ArticleSex Offender Recidivism Revisited Review ofRecen.docx

Article
Sex Offender Recidivism Revisited: Review of
Recent Meta-analyses on the Effects of Sex
Offender Treatment
Bitna Kim
1
, Peter J. Benekos
2
, and Alida V. Merlo
1
Abstract
The effectiveness of sex offender treatment programs continues
to generate misinformation and disagreement. Some literature
reviews conclude that treatment does not reduce recidivism
while others suggest that specific types of treatment may
warrant
optimism. The principal purpose of this study is to update the
most recent meta-analyses of sex offender treatments and to
com-
pare the findings with an earlier study that reviewed the meta-
analytic studies published from 1995 to 2002. More importantly,
this study examines effect sizes across different age populations
and effect sizes across various sex offender treatments. Results
of
this review of meta-analyses suggest that sex offender
treatments can be considered as ‘‘proven’’ or at least
‘‘promising,’’ while age

of participants and intervention type may influence the success
of treatment for sex offenders. The implications of these
findings
include achieving a broader understanding of intervention
moderators, applying such interventions to juvenile and adult
offenders,
and outlining future areas of research.
Keywords
offenders, sexual assault, recidivism, intervention
Introduction
The topic of sex offenders generally elicits fear and anxiety
from the public and contributes to punitive policies aimed at
harsh, exclusionary punishments. The perspective that commu-
nities need to be protected from sex offenders through incar-
ceration and surveillance often overshadows the prospects
that treatment can also provide public safety. In their study,
Kernsmith, Craun, and Foster (2009) found that citizen respon-
dents who reported higher levels of fear of sex offenders were
more supportive of registration requirements for sex offenders.
Levenson, Brannon, Fortney, and Baker (2007) also reported
that public perceptions of sex offenders reflect public anxiety
and support for community protection.

Although negative attitudes toward sex offenders are not
reflective of all countries, cultural differences and historical
context can account for less punitive public responses. For
example, McAlinden (2012) found that therapeutic interven-
tions for sex offenders were more prevalent in European coun-
tries than in England and Wales. She attributes this to a more
scientific and medical approach to sex offending across Europe
and less emphasis on ‘‘sexual abuse as a moral, legal, and social
problem’’ (p. 170). Nevertheless, the sex offender problem has
become more serious across Europe and policies reflect a shift
toward more punitive attitudes and sanctions (McAlinden,
2012). Not only in European countries but also in the United
States, one of the misgivings about how to respond to sex
offenders concerns the effectiveness of treatment.
In this article, the authors address the treatment issue by
updating the meta-meta-analytic study of Craig et al. (2003)
on sexual offender treatment. This study augments the original
work of Craig et al. by incorporating more recent meta-analytic

studies in the analysis. In this research, all salient meta-analytic
sex offender treatment studies from 1995 to 2010 were
included. The purpose of this study is to systematically review
what is known about the effectiveness of sex offender treat-
ments based on results of extant meta-analyses of different
types of treatment for sex offenders.
Furthermore, the study examines the issue of treatment spe-
cificity and which treatment strategies are effective for adult
versus juvenile offenders. Juvenile offenders who commit sex
offenses can evoke more alarm and fear among the public and
prosecutors because age is viewed as an aggravating character-
istic that can contribute to reoffending. When the prosecutors
emphasize public safety, this not only reinforces fears, but also
justifies more punitive rather than therapeutic responses.
Michels reports that prosecutors can take the position that
1
Department of Criminology, Indiana University of
Pennsylvania, Indiana, PA,
USA

2 Criminal Justice Department, Mercyhurst University, Erie,
PA, USA
Corresponding Author:
Bitna Kim, Department of Criminology, Indiana University of
Pennsylvania,
Indiana, PA 15705, USA.
Email: [email protected]
TRAUMA, VIOLENCE, & ABUSE
2016, Vol. 17(1) 105-117
ª The Author(s) 2015
Reprints and permission:
sagepub.com/journalsPermissions.nav
DOI: 10.1177/1524838014566719
tva.sagepub.com
http://www.sagepub.com/journalsPermissions.nav
http://tva.sagepub.com
http://crossmark.crossref.org/dialog/?doi=10.1177%2F15248380
14566719&domain=pdf&date_stamp=2015-01-08
juvenile sex offenders are the ‘‘worst of the worst’’ because
they
are more prone to reoffending and ‘‘therefore too dangerous to
release’’ (2012, { 9). This demonizing of juvenile sex offenders
reflects a concern that this population cannot be effectively
treated, that they are at greater risk of recidivism, and that they
present a threat to public safety. Although these views are gen-

erally inaccurate, they do impact public reaction and
prosecutor-
ial responses (Chaffin, 2008; Letourneau & Miner, 2005).
This study assesses the effectiveness of sex offender treat-
ment programs and includes 11 meta-analytic studies, 6 of
which were included in the Craig et al. (2003) study and 5 of
which are more recent. Cohen’s d was reported to aid in the
interpretation of effect sizes. Definitions of small (d ¼ .20),
medium (d ¼ .50), and large (d ¼ .80) effects were based on
Cohen’s (1988) guide and based on effect sizes encountered
in the behavioral sciences (Cooper, 2010). These guides are
most appropriately employed ‘‘when no better basis for esti-
mating the effect size is available’’ (Cohen, 1988, p. 25).
Two other descriptors of research results related to program
evaluations that have recently received attention among some
social scientists are ‘‘proven’’ and ‘‘promising’’ (Cooper,
2010). Among different guides for magnitude labels of proven
and promising, the Promising Practices Network (PPN) is con-
sidered as credible by associating the terms such as proven and

promising with the solid evidence criteria (e.g., type of out-
comes affected, substantial effect size, statistical significance,
comparison groups, sample size, and availability of program
evaluation documentation (Cooper, 2010; PPN, 2007).
According to the PPN (2007), in order for a program to be
labeled proven, the associated evidence must meet the follow-
ing criteria: ‘‘(1) the program must directly affect one of the
indicators of interest; (2) at least one outcome is changed by
20%, d ¼ .25, or more; (3) at least one outcome with a substan-
tial effect size is statistically significant at the 5% level; (4) the
study design used a convincing comparison group to identify
program impacts, including studies that used random assign-
ment or some quasi-experimental designs; (5) the sample size
of the evaluation exceeds 30 in both the treatment and compar-
ison groups; and (6) the report is publicly available’’ (Cooper,
2010, p. 209). An intervention would be labeled promising if it
measured the outcomes of most interest and used rigorous
designs and revealed a smaller effect size (e.g., an associated
change in outcome of more than 1%) that PPN requires for a
program to be considered proven (Cooper, 2010, p. 209). One

purpose of this study is to determine whether the current evi-
dence supports a conclusion that sex offender treatment is pro-
ven or promising. This study utilizes both Cohen’s (1988)
guide and the PPN (2007) guide to convey proven and promis-
ing findings of sex offender treatments.
Sex Offender Treatments
Cognitive Behavioral Therapy
The treatment foundation that is used in many sex offender pro-
grams is cognitive behavioral therapy (CBT) and relapse
prevention (Baker, 2012; Brandes & Cheung, 2009; Center
for Sex Offender Management, 2006; McGrath, Cumming,
Burchard, Zeoli, & Ellerby, 2009; Worling & Langton, 2012).
Based on their survey of 1,379 programs in the United States
and Canada, McGrath et al. (2009) reported that the cognitive
behavioral model was in the top three choices for most adult
and adolescent programs (86%) and relapse prevention was
in the top two choices for 50% of the programs.
CBT combines two psychotherapies to address thoughts and

beliefs as well as behaviors and actions (Development Services
Group, Inc., 2009). The cognitive focus is on assumptions and
attitudes that contribute to dysfunctional thinking that rein-
forces patterns of unacceptable or inappropriate behaviors. The
behavioral component emphasizes actions and settings that
contribute to patterns of behavior. This problem-focused
approach helps sex offenders learn new skills and develop com-
petencies in maintaining appropriate behaviors. CBT confronts
rationalizations about behavior and provides skills to control
sexual impulses. Similarly, relapse prevention is also a cogni-
tive approach that helps sex offenders regulate their own beha-
viors by recognizing internal and external risks and learning to
manage their behaviors.
In his review of CBT, Greenwald (2009) described struc-
tured intervention strategies that improve interpersonal
problem-solving skills and facilitate more effective communi-
cation skills. By developing self-management skills that recog-
nize social cues and maladaptive behaviors, treatment provides

more constructive ways of thinking and understanding the con-
sequences of behavior. Corson (2010) also noted that social and
life skills training and cognitive restructuring are characteris-
tics of CBT. Essentially, treatment programs include various
strategies that focus on correcting thoughts, feelings, and beha-
viors that promote inappropriate behaviors and replacing them
with self-directed behavioral skills that maintain prosocial
beliefs and behaviors.
As previously noted, CBT presents strategies that are effec-
tive in cognitive restructuring that improve victim empathy and
complement relapse prevention (Craig, Browne, & Stringer,
2003; Hanson, Bourgon, Helmus, & Hodgson, 2009). Galla-
gher, Wilson, Hirschfield, Coggeshall, and MacKenzie
(1999) noted that cognitive behavioral treatment is not only
broadly supported in the literature, but also in their meta-
analytic study of 25 studies, in which they found that ‘‘ . . .
cognitive behavioral programs are effective in reducing the
recidivism of treated offenders’’ (p. 27). In addition, Marshall
and McGuire (2003) found supporting evidence that treatment

of sex offenders is effective. In reporting that recidivism among
sexual offenders is lower than among other offenders, Mann,
Hanson, and Thornton (2010) observe that this contradicts
common beliefs.
CBT is also the most prevalent treatment approach for ado-
lescent sex offenders and has wide support. For example, in
their
meta-analytic study, Reitzel and Carbonell (2006) found that
cognitive behavioral approaches were the most effective for
juvenile offenders. The cognitive behavioral treatment–relapse
prevention (CBT-RP) approach to treatment underscores
106 TRAUMA, VIOLENCE, & ABUSE 17(1)
changing ‘‘thoughts, behaviors, and arousal patterns of juvenile
sex offenders’’ (Fanniff & Becker, 2006, p. 273).
Generally, programs that use CBT-RP to work with adoles-
cent sex offenders motivate them to reject their thinking errors
and to identify situations and ideations that precipitate inap-

propriate behaviors. In addition, adolescents learn to recognize
the connection between their emotions and behaviors. (Bourke
& Donohue, 1996; Hall, 1995; Hunter & Santos, 1990; Lipsey,
2009; Marques, Wiederanders, Day, Nelson, & van Ommeren,
2005). Since juveniles are still maturing, developing, and
experimenting, there is an expectation that they are more
responsive to cognitive restructuring and skills development.
This is consistent with findings that juvenile sex offenders who
receive treatment have low rates of reoffending (Baker, 2012;
Center for Sex Offender Management, n.d.; Reitzel & Carbo-
nell, 2006).
Multisystemic Therapy
Another promising approach for treating sex offenders is multi-
systemic therapy (MST) (Borduin, Schaeffer, & Heiblum,
2009; Fanniff & Becker, 2006; Henggeler, 2012; Huey, Heng-
geler, Brondino, & Pickrel, 2000; MST Associates, n.d.). MST
was originally developed by Scott Henggeler as a family-based
treatment program for antisocial children and serious delin-

quent offenders. The emphasis of MST was on working with
families to improve monitoring, supervising, and disciplining
youth, and on reducing deviant peer affiliations (MST Services,
n.d.). MST has been adapted and has demonstrated effective-
ness in treating adolescent socialization issues and interperso-
nal relations (Crime
Solution
s, n.d.; Henggeler, 2012). The
intervention is provided at home or in the community and
focuses on interrupting the sexual assault cycle by working
with the offender and his family to develop a safety plan, by
empowering the family with skills and resources to more effec-
tively parent, and by targeting treatment toward individual
and family risk factors for sexual and nonsexual delinquency

(Fanniff & Becker, 2006; Henggeler, 2012).
Borduin et al. (2009) reported that juvenile sexual offenders
treated with MST had lower recidivism rates than offenders
receiving ‘‘usual’’ community services. Multiple randomized
controlled trials of MST provided to juvenile sex offenders
have found reductions in recidivism, problematic sexual beha-
vior, and out-of-home placements (Letourneau et al., 2009). In
their meta-analytic study, Walker, McGovern, Poey, and Otis
(2004, p. 289) found that MST appeared promising and they
recommended that future research on adolescent sexual offen-
der treatment ‘‘test the effectiveness of CBT against that of
multisystemic therapy.’’
Using their findings from a meta-analysis and distinguishing

between specialist and generalist sex offenders, Pullman and
Seto (2012) recommended both MST and CBT in order to
achieve more effective treatment outcomes. They concluded
that using MST and CBT to focus on sexual self-regulation
results in lower recidivism for specialist adolescent sex offen-
ders than using MST alone.
Additional Sex Offender Treatments
Sex offender therapy can also include medical interventions
that are either physical or chemical. Surgical procedures denote
mechanical castration, and chemical castration refers to hormo-
nal drugs such as antiandrogen, which are used to reduce sexual
arousal (Pray, 2002, p. 99). Gallagher et al. (1999) reported that

cognitive behavioral treatment (or other psychological treat-
ment) is sometimes used in conjunction with hormonal treat-
ment such as Depo-Provera, which reduces physiological
drive to engage in deviant behavior (Gallagher, Wilson,
Hirschfield, Coggeshall, & MacKenzie, 1999, p. 25).
In his study of hormonal treatments, Hall (1995) found that
effect sizes in studies that used a cognitive behavioral approach
were not significantly different from those that employed hor-
monal treatments. Hall performed a meta-analysis of 12 pri-
mary studies and found that both cognitive behavioral and
hormonal treatments were effective. However, the refusal and
discontinuation rates of hormonal treatment participants is con-
siderably higher compared to cognitive behavioral treatment

participants, and Hall suggests that this may indicate that cog-
nitive behavioral treatment is more advantageous (p. 807).
More recently, Rice and Harris (2011) also considered the
effectiveness of androgen deprivation therapy (ADT) to reduce
sexual recidivism. In describing the outcomes of surgical and
chemical treatment, the authors acknowledge that voluntary
subjects and weak methodology limit confidence in the out-
comes. Although some studies comparing volunteers with refu-
sers report favorable outcomes using pharmacological ADT,
the authors identify sufficient concerns to conclude that ‘‘ADT
cannot serve as a guarantee against sexually violent recidi-
vism’’ (p. 325). In the cases of men who volunteer and request

ADT, sexual recidivism may be reduced but this may be more
indicative of the characteristics of volunteers rather than the
effects of ADT (p. 328).
In addition to qualified conclusions about the effectiveness
of ADT, the authors recognize legal and ethical issues that sur-
round the use of castration. For example, long-term effects of
ADT on health, sexual behavior, and sexual recidivism remain
a concern among researchers and therapists. In spite of the sup-
port that androgen reduction therapy receives from some thera-
pists, the differential effects experienced by sex offenders and
the methodological limitations of many studies lead Rice and
Harris to conclude that ‘‘Clearly, much more research is needed
before ADT has a sufficient scientific basis to be relied upon as

a principal component of sex offender treatment’’ (p. 328).
Although it is more controversial, surgical castration can be
used in concert with other types of treatment, including psycho-
logical approaches. Although the operation is performed infre-
quently, it has been utilized in Western Europe and in the
United States. In one study of German offenders, Wille and
Beier (1989) found that the surgically castrated offenders
(volunteers) were more likely to refrain from further sexual
offending than offenders who had applied for the surgery but
were denied approval or withdrew their request (Gallagher
et al., 1999, p. 25). Due, in part, to the dearth of studies on this
Kim et al. 107

treatment approach and the lack of a similar control group in
the Wille and Beier study, researchers are reluctant to embrace
its effectiveness (Eher & Pfäfflin, 2011).
Current Study
Although several narrative reviews of sex offender interventions
exist, the most useful are meta-analyses that quantitatively
synthesize the literature. Meta-analyses are characterized by a
number of strengths, including (1) exhaustive literature
searches, (2) an ability to synthesize large literature, (3) a focus
on precise effect sizes rather than solely on statistical signifi-
cance, and (4) an ability to empirically test moderators of study
outcomes and help understand why certain studies had stronger

effects than others (Noar, 2008). Given that the literature of sex
offender treatment has continued to grow at a rapid pace, these
more recent meta-analyses have taken advantage of more
sophisticated analyses that larger literature permit (Noar, 2008).
Craig et al. (2003) previously reviewed six meta-analytic
studies that were published from 1995 to 2002 (Alexander,
1999; Aos, Phipps, Barnoski, & Lieb, 2001; Gallagher et al.,
1999; Hall, 1995; Hanson et al., 2002; Polizzi, MacKenzie,
& Hickman, 1999) and concluded that there were positive
treatment effects in reducing sexual offense recidivism. The
principal purpose of this study is to update the most recent
meta-analyses of sex offender treatments and compare the

findings with those of Craig et al. (2003). This is a replication
of the earlier Craig et al. (2003) study with an expanded sample
of meta-analyses. In addition, this study extends the earlier
review by examining and comparing: (1) effect sizes across the
meta-analytic literature, (2) effect sizes across different target
populations (adolescents vs. adults) in order to examine how
sex offender treatments have performed across populations, and
(3) effect sizes across different types of sex offender
treatments.
Method
Search Strategy and Inclusion Criteria
To comprehensively identify meta-analysis studies on sex
offender treatment, the authors conducted a search of a number

of online databases in which criminal justice-related meta-
analyses might plausibly be reported. The intent was to locate
all meta-analyses of sex offender treatments published in peer-
reviewed journals that were available (in print or electronic
form) or in dissertation databases and met criteria for this
review (Noar, 2008).
The search looked for any mention in the title, the abstract,
or the keyword list of the words ‘‘meta-analysis,’’
‘‘quantitative
review,’’ and ‘‘systematic review,’’ paired with any of the fol-
lowing terms: sex offender treatment or sex offender interven-
tion. The specific databases used were: Criminal Justice
Abstracts, Sociological Abstracts, PsychINFO, MEDLINE,

Social Science Abstracts, Psychology and Behavioral Science
Collections, and Current Contents. In addition, computer and
manual searches identified listings of unpublished materials
(Dissertation Abstracts International, ERIC). The reference
lists of those articles retrieved from each of the databases were
scanned to identify additional studies that may have used
meta-analytic procedures (Lundahl, Taylor, Stevenson, &
Roberts, 2008). The abstracts of likely references were
reviewed to confirm that they used meta-analysis, and an
attempt was made to obtain copies of each of the likely candi-
dates (Wells, 2009).
Meta-analyses were included in the review if they: (1) con-
ducted a meta-analysis (quantitative research synthesis) of

formally developed and evaluated sex offender treatments tar-
geting recidivism; (2) were focused on a defined target popu-
lation of adolescent and adult sex offenders; and (3) examined
outcome variables of sexual recidivism, violent recidivism, or
any recidivism. As a result of these search strategies and
inclusion criteria, a final set of 11 meta-analyses were
included in the current review. Of the 11 meta-analyses, 5
studies were published since 2002 and not included in Craig
et al. (2003).
Effect Size Conversion
Effect size essentially refers to the magnitude of the ‘‘effect’’
of
the program on recidivism (Cohen, 1988). Bigger program

effects (impacts) imply that the program had a greater effect
than smaller effect sizes. The meta-analyses included in this
review used differing effect size indicators. In order to provide
a common metric for interpretation and comparison across all
meta-analyses, effect sizes and confidence intervals in the odds
ratios and r meta-analyses were converted to d using the fol-
lowing equations (Ellis, 2010):
d ¼
2r
ffiffiffi
1
p
�r2
and d ¼ log odds ratio�
ffiffiffi

3
p
p
Negative effect size indicates recidivism reduction among
intervention participants. One arbitrary criterion used to
determine what constitutes a big effect size as opposed to a
smaller one is that effect sizes of .20 are small, .50 are
medium, and .80 or higher are large (Polizzi et al., 1999).
Cohen (1988) suggests that a small effect of d ¼ .20 is typical
of those found in personality, social, and clinical psychology,
while a large effect as d ¼ .80 is more likely to be found in
sociology, economics, and experimental or physiological psy-
chology (Cooper, 2010).
Although d is probably one of the best known effect size

indexes, a more compelling way to provide a translation of the
effects of discrete interventions on dichotomous outcomes
(e.g., success or recidivism) is to present the results in a bino-
mial effect size display (BESD) (Cooper, 2010). Developed by
Rosenthal and Rubin (1982), the BESD is a 2 � 2 contingency
table where the rows correspond to the independent variable
(e.g., treatment and control) and the columns correspond to any
dependent variable that can be dichotomized (e.g., success or
recidivism). For any given correlation (r), the success rate for
the treatment group is calculated as (.50 þ r/2), while the suc-
cess rate for the control group is calculated as (.50 � r/2).
108 TRAUMA, VIOLENCE, & ABUSE 17(1)

In this study, to use the BESD for a standardized mean dif-
ference effect size of d, the effect size of d was converted into
the correlational equivalent using the formula, r ¼ d=
p
4 þ d2
(Lipsey & Wilson, 2001). For example, d ¼ .60 is converted to
the correlation effect size of .30. So, the value in the success-
treatment cell is .65 (or .50 þ .30/2) and the value in the
success-control cell is .35 (or .50 � .30/2). The BESD shows
that success was observed for nearly two thirds of the people
who undertook treatment but only a little over one third of
those in the control group (Ellis, 2010). The difference between
the two groups is 30 percentage points, meaning that those who
took the treatment saw an 86% improvement in their success
rate (representing the 30 percentage point gain divided by the
35-point baseline; Ellis, 2010). It is easier to comprehend the

magnitude of a relationship if it is expressed as a difference
between a 65% and a 35% success rate than if it is expressed
as a correlation effect size of .30 or a standardized mean effect
size of .60.
Results
Table 1 lists characteristics for each of the 11 meta-analyses.
As can be seen, to date three meta-analyses (Gallagher et al.,
1999; Hanson et al., 2002, 2009) included studies conducted
both inside the United States and outside the United States, and
the remaining eight meta-analyses included only American
studies on sex offender treatment. Six meta-analyses included
in Craig et al. (2003) examined the research on sex offender
treatments from as early as 1943 (Doshay, 1969/1943) and as

late as 2000 (Borduin, Schaeffer, & Heiblum, 2000; Hanson
& Nicholaichuk, 2000; Looman, Abracen, & Nicholaichuk
2000; McGuire, 2000; Nicholaichuk, Gordon, Gu, & Wong,
2000; Walker, 2000), while the newly added five meta-
analyses in the current review have examined the research as
late as 2009 (Borduin et al., 2009). In the current review, the
term ‘‘study’’ (represented by the letter k) is used to refer to the
primary intervention trials. This set of meta-analyses typically
treated each research trial as one study (deriving one effect size
from each report), although in some cases trials only reported
data in subgroups (e.g., separately for adolescents and adults),
leading meta-analysts to treat those separate groups as different

‘‘studies’’ (deriving multiple effect sizes from a single report;
Noar, 2008). Using this definition, these meta-analyses have
included as few as 9 studies with a cumulative N ¼ 2,986 (Reit-
zel & Carbonell, 2006) and as many as 79 studies with a cumu-
lative N ¼ 10,988 (Alexander, 1999), with a median of k ¼ 22
primary studies.
Efficacy of Sex Offender Treatment
Table 2 is a summary of effect size indices across study out-
comes in the meta-analyses. The effect sizes for the recidivism
measures are listed in the third and fourth columns of the table.
Results from all meta-analyses favored the treatment group. All
effect sizes reported are from fixed effects analyses except for
Gallagher et al. (1999); Hanson, Bourgon, Helmus, and Hodg-
son (2009); and Lösel and Schmucker (2005).

Results indicated that every meta-analysis (Alexander,
1999; Aos et al., 2001; Gallagher et al., 1999; Hall, 1995; Han-
son et al., 2002; Polizzi et al., 1999) examined in Craig et al.
(2003) found significant effects, and the mean effect size was
d ¼ �.20 (range �.11 to �.43), suggesting the sex offender
treatments produced an overall 10% reduction in recidivism.
The weakest effect was found in Aos et al. (2001), which
synthesized the outcomes of the cognitive behavioral treatment
(k ¼ 25), psychotherapy (k ¼ 6), behavioral treatment (k ¼ 5),
chemical treatment (k ¼ 3), and surgical treatment (k ¼ 2) for
adults in the United States (see Table 1). The strongest effect
size was found in Gallagher et al. (1999), which synthesized the
outcomes of both psychological therapies (k ¼ 20) and surgical
castration (k ¼ 1) and chemical castration/supplemental com-
ponent (k ¼ 4) for adolescents and adults in the United States
(k ¼ 14), Canada (k ¼ 10), and Germany (k ¼ 1; see Table 1).

The more recent five meta-analyses (Hanson et al., 2009;
Lösel & Schmucker, 2005; Pray, 2002; Reitzel & Carbonell,
2006; Walker, McGovern, Poey, & Otis, 2004) were included
in the current review. Results of these five meta-analyses indi-
cated that every meta-analysis found significant effects, and the
mean effect size was d ¼�.36 (range �.15 to �.80), suggest-
ing that the sex offender treatments produced an overall 22%
reduction in recidivism. This average effect size of the updated
sample of meta-analyses is 1.77 times bigger than the average
effect size of Craig et al.’s (2003) sample. The weakest effect
size was found in Pray’s (2002) dissertation that synthesized
the outcomes of psychological treatments (k ¼ 10; see Table
1). The strongest effect size was found in …

SPECIAL SECTION: SEXUAL HEALTH IN GAY AND
BISEXUAL MEN
Complexity of Childhood Sexual Abuse: Predictors of Current
Post-
TraumaticStressDisorder,MoodDisorders,SubstanceUse,andSexu
al
Risk Behavior Among Adult Men Who Have Sex with Men
Michael S. Boroughs1,2 • Sarah E. Valentine1,2 • Gail H.
Ironson3 • Jillian C. Shipherd4,5 •
Steven A. Safren1,2,6 • S. Wade Taylor6,7 • Sannisha K.
Dale1,2, • Joshua S. Baker6 •
Julianne G. Wilner1 • Conall O’Cleirigh1,2,6
Received: 11 August 2014/Revised: 7 April 2015/Accepted: 10
April 2015/Published online: 10 July 2015
� Springer Science+Business Media New York 2015
Abstract Men who have sex with men (MSM) are the group

mostatriskforHIVandrepresentthemajorityofnewinfections
intheUnitedStates.Ratesofchildhoodsexualabuse(CSA)among
MSM have been estimated as high as 46%. CSA is associated
with increased risk of HIV and greater likelihood of HIV sexual
risk behavior. The purpose of this study was to identify the
relationships between CSA complexity indicators and mental
health, substance use, sexually transmitted infections, and HIV
sexual risk among MSM. MSM with CSA histories (n=162)
whowerescreenedforanHIVpreventionefficacytrialcompleted
comprehensive psychosocial assessments. Five indicators
ofcomplexCSAexperienceswerecreated:CSAbyfamilymember,
CSA withpenetration,CSA withphysicalinjury,CSA withintense
fear,andfirstCSAinadolescence.Adjustedregressionmodelswere

used to identify relationships between CSA complexity and
outcomes.ParticipantsreportingCSAbyfamilymemberwere
at 2.6 odds of current alcohol use disorder (OR 2.64: CI
1.24–5.63), two times higher odds of substance use disorder
(OR 2.1: CI 1.02–2.36), and 2.7 times higher odds of reporting
anSTIinthepastyear(OR2.7:CI1.04–7.1).CSAwithpenetration
wasassociatedwithincreasedlikelihoodofcurrentPTSD(OR
3.17: CI 1.56–6.43), recent HIV sexual risk behavior (OR 2.7:
CI 1.16–6.36), and a greater number of casual sexual partners
(p= 0.02). Both CSA with Physical Injury (OR 4.05: CI 1.9–
8.7) and CSA with Intense Fear (OR 5.16: CI 2.5–10.7) were
related to increased odds for current PTSD. First CSA in ado-

lescencewasrelatedtoincreasedoddsofmajordepressivedis-
order.Thesefindings suggest thatCSA,with one ormorecom-
plexities,createspatternsofvulnerabilitiesforMSM,includingpost-
traumaticstressdisorder,substanceuse,andsexualrisktaking,
and suggests the need for detailed assessment of CSA and the
development of integrated HIV prevention programs that
address
mental health and substance use comorbidities.
Keywords Men who have sex with men (MSM) �
Childhoodsexualabuse(CSA)�PTSD�HIV�Sexualorientation
Introduction
Childhood Sexual Abuse: Mental Health and Sexual
Health Consequences
Intheextantliterature,childhoodsexualabuse(CSA)hasemerged

asanon-specificriskfactorforarangeofnegativehealthandmen-
talhealthsequelaeinadults.Forinstance,CSAhasbeenassociated
withmentalhealthproblemssuchasdepressionandpost-traumatic
stress disorder (PTSD), as well as substance use disorders
(e.g., Browne & Finkelhor, 1986; Maniglio, 2010; Neu-
mann, Houskamp, Pollock, & Briere, 1996; Suvak, Brogan,
& Shipherd, 2012). In addition to mental health and substance
abuseproblems,CSAhasbeenassociatedwithsexualriskbehav-
ior,sexualdysfunction,andinterpersonaldifficulties(i.e.,impaired
& Conall O’Cleirigh
[email protected]
1
Department of Psychiatry, Massachusetts General Hospital, One
Bowdoin Square, 7th Floor, Boston, MA 02114, USA

2
Department of Psychiatry, Harvard Medical School, Boston,
MA,
USA
3
Department of Psychology, University of Miami, Coral Gables,
FL, USA
4
National Center for PTSD –Women’s Health Sciences, Division,
VA Boston Healthcare System, Boston, MA, USA
5
Department of Psychiatry, Boston University School of
Medicine,
Boston, MA, USA
6
The Fenway Institute, Fenway Health, Boston, MA, USA

7
DepartmentofSocialWork,WheelockCollege,Boston,MA,USA
123
Arch Sex Behav (2015) 44:1891–1902
DOI 10.1007/s10508-015-0546-9
http://crossmark.crossref.org/dialog/?doi=10.1007/s10508-015-
0546-9&domain=pdf
http://crossmark.crossref.org/dialog/?doi=10.1007/s10508-015-
0546-9&domain=pdf
social cognitions, emotional lability, and poor interpersonal
relatedness) amongadults (e.g.,Neumann et al., 1996; Van
Bruggen,Runtz,&Kadlec,2006).Severalstudieshaverevealed
anassociationbetweenCSAandsexualriskvariablesincluding
unprotectedsex,sexwithmultiplepartners,andengaginginsex

trading among women (Arriola, Louden, Doldren, & Forten-
berry, 2005; Fargo, 2009; Gidycz, Coble, Latham, & Layman,
1993; Kaltman, Krupnick, Stockton, Hooper, & Green, 2005;
Suvak et al., 2012).
Childhood Sexual Abuse Among Gay, Bisexual,
and Other Sexual Minority Men
Although much of the extant literature has focused on the vic-
timizationofwomen,estimatesofCSAamonggayandbisexual
men reach as high as 47% (Arreola, Neilands, Pollack, Paul, &
Catania, 2008; Lenderking et al., 1997; Mimiaga et al., 2009;
O’Cleirigh, Safren, & Mayer, 2012). As a group, gay and
bisexual adults report more childhood psychological and phy-

sicalabusebyparentsandcaretakers(i.e.,familymembers)than
theirheterosexualsiblingsofthesamesex,andmoreCSA(Balsam,
Rothblum, & Beauchaine, 2005). In a study of young gay and
bisexual men (ages 15–22), 68% of the sample reported expe-
riencing verbal and physical violence victimization from family
members(Koblinetal.,2006).Agrowingbodyofresearchongay
and bisexual men’s health has revealed correlates of CSA that
parallel those first established among women. Specifically, gay
and bisexual men with CSA histories are more likely to expe-
riencenegativeemotional,cognitive,andinterpersonaloutcomes
as adults, including depression, suicidal ideation, substance
abuse,
andsexualrisk-takingbehaviorcomparedtogayand bisexual men

withoutCSAhistories(Bartholowetal.,1994;Brennan,Heller-
stedt, Ross,& Welles,2007; Kalichman,Gore-Felton, Benotsch,
Cage, & Rompa, 2004; Lloyd & Operario, 2012; Relf, 2001b;
Stalletal.,2003).Further,theseearlyexperiencesofvictimization
appear to put gay and bisexual men at increased risk for subse-
quent experiences of violence and abuse in adulthood, including
increasedriskofvictimizationintheiradultromanticrelationships
(Balsam, Lehavot, & Beadnell, 2011; Balsam et al., 2005;
Koblin
et al., 2006; Lalor & McElvaney, 2010).
Childhood Sexual Abuse in the Context of HIV Risk
and Prevention
Among gay, bisexual, and other men who have sex with men

(herein MSM for each of these groups), CSA history has been
consistently associated with increased risk for HIV acquisition
(Limetal.,2010;Lloyd&Operario,2012;Mimiagaetal.,2009;
O’Cleirigh et al., 2011; Stall et al., 2003). In addition, CSA has
been linked to a variety of sexual risk behaviors among MSM
including unprotected anal sex with a non-primary partner,
serodiscordant unprotected anal sex, sex with multiple partners,
and sex in exchange for money or drugs (Bartholow et al., 1994;
Brennan et al., 2007; Carballo-Diéguez & Dolezal, 1995;
Kalichman et al., 2004; Lenderking et al., 1997; O’Leary,
Purcell, Remien, & Gomez, 2003; Paul, Catania, Pollack, &
Stall,
2001;Relf,2001a;Stalletal.,2003).Theexperienceofviolencein

MSM’sadultromanticrelationshipshasalsobeenassociatedwith
unprotected sex and HIV acquisition (Merrill & Wolfe,
2000; Nieves-Rosa, Carballo-Dieguez, & Dolezal, 2000; Relf,
Huang,Campbell,&Catania,2004).Thus,MSMareatincreased
risk of HIV acquisition both in primary and non-primary sexual
relationships.Inprimaryrelationships,MSMwithCSAhistories
are morelikelytoreport feelingunsaferequestingthattheir abu-
sive partners use barrier protection (Heintz & Melendez, 2006).
Preliminary evidence from HIV prevention trials suggests that
sexual risk reduction interventions may be less effective for
MSMwhohaveCSAhistories(Crepazetal.,2006;Mimiaga
et al., 2009; Safren, Reisner, Herrick, Mimiaga, & Stall, 2010).
Thesefindingssuggesttheneedforbetterunderstandingofthose

constructs linking CSA to sexual risk that may be achieved by
more nuanced assessment of CSA.
Assessment of Childhood Sexual Abuse
The correlation between CSA and HIV risk is well established
among MSM, although exact mechanisms remain unclear. One
of the main limitations of the current literature is that the way
in
which CSA is operationalized (often as a binary indicator) dis-
counts the within-group heterogeneity of experiences. Defining
CSAinthiswaydilutesthelivedexperiencesofvictimsforwhom
CSAcanrepresentasingle-eventthatisincongruentwithinasur-
vivor’s context (‘‘an anomaly’’); or, CSA may represent just
one

event in the context of pervasive interpersonal abuse and
neglect
(‘‘the norm’’). In support of a more nuanced conceptualization
of
CSA, previous researchers have highlighted the importance of
abusecharacteristicssuchasduration,ageoffirstexperience,use
ofthreatorharm,andabuseinvolvingpenetration,inunderstand-
ingpost-traumaticadjustment,includingcopingstyleandriskfor
mental health and substance abuse problems (Cloitre & Rosen-
berg, 2006; Merrill, Guimond, Thomsen, & Milner, 2003).
Inaddition, recent researchon the nuance of definingand char-
acterizing CSA experiences among MSM suggests that there
may
be additional considerations when defining CSA for this popula-

tion. For instance, some researchers have taken a closer
examina-
tionoftheconsequencesofchildhoodsexualexperiencewitholder
partners (i.e., partners prior to the age of 13 who are at least
four
years older) among MSM (e.g., Arreola et al., 2008; Carballo-
Dieguez, Balan, Dolezal, & Mello, 2012). Carballo-Dieguez et
al.
only define the subset of these experiences, namely, experiences
where the child felt emotionally or physically hurt as a result of
CSA.Carballo-Dieguezetal.suggestthatnotenoughattentionhas
been paid to the perceptions of survivors of the events, such as
whether or not men choose to label these childhood sexual expe-
riencesasabuse(Carballo-Dieguez&Dolezal,1995;seealsoRind,

Tromovitch, & Bauserman, 1998).
1892 Arch Sex Behav (2015) 44:1891–1902
123
AfewstudieshavefoundthatonlyMSMwhoperceivedforce
orcoercionaspartoftheirchildhoodsexualexperiencesreported
poor adjustment, including depression and suicidal ideation
(Arreola et al., 2008; Stanley, Bartholomew, & Oram, 2004).
Importantlythough,MSMwhoreportedchildhoodsexualexpe-
rienceswitholderpartners(withandwithoutforce/coercion) were
more likely to engage in HIV sexual risk behaviors compared to
MSM without these experiences (Arreola et al., 2008). It is also
importanttonoteherethattheseauthorsrelyonadultretrospective

perceptions experiences from childhood, and do not adequately
acknowledgehoweasilytheseperceptionscanbedistortedbypost-
traumatic sequelae, such as guilt or denial (for detailed
summary
of this argument, see Dallam et al., 2001; Ondersma, Chaffin,
Berliners, Cordon, & Goodman, 1998). Althoughmostadults
who experienced CSA do not go on to have negative sequelae,
thisdoesnotmeanthatadult–childsexisnotharmfultochildren
(Dallam et al., 2001; Ondersma et al., 1998). Further, a recent
study on the labeling of CSA experiences, among HIV-positive
MSM, suggests that negative mental health sequelae are present
regardless of how the survivor labels the experience (Valentine
&Pantalone,2013).Despitewidedisagreementinthefield,these

findings highlight that it is important to distinguish between
forced/coercive sex and consensual sex when reporting findings
regarding childhood sexual experiences, and this is particularly
truewhendiscussingthechildhoodsexualexperiencesofMSM.
These nuances and characteristics are thought to represent
CSA complexities that warrant further study. Five dimensions,
orcomplexityindicators,wereinvestigatedinthisstudybecause
they may contribute to making the traumatic experience more
difficult given their association with greater disturbance and
impact upon functioning, and because they may predict distress
ordisturbanceintoadulthoodcomplicatingassessmentandtreat-
ment. Thus, we define complexity indicators as those character-

istics, supported by previous work, that influence negative
health
outcomes and complicate assessment and treatment of sexual
trauma for MSM.
Thereiscurrentlynogoldstandardforthemeasurementof
CSAcomplexity,although researchersagreethatfrequencyand
intensity of abuse, current functioning, and context of CSA
matters when attempting to characterize post-abuse adjustment
(Casey & Nurius, 2005; Kaysen, Rosen, Bowman, & Resick,
2010; Loeb, Gaines, Wyatt, Zhang, & Liu, 2011; Zink, Klesges,
Stevens, & Decker, 2009). Given the evidence demonstrated in
theliterature,webelievethattheCSAcomplexityissignificantly
influential in risk for impaired mental health, substance use,
and

sexualrisktaking.Theseoutcomesareofparticularinterestbecause
of their influence in the adult mental health and adult
adjustment
particularly among MSM with CSA histories. However, depres-
sion (Koblin et al., 2006; Mustanski, Newcomb, Du Bois,
Garcia,
& Grov, 2011; O’Cleirigh et al., 2013), PTSD (El-Bassel,
Gilbert,
Vinocur, Chang, & Wu, 2011; Ibañez, Purcell, Stall, Parsons, &
Gómez, 2005; Reisner, Mimiaga, Safren, & Mayer, 2009), and
substance use (e.g., Skeer et al., 2012) have each independently
been identified as predictors of sexual risk for HIV among MSM
regardless of CSA history.
The relationship between CSA complexity indicators, sexu-

allytransmittedinfections,andHIVsexualriskbehaviormayalso
helptospecifyaspectsoftheCSAexperiencethatserveaspoten-
tiatorsoftheproximalrisksforHIVinfectionamongMSM.Thus,
thecurrentstudyexaminedtherelationshipsbetweenempirically
derivedindicatorsofCSAcomplexity(i.e.,CSAbyafamily
member, CSA with penetration, CSA with physical injury, CSA
withintensefear,orfirstCSAinadolescence)andadultfunction-
ing,includingmentalhealth,substanceuse,andsexualrisktaking
withanexpectationthatthecomplexityofCSAwillimpactthese
outcomes among MSM.
Method
Participants

Datawerecollectedasapartofacomprehensiveassessmentfrom
amulti-siterandomizedclinicaltrialfromHIV-uninfectedMSM
(n = 162) that reported sexual risk and had a history of CSA
beforeage17.ThestudysiteswerelocatedinBoston,MA,and
Miami, FL. The average age was M = 39.4, SD= 11.8 (range
19–67).Thesamplewas66.1%EuroAmerican,22.6%African
American,3.6%Asian/PacificIslander,3.6%NativeAmerican,
with 27.8% identifying as Latino distributed across racial cate-
gories. Sexual orientation was assessed resulting in a sample
that
identified as 61 % gay, 27 % bisexual, 9 % unsure, and 3 %
heterosexual. The majority of the sample (81 %) experienced
multiple episodes of CSA before age 13, while 51% reported

experiencingsexualabusebetweenages13and17.Asignificant
minority (43%) of participants reported CSA across both age
ranges (see Table1).
Procedure
Recruitment
Recruitment was accomplished via outreach including at bars,
clubs, and cruising areas, community outreach, and advertising.
Recruitment for the study was done in conjunction with recruit-
ment for other, ongoing studies, and health promotion activities
todecreasestigmaandprotectindividualswhospokewithstudy
stafffrombeingidentifiedbyothersinthevenueassomeonewho
experienced sexual abuse in childhood.
Study Procedure

Following recruiting procedures, prospective participants were
screened by trained clinical staff via a structured questionnaire.
Arch Sex Behav (2015) 44:1891–1902 1893
123
Those who self-identified as HIV-negative were considered for
participation in the study, confirmed via rapid testing. All study
participantscompletedacomprehensivebaselineassessment
that included a thorough psychiatric evaluation, HIV and other
STItesting,andcomputer-basedpsychosocialassessments.Par-
ticipants responded to survey questions directly into a computer
because of the preponderance of studies that reveal that partici-

pants are more likely to disclosure sensitive information in this
manner (Des Jarlais et al., 1999; Metzger et al., 2000; Millstein,
1987;Navalineetal.,1994;O’Reilly,Hubbard,Lessler,Biemer,
& Turner, 1994; Turner et al., 1998; Wilson, Genco, & Yager,
1985).Inordertobeincludedinthestudy,participantshadto(1)
identify as a biological man who has sex with men age 18 or
older,(2)reportsexualcontactbeforetheageof13withanadult
oraperson5yearsolder,orsexualcontactbetweentheagesof13
and16inclusivewithaperson10yearsolder(oranyagewiththe
threatofforceorharm),(3)reportmorethanoneepisodeofunpro-
tected anal or vaginal intercourse within the past three months,
and (4) be HIV uninfected. Participants were excluded if all
episodesofunprotectedanalorvaginalintercourseoccurred

withonlyasingle,primary,HIV-negativepartner.Allprocedures
were IRB-approved.
Measures
Demographics
Theseincludedself-reportedage,race,ethnicity(independentof
racialcategory),income,relationship/maritalstatus,andedu-
cational attainment.
Assessment of Childhood Sexual Abuse
The parameters of CSA were assessed through a clinician-ad-
ministered interview adapted from previous work in HIV treat-
ment and prevention and used previously to assess sexual abuse
in a variety of medical populations (Leserman et al., 1997; Le-

serman, Li, Drossman, & Hu, 1998) including those HIV in-
fected(Lesserman,Ironson,&O’Cleirigh,2006).Theinterview
provided standardized questions that assessed sexual abuse
history comprised of 20 closed-ended questions predominantly
requiringyes/noanswers.CSAwasassessedacrosstwoageranges
0–12 years old and 13–16 years old. CSA is indicated in the
younger age range with any unwanted sexual contact report-
ed with someone 5 or more years older. In the older age range,
CSAwasindicatedifwithanysexualcontactreportedwithsome-
one 10years older or with some one of any age if there was the
threatofforceorharm.CSAwasindicatedifanyofthefollowing
occurred: genital touching, being touched, or penetrative inter-
course(i.e.,vaginaloranalpenetration).Thismeasureofunwanted

sexual contact was adapted from earlier research (Kilpatrick,
1992). All items on the measure asked about unwanted sexual
contact.Tomeetcriteriaforsexualabuse,theremustbeclearforce
or threat of harm for adolescents with a perpetrator less than 10
yearsolder;however,inchildren(13years),thethreatofforceor
harmisimpliedbya5-yearagedifferentialbetweenthevictimand
perpetrator.
CSA Complexity Indicators
Each of these CSA characteristics was coded dichotomously
indicating the presence or absence of the indicator.
Table1 Participant characteristics
Participant sample (N=162)
n %

Race Euro American 111 66.1
African American 38 22.6
Asian/Pacific Islander 6 3.6
Native American 6 3.6
Ethnicity Latino 45 27.8
Income $10,000 per year 50 30.2
[$60,000 per year 30 18.6
Educational attainment Some High School 10 6.2
High School Diploma 40 24.7
Some College 58 35.8
College Graduate 27 16.7
Some Graduate or above 27 16.7
Relational status Partnered 50 30.4

Single 112 69.6
Age M (SD) 39.4 (11.8)
1894 Arch Sex Behav (2015) 44:1891–1902
123
CSA by Family Member Participants were asked to identify
theirrelationshiptotheperpetrator(s),withapositivecodeinthis
category if the participant reported any CSA perpetrated by a
parent, stepparent, guardian, brother, other family member, or
other adult living in the family home.
CSA with Penetration was indicated if the participant repor-
tedthatpenetrativesexoccurredasdescribedaboveduringeither

age range.
CSA with Physical Injury was assessed via one question that
asked‘‘during any of the abuse experiences did you suffer ‘no
physicalinjuries,’‘minorphysicalinjuries’(scrapesandbruises),
or‘majorphysicalinjuries’(injuriesrequiringmedicalatten-
tion).’’CSAwithphysicalinjurywasindicatedifminorormajor
physical injury was reported.
CSA with Intense Fear was assessed through the question
‘‘Duringthe worst episode were youafraidthatyoumightbe
killed or seriously injured.’’
First CSA in Adolescence Participants’ CSA experiences
were assessed within two age ranges, one prior to their 13th
birthday and the other from age 13 through age 16. Partici-

pants who reported their first CSA experience during the
older age range were coded in this category.
Post-Traumatic Stress Symptom Assessment
Structured Clinical Interview for DSM-IV Axis I Disorders
(SCID-IV; Spitzer, Gibbon, & Williams, 1997)
Only the section on PTSD was used to provide an independent
assessment of current PTSD diagnosis and symptoms.
Sexual Risk Assessment
HIV sexual risk behavior was defined as insertive or receptive
anal or vaginal intercourse without a condom with any casual
partner or with any primary partner who had not specifically
disclosed that he/she was HIV uninfected and reported a recent

(past 3months) negative HIV test result. The number of HIV
sexual risk acts in the previous 3months as defined above was
summeddichotomizedatthemeantoreflecthighandlowsexual
risk. As recent sexual risk was one of the inclusion criteria in
order to enroll inthe study, thisconstruct lacksvariabilityinthat
no one reported zero risk episodes. The data were also heavily
skewed at the upper end of the range. To account for these
characteristics in the distribution, the distribution of sexual risk
behaviorwasdichotomizedatthemeantodistinguishthosewith
higher levels of recent sexual risk behaviors.
Sexually Transmitted Infections
As part of the self-report assessment, participants were asked if
theyhadbeendiagnosedwithanSTI inthe past 12months.This

generated a dichotomous variable.
Distress Assessment
The Mini-International Neuropsychiatric Interview (M.I.N.I.;
Sheehan et al., 1998)
TheMINIisashortstructureddiagnosticinterviewthathasgood
reliabilityandvaliditythatiscomparabletotheStructuredClinical
InterviewforDSM-IV(SCID-IV)(Sheehanetal.,1998).This
assessmentwascompletedwitheachparticipantbyatrainedInde-
pendentAssessoratthebaselineevaluationtoprovideinformation
on the presence of major mental illness (e.g., untreated severe
mood disorders, psychotic disorders), which is one of the exclu-
sioncriteria,andassistwithprovidingdiagnosisofothermoodor

substanceusedisorders.MajorDepressiveDisorderwasscoredas
present for anyone meeting diagnostic criteria for major depres-
siveepisodeatanytimeupto2weekspriortothebaselineassess-
ment.AnySubstanceUseDisorderwasscoredaspresentforthose
meeting diagnostic criteria for either substance abuse or depen-
denceacrossanyofthesubstancecategoriesinthepast12months.
Similarly,anyAlcoholUseDisorderwasscoredaspresentforeach
participant who met diagnostic criteria for either alcohol abuse
or
dependence in the past 12months.
Data Analysis
The demographics and background information provided in
Table 1 were generated through frequency counts, percent-

ages, and the calculation of means and standard deviations. The
interrelationships between the CSA complexity indicators were
examinedusingunadjustedlogisticregressions.Therelationships
between the CSA complexity indicators and the adult mental
health,substanceuse,andsexualhealthoutcomeswereestimated
using logistic regressions adjusted for age, race, education
level,
andtheabsenceorpresenceofadiagnosisofcurrentPTSD.Cur-
rent PTSD was included as a covariate to identify the magnitude
and significance of these relationships over and above what is
contributed by PTSD. The magnitude and significance of these
relationships are provided by the odds ratios and the associated
95% confidence interval. In one instance, the outcome variable

wascontinuous,i.e.,numberofcasualsexualpartners,andlinear
regressionsmodelswereusedwiththeidenticalcovariatesusedin
the logistic regression models. For the continuous outcome, the
tstatistic,degreesoffreedom,andthepvalueassociatedwiththe
CSA complexity predictor are reported. For the models predict-
ing current PTSD, PTSD was omitted from the list of covariates.
Results
Background Characteristics
Thetotalnumberofsexualpartnersintheprevious3-monthperiod
wasM=7.9,Median=5(range1–50),andtheHIVstatusofmale
Arch Sex Behav (2015) 44:1891–1902 1895
123

andfemalesexualpartnerswasoftenunknown.Themajorityofthe
sample reported male sexual partners exclusively (68.7%), fol-
lowed by both male and female partners (29.5%), and just 1.8%
reported female sexual partners exclusively over the previous
3-month period.
Examination of Outcome Data
Each of the outcomes of interest was descriptively examined.
Given the full sample, sexual risk behavior was M=7.52, SD=
12.43 suggesting an average of 7–8 partners in the past 3month
period. For the other outcomes interest, a sizable number of par-
ticipantshadcurrentPTSD(46%),anymooddisorder(40%),or
any alcohol use disorder (36 %). A smaller number of par-

ticipants reported an STI (17 %).
Interrelationships Between CSA Complexity Indicators
The strongest relationships were observed between CSA with
physical injury and CSA with penetration (OR 11.8: CI 4.4–
31.8) and between CSA with physical injury and CSA with
intensefear(OR9.4:CI4.3–20.5).FirstCSAinadolescencewas
significantly associated with increased odds of CSA with pen-
etration(OR4.1:CI2.1–8.3),CSAwithphysicalinjury(OR3.0:
CI 1.4–6.6), and CSA with intense fear (OR 2.3: CI 1.2–4.7).
Allbuttwooftheindicatorsweresignificantlyrelatedtoeach
other.CSAwithpenetrationwasnotsignificantlyrelatedtoCSA
by family member and neither was first CSA in adolescence
significantly related to CSA by family member. The complete

matrix of these interrelationships is presented in Table2.
Relationships between CSA Complexity Indicators
and Psychological and Health/Risk in Adulthood
Those reporting CSA with physical injury had more than four
times higher odds (OR 4.05: CI 1.90–8.70) tobe diagnosedwith
current PTSD than those who reported no physical injury. CSA
withinjurywasnotsignificantlyassociatedwithanyoftheother
outcomes under investigation (See Table3a, b for full results).
Similarly, CSA with penetration was significantly associated
with more than three times higher odds of being diagnosed with
current PTSD (OR 3.17: CI 1.56–6.43). CSA with penetration
wasalsoassociatedwithnearlythreetimeshigheroddsofreport-

ing very high levels unprotected anal intercourse in the past 3
months (OR 2.72: CI. 1.16–6.36) and with a higher number of
casual sexual partners in the past 3months.
ThosereportingCSAbyfamilymemberhad2.6timeshigher
odds(OR2.64:CI1.24–5.63)ofbeingdiagnosedwithanalcohol
usedisorderandmorethantwicetheodds(OR2.1:CI1.02–4.36)
of being diagnosed with a current substance use disorder. CSA
byfamilymemberwasnotsignificantlyassociatedwithincreased
risk of current mood disorder, current PTSD, or increased
sexual
riskforHIV.ThosereportingCSAwithphysicalinjuryhadnearly
threetimeshigheroddsinreportingasexuallytransmitteddisease
inthepastyear(OR2.7:CI1.04–7.10).ThosewhoreportedCSA

withintensefear(i.e.,fearofbeingkilledorseriouslyinjured)had
morethanfivetimeshigheroddsinmeetingdiagnosticcriteriafor
current PTSD than those who did not (OR 5.15: CI 2.5–10.7).
CSAwithintensefearwasnotsignificantlyassociatedwithanyof
the other adult outcomes. See Table3a, b for full results.
ThosewhoreportedfirstCSAinadolescencewerelesslikely
to meet criteria for major depressive disorder compared to those
who had first been abused during childhood. Despite its strong
relationshiptoallbutoneoftheotherCSAcomplexityindicators
first CSA in adolescence was not significantly related to any of
the other adult outcomes.
The reference group for each of these analyses is gay, bisex-
ual, other MSM with CSA histories, but who did not experience

each of the complexity indicators.
Discussion
This is the first study, of which we are aware, to link indices of
CSAcomplexitytoincreasedrisk for mental health,alcoholand
substance use disorders, and to increased risk for sexually trans-
mitted infections, and sexual risk for HIV, among adult MSM
overandabovewhatcanbeascribedtodiagnosticlevelsofPTSD.
Bothalcoholandothersubstanceusedisorderswerepredictedby
a history of CSA by family member. This category was also sig-
nificantly associated with a participant self-report of at least
one
sexuallytransmittedinfectioninthepastyear.Thus,therelational
Table2 Interrelationships between CSA complexity indicators

CSA complexity
indicators
% (n) CSA with injury CSA with
penetration
CSA by family
member
CSA with intense
fear
First CSA
in adolescence
CSA with physical injury 31.1 (52) – 11.8 (4.4–31.8) 2.0 (1.01–
3.9) 9.4 (4.3–20.5) 3.0 (1.4–6.6)
CSA with penetration 58.3 (98) – 1.4 (0.97–2.0) 6.1 (3.0–12.6)

4.1 (2.1–8.3)
CSA by family member 31.5 (53) – 1.95 (1.01–3.8) 0.6 (.30–
1.2)
CSA with intense fear 41.7 (70) – 2.3 (1.2–4.7)
First CSA in adolescence 61.3 (103) –
Expressed as unadjusted Odds Ratio (95% Confidence Interval)
Odds ratios that are significant at p.05 or less are indicated in
bold
1896 Arch Sex Behav …
Sexual Abuse: A Journal of
Research and Treatment
2016, Vol. 28(4) 340 –359
© The Author(s) 2014
Reprints and permissions:

DOI: 10.1177/1079063214535819
sax.sagepub.com
Article
Adverse Childhood
Experiences in the Lives
of Male Sex Offenders:
Implications for Trauma-
Informed Care
Jill S. Levenson1, Gwenda M. Willis2,
and David S. Prescott3
Abstract
This study explored the prevalence of childhood trauma in a
sample of male sexual
offenders (N = 679) using the Adverse Childhood Experience
(ACE) scale. Compared
with males in the general population, sex offenders had more
than 3 times the odds
of child sexual abuse (CSA), nearly twice the odds of physical
abuse, 13 times the

odds of verbal abuse, and more than 4 times the odds of
emotional neglect and
coming from a broken home. Less than 16% endorsed zero
ACEs and nearly half
endorsed four or more. Multiple maltreatments often co-
occurred with other types
of household dysfunction, suggesting that many sex offenders
were raised within
a disordered social environment. Higher ACE scores were
associated with higher
risk scores. By enhancing our understanding of the frequency
and correlates of
early adverse experiences, we can better devise trauma-
informed interventions that
respond to the clinical needs of sex offender clients.
Keywords
adverse childhood experiences, sexual offender treatment,
trauma-informed care
1Barry University School of Social Work, Miami Shores, FL,
USA
2The University of Auckland, New Zealand
3Becket Family of Services, Falmouth, ME, USA

Jill Levenson, PhD, LCSW, Associate Professor, Barry
University School of Social Work, 11300 NE, 2nd
Ave, Miami Shores, FL 33161 USA.
535819 SAXXXX10.1177/1079063214535819Sexual
AbuseLevenson et al.
research-article2014
14535819&domain=pdf&date_stamp=2014-05-28
Levenson et al. 341
Over the past few decades, researchers have established that the
prevalence of early
traumatic experiences such as child maltreatment and family
dysfunction is far greater
than previously recognized (Centers for Disease Control and
Prevention [CDC],
2013b). Multiple types of adversity are often present and
research has demonstrated
that cumulative experiences of childhood trauma lead to
alarming increases in the risk

for a range of health and social problems (Anda, Butchart,
Felitti, & Brown, 2010;
Felitti, 2002; Felitti et al., 1998). Emerging evidence also
suggests that early traumatic
experiences are common in the lives of sexual offenders
(Jespersen, Lalumière, &
Seto, 2009; Reavis, Looman, Franco, & Rojas, 2013). A clear
understanding of the
scope and impact of early adversity is important in the
development of treatment inter-
ventions and social policy (Anda et al., 2010; Anda et al., 2006;
Felitti et al., 1998).
Trauma, by definition, is any extraordinary event (experienced
or witnessed) that
threatens an individual’s physical or psychological well-being
and challenges his or
her coping skills (American Psychiatric Association, 2000,
2013; Whitfield, 1998).
The Adverse Childhood Experiences (ACE) study, a
collaborative research project
between the U.S. CDC and Kaiser Permanente (a network of
health care organiza-
tions), produced staggering evidence of the pervasive and
enduring nature of early

trauma (CDC, 2013b). Beginning in 1997, the ACE study
collected data about child-
hood adversity and its relationship to adult health outcomes
from 17,337 participants
who sought health services from Kaiser Permanente (Felitti et
al., 1998).
Notwithstanding an underrepresentation of ethnic minorities and
lower socioeconomic
classes, the results of this project were remarkable for their
revelation of the frequency
and negative correlates of child maltreatment and household
dysfunction. More than
28% of the participants reported childhood physical abuse, 11%
were emotionally
abused, and 21% had been sexually abused. Women were more
likely to report sexual
(25%) and emotional (13%) abuse than men (16% and 8%,
respectively), and men
were slightly more likely to have been physically abused.
Nearly one quarter of the
respondents had been physically or emotionally neglected.
Household dysfunction
was also common; 13% had witnessed domestic violence in the
home, 27% experi-
enced parental substance abuse, 19% had a parent who was

depressed, mentally ill, or
attempted suicide, and 23% came from homes in which the
parents were separated or
divorced. Nearly 5% reported that a family member had gone to
prison (CDC, 2013b).
More than two thirds of the participants reported experiencing
at least one adverse
event before they turned 18 years (CDC, 2013b). Multiple forms
of child maltreatment
and household dysfunction were interrelated; the presence of a
single ACE factor more
than doubled the odds of reporting additional ACEs (Dong,
Anda, Dube, Giles, &
Felitti, 2003; Dong et al., 2004). As the number of childhood
adverse experiences
increases, the risk for myriad health, mental health, and
behavioral problems in adult-
hood also grows in a cumulative fashion (Anda et al., 2006;
Dube, Anda, Felitti,
Edwards, & Williamson, 2002; Felitti, 2002; Felitti et al.,
1998). For instance, as ACE
scores increase, so does the likelihood of adulthood substance
abuse, suicide attempts,
depression, smoking, heart and pulmonary diseases, fetal death,

obesity, liver disease,
intimate partner violence, early initiation of sexual activity,
promiscuity, sexually
transmitted diseases, and unintended pregnancies (CDC, 2013a;
342 Sexual Abuse 28(4)
ACE research has clearly and consistently demonstrated the
negative impact of early
trauma on behavioral, medical, and social well-being in
adulthood (Anda et al., 2010;
ACEs and Criminal Offenders
A history of child abuse is common among criminal offenders.
Prevalence rates can
vary depending on how child abuse is defined in an interview or
survey, and male
prisoners in particular may underreport child abuse due to
normalized perceptions of
victimizing behavior or fears of appearing vulnerable. Several

studies have reported
higher rates of physical and sexual abuse in inmates compared
with the general popu-
lation (Courtney & Maschi, 2013; Harlow, 1999; Maschi,
Gibson, Zgoba, & Morgen,
2011; Weeks & Widom, 1998). Household dysfunction is also
common among inmates
and often co-occurs with child maltreatment. Prisoners
frequently report witnessing
violence in childhood and many experienced the death of a
family member, parental
separation or abandonment, or parental substance abuse
(Courtney & Maschi, 2013;
Haugebrook, Zgoba, Maschi, Morgen, & Brown, 2010; Maschi
et al., 2011). Harlow
(1999) found that approximately 40% of prisoners reported out-
of-home foster care
placement in childhood and many had an incarcerated family
member. Abused prison-
ers were more likely than nonabused prisoners to be serving a
sentence for a homicide,
violent offense, or sexual crime (Harlow, 1999).
A study of adverse childhood events among more than 700
California inmates

using a scale very similar to the ACE survey revealed that 28%
were emotionally or
physically neglected and 45% were physically or sexually
abused (Messina, Grella,
Burdon, & Prendergast, 2007). Household dysfunction was also
common, with nearly
half reporting domestic violence in their childhood homes, 43%
reporting parental
separation, 37% having an incarcerated family member, 14%
experiencing placement
in foster care, and half stating that a parent abused substances.
Only 13% of the total
sample reported zero adverse events, while approximately 30%
reported four or more.
There were strong correlations between nearly all categories.
Collectively, research
findings reviewed demonstrate that childhood adversity is
associated with adult crimi-
nality, particularly interpersonal violence, and that greater
exposure to adverse events
significantly increases the likelihood of mental health problems
and serious involve-
ment in drugs and crime (Harlow, 1999; Messina et al., 2007).
ACEs and Sexual Offenders

Although it has been commonly hypothesized that most sexual
offenders are former
victims, studies have varied widely in their findings of the
prevalence of early moles-
tation among sexual perpetrators. An early survey found that
63% of incarcerated sex
offenders reported being sexually abused as children or being
pressured into sexual
activity by an adult (Groth, 1979). A subsequent meta-analysis
of empirical studies
containing a total of 1,717 subjects found that 28% of sex
offenders reported a history
of childhood sexual abuse (Hanson & Slater, 1988). This figure
is substantially greater
Levenson et al. 343
than the 16% to 17% rate of sexual victimization of males in the
general population
(CDC, 2013b; Hunter, 1990). Hindman (1988) offered
surprising findings when she
polygraphed 129 sex offenders in treatment about their reported

sexual histories. The
results showed that although 67% of offenders initially reported
being sexually abused
as children, when polygraphed the number dropped to 29%,
suggesting that some men
may fabricate or exaggerate early childhood trauma in an
attempt to rationalize their
behavior or gain sympathy from therapists (Hindman, 1988;
Hindman & Peters, 2001).
Studies using multiple methodologies have found higher
prevalence rates among sex-
ual offenders, and how a researcher asks relevant questions
(e.g., the use of emotion-
ally laden terms such as abuse) can influence results (Simons,
2007).
In a study administering the ACE questionnaire to child
abusers, domestic violence
offenders, sex offenders, and stalkers (n = 151), it was found
that these offenders as a
group had significantly higher rates of ACEs than men in the
general population
(Reavis et al., 2013). Only 9.3% of the sample reported no
adverse events in child-
hood, compared with 38% of the male sample in the ACE study.

As well, 48% reported
four or more adverse experiences, compared with 9% of the men
in the ACE study.
Sex offenders in particular had significantly higher ACE scores
than the general popu-
lation (Reavis et al., 2013). Weeks and Widom (1998) also
found higher rates of mal-
treatments in male sex offenders, with 26% reporting sexual
abuse in childhood, 18%
reporting neglect, and two thirds revealing childhood physical
abuse.
A meta-analysis of 17 studies compared rates of sexual and
other forms of abuse
reported in a combined sample of 1,037 sex offenders and 1,762
non–sex offenders
(Jespersen et al., 2009). The authors also analyzed the
prevalence of different forms of
abuse in 15 studies that compared sex offenders who assaulted
adults (n = 962) with
those with child victims (n = 1,334). Most of the studies
revealed that sexual abuse,
physical abuse, and neglect were common among sex offenders.
Sex offenders were
more than 3 times more likely to have been sexually abused

than non–sex offenders
but not more likely to have been physically abused. Sex
offenders against children
were more likely to have been sexually abused but those who
assaulted adults were
more likely to have experienced physical abuse in childhood.
The neurodevelopmental pathway from childhood adversity to
adult behavior is an
enormously complex biopsychosocial process. Environmental
stressors stimulate the
overproduction of stress-related hormones associated with fight-
or-flight responses,
inhibiting the growth and connection of neurons and
contributing to lasting effects
such as affective dysregulation, deficits in social attachment,
and cognitive problems
(Anda et al., 2010; Anda et al., 2006; Creeden, 2009). These
social, emotional, and
cognitive impairments often result in adoption of high-risk
behaviors as coping strate-
gies to relieve distress, culminating, for many people, in the
development of illnesses,
disabilities, psychosocial problems, and premature mortality at
rates higher than in the

general population (Felitti et al., 1998).
In summary, early childhood maltreatment and family
dysfunction are common in
the general population. Adverse experiences are associated with
poorer health, mental
health, and behavioral outcomes, and cumulative trauma
dramatically increases the
odds of medical and psychosocial problems as well as
addictions (Anda et al., 2006;
Dong et al., 2003; Dong et al., 2004; Dube et al., 2005; Felitti
et al., 1998). Criminal
populations, including sexual offenders, are even more likely
than the general popula-
tion to have a history of early trauma. Reavis et al. (2013)
opined that given the preva-
lence of early maltreatment in the histories of sex offenders, it
is perhaps unsurprising
that offense-specific models of sex offender treatment have
produced mixed results in

terms of effectiveness. They suggested that treatment programs
should more strongly
emphasize the role of early trauma in self-regulation and
attachment. It is important to
understand the frequency and role of these early experiences in
the development of
sexual offending and to use that knowledge to inform treatment
protocols.
Purpose of the Current Study
The purpose of this study was to explore the prevalence of
ACEs in a large sample of
male sexual offenders and to compare findings with rates of the
same experiences for
males in the general population. It was hypothesized that the
sex offenders would have
higher rates of early adverse experiences than males in the
general population. The
study also sought to explore differences in ACE scores between
different types of
sexual offenders and to examine ACE scores in relation to
recidivism risk. By enhanc-
ing our understanding of the frequency and correlates of child
maltreatment and

household dysfunction, we can better devise clinical
interventions that respond to the
needs of sex offender clients.
Method
Participants
A nonrandom sample of participants was surveyed in civil
commitment (28%) and
outpatient (72%) sex offender treatment programs across the
United States. The pro-
grams were recruited through a solicitation on the professional
listserv of the
Association for the Treatment of Sexual Abusers. Therapists
who responded to the
solicitation agreed to become data collection sites, and they in
turn invited their clients
to participate in the survey. Most outpatient programs serve
clients who have been
ordered to attend treatment by the court as part of their
probation requirements follow-
ing a criminal conviction or as part of their Family Court case
plan following a finding
of sexual abuse in a child protective services investigation.

Participating programs
included sex offenders from New Jersey, Illinois, Texas,
Florida, Georgia, Maryland,
Montana, Washington, and Maine. All clients attending
treatment at the outpatient or
inpatient facilities (n = approximately 970) were invited to
participate in the project,
and a total of 709 clients voluntarily agreed to participate.
Thus, the response rate was
approximately 73%.
The sample for the current study consisted of 679 adult male
sex offenders.
Although females participated in the study, they were excluded
from these analyses
and those data will be reported elsewhere. Sample demographics
are described in
Table 1. The majority of participants were White (67%) and
most (71%) were between
Levenson et al. 345
30 and 60 years of age, with 20% younger than age 30 (7% were

18-25) and 9.6%
older than age 60. Approximately 62% of the sample had
completed high school or
general equivalency diploma (GED), and 19.6% identified
themselves as college grad-
uates. About 59% earned less than $30,000 per year in the last
year they earned income.
Nearly half of the sample had never been married, 16% were
currently married, and
34% were divorced or separated.
Table 2 describes participant, offense, and victim
characteristics. Participants had
been arrested for a variety of sexual crimes; two thirds reported
that their index offense
involved sexual contact with a minor, and 9% reported sexual
assault of an adult.
About 9% said they had been arrested for a child pornography
offense, 7% for Internet
solicitation, 3% for exposure of genitals, and less than 1% for
voyeurism. Participants
were asked a series of questions about victim characteristics,
taking into account their
index offending, any prior offending, and any undetected
offending. Most participants

reported that they had offended against female victims, about
one third reported that
they had victimized strangers, and more than half said they
offended against prepubes-
cent children (percentages do not add up to 100% because some
endorsed multiple
Table 1. Sample Demographics.
Demographic categories % (N = 679)
Race
White 67
Minority 32
Age (years)
18-30 20
31-40 21
41-50 30
51-60 20
Older than 60 9
Marital status
Never married 47
Married 16
Divorced/separated 34
Widowed 3

Education
Not high school graduate 18
High school graduate or GED 63
College graduate or higher 19
Income
Less than $20,000 42
$20,000-$29,999 17
$30,000-$49,999 20
$50,000+ 21
Note. GED = general equivalency diploma.
categories). It should be noted that although most sex offenses
involve perpetrators
and victims who are known to each other (Bureau of Justice
Statistics, 1997, 2010),
28% of this sample was civilly committed and was more likely
to have a stranger vic-
tim. When asked whether they had ever had a stranger victim,
62% of the civilly com-
mitted offenders endorsed “yes” compared with 25% of the

outpatients. Most
participants (69%) reported that they had been arrested once for
a sex crime, 19%
twice, and approximately 12% reported three or more sex crime
arrests. Consistent
with statutory language used to determine whether a person
meets criteria for civil
commitment, civilly committed sex offenders had a higher mean
number of sex crime
arrests (2.3, SD = 1.5) than outpatients (1.2, SD = .79). The
median length of time in
treatment was 30 months (mode = 24, M = 50, SD = 53).
Participants were asked to disclose their total number of victims
(including offenses
they had not been arrested for), and they reported a median
number of two victims
(mode = 1, M = 20, SD = 172). One participant reported more
than 3,000 victims and
2 participants reported more than 1,000 victims, whereas 82%
reported 10 victims or
less and 67% reported 3 or less. Because outliers can skew
measures of central ten-
dency, the 5% trimmed mean number of victims was calculated
(excluding the 5%

highest and lowest values), and was found to be six. It should
be noted that noncontact
Table 2. Offender, Offense, and Victim Characteristics.
Valid n M/%
Female victim 681 77%
Male victim 676 28%
Family victim 677 40%
Unrelated victim 677 48%
Stranger victim 681 35%
Victim younger than 12 years 683 52%
Teen victim 675 56%
Adult victim 673 29%
Total sex crime arrests 684 1.58
Total victims 636 20.32a
Ever used force 682 23%
Ever used weapon 689 9%
Ever caused injury 687 9%
Total non–sex arrests 685 1.50
Months in Tx 645 50.09
On probation 666 61%
Months on probation 400 45.21

Lifetime months in prison 670 85.25
Lifetime months on probation 637 47.31
Note. Percentages may not add up to 100% because some
categories were not mutually exclusive.
aThe average number of victims was skewed due to a few high-
value outliers. Median number of
victims = 2 and mode = 1;Tx=Treatment.
Levenson et al. 347
offenders such as exhibitionists were included in the sample,
perhaps accounting for
some of the outlying cases. Exhibitionism is known to be highly
compulsive and repet-
itive and some men have engaged in the behavior thousands of
times (McGrath, 1991;
Morin & Levenson, 2008).
Instrumentation
A survey was developed by the principal investigator for the
purpose of collecting data

on the prevalence of early trauma. The first section of the
survey consisted of the ACE
scale (CDC, 2013b), a 10-item dichotomous (yes/no) scale in
which participants
endorse certain experiences prior to 18 years of age: abuse
(emotional, physical, and
sexual), neglect (emotional and physical), and household
dysfunction (domestic vio-
lence, unmarried parents, and the presence of a substance-
abusing, mentally ill, or
incarcerated member of the household). One’s ACE score
reflects the total number of
adverse experiences endorsed by that individual. The ACE
categories were developed
using items adapted from earlier studies: the Conflict Tactics
Scale (Straus, Gelles, &
Smith, 1990), the Child Trauma Questionnaire (Bernstein et al.,
1994), and questions
from a survey about sexual abuse (Wyatt, 1985).
The second section of the survey asked questions about offense
history using
forced-choice categorical responses to ensure anonymity.
Questions about the nature
of the sex offenses committed were asked, such as victim age,

gender, and relation-
ship, as well as the number of prior arrests. No information that
could potentially
identify offenders or victims was sought.
Data Collection
Federal guidelines for human subject protection were followed
and the project was
approved by an Institutional Review Board. Clients were invited
to complete the anon-
ymous survey during regularly scheduled group therapy sessions
at participating data
collection sites. Clients were instructed not to write their names
on the survey, and to
place the completed survey in a sealed box with a slot opening.
Informed consent was
provided in writing and explained verbally, however, to protect
anonymity, partici-
pants were not required to sign a consent document. Completion
of the survey was
considered to imply informed consent to participate in the
project.
Analyses

Descriptive statistics are reported for each of the survey items.
Binomial analyses, t
tests, and odds ratios (OR) were used to examine differences
between groups, and
bivariate correlations were used to examine relationships
between variables.
Results
Figure 1 depicts the proportion of participants endorsing “yes”
to each ACE item.
Child maltreatment and household dysfunction were common,
with more than half
of the participants endorsing verbal abuse and parental
separation or divorce (53%
and 54%, respectively), nearly half reporting household
substance abuse (47%),
and greater than one third of participants endorsing childhood
physical abuse

(42%), sexual abuse (38%), and emotional neglect (38%).
Figure 2 shows the dis-
tribution of ACE scores. Slightly less than 16% said that they
experienced no ACEs
and nearly half endorsed four or more. The mean ACE score
was 3.5 (median = 3,
SD = 2.74).
Table 3 shows each ACE item exactly how it was presented to
participants, as well
as the proportion endorsing each item compared with the
prevalence in the original
CDC male sample. In each category, the sex offenders reported
higher prevalence rates
than the general male population, and binomial tests revealed
that all differences were
statistically significant (p < .001).
ORs are used to compare the relative odds of the occurrence of
an event (e.g., child-
hood sexual abuse) in one group with the odds of occurrence of
the same event in
another group (Szumilas, 2010). ORs in the current analysis
were calculated as
described in the following cogent example:

. . . If 25 out of 100 sex offenders have a history of sexual
abuse, their odds of having a
sexual abuse history are 25/75, or 0.33; if 10 of 100 of non-sex
offenders have a similar
history, their odds are 10/90, or 0.11. The OR for this
comparison is thus 0.33/0.11, or 3.0.
An odds ratio of 1.0 represents the absence of a group
difference whereas an odds ratio
0%
10%
20%
30%
40%
50%
60% 53%
42%

38% 38%
16%
54%
24%
47%
26%
23%
Figure 1. Percentage of male sex offenders endorsing ACE
items (N = 679).
Note. ACE = Adverse Childhood Experience; DV = domestic
violence.
Levenson et al. 349
greater than 1.0 means a greater prevalence of abuse in the first
group; an odds ratio
smaller than 1.0 means a lower prevalence of abuse in the first

group. (Jespersen et al.,
2009, p. 182)
In the current analysis, results revealed that sex offenders were
more likely to expe-
rience all ACE items compared with males in the general
population (see Table 3).
As shown in Table 4, correlations between ACE items were all
positive and signifi-
cant, suggesting that child maltreatment occured in household
environments in which
a variety of dysfunctions were often present. The correlation
between verbal abuse and
physical abuse, r = .67, corresponded to a large effect size
(Cohen, 1988). Correlations
demonstrating a medium effect size included domestic violence
and physical child
abuse, r = .41, emotional neglect and verbal abuse, r = .41, and
emotional neglect and
physical abuse, r = .42.
Higher ACE scores were significantly correlated with lower
educational attain-
ment, r = −.26; p < .01, lower income, r = −.25; p < .01, and

more arrests for nonsexual
offenses, r = .29; p < .01. ACE scores had no significant
correlation with the number
of sex crime arrests or the number of total victims. Those with
victims younger than
12 years of age had significantly higher mean ACE scores than
those with older vic-
tims, 4.2 versus 2.9; t = −6.133, p < .001. Higher mean ACE
scores were also found in
the groups of sex offenders who said that they had used force or
violence in the com-
mission of a sex offense, 4.9 versus 3.2; t = −7.043, p < .001,
used a weapon in a sex
crime, 5.3 versus 3.4; t = −4.863, p < .001, or who injured a
victim in a sex crime, 5.4
versus 3.4; t = −5.435, p < .001. Higher mean ACE scores were
found for sex offenders
with contact sex offenses versus noncontact sex offenses, 3.4
versus 2.2; t = 4.069,
0%
5%
10%

15%
20%
25%
30%
35%
40%
45%
50%
0 1 2 3 4+
15.6%
13.7% 12.8% 12.3%
45.7%
Figure 2. Distribution of ACE scores (N = 679).

Note. ACE = Adverse Childhood Experience.
p < .01. No significant differences were found in ACE scores
between those with only
adult victims versus those with at least one minor victim, or for
those with only extra-
familial victims versus those with at least one family victim.
Table 3. ACE Item Comparisons Between Sex Offenders and
Males in CDC Sample.
ACE questions: While you were growing up, in your
first 18 years of life . . .
Sex
offenders
Male CDC
sample
Odds

ratio (N = 679) (n = 7,970)
1. Did a parent or other adult in the household often
or very often swear at you, insult you, put you
down, or humiliate you? Or, act in any way that
made you afraid that you might be physically hurt?
53.3%*** 7.6% 13.88
2. Did a parent or other adult in the household often
or very often push, grab, slap, or throw something
at you? Or, ever hit you so hard that you had marks
or were injured?
42.2%*** 29.9% 1.71
3. Did an adult or person at least 5 years older than
you ever touch or fondle you or have you touch his
or her body in a sexual way? Or, attempt or actually
have oral, anal, or vaginal intercourse with you?
38%*** 16% 3.22
4. Did you often or very often feel that no one in your
family loved you or thought you were important or

special? Or, your family did not look out for each
other, feel close to each other, or support each
other?
37.6%*** 12.4% 4.26
5. Did you often or very often feel that you did not
have enough to eat, had to wear dirty clothes, and
had no one to protect you? Or, your parents were
too drunk or high to take care of you or take you
to the doctor if you needed it?
15.9%*** 10.7% 1.58
6. Were your parents ever separated or divorced? 54.3%***
21.8% 4.26
7. Was your mother or stepmother often or very
often pushed, grabbed, slapped, or had something
thrown at her? Or, sometimes often or very often
kicked, bitten, hit with a fist, or hit with something
hard? Or, ever repeatedly hit at least a few minutes
or threatened with a gun or knife?
24%*** 11.5% 2.43

8. Did you live with anyone who was a problem
drinker or alcoholic or who used street drugs?
46.7%*** 23.8% 2.81
9. Was a household member depressed or mentally ill,
or did a household member attempt suicide?
25.9%*** 14.8% 2.01
10. Did a household member go to prison? 22.6%*** 4.1% 6.83
Note. ACE = Adverse Childhood Experience; CDC = Centers
for Disease Control and Prevention.
***Frequencies endorsed by the sex offenders were compared
with those observed in the CDC male
sample using binomial nonparametric tests and all showed
significant differences between groups
(p < .001). SPSS does not produce coefficients for one-sample
binomial tests.
Levenson et al. 351

Finally, a simulated risk score was devised for each offender by
tabulating the num-
ber of risk factors known to be associated with sexual
recidivism and found in the
Static-99R, the most well-researched and commonly used risk
assessment instrument
in North America (Hanson & Morton-Bourgon, 2005; Hanson &
Thornton, 1999,
2000; Helmus, Thornton, Hanson, & Babchishin, 2012). Age
was coded by the follow-
ing categories: 18 to 25 = 1, 26 to 40 = 0, > 40 = −1 (due to the
way data were col-
lected, categorical breakdowns were similar but did not
precisely correspond to those
in the Static-99R; Helmus et al., 2012). The remaining risk
factors were coded as 1 =
yes and 0 = no: unmarried (never married), …
International Journal of
Offender Therapy and
Comparative Criminology

2016, Vol. 60(4) 371 –396
© The Author(s) 2014
Reprints and permissions:
DOI: 10.1177/0306624X14553227
ijo.sagepub.com
Article
“They Treat Us Like Human
Beings”—Experiencing a
Therapeutic Sex Offenders
Prison: Impact on Prisoners
and Staff and Implications for
Treatment
Nicholas Blagden1, Belinda Winder1,
and Charlie Hames1
Abstract
Research evidence demonstrates that sex offender treatment
programmes (SOTPs)

can reduce the number of sex offenders who are reconvicted.
However, there
has been much less empirical research exploring the experiences
and perspectives
of the prison environment within which treatment takes place.
This is important,
particularly for sexual offenders, as they often face multiple
stigmas in prison. This
study used a mixed-methods approach to explore the
experiences of prisoners and
staff at a therapeutically orientated sexual offenders’ prison to
understand whether
the prison environment was conducive to rehabilitation. The
quantitative strand of
the research sampled prisoners (n = 112) and staff (n = 48) from
a therapeutically
orientated sex offenders prison. This strand highlighted that
both prisoners and staff
had positive attitudes toward offenders and high beliefs that
offenders could change.
Importantly, the climate was rated positively and, in particular,
participants had very
high ratings of “experienced safety.” The qualitative strand of
the research consisted
of semistructured interviews with prisoners (n = 15) and a range

of prison staff (n =
16). The qualitative analysis revealed positive prisoner views
toward staff relationships,
with most participants articulating that the prison and its staff
had contributed to
positive change in prisoners. Crucially, the environment was
perceived as safe and
allowed prisoners “headspace” to work through problems and
contemplate change.
1Nottingham Trent University, UK
Nicholas Blagden, Sexual Offences Crime and Misconduct Unit,
Division of Psychology, Nottingham
Trent University, Chaucer Building, Burton Street, Nottingham
NG1 4BU, UK.
553227 IJOXXX10.1177/0306624X14553227International
Journal of Offender Therapy and Comparative
CriminologyBlagden et al.
research-article2014
mailto:[email protected]

X14553227&domain=pdf&date_stamp=2014-10-09
372 International Journal of Offender Therapy and Comparative
Criminology 60(4)
This research offers some support to the notion that context is
important for sex
offender rehabilitation.
Keywords
sex offender rehabilitation, rehabilitative climate, therapeutic
climate, mixed methods
Introduction
Research has demonstrated that sex offender treatment
programmes (SOTPs) can
reduce the number of sex offenders who are reconvicted
(Hanson et al., 2002; Losel &
Schmucker, 2005). Specifically, programmes that take a risk–
need–responsivity
approach have been found to be the most successful (Hanson,
Morton, Helmus, &
Hodgson, 2009). However, although there is now an established

body of knowledge
regarding sex offender treatment effectiveness, there has yet to
be any significant
attention paid to the environment/context in which treatment
takes place (Ware, Frost,
& Hoy, 2010). Research on the broader environment is limited,
primarily focusing on
“within treatment” climate (see Beech & Hamilton-Giachritsis,
2005; Beech & Scott
Fordham, 1997). For example, findings from Beech and
Hamilton-Giachritsis (2005)
affirmed that a therapeutic climate was related to treatment
outcome, with therapists’
attitudes and goals having an impact on treatment effectiveness.
However, there has been no systematic empirical research
exploring the climate of
the prison where such treatment takes place. Indeed, Woessner
and Schwedler (2014)
asserted that “few researchers have ventured to question
whether therapeutic prisons
actually provide a therapeutic climate” (p. 4). This is surprising,
given the amount of
research that has found that social and therapeutic climate
influences a variety of clini-

cal and organisational outcomes related to staff and patients in
forensic mental health
services (Tonkin et al., 2012; Willets, Mooney, & Blagden,
2014). It is also potentially
important as the broader prison environment could either
facilitate or interfere with
treatment intervention. This is particularly relevant for sexual
offenders as they face
multiple stigmas in prison, occupy the lower rungs of the prison
hierarchy, and experi-
ence hostility and anxiety on a daily basis (Schwaebe, 2005).
For these reasons,
Schwaebe (2005) stressed the need to understand the context of
sexual offenders in
prison and the context of their treatment to understand the
limits of treatment gain in
prison-based programmes. This broader environment is typically
overlooked, despite
research finding that status in prison is a factor in sex offender
treatment refusal
(Mann, Webster, Wakeling, & Keylock, 2013). A prison’s social
environment has been
found to be important for shaping behaviour and is central to
the extent to which treat-
ment gains are sustained and generalised (Ward, Day, Howells,

& Birgden, 2004). If
sexual offenders find themselves in prisons that are
characterised by suspicion, hostil-
ity, and guardedness, this will impair treatment outcome and
may make prisoners less
likely to volunteer for programmes (Ward et al., 2004).
Antitherapeutic prison envi-
ronments have been found to negatively affect on treatment
readiness and programme
outcome (Schalast, Redies, Collins, Stacey, & Howells, 2008;
Ward et al., 2004).
Blagden et al. 373
Thus, the prison climate, whether therapeutic (or not), and the
attitudes of staff (thera-
pists, prison officers, and general staff) within the prison could
play a pivotal role in
the successful treatment and rehabilitation of offenders.
In many jurisdictions, sexual offenders are often isolated for
their own protection
due to the dangers they face. In England and Wales, this often

means segregation onto
“vulnerable prisoner units” (VPUs) or transfer to prisons that
deal predominantly with
sex offenders. However, even in specialised units, sex offenders
still experience threats
and fear from other prisoners and, at times, staff (see, for
example, O’Donnell &
Edgar, 1998). There is a clear international gap in the literature
regarding sex offend-
ers’ experiences of prison climate/environment. This is
important as there are growing
concerns that rehabilitative programmes and practice are being
compromised by inef-
fective correctional environments, staff drift, organisational
resistance, degree to
which therapeutic integrity is maintained, and the quality of
programme implementa-
tion (Day, Casey, Vess, & Huisy, 2012; Smith, Cullen, &
Latessa, 2009).
Evidence from the therapeutic community (TC) literature
highlights the importance
of context and environment for offender rehabilitation. Jensen
and Kane (2012) found
that completing a TC had a significant effect on reducing the

likelihood of rearrest for
prisoners. Marshall (1997) conducted a large-scale evaluation of
the effectiveness of
TCs for sexual offenders. In his 4-year follow-up, he found that
18% of treated offend-
ers (with two or more previous convictions for sexual offences)
were reconvicted
compared with 43% of untreated sexual offenders. Such
environments have been
found to also bolster treatment goals and targets and contribute
to prosocial modelling.
TCs have been found to have a positive effect on self-identity
and enable prisoners to
construct positive identities (Miller, Sees, & Brown, 2006); they
improve quality of
life for prisoners within the institution (Shefer, 2010), effect
personality change, and
prisoners are less likely to receive an adjudication within the
prison (Newton, 1998).
This has led some to argue that TCs, or at least environments
that have an explicit
therapeutic focus, are the ideal environments for “doing” sexual
offender treatment
(Ackerman, 2010; Ware, Frost, & Hoy, 2010).

Prison Climate and Potential Correlates of a Prison Climate
The definition of prison climate is, at times, ambiguous with
some using terms like
“culture” and “climate” interchangeably (Day et al., 2012). A
good prison social cli-
mate can be characterised as being supportive, offering a safe
environment and oppor-
tunities for personal growth and development (van der Helm,
Stamms, & van der
Laan, 2011). Schalast et al. (2008) proposed that a social and
therapeutic climate is the
extent to which the climate is perceived as supportive of
therapy and therapeutic
change. This incorporates whether mutual support is typically
seen as characteristic of
the prison environment and the level of tension, perceived
threat of fear, aggression,
and violence within the prison. We contend that a rehabilitative
climate of a prison can
be understood as the prison’s social climate coupled with the
prison’s culture, philoso-
phy, and fitness for purpose in relation to reducing reoffending.
These critical aspects
of a prison are likely to have a direct impact on the

effectiveness of rehabilitative
374 International Journal of Offender Therapy and Comparative
Criminology 60(4)
measures, behaviour, and personal change and consequently the
effectiveness of the
prison in reducing reoffending.
The climate of a prison is related to aspects of prison life. A
key component for any
prison climate would appear to be prisoner–staff relationships.
For example, it has
been argued that positive attitudes and beliefs about change in
prison staff and prison-
ers are vital for fostering effective offender rehabilitation and
promoting change in
offending behaviour (see, for example, Hogue, 1993; Kjelsberg
& Loos, 2008; Lea,
Auburn, & Kibblewhite, 1999). This has been found to be
important for sex offenders
as positive attitudes by prison staff toward sex offenders have
been found to facilitate

ArticleSex Offender Recidivism Revisited Review ofRecen.docx

ArticleSex Offender Recidivism Revisited Review ofRecen.docx

Recomendados

Recomendados

Mais conteúdo relacionado

Semelhante a ArticleSex Offender Recidivism Revisited Review ofRecen.docx

Semelhante a ArticleSex Offender Recidivism Revisited Review ofRecen.docx (20)

Mais de ssusera34210

Mais de ssusera34210 (20)

Último

Último (20)

ArticleSex Offender Recidivism Revisited Review ofRecen.docx