Prison Privatization: A Meta-Analysis

Running Head: PRISON PRIVATIZATION: META-ANALYSIS

Prison Privatization:
A Meta-Analysis of Cost Effectiveness and Quality of Confinement Indicators

Brad Lundahl, PhD
Chelsea Kunz
Cyndi Brownell
Norma Harris, PhD
Russ Van Vleet, MSW

Utah Criminal Justice Center
College of Social Work
University of Utah

April 26, 2007

1

Prison Privatization: A Meta-Analysis

2

Abstract
Opinions vary about privatizing prisons. In an effort to provide an empirical base from
which decisions about privatization might be made, we conducted a meta-analysis of reports on
head-to-head comparisons between an identifiable privately managed and publicly managed
prison(s). Our search identified 12 studies. Indicators of cost effectiveness and confinement
quality were assessed. Results suggest privately managed prisons provide no clear benefit or
detriment. Cost savings from privatizing prisons are not guaranteed and appear minimal. Quality
of confinement is similar across privately and publicly managed systems, with publicly managed
prisons delivering slightly better skills training and having slightly fewer inmate grievances.

Prison Privatization: A Meta-Analysis

3

Introduction
Government leaders seek to establish and manage programs that best meet citizens’
needs. Therefore, it is sensible that government leaders investigate whether privately managed
prison systems represent an improvement over publicly managed prisons. Investigating the
potential benefits of privatization is timely because imprisonment rates across the United States
are increasing which pressures budgets and prison facilities (Camp, 1997). Privatization of
services that are typically managed by public entities is seen as a means to deliver higher quality
services at reduced costs through encouragement of competition. Privatizing prisons may be one
mechanism to reduce pressures currently facing prison systems and has been used in 31 states
within the U.S. (Blakely & Bumphus, 2004).
Privatizing prisons is, however, controversial. Proponents argue privatization will lead to
innovative, cost effective, and high-quality programs. Dissidents argue privatization violates
important ethical principles, carries considerable risk, and does not improve upon publicly
managed systems. The intensity of the debate suggests that a clear, obvious choice about
privatizing prisons does not exist. In addition, the stakes of prison privatization are high because
multiple stakeholders will be influenced by the decision. At a broad level, public safety is a
concern as is responsible spending of tax revenue. At an individual level, inmates and prison
management staff may be positively or negatively influenced by such decisions.
Privatization of prisons has a history of over 22 years (Lanza-Kaduce, Parker, & Thomas,
1999). Thus, policy makers currently investigating whether prison privatization best meets its
citizens need have the benefit of learning from the pioneering efforts of others. In this study we
compared publicly managed and privately managed prisons on two outcomes, cost effectiveness

Prison Privatization: A Meta-Analysis

4

and delivery of quality services, through a quantitative literature review known as a metaanalysis. Our literature search and meta-analysis provide two benefits. First, we report on the
conclusions of previous literature reviews that we identified in our literature search. Second, and
more important, we present what we believe to be the most extensive quantitative literature
review of the very best studies investigating prison privatization – those that compared privately
managed and publicly managed prisons. Prior to presenting how we conducted our study and the
findings, we offer a brief introduction and history of prison privatization and the controversy it
has engendered.
What is prison privatization?
Prison privatization involves a business contracting with a branch of the government to
operate a prison facility. Many of the large businesses operating prisons today are publicly traded
companies (Chang & Thompkins, 2002). Private companies generally charge the government a
daily rate per inmate to cover investment, operating costs, and profit. Under this rate, private
companies supply many or most of the services needed to operate a prison system, including
guards, staff, food, program costs, medical care (partial), and other services. Private companies
may also build new facilities without direct tax expenditures or public bonds (Lanza-Kaduce et
al., 1999).
The trend to privatize prisons began in earnest in 1984 when Hamilton County,
Tennessee and Bay County, Florida entered into contracts with the private sector (Anderson,
2000). By the end of 1988, 20 federal, state, and local level privately operated detention facilities
were in operation in nine states. In the 1980s the size of the average facility under private
contract swelled from the initial experimental 80-bed detention facilities to 500 and 600-bed
facilities. The type and classification of the privately operated facilities changed from

Prison Privatization: A Meta-Analysis

5

predominantly low security to many facilities with medium and maximum-security levels and
several facilities which housed inmates at all levels of security (Calvert, 1999). As of 1990, there
were 15,000 private prison beds in existence; by 1996 this number had increased by 435%
(Blakely & Bumphus, 2004).
By the year 2000, privately operated prisons held more than 101,000 federal, state and
local jail inmates. At the time, this represented 11% of all federal inmates, 6% of all state
inmates, and 2% of local jail inmates. The majority of private prison operations (63%) were
concentrated in the southern states. Texas has the highest private prison capacity with 30,000
beds and the largest number of inmates (14,000) housed in private facilities (Chang &
Thompkins, 2002). As of 2004, 31 states use 158 private correctional facilities designed to house
122,871 inmates (Blakely & Bumphus, 2004).
Controversy surrounding prison privatization
Changing to a privately managed prison system is controversial and seems to have been
polarizing in many instances. A healthy percentage of the articles we examined for this metaanalysis, not all of which were peer-reviewed, took clear positions about the value and/or risks of
privatization while seeming to “sell” their position. As an example, there are situations where
opposite conclusions about the same data are reached (OPPAGA, 1997).
We highlight the arguments advanced by proponents and critics of prison privatization in
Table 1. Two limitations need to be considered in reviewing this table. First, these
summarizations were gathered as we read articles to determine whether they meet our inclusion
criteria and do not reflect a systematic or scientific compilation of the varied arguments or
positions on prison privatization. Thus, we may have missed arguments or concerns advanced by
advocates and critics. Second, in our effort to highlight the arguments we do not provide the

Prison Privatization: A Meta-Analysis

6

detail needed to fully understand some of the positions. Despite these limitations, we believe
Table 1 provides a summary of the positions, arguments, and concerns typically advanced by
proponents and critics of prison privatization.

[INSERT TABLE 1 ABOUT HERE]

Previous Literature Reviews
In our literature search, we found two systematic reviews and one meta-analysis on
prison privatization. Two of the studies (Perrone & Pratt, 2003; Pratt & Maahs, 1999) published
in peer-reviewed sources. The third (Segal & Moore, 2002) was published, without obvious blind
peer review, by the Reason Public Policy Institute which appears to be a think tank that supports
privatization.
Pratt and Maahs (1999) conducted a meta-analysis investigating whether privately
managed prisons are more cost effective than public prisons. They identified 24 studies that
examined cost-effectiveness from 33 evaluations of private and public prisons. In their metaanalysis the prisons compared were not matched and did not involve head-to-head comparisons.
Pratt and Maahs found that, at first glance, privately managed prisons appeared more cost
effective than publicly managed sites, on average $2.45 U.S. dollars less per prisoner per day.
However, these authors also found that the best predictors of cost were number of inmates served
(r = -.345), age of the physical facility (r = .511), and security level (r = .347). After considering
these factors, locus of management (i.e., private or public) did not significantly predict cost
effectiveness (r < .05). In their concluding remarks the authors state “Although specific
privatization policy alternatives may result in modest cost savings… relinquishing the

Prison Privatization: A Meta-Analysis

7

responsibility of managing prisons to the private sphere is unlikely to alleviate much of the
financial burden on state correctional budgets” (pp. 367-368).
Through a non-quantitative review Perrone and Pratt (2003) investigated whether
privately managed prisons, compared to publicly managed prisons, would perform better with
regard to two issues: quality of confinement and cost effectiveness. These authors found nine
studies that assessed the relative quality of private versus public prisons. That is, in each of their
nine studies an identified privately managed prison (or small group of privately managed
prisons) was compared to an identified public prison (or a small group of publicly managed
prisons). Although all of the studies were matched with regard to security level, in four of the
nine studies the private prison facility was newer than that of the public facility and none of the
comparisons involved a public facility that was newer than the compared private prison. With
regard to capacity, 3 of the studies involved public facilities with larger capacity compared to 1
study where the private facility was larger. Drawing confident inferences about the impact of
prison privatization is difficult when variables other than management could account for such
differences in outcomes. Despite the fact that prison sites were not well matched, Perrone and
Pratt compared private and public prisons on the following indicators of quality: condition (e.g.,
clean), management (e.g., staff stress and burnout), inmate activity (e.g., educational and
vocational training), safety (e.g., assaults on inmates, staff), security (e.g., escapes), order (e.g.,
disturbances), and care (e.g., medical attention). From their analyses, the authors noted
“comparisons of the quality of confinement between public and private prisons is inconclusive”
(p. 309). With regard to cost effectiveness, Perrone and Pratt found that the daily median per
diem for private prisons was approximately $3.40 U.S. dollars cheaper than publicly managed
facilities. However, the authors also note that only 2 of the 9 comparisons offered firm

Prison Privatization: A Meta-Analysis

8

conclusions with regard to cost effectiveness. Of these, one comparison favored public
management and the other private management. In their concluding section, the authors indicated
that neither privately or publicly managed prisons can boast a clear advantage in cost
effectiveness or quality. We note that the conclusions Perrone and Pratt (2003) made were
technically the result of a systematic review, not a meta-analysis. That is, with the exception of
providing the average cost savings the authors did not extract and combine effect sizes from the
9 identified studies.
In the third systematic review, Segal and Moore (2002) reviewed numerous studies on
privatization and reported the overall findings from each study. The authors did not report
parameters for which studies were included in their analyses. With regard to cost effectiveness,
Segal and Moore noted that in 22 of the 28 studies they reviewed, privately managed prisons
were more cost effective an average savings of 12.38 % (Standard Deviation = 8.53%).These
authors also reported that that 11 of the 16 indicators of quality of confinement favored privately
managed prisons. Segal and Moore cite their findings as a strong rationale for privatization. It is
our opinion, however, that Segal and Moore’s analyses are suspect for several reasons. First,
these authors report no instances where a publicly managed prison was more cost effectiveness
than a privately managed prison, which stands in contrast to the other reviews and our own
analyses (see below). Second, Segal and Moore provided a relatively less sophisticated analysis
compared to the other reviews; for each study reviewed they provide, on average, a one
paragraph summary giving data but no indication on the internal validity of the reviewed studies.
This review style leaves open questions of potential confounds to the inferences made. Third, the
authors’ affiliation is with a group that is clearly aligned with promoting privatization.

Prison Privatization: A Meta-Analysis

9

Conclusions from the Perrone and Pratt (2003), Pratt and Maahs (1999), and Segal and
Moore (2002) studies diverge significantly. The two studies authored by Pratt and colleagues
concluded that the evidence about the proposed benefit of privatization is inconclusive whereas
Segal and Moore reported “there is clear and significant evidence that private prisons actually
improve quality” and “private prisons are providing quality services, while remaining costefficient and providing significant cost savings” (p. 14). Our opinion, which we believe is
scientifically defensible, is that Pratt and colleagues engaged in a more sophisticated approach to
reviewing studies by considering issues related to internal validity and, therefore, provide a more
accurate perspective on prison privatization.
Our Study
To date it appears that only one meta-analytic review on prison privatization has been
published: the Pratt and Maahs (1999) study on cost-effectiveness which used a regression
model. To help fill this void, we conducted a meta-analysis looking at cost effectiveness and
quality using a group-differences model. Our study offers two advantages to the other published
reviews. First, we conducted a meta-analysis on indicators of confinement quality in addition to
cost effectiveness. Second, our investigation relied upon “gold standard” reports – those that
compared identified privatized systems to identified public systems and presented data which
could translate into effect size statistics.
Meta-analysis is a research design that extracts and summarizes the quantitative findings
from published studies or completed reports. In this study, the numeric findings from 12
published reports on the relative effectiveness of privately operated prisons, compared to
publicly operated prisons, were summarized. Meta-analysis is a powerful research design,
compared to simple reviews, because it (a) combines a body of findings to provide an overall

Prison Privatization: A Meta-Analysis

10

estimate, (b) uses objective procedures to produce numeric indicators on the strength of an
intervention or relationship, (c) avoids considerable subjectivity by detailing precise rules and
steps taken, and (d) allows for replication because findings are based on quantitative data not
subjective interpretations of the data.
Method
In general, a meta-analytic review follows the following 8 steps: (1) developing a
research question, (2), determining criteria for study inclusion and exclusion, (3) developing a
strategy to identify the relevant literature, (4) securing the relevant literature, (5) coding studies
and extracting effect sizes, (6) combining effect sizes and investigating moderator effects
(results), (7) interpreting the results, and (8) presenting the findings. The description of our study
is organized around these steps.
Research Question
We compared privately managed and publicly managed prisons on two domains: cost
effectiveness and quality of care. Although we approached this study from an objective position,
testing research questions requires the establishment of a research question and null hypothesis.2
For the sake of convenience we structured our questions in an orientation that favored privately
managed prisons; that is, we temporarily adopted the assumption that privatization would
produce more desirable outcomes. Thus, a positively valenced (+) effect size represented an
advantage for privately managed systems and a negatively (-) valenced effect size represented an
advantage for publicly managed systems. These valences could have been reversed without
changing the interpretations and, therefore, have no influence on the outcome.
Two questions focused on cost effectiveness: Are privately managed prisons more cost
effective than publicly managed prisons? How much more cost effective are privately managed

Prison Privatization: A Meta-Analysis

11

prisons compared to publicly managed prisons? The measure of cost effectiveness was the
average savings per-prisoner per day. The next two questions centered on quality of services
rendered. Are privately managed prisons delivering higher quality services compared to publicly
managed prisons? How much better are publicly managed prisons at delivering quality services
compared to publicly managed facilities?
Inclusion and Exclusion Criteria
Determining which studies will be included in a meta-analysis strongly influences
confidence levels in the findings. A typical phrase used in research is “garbage in, garbage out.”
If low quality studies are included, the findings are suspect because there is a high chance that
alternative explanations may provide a better explanation. Because the stakes are high in making
a decision to privatize, we only included high quality studies. Our assessment of the literature on
privatization is that high quality studies involve those that directly compared a specific,
identifiable private prison(s) with a closely matched, identifiable public prison(s). While
randomization or exact matching are clearly superior designs and produce superior outcomes,
pragmatic and ethical considerations all but prohibit such designs in this area. A lower quality
study would be one that reports group averages from privately and publicly managed prisons that
cannot be identified. For example, Blakely and Bumphus (2003) presented national findings.
While this type of study is very valuable for certain questions, it does not permit a rigorous
comparison of privately and publicly managed prisons.
For inclusion in our meta-analysis, studies needed to meet six criteria. First, studies had
to report on a comparison study of a privately managed and publicly managed prison(s) that
could be identified. To be identified the study simply had to name the facilities and their
geographic location. Studies that reported findings from an evaluation of only a private or a

Prison Privatization: A Meta-Analysis

12

public prison were excluded. Second, studies had to provide statistical data from their analyses
that could be transformed into an effect size or percentage of cost effectiveness. Third, studies
had to report on their own analyses from primary or secondary data. That is, we excluded studies
who simply commented on another report or study (there are many such studies). Fourth, we
only accepted studies targeting prisons for adults. Fifth, we limited our study to reports or
publications that could be found through electronic databases and reference sections of such
reports. Sixth, we excluded articles that reported hypothetical data or projections rather than
actual findings. Determining whether a study fit these six criteria was an iterative process and is
described below.
Identifying the Retrieving Relevant Literature
We utilized three search strategies: electronic data bases, searching reference lists, and
consultation with an expert in adult corrections.3 Most of our efforts were directed to the
electronic data bases and searching references of key studies (i.e., the previously mentioned
reviews). In an effort to not miss key studies from the electronic database search, we used the
following broad search terms: prison, privatization, privatisation (British or Canadian spelling),
correction, and jail connected by the term “or.” The following data bases were searched:
Criminal Justice Abstracts, ERIC, PsychInfo, CSA Social Services Abstracts, CSA Sociological
Abstracts, Recent References Related to the Social Sciences/Humanities, Academic Search
Premier, PsycARTICLES, Psychology and Behavioral Sciences Collection, and Family &
Society Studies Worldwide. We limited our search to reports published between 1980 and
October of 2006 that were written in English. This strategy yielded approximately 1,110 articles.
The abstracts of these articles were then read with two guiding screening criteria for
inclusion, that they: (a) provided an investigation into prison privatization, and (b) provided

Prison Privatization: A Meta-Analysis

13

statistical data rather than theory or opinion. This strategy identified 259 articles that were then
retrieved using interlibrary loan and other retrieving strategies. Of these, 44 were never
considered in the final analyses because they were either duplicates of already retrieved articles,
only accessible through microfiche or other media that was inaccessible, or could not be found in
any libraries or public journals. This left 215 articles that were then screened based on phase two
of the article selection process. At this stage we also identified 53 studies by reviewing the
reference sections of the reviews previously mentioned and we reviewed 48 studies provided by
a local expert.3 Thus, 316 articles were secured and reviewed based on the inclusion criteria. Of
these, only 12 articles met all criteria.
Coding Studies
Studies varied on several dimensions and needed to be organized to complete analyses.
We limited coding to variables believed to have a relationship to the outcomes of interest or to
variables that may influence the internal validity of a study. We certainly did not code the
universe of possible variables; prison management systems are very complex and not all studies
follow a standardized approach to measuring inputs and outputs.
In an effort to provide a profile of the prisons being compared we coded characteristics of
the prisons being compared. The results can be found in Table 2. All studies that met the
inclusion criteria were independently double coded by a combination of the first 3 authors.
Disagreements were resolved through reviewing the articles and discussion.

[INSERT TABLE 2 ABOUT HERE]

Prison Privatization: A Meta-Analysis

14

Information about how data was collected is important in that it provides a possible
indicator of objectivity. Five categories of data source were coded: official records (OR) which
included financial audits and state or federal reports; agency records (AR) which included prison
records or year-end reports or surveys; staff interview (SI) which involved data based on reports
from prison staff members; inmate interviews (II) which involved comments or impressions from
inmates; and lastly researcher observations (RO) which included impressions from the report
authors or researchers assigned to compare the prisons.
The number of prisons compared being compared was also coded. Some studies
compared one publicly managed prison to one privately managed prison (e.g., Brown, 1994).
Other studies compared several specific prisons. For example, Archambeault (1996) compared 1
government or publicly managed prison with 2 privately managed prisons. Security level details
the classification level of the inmates incarcerated. As can be seen in Table 2, each study
compared prisons that were matched on security level. The levels we coded included: minimum,
medium, mixture (combination of minimum and medium security), and closed/high-security.
Some of the prisons only housed males or females, while others housed both genders;
thus we coded inmate gender. As prison age has been shown to predict costs (Pratt & Maahs,
1999), we coded such information when it was available. Next, we coded facility type which
provides information about whether the prison belonged to the federal, state, or county system.
The number of inmates housed in a prison, prison capacity, was also coded as it has been
shown to be related to effectiveness indicators (Pratt & Maahs, 1999). We also coded whether
the comparison of prisons was contemporaneous; that is we examined whether the time frame of
the comparison was equal across the private and public systems. As can be seen in Table 2, all

Prison Privatization: A Meta-Analysis

15

comparisons were contemporaneous. Lastly, we examined whether the facility was constructed
under private management or public management.
Dependent Variables
Two broad-based outcomes were assessed: cost effectiveness and quality of confinement.
Understandably, the studies included in this meta-analysis investigated different outcomes of
interest and used dissimilar instruments to measure such outcomes. Such variability is considered
to be both a strength and weakness of meta-analyses. As a strength, the diversity of outcomes
and measurement tools provides some protection against relying on a sole indicator. As a
weakness, not all instruments measure exactly the same construct and some outcome classes
include varied indicators – factors which limit precision.
Cost effectiveness. Formulas to derive cost effectiveness, generally a per diem amount per
inmate, varied with regard to what factors were considered. Our first preference for calculating
percentage of savings was to utilize raw data. This was done by determining the difference
between a publicly and privately operated prison and then dividing this value by the lower of the
two per diem rates. When multiple indicators of cost effectiveness were provided, an average
was taken based on the expectation that this would provide the most stable estimate of costs
(Howell, 1997). If raw data was not available, we simply reported what the authors listed. Again,
if multiple indicators were presented, an average was taken. We note that some studies included
the costs associated with government monitoring and/or facility construction expenses while
others did not (see Table 3).
Quality of confinement. In order to organize and consolidate the many indicators of
confinement quality, outcome classes were created. Logan (1992) identified eight dimensions of
confinement quality: security, safety, order, care, activity, justice, conditions, and management.

Prison Privatization: A Meta-Analysis

16

We slightly adapted this model to more easily accommodate the indicators we encountered. This
was done by consensus. Four of the five authors jointly reviewed and decided which outcome
class best captured each indicator for which an effect size was calculated. This process resulted
in nine outcome classes which are described below.
Public safety is made up of indicators of prisoner escape rates and visitors being harmed.
Prison safety is made up of indicators of harm to prison staff or inmates arising from violence or
disciplinary action. Order was a broad category that involved indicators such as compliance with
prison rules and regulations, drug use within the prison, protection from communicable diseases,
exposure to medical risks, and suicide prevention. Heath care is made up of indicators reflecting
delivery of medical, dental, mental health, and/or drug and alcohol counseling. Skills training
reflects delivery or availability of programs designed to equip inmates with useful life skills,
such as general education classes, work opportunities, job skill development, and/or career
planning. A miscellaneous inmate benefits category was developed to capture measurements of
some combination of health care and skills training that could not be separated because of their
data presentation or other generic benefits such as physical fitness time, leisure time, or visitation
opportunities. Several of the studies reported on inmate grievances comprised of complaints to
the ARPS or civil suits. A high number of grievances is believed to reflect inmate dissatisfaction
with prison conditions. Facility conditions reflects indicators of prison cleanliness, nutrition
being offered, and satisfaction. Lastly, some studies sampled prison staff for their reflections of
employee morale, job satisfaction, or looked at employee turnover rates; such variables were
collapsed into the employee work climate outcome.
Studies often reported several indicators that fell within a single outcome class. Consider,
for example, the outcome class of safety. Archambeault et al. (1996) report highly detailed

Prison Privatization: A Meta-Analysis

17

information such as inmate assault on other inmates resulting in (a) serious injury [d = .071], (b)
no injury [d = -.119], or (c) some injury [d = -.094]. In this case, effect sizes [reported above as
d] were averaged, d = -.047, and this average was advanced for the “safety” grouping. This
practice follows best practices in meta-analyses (Lipsey & Wilson, 2001).
Effect Size Calculations
Cohen’s d was used as the effect size (Lipsey & Wilson, 2001). An effect size is a
statistic that represents impact strength or magnitude. As a guide, a d in the 0.20 range is
considered small, though significant, a d in the 0.50 is considered to be moderate in magnitude,
and a d in the range of 0.80 is considered to be large (Cohen, 1988). Effect sizes can be
calculated from various types of data, including means, proportions, frequencies, and p-values
(Cooper & Hedges, 1994; Lipsey & Wilson, 2001). Effect size computations and summary
analyses were done using DSTAT, a meta-analytic software program (Johnson, 1993).
One methodological issues arose in calculating effect sizes: several studies failed to
clearly present full information needed to calculate effect sizes with confidence. At times
assumptions were made to calculate effect sizes, which slightly undermines the confidence we
have in this report. For example, the number of participants involved in a given comparison often
had to be estimated from other information. In an effort to be transparent, we note such instances
in Table 3 and will make available a detailed account of how effect sizes were calculated. When
assumptions were made to calculate an effect size (see Table 3), a conservative approach was
followed. Such decisions are common in meta-analyses and we followed authoritative
recommendations (Cooper & Hedges, 1994).

Prison Privatization: A Meta-Analysis

18

One of the values in the public safety outcome class was an outlier (Bowery, 1996). To
control the undue leverage it would have on the average, this value was Windorized by adjusting
it to the next highest value.
Results
Eight of the 12 studies that met inclusion criteria provided information on cost
effectiveness (see Table 3). Half of these 8 revealed that privately managed prisons outperformed
publicly managed prisons, with a range of cost effectiveness from 4.6% to 15.2%. Of the
remaining 4 studies, 2 showed that publicly managed prisons were more cost effective than their
privately managed counterparts (10.0% and 14.2%). The remaining 2 studies revealed a
statistical tie: that is neither system outperformed the other. Thus, 50% of the time privately
managed prisons showed a financial advantage over publicly managed prisons, while publicly
managed prisons showed an advantage only 25% of the time. The average cost savings across all
8 studies was 2.2% (SD = 11.5%) favoring privately managed prisons.

[INSERT TABLE 3 ABOUT HERE]

Several patterns can be detected from examining the results from indicators of quality of
confinement. First, no more than half of the studies contributed effect sizes to any one construct.
This situation tends to undermine confidence in inference making. Second, most effect sizes in
Table 3 are very near to zero (“0.00”). This suggests that, in general, there is not much difference
between privately and publicly managed prison systems. Third, effect size valences are not
predominately positive or negative. With these patterns in mind, we briefly discuss each outcome
class.

Prison Privatization: A Meta-Analysis

19

Publicly managed prisons tended to perform better with regard to public safety. Five of
the 6 studies that presented data in this area favored public prisons, although the effect sizes are
small. Similarly, the average effect size across the six studies was small, -.04, but may represent
concern when considering the scope of the prison system across the nation and the magnitude of
risks associated with public safety issues. With regard to prison safety within a prison’s walls, 3
of the 5 studies revealed privately managed prisons were safer; the other 2 favored publicly
managed systems. The overall or average effect, however was nil. Of the 5 studies which
provided indicators related to prison order, 4 showed slight advantages to privately managed
prisons. The 1 publicly managed prison that outperformed its private counterpart, revealed a
rather large advantage (i.e., d = -.12), which brought the overall effect to nil.
Indicators of health care delivery suggest no real advantage or disadvantage from private
management. By contrast, each of the 3 publicly managed systems who reported data on skills
training outperformed privately managed prisons. Of these, 2 showed sizeable effect sizes while
1 was near zero, resulting in an overall effect size of -.10. There was no observed advantage for
miscellaneous benefits for publicly or privately managed systems. However, a slight advantage
was noted in the grievance category which favored publicly managed prisons. Here, 3 of the 4
studies reporting on this dimension favored public managed systems with an overall effect size
of -.07. Only 1 study reported on facility conditions, which favored privately managed prisons –
though the effect size was small, d = .02. The last category, employee work conditions, showed a
slight advantage to privately managed prisons with 3 of the 4 studies showing positive effect
sizes. Again, the overall advantage was very small, .03.
In addition to rather small effect sizes, the distribution of positive and negative valences
was balanced. There were 45 outcome indicators overall. Of these, 21 (47%) favored privately

Prison Privatization: A Meta-Analysis

20

managed prisons, 20 (44%) favored publicly managed prisons, and 4 (9%) favored neither. Of
the 10 summaries of these 45 values 50% favored publicly managed prisons, 30% favored
privately managed systems, and 20% showed no difference.
Discussion
Our conclusion is that prison privatization provides neither a clear advantage nor
disadvantage compared to publicly managed prisons. Cost savings from privatization are not
guaranteed and quality of services is not improved. Across the board effect sizes were small, so
small that the value of moving to a privately managed system is questionable. An empirical
argument against privatization may be made based on the finding that publicly managed prisons
tended to provide better skills training programs and seemed to generate fewer complaints or
grievances. However, improved training programs may not result in benefits to inmates or
society, and the number of grievances filed may not accurately reflect the quality of life in
prison.
Understanding the interpretation of small effect sizes based on prison management is
difficult given the complex issues and goals involved in the criminal justice system. The largest
average effect size was found for skills training where publicly managed prisons outperformed
privately managed prisons. The average effect size (d = .10) suggests that publicly managed
prisons are 4% better at skills training (Lipsey & Wilson, 2001, p. 153). An example may help in
interpreting the impact of this level. If a matched comparison was made between a privately and
publicly prison, for every 100 inmates in a privately managed prison who received skills training
there would be 104 from the public section. Does this level of advantage represent a significant
value? It depends on (a) the presumed or proven relationship between skills training and healthy
living and (b) the scope or number of prisoners affected. That is, does skills training result in

Prison Privatization: A Meta-Analysis

21

lowered recidivism, better employment, or fewer social and emotional difficulties? If there is a
relationship, then the value would be significant for the additional 4 inmates (per 100) and their
families who received skills training. The benefits to society as a whole would need to be
considered in relation to the costs and benefits of skills training. Unfortunately, our lack of
expertise in this area limits our ability to comment authoritatively on these issues. That said, our
evidence suggests that the costs of publicly managed prisons is not statistically different from
privately managed systems. Thus, the 4% benefit in skills training may be “free.”
How do our findings compare with the previous reviews? Our findings on cost savings
concur with the meta-analysis done by Pratt and Maahs (1999), that minimal, if any, advantages
are realized through privatization. Our findings on quality concur with the review conducted by
Perrone and Pratt (2003), that the data are equivocal. Our findings differ with those of Segal and
Moore (1998) on both accounts. The “score” of findings from these four reviews is 3 to 1,
favoring an interpretation of minimal or no benefit from privatization. This finding is similar to
the experience auditors in the State of Florida had when trying to evaluate the value of
privatization. This group indicted that despite being able to compare their 5 privately managed
prisons with their public counterparts, it was not possible to determine whether there was a
benefit or liability from privatization (OPPAGA, 1997).
What recommendations can we offer? The data we reviewed do not support a move
toward privatization. Similarly, the data do not clearly discourage privatization despite a slight
advantage for publicly managed prisons in skills training. How should decisions be made when
clear evidence does not exist? Arguments in favor of privatization will fall back on ideology.
Arguments against privatization will likely center on ideology and the lack of support for
privatization.

Prison Privatization: A Meta-Analysis

22

Limitations.
Like all studies, ours contains several limitations which need to be considered in judging
the implications and inferences. First, although we did a broad literature search our findings are
based on only 12 studies. However, we believe that to date, this is the most comprehensive
review and represents the best available evidence. Second, the literature on prison quality and
cost does not provide a standardized approach to measuring outcomes. A criticism of metaanalysis is that varied indicators of a given construct are combined when there may not be
sufficient scientific rationale for doing so. On the other hand, it can be argued that relying on a
single indicator of a given construct is akin to not diversifying one’s investment portfolio. Our
assessment of the literature is that too much variability exists, making interpretations of our
findings difficult. That said, some form of objective measurement of constructs, in our opinion,
is better than subjective judgment. Third, at times we made some minor assumptions in
calculating effect sizes because sufficient data was not always present. To protect against being
subjective, we documented how each effect size was calculated and related decisions. This
information can be obtained from the first author. Moreover, we examined the impact of making
such assumptions and believe that they would have in no way changed the pattern of findings.
Fourth, while we performed a comprehensive search for studies detailing head-to-head
comparisons of privately and publicly managed prisons, we are not sure that all were found. We
do not believe the net effect of these limitations is serious because we believe we executed a high
quality meta-analysis. The limitations reflect the rather poor literature base on prison
privatization and the complexity of the questions.
Prior to concluding, we present findings from a unique model that involved a blending of
private and public entities. Shichor (1999) discussed how California allowed several public

Prison Privatization: A Meta-Analysis

23

facilities to be managed by municipalities or small cities with weak economic bases to bolster
budgets. This system appeared to produce results that were much more desirable than traditional
privately managed prisons. While firm conclusions cannot be made from a single study, it is
instructive to know that models beyond full privatization exist which may blend the arguments of
both sides on this debate.

Prison Privatization: A Meta-Analysis

24

References
Note. Articles noted with an “*” were included in the meta-analysis.
Abt Associates Inc. (2005). Contracting for Imprisonment in the Federal Prison System. Cost
and Performance of the Privately Operated Taft Correctional Institution. Cambridge,
MA: Douglas C. McDonald, Ph.D. & Kenneth Carlson.
Anderson, G. M. (2000). Prisons for Profit. America, 183-16, 12-16.
*Archambeault, W. G. Ph.D. & Deis, D., R. Jr., Ph.D. (1996). Cost Effectiveness Comparisons
of Private versus Public Prisons in Louisiana: A Comprehensive Analysis of Allen,
Avoyelles, and Winn Correctional Centers. Baton Rouge, LA: Louisiana State
University, School of Social Work, Office of Correctional Services.
Blakely, C. R. & Bumphus, V. W. (2004). Private and Public Sector Prisons—A Comparison of
Select Characteristics. Federal Probation, 68-1, 27-31.
*Brown, A. (1994). Economic and Qualitative Aspects of Prison Privatization in Queensland
(Australia). Annandale, New South Wales: Pluto Press.
*Bowery, M. (1996). NSW Department of Corrective Services. (1996). Private Prisons in NSW:
Junee – Year Two. (Research Publication No. 35, May 1996). Sydney, New South
Wales.

Calvert-Hanson, L. S. (1991). The Privatization of Corrections Movement: A Decade of
Change. Journal of Contemporary Criminal Justice, 7-1, 1-20.
Camp, C. & Camp, G. (1997). The Corrections Yearbook. Criminal Justice Institute. 1997, 11.
Cikins, W., I. (1986).Privatization of the American Prison System; An idea whose time has
come? Notre Dame Journal of Law and Ethics. 2, 445-64.

Prison Privatization: A Meta-Analysis

25

Chang, T.F.H. & Thompkins, D. E. (2002). Corporations Go to Prisons: The Expansion of
Corporate Power in the Correctional Industry. Labor Studies Journal, 22-1, 45-69.
Cohen, J. (1988). Statistical power analysis for the behavioral sciences (2nd ed.). Hillsdale, NJ:
Erlbaum.
Cooper, H., & Hedges, L. V. (1994). The handbook of research synthesis. New York: Sage.
Culp, R. F. (2005). The Rise and Stall of Prison Privatization: An Integration of Policy Analysis
Perspectives. Criminal Justice Policy Review, 16-4, 412-442.
*Drowata, C. & Don S. Select Oversight Committee on Corrections. (1995). Comparative
Evaluation of Privately-managed CCA Prison (South Central Correctional Center) and
State-managed Prototypical Prisons (Northeast Correctional Center, Northwest
Correctional Center).
*Greene, J. (1999). Comparing Private and Public Prison Services and Programs in Minnesota:
Findings from Prisoner Interviews. Current Issues in Criminal Justice, 11-2, 202-225.
*Hatry, H. P., Brounstein, P. J. & Levinson, R.B. (1989). Comparison of Privately
and Publicly Operated Corrections Facilities in Kentucky and Massachusetts.
Howell, D. C. (1997). Statistical methods for psychology (4th Ed.). Belmont, CA: Wadsworth
Publishing Company.
Johnson, B. T. (1993). DSTAT 1.10: Software for the meta-analytic review of literatures
[Computer software and manual]. Hillsdale, NJ: Erlbaum.
Lanza-Kaduce, L., Parker, K., & Thomas, C. (1999). A comparative recidivism analysis of
releases from private and public prisons. Crime & Delinquency. 45, 28-47.
Lipsey, M. W., & Wilson, D. B. (2001). Practical meta-analysis. Thousand Oaks, CA: Sage.

Prison Privatization: A Meta-Analysis

26

Logan, C. H. (1992). Well Kept: Comparing Quality of Confinement in Private and Public
Prisons. The Journal of Criminal Law and Criminology, 83-3, 577-613.
*Maximus, Inc. (2006). Report on the Evaluation of Arizona Department of Corrections’
Operating Per Capital Cost Report and Private Prison Cost Mode. Phoenix, AZ: Dora
Schriro.
*McDonald, D. C. Ph.D. & Kenneth Carlson Abt Associates Inc. (2005). Contracting for
Imprisonment in the Federal Prison System. Cost and Performance of the Privately
Operated Taft Correctional Institution. Cambridge, MA: Douglas C. McDonald, Ph.D. &
Kenneth Carlson.
Moore, A. (1998).Private prisons: Quality corrections at a lower cost. Reason Public Policy
Institute Study No. 240. 18.
*OPPAGA Private Prison Review. (2000). South Bay Correctional Facility Provides Saving and
Success; Room for Improvement. (Report No. 99-39, March 2000). Tallahassee, FL:
Office of Program Policy Analysis and Government Accountability.
Perrone, D., & Pratt, T. C. (2003). Comparing the quality of confinement and cost-effectiveness
of public versus private prisons: What we know, why we do not know more, and where to
go from here. The Prison Journal, 83, 301-322.
Pratt, T. C., & Maahs, J. (1999). Are private prisons more cost-effective than public prisons? A
meta-analysis of evaluation research studies. Crime & Delinquency, 45, 358-371.
Sechrest, D., & Shichor, D. (1993). Corrections goes public (and private) in California. Federal
Probation. 57, 3-8.

Prison Privatization: A Meta-Analysis

27

Segal, G. F., & Moore, A. T. (2002). Weighing the watchmen: Evaluating the costs and benefits
of outsourcing correctional services. Part II: Reviewing the literature on cost and quality
comparisons. Reason Public Policy Institute.
Select Oversight Committee on Corrections. (1995). Comparative Evaluation of Privatelymanaged CCA Prison (South Central Correctional Center) and State-managed
Prototypical Prisons (Northeast Correctional Center, Northwest Correctional Center).
City, ST: Clair Drowata, Claire & Don Stoughton.
*Sellers, M.P. (1989). Private And Public Prisons: A Comparison Of Costs, Programs And
Facilities. International Journal of Offender Therapy and Comparative Criminology, 33,
241 - 256.
Shichor, D. (1999). Privatizing Correctional Institutions: An Organizational Perspective. The
Prison Journal, 79-2, 226-249.
*Thomas, C. Ph.D. (August, 1997). Comparing the Cost and Performance of Public and Private
Prisons in Arizona: An Overview of the Study and Its Conclusions. Retrieved December,
2006, from http://www.anarchistblackcross.org/content/essays/articles/priv/ariz.htm.
Urban Institute. (1989). Comparisons of privately and publicly operated corrections facilities in
Kentucky and Massachusetts. Urban Institute.

Prison Privatization: A Meta-Analysis

28

Notes
1

This study was undertaken to provide information that the State of Utah’s Legislature

might use in making a decision to privatize Utah’s prison system.

2

To assist readers of this report judge the findings, we make the following disclosure. The

primary investigator, Brad Lundahl, PhD, and his research team did not begin this project with
an opinion about the value, or lack thereof, of prison privatization. Dr. Lundahl initiated and
completed this study conditioned upon an agreement that an objective and independent
investigation would proceed and that all results would be presented regardless of whether such
results would support or not support a move toward prison privatization. Dr. Lundahl was invited
to conduct a meta-analysis on prison privatization outcomes by Mr. Russ Van Vleet, co-Director
of the Utah Criminal Justice Center (UCJC) as a means of informing the State of Utah’s
legislative body about the potential risks and rewards of prison privatization. Dr. Lundahl did not
receive funding from the UCJC, although two master-level graduate research assistants who
worked on this project were paid through the UCJC. Dr. Lundahl has not previously researched
issues related to criminal justice and will not likely research privatization in the future as his
research agenda is far-a-field from this area. Thus, he truly comes to the question of privatization
without a biased history and without pressure to bias the results.

3

Mr. Cliff Butter from the State of Utah’s Department of Corrections served as an outside

consultant. Mr. Butter provided us with several articles, none of which met final the 5 inclusion
criteria.

Prison Privatization: A Meta-Analysis

29

Table 1
Arguments regarding prison privatization

Argument In Favor of Privatization
1. Cost Effectiveness
a. Innovative practices are developed to contain costs
b. Reduce labor costs through improved scheduling and management
c. Negotiate for lower prices more effectively
d. Respond to problems and opportunities faster because they are not encumbered
by the “red tape” and restrictions common to government bureaucracies
e. Construct buildings faster and more efficiently
f. Pay taxes to government because they are a business
2. Restraining Costs System-Wide
a. Increase market competition which lowers overall prison costs
3. Provide Higher Quality of Confinement
a. Provide high quality services to avoid inmate grievances/legal action
b. Implement programs in an efficient and streamlined manner
c. Emphasize quality to secure repeat contracts and a positive public image
4. Government Abetment
a. Reduce the pressure and costs of overcrowding in the public prison system
5. Privatization has worked in other sectors and should work for prison management
Arguments Against Privatization
1. Ethical Conflicts of Interest
a. Pressure to show profit may lead to compromised service quality (e.g., lower staff
to inmate ratio, fewer rehabilitative services, reduced range of services, etc)
b. Potential for abuse when a profit making company has authority to restrict basic
civil liberties
c. Profit motives may supersede the interests of the public and inmates
d. Fear of negative financial consequences may lead to nondisclosure of problems
(e.g., may adopt in-house grievance system that circumvents traditional judicial
system)
e. Creates an environment that may support corruption such as bid rigging, bribes,
kickbacks
2. Lobbying
a. Private companies may attempt to influence legislation that would favor profit
(e.g., change laws that drive up incarceration rates and sentences)

Prison Privatization: A Meta-Analysis

30

Table 1 (continued)

3. Workforce Quality May Suffer And Degrade Overall Quality
a. Discourage unionization
b. Pay lower salaries and benefits, which leads to lower morale and may promote
corruption
c. Low morale among employees increases turnover rates, which can compromise
basic prison functioning (e.g., security, safety)
4. Cost Effectiveness Defense
a. Large economy of scale also applies to the government
b. Cost comparisons with private companies overlook hidden costs (e.g.,
government monitoring, triage, major medical costs)
c. Accountable to the public for approval of costs
5. Long Term Obligations
a. If the private company goes out of business, the government retains liability
b. The government is responsible for building depreciation
c. The state is in a poor bargaining position when their prison is overcrowded or at
maximum capacity and the private company’s contract is up for renewal

Prison Privatization: A Meta-Analysis

31

Prison Privatization: A Meta-Analysis

32

Table 2
Characteristics of Compared Prisons
Study name

Information
source

Number of
prisons
compared

Security
level

Inmate
gender

Prison age
(in months)

Facility
type

Prison
capacity

Same
time
frame?

Construction
managed by
private company

Archambeault
1996

RO,SI,
II,OR

G=1
P=2

NR

Mix

State

Yes

No

Bowery

AR, SI

G=3

Mix

NR

G1=204
P1=192
P2=192
G=NR

Yes

Yes

Mix

NR

Yes

No

Medium

Mix

Yes

Yes

Medium

Male

1996
Brown
1992
Drowota
1995
Greene
1999

Hatry
1989
Logan
1992
Maximus
2006
McDonald
2005
OPPAGA
2000
Sellers
Part 1

OR
OR, II,
SI, RO
SI

P=1
G=1
P=1
G=2
P=1
G=3
P=1

P=168
G=NR
P=204
NR

State
State
State

P1=1,474
G1=1,474
G2=1,474
3068/2yrs
1,481/2nd
yr
G=NR
P=244
NR

State

NR

Yes

No

State

NR

Yes

Yes

State/
Federal

G1=609
G2=NR
P=200
NR

Yes

Yes

Yes

NR

Federal/

AR, SI,
II, RO
OR, SI,
II

G=1
P=1
G=2
P=1

Minimum

Mix

Mix

Female

AR

G=1
P=3
G=Survey
P=1

Mix

NR

G1=348
G2=228
G3=204
P1=156
G=240
P=NR
G=48
G=NR
P=0
NR

Minimum

G=Mix
P=Male

G=N/A
P=108

Federal

G=N/A
P=2,048

Yes

NR

Close

Male

State

Yes

Mix

G=1,093
P=1,318
G=95
P=106

Yes

High

G=144
P=120
NR

Yes

NR

OR

AR
SI, II,
RO

G=1
P=2
G=1
P=1

State

County

Prison Privatization: A Meta-Analysis
Sellers
Part 2
Thomas
1997
Date 3-29-07

SI, II,
RO
OR

G=1
P=1
G=Survey
P=1

High

Mix

Minimum

Mix

G=240
P=264
G=N/A
P=144

State
State

33
G=87
P=350
NR

Yes

Yes

Yes

NR

Notes. SI = staff interview. OR = official records. AR = agency records. II = inmate interviews. RO = research observations. G =
Government managed facility. P = Privately managed facility. NR = Not reported.

Prison Privatization: A Meta-Analysis

34

Table 3
Percent Economic Savings and Effect Size for Quality Performance of Private versus Public Prisons
Cost
Saving1

Public
Safety

Prison
Safety1

Prison
Order

Archambeault,
3
1996

13.0%

-.04

.05

.03

Bowery, 1996

–

-.384

-.06

Brown, 1994

0.0%1

Study

Health
Care

Skills
Training

Misc
Benefits

Griev- Facility
ance
Conditions

Employee
Work condition

–

-.01

–

-.07

–

.10

.05

–

-.12

-.15

-.05

–

–

–

–

–

–

–

–

–

–

–

–

–

–

-.01

Drowota, 1995

–

-.04

-.11

–

.01

–

Greene, 1999 -

–

–

–

–

-.13

-.17

-.06

–

–

–

–

–

.03

–

–

.00

–

–

.01

.02

.03

.02

.00

–

.01

.01

.02

.03

–

–

–

–

–

–

–

–

–

-.07

.09

-.12

–

–

–

-.18

–

–

-.06

–

–

.06

–

–

–

–

–

–

–

–

–

–

0.16

–

–

–

15.2%1

–

–

–

–

–

–

–

–

–

2.2% (11.5)

-.043 (.03)

.00 (.08)

Hatry, 1989 -

-10.0%1

Logan, 1992

–

Maximus, 2006

-14.2%1

Mcdonald, 2005

9.0%1

OPPAGA, 2000

4.6%

3

Sellers, 1989

Thomas, 1997
Mean (SD)

1

0.0% 1,2

.00 (.07) - .02 (.08)

-.10 (.08)

-.01 (.11)

-.07 (.08)

.02 (n/a)

.03 (.05)

Prison Privatization: A Meta-Analysis

35

Note. SD = Standard Deviation. Effect sizes were calculated so that a positive valence reflects an advantage for privately managed
prisons and a negative valence reflects an advantage for publicly managed prisons. 1Controlled for or made adjustments for
administrative costs or costs of managing a private sector prison. 2We did not average in difference of Silverdale versus Warren
County because sites not properly matched and a huge difference, in favor of private (126% better) would have been an outlier and
given undue weight to one study. 3In computing the average, the original value was Windorized to the next highest value within a
column to limit the undue influence of an outlier.