Salisbury 2008

Crime & Delinquency
Article Volume 55 Number 4

October 2009 550-585
© 2009 SAGE Publications
The Predictive Validity of a 10.1177/0011128707308102
http://cad.sagepub.com
Gender-Responsive hosted at
http://online.sagepub.com
Needs Assessment
An Exploratory Study
Emily J. Salisbury
Portland State University
Patricia Van Voorhis
University of Cincinnati
Georgia V. Spiropoulos
California State University, Fullerton
Risk assessment and classification systems for women have been largely
derived from male-based systems. As a result, many of the needs unique to
women are not formally assessed or treated. Emerging research advocating a
gender-responsive approach to the supervision and treatment of women offenders
suggests that needs such as abuse, mental health, substance abuse, relationship
difficulties, self-esteem, self-efficacy, and parenting issues are important treat-
ment targets. Although these needs may be highly prevalent among women
offenders, they have not been adequately tested to determine their relationships
with future offending. In response, the present study sought to understand
whether gender-responsive needs contributed as risk factors to poor prison
adjustment and community recidivism. Additionally, several types of risk
assessment models were explored to determine whether gender-responsive
needs enhanced the validities of currently established risk classification.
Authors’ Note: The authors would like to thank Phyllis Modley from the National Institute
of Corrections, Sammie Brown from the South Carolina Department of Corrections, and Scott
Hromas from the Colorado Department of Corrections for their hard work and assistance on
this research project. This research was funded by the National Institute of Corrections under
cooperative agreement 99P03GIL0 with the Center for Criminal Justice Research at the
University of Cincinnati. Points of view or opinions stated in this article are those of the
authors and do not necessarily represent the official position or policies of the U.S.
Department of Justice or National Institute of Corrections. An earlier version of this article
was presented at the Annual Meeting of the American Society of Criminology, Nashville, TN,
November 16 to 20, 2004. Please address correspondence to Pat Van Voorhis, Division of
Criminal Justice, University of Cincinnati, P.O. Box 210389, Cincinnati, OH 45221-0389.
550
Downloaded from cad.sagepub.com at SIMON FRASER LIBRARY on March 16, 2015

Salisbury et al. / Gender-Responsive Needs Assessment 551
systems (i.e., a state’s institutional custody scale and the Level of Service
Inventory-Revised). Patterns of results differed across prison and community
outcomes, with some gender-responsive needs contributing to more valid risk
assessment systems. As a pilot study, the results, although mixed, appear to
support continued research on this topic.
Keywords: women offenders; risk-needs assessment; gender-responsive
C orrectional researchers and practitioners focus their attentions on men

largely because they comprise the majority of convicted offenders in
the United States. Indeed, women offenders represent only 7% of the U.S.
prison population (Bureau of Justice Statistics, 2005). However, recent sta-
tistics indicate that the female prison population is growing at an alarming
rate—faster than the male prison population. Since 1995, the total number
of incarcerated women has increased 53%, whereas the total number of
incarcerated men has increased 32% (Bureau of Justice Statistics, 2005).
This escalation draws attention to current state and federal practices for
assessing and classifying women offenders in both institutional and com-
munity corrections. Such classification systems involve the use of statisti-
cally derived, actuarial assessments to achieve two purposes (Van Voorhis,
2004). The first, and largely viewed to be the most important purpose, is to
predict an offender’s likelihood of recidivism or an inmate’s likelihood of
serious misconducts. Ultimately, the resulting risk score determines the
custody level of one’s prison assignment if incarcerated or level of com-
munity supervision if on probation or parole. The second purpose of cor-
rectional classification is to identify needs that must be addressed, or
programmed for, to meet basic needs, change offender behavior, or assure
humane prison adjustment (Clements, McKee, & Jones, 1984).
Given their import to the lives of offenders, it is extremely unfortunate
that most correctional classification systems were originally developed for
men and subsequently applied to women with little regard for their validity
or appropriateness (Bloom, Owen, & Covington, 2003; Chesney-Lind,
1997; Morash, Bynum, & Koons, 1998; Van Voorhis & Presser, 2001). With
respect to prison classification, for example, a recent national survey of
state correctional classification directors found that 36 states had not vali-
dated their institutional classification systems on their female inmates (Van
Voorhis & Presser, 2001). In addition, the directors opined that many of the
prison risk assessment systems, or custody classification systems,1 resulted
in “overclassification” of women. In other words, women were held at higher
custody levels than warranted by their behavior. Finally, many directors

552 Crime & Delinquency
felt that the current generation of classification systems failed to address the
unique needs of women offenders, particularly those pertaining to mental
health, children and parenting, relationships, self-esteem, and abuse (Van
Voorhis & Presser, 2001).
Risk assessments designed for community correctional agencies gener-
ate similar concerns. Distinct from the institutional risk assessments in that
needs and criminal history factors are combined into a single instrument,
these assessments are faulted by an emerging literature on gender-responsive
programming (Bloom et al., 2003) for their failure to tap needs optimally
relevant to women offenders. The omission of gender-responsive factors
from current community assessments is attributed to the fact that they too
were constructed from research samples of male offenders (Blanchette,
2004; Blanchette & Brown, 2006; Brennan, 1998; Brennan & Austin, 1997;
Farr, 2000; Reisig, Holtfreter, & Morash, 2006). As a result, the gender-
responsive factors (e.g., abuse or trauma, parenting, mental health, rela-
tionships, self-esteem) have not been adequately tested to determine
whether they are risk factors for future offending.
This article empirically examines the issue of whether current correc-
tional classification instruments are valid and relevant to women offenders.
We explore this issue with respect to both institutional and community cor-
rections. In doing so, a number of questions are addressed. First, are cur-
rent prison and community risk assessments predictive of appropriate
offense-related outcomes (e.g., serious prison misconducts in the case of
prison risk assessment and new offenses and technical violations in the case
of community risk assessment)? Second, might the emerging gender-
responsive needs also be considered risk factors for future offending?
Third, might consideration of gender-responsive factors improve on the
current risk assessment models for women? In this sense, we compare the
predictive strength of commonly used predictors2 to the array of women’s
needs emerging from the gender-responsive literature (Belknap & Holsinger,
2006; Bloom et al., 2003; Brennan, 1998; Covington, 1998; Funk, 1999;
Holtfreter & Morash, 2003; McClellan, Farabee, & Crouch, 1997; Morash
et al., 1998; Owen, 1998).
Risk, Needs, and Correctional Policy
Historically, correctional classification differentiated the assessment of

risk from the assessment of needs (Van Voorhis, 2004). Early risk assess-
ment scores represented a numerical sum of statistically derived predictors

(risk factors) pertaining to static, current offense, and criminal history mea-
sures (Bonta, 1996). Needs assessments, on the other hand, briefly screened
for the presence of educational, employment, substance abuse, mental
health, family, financial, or medical needs. Our research is embedded in a
more recent innovation in classification—the discovery that many of these
needs are also risk factors, predictive of prison misconducts, technical vio-
lations, and new offenses (Andrews, Bonta, & Hoge, 1990). As such, the
most recent generation of risk assessment instruments are now composed
of both offense-based predictors (e.g., measures of seriousness of the current
offense and prior record) and needs that are also known to be predictive of
new offenses or correctional misconducts (Bonta, 1996). Consequently, the
latest generation of risk assessments, called dynamic risk or needs assessment
instruments, serve a dual function; they assess risk and direct correctional
practitioners to the needs that contribute to an offender’s prospects for future
offending.
Community correctional agencies have moved more quickly to dynamic
risk assessment instruments than prisons, which stay wedded to the static
risk assessments comprised of measures of criminal history. However, with
the advent of prisoner re-entry initiatives (Petersilia, 2003; Travis, 2005)
and the commensurate policy that planning for prison release begin at
prison entry with a strong understanding of offender needs as they pertain
to risk upon release, a number of states are implementing dynamic risk or
needs assessments in prisons. New re-entry models, such as the National
Institute of Corrections (NIC) Transition from Prison to Community
Initiative (http://nicic.org/Library/017520), encourage the use of dynamic
risk assessment tools even though offenders are incarcerated. Although the
purpose of such use is for planning release decisions rather than custody
decisions, some studies have nevertheless observed that the dynamic
instruments also predict institutional misconducts (Bonta, 1989; Bonta &
Motiuk, 1987, 1990, 1992; Kroner & Mills, 2001; Motiuk, Motiuk, &
Bonta, 1992; Shields & Simourd, 1991). The most well-known dynamic
assessments of this kind are the Northpointe COMPAS (Northpointe
Institute for Public Management, 1997) and the Level of Service Inventory-
Revised (LSI-R; Andrews & Bonta, 1995).
These dynamic risk or needs assessment models relate well to current
correctional priorities. In most correctional policy circles, risk is cited as
the driving force behind correctional budgets, institutional construction,
and correctional programming (Cullen, Fisher, & Applegate, 2000; Feeley
& Simon, 1992). It follows, then, that risk factors, or predictors of recidi-
vism, must be given priority for treatment dollars over factors that are not

predictors of recidivism. This paradigm is supported by a series of widely

acclaimed meta-analyses of experimental studies of correctional rehabilita-
tion programming (Andrews, Zinger et al., 1990; Gendreau, Little, &
Goggin, 1996; Izzo & Ross, 1990; Lipsey, 1992). Two principles—the risk
principle and the needs principle (Andrews, Bonta, & Hoge, 1990; Andrews,
Zinger et al., 1990)—emerged from the meta-analytic research and have
come to have a profound impact on correctional practice and funding pri-
orities. The risk principle builds from findings showing that the most effec-
tive programs (those achieving the greatest reductions in recidivism) were
those that directed intensive services to medium- and high-risk clients
(Andrews, Zinger et al., 1990; Bonta, Wallace-Capretta, & Rooney, 2000;
Lipsey, 1992; Lipsey & Wilson, 1998; Lowenkamp & Latessa, 2002, 2005;
Lowenkamp, Latessa, & Holsinger, 2006).3
The needs principle illustrates that reductions in recidivism can be
achieved only through targeting dynamic risk factors defined as needs that
are correlated with future offending (Andrews, Bonta, & Hoge, 1990; Andrews,
Zinger et al., 1990). Of crucial importance to women offenders, especially
in light of the fact that the needs principle is a policy directive, is the issue
that needs targeted for treatment are essentially those embedded within the
current generation of dynamic risk assessments. These dynamic risk factors
include such needs as antisocial attitudes, criminal peers, substance abuse,
education, employment, satisfaction with family life, and financial well-
being. Additionally, it has been argued that the “big four” risk factors
(i.e., antisocial attitudes, peers, personality, and criminal history) are the
strongest predictors of recidivism and therefore should be the primary focus
of most correctional programs (Andrews & Bonta, 2003). In this context, it
is widely accepted that correctional programs should target the traditional
needs listed above through social learning and cognitive-behavioral treat-
ment modalities (Andrews & Bonta, 2003; Andrews, Zinger et al., 1990;
Antonowicz & Ross, 1994; Garrett, 1985; Gendreau, 1996; Izzo & Ross,
1990; Lipsey, 1992; Lipsey, Chapman, & Landenberger, 2001; Lösel, 1995;
Pearson, Lipton, Cleland, & Yee, 2002).
Thus, it is assumed by many that the empirical search for the dynamic risk
factors that should be included in risk assessment instruments is over. After
all, the dynamic risk factors and the principles discussed above were identi-
fied in a series of rigorous meta-analyses (Andrews, Zinger et al., 1990;
Garrett, 1985; Gendreau et al., 1996). Until recently, the fact that these meta-
analyses contained very few studies of women offenders did not dissuade
scholars or practitioners from assuming that women’s assessments and
programs should look like men’s assessments and programs (Chesney-Lind,

2000). In support, a meta-analysis that sampled a much smaller number of

studies found that the factors enumerated above predicted new offenses and
other adverse outcomes for women (Dowden & Andrews, 1999). Additional
studies have found the dynamic risk assessment instruments to be valid
for women (Andrews, Dowden, & Rettinger, 2001; Blanchette & Brown,
2006; Coulson, Ilacqua, Nutbrown, Giulekas, & Cudjoe, 1996; Holsinger,
Lowenkamp, & Latessa, 2003), but others reported areas of divergence
(Blanchette, 2005; Law, Sullivan, & Goggin, in press; Olson, Alderden, &
Lurigio, 2003; Reisig et al., 2006). It is noteworthy, however, that few stud-
ies speak directly to the recommendations from the gender-responsive lit-
erature. A critical question, asked mostly by nonsupportive stakeholders,
concerns whether the gender-responsive attributes of trauma, dysfunctional
relationships, self-esteem, self-efficacy, parenting, and other family issues
are risk factors or simply problems that are highly prevalent in the troubled
lives of women offenders but not related to future offending (Blanchette &
Brown, 2006). If the latter, many would concur that these issues are not
important from a policy standpoint.
Classification and Women Offenders
What are the implications of this situation for women offenders? Clearly
there is no lack of policy governing the treatment of women offenders
because essentially it is the same policy governing the treatment of men
sans the supporting research. Women offenders did not factor into the
development of any currently-used risk assessment instruments. This situa-
tion is most egregious with respect to prison custody classification sys-
tems—those risk assessment systems that continue to use static, current
offense, and criminal history predictors of prison misconducts. As noted
above, as of 2001, approximately 36 states had not validated their custody
assessments for women offenders (Van Voorhis & Presser, 2001). Contrary
to ethical guidelines in the fields of psychology and education (American
Association of Correctional Psychologists Standards Committee, 2000;
American Psychological Association, 1992), women were assigned to dif-
ferent custody levels, including maximum, on the basis of criteria that were
not known to be related to security concerns for imposing adverse condi-
tions, such as greater distances from home, fewer privileges, and more aus-
tere environments, for purposes of security (Hardyman & Van Voorhis,
2004; Van Voorhis & Presser, 2001).

Flaws in the custody assessments have appeared apart from validation

research. In the absence of empirical studies, focus groups with adminis-
trators and female inmates have observed repeatedly that troubled inmates
make more difficult adjustments than inmates with high custody scores
(Hardyman & Van Voorhis, 2004).
Even in those few studies where risk assessment (custody) instruments
proved valid for women, women at higher custody levels incurred less seri-
ous misconducts than men at the same level. For example, the few compar-
ative studies of male and female prison populations available found that the
proportion of women in maximum custody who incurred serious prison mis-
conducts more closely approximated the proportion of medium custody men
who committed such acts (Brennan, 1998; Fowler, 1993; Hardyman & Van
Voorhis, 2004). Overclassification occurs with the dynamic risk or needs
instruments as well. In one study, for example, the difference in recidivism
rates for high-risk men was 10% higher than the rate for high-risk women
(Washington State Institute for Public Policy, 2003). Correcting the problem
through adjustment to cut points differentiating risk levels, so that men and
women have similar behavioral outcomes, greatly reduces the number of
women appropriate for maximum custody (Harer & Langan, 2001), with
implications that are likely to be difficult to sell to correctional officials.4
A final implication of the current classification practices for women
stems from the fact that popular dynamic risk assessments do not converge
with the emerging gender-responsive literature. If it were not for the fact
that risk, as embodied in terms such as risk management, security, and com-
munity safety, is the central guiding force of correctional policy (Cullen
et al., 2000) and correctional classification both of men and women (Feeley
& Simon, 1992; Van Voorhis & Presser, 2001), the field might be content to
simply observe that women have unique needs that require treatment and to
proceed to develop new programs for women—programs to address par-
enting, trauma, relationships, and depression, for example. Yet the policy
imperatives of the risk and need principles have accorded priority to those
dynamic risk factors that have been identified through research on predom-
inately male offenders. There appears to be no place for gender-responsive
factors in this paradigm, and in this context, their import is likely to be
overshadowed by assessments that ignore them.
In response, the present study administered a number of assessment instru-
ments to a cohort of women offenders upon their admission to the Department
of Corrections (DOC) in a western state. These assessments included the
state’s Intake Custody Classification Instrument, the LSI-R (Andrews &
Bonta, 1995), and a series of scales designed to tap the aforementioned

gender-responsive measures (Van Voorhis, Pealer, Spiropoulos, & Sutherland,

2001). The women were followed up at 6 months while incarcerated and for
up to 44.2 months upon their release in the community for purposes of obtain-
ing outcome variables pertaining to prison misconducts, community recidi-
vism, and violations of supervision conditions.
Women’s Unique Needs
The emerging literature on gender-responsive strategies for women

offenders (Berman, 2005; Bloom et al., 2003) offers much support for the
belief that had we started with women, the current generation of risk or
needs assessments might look quite different from the status quo. Women
offenders have increasingly become understood as a unique population,
evidencing different pathways to crime in comparison to men (Bloom et al.,
2003; Daly, 1992, 1994; Owen, 1998; Reisig et al., 2006; Richie, 1996).
These pathways acknowledge the following needs: (a) extensive traumatic
and abusive histories; (b) experiences of acute mental illness, most typi-
cally major mood disorders (i.e., depression, anxiety, PTSD); (c) issues
with self-esteem and self-efficacy; (d) dysfunctional relationships, espe-
cially with intimate others; (e) overwhelming parental responsibilities; and
(f) substance abuse, often to self-medicate emotional or physical pains.
Victimization and Abuse

Research has shown that women under correctional supervision are
more likely to experience physical and sexual abuse as children and adults
than male offenders or women in the general population (Bureau of Justice
Statistics [BJS], 1999; McClellan et al., 1997). The BJS (1999) reported that
percentages of female offenders reporting physical abuse at some point in
their lives ranged from 32% to 47%, depending on the type of correctional
population examined. For male offenders, percentages ranged from 6% to
13%. Even more staggering were reports of sexual abuse—for women,
figures were between 23% to 39%; for men, they fell between 2% to 6%
(BJS, 1999). The rates of abuse reported by BJS may be conservative esti-
mates, as other studies found rates of physical abuse among women as high
as 75% (Browne, Miller, & Maguin, 1999; Greene, Haney, & Hurtado,
2000; Owen & Bloom, 1995).
Research exploring associations between various victimization types
and crime offer somewhat inconsistent conclusions. Although evidence is

mounting in support of the relationship between child maltreatment and

delinquency in young girls (Hubbard & Pratt, 2002; Siegel & Williams,
2003; Widom, 1989; but also see Giordano, Deines, & Cernkovich, 2006),
the link between victimization experiences (both child and adult) and
women’s recidivism seems much less conclusive. Previous research either
revealed no significant association once other relevant variables were con-
trolled (Bonta, Pang, & Wallace-Capretta, 1995; Loucks & Zamble, 1999;
Rettinger, 1998), or significant and/or near-significant negative associa-
tions, implying that abused women were less likely to engage in future
criminal activity than those who were not abused (Blanchette, 1996; Bonta
et al., 1995). Additionally, Lowenkamp, Holsinger, and Latessa (2001)
showed that the prevalence of child abuse alone did not contribute to more
explained variation in recidivism than by the LSI-R alone. A potential lim-
itation of several of these studies is that women offenders were asked about
childhood abuse in an interview format, which could have resulted in
underreporting (Browne et al., 1999).
In contrast, prospective studies of young girls followed into adulthood
revealed evidence supporting the link between childhood victimization and
adult criminal behavior. Comparing abused and nonabused girls’ juvenile
and adult records, Widom (1989) found that girls who were abused and
neglected were significantly more likely to have both a formal juvenile
delinquency record and a formal adult criminal record compared to girls
who were not abused nor neglected.5 Similar results were also found by
Siegel and Williams (2003) who studied girls struggling with sexual abuse
histories.
A recent meta-analysis of the female offender risk prediction research
indicated some empirical support for the childhood victimization–recidivism
relationship. Law, Sullivan, and Goggin (in press) found a moderately sig-
nificant mean effect size when predicting general community recidivism
from measures of child abuse (k = 11, Mz+ = .16, CIMz+ = .12 to .19). The
mean effect size for adult victimization and general community recidivism
fell into a significant, but unreliable, range; that is, the confidence interval
was too wide, suggesting there were not enough primary studies to create a
reliable estimate. Notably, Law et al.’s results further suggested that the vic-
timization relationships may be dependent on the type of recidivism pre-
dicted: With child abuse, significant mean effect sizes were exhibited only
with general community recidivism and not with measures of institutional
adjustment.
Overall, the relationship between abuse and criminal behavior in adult
women is mixed, likely as a result of studies (a) utilizing various techniques

to measure victimization and (b) predicting different types of recidivism.

Moreover, current meta-analyses are not entirely conclusive because of the
limited number of available studies.
Mental Health
The mental health needs of female offenders differ substantially from
those of male offenders. Depression, anxiety, and self-injurious behavior are
more prevalent among female than male populations (Belknap & Holsinger,
2006; Bloom et al., 2003; McClellan et al., 1997; Peters, Strozier, Murrin, &
Kearns, 1997). Disorders commonly seen with women offenders include
major mood disorders such as depression and bipolar disorder, as well as
panic, posttraumatic stress, and eating disorders (Bloom et al., 2003). Most
importantly, women suffer from several co-occurring mental health needs
such as depression and substance abuse (Bloom et al., 2003; Holtfreter &
Morash, 2003; Owen & Bloom, 1995) at rates that are nearly four times the
rate for men (Blume, 1997). Similarly, phobic disorders were observed at
more than twice the rate, and panic disorders at three and a half times the
rate for men (Blume, 1997).
In some accounts, mental health needs are categorized as responsivity fac-
tors rather than dynamic risk factors. In such discussions, depression, anxi-
ety, and other psychological issues are considered needs, which should be
accommodated for a variety of reasons but not necessarily for reducing future
criminal behavior (Andrews, Bonta, & Hoge, 1990). After reviewing the pre-
diction literature on mental health, Blanchette and Brown (2006) concluded
that “personal distress, mental ability, and mental health variables are not
strongly associated with women’s likelihood of recidivism” (p. 105).
However, two significant problems afflict most research in this area.
First, traditional mental health domains on risk assessment instruments are
driven largely by the offender’s exhibition of severely psychotic behavior.
Major mood disorders, such as those frequently seen with women, can be
overlooked if they have not been previously diagnosed and recorded. Better
measures of women’s mental health issues are needed, namely behaviorally
specific indicators of depression, anxiety, and PTSD. In this context, stress,
depression, fearfulness, and suicidal thoughts and attempts have shown to
be strong predictors of women’s recidivism (Benda, 2005; Blanchette &
Motiuk, 1995; Brown & Motiuk, 2005), though not for men’s recidivism
(Benda, 2005).
Second, prediction studies frequently aggregate mental illness indicators
into broad mental health domains that could potentially confound relevant

associations. For example, results from Law et al.’s (in press) meta-analysis
suggested that women offenders’ mental health aspects are significantly
related to both institutional (k = 26, Mz+ = .07, CIMz+ = .02 to .11) and com-
munity outcomes (k = 13, Mz+ = .09, CIMz+ = .06 to .12). Although these
mean effect sizes are relatively weak in strength, the study’s mental health
domains reflected an amalgam of heterogeneous indicators of mental ill-
ness. This method of aggregation could mask important relationships
between specific types of mental illness and recidivism. Essentially, the
psychological issues specific to women offenders have not been rigorously
tested for their import to the task of risk assessment.
Dysfunctional Relationships
Relationships are certainly of great importance to all people, but they are
critically so for women. According to relational theory, a woman’s identity,
self-worth, and sense of empowerment are said to be defined by the quality
of relationships she has with others (Gilligan, 1982; Kaplan, 1984; Miller,
1976; Miller & Stiver, 1998). However, because of the high rates of abuse,
trauma, and neglect experienced by female offenders, their ability to rec-
ognize and achieve healthy, mutually empowering relationships is severely
limited (Covington, 1998). Indeed, women offenders often engage in co-
dependent relationships that facilitate their criminal behavior (Koons, Burrow,
Morash, & Bynum, 1997; Richie, 1996). Extricating themselves from dys-
functional relationships appears to be quite difficult. If forced into a choice of
either being abandoned (or abused) by their intimate partner or engaging in
criminal behavior to secure his needs, the decision often becomes an easy one
for women (Richie, 1996), one which is tied to the continued fulfillment of a
multitude of needs (e.g., economic, housing, parental, addictive, etc.).
Relational theory (Miller, 1976) generally speaks to the treatment
modalities that would be most effective with women, but it remains largely
silent on theoretical explanations of female offending, other than to inform
a pathways perspective. However, one plausible proposition gleaned from
the theory is that females are less inclined to engage in criminal behavior
because it threatens crucial relationships in their lives (Blanchette &
Brown, 2006). This explanation, however, may only pertain to women with
strong prosocial relationships, because the same relational attachment
process might also explain women’s increased participation in crime if they
are engaged in antisocial relationships. With so few studies of the impact of
relationships on criminal behavior, these matters are far from resolved. In
fact, one study revealed that relationships with intimate partners had both

positive and negative influences on women offenders (Benda, 2005). In this

study, criminal partners played a significant role in women’s recidivism
(more so than men’s recidivism); at the same time, women’s desistance was
significantly related to satisfying intimate relationships.
During the preliminary phase of the current study, women’s focus
groups frequently related that they feared the consequences of becoming
involved with uncaring, antisocial men upon their release (Van Voorhis
et al., 2001). Notwithstanding the paucity of research on the topic, a focus
on relationships has been shown to be an important characteristic of
promising correctional treatment programs for women (Koons et al., 1997).
Self-Esteem & Self-Efficacy

A significant amount of research has addressed whether self-esteem is a
dynamic risk factor. Results from these studies overwhelmingly indicated
as that low self-esteem, which was often aggregated into a category denoted
personal distress, was not a risk factor for recidivism and that programs
targeting self-esteem were not promising (Andrews & Bonta, 2003).
Furthermore, some programs that aimed to increase offenders’ sense of
self-worth actually increased the likelihood of recidivism (Andrews, 1983;
Andrews, Bonta, & Hoge, 1990; Gendreau et al., 1996; Wormith, 1984).
Again, the vast majority of these studies focused on male offenders.
Self-esteem, a concept related to empowerment, is often discussed in the
gender-responsive literature and has been targeted by a number of correc-
tional programs for women. Empowerment denotes the process of increas-
ing women’s self-esteem and internal locus of control (i.e., the belief that
their lives are under their own power and control; Task Force on Federally
Sentenced Women, 1990). Concepts of self-worth are often cited by cor-
rectional treatment staff, researchers, and women offenders themselves as
critical to their desistance (Carp & Schade, 1992; Case & Fasenfest, 2004;
Chandler & Kassebaum, 1994; Koons et al., 1997; Morash et al., 1998;
Prendergast, Wellisch, & Falkin, 1995; Schram & Morash, 2002; Task
Force on Federally Sentenced Women, 1990).
The psychological literature puts forward a large body of knowledge
showing negative associations between women’s abusive experiences and
self-esteem among women in the general population (Aguilar & Nightingale,
1994; Cascardi & O’Leary, 1992; Clements, Ogle, & Sabourin, 2005;
Clements, Sabourin, & Spiby, 2004; Orava, McLeod, & Sharpe, 1996; Resick,
1993; Williams & Mickelson, 2004; Zlotnick, Johnson, & Kohn, 2006).
However, whether women’s self-esteem, in turn, is related to recidivism is

understudied, although one meta-analysis (12 effect sizes) showed an asso-

ciation between female offenders’ low self-esteem and antisocial behavior
(Larivière, 1999).
Self-efficacy, a distinct concept from self-esteem, can be defined as a
person’s confidence in achieving specific goals. Similar to current evidence
with self-esteem, the general evidence-based, risk prediction research cate-
gorizes low self-efficacy as a personal distress factor, which again has been
shown to be of minimal import in predicting recidivism based on studies
conducted with male offenders. Little is known about the importance of
self-efficacy to recidivism with women offenders, but it has been suggested
as playing a major role (Rumgay, 2004). Additionally, improved self-efficacy
through skills enhancement is advocated as a central, critical element of
gender-responsive treatment programming (Bloom et al., 2003; Bloom,
Owen, & Covington, 2005).
Parental Stress
Nearly 71% of women under correctional supervision have at least one
child under the age of 18, with an average of 2.11 children (BJS, 1999).
This, coupled with women’s economic marginalization and substance
abuse, often leads to stress and overwhelmed feelings about being able to
take care of and provide for their children (Greene et al., 2000). Maternal
demands may contribute to recidivism based on the fact that many women
offenders also have (a) financial difficulties in providing for themselves and
their children, (b) substance abuse problems, and (c) minimal assistance. In
support, some studies with mothering offenders have detected a relation-
ship between parental stress and crime (Ferraro & Moe, 2003; Ross,
Khashu, & Wamsley, 2004). Similarly, Bonta et al. (1995) found that
women offenders who were parenting children alone were significantly
more likely to be reconvicted than women raising children with partners
(51.7% vs. 22.2%, χ2 = 4.01, p < .05).
Incarcerated mothers have received the majority of research attention on
parental stress, leaving limited data on the vast number of mothers in com-
munity corrections. Much of this inmate mother research, including
Baunach’s (1985) groundbreaking work, Mothers in Prison, focused on the
effects of incarceration on mothers and their children, as well as the practi-
cal concerns surrounding visitation and custody issues (Clark, 1995; Enos,
2001; Kampfner, 1995; Kazura, 2001). More recent studies investigated the
relationship between child contact and women’s prison adjustment, finding
that stress associated with limited contact was related to higher levels of
mental illness (Houck & Loper, 2002; Tuerk & Loper, 2006).

Parental stress is perhaps at its greatest among women who are threat-
ened with the loss of child custody, a fairly common occurrence since the
passage of the Adoption and Safe Families Act of 1997. Although it is
assumed that the loss of children can be the result of arrest and incarcera-
tion, Ross et al. (2004) found that 85% of maternal arrests occurred after
rather than prior to child placement. These mothers, most of whom had
criminal records, were likely having difficulty maintaining their parental
responsibilities as inferred by the placement of their children into foster
care. Losing their children became a devastating event, creating a down-
ward spiral that often led to criminal charges for drug use (56%).
The Present Study
The present study sought to understand whether the gender-responsive

needs contributed (as risk factors) to poor prison adjustment and commu-
nity recidivism. In 1999, the Prisons Division of NIC entered into two
cooperative agreements to develop improved strategies for classifying
incarcerated women offenders (Hardyman & Van Voorhis, 2004). One of
these studies, conducted by the University of Cincinnati, examined classi-
fication procedures for female inmates sentenced to the DOC in a western
state. That project examined the comparative validity of three different clas-
sification models: (a) static risk assessment (the custody classification
model in use at the time of the project), (b) a dynamic risk or needs model,
and (c) a dynamic model incorporating gender-responsive measures of par-
enting, self-esteem, self-efficacy, relationships, mental health, adult victim-
ization, and child abuse (Van Voorhis et al., 2001).
To assess these gender-responsive needs, an instrument (referred to as a
trailer) was created by integrating both well-established scales and newly
constructed composite scales (details below). This self-report instrument
was administered at prison intake, along with the state’s static (custody)
risk instrument. Measures of additional dynamic risk factors (e.g., anti-
social attitudes, antisocial peers, mental health, substance abuse, education,
employment, financial, accommodations, and use of leisure time) were
obtained from the LSI-R (Andrews & Bonta, 1995) and were administered
as a part of the participants’ presentence investigation.
This article presents findings from the original study as well as a follow-
up study of the participants on their release to parole. Data for the original
validation study were collected on all women admitted to the state DOC
between October 10, 2000 and January 8, 2001, creating an intake cohort

of 156 women offenders. Institutional adjustment (i.e., serious prison mis-

conducts) measures were collected 6 months after intake. The present study
extended the follow-up period of the original cohort to include up to 44.2
months of time in the community following the release of 85.9% of the
sample (N = 134). As might be apparent from these sample sizes, this
research was preliminary in nature and served as a pilot study for a much
larger research initiative currently underway at the University of Cincinnati.
Method
Participants
Criminal history, classification, prison misconducts, and recidivism data
for the intake and released participants are presented in Table 1. Half of the
women in the intake sample were White (53.2%), 28.8% were Black, and
16.0% were Hispanic. The mean age of the sample at admission was 34.6
years. Convictions for the original sample were primarily for property
(28.4%) and drug-related offenses (43.9%). On entering the prison, the
majority were placed on minimum or minimum-restrictive custody levels
(76.3%), whereas only 23.1% were on medium custody. Only one woman
was held at close supervision. Twenty-eight of the women released (20.9%)
had at least one new re-arrest for either a felony or misdemeanor, and 47
women had at least one technical violation (35.1%). Thus, 73 women
(54.5%) were classified as having failed (committing either a new crime or
technical violation) in the community.
Although the release sample contains 22 fewer inmates than the intake
sample, age (M = 34.2) and race distributions were similar to those for the
intake sample. Table 1 also shows that offense-related characteristics of
the release sample were similar to those for the original intake sample.
There were no significant differences between the release sample and the
original, intake samples on any of these measures.
Measures
Recidivism data were compiled by researchers at the state’s DOC. Two
measures of recidivism—(a) new crimes and (b) technical violations while
on parole—were obtained from the state’s information center and National
Crime Information Center databases. The exact nature of these offenses and
violations was, unfortunately, not available. Both dependent variables
were categorized into incidence (frequency) and prevalence (presence and
absence) measures.

Table 1
Frequency and Percentage Distribution of Demographic, Criminal
History, Classification, Prison Adjustment, and Recidivism Measures
Intake Sample Release Sample
Variable N % N %
Number Representative of Sample 156 100.0 134 100.0
Race
White 83 53.2 68 50.7
Black 45 28.8 44 32.8
Hispanic 25 16.0 20 14.9
Native American 3 1.9 2 1.5
Most serious conviction charge
Burglary 4 2.6 3 2.2
Assault 13 8.3 7 5.2
Robbery 2 1.3 1 0.7
Theft 28 18.1 27 20.1
Escape/attempt escape 13 8.4 11 8.2
Forgery/fraud 12 7.7 11 8.2
Attempt/possession of drugs 40 25.8 35 26.1
Distribute/sell drugs 28 18.1 25 18.7
Other 15 9.7 14 10.4
Maximum sentence length
Less than 24 months 48 30.8 47 35.1
25 to 48 months 62 39.7 57 42.5
49 to 120 months 42 26.9 30 22.4
More than 120 months 4 2.6 0 0.0
Prior felonies
None 80 51.3 68 50.7
One 30 19.2 29 21.6
Two 23 14.7 19 14.2
Three or more 23 14.7 18 13.4
Prior incarcerations
No 128 82.1 111 82.8
Yes 28 17.9 23 17.2
Number representative of sample 156 100.0 134 100.0
Current custody level at intake
Minimum 36 23.1 25 19.2
Minimum-restrictive 83 53.2 67 51.5
Medium 36 23.1 37 28.5
Close 1 0.6 1 0.8
Number of serious disciplinaries
while incarcerated (6 months)
None 129 82.7 112 83.6
One 17 10.9 13 9.7
Two or more 10 6.4 9 6.7
Mean: 0.3 serious disciplinaries
(continued)

Table 1 (continued)
Intake Sample Release Sample
Variable N % N %
Number of Rearrests NA NA
None 106 79.1
One 23 17.2
Two 5 3.7
Mean: 0.3 new crimes
Number of Postrelease Technical Violations NA NA
None 87 64.9
One 43 32.1
Two 4 3.0
Mean: 0.4 technical violations
Any Postrelease Failure NA NA
No 61 45.5
Yes 73 54.5
LSI-R Risk Categories
High (41 or higher) 29 18.6 25 18.7
Medium High (34-40) 56 35.9 49 36.6
Moderate (24-33) 48 30.8 41 30.6
Low Moderate (14-23) 21 13.5 18 13.4
Low (13 or less) 2 1.3 1 0.0
Note: The mean age for the intake sample is 34.6 years. The mean age for the release sample
is 34.2 years.
Time at risk was calculated by subtracting either (a) the first failure date
or (b) the last date of data collection (July 15, 2004) from the date of release,
whichever came first. Women’s time at risk ranged from one day (e.g., fail-
ure to report) to 44 months. The average time at risk for women who failed
on parole was 17 months. Women who succeeded on parole had an average
time at risk of approximately 20 months. In all analyses, a significance level
of .10 was chosen because of the exploratory nature of this study.
The scales created below, with the exception of the LSI-R and the state’s
mental health measure, were created through factor analysis using either (a)
principle component extraction with a varimax rotation or (b) maximum like-
lihood extraction with a quartimax rotation, depending on the scale. Item
analysis and more detailed psychometric results are provided in the original
final report (Van Voorhis et al., 2001) and are available from the authors.
Level of Service Inventory-Revised. The LSI-R (Andrews & Bonta,

1995) is a well-established dynamic risk or needs assessment consisting of
a semistructured interview, corroborated by a review of official records. The
54-item scale measures 10 distinct domains, including criminal history,

education or employment, financial situation, family or marital relation-

ships, accommodation, use of leisure time, companions, alcohol or drug
use, emotional or personal, and attitude or orientations. The mean LSI-R
total score was 33.4 (minimum = 10, maximum = 48). Table 2 presents the
mean, standard deviation, and range of LSI-R subscale scores, and the fol-
lowing gender-responsive scales.
Institutional Risk Assessment (Custody Classification scale). This is an

institutional classification used in many states. The scale is the sum of the fol-
lowing items: (a) history of institutional violence; (b) severity of the current
offense; (c) multiple convictions; (d) severity of prior convictions; (e) escape
history; (f) current or pending detainers; (g) prior felony convictions; and
(h) duration of sentence. Custody scores ranged from 2 to 22 (M = 10.7).
Mental health. Two measures of mental health were available to this

study: (a) the LSI-R emotional or personal scale and (b) a 5-point scale
developed for the DOC, combining the results of the Millon Clinical
Multiaxial Inventory (Millon, 1997) and symptoms called to the attention
of DOC personnel. Psychometric details of the DOC measure were not
available at the time of the study; however, correlations with the LSI-R
emotional or personal scale (r = .53, p < .001) revealed its construct valid-
ity. Unfortunately, neither measure maps onto the gender-responsive litera-
ture in an ideal manner. They are both global measures of functioning,
which combine varied mental health diagnoses into one scale.
Rosenberg Self-Esteem scale. The Rosenberg Self-Esteem scale (Rosenberg,

1979) consists of 10 items using a 3-point Likert-type answer format. It has
been tested in a variety of settings and found to have strong psychometric
properties (Dahlberg, Toal, & Behrens, 1998; Rosenberg, 1979). High
scores on this scale reflect favorable levels (high) of self-esteem.
Sherer Self-Efficacy scale. The Sherer Self-Efficacy scale (Sherer et al.,

1982) is a 17-item scale using a 3-point Likert-type answer format. High
scores reflect high self-efficacy.
Relationship scale. The purpose of this scale was to identify women who
were experiencing relationship difficulties resulting in a loss of personal
power. A number of sources from the substance abuse literature use the
term co-dependency to describe such difficulties (Beattie, 1987; Bepko &
Krestan, 1985; Woititz, 1983). We recognize, however, that this construct
has not been widely researched.

Table 2
Mean, Standard Deviation, and Range of LSI-R and
Gender-Responsive Need
Minimum
to Maximum
Risk Factor M SD Score
Intake Custody Scale 10.7 4.6 2-22

LSI-R Total Score 33.4 7.6 10-48
LSI-R Subscale Scores
Criminal history 6.0 1.7 1-9
Education/work 5.3 2.9 0-10
Financial 1.5 0.6 0-2
Family/marital 1.8 1.2 0-4
Accommodations 2.3 1.0 0-3
Use of leisure time 2.0 0.3 0-2
Alcohol/drugs 6.0 2.8 0-9
Companions 3.6 1.0 0-5
Emotional/personal 1.7 1.3 0-5
Attitude 3.4 1.0 0-4
Gender-Responsive Needs
Self-esteem (alpha = .89) 24.0 4.8 10-30
Self-efficacy (alpha = .88) 43.5 6.1 28-51
Relationships (alpha = .78) 21.3 3.9 12-27
Total adult victimization (alpha = .92) 24.2 14.6 0-54
Adult emotional abuse (alpha = .96) 3.6 4.0 0-10
Adult physical abuse (alpha = .96) 12.5 7.7 0-26
Adult harassment (alpha = .95) 8.1 5.7 0-21
Total child abuse (alpha = .96) 16.5 12.0 0-43
Serious physical child abuse (alpha = .96) 9.3 6.5 0-22
Parental stress (alpha = .77) 33.4 4.7 21-44
Note: LSI-R = Level of Service Inventory-Revised.
The 15-item Likert-type questionnaire contained questions that were

influenced by, but not identical to, scales developed by Fischer, Spann,
and Crawford (1991; Spann-Fischer Codependency scale), Roehling and
Gaumond (1996; Codependent Questionnaire), and Crowley and Dill
(1992; Silencing the Self scale). Factor analysis revealed that the factor
accounting for the largest proportion of explained variance (21.1%; eigen-
value = 3.2) tapped items pertaining to loss of a sense of self in relationships,
neglect of self, worry over what others thought, and a greater tendency to
incur legal problems when in an intimate relationship than when not in one.
High scores indicated low codependency.

Parental stress. Modifications were made to a 20-item Likert-type scale

developed by Avison, Turner, and Noh (1986). Factor analysis revealed a
single factor containing 12 items that reflected a woman who felt over-
whelmed by her parental responsibilities and included items pertaining to
child management skills and the extent of support offered by family
members (n = 82; explained variance = 31.8%; eigenvalue = 3.8). High
scores on this scale denote less stress.
Adult Victimization and Child Abuse scales. Items contained in both the
Adult Victimization and the Child Abuse scales were informed by Belknap,
Fisher, and Cullen (1999), Campbell, Campbell, King, Parker, and Ryan
(1994), Coleman (1997), Holsinger, Belknap, and Sutherland (1999),
Murphy and Hoover (1999), Rodenberg and Fantuzzo (1993), and Shepard
and Campbell (1992).
The Adult Victimization scale contained 54 behavioral indicators of
abuse and victimization. Respondents were asked to mark one of three
response choices for each of the 54 items that included: (a) never, (b) less
than five times, and (c) more than five times. Factor analysis of the scale
revealed three factors: (a) emotional abuse, consisting of 10 items explaining
18.7% of the variance (eigenvalue = 9.9); (b) physical abuse, containing 13
items explaining an additional 17.2% of the variance (eigenvalue = 9.1); and
(d) harassment, containing 11 items explaining an additional 15.8% of the
variance (eigenvalue = 8.4). A total summary scale was also created.
The Child Abuse scale contained 24 behavioral indicators of abuse and
had the same response choices as the Adult Victimization scale. Factor
analysis of the scale indicated two factors, both related to physical abuse.
The first factor depicted more serious forms of physical abuse and
explained 33.3% of the variance (eigenvalue = 8.3). Less serious forms of
physical abuse explained 25.1% of the variance (eigenvalue = 6.3). The
Serious Physical Abuse scale and a Total scale combining the two physical
abuse scales were further investigated.
Results
Results of the study are shown in Tables 3 to 5 below. We begin with a

bivariate analysis of the impact of independent measures on prison and
community outcomes, shown in Table 3. Although none of the correlates is
particularly strong, all of the outcomes, prison and community, were far
more likely to have been impacted by needs than by the custody risk scale,

the classification model currently in use in many prisons. Static criminal

history items also comprise the criminal history scale of the LSI-R; it too
was only modestly related to technical violations and to the composite com-
munity outcome measure, any failure.
Patterns differed across prison and community outcomes. A considera-
tion of community outcomes finds the LSI-R (total scale) a stronger pre-
dictor of community outcomes (technical violation and any failure) than
prison misconducts. Additionally, several of the LSI-R subscales correlated
with technical violations and any failure. Findings were less apparent with
respect to the arrest data. We see in correlations for the LSI-R subscales a
pattern that suggests something other than the commonly asserted influence
of the “Big 4” (Andrews & Bonta, 2003). For example, we did not find anti-
social attitudes, antisocial companions, or criminal history6 among the
strongest predictors of either the prison or community outcomes. Instead,
risk for women was more strongly influenced by financial, education, liv-
ing conditions, and clearly substance abuse. Additionally, two variables—
emotional or personal and use of leisure time—correlated with outcomes in
the opposite direction for unknown reasons.
Table 3 also reports that several gender-responsive scales were correlated
with prison misconducts and recidivism. Child abuse, relationship issues,
self-efficacy, and adult emotional abuse were associated with serious mis-
conducts in prison. Childhood victimization—a predictor of prison miscon-
duct (r = .16, p < .05)—was not associated with community outcomes once
these women were released. At the same time, factors pertaining to
adult emotional abuse (r = .20, p < .05), harassment (r = .15, p < .05), and
a summary measure of victimization (r = .17, p < .05) did not influence prison
adjustment but did so once women returned to their communities. Parental
stress contributed to technical violations.
Neither of the mental health variables were related to outcome measures
in meaningful ways. In earlier analyses of the intake sample, the DOC mea-
sure correlated with aggressive prison misconducts (r = .22, p < .01). The
LSI-R measure did as well (r = .16, p < .05).
Informative, yet contradictory, results were found with regard to self-
efficacy and abuse when comparing the institutional versus community
outcomes. Women who reported more self-efficacious characteristics
(i.e., self-confidence) were significantly more likely to incur more serious
misconducts in prison (r = .14, r = .15, p < .05, for prevalence and inci-
dence data, respectively). Yet once released into the community, self-efficacy
seemed to become a protective factor for these women, where more confident
women had fewer technical violations than less confident women (r = –.13,

Table 3
Relationships Between Tested Risk Factors and Prison Misconducts, Technical Violations,
Rearrest, and Any Failure Under Community Supervision (Pearson r, one-tailed)
Prison Outcome Postrelease Community Outcomes

Serious Prison Technical
Misconducts Violations Rearrest Any Failure
Risk Factor Y/N Number Y/N Number Y/N Number Y/N
Custody Risk Scale (original)

LSI-R Total Score .12* .16** .18** .20*** .21***
Criminal history .15**
Education/Employment .13* .13* .12* .18** .14** .12* .24***
Financial .19** .19** .11*
Family/marital
Accommodations .19** .21***
Use of leisure time .12* .12* –.19** –.14*
Alcohol/drugs .12* .15** .22*** .21*** .24***
Companions .14** .13* .13* .14* .12*
Emotional/personal –.20*** –.20*** .11*
Attitude/orientation
Mental health (high = need for treatment)
Self-esteem (high = high self-esteem)
Self-efficacy (high = high self-efficacy) .14** .15** –.13* –.12*
Relationships (high = low codependency) –.15** –.18**

Parental stress (high = no stress) –.18**
Adult Victimization (high = abuse) .17** .16**
Adult emotional abuse .14* .22*** .20**
Adult physical abuse
Adult harassment .15* .14* .16**
Child abuse (high = abuse) .16**
Child physical abuse .19**
571
Note: Y/N = prevalence data; number = frequency data; only significant correlations are shown.
*p < .10. **p < .05. ***p < .01.
Table 4
Partial Correlations (Controlling for Months at Risk) Between Tested
Risk Factors and Technical Violations, Rearrest, and Any Failure
Under Community Supervision (Pearson r, one-tailed)
Any
Technical Violations Rearrest Failure
Risk Factors Y/N Number Y/N Number Y/N
Custody Risk Scale (original)

LSI-R Total Score .14** .17** .18**
Criminal history .13* .14** .14**
Education/employment .13* .15** .13* .20***
Financial .16** .16**
Family/marital
Accommodations .17** .19***
Use of leisure time –.19** –.14*
Alcohol/drugs .20*** .19*** .23***
Companions .11*
Emotional/personal –.20*** –.20*** .11*
Attitude/orientation
Mental health (high = need for treatment)
Self-esteem (high = high self-esteem)
Self-efficacy (high = high self-efficacy)
Relationships (high = low codependency)
Parental stress (high = no stress) –.15*
Adult victimization (high = abuse) .16** .14**
Adult emotional abuse .20*** .18**
Adult physical abuse
Adult harassment .14** .13* .13*
Child abuse (high = abuse)
Child physical Abuse
Note: Y/N = prevalence data; number = indicates frequency data; only significant correla-
tions are shown.
*p < .10. **p < .05. ***p < .01.
p < .10; see Table 3). However, this relationship, weak to begin with,
dropped out of significance once time at risk was controlled (see Table 4).
Gender-responsive needs, such as self-esteem, mental health, and rela-
tionships, were not significantly correlated with the community recidivism
data. In fact, mental health and self-esteem were not significantly related to
any of the correctional outcomes. Moreover, Table 4 finds that some of the
less strongly correlated factors (e.g., use of leisure time) and antisocial

Table 5
A Comparison of Selected Risk Assessment Models (Pearson r, one-tailed)
Serious Prison Technical Any
Misconducts Violations Rearrest Failure
Risk Factor Y/N Number Y/N Number Y/N Number Y/N
Static Criminal History and Current Offense Predictors

Custody scalea
LSI-R Prior History scale .13* .15** .15**
Risk/needs Scale
Modified Custody scaleb .14** .20*** −.15** −.11* .16** .14**
LSI-R Total scale .12* .16** .18** .20*** .21***
Risk/needs Plus Gender Responsive—Institutional
Modified custody and institutional .26*** .29*** .13* .16**
gender-responsive predictorc
LSI-R & institutional gender responsive predictorsd .18** .21*** .16** .19*** .21***
Risk/needs Plus Gender Responsive—Community
LSI-R + Abusee .12* .17** .18** .21*** .22***
Optimal factorsf .18** .14* .25*** .27*** .29***
Note: LSI-R = Level of Service Inventory-Revised; Y/N = indicates prevalence data; number = indicates frequency data.
a. Scale includes factors pertaining to history of institutional violence, severity of current offense, multiple convictions, severity of prior convictions,

prior escapes, prior felonies, age, detainers, and time to serve.
b. Scale includes all factors in the custody scale plus needs pertaining to substance abuse, employment, and education.
c. Scale includes all factors in the modified custody scale plus needs pertaining to relationships, mental health, and child abuse.
d. Scale includes all factors in LSI-R plus needs pertaining to relationships, mental health, and child abuse.
e. Adds total adult abuse scale to LSI-R total scale.
573
f. Adds factors found to be predictive of technical violations or arrests, including criminal history, adult abuse, education/employment, financial,
accommodations, alcohol/drugs, and antisocial companions.
*p < .10. **p < .05. ***p < .01.
companions disappeared or became considerably attenuated once time at

risk was controlled.
How do these factors come together in the form of a composite risk
assessment model? This question was addressed through the examination
of eight different models, which are shown in Table 5. All of the models
sum predictors discussed up to this point according to established or pro-
posed (gender-responsive) approaches to risk assessment. The first two
(under the heading Static Criminal History and Current Offense Predictors)
formulate risk assessment scales that conform to second-generation models
(Bonta, 1996), where risk is predicted by summing scales pertaining to sta-
tic current offense or prior criminal history. As noted above, such models
are common to institutional corrections and were used for some time in
community corrections, parole in particular (see Hoffman, 1994), but have
been gradually replaced by the dynamic risk or needs models. As can be
seen, they do not predict serious misconducts or new offenses for women
offenders. Only a modest correlation between the LSI-R criminal history
scale and technical violations was found (r = .15, p < .05; r = .13, p < .10).
Once the scales begin to incorporate dynamic risk factors, however, this
picture changes. Two models—the Modified Custody scale and the LSI-R
Total scale—consist of static criminal history factors added to dynamic risk
factors that have been included on community and some institutional risk
assessment models for at least the past 15 years. Shown in the third and
fourth rows, these models improve considerably on the static offense-based
models. The LSI-R, in particular, is significantly related to both prison and
community outcomes.
Both the Modified Custody scale and the LSI-R result in better predic-
tors of serious prison misconducts when augmented by gender-responsive
factors pertaining to mental health, relationships, and childhood abuse. This
was not the case, however, in predicting community outcomes. Most
notably, the addition of these factors did not contribute to the predictive
merits of the LSI-R over and above what was found for the LSI-R alone.
Table 5 also reports that neither the custody risk scale nor the two mod-
ified custody scales are strongly predictive of community outcomes. This is
important to note because custody classification systems are often used to
inform community release decisions (e.g., work furloughs, early release,
etc.). The LSI-R performed much better for this purpose. As shown in the
final set of models, it did not improve with the addition of adult abuse, the
only variable noted to be significantly related to a community outcome
variable (see Tables 3 and 4).

This is not to suggest, however, that the LSI-R was the optimal model.
Indeed, removal of three of the nonpredictive LSI-R domains (e.g., emo-
tional or personal, attitudes, and use of leisure time), as well as the substi-
tution of the DOC mental health variable, for a better prediction of clinical
mental health diagnoses, and the addition of the abuse variable created the
more favorable community model of the eight. Controlling for time at risk
did not change the pattern of findings shown in Table 5.
Discussion
This study explored whether an array of gender-responsive needs might

contribute in meaningful ways to the institutional and community classifi-
cation of women offenders. It also examined a more fundamental question:
Is there evidence to suggest that the gender-responsive factors are risk fac-
tors? This was a pilot study, and the results, although mixed, appear to sup-
port continued research on this topic. Just the same, a number of limitations
should be noted. First, the intake sample (N = 156) and the release sample
(N = 134) were small but adequate to the nature of the analysis. Second, the
study measures do not optimally map onto all of the women’s needs identi-
fied in the emerging gender-responsive literature. Two mental heath vari-
ables, for example, aggregate various diagnoses into one measure, whereas
the literature speaks primarily to depression, dual diagnoses, and trauma.
Gender-responsive authors also identify a number of factors we did not test,
including family and relationship conflict as well as housing safety, and a
number of resiliency measures (e.g., support, financial assets). Third, short
follow-up periods may have attenuated the findings. Fourth, many of the
gender-responsive needs and LSI-R needs are dynamic in nature. Because the
needs assessment and LSI-R were administered at prison intake and pretrial,
respectively, they may have changed during the course of the follow-up
period. More proximate measures likely would have produced stronger find-
ings (see Law, 2004). Finally, findings suggesting that the gender-responsive
variables contribute somewhat to existing classification models in no way
suggests that adding patches to existing assessments is the best way to incor-
porate these factors into the assessment technology. It is still possible that
some of the gender-neutral variables (e.g., associates, attitudes, accommoda-
tion) might themselves be better framed for women’s lives; an instrument
designed specifically for women is worthy of consideration.
Even so, the study puts forth some meaningful results. Most important,
perhaps, are findings that child abuse and relationships are associated with

prison adjustment. Additionally, adult victimization, limited self-efficacy,

and parental stress appear to be risk factors for women upon release. With
correctional policy giving strong priority to the treatment of risk factors,
gender-responsive proponents face a daunting struggle advocating for
issues that are prevalent and unfortunate but unrelated to future offending
(Blanchette & Brown, 2006). As in fields of medicine, public health, mental
health, and an array of additional social services, a stronger case can be
made to fund interventions for risk factors than for funding unfortunate
conditions. Just the same, these findings join an extraordinarily small
number of studies on the topic. Especially lacking in this regard is evidence
that the treatment of gender-responsive needs will reduce recidivism.
Examination of the patterns of these results also questions current
evidence-based perspectives on offender rehabilitation as it pertains to
women offenders. The prevailing knowledge base underscores the importance
of interventions targeted to criminal thinking and antisocial companions
(Andrews & Bonta, 2003). The pattern of findings for these women, especially
for the release sample, appears to point to interventions for substance abuse,
education, employment, poverty, victimization, and living conditions.
It is also important to emphasize that these findings shed some doubt on
the wisdom of classifying women solely according to offense-related
attributes. Although it is true that classifying according to static, criminal
history factors is largely confined to institutional corrections, the policy,
nevertheless, is likely to be inflicting unwarranted costs on female inmates.
Indeed, the institutional (custody) risk assessments affect inmates in a mul-
titude of ways because they determine living conditions, distance from
home communities, work assignments, access to programs, and in some
cases early release decisions. Yet we see from this study that needs—both
those identified by the LSI-R and the gender-responsive factors—were far
more predictive of both community recidivism and institutional adjustment
than offense-related variables.
In a number of instances, findings appeared to be conditioned upon envi-
ronment. The juxtaposition of results relative to abuse, self-efficacy,
parental stress, accommodations, and financial well-being appear espe-
cially meaningful in this regard. Many of these—adult abuse, parental
stress, accommodations, financial well-being, education or employment,
and substance abuse—were more important in the community than in insti-
tutions where women were shielded from some of the adversities associated
with these issues. Histories of abuse (child abuse) and mental health were
more troublesome within institutional settings. Modest correlations sug-
gested that for some women, self-confidence caused problems in prison,

likely for its propensity to irritate prison officials, but later helped to
insulate women from new offenses in the community. The fact that the
relationship factor (codependency) predicted in prison but not in the com-
munity may also be attributable to environmental issues. A number of the
serious misconducts involved inmates’ relationships with other inmates
(e.g., fighting with other inmates over a significant other) and therefore
may well have been relevant to measures depicting one’s personal power
while in such relationships. At the same time, parole conditions place
limitations on women’s relationships on release, especially with regard to
antisocial relationships. In another sense, however, this research simply
may not have tapped some of the many ways in which relationships affect
women’s lives. The LSI-R measure of family or marital issues, for example,
is somewhat constrained by social learning factors pertaining to antisocial
influences. It and our measures likely did not adequately map onto dimen-
sions of support, safety, and conflict.
Although we are encouraged by these findings, continued observation of
a new set of risk factors for women will leave much to be sorted out by
policy makers and correctional leaders. Care will need to be taken to assure
that assessments built from findings such as these triage women according
to treatment needs rather than to punishment. For example, feminist scholars
have criticized proponents of evidence-based, best practices, as well as the
authors of dynamic risk assessment instruments for elevating women’s cus-
tody according to their problems rather than the nature of their offenses
(Hannah-Moffat, 2004). Against this prospect, decisions will have to be
made about how we use needs-based risk assessment models and how we
target new risk factors such as abuse, depression, and parental stress.
Resolutions to these issues will need to carefully match risk levels to the
realities of women’s offending. Their rates of recidivism and serious
misconducts are comparatively low in comparison to men. Failure to
accommodate this in the establishment of thresholds for determining risk
levels is likely to further exacerbate problems with overclassification
(Brennan, 1998; Hardyman & Van Voorhis, 2004). Finally, a very careful
delineation of treatment implications will need to follow from assessments
giving more focus to mental health issues and adversity. Sources have iden-
tified a number of potential misuses pertinent to the needs themselves.
Mandatory treatment of abuse victims, increased difficulties with child pro-
tective agencies, and overmedication are clearly far from the treatment rec-
ommendations of the proponents of gender-responsive programming but
are just some of the ways in which accommodating the gender-responsive
risk factors may encounter unintended consequences. Potential mistakes,
however, do not diminish the import of women’s needs.

Notes
1. For ease of presentation, the article refers to both the prison and the community correc-
tional risk prediction instruments as risk assessment instruments. In practice, only the parole
and probation prediction instruments are referred to as risk assessments, whereas correctional
prediction instruments are termed custody assessments. Nevertheless, they are similar in that
both emerged from prediction research that constructed assessments compiling the predictors
or risk factors associated with an outcome behavior…recidivism in the case of parole or pro-
bation samples or serious prison misconducts in the case of prison samples.
2. As will be explained in more detail, the newest generation of risk assessment instru-
ments focus on predictors pertaining to current and criminal history, criminal thinking, crimi-
nal associates, substance abuse, personal distress, residential stability, use of leisure time, and
family issues. They ignore other matters such as parental stress, relationship issues, trauma
and abuse, self-esteem, and self-efficacy, which are core components of the gender-responsive
literature.
3. The risk principle is more than a recommendation for triaging offenders; there is evi-
dence that part of the risk effect is attributable to the fact that intensive interventions introduce
low-risk offenders to criminogenic influences and interrupt family, employment, and other
sources of prosocial stability. Assumptions that the risk principle intended to prevent low risk
offenders from treatment for serious conditions are not entirely accurate. Moreover, a recent
analysis of the needs evidenced by offenders classified as low risk on the LSI-R (Van Voorhis,
Salisbury, & Wright, 2006) found very few low-risk women (4%) with diagnosed needs.
4. Valid assessments do not insure against overclassification; even though women may be
accurately classified relative to each other in terms of risk, their risk may nevertheless not be
comparable to the risk posed by men. Assuming that an assessment is valid, a number of
options exist for reducing overclassification. Full discussion is beyond the scope of the present
study (but see Hardyman & Van Voorhis, 2004).
5. Widom (1989) found similar results for boys.
6. The fourth variable among the big four—criminal personality—is not contained on the
Level of Service Inventory-Revised.
References
Aguilar, R. J., & Nightingale, N. N. (1994). The impact of specific battering experiences on
the self-esteem of abused women. Journal of Family Violence, 9, 35-45.
American Association of Correctional Psychologists Standards Committee. (2000). Standards
for psychology services in jails, prisons, and correctional facilities and agencies. Criminal
Justice and Behavior, 27, 433-493.
American Psychological Association. (1992). Ethical principles of psychologists and code of
conduct. American Psychologist, 47, 1597-1611.
Andrews, D. A. (1983). Assessment of outcome in correctional samples. In M. J. Lambert,
E. R. Christensen, & S. S. DeJulio (Eds.), Assessment of psychotherapy outcome (pp. 160-201).
New York: Wiley.
Andrews, D. A., & Bonta, J. (1995). Level of Service Inventory-Revised. North Tonawanda,
NY: Multi-Health Systems.
Andrews, D. A., & Bonta, J. (2003). The psychology of criminal conduct (3rd ed.). Cincinnati,
OH: Anderson.

Andrews, D. A., Bonta, J., & Hoge, R. D. (1990). Classification for effective rehabilitation:
Rediscovering psychology. Criminal Justice and Behavior, 17, 19-52.
Andrews, D. A., Dowden, C., & Rettinger, L. J. (2001). Special populations within corrections.
In J. A. Winterdyck (Ed.), Corrections in Canada: Social reactions to crime (pp. 170-212).
Toronto: Prentice Hall.
Andrews, D. A., Zinger, I., Hoge, R. D., Bonta, J., Gendreau, P., & Cullen, F. T. (1990). Does
correctional treatment work? A clinically-relevant and psychologically-informed meta-
analysis. Criminology, 28, 369-404.
Antonowicz, D., & Ross, R. (1994). Essential components of successful rehabilitation
programs for offenders. International Journal of Offender Therapy and Comparative
Criminology, 38, 97-104.
Avison, W., Turner, R., & Noh, S. (1986). Screening for problem parenting: Preliminary evi-
dence on a promising instrument. Child Abuse & Neglect, 10, 157-170.
Baunach, P. J. (1985). Mothers in prison. New Brunswick, NJ: Transaction.
Beattie, M. (1987). Codependent no more. New York: Harper/Hazelden.
Belknap, J., Fisher, B., & Cullen, F. T. (1999). The development of a comprehensive measure
of the sexual victimization of college women. Violence Against Women, 5, 185-214.
Belknap, J., & Holsinger, K. (2006). The gendered nature of risk factors for delinquency.
Feminist Criminology, 1, 48-71.
Benda, B. B. (2005). Gender differences in life-course theory of recidivism: A survival analysis.
International Journal of Offender Therapy and Comparative Criminology, 49, 325-342.
Bepko, C., & Krestan, J. (1985). The responsibility trap: A blueprint for treating the alcoholic
family. New York: Macmillan.
Berman, J. (2005). Women offender transition and reentry: Gender responsive approaches to
transitioning women offenders from prison to the community. Washington, DC: USDOJ,
National Institute of Corrections.
Blanchette, K. (1996). The relationship between criminal history, mental disorder, and recidi-
vism among federally sentenced female offenders. Unpublished master’s thesis, Carleton
University, Ottawa, Ontario, Canada.
Blanchette, K. (2004). Revisiting effective classification strategies for women offenders in
Canada. Feminism & Psychology, 14, 231-236.
Blanchette, K. (2005). Field-test of a gender-informed security reclassification scale for
female offenders. Unpublished doctoral dissertation, Carleton University, Ottawa, Ontario,
Canada.
Blanchette, K., & Brown, S. L. (2006). The assessment and treatment of women offenders: An
integrative perspective. Chichester: Wiley.
Blanchette, K., & Motiuk, L. L. (1995). Female offender risk assessment: The case manage-
ment strategies approach. Poster session presented at the Annual Convention of the
Canadian Psychological Association, Charlottetown, Prince Edward Island.
Bloom, B., Owen, B., & Covington, S. (2003). Gender-responsive strategies: Research practice
and guiding principles for women offenders. Washington, DC: U.S. Department of Justice,
National Institute of Corrections.
Bloom, B., Owen, B., & Covington, S. (2005). Gender-responsive strategies for women
offenders: A summary of research, practice, and guiding principles for women offenders.
Washington, DC: U.S. Department of Justice, National Institute of Corrections.
Blume, S. (1997). Women: Clinical aspects. In J. Lowinson et al. (Eds.), Substance abuse:
A comprehensive textbook (pp. 645-654). Baltimore, MD: Williams and Wilkins.
Bonta, J. (1989). Native inmates: Institutional responses, risks, and needs. Canadian Journal
of Criminology, 31, 49-62.

Bonta, J. (1996). Risk-needs assessment and treatment. In A. T. Harland (Ed.), Choosing

correctional options that work: Defining the demand and evaluating the supply (pp. 18-32).
Thousand Oaks, CA: Sage.
Bonta, J., & Motiuk, L. L. (1987). The diversion of incarcerated offenders to correctional
halfway houses. Journal of Research in Crime and Delinquency, 24, 302-323.
Bonta, J., & Motiuk, L. L. (1990). Classification to correctional halfway houses: A quasi-
experimental evaluation. Criminology, 28, 497-506.
Bonta, J., & Motiuk, L. L. (1992). Inmate classification. Journal of Criminal Justice, 20, 343-353.
Bonta, J., Pang, B., & Wallace-Capretta, S. (1995). Predictors of recidivism among incarcer-
ated female offenders. The Prison Journal, 75, 277-294.
Bonta, J., Wallace-Capretta, S., & Rooney, J. (2000). A quasi-experimental evaluation of an
intensive rehabilitation supervision program. Criminal Justice and Behavior, 27, 312-329.
Brennan, T. (1998). Institutional classification of females: Problems and some proposals for
reform. In R. T. Zaplin (Ed.), Female offenders: Critical perspectives and effective inter-
ventions (pp. 179-204). Gaithersburg, MD: Aspen.
Brennan, T., & Austin, J. (1997). Women in jail: Classification issues. Washington, DC: U.S.
Department of Justice, National Institute of Corrections.
Brown, S. L., & Motiuk, L. L. (2005). The Dynamic Factor Identification and Analysis (DFIA)
component of the Offender Intake Assessment (OIA) process: A meta-analytic, psychometric,
and consultative review (Research Report No. R-164). Ottawa, Ontario: Correctional Service
Canada.
Browne, A., Miller, B., & Maguin, E. (1999). Prevalence and severity of lifetime physical and
sexual victimization among incarcerated women. International Journal of Law and
Psychiatry, 22, 301-322.
Bureau of Justice Statistics. (1999). Special report: Women offenders. Washington, DC: U.S.
Department of Justice.
Bureau of Justice Statistics. (2005). Prisoners in 2004. Washington, DC: U.S. Department of
Justice.
Campbell, D., Campbell, J., King, C., Parker, B., & Ryan, J. (1994). The reliability and factor
structure of the Index of Spouse Abuse with African-American women. Violence and
Victims, 9, 259-274.
Carp, S. V., & Schade, L. (1992, August). Tailoring facility programming to suit female
offender’s needs. Corrections Today, pp. 154-158.
Cascardi, M., & O’Leary, K. D. (1992). Depressive symptomatology, self-esteem, and self-
blame in battered women. Journal of Family Violence, 7, 249-259.
Case, P., & Fasenfest, D. (2004). Expectations for opportunities following prison education:
A discussion of race and gender. Journal of Correctional Education, 55, 24-39.
Chandler, S. M., & Kassebaum, G. (1994). Drug-alcohol dependence of women prisoners in
Hawaii. Affilia, 9, 157-170.
Chesney-Lind, M. (1997). The female offender: Girls, women, and crime. Thousand Oaks,
CA: Sage.
Chesney-Lind, M. (2000). What to do about girls? Thinking about programs for young women.
In M. McMahon (Ed.), Assessment to assistance: Programs for women in community
corrections (pp. 139-170). Lanham, MD: American Correctional Association.
Clark, J. (1995). The impact of the prison environment on mothers. The Prison Journal, 75,
306-329.
Clements, C. M., McKee, J., & Jones, S. (1984). Offender needs assessments: Models and
approaches. Washington, DC: U.S. Department of Justice, National Institute of Corrections.

Clements, C. M., Ogle, R., & Sabourin, C. M. (2005). Perceived control and emotional status
in abusive college student relationships: An exploration of gender differences. Journal of
Interpersonal Violence, 20, 1058-1077.
Clements, C. M., Sabourin, C. M., & Spiby, L. (2004). Dysphoria and hopelessness following
battering: The role of perceived control, coping, and self-esteem. Journal of Family
Violence, 19, 25-36.
Coleman, F. L. (1997). Stalking behavior and the cycle of domestic violence. Journal of
Interpersonal Violence, 12, 420-432.
Coulson, G., Ilacqua, G., Nutbrown, V., Giulekas, D., & Cudjoe, F. (1996). Predictive utility
of the LSI for incarcerated female offenders. Criminal Justice and Behavior, 23, 427-439.
Covington, S. (1998). The relational theory of women’s psychological development: Implications
for the criminal justice system. In R. T. Zaplin (Ed.), Female offenders: Critical perspectives
and effective interventions (pp. 113-128). Gaithersburg, MD: Aspen.
Crowley, J. D., & Dill, D. (1992). Silencing the Self scale. Psychology of Women Quarterly,
16, 97-106.
Cullen, F. T., Fisher, B. S., & Applegate, B. K. (2000). Public opinion about punishment and
corrections. In M. Tonry (Ed.), Crime and justice: A review of research (Vol. 27, pp. 1-79).
Chicago: University of Chicago Press.
Dahlberg, L., Toal, S., & Behrens, C. (1998). Measuring violence-related attitudes, beliefs,
and behaviors among youth. Washington, DC: U.S. Department of Health and Human
Services.
Daly, K. (1992). Women’s pathways to felony court: Feminist theories of lawbreaking and
problems of representation. Southern California Review of Law and Women’s Studies, 2, 11-52.
Daly, K. (1994). Gender, crime, and punishment. New Haven, CT: Yale University Press.
Dowden, C., & Andrews, D. A. (1999). What works for female offenders: A meta-analytic
review. Crime & Delinquency, 45, 438-452.
Enos, S. (2001). Mothering from the inside: Parenting in a women’s prison. Albany, NY:
SUNY Press.
Farr, K. A., (2000). Classification for female inmates: Moving forward. Crime & Delinquency,
46, 3-17.
Feeley, M. M., & Simon, J. (1992). The new penology: Notes on the emerging strategy of
corrections and its implications. Criminology, 30, 449-474.
Ferraro, K. J., & Moe, A. M. (2003). Mothering, crime, and incarceration. Journal of Contemporary
Ethnography, 32, 9-40.
Fischer, J., Spann, L., & Crawford, D. (1991). Measuring codependency. Alcoholism Treatment
Quarterly, 8, 87-99.
Fowler, L. T. (1993). What classification for women? In Classification: A tool for managing
today’s offenders. Laurel, MD: American Correctional Association.
Funk, S. J. (1999). Risk assessment for juveniles on probation: A focus on gender. Criminal
Justice and Behavior, 26, 44-68.
Garrett, C. (1985). Effects of residential treatment on adjudicated delinquents: A meta-analysis.
Journal of Research on Crime and Delinquency, 22, 287-308.
Gendreau, P. (1996). The principles of effective intervention with offenders. In A. T. Harland
(Ed.), Choosing correctional options that work: Defining the demand and evaluating the
supply (pp. 117-130). Thousand Oaks, CA: Sage.
Gendreau, P., Little, T., & Goggin, C. (1996). A meta-analysis of the predictors of adult
offender recidivism: What works! Criminology, 34, 575-607.
Gilligan, C. (1982). In a different voice: Psychological theory and women’s development.
Cambridge, MA: Harvard University Press.

Giordano, P. C., Deines, J. A., & Cernkovich, S. A. (2006). In and out of crime: A life-course
perspective on girls’ delinquency. In K. Heimer & C. Kruttschnitt (Eds.), Gender and crime:
Patterns in victimization and offending (pp. 17-40). New York: New York University Press.
Greene, S., Haney, C., & Hurtado, A. (2000). Cycles of pain: Risk factors in the lives of incar-
cerated mothers and their children. The Prison Journal, 80, 3-23.
Hannah-Moffat, K. (2004). Gendering risk at what cost: Negotiations of gender and risk in
Canadian women’s prisons. Feminism and Psychology, 14, 243-249.
Hardyman, P. L., & Van Voorhis, P. (2004). Developing gender-specific classification systems
for women offenders. Washington, DC: U.S. Department of Justice, National Institute of
Corrections.
Harer, M., & Langan, N. (2001). Gender differences in predictors of prison violence:
Assessing the predictive validity of a risk classification system. Crime and Delinquency,
47, 513-536.
Hoffman, P. (1994). Twenty years of operational use of a risk prediction instrument: The
United States Parole Commission’s Salient Factor Score. Journal of Criminal Justice, 22,
477-494.
Holsinger, A. M., Lowenkamp, C. T., & Latessa, E. J. (2003). Ethnicity, gender, and the Level
of Service Inventory-Revised. Journal of Criminal Justice, 31, 309-320.
Holsinger, K., Belknap, J., & Sutherland, J. (1999, November). Assessing the gender specific
program and service needs for adolescent females in the juvenile justice system. Paper pre-
sented to the State of Ohio’s Office of Criminal Justice Services.
Holtfreter, K., & Morash, M. (2003). The needs of women offenders: Implications for correc-
tional programming. Women & Criminal Justice, 14, 137-160.
Houck, K. D. F., & Loper, A. B. (2002). The relationship of parenting stress to adjustment
among mothers in prison. American Journal of Orthopsychiatry, 72, 548-558.
Hubbard, D. J., & Pratt, T. C. (2002). A meta-analysis of the predictors of delinquency among
girls. Journal of Offender Rehabilitation, 34(3), 1-13.
Izzo, R., & Ross, R. (1990). Meta-analysis of rehabilitation programs for juvenile delinquents:
A brief report. Criminal Justice and Behavior, 17, 134-142.
Kampfner, C. J. (1995). Post-traumatic stress reactions in children of imprisoned mothers. In
K. Gabel & D. Johnson (Eds.), Children of incarcerated parents (pp. 21-30). New York:
Lexington.
Kaplan, A. G. (1984). The “self in relation”: Implications for depression in women
(Publication No. 14). Wellesley, MA: Stone Center.
Kazura, K. (2001). Family programming for incarcerated parents: A needs assessment among
inmates. Journal of Offender Rehabilitation, 32, 67-83.
Koons, B. A., Burrow, J. D., Morash, M., & Bynum, T. (1997). Expert and offender percep-
tions of program elements linked to successful outcomes for incarcerated women. Crime
and Delinquency, 43, 515-532.
Kroner, D. G., & Mills, J. F. (2001). The accuracy of five risk appraisal instruments in pre-
dicting institutional misconduct and new convictions. Criminal Justice and Behavior, 28,
471-489.
Larivière, M. (1999). The relationship between self-esteem, criminality, aggression, and vio-
lence: A meta-analysis. Unpublished comprehensive examination, Carleton University,
Ottawa, Ontario, Canada.
Law, M. (2004). A longitudinal follow-up of federally sentenced women in the community:
Assessing the predictive validity of the dynamic characteristics of the Community Intervention
Scale. Unpublished doctoral dissertation, Carleton University, Ottawa, Ontario.

Law, M. A., Sullivan, S. M., & Goggin, C. (in press). Security classification measures for
female offenders and predictors of female criminal conduct: A literature review. Ottawa,
Ontario: Correctional Service Canada.
Lipsey, M. (1992). Juvenile delinquency treatment: A meta-analytic inquiry into the variability
of effects. In T. Cook et al. (Eds.), Meta-analysis for explanation (pp. 83-127). New York:
Russell Sage.
Lipsey, M., Chapman, G., & Landenberger, N. (2001). Cognitive-behavioral programs for
offenders. Annals of the American Academy of Political and Social Science, 578, 144-157.
Lipsey, M., & Wilson, D. (1998). Effective intervention for serious juvenile offenders: A syn-
thesis of research. In R. Loeber & D. Farrington (Eds.), Serious and violent juvenile
offenders: Risk factors and successful interventions (pp. 313-345). Thousand Oaks, CA:
Sage.
Lösel, F. (1995). Increasing consensus in the evaluation of offender rehabilitation? Lessons
from recent research synthesis. Psychology, Crime, & Law, 2, 19-39.
Loucks, A., & Zamble, E. (1999). Predictors of recidivism in serious female offenders: Canada
searches for predictors common to both men and women. Corrections Today, 61, 26-31.
Lowenkamp, C. T., Holsinger, A. M., & Latessa, E. J. (2001). Risk/need assessment, offender
classification, and the role of childhood abuse. Criminal Justice and Behavior, 28, 543-563.
Lowenkamp, C. T., & Latessa, E. J. (2002). Evaluation of Ohio’s community-based correc-
tional facilities and halfway house programs. Cincinnati, OH: University of Cincinnati.
Lowenkamp, C. T., & Latessa, E. J. (2005). Increasing the effectiveness of correctional pro-
gramming through the risk principle: Identifying offenders for residential placement.
Criminology & Public Policy, 4, 263-290.
Lowenkamp, C. T., Latessa, E. J., & Holsinger, A. M. (2006). The risk principle in action:
What have we learned from 13,676 offenders and 97 correctional programs? Crime &
Delinquency, 52, 77-93.
McClellan, D. S., Farabee, D., & Crouch, B. M. (1997). Early victimization, drug use, and
criminality: A comparison of male and female prisoners. Criminal Justice and Behavior,
24, 455-476.
Miller, J. B. (1976). Toward a new psychology of women. Boston: Beacon Press.
Miller, J. B., & Stiver, I. P. (1998). The healing connection: How women form relationships in
therapy and life. Wellesley, MA: Wellesley Centers for Women.
Millon, T. (1997). The Millon Clinical Multiaxial Inventory-III Manual (2nd ed.). Minneapolis:
National Computer Systems.
Morash, M., Bynum, T. S., & Koons, B. A. (1998). Women offenders: Programming needs and
promising approaches. Washington, DC: U.S. Department of Justice, National Institute of
Justice.
Motiuk, M. S., Motiuk, L. L., & Bonta, J. (1992). A comparison between self-report and
interview-based inventories in offender classification. Criminal Justice and Behavior, 19,
143-159.
Murphy, C., & Hoover, S. (1999). Measuring emotional abuse in dating relationships as a mul-
tifactorial construct. Violence and Victims, 14, 39-53.
Northpointe Institute for Public Management. (1997). Correctional Offender Management
Profile for Alternative Sentences. Traverse City, MI: Northpointe Institute for Public
Management.
Olson, D. E., Alderden, M., & Lurigio, A. J. (2003). Men are from Mars, women are from
Venus, but what role does gender play in probation recidivism? Justice Research and
Policy, 5, 33-54.

Orava, T. A., McLeod, P. J., & Sharpe, D. (1996). Perceptions of control, depressive sympto-
matology, and self-esteem of women in transition from abusive relationships. Journal of
Family Violence, 11, 167-186.
Owen, B. (1998). In the mix: Struggle and survival in a woman’s prison. Albany, NY: State
University of New York Press.
Owen, B., & Bloom, B. (1995). Profiling the needs of California’s female prisoners: A needs
assessment. Washington, DC: U.S. Department of Justice, National Institute of Corrections.
Pearson, F. S., Lipton, D. S., Cleland, C. M., & Yee, D. S. (2002). The effects of behav-
ioral/cognitive-behavioral programs on recidivism. Crime & Delinquency, 48, 476-496.
Peters, R. H., Strozier, A. L., Murrin, M. R., & Kearns, W. D. (1997). Treatment of substance
abusing jail inmates: Examination of gender differences. Journal of Substance Abuse
Treatment, 14, 339-349.
Petersilia, J. (2003). When prisoners come home: Parole and prisoner reentry. New York:
Oxford University Press.
Prendergast, M., Wellisch, J., & Falkin, G. (1995). Assessment of services for substance-abusing
women offenders and correctional settings. The Prison Journal, 75, 240-256.
Reisig, M. D., Holtfreter, K., & Morash, M. (2006). Assessing recidivism risk across female
pathways to crime. Justice Quarterly, 23, 384-405.
Resick, P. A. (1993). The psychological impact of rape. Journal of Interpersonal Violence, 8,
223-255.
Rettinger, J. L. (1998). A recidivism follow-up study investigating risk and need within a
sample of provincially sentenced women. Unpublished doctoral dissertation, Carleton
University, Ottawa, Ontario, Canada.
Richie, B. (1996). The gendered entrapment of battered, Black women. London: Routledge.
Rodenberg, F., & Fantuzzo, J. (1993). The measure of wife abuse: Steps toward the develop-
ment of a comprehensive assessment technique. Journal of Family Violence, 8, 203-228.
Roehling, P., & Gaumond, E. (1996). Reliability and validity of the Codependent Questionnaire.
Alcoholism Treatment Quarterly, 14, 85-95.
Rosenberg, M. (1979). The concept of self. New York: Basic Books.
Ross, T., Khashu, A., & Wamsley, M. (2004). Hard data on hard times: An empirical analysis
of maternal incarceration, foster care, and visitation. New York: Vera Institute of Justice.
Rumgay, J. (2004). Scripts for safer survival: Pathways out of female crime. The Howard
Journal of Criminal Justice, 43, 405-419.
Schram, P. J., & Morash, M. (2002). Evaluation of a life skills program for women inmates in
Michigan. Journal of Offender Rehabilitation, 34, 47-70.
Shepard, M., & Campbell, J. (1992). The Abusive Behavior Inventory: A measure of psycho-
logical and physical abuse. Journal of Interpersonal Violence, 7, 291-305.
Sherer, M., Maddus, J., Mercandante, B., Prentice-Dunn, S., Jacobs, B., & Rogers, R. (1982).
The Self-Efficacy Scale: Construction and validation. Psychological Reports, 51, 663-671.
Shields, I. W., & Simourd, D. J. (1991). Predicting predatory behavior in a population of incar-
cerated young offenders. Criminal Justice and Behavior, 18, 180-194.
Siegel, J. A., & Williams, L. M. (2003). The relationship between child sexual abuse and
female delinquency and crime: A prospective study. Journal of Research in Crime and
Delinquency, 40, 71-94.
Task Force on Federally Sentenced Women. (1990). Creating choices: The report of the Task
Force on Federally Sentenced Women. Ottawa, ON: Ministry of the Solicitor General Canada.
Travis, J. (2005). But they all come back: Facing the challenges of prisoner reentry.
Washington, DC: Urban Institute Press.

Tuerk, E. H., & Loper, A. B. (2006). Contact between incarcerated mothers and their children:
Assessing parenting stress. Journal of Offender Rehabilitation, 43, 23-43.
Van Voorhis, P. (2004). An overview of offender classification systems. In P. Van Voorhis,
M. Braswell, & D. Lester (Eds.), Correctional counseling and rehabilitation (5th ed.,
pp.133-160). Cincinnati, OH: Anderson.
Van Voorhis, P., Pealer, J., Spiropoulos, G., & Sutherland, J. (2001). Validation of offender
custody classification and needs assessment systems for incarcerated women offenders in
the (Unidentified State) Department of Corrections. Washington, DC: U.S. Department of
Justice, National Institute of Corrections.
Van Voorhis, P., & Presser, L. (2001). Classification of women offenders: A national assessment
of current practices. Washington, DC: U.S. Department of Justice, National Institute of
Corrections.
Van Voorhis, P., Salisbury, E. & Wright, E. (2006). Is there a bridge between “What Works”
and “Gender Responsive”? Presented at the Annual Meeting of the American Society of
Criminology, Los Angeles, CA.
Washington State Institute for Public Policy. (2003). Washington’s Offender Accountability
Act: An analysis of the Department of Corrections risk assessment. Olympia, WA:
Washington State Institute for Public Policy.
Widom, C. S. (1989). Child abuse, neglect, and violent criminal behavior. Criminology, 27, 251-271.
Williams, S. L., & Mickelson, K. D. (2004). The nexus of domestic violence and poverty.
Violence Against Women, 10, 283-293.
Woititz, J. (1983). Adult children of alcoholics. Hollywood, FL: Health Communications.
Wormith, J. S. (1984). Attitude and behavior change of correctional clientele: A three year
follow-up. Criminology, 23, 326-348.
Zlotnick, C., Johnson, D. M., & Kohn, R. (2006). Intimate partner violence and long-term
psychosocial functioning in a national sample of American women. Journal of Interpersonal
Violence, 21, 262-275.
Emily J. Salisbury is an assistant professor in the Division of Criminology and Criminal

Justice at Portland State University. Her current research interests involve gendered criminal
pathways of women offenders.
Patricia Van Voorhis is a professor in the Division of Criminal Justice at the University of
Cincinnati and director of the University of Cincinnati’s Corrections Institute. She is the prin-
cipal investigator of the National Institute of Corrections Women Offender Classification
Project.
Georgia V. Spiropoulos is assistant professor of Criminal Justice at California State

University, Fullerton. Her research and teaching interests focus on correctional rehabilitation
and risk assessment. She is currently examining the influence of race on correctional treatment
effectiveness.

Salisbury 2008

Uploaded by

Copyright:

Available Formats

Salisbury 2008

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Salisbury 2008

Uploaded by

Copyright:

Available Formats

Crime & Delinquency

Article Volume 55 Number 4

Downloaded from cad.sagepub.com at SIMON FRASER LIBRARY on March 16, 2015

Keywords: women offenders; risk-needs assessment; gender-responsive

C orrectional researchers and practitioners focus their attentions on men

Downloaded from cad.sagepub.com at SIMON FRASER LIBRARY on March 16, 2015

Risk, Needs, and Correctional Policy

Historically, correctional classification differentiated the assessment of

Downloaded from cad.sagepub.com at SIMON FRASER LIBRARY on March 16, 2015

Downloaded from cad.sagepub.com at SIMON FRASER LIBRARY on March 16, 2015

predictors of recidivism. This paradigm is supported by a series of widely

Downloaded from cad.sagepub.com at SIMON FRASER LIBRARY on March 16, 2015

2000). In support, a meta-analysis that sampled a much smaller number of

Classification and Women Offenders

Downloaded from cad.sagepub.com at SIMON FRASER LIBRARY on March 16, 2015

Flaws in the custody assessments have appeared apart from validation

Downloaded from cad.sagepub.com at SIMON FRASER LIBRARY on March 16, 2015

gender-responsive measures (Van Voorhis, Pealer, Spiropoulos, & Sutherland,

Women’s Unique Needs

The emerging literature on gender-responsive strategies for women

Victimization and Abuse

Downloaded from cad.sagepub.com at SIMON FRASER LIBRARY on March 16, 2015

mounting in support of the relationship between child maltreatment and

Downloaded from cad.sagepub.com at SIMON FRASER LIBRARY on March 16, 2015

to measure victimization and (b) predicting different types of recidivism.

Downloaded from cad.sagepub.com at SIMON FRASER LIBRARY on March 16, 2015

Downloaded from cad.sagepub.com at SIMON FRASER LIBRARY on March 16, 2015

positive and negative influences on women offenders (Benda, 2005). In this

Self-Esteem & Self-Efficacy

Downloaded from cad.sagepub.com at SIMON FRASER LIBRARY on March 16, 2015

understudied, although one meta-analysis (12 effect sizes) showed an asso-

Downloaded from cad.sagepub.com at SIMON FRASER LIBRARY on March 16, 2015

The Present Study

The present study sought to understand whether the gender-responsive

Downloaded from cad.sagepub.com at SIMON FRASER LIBRARY on March 16, 2015

of 156 women offenders. Institutional adjustment (i.e., serious prison mis-

Downloaded from cad.sagepub.com at SIMON FRASER LIBRARY on March 16, 2015

Downloaded from cad.sagepub.com at SIMON FRASER LIBRARY on March 16, 2015

Level of Service Inventory-Revised. The LSI-R (Andrews & Bonta,

Downloaded from cad.sagepub.com at SIMON FRASER LIBRARY on March 16, 2015

education or employment, financial situation, family or marital relation-

Institutional Risk Assessment (Custody Classification scale). This is an

Mental health. Two measures of mental health were available to this

Rosenberg Self-Esteem scale. The Rosenberg Self-Esteem scale (Rosenberg,

Sherer Self-Efficacy scale. The Sherer Self-Efficacy scale (Sherer et al.,

Downloaded from cad.sagepub.com at SIMON FRASER LIBRARY on March 16, 2015

Intake Custody Scale 10.7 4.6 2-22

Note: LSI-R = Level of Service Inventory-Revised.

The 15-item Likert-type questionnaire contained questions that were

Downloaded from cad.sagepub.com at SIMON FRASER LIBRARY on March 16, 2015

Parental stress. Modifications were made to a 20-item Likert-type scale

Results of the study are shown in Tables 3 to 5 below. We begin with a

Downloaded from cad.sagepub.com at SIMON FRASER LIBRARY on March 16, 2015

the classification model currently in use in many prisons. Static criminal

Downloaded from cad.sagepub.com at SIMON FRASER LIBRARY on March 16, 2015

Prison Outcome Postrelease Community Outcomes

Custody Risk Scale (original)

Downloaded from cad.sagepub.com at SIMON FRASER LIBRARY on March 16, 2015

Risk Factors Y/N Number Y/N Number Y/N