)
Average age (SD; 0.00%)	35.84 (10.63)	35.89 (10.07)
High school (5.88%)	806 (20.85%)	381 (20.30%)
Gang affiliation (0.00%)	936 (22.95%)	423 (20.90%)
Offense/incarceration
Mean offense degree (SD; 0.41%)	5.49 (0.02)	5.52(0.02)
Mean level of supervision inventory (SD; 1.98%)	30.41 (0.11)	30.31 (0.15)
Average final sentence (days; 0.00%)	880.88 (17.59)	177.08 (1.91)
Mean number of priors (SD; 0.00%)	1.38 (0.01)	1.42 (0.02)
Needs
Mean medical (SD; 0.00%)	1.80 (0.01)	1.70 (0.02)
Mean mental health (SD; 0.00%)	2.21 (0.01)	2.14 (0.01)
Mean substance abuse (SD; 0.34%)	3.28 (0.02)	3.42 (0.02)
Mean academic (SD; 0.00%)	2.64 (0.02)	2.52 (0.02)
Mean vocational (SD; 0.00%)	2.50 (0.02)	2.38 (0.02)
Mean self-destructive (SD; 0.52%)	1.19 (0.01)	1.12 (0.01)
Mean sex offender (SD; 0.00%)	1.83 (0.02)	1.12 (0.01)
Mean anger (SD; 0.25%)	1.87 (0.02)	1.82 (0.02)
Mean developmental (SD; 0.29%)	1.22 (0.01)	1.14 (0.01)

As with most private facilities, this facility is designed to serve only certain inmates. In particular, it is designed to serve medium-security inmates with low- to moderate-needs levels. Thus, information regarding the inmates’ needs was also included in the matching criteria. In particular, information regarding medical, mental health, substance abuse, academic, vocational, anger, developmental, self-destructive tendencies, and sexual offender levels were included to match with comparable public inmates. These needs all ranged from one to five and the distributions can be found in Table 2. Given that this information is not always collected and due to the time frame selected for analysis, some of the needs indicators had substantial missing data (upward of 25%). Thus, for offenders who had been assessed on a prior date, the most recent information was included. With regard to overall missing data, as shown in Table 2, across all indicators, missing data were greatest for education (5.88%), followed by level of supervision (1.98%). Among the needs variables, missing data were less than 1% of the sample.

Table 2.

Balance Statistics.

	Unmatched				Matched
	StateM/%	PrivateM/%	t testt	SB%	StateM/%	PrivateM/%	t testt	SB%
Age	35.84	35.90	0.19	0.5	35.33	35.44	0.31	1.0
Black	16.06%	20.8%	4.58**	12.2	22.78%	20.43%	−1.71^†	−6.1
Other race	3.43%	3.11%	−0.66	−1.8	3.59%	3.03%	−0.94	−3.2
Latino	33.77%	26.14%	−6.07**	−16.7	26.82%	25.14%	−1.15	−3.7
Education—High school	20.85%	20.30%	−0.48	−1.4	20.93%	20.54%	−0.29	−1.0
Gang affiliation	22.95%	20.90%	−1.82^†	−5.0	21.66%	19.87%	−1.32	−4.3
Offense degree	5.49	5.52	1.28	3.5	5.50	5.48	−0.59	−2.0
Level of supervisory inventory	30.41	30.31	−0.51	−1.4	30.49	30.19	−1.42	−4.5
Sentence length (in months)	29.36	5.90	−28.15**	−88.4	5.79	5.94	1.50	0.6
Total number of incarcerations	1.38	1.42	1.71^†	4.6	1.35	1.33	−0.86	−2.7
Needs—Medical	1.80	1.70	−4.62**	−12.8	1.72	1.69	−1.12	−3.7
Needs—Mental health	2.21	2.14	−3.50**	−9.7	2.14	2.15	0.54	1.7
Needs—Substance abuse	3.28	3.42	4.62**	12.7	3.38	3.40	0.67	2.1
Needs—Academic	2.64	2.52	−4.04**	−11.1	2.51	2.51	−0.06	−0.2
Needs—Vocational	2.50	2.38	−4.12**	−11.3	2.35	2.36	0.25	0.8
Needs—Anger	1.87	1.82	−1.83^†	−5.1	1.83	1.81	−0.53	−1.8
Needs—Developmental disability	1.22	1.14	−5.49**	−15.7	1.16	1.14	−1.52	−4.6
Needs—Self-destructive	1.19	1.12	−4.44**	−12.6	1.13	1.12	−0.22	−0.7
Needs—Sex offender	1.83	1.12	−20.69**	−63.7	1.09	1.12	1.82^†	2.6

†

p<.10, *p<.05, **p<.01.

Analysis

Ideally, to measure the impact of placement on recidivism, one would conduct an experiment and randomly assign inmates to one of the facilities. However, in situations where this is impractical or impossible, quasi-experimental methods are useful in that they can approximate the use of random assignment to eliminate selection bias. To that end, this analysis involved matching through propensity scores to make the groups look statistically similar (Apel & Sweeten, 2010). Using logistic regression, the analysis proceeds by estimating a propensity score that is the probability of being assigned to the treatment condition based on observable variables (Rosenbaum & Rubin, 1983). Matching participants based on their propensity scores, those two individuals have relatively the same likelihood of receiving the treatment (assignment to the private facility). This analysis employed a 1 to 1 nearest neighbor matching approach with a caliper of .05.¹ In addition, these individuals were matched with replacement (meaning control groups participants can be included more than once in the model). Although initially unintuitive, the use of nonreplacement has been questioned because it may force a match between individuals with quite different propensity scores and the model may be influenced by the order of the matched treatment cases (Rosenbaum, 1995).² At the end of the matching procedure, once the models are balanced,³ the two groups look statistically similar and the treatment effect of facility can be estimated.

Results

The results of the analyses are reported in two parts. First, information regarding the propensity score analyses is provided for both the measure of overall recidivism and recidivism arising from a new offense. Second, the hazard models depicting the differences in the recidivism curves and time are presented for both measures of recidivism.

Turning first toward the results of the overall measure of reincarceration, Table 2 displays the pre- and postmatching statistics for these analyses. As shown, prior to matching, the treatment and comparison groups were statistically different on most of the covariates. Following matching, none of the differences were statistically different. Furthermore, the standardized bias (SB) was typically less than 3%. Taken together, this analysis suggests that there was substantial bias between the groups prior to matching and that bias was successfully reduced after matching.

Turning to differences in recidivism, Figure 1 depicts the pre- and postmatching distributions of recidivism for inmates who went to each facility. As shown, prior to matching, 46.8% of the treatment and 61.3% of the comparison group recidivated, and the difference is statistically significant (t = −10.33, p < .01). However, after matching on relevant covariates, the difference was reduced to nonsignificance.

Figure 1.

Overall recidivism.

Figure 2 depicts the pre- and postmatching distributions of recidivism when it is operationalized as a new crime. As shown, approximately 14% of those in the private facility committed a new crime compared with 9% prematching and 8% postmatching. There is a statistically significant difference after matching with regard to the frequency of recidivism (t = 4.42, p < .01). Those assigned to the private facility were more likely to commit a new offense.

Figure 2.

Recidivism for a new offense.

Also of interest is whether the length of time to recidivate is different between the two groups. Survival analysis, also called time-to-event analysis, allows for the examination of whether the two groups differ in the length of time to recidivating. Survival analysis, as opposed to linear regression with time being the dependent variable, is preferable with data that are censored because the analytic models take into account that not everyone has experienced the event; however, some may experience the event at a time outside the data parameters. There are three main components to the data for this analysis. First, there is the event, which in this study is recidivism. Inmates are coded for experiencing the event if they have reoffended during the time period in the data. Second, there is the time to event, which is the number of days since the inmates’ release until they recidivate. Finally, for those who have not recidivated before the data period ends, their time variable is coded for the approximate date of when the data were formed (August 1, 2013).⁴ These cases are considered right censored, as it is unknown whether the event would happen given a longer time period.

The final propensity score analysis was used to obtain a sample for survival analysis. This results in a total sample of 2,559 (1,782 in the treatment group and 777 in the comparison group). The time span where inmates recidivate the most is within the first year. By the end of the time period, the survival trend is relatively stable, with a little more than 50% of the sample not recidivating after the study period. The differences between the groups is insignificant (χ² = 1.52, p > .05).

Discussion

Using “return to prison” as the measure of recidivism for the project, we employed a quasi-experimental design (propensity score matching) and survival analyses to evaluate the effects of participation in reentry programming on the subsequent recidivism of program participants relative to an equivalent control group (identified from the Colorado Department of Corrections’ data). An important finding, noted by other scholars examining the success of inmates leaving both private and public prisons, is the high rate of recidivism among all Colorado Department of Corrections (CDOC) inmates (both those in state prison and those who were sent to private prison for part of the sentence). Approximately half of all inmates were found to recidivate and among those who do, the majority are for a technical violation compared with a new crime. This finding is an important one given the large number of inmates who will be paroled via both public prisons and private and public reentry centers (Henrichson & Delaney, 2012). This suggests that a major issue with recidivism may not necessarily be the committing of new offenses, but difficulties adhering to the current postrelease model of supervision. This points to the need for the expansion and modification of services and programming to address the factors leading to technical violations, including continued substance use and abuse and failure to comply with parole conditions. Future research should examine the factors that specifically relate to the reduction of technical violations of offenders post release. One avenue for future research may be to examine community-based programs and facilities and their effectiveness in reducing recidivism after release from private and public prisons. Innovative programs that provide inmates reentering the community resources and supervision to facilitate their success should be examined.

Our findings initially suggest that those from the private reentry center recidivate more frequently and in less time. Yet, proper analysis must take into consideration that assignment to these facilities is not random and, thus, any treatment effect for the type of facility is potentially confounded with factors related to placement. Ideally, to measure the impact of placement on recidivism, one would conduct an experiment and randomly assign inmates to one of the facilities. However, in situations where this is impractical or impossible, quasi-experimental methods are useful, in that, they can approximate the use of random assignment to eliminate selection bias. To that end, this analysis involved matching through propensity scores to make the groups look statistically similar. Our initial analyses included everyone in the sample, and the treatment was defined as anyone who had ever attended the private reentry center, regardless of duration or time. We found that approximately 68% of those who attended the private reentry center recidivated compared with approximately 45% who did not.⁵ This difference is statistically significant. Postmatching the difference is slightly reduced, but still significant. Yet, given that inmates may spend some time at the private reentry center and some time in prison (they are not mutually exclusive sentences) and the private reentry center includes programming aimed at reducing recidivism, it is important to consider the length of stay at the facility (Bales et al., 2005). We, therefore, investigated inmates sent to the private reentry center for more than 75% of their sentence and released after 2010. Our findings indicate that the differences in recidivism between the CDOC and the private reentry center group were not significant, with approximately 40% of each group recidivating. Given that the private reentry center is not designed to accommodate inmates with severe needs (e.g., mental health needs) or that an extensive criminal history can affect the probability of recidivism through collateral consequences, our analyses were further restricted to individuals whose current offense was their only offense and to those inmates who met the eligibility criteria of the private reentry center. Once again, the distribution of recidivism is comparable both before and after matching with no significant differences between the CDOC and the private reentry center groups. Our survival analyses also indicate that the private reentry center inmates are no more likely to recidivate faster than those assigned to state and private prisons. Our findings are consistent with other research comparing offender outcomes among those from private versus state facilities (Bales et al., 2005; Taxman, Rexroat, Shilton, Mericle, & Lerch, 2010). As expected, and with other research on recidivism, the longer inmates are out of the facility, the lower the probability of recidivating.

There are a number of challenges to reentry for inmates who have served time in prison, and it may be that brief periods in reentry centers and facilities may not be able to overcome or fully address these challenges. As Shilton, Rexroat, Mericle, and Taxman (2010) noted,

The model of prison-reentry in a separate facility closer to the offender’s home-supervised release is considered the optimum strategy to reduce the likelihood of recidivism. The reentry period of time is designed to stabilize the offender by beginning the process of obtaining employment, relations with family members, and reentry into the community. Very little is known about what is actually going on during the reentry process or how reentry centers are currently functioning. A pressing need in the field is to ensure that these reentry centers achieve the objectives of fostering better outcomes and performance on supervised release, and assisting the offender in becoming a productive citizen. (p. 2)

It is therefore clear that issues around education, employment, relations with family members, and the full reentry into the community are complicated issues for those who involved in the criminal justice system. First and foremost, among incarcerated individuals, family and intimate relationships are already affected by criminality and likely further weakened by all types of incarceration (Gaes & Kendig, 2002). Although reentry centers likely address a number of inmate needs, convicted felons will have difficulty finding employment and, again, this is likely to be aggravated by all prison and incarceration experiences (McLean & Thompson, 2007). In addition, the services, programming, and resources provided by reentry centers are likely just the first step needed in fully preparing inmates for the return to the community. This includes issues related to continued abstinence from alcohol and drugs, mental health needs, housing, employment, and education (Gaes & Kendig, 2002). Inmates are also likely to experience challenges related to housing, not only assistance in finding affordable rental housing but they also experience the barriers and stigma of being an ex-offender. Although private and public facilities have education participation requirements for their inmates, ex-offenders’ lower rates of literacy likely hinder the transition into the community. If we are to prepare inmates for reentry, facilities will need to address both the educational and occupational needs of offenders in an ever-competitive job market (Gaes & Kendig, 2002). McLean and Thompson (2007) found that many inmates believe that having a job is an important factor in their path to reentry, remaining out of criminality, and staying out of prison. Yet, many do not have employment lined up after their release, few receive high-quality employment-related training in prison, and only a small number participated in work-release employment while incarcerated that would otherwise have a positive impact on the likelihood of employment post release (McLean & Thompson, 2007).

Finally, future research should assess the specific characteristics of reentry centers that positively affect particular barriers post release that offenders face to reduce recidivism. As Wright (2010) noted, “Private prisons present an opportunity to shift the guiding paradigm while still retaining some of the driving forces in modern corrections” (p. 79). Whereas studies of recidivism are important, future research should endeavor to use private prisons as a way to encourage rehabilitative programming and examine what works with regard to corrections.

Limitations

There are a few limitations that should be discussed. Although this study was the first to examine the differences in facilities in Colorado, because the data were provided by the Colorado Department of Corrections, our measure of recidivism is limited to those who are recaptured into the Colorado system. Furthermore, we operationalized recidivism to be reincarceration, which corresponds to the Colorado Department of Corrections’ definition of recidivism. However, it should be noted that not all recidivists are reincarcerated. Therefore, this study underestimates the true recidivism rate. And, although we disaggregated recidivism to examine differences between the facilities in new offenses, it was beyond the scope of this article to look at the severity of the recidivism offense or the specific type of offense (e.g., drug vs. violent). However, it is possible that these facilities differ with regard to the effectiveness at reducing certain types of recidivism, and those differences may be the result of the different programs offered at the facilities.

This analysis was restricted to men, but previous analyses point to positive results for private facilities that serve female offenders (Farabee & Knight, 2002). This may be the result of gender-specific programming offered in these facilities. Future research should continue to explore gender differences in the effectiveness of private prisons.

Also with regard to programming, as Bales and colleagues (2005) noted in their analysis, research should endeavor to examine who participated and completed the programs available to inmates. Although this study provided an analysis of a facility that is state mandated to provide prerelease programming, the extent to which inmates meaningfully participated and completed the programming is unknown.

Conclusion

Previous research on the success of privately run correctional facilities in reducing offender recidivism is inconclusive (Bales et al., 2005; Bayer & Pozen, 2005; Spivak & Sharp, 2008). Addressing the issue that some studies have lacked rigorous statistical methods, the current study applied a quasi-experimental design as well as propensity score matching to allow for a causal analysis of recidivism reduction postincarceration in either private or public correctional facilities. Those released from the private facility are more likely to return to prison for a new offense; however, the overall rate of recidivism and the time to return to prison were comparable between the facilities. Overall, this research suggests that these facilities are comparable. Future research should focus on the identification, application, and measurement of the components that reduce recidivism and facilitate successful reentry for various types of offenders.

Footnotes

Acknowledgements

The authors thank the Colorado Department of Corrections and Community Education Centers for their assistance and invaluable feedback on this project.

Authors’ Note

The authors are entirely responsible for the research and results reported in this article, and their position or opinions do not necessarily represent those of the University of Colorado, Colorado Springs, the University of South Florida, the University of Louisiana at Lafayette, the Colorado Department of Corrections, or Community Education Centers.

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) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The project was supported by contract awarded to Dr. Catherine Kaukinen by Community Education Centers.

Notes

Author Biographies

Rachael A. Powers, PhD, is an associate professor in the Department of Criminology, University of South Florida. Her research interests focus on violent victimization, public perceptions of crime, and gender.

Catherine Kaukinen, PhD, is professor and chair in the Department of Criminal Justice, University of Central Florida. Her research interests include intimate partner violence, risk and protective factors for violent victimization, and campus-based violence evaluation, prevention, and intervention.

Michelle Jeanis, PhD, is an assistant professor in the Department of Criminal Justice, University of Louisiana at Lafayette. Her research interests include missing persons, juvenile runaway behavior, and the representation of crime in the media.

References

Apel

Sweeten

(2010). Propensity score matching in criminology and criminal justice. In Piquero

A. R.

Weisburd

(Eds.), The handbook of quantitative criminology (pp. 543-562). New York, NY: Springer.

Bales

W. D.

Bedard

L. E.

Quinn

S. T.

Ensley

D. T.

Holley

G. P.

(2005). Recidivism of public and private state prison inmates in Florida. Criminology & Public Policy, 4(1), 57-82.

Bayer

Pozen

D. E.

(2005). The effectiveness of juvenile correctional facilities: Public versus private management. Journal of Law & Economics, 48, 549-589.

Becker

Ichino

(2002). Estimation of average treatment effects based on propensity scores. The Stata Journal, 2, 358-377.

Chang

T. F.

Thompkins

D. E.

(2002). Corporations go to prisons: The expansion of corporate power in the correctional industry. Labor Studies Journal, 27(1), 45-69.

Community Education Centers. (2015). The Cheyenne Mountain Re-Entry Center History and Services. Retrieved from http://www.cecintl.com/facilities_rr_co_004.html

Duwe

Clark

(2013). The effects of private prison confinement on offender recidivism: Evidence from Minnesota. Criminal Justice Review, 38, 375-394.

Farabee

Knight

(2002). A comparison of public and private prisons in Florida: During- and post-prison performance indicators. Los Angeles, CA: Query Research.

French

M. T.

Fang

Fretz

(2010). Economic evaluation of a prerelease substance abuse treatment program for repeat criminal offenders. Journal of Substance Abuse Treatment, 38, 31-41.

10.

Fretz

Heilbrun

K. S.

Brown

(2005). Outcome research as an integral component of performance-based offender treatment. Corrections Compendium, 29(4), 1-4.

11.

Gaes

G. G.

Kendig

(2002, January). The skill sets and health care needs of released offenders. Paper presented at the From Prison to Home Conference. Available from http://www.urban.org

12.

Geiger

(2017). U.S. private prison population has declined in recent years. Retrieved from http://www.pewresearch.org/fact-tank/2017/04/11/u-s-private-prison-population-has-declined-in-recent-years/

13.

Greene

(2000, May). Prison privatization: Recent developments in the United States. Paper presented at the International Conference on Penal Abolition. Toronto, Canada. Available from http://www.iprt.ie/files/international/greene_icopa.pdf

14.

Heilbrun

DeMatteo

Fretz

Erickson

Gerardi

Halper

(2008). Criminal recidivism of female offenders: The importance of structured, community-based aftercare. Corrections Compendium, 33(2), 1-2.

15.

Henrichson

Delaney

(2012). The price of prisons: What incarceration costs taxpayers. Federal Sentencing Reporter, 25, 68-80.

16.

Kaeble

Glaze

(2016). Correctional populations in the United States, 2015. Washington, DC: Bureau of Justice Statistics.

17.

Kleis

K. M.

(2010). Facilitating failure: Parole, reentry, and obstacles to success. Dialectical Anthropology, 34, 525-531.

18.

Lanza-Kaduce

Maggard

(2001, October). The long term recidivism of public and private prisoners. Paper presented at the National Conference of the Bureau of Justice Statistics and Justice Research and Statistics Association, New Orleans, LA.

19.

Lanza-Kaduce

Parker

K. F.

Thomas

C. W.

(1999). A comparative recidivism analysis of releasees from private and public prisons. Crime & Delinquency, 45, 28-47.

20.

Lee

(2012). Private prisons and community corrections. In Price

B. E.

Morris

J. C.

(Eds.), Prison privatization: The many facets of a controversial industry (pp. 223-240). Santa Barbara, CA: ABC-CLIO.

21.

Lundahl

B. W.

Kunz

Brownell

Harris

Van Vleet

(2009). Prison privatization: A meta-analysis of cost and quality of confinement indicators. Research on Social Work Practice, 19, 383-394.

22.

McLean

R. L.

Thompson

M. D.

(2007). Repaying debts. New York: Council of State Governments Justice Center.

23.

Nagin

D. S.

Cullen

F. T.

Jonson

C. L.

(2009). Imprisonment and reoffending. Crime and Justice, 38, 115-200.

24.

Rosenbaum

P. R.

(1995). Observational studies. New York, NY: Springer.

25.

Rosenbaum

P. R.

Rubin

D. B.

(1983). The central role of the propensity score in observational studies for causal effects. Biometrika, 70, 41-55.

26.

Shilton

Rexroat

Mericle

Taxman

F.S.

(2010). What Works in Residential Reentry Centers. Report 1: What is the impact of “Performance Contracting” on Offender Supervision Services? (Cooperative Agreement No.# 07EPI16GJT3). Fairfax, VA: Federal Bureau of Prisons and the National Institute of Corrections. Retrieved from https://www.gmuace.org/documents/publications/2010/Monograph%201.pdf

27.

Spivak

A. L.

Sharp

S. F.

(2008). Inmate recidivism as a measure of private prison performance. Crime & Delinquency, 54, 482-508.

28.

Taxman

F.S.

Rexroat

Shilton

Mericle

Lerch

(2010). What works in residential reentry centers. Executive Overview. Advancing Correctional Excellence! Fairfax, VA: George Mason University. Available from https://www.gmuace.org/documents/publications/2010/Overview%20Document.pdf

29.

Wells

O’Keefe

Allen

(2013). Annual report concerning the status of private contract prisons. Retrieved from http://www.doc.state.co.us/sites/default/files/opa/PPMU%20Annual%202013.pdf

30.

Wojtowicz

J. P.

Liu

(2006). Seven-year analysis of walk-away rates of the New Jersey halfway house program. Corrections Compendium, 31(6), 6-23.

31.

Wright

K. A.

(2010). Strange bedfellows? Reaffirming rehabilitation and prison privatization. Journal of Offender Rehabilitation, 49, 74-90.