working with female offenders what does science tell us? kimberly gentry sperber, ph.d
TRANSCRIPT
Working With Female Offenders
What Does Science Tell Us?
Kimberly Gentry Sperber, Ph.D.
Setting the Stage Growing number of women under
correctional supervision Increases in correctional supervision
greater for women than for men Number of women under correctional
supervision increased 81% from 1990-2000; number of men increased 45%.
85% of women under correctional supervision are in the community.
Setting the Stage
Major factor in prison growth for women = drug offenses
Major factor in prison growth for men = violent offenses
What Does This Mean for Practitioners?
More women entering into the system.
Women entering into the system faster than men.
More research being done on women. Push to implement evidence-based
approaches for all offenders More pressure to implement gender-
responsive approaches with females.
So Where Do We Start?
Getting to Know the Women
Characteristics of Female Offenders
More likely to have experienced physical and sexual abuse Estimates range from less than 40% to over
80% depending on the study and the setting.
Depression, anxiety, self-harm more prevalent in women
Dual diagnoses more prevalent More likely to be primary caregiver of a
minor child
Characteristics of Female Offenders
Less likely to have been convicted of a violent offense
When they do commit violent offenses, more likely to involve an intimate or family member (true of girls as well)
More likely to have been convicted of property or drug offense.
Girls more likely to be charged with status offenses. Women more likely to abuse drugs; men more
likely to abuse alcohol (BJS data). Women’s substance abuse has been shown to be
correlated to trauma and mental health issues. Often more likely to be unemployed.
If they are employed, low-skill, low-pay jobs.
Characteristics of Female Offenders
Women often have different physical health issues For example, female inmates more
likely to be HIV positive. Women more likely to have certain
types of cancer (implications for residential settings)
Women have higher rates of STD’s
Characteristics of Female Offenders
Setting Matters Women on probation more likely to have
committed a property offense. Women in prison more likely to have
committed a drug or violent offense. Incarcerated women more likely than men
to have had a family member incarcerated.
Incarcerated women typically have shorter criminal histories than incarcerated men.
Pathways Into Crime
Runaways Substance Abuse Poverty Relationships
Applying the Traditional Model to Females
Risk, Need, Responsivity, Treatment
Risk Principle
Predicting future criminal behavior Matching levels of
treatment/services to the risk level of the offender
Tells us who to target
Factors Correlated to RiskFactor Male FemaleLower class origins .04 (58) .03 (12)
Personal distress/Psychopathology .09 (157) .08 (19)
Personal education/Vocational achievement .11 (96) .13 (7)
Parental/family factors .16 (180) .16 (43)
Temperament/Misconduct/ Personality .18 (461) .23 (38)
Antisocial attitudes/associates .21 (113) .23 (12)
Overall .16 (1065) .16 (131)
Factors Correlated to Risk
Factor Female MaleLower Class Origins .07 .06
Personal Distress/Psychopathology .10 .09
Family Structure/Parent Problems .07 .09
Minor Personality Variables .18 .22
Poor Parent-Child Relations .20 .22
Personal Educ/Vocational Achievement
.24 .23
Temperament/Misconduct/Self-Control
.35 .36
Antisocial Attitudes/Associates .39 .40
The LSI-R and Women Does it work? At least 6 studies demonstrating that
the LSI/LSI-R predicts re-offending in women: Coulson et al. (1996) found the tool
predicted recidivism for women with sentences of less than 2 years.
McConnell (1996) found the tool predicted recidivism for women with sentences of more than 2 years.
Retrospective study
The LSI-R and Women Lowenkamp et al. (2001) found that
the tool predicted reincarceration for both males and females.
Palmer and Hollin (2007) found that the tool predicted reconviction for a sample of English prisoners. Women scored higher on 5 subscales –
accommodations, companions, family/marital, substance abuse, and emotional/personal.
The LSI-R and Women
Folsom and Atkinson (2007) found that the tool predicted recidivism for women with sentences of more than 2 years. Age at 1st arrest and # prior convictions
were strongest predictors. Hubbard (2007) found that the LSI-R
predicted recidivism for both men and women (arrest and incarceration).
The LSI-R and Women
At Talbert House: One study demonstrating that LSI-R
scores positively correlated with recidivism.
Sample comprised of adult women from 3 treatment settings: Residential drug court program Halfway house Jail-based treatment program
Second study demonstrating that higher LSI-R scores predict AWOL’s in halfway house sample.
LSI-R and Women One study that found the LSI-R did not
perform well for women (Reisig et al., 2006): LSI-R misclassified a significant number of
women who were economically marginalized
Misclassified a significant number of drug-connected women
More likely to correctly classify women who looked “male” in terms of offending (behaviors and context)
Comparing the LSI-R to Gender-Responsive Tools
Study comparing traditional risk tools to gender-responsive risk tools (Van Voorhis et al., 2008).
Most of the traditional risk variables were predictive across most of the samples
Criminal attitudes was the least consistent predictor across the samples
Many of the gender-responsive risk variables were also predictive, although there was some variation depending on setting (e.g., community versus institution).
Adding the gender-responsive items to the traditional tools increased predictive power.
Conclusions About the Risk Principle With Women
Women differ in terms of risk to reoffend (as do men)
Service delivery is aided when we can classify women into smaller more homogeneous groups based on shared characteristics (as with men)
The LSI-R does appear to predict re-offending in women (as it does for men)
The LSI-R (and tools like it) are likely NOT capturing all of the variables predictive of reoffending for women;
Therefore, supplemental assessments are likely warranted.
Need Principle Distinguishes between criminogenic needs and
non-criminogenic needs Dynamic risk factors become the criminogenic
needs By definition, these are factors that can change Since they are criminogenic, changing them
changes the probability of re-offending Criminogenic needs become the treatment
targets Tells us what to target
Traditional Criminogenic Needs
Criminogenic Antisocial
Attitudes Antisocial Peers Substance Abuse Employment
Non-Criminogenic Anxiety Low Self-esteem Medical Needs Creative Abilities Physical
Conditioning
Traditional Criminogenic Needs Antisocial Attitudes
Sperber and Smith (forthcoming) – 9 of the 11 subscales of the How I Think Questionnaire positively correlated with the LSI-R score in a sample of halfway house women.
Current clinical norms for the tool not relevant for women, however.
Van Voorhis et al. (2008) – antisocial attitudes not consistently correlated with recidivism (varied by sample).
Antisocial Peers For men, peers often provide a learning opportunity. For
women, antisocial peers are often intimate partners, or a reflection of unhealthy relationships (Van Voorhis).
Regardless of the manifestation, relationships with antisocial others puts both men and women at risk.
Traditional Criminogenic Needs
Substance Abuse Research tends to demonstrate that this is a
strong correlate of offending for women. Often commit crimes to support habit.
Employment Focus is on ability to secure jobs that pay a
decent wage and allow the woman to support herself, and often her minor children.
Housing Additional consideration of housing safety
Criminogenic Need or Responsivity?
Depression Hubbard (2007) – depression was not
correlated with recidivism for both males and females.
Van Voorhis et al. (2008) – depression correlated to recidivism in both community and institutional settings for females.
Benda (2005) – depression correlated to recidivism for women but not for men.
Criminogenic Need or Responsivity?
Self-Esteem Lariviere (1999) found relationship between
low self-esteem and antisocial behavior in women.
Hubbard (2007) – low self-esteem was not correlated to recidivism for both males and females.
Sperber and Carter (forthcoming) – self-esteem was not correlated to recidivism in a sample of adult women.
Self-esteem was correlated to LSI-R score, however (lower self-esteem, higher LSI-R score).
Self-efficacy was correlated to recidivism (higher self-efficacy, lower recidivism)
Self-Esteem vs. Self-Efficacy
GRT contains scales for both self-esteem and self-efficacy
Research question was about the relative importance of self-esteem versus self-efficacy as a predictor of recidivism for women.
Defining Self-Esteem
An individual's sense of his or her value or worth.
The most broad and frequently cited definition of self-esteem within psychology is Rosenberg's (1965), who described it as a favorable or unfavorable attitude toward the self.
Defining Self-Efficacy
The concept of self-efficacy lies at the center of Bandura’s social cognitive theory.
Self-efficacy is a person’s belief in his or her ability to succeed in a particular situation.
GRT at Talbert House Participated in validation study Sample comprised of 147 women
residing in 4 community-based correctional programs Halfway House Residential Drug Court Program Jail-Based Assessment Program Jail-Based Treatment Program
Measures LSI-R Demographics/Clinical Attributes GRT
Self-Esteem Self-Efficacy Relationships Abuse Parenting
Recidivism New charge New conviction
Bivariate Results
Significant positive correlation between self-esteem and self-efficacy.
Both self-esteem and self-efficacy positively significantly correlated with relationships.
Both self-esteem and self-efficacy negatively correlated with abuse.
Bivariate Results
Both self-esteem and self-efficacy were significantly negatively correlated with a number of LSI domains: Education/Employment Leisure/Recreation Alcohol/Drugs Emotional/Personal
Both also negatively correlated with overall LSI composite risk score.
Bivariate Results
Self-Efficacy also significantly negatively correlated with 2 additional LSI domains: Companions Accommodations
Bivariate Correlations With Outcomes
Self-esteem was not significantly correlated to new charge or new conviction post-discharge.
Self-efficacy was significantly correlated with both new charge and new conviction post-discharge.
Women with higher self-efficacy were less likely to recidivate.
Conclusions Self-esteem was significantly correlated
with a number of criminogenic risk domains but not with actual offending. Appears that self-esteem may impact
recidivism indirectly through its impact on these domains.
Self-efficacy was significantly correlated with a larger number of criminogenic risk domains and actual offending. Appears that there is a direct relationship
between self-efficacy and offending in women.
Implications for Practitioners
Study offers some support for idea that self-esteem operates as a responsivity factor, while self-efficacy may be a criminogenic need.
Self-esteem versus self-efficacy: Self-efficacy is easier to target for change
through traditional CBT programs which focus on increasing skills (i.e., increased skills likely lead to improved confidence in one’s ability).
Improvements in self-efficacy are likely to net gains in self-esteem as well given they are positively correlated.
Criminogenic Need or Responsivity?
Lowenkamp et al. (2001) – abuse not related to recidivism for males or females.
UC study examining relationship between child abuse and recidivism for adolescent sample also found no relationship.
Other studies have found a correlation between prior abuse and recidivism for women, however.
Talbert House Abuse Data
GRT validation sample
Abuse correlated with: New Charge New Conviction Days to Re-Conviction
Conclusions About Criminogenic Needs
Focusing on traditional criminogenic needs is still warranted.
Need to understand that they may manifest or originate differently in women than in men.
Need to widen pool of treatment targets to include items that have traditionally been characterized as responsivity issues.
Approaches to Working With Females
CBT Promoted as most effective method for
reducing recidivism. Teaches people to change antisocial
attitudes, to make more responsible choices, how to problem-solve, etc.
Uses skill-based exercises Some have questioned utility for
women
Approaches to Working With Females
Meta-Analysis by Andrews and Dowden (1999) on the principles of effective interventions and females.
Found that programs that followed these principles had greater recidivism reductions. Targeted higher risk offenders for services Focused on traditional criminogenic needs Utilized approaches based on CBT and
structured social learning
Approaches to Working With Females
Study of 4 residential programs using CBT by Hubbard (2007) 2 male, 2 female (Talbert House programs) After controlling for risk, women performed
better than men in terms of recidivism Possible conclusion is that women responded more
favorably to CBT than men Other variables were not predictive of
recidivism: Depression Self-Esteem IQ
Approaches to Working With Females
Relational Approaches Premise that women’s maturity is built
on motivation to build a sense of connection with others.
Self-worth is based on connections to others.
Growth occurs when these relationships are mutual, empathetic, and empowering
Growth is stunted when relationships are abusive or nonmutual.
Approaches to Working With Females
Relational Approaches Significant for women due to amount
of disconnection and violations in their relationships.
Need healthy relationships to change and grow.
Relational context is seen as important in understanding why women offend, their motivations for change, and their needs for reintegration.
Conclusions About Approaches CBT and relational approaches are not
mutually exclusive Relationships can be a focus of CBT
programming (focus on how women choose friends and intimates and consequences of those choices so they can make better choices)
Programs can help foster healthy relationships for women
Examples From the Field Passages Program for Girls
Residential Substance Abuse Program Primary modality is CBT (Corrective Thinking) Milieu management model is Teaching Family
Model (TFM) – social learning theory Recently adopted VOICES curriculum - relational Provides MH counseling Used to provide groups specific to relationships
and self-esteem; re-examining these services. Provides some family therapy and case
management
Relationship Between Intermediate Outcomes and Recidivism
Preliminary Results for Successful Completers Increased self-esteem = 62.5%
Change from pre to post statistically significant
Increased self-efficacy = 61.4% Change from pre to post statistically significant
Reduced antisocial attitudes = 82.5% Change from pre to post statistically significant
Significant changes in family functioning: Cohesion Conflict Intellectual-Cultural Orientation, Moral-Religious Emphasis Organization
Examples From the Field Hamilton County Probation GEMS Unit
Girls Empowered and Motivated to Succeed Girls only caseload Caseloads capped at 25
Staff trained in gender responsivity Provide more intensive interaction with
PO and for longer periods of time Provide girls-only groups (Girls Circle) Focus is on relationship building Specialized risk assessment (HRAF)
Examples From the Field MCC Women’s Center CHANGE Program Offer traditional CBT groups (T4C) Offer Moving On curriculum (Van Dieten);
developed specifically for women DSAT program for women substance abusers Offer Seeking Safety groups for trauma Offer parenting classes Offer college classes Work programs are breaking gender
stereotypes and training women in professions that have higher chance of earning a good wage
Examples From the Field Unique aspects of the unit
Women housed separately from the men. Women were involved in the design of the new
unit from the beginning Women were given tours of the new unit as it was
being built/designed Women were on committees to help structure the
new unit and still run these committees for ongoing operations
Weekly town hall meetings with residents and staff
Residents provide tours to visitors to the unit. All of these give a voice to the women and
represent a relational method of operation. They also help build self-efficacy and self-esteem
of the women.
Operational Issues for Discussion Negative stereotypes of women
Often perceived as more difficult to work with Staff gender Risk management issues
Vulnerability to sexual misconduct of staff Vulnerability to being triggered by staff conduct Some needed services may be outside scope of practice
Training requirements How to choose curricula Resource allocation for traditional model versus
gender-responsive model Resource allocation for services for men and
women (does equitable mean equal?) Reentry considerations
Conclusions There is support for use of the traditional RNRT
model with female offenders. There is a growing body of research that also
recognizes that certain responsivity variables may actually be criminogenic needs for females and therefore warrant more attention.
The application of specific treatment models or curricula is still under-studied for females.
Agencies need to build a strong base of service delivery drawing from both the traditional approach and a gender-responsive approach.