Predictors of Recidivism Over 3 Years Among Substance-Using Women Released From Jail

Abstract

This study examines predictors of recidivism over 3 years for 624 women released from a county jail using a comprehensive range of standardized measures derived from gender-responsive and gender-neutral criminogenic recidivism models. Although more than a dozen factors were related to recidivism in the univariate analysis, the multivariate analysis shows that recidivism can be reliably predicted (area under the curve = 0.90) with just four factors: age, no custody of children, substance use frequency, and number of substance problems. Exploratory analysis of women who recidivated in post-release months 1 to 3, 4 to 12, and 13 to 36 revealed that the effects of several variables (age, super optimism, and number of weeks in the jail treatment program) were dependent on the time elapsed since release from jail, whereas others (substance use and custody) had persistent effects over time. These findings support the development of re-entry services tailored for female offenders who address both gender-responsive and gender-neutral criminogenic risk factors.

Keywords

women offenders predicting recidivism recidivism substance use disorders criminal thinking

Arrest and incarceration rates have dramatically increased for women during the past two decades, particularly for drug-related offenses (Mauer, Potler, & Wolf, 1999). A recent Department of Justice report shows that although the overall population of inmates in county and city jails decreased from 2010 to 2013, the number and proportion of female inmates increased by 10.9% (Minton & Golinelli, 2014). As more attention is focused on understanding the factors related to successful outcomes for women who are released from in-custody settings (Lewis, 2006), the question of whether existing models of risk for recidivism are applicable to women remains open to debate (Braithwaite, Treadwell, & Arriola, 2005; Smith, Cullen, & LaTessa, 2009; Spjeldnes & Goodkind, 2009; Van Voorhis, 2005; Van Voorhis, Wright, Salisbury, & Bauman, 2010).

This article examines the degree to which risk factors measured at entry to jail predict recidivism over 3 years post release among a sample of women who were incarcerated in a large, urban jail and participated in an in-custody drug treatment program. The review of core predictors was derived from both gender-neutral criminogenic and gender-responsive models of recidivism risk. Next, we examine the association of these risk factors with recidivism over a period of 36 months. Because the strength of predictors measured at a specific point in time may change over time, we examine the effects of baseline risk factors across various time intervals (1-3, 4-12, 13-36 months) following release. Our intent is to determine the relative and temporal influences of baseline predictors derived from both gender-neutral/criminogenic and gender-responsive models of recidivism among a sample of women offenders with substance use disorders.

Criminogenic Models of Risk

Models of criminogenic risk for reoffending focus on criminal history (e.g., age of initiation of criminal behavior at first arrest), type of offense, and attitudes that support criminal behavior, referred to as “criminal thinking.” Criminal thinking is hypothesized to be a predictor of criminal behavior; changes in criminal thinking may lead to changes in criminal behavior (Walters, 2006). Criminal thinking styles have been associated with aggression, impulsivity, lack of empathy, violent criminal history, antisocial personality, and inmate misconduct during incarceration (Taxman, Rhodes, & Dumenci, 2011).

The criminogenic approach has led to the development of a correctional treatment paradigm embodied in the Risk-Needs-Responsivity (RNR) model that guides assessment and treatment planning for offenders in correctional settings (Center for Advancing Correctional Excellence [CACE], 2014). The extent to which these models apply to women offenders, however, is unclear (Van Voorhis, 2012). Some measures of criminal thinking do not distinguish between male and female offenders despite substantial gender differences in patterns of criminal behavior involvement. One study of participants in a boot camp intervention found that criminal peer associations and aggressiveness were more strongly associated with recidivism over 5 years in men than in women. In contrast, among women, stress, depression, suicidality, and childhood and recent abuse predicted faster time to recidivism, whereas having more children and better relationships with partners, friends, and family decreased their risk of recidivism (Benda, 2005). Women offenders are more likely to experience shame and guilt related to their behavior, which might mitigate other criminogenic factors (Tangney, Stuewig, Mashek, & Hastings, 2011).

Another study of offenders participating in drug treatment found lower levels of criminal thinking and higher rates of “psychosocial dysfunction” in women than in men (Staton-Tindall et al., 2007). Similarly, a meta-analysis of five large studies found that a risk-needs assessment was generally “gender neutral” across eight risk/needs domains, although the substance abuse factors were more strongly related to the recidivism of female offenders than to that for males (Andrews et al., 2012).

Gender-Responsive Models of Risk

Gender-responsive models of risk stress those factors that may be differentially influential for women, such as mental health status. A majority of incarcerated women (in both prisons and jails) meet criteria for lifetime as well as current mental health and substance use disorders (James & Glaze, 2006; Jordan, Schlenger, Fairbank, & Caddell, 1996; B. M. Pelissier & O’Neil, 2000; Teplin, Abram, & McClelland, 1996). Most have participated in mental health or substance abuse treatment programs prior to their incarceration (Jordan et al., 2002). Moreover, female offenders have higher rates of co-occurring mental and substance use disorders than male offenders (James & Glaze, 2006), particularly mood and anxiety disorders (B. Pelissier & Jones, 2005). One study of jail detainees found higher prevalence of medical conditions, psychiatric disorders, and drug use disorders among women, whereas men had higher rates of alcohol use disorders (Beck, 2012; Binswanger et al., 2010; Maruschak, 2008). A multisite study of jail detainees found that 14.5% of men and 31.0% of women had current, serious mental disorders (Steadman, Osher, Robbins, Case, & Samuels, 2009).

Trauma and abuse increase risk for offending (Hollin & Palmer, 2006), and one quarter to one half of female inmates in state and federal correctional facilities have histories of childhood physical or sexual abuse (General Accounting Office, 1999; Harlow, 1999; Sarteschi & Vaughn, 2010). One study of female offenders in a prison-based treatment program found that they reported significantly higher rates of childhood abuse, neglect, and household dysfunction, leading to negative health outcomes, compared with adult women sampled from a health maintenance organization (Messina & Grella, 2006). In another study that compared women in prison with a matched sample of women in the general population, the former had greater odds for all types of traumas examined, ranging from approximately twice the odds for the sudden death of a family member or a friend to approximately four times the odds of physical or sexual trauma (Grella, Lovinger, & Warda, 2013).

Experiences of trauma and abuse can be precursors to other factors that precipitate women’s involvement in criminal behavior, such as risky sexual behavior and substance abuse. As with women in the general population, numerous studies of women in the criminal justice system have shown that childhood abuse and trauma are associated with elevated levels of mental health problems, including posttraumatic stress disorder (PTSD), substance use disorders, and other behavioral problems (Messina & Grella, 2006; Mullings, Hartley, & Marquart, 2004; Mullings, Marquart, & Brewer, 2000; Zlotnick, 1997). Women might seek and use drugs (in part) to relieve trauma-related distress, leading to their increased likelihood of developing substance use disorders as well as committing drug and drug-related crimes (Anumba, DeMatteo, & Heilbrun, 2012). Recent evidence supports a mediation model, in which a history of victimization and trauma increases vulnerability to stress and mental health problems, which in turn predicts criminal behavior involvement (Anumba et al., 2012; Salisbury & Van Voorhis, 2009; Salisbury, Van Voorhis, & Spiropoulos, 2009). Thus, women’s criminal involvement can be deeply rooted in their histories of childhood trauma and related problems in adulthood, including maladaptive coping, emotional dysregulation, lack of social or family support, abusive relationships, and substance abuse (El-Bassel et al., 1996).

Access to Treatment and Other Resources

Factors associated with poor recidivism outcomes among women offenders include a lack of pre-release planning that provides referrals to needed services in the community (van Olphen, Eliason, Freudenberg, & Barnes, 2009); lack of access to housing and employment (Adams, Leukefeld, & Peden, 2008); lack of linkage with community-based treatment (Alemagno, 2001); and pressures (or the choice) to return to former relationships, neighborhoods, or social networks that precipitate relapse to drug use or crime (Schram, Koons-Witt, Williams, & McShane, 2006; Shaffer, Hartman, & Listwan, 2009). As it is for men, unemployment is a robust predictor of recidivism in studies of women offenders (Matheson, Doherty, & Grant, 2011).

Women who are released from jail with severe psychosocial problems are at high risk for recidivism in the absence of re-entry programs (Singer, Bussey, Song, & Lunghofer, 1995). In a study of detainees leaving jail in New York City, women were more likely to be homeless, use illicit drugs, report drug charges at index arrest, have health problems, and be parents when compared with men (Freudenberg, Moseley, Labriola, Daniels, & Murrill, 2007), leading to more complex re-entry needs. In one study of a community-based sample, substance abuse and risky sexual behaviors were strongly associated with having been incarcerated, having sexual partners who had recently been incarcerated (Khan et al., 2008), and having been diagnosed with a sexually transmitted disease (STD; Rogers et al., 2012). Although these associations were robust for men and women, the associations were particularly strong for women.

Similarly, among women released from prison, returning to drug use and criminal behavior involvement may be attributable to the lack of access to (or lack of sufficient) substance abuse treatment and wraparound services that address their specific service needs (Grella & Greenwell, 2007; Guydish et al., 2011; Oser, Knudsen, Staton-Tindall, & Leukefeld, 2009). In one study, women offenders were more likely to receive services for mental health and substance abuse problems while incarcerated than when they were in the community (Blitz, Wolff, & Paap, 2006). Compounding matters further, they often returned to impoverished communities that were lacking in available services (Blitz, Wolff, Pan, & Pogorzelski, 2005). Even though these women can be stabilized through services provided during incarceration, their inability to successfully access substance use and mental health services after release from jail or prison suggests that their stabilized condition will likely deteriorate during community re-entry.

Current Study

This study examines a conceptual model of risk for recidivism over time using variables measured at intake into jail among women offenders released from jail, following their participation in an in-custody drug treatment program. Variables included in the model are derived from two models of risk. First, gender-neutral factors that typically apply to both men and women are included, such as socioeconomic status (e.g., employment, housing, education, income sources), environmental or community-related factors (e.g., neighborhood crime or poverty), and substance use history and severity. The model also includes factors that are derived from the criminogenic model, such as the history and nature of criminal behavior involvement and indicators of criminal thinking and other personality disorders. Second, factors believed to be especially relevant to women offenders, derived from the gender-responsive model, were also examined. These include parental status (e.g., dependent children, child welfare involvement, loss of child custody) and family functioning, mental health status (symptoms of depression, anxiety, impaired cognitive functioning, behavioral disorders), trauma history (type of exposure, severity, duration, trauma-related symptoms), and relational factors (e.g., substance use or criminal behavior involvement among family/social network).

In addition, our conceptual model posits that the influence of some baseline risk factors varies over time, with those related to basic subsistence needs (e.g., stable housing, employment, income) having greater influence during the initial re-entry period. In contrast, risk factors that are trait-based (e.g., personality characteristics, criminal thinking) or indicative of chronic problems (e.g., child welfare involvement, psychological distress stemming from history of trauma exposure, mental health disorders) would have a more continuous influence on recidivism over time.

The current study sample is drawn from a large study that examined the effectiveness of an early re-intervention model designed to detect and respond to risks for relapse and recidivism following release to the community (Scott & Dennis, 2012). That study demonstrated that relapse rates were high within the initial 90-day period following release from jail; 68% of participants relapsed within 30 days, increasing to 77% within 90 days (Scott & Dennis, 2012). Hence, the analytic model in this current study examines the effects of baseline predictors on recidivism at different time periods: in the initial 90-day period following release, in the subsequent period from 4 to 12 months, and in the remaining period up to 36 months.

Method

Data Source

Data for this article are from the Recovery Management Checkups for Women Offenders (RMC-WO) Experiment (Scott & Dennis, 2012), which included an evaluation of the risk of recidivism for women entering a substance use treatment program run by Cook County Jail. The program focuses on women with substance use issues. More than 90% of the women self-reported criteria that are presumptive of a lifetime substance use disorder, with 74% in the severe range, 9% in the moderate range, and 7% in the mild range. A mix of severity is not unusual for a treatment program.

Participation was voluntary after providing informed consent and conducted in accord with the standards of the Committee on Human Experimentation of the institution in which the experiments were done or in accord with the Helsinki Declaration of 1975. It was also conducted under the supervision of Chestnut Human Subject Institutional Review Board and an independent Data Safety Monitoring Board. All data were collected by independent research staff in confidential interviews and operated under a federal certificate of confidentiality to prevent any subsequent forced disclosure of their answers. Participants were provided with refreshments, allowed to take breaks, and provided an incentive. In addition, the process provided a break from typical jail routine.

Participants

Participants were recruited from the Department of Women’s Justice Services (DWJS) in Illinois’ Cook County Jail, which operates jail- (residential) and furlough- (outpatient) based treatment programs for women offenders with drug problems and nonviolent charges. Cook County Jail is the largest single site jail, and DWJS is one of the largest jail-based treatment programs for women in the United States.

The target population consisted of adult female offenders who were re-entering the community from a county jail substance abuse treatment program. Women were deemed ineligible if they had not used substances in the 90 days before detention, had no substance use disorder symptoms in the year before detention, were below age 18, lived or planned to move outside Chicago within the next 12 months, were fluent in neither English nor Spanish, were cognitively unable to provide informed consent, or were released before their 14th day in DWJS. A total of 866 women were eligible, and 810 (93%) agreed to participate and completed the initial interview. Women who were transferred from the jail to a prison in the Illinois Department of Corrections (n = 186) were excluded due to the extended length of time prior to their release. The final study sample was comprised of 624 women.

The women were predominately African American (83%), between the ages of 21 and 49 (83%), never married (72%), mothers (63%), not employed (84%), and they self-reported criteria sufficient for substance use disorders in their lifetime (90%). Many also had issues around having below a ninth-grade reading level (48%) or mental health disorders (45%). While most were first arrested before the age of 26 (63%) and had previously been incarcerated (99%), few were in the high range on the lifetime criminal screening form (18%) or General Crime Thinking Scale (7%).

Measures

Dependent variable

Recidivism data were drawn from two administrative data systems: Cook County’s Incarceration Management and Cost system, and the State of Illinois’ Law Enforcement Agencies Data System. Recidivism was defined as any new charge after release found in either database. Time to new charges was calculated as the number of days from release from DWJS until the first new charge and was right-censored at the end of Year 3. Time to recidivism was categorized into the following groups: months 1 to 3, 4 to 12, 13 to 36, and no recidivism. Figure 1 shows the hazard curve of time to recidivism in the 36-month period post release with grid lines showing these group cut points, and with no recidivism representing the 30% remaining above the curve at 36 months.

Figure 1:

Hazard Curve for Recidivism (n = 624)

Baseline predictors

Predictors tested in the univariate and multivariate models were selected based on the gender-neutral criminogenic (e.g., RNR) and gender-responsive models of recidivism. A detailed description of all the variables tested in the analyses is available in the appendix. Several standardized assessments were contained in the baseline assessment. These include the following:

A modified version of the Global Appraisal of Individual Needs (GAIN; Dennis, Titus, White, Unsicker, & Hodgkins, 2003) was administered to participants at study intake and contained a broad assessment of background characteristics and psychosocial functioning. Composite scales include the Substance Frequency Scale (SFS; α = .80), Substance Problem Scale (SPS; α = .91), and Illegal Activity Scale (α = .95).

The Family Effectiveness Measure (FEM; McCreary et al., 2013) consists of two scales. The first is a 20-item measure of effective family functioning (EFF; for example, family members help each other out, do things for each other, trust each other). This scale has good internal consistency (α = .96). The second scale is a 16-item measure of ineffective family functioning (IFF; for example, family members break promises, make kids feel bad, find fault with each other; α = .85). Items are summed for a total score for both scales.

Lifestyle Criminality Screening Form (LCSF; Walters & McDonough, 1998; Wexler, Melnick, & Cao, 2004). A total score is derived by adding the scores on 14 items contained in four subscales: irresponsibility, self-indulgence, interpersonal intrusiveness, and social rule breaking.

Psychological Inventory of Criminal Thinking Styles (PICTS; Walters, 2002, 2006, 2007, 2011a, 2011b; Walters & Elliott, 1999; Walters & McDonough, 1998) has well-established reliability and validity for assessing discrete styles of criminal thinking. Subscales include the following: Entitlement, Discontinuity, Current Criminal Thinking, Historical Criminal Thinking, Proactive Criminal Thinking, Reactive Criminal Thinking, Infrequency Factor, Self-Assertion/Deception Factor, and Confusion.

Index arrest and treatment episode: These measures included the most recent charge at admission to DWJS and time (in weeks) in DWJS-based treatment.

Statistical Analysis

Time to recidivism was assessed using Cox regression survival analysis (Cox & Oakes, 1984). This analysis takes into account individual differences in the amount of time observed and censoring when there was no occurrence of the event after 3 years. Odds ratios (ORs) significantly greater than 1.00 indicate that a given predictor is associated with a greater hazard of recidivism and therefore predicts a shorter time to recidivism. ORs significantly less than 1.00 indicate that a given predictor is associated with a lower hazard and therefore predicts a longer time to recidivism. Each predictor was first analyzed individually in the univariate models to determine whether there was a relationship between the predictor and recidivism. Any predictors that were significantly related to recidivism (p < .05) were then included in a multivariate Cox regression model, using a backward stepwise method of entering predictors to derive the final model.

To examine effects of the baseline predictors on recidivism over time, multinomial logistic regression was used to predict recidivism at months 1 to 3, 4 to 12, and 13 to 36, with “no recidivism” as the referent group. As with the survival analysis described above, we first tested each predictor in separate models and then ran a backward stepwise multivariate multinomial logistic regression on the significant predictors from the univariate analyses to derive the final model, retaining significant variables (p < .05).

Receiver operating characteristic (ROC) curve analysis examined how well the final Cox regression model predicted recidivism. The area under the curve (AUC) determined whether the model was significantly different from chance (AUC = 0.50) in predicting recidivism over 36 months. For the multivariate multinomial logistic regression, the agreement rate and kappa between the actual recidivism groups and the predicted recidivism groups were examined. All the analyses were run using IBM SPSS Statistics (version 20.0).

Missing data

Across the predictors, 81% of the 100 predictors had less than 1% of cases with missing values and 96% had less than 5% missing. The literacy level variable was missing for about 21% of the cases and was attributable to missing pages in some copies of the instrument. These missing data were random and were imputed from the regression analyses. The other three predictors with more than 5% missing were as follows: EFF, IFF, and illegal activity scale. Participants skipped out of the two family functioning measures if they reported not having anyone they considered to be their family. It was unclear whether or not values for the illegal activity scale were missing at random. Thus, these measures were retained as is and used only in the univariate analyses because of the loss of sample size that would result in the multivariate analysis due to listwise deletion. In the final multivariate models, n = 602 (96% of the 624 women) were included in the analyses.

Results

Participant Characteristics

Participant characteristics for the study sample overall, as well as broken up by the four-group recidivism variable, are shown in Table 1. Overall, 70.4% of the study sample returned to custody at some point during the 36 months: 18.8% within the first 3 months, 32.4% within 4 to 12 months, and 19.2% within 13 to 36 months; 29.6% did not recidivate. The recidivism groups differed by age, with higher percentages of the older age groups in the 13 to 36 month and no recidivism groups. Overall, 42% of participants had retained custody of one or more of their children, 21% had lost custody of their children, and 37% had no dependent children; however, a greater proportion of participants who had lost custody of their children were in one of the recidivism groups (vs. no recidivism). Significantly fewer women in the no recidivism group reported lifetime dependence on opioids. Age at first arrest was significantly associated with recidivism, with greater proportions of the older age groups at first arrest (age 18-25 and 26+) in the no recidivism group. There was also statistical significance for lifetime dependence on drugs other than alcohol, marijuana, cocaine, or opiates, but this was determined to stem from only two women who had new charges within the first 3 months post release.

Table 1:

Client Demographics and Characteristics

		Time to recidivism groups
	Total sample	1-3 months	4-12 months	13-36 months	No recidivism
	(N = 624; %)	(n = 117; %)	(n = 202; %)	(n = 120; %)	(n = 185; %)	χ²	p value
Race
African American	83	75	85	86	83	13.60	.131
Caucasian	9	12	9	8	6
Hispanic	5	6	4	5	5
Mixed/other	4	7	2	2	7
Age
18-20	7	10	6	8	6	25.04	.014*
21-29	22	26	19	22	22
30-39	26	26	33	23	20
40-49	35	33	37	33	36
50+	11	6	6	15	16
Marriage status
Married/living as	12	10	10	14	15	2.82	.833
Separated/divorced/widowed	16	16	15	16	17
Never married	72	74	74	70	69
No. of children below age 21
None	37	31	38	32	42	16.03	.065
1	17	22	13	24	14
2	18	15	17	20	20
3+	28	32	32	25	24
Custody
No children	37	31	38	32	42	21.04	.011*
Has custody	34	26	34	36	36
No custody	21	35	20	23	14
Mixed custody	8	9	8	9	8
Employment
Full-time	10	10	9	8	14	8.82	.183
Part-time	6	3	5	9	5
Not employed/other	84	87	86	83	81
Literacy level
Less than 6th grade	14	14	12	15	17	5.44	.490
7th-8th grade	33	39	36	28	30
9th grade+	52	47	52	57	53
Substance dependence (lifetime)
Alcohol	14	14	13	15	16	0.71	.874
Marijuana	10	8	12	6	12	5.24	.154
Cocaine	42	51	38	40	41	5.92	.118
Opiate	44	48	50	48	35	10.49	.015*
Other drugs	0	2	0	0	0	8.70	.036*
Mental health disorders (past year)
Any co-occurring	45	49	42	47	43	1.67	.639
Mood disorder	33	37	32	36	31	1.91	.589
Generalized anxiety	12	14	12	8	14	2.09	.554
Traumatic stress	20	21	20	21	20	0.14	.989
Attention deficit/hyperactivity disorder	17	21	15	16	18	2.34	.504
Conduct disorder	16	20	16	12	17	2.97	.397
Borderline personality disorder	38	37	41	35	38	1.43	.696
Antisocial personality disorder	10	14	8	8	10	2.69	.446
Age of first arrest
<18	20	28	17	19	18	12.82	.048*
18-25	43	42	48	36	41
26+	37	29	35	45	41
Times incarcerated (lifetime)^a
0	1	1	1	2	2	12.74	.395
1	29	21	30	27	34
2	19	20	15	20	22
3-9	44	51	46	43	38
10+	7	8	9	8	5
High on lifestyle criminality screening form	18	24	18	18	16	3.08	.373
High on general crime thinking (PICTS)	7	11	7	8	4	5.02	.160

Note. PICTS = Psychological Inventory of Criminal Thinking Styles.

Prior to the arrest for this study.

Recidivism Analysis Using Cox Regression Models

Table 2 shows the means and standard deviations for continuous predictors as well as the overall percentages for the categorical predictors used in the analyses. This table also displays the results of the univariate and multivariate Cox regression analyses, with ORs and 95% confidence intervals (CIs). Continuous predictors (with the exception of those expressed in number of times or weeks) were converted to z scores, resulting in ORs that represent the change in likelihood for each standard deviation–sized change. The univariate analyses showed that older age (OR = 0.89), better family functioning (OR = 0.90), and older age at first arrest (OR = 0.98) all significantly decreased the odds of recidivating over the next 3 years. Factors that significantly increased the odds of recidivism included: not having custody of children (OR = 1.60) or having children in foster care (OR = 1.72); having higher substance frequency (OR = 1.18), more substance problems (OR = 1.17), and more lifetime times incarcerated (OR = 1.02); and having higher scores on the LCSF (OR = 1.11), the Social Rule Breaking Scale (OR = 1.10), the PICTS infrequency factor (OR = 1.10), and self-assertion/deception factor (OR = 1.11). Although literacy level was not a significant predictor in the univariate runs, we included this variable in a separate analysis for participants who had valid responses (excluding the replaced missing values); results were similar to those conducted with the larger sample.

Table 2:

Cox Regressions Predicting Time to Recidivism 3 Years Post Release

			Cox regression
	Total		Univariate		Multivariate
	n	% or M (SD)	OR	95% CI	OR	95% CI
Demographics
African American (vs. other)	624	83%	0.90	[0.71, 1.15]
Age (in decades)	624	3.67 (1.03)	0.89*	[0.82, 0.97]	0.83*	[0.75, 0.92]
Married (vs. not)	623	12%	0.82	[0.61, 1.10]
Number of children below age 21	620	1.80 (2.03)	1.03	[0.99, 1.08]
Lived with own children in past year	623	40%	0.83	[0.69, 1.01]
Lived with other children in past year	623	12%	1.05	[0.79, 1.39]
Custody
No children	229	37%	1.00	N/A
No custody of own children	132	21%	1.60*	[1.24, 2.04]	1.56*	[1.21, 2.01]
Custody of any of own children	259	42%	1.02	[0.82, 1.27]	1.01	[0.81, 1.26]
Any child in foster care past 90 days	621	3%	1.72*	[1.01, 2.94]
Percent of poverty index	615	155.68 (250.1)	1.05	[0.96, 1.13]
Days worked (past 90)	615	7.90 (20.3)	0.91	[0.83, 1.01]
Literacy grade level
6th or less	90	14%	0.90	[0.68, 1.20]
7th-8th	208	33%	1.14	[0.93, 1.40]
High school	326	52%	1.00	N/A
Family
Need help with family’s substance use	619	8%	0.95	[0.67, 1.35]
History of family alcohol problems	622	64%	0.94	[0.77, 1.14]
History of family drug problems	621	65%	0.94	[0.77, 1.14]
History of family psychological problems	616	27%	0.90	[0.72, 1.11]
Days of alcohol use by others in home	621	22.30 (36.71)	0.92	[0.91, 1.09]
Days of drug use by others in home	621	22.53 (37.50)	0.99	[0.90, 1.09]
Effective family functioning^a	564	40.56 (11.82)	0.90	[0.81, 0.99]
Ineffective family functioning^a	564	13.56 (7.02)	0.95	[0.86, 1.04]
Social peer riskiness
Environmental Risk Scale	623	37.59 (7.65)	1.06	[0.97, 1.17]
Living Environmental Risk Index	620	11.46 (3.84)	1.00	[0.91, 1.09]
Vocational Environmental Risk Index	621	11.86 (1.61)	1.02	[0.93, 1.11]
Social Environmental Risk Index	620	14.26 (5.10)	1.10	[0.99, 1.20]
Substance Involvement Index	618	4.01 (4.34)	1.01	[0.98, 1.02]
Illegal Involvement Index	623	1.39 (2.09)	1.03	[0.99, 1.08]
Substance use
Need help with your substance use	620	30%	1.14	[0.93, 1.40]
Age of first substance use	617	17.06 (5.81)	0.99	[0.98, 1.01]
Substance Frequency Scale (90 days)	624	0.36 (0.21)	1.18*	[1.08, 1.30]	1.12*	[1.00, 1.25]
Substance Problem Scale (past year)	624	9.18 (4.97)	1.17*	[1.06, 1.28]	1.14*	[1.01, 1.28]
Mental health (past year)
Internal Mental Distress Scale	624	7.49 (9.06)	0.99	[0.90, 1.09]
Somatic Symptom Index	618	1.00 (1.36)	0.97	[0.88,1.07]
Depressive Symptom Scale	622	2.66 (3.04)	1.02	[0.93, 1.12]
Homicidal/Suicidal Thought Scale	621	0.14 (0.57)	1.01	[0.92, 1.10]
Anxiety and Fear Symptom Scale	621	1.76 (2.75)	0.96	[0.87, 1.06]
Do psychological problems go away with substance use	618	30%	1.13	[0.92, 1.38]
Psychological problems get worse with substance use	618	21%	1.02	[0.81, 1.29]
Happen only after substance use	615	10%	1.29	[0.95, 1.73]
Problems happen even if not using	615	44%	1.02	[0.84, 1.23]
General Victimization Scale	623	3.95 (3.79)	1.05	[0.96, 1.16]
Traumatic Symptom Scale	622	1.97 (3.45)	1.00	[0.91, 1.10]
Age of first trauma exposure
Never	92	15%	1.00	N/A
Below 18	167	28%	0.89	[0.66, 1.20]
Adult	347	57%	0.90	[0.68, 1.18]
Trauma exposure count	622	5.29 (4.18)	0.98	[0.89, 1.07]
Trauma Exposure Severity Subscale	622	10.36 (9.08)	0.98	[0.89, 1.07]
Recency family member sent to jail
Never	417	68%	1.00	N/A
1+ years ago	129	21%	0.97	[0.76, 1.23]
Past year	64	11%	0.95	[0.69, 1.30]
Behavior Complexity Scale	622	4.55 (7.00)	1.04	[0.94, 1.14]
Attention Deficit/Hyperactivity Disorder Scale	622	3.31 (5.23)	1.05	[0.95, 1.15]
Inattentive Disorder Scale	622	2.05 (3.16)	1.06	[0.96, 1.16]
Hyperactivity–Impulsivity Scale	622	1.26 (2.38)	1.03	[0.94, 1.13]
Conduct Disorder Scale	622	1.24 (2.22)	1.01	[0.92, 1.10]
SCID-II
Impulsive personality symptoms	623	5.71 (4.88)	1.03	[0.94, 1.13]
Borderline personality symptoms	623	2.45 (2.18)	1.01	[0.92, 1.11]
Conduct disorder symptom	621	1.09 (2.15)	1.00	[0.91, 1.10]
Antisocial personality symptoms	621	2.19 (1.74)	1.06	[0.97, 1.16]
NEO-I
NEO global score	623	131.21(10.08)	0.99	[0.90, 1.09]
Neuroticism	623	21.32 (5.73)	1.02	[0.93, 1.12]
Extraversion	623	26.75 (4.66)	1.03	[0.93, 1.13]
Openness	623	25.58 (3.66)	1.01	[0.92, 1.10]
Agreeableness	623	27.46 (4.51)	0.98	[0.89, 1.07]
Conscientiousness	623	30.09 (5.34)	0.96	[0.88, 1.05]
Criminal history
Age at first arrest	617	24.95 (8.91)	0.98*	[0.97, 0.99]
Times incarcerated	619	4.00 (5.07)	1.02*	[1.01, 1.04]
Crime and Violence Scale	623	3.55 (4.03)	1.06	[0.96, 1.16]
General Conflict Tactic Scale	621	2.46 (2.80)	1.04	[0.94, 1.14]
General Crime Scale	621	1.10 (1.98)	1.06	[0.97, 1.16]
Illegal Activity Scale	591	0.19 (0.29)	1.10	[1.00, 1.22]
LCSF
Total score	624	7.44 (2.33)	1.11*	[1.01, 1.22]
Irresponsibility score	615	2.17 (1.28)	1.06	[0.98, 1.14]
Self-indulgence score	624	3.23 (0.99)	1.03	[0.93, 1.13]
Criminal thinking (PICTS)
GCT	622	112.80 (30.68)	1.08	[0.99, 1.19]
MO	622	12.10 (4.21)	1.05	[0.96, 1.16]
CO	622	14.37 (5.00)	1.09	[0.99, 1.20]
EN	622	13.02 (4.02)	1.04	[0.95, 1.14]
PO	622	12.35 (4.38)	1.06	[0.97, 1.16]
SN	622	16.11 (3.80)	1.07	[0.97, 1.17]
Interpersonal intrusiveness score	616	0.20 (0.56)	1.04	[0.89, 1.23]
Social rule breaking score	617	1.90 (1.02)	1.10*	[1.01, 1.21]
SO	622	14.14 (4.07)	1.09	[0.99, 1.19]
CI	622	15.58 (4.92)	1.06	[0.97, 1.17]
DS	622	15.11 (5.09)	1.09	[1.00, 1.20]
CUR	622	24.72 (8.03)	1.08	[0.99, 1.19]
HIS_p	622	20.66 (6.96)	1.09	[0.99, 1.20]
P	622	71.71 (22.67)	1.08	[0.99, 1.19]
R	622	82.52 (26.58)	1.09	[0.99, 1.20]
PRB	622	19.37 (6.33)	1.09	[0.99, 1.20]
INF	622	13.23 (4.18)	1.10*	[1.01, 1.20]
AST	622	16.67 (5.75)	1.11*	[1.01, 1.22]
DNH	622	21.13 (4.70)	1.00	[0.92, 1.10]
CF_R	622	11.54 (3.85)	1.07	[0.98, 1.17]
DF_R	622	22.14 (4.62)	0.96	[0.88, 1.05]
FOC	622	14.68 (5.37)	1.02	[0.92, 1.12]
Charges at baseline^b
Violent crime	624	5%	1.00	[0.65, 1.55]
Property crime	624	25%	1.02	[0.82, 1.26]
Substance crime	624	62%	1.00	[0.82, 1.21]
Prostitution	624	6%	1.33	[0.90, 1.96]
Criminal justice violation	624	4%	1.26	[0.79, 1.99]
Other crime	624	8%	0.80	[0.56, 1.15]
Weeks in DWJS	621	12.20 (10.99)	0.99 [0.98, 1.001]

Note. OR = odds ratio; CI = confidence interval; SCID-II = Structured Clinical Interview for Diagnostic and Statistical Manual of Mental Disorders (DSM) for Axis II; NEO-I = Neuroticism–Extroversion–Openness Inventory; LCSF = Lifestyle Criminality Screening Form; PICTS = Psychological Inventory of Criminal Thinking Styles; GCT = general crime thinking; MO = mollification; CO = cutoff; EN = entitlement; PO = power orientation; SN = sentimentality; SO = super optimism; CI = cognitive indolence; DS = discontinuity; CUR = current criminal thinking; HIS_p = historical criminal thinking; P = proactive criminal thinking; R = reactive criminal thinking; PRB = problem avoidance factor; INF = infrequency factor; AST = self-assertion/deception factor; DNH = denial of harm factor; CF_R = confusion-revised; DF_R = defensiveness-revised; FOC = fear-of-change; DWJS = Department of Women’s Justice Services; N/A = not applicable.

Only asked if participant reported having anyone they considered to be family.

Charges from arrest at baseline are not mutually exclusive groups.

p < .05.

The same predictors were examined using a backward stepwise multivariate Cox regression model to determine the unique predictors of time to recidivism over the 36-month period. The predictors that remained in this model were age (OR = 0.83), no custody of children (OR = 1.56), substance frequency (OR = 1.12), and substance problems (OR = 1.14). The ROC analysis of the actual recidivism variable by the predicted survival function revealed the AUC to be 0.90 (95% CI = [0.88, 0.92]). As a check, we ran the multivariate analysis again and included the FEM and illegal activity scale, neither of which remained in the model or affected the outcome. Thus, here we are presenting only the final multivariate model without those two measures.

Recidivism Analysis Using Multinomial Logistic Regression Models

Univariate multinomial logistic regression models were used to examine the effects of baseline predictors on recidivism outcomes in different time periods. The means or percentages of the predictors by the four recidivism groups, as well as the ORs and 95% CIs for the univariate multinomial logistic analyses, are reported in Table 3. Generally, the same predictors that were significant in the Cox regressions were also significant for predicting recidivism in the first 3 months compared with the no recidivism group, with the exception that having children in foster care did not predict recidivism in the first 3 months. Other predictors of recidivism in the first 3 months included: the illegal activity scale (OR = 1.28); from PICTS, the cutoff measure (OR = 1.30), discontinuity (OR = 1.29), current criminal thinking (OR = 1.26), historical criminal thinking (OR = 1.33), proactive criminal thinking (OR = 1.30), reactive criminal thinking (OR = 1.28), confusion-revised (OR = 1.25); prostitution charge at baseline arrest (OR = 2.53); and weeks in the DWJS system (OR = 0.97). The GAIN SFS (OR = 1.29) and SPS (OR = 1.28) and having other charges (mostly ordinance violations) at baseline (OR = 0.43) were predictors of any recidivism in months 4 to 12. Having no custody of one’s children (OR = 2.22) was the only predictor of recidivism in months 13 to 36.

Table 3:

Baseline Characteristics and Odds of Time to Recidivism (Relative to No Recidivism)

		Recidivism groups			No recidivism	Univariate logistic regression
Risk factors	n	1st 3 months	4-12 months	13-36 months		1st 3 months		4-12 months		13-36 months
Risk factors	n	1st 3 months	4-12 months	13-36 months		OR	95% CI	OR	95% CI	OR	95% CI
N	624	117	202	120	185
Demographics
African American (vs. other)	624	75%	85%	86%	83%	0.64	[0.36, 1.12]	1.15	[0.67, 1.98]	1.27	[0.67, 2.40]
Age (in decades)	624	3.44	3.65	3.71	3.81	0.71*	[0.56, 0.89]	0.86	[0.70, 1.04]	0.90	[0.72, 1.13]
Married (vs. not)	623	10%	10%	14%	15%	0.67	[0.32, 1.38]	0.67	[0.37, 1.25]	0.98	[0.51, 1.88]
Number of children below age 21	620	1.98	1.93	1.59	1.67	1.08	[0.96, 1.21]	1.07	[0.97, 1.18]	0.98	[0.87, 1.11]
Lived with own children (past year)	623	32%	39%	42%	44%	0.62	[0.38, 1.00]	0.83	[0.55, 1.24]	0.93	[0.58, 1.48]
Lived with other children (past year)	623	8%	13%	15%	10%	0.73	[0.32, 1.67]	1.35	[0.72, 2.52]	1.56	[0.78, 3.11]
Custody
No children	229	31%	38%	32%	42%	1.00	N/A	1.00	N/A	1.00	N/A
No custody of own children	132	35%	20%	23%	14%	3.47*	[1.83, 6.55]	1.62	[0.90, 2.93]	2.22	[1.14, 4.32]
Custody of any of own children	259	35%	42%	45%	44%	1.07	[0.62, 1.85]	1.06	[0.69, 1.65]	1.34	[1.14, 4.32]
Any child in foster care (90 days)	621	5%	4%	1%	2%	3.28	[0.80, 13.38]	2.20	[0.56, 8.64]	0.52	[0.05, 5.04]
Percent of poverty index	615	201.07	144.36	199.82	118.77	1.18	[0.94, 1.47]	1.00	[0.79, 1.27]	1.04	[0.81, 1.35]
Days worked (past 90)	615	6.16	6.50	8.99	9.82	0.83	[0.65, 1.06]	0.85	[0.70, 1.04]	0.97	[0.78, 1.20]
Literacy grade level
6th or less	90	14%	12%	15%	17%	0.92	[0.46, 1.83]	0.75	[0.42, 1.36]	0.84	[0.43, 1.62]
7th-8th	208	39%	36%	28%	30%	1.46	[0.88, 2.44]	1.20	[0.77, 1.87]	0.88	[0.52, 1.48]
High school	326	47%	52%	57%	53%	1.00	N/A	1.00	N/A	1.00	N/A
Family
Need help with family’s substance use	619	6%	9%	8%	9%	0.68	[0.27, 1.71]	1.03	[0.51, 2.09]	0.86	[0.37, 2.01]
History of family alcohol problems	622	62%	62%	67%	65%	0.87	[0.54, 1.40]	0.87	[0.57, 1.31]	1.08	[0.67, 1.76]
History of family drug problems	621	66%	62%	66%	66%	0.98	[0.60, 1.60]	0.81	[0.54, 1.23]	0.98	[0.60, 1.59]
History of family psychological problems	616	28%	26%	23%	30%	0.92	[0.55, 1.53]	0.80	[0.51, 1.25]	0.68	[0.40, 1.16]
Days of alcohol use by others in home	621	27.79	19.02	21.53	22.89	1.13	[0.91, 1.41]	0.90	0.73, 1.10]	0.96	[0.76, 1.22]
Days of drug use by others in home	621	27.03	20.38	20.68	23.21	1.10	[0.88, 1.37]	0.93	[0.76, 1.13]	0.93	[0.74, 1.18]
Effective family functioning^a	564	38.74	39.61	41.84	41.93	0.76*	[0.60, 0.97]	0.82	[0.66, 1.01]	0.99	[0.77, 1.27]
Ineffective family functioning^a	564	13.57	13.20	13.42	14.05	0.94	[0.74, 1.19]	0.89	[0.72, 1.09]	0.92	[0.72, 1.16]
Social peer riskiness
Environmental Risk Scale	623	38.34	37.64	37.65	37.02	1.19	[0.94, 1.50]	1.09	[0.89, 1.33]	1.09	[0.86, 1.37]
Living Environmental Risk Index	620	11.63	11.31	11.55	11.45	1.05	[0.83, 1.32]	0.96	[0.79, 1.18]	1.03	[0.81, 1.29]
Vocational Environmental Risk Index	621	11.94	11.87	11.80	11.84	1.06	[0.84, 1.33	1.02	[0.83, 1.24]	0.97	[0.77, 1.23]
Social Environmental Risk Index	620	14.78	14.47	14.22	13.74	1.23	[0.97, 1.55]	1.16	[0.95, 1.42]	1.11	[0.87, 1.40]
Substance Involvement Index	618	4.20	3.89	4.17	3.91	1.02	[0.96, 1.07]	1.00	[0.95, 1.05]	1.01	[0.96, 1.07]
Illegal Involvement Index	623	1.51	1.49	1.32	1.24	1.07	[0.95, 1.19]	1.06	[0.96, 1.17]	1.02	[0.91, 1.14]
Substance use
Need help with your substance use	620	35%	30%	30%	27%	1.44	[0.87, 2.38]	1.15	[0.74, 1.79	1.17	[0.70, 1.95]
Age of first substance use	617	16.53	16.92	17.66	17.15	0.98	[0.94, 1.02]	0.99	[0.96, 1.03]	1.01	[0.98, 1.05]
Substance Frequency Scale	624	0.41	0.38	0.32	0.33	1.52*	[1.20, 1.93]	1.29	[1.06, 1.58]	0.98	[0.78, 1.24]
Substance Problem Scale	624	10.27	9.67	8.40	8.45	1.46*	[1.15, 1.85]	1.28*	[1.05, 1.56]	0.99	[0.79, 1.24]
Mental health (past year)
Internal Mental Distress Scale	624	8.15	7.11	7.32	7.61	1.06	[0.85, 1.33]	0.95	[0.77, 1.16]	0.97	[0.77, 1.22]
Somatic Symptom Index	618	1.15	0.88	0.91	1.08	1.05	[0.84, 1.31]	0.86	[0.71, 1.06]	0.89	[0.70, 1.12]
Depressive Symptom Scale	622	2.87	2.55	2.77	2.58	1.10	[0.87, 1.38]	0.99	[0.81, 1.21]	1.07	[0.85, 1.34]
Homicidal/Suicidal Thought Scale	621	0.17	0.11	0.18	0.12	1.08	[0.86, 1.35]	0.98	[0.78, 1.22]	1.09	[0.88, 1.36]
Anxiety and Fear Symptom Scale	621	1.89	1.64	1.63	1.89	1.00	0.80, 1.25]	0.92	[0.75, 1.12]	0.91	[0.72, 1.15]
Do psychological problems go away with use	618	31%	30%	32%	26%	1.24	[0.74, 2.07]	1.22	[0.78, 1.90]	1.33	[0.80, 2.20]
Psychological problems get worse with use	618	25%	20%	16%	21%	1.21	[0.70, 2.09]	0.94	[0.57, 1.54]	0.70	[0.38, 1.29]
Happen only after use	615	13%	10%	12%	7%	2.07	[0.93, 4.60]	1.49	[0.70, 3.16]	1.90	[0.84, 4.26]
Problems happen even if not using	615	43%	41%	51%	42%	1.06	[0.66, 1.70]	0.97	[0.65, 1.46]	1.43	[0.90, 2.28]
General Victimization Scale	623	4.51	3.91	3.66	3.82	1.20	[0.95, 1.50]	1.02	[0.84, 1.25]	0.96	[0.76, 1.21]
Traumatic Symptom Scale	622	2.09	1.94	1.91	1.97	1.03	[0.82, 1.30]	0.99	[0.81, 1.21]	0.98	[0.78, 1.24]
Age of first trauma exposure
Never	92	17%	17%	11%	14%	1.00	N/A	1.00	N/A
Below 18	167	33%	21%	32%	28%	1.02	[0.49, 2.11]	0.62	[0.32, 1.18]	1.45	[0.66, 3.19]
Adult	347	50%	62%	57%	58%	0.76	0.39, 1.50]	0.88	[0.50, 1.57]	1.25	[0.60, 2.61]
Trauma exposure count	622	5.48	4.99	5.55	5.32	1.01	[0.96, 1.07]	0.98	[0.93, 1.03]	1.01	[0.96, 1.07]
Trauma Exposure Severity Subscale	622	10.91	9.45	11.01	10.57	1.04	[0.83, 1.30]	0.88	[0.72, 1.08]	1.05	[0.84, 1.31]
Recency family member in jail
Never	417	66%	72%	65%	68%	1.00	N/A	1.00	N/A
1+ years ago	129	23%	19%	23%	22%	1.12	[0.64, 1.98]	0.82	[0.49, 1.36]	1.09	[0.62, 1.94]
Past year	64	11%	9%	12%	11%	1.11	[0.52, 2.37]	0.82	[0.41, 1.62]	1.21	[0.57, 2.55]
Behavior Complexity Scale	622	5.49	4.36	4.13	4.45	1.15	[0.92, 1.43]	0.99	[0.81, 1.21]	0.95	[0.75, 1.21]
Attention Deficit/Hyperactivity Disorder Scale	622	4.04	3.14	3.10	3.17	1.17	[0.94, 1.46]	1.00	[0.81, 1.22]	0.99	[0.78, 1.25]
Inattentive Disorder Scale	622	2.53	1.92	2.04	1.91	1.20	[0.96, 1.50]	1.00	[0.82, 1.23]	1.04	[0.83, 1.32]
Hyperactivity–Impulsivity Scale	622	1.52	1.22	1.06	1.26	1.10	[0.89, 1.38]	0.98	[0.81, 1.20]	0.91	[0.72, 1.16]
Conduct Disorder Scale	622	1.45	1.22	1.03	1.29	1.07	[0.86, 1.33]	0.97	[0.80, 1.18]	0.88	[0.68, 1.13]
SCID-II
Impulsive personality symptom	623	6.12	5.79	5.29	5.62	1.11	[0.88, 1.39]	1.03	[0.85, 1.26]	0.93	[0.73, 1.18]
Borderline personality symptoms	623	2.41	2.58	2.26	2.44	0.99	[0.78, 1.24]	1.07	[0.87, 1.30]	0.92	[0.73, 1.16]
Conduct disorder symptoms	621	1.42	0.99	0.81	1.14	1.11	[0.90, 1.37]	0.93	[0.76, 1.14]	0.83	[0.64, 1.08]
Antisocial personality symptoms	621	2.29	2.22	2.22	2.07	1.14	[0.90, 1.43]	1.09	[0.89, 1.33]	1.09	[0.87, 1.37]
NEO-I
NEO global score	623	130.81	131.37	131.03	131.39	0.94	[0.75, 1.19]	1.00	[0.82, 1.22]	0.96	[0.77, 1.21]
Neuroticism	623	21.24	21.63	21.03	21.23	1.00	[0.79, 1.26]	1.07	[0.88, 1.31]	0.96	[0.77, 1.22]
Extraversion	623	26.90	26.87	26.44	26.73	1.04	[0.82, 1.31]	1.03	[0.84, 1.26]	0.94	[0.74, 1.19]
Openness	623	25.89	25.26	25.87	25.56	1.10	[0.87, 1.38]	0.92	[0.75, 1.13]	1.09	[0.87, 1.37]
Agreeableness	623	27.41	27.45	27.34	27.58	0.96	[0.76, 1.21]	0.97	[0.80, 1.19]	0.95	[0.75, 1.19]
Conscientiousness	623	29.38	30.15	30.34	30.29	0.84	[0.67, 1.06]	0.97	[0.80, 1.19]	1.01	[0.80, 1.27]
Criminal history
Age at first arrest	617	23.22	24.25	26.22	25.97	0.96*	[0.94, 0.99]	0.98	[0.96, 1.00]	1.00	[0.98, 1.03]
Times incarcerated	619	4.75	4.34	3.75	3.32	1.07*	[1.01, 1.13]	1.06*	[1.003, 1.12]	1.03	[0.97, 1.10]
Crime and Violence Scale	623	4.20	3.49	3.32	3.37	1.21	[0.97, 1.52]	1.03	[0.84, 1.27]	0.99	[0.77, 1.26]
General Conflict Tactic Scale	621	2.71	2.45	2.36	2.39	1.12	[0.89, 1.41]	1.02	[0.84, 1.26]	0.99	[0.78, 1.26]
General Crime Scale	621	1.49	1.03	0.94	1.01	1.23	[0.99, 1.52]	1.01	[0.82, 1.25]	0.96	[0.74, 1.24]
Illegal Activity Scale	591	0.25	0.20	0.15	0.17	1.28*	[1.004, 1.62]	1.12	[0.90, 1.39]	0.88	[0.66, 1.16]
LCSF
Total score	624	7.90	7.44	7.38	7.19	1.36*	[1.07, 1.72]	1.12	[0.91, 1.38]	1.09	[0.86, 1.38]
Irresponsibility score	615	2.30	2.19	2.17	2.05	1.17	[0.97, 1.40]	1.09	[0.93, 1.27]	1.07	[0.90, 1.29]
Self-indulgence score	624	3.32	3.19	3.23	3.21	1.12	[0.89, 1.42]	0.99	[0.81, 1.21]	1.03	[0.82, 1.30]
Interpersonal intrusiveness score	616	0.23	0.18	0.25	0.17	1.20	[0.80, 1.82]	1.02	[0.69, 1.51]	1.24	[0.83, 1.86]
Social rule breaking score	617	2.10	1.94	1.81	1.80	1.34*	[1.07, 1.69]	1.15	[0.94, 1.40]	1.01	[0.80, 1.27]
Criminal thinking (PICTS)
GCT	622	117.97	113.19	110.12	110.83	1.25	[1.00, 1.57]	1.08	[0.89, 1.32]	0.98	[0.77, 1.23]
MO	622	12.38	12.11	12.23	11.84	1.14	[0.90, 1.43]	1.07	[0.87, 1.31]	1.10	[0.87, 1.38]
CO	622	15.31	14.39	14.03	13.98	1.30*	[1.03, 1.63]	1.09	[0.89, 1.33]	1.01	[0.80, 1.28]
EN	622	13.66	12.98	12.58	12.94	1.18	[0.95, 1.48]	1.01	[0.83, 1.23]	0.91	[0.72, 1.15]
PO	622	13.05	12.34	11.96	12.18	1.20	[0.97, 1.50]	1.04	[0.85, 1.27]	0.95	[0.75, 1.20]
SN	622	16.42	16.33	15.79	15.89	1.14	[0.91, 1.43]	1.12	[0.92, 1.36]	0.97	[0.78, 1.22]
SO	622	15.10	14.12	13.52	13.97	1.30*	[1.04, 1.63]	1.04	[0.85, 1.27]	0.89	[0.69, 1.13]
CI	622	16.07	15.71	15.24	15.34	1.16	[0.92, 1.46]	1.08	[0.88, 1.32]	0.98	[0.78, 1.23]
DS	622	15.99	15.20	14.77	14.68	1.29*	[1.02, 1.62]	1.11	[0.91, 1.36]	1.02	[0.87, 1.28]
CUR	622	25.97	24.77	24.38	24.09	1.26*	[1.001, 1.58]	1.09	[0.89, 1.33]	1.04	[0.82, 1.31]
HIS_p	622	22.25	20.59	19.93	20.22	1.33*	[1.06, 1.67]	1.06	[0.86, 1.30]	0.96	[0.75, 1.22]
P	622	76.55	71.61	69.21	70.37	1.30*	[1.04, 1.63]	1.06	[0.86, 1.30]	0.95	[0.75, 1.20]
R	622	86.84	82.77	81.41	80.24	1.28*	[1.02, 1.61]	1.10	[0.90, 1.35]	1.05	[0.83, 1.32]
PRB	622	20.17	19.47	19.31	18.79	1.24	[0.99, 1.57]	1.12	[0.91, 1.36]	1.09	[0.86, 1.37]
INF	622	13.87	13.22	13.31	12.79	1.29*	[1.02, 1.61]	1.12	[0.91, 1.38]	1.14	[0.90, 1.45]
AST	622	17.99	16.72	16.08	16.18	1.36*	[1.08, 1.71]	1.10	[0.90, 1.35]	0.98	[0.77, 1.25]
DNH	622	21.43	21.21	20.52	21.26	1.04	[0.83, 1.30]	0.99	[0.82, 1.20]	0.85	[0.68, 1.07]
CF_R	622	12.24	11.41	11.45	11.29	1.25*	[1.01, 1.56]	1.03	[0.84, 1.27]	1.04	[0.83, 1.32]
DF_R	622	21.52	22.05	22.91	22.14	0.87	[0.69, 1.10]	0.98	[0.80, 1.20]	1.18	[0.94, 1.49]
FOC	622	15.27	14.51	14.46	14.65	1.12	[0.89, 1.41]	0.97	[0.80, 1.19]	0.97	[0.77, 1.22]
Charges at baseline^b
Violent crime	624	4%	4%	6%	4%	0.99	[0.32, 3.10]	1.03	[0.39, 2.73]	1.37	[0.48, 3.88]
Property crime	624	28%	25%	22%	26%	1.12	[0.67, 1.89]	0.94	[0.59, 1.49]	0.79	[0.46, 1.36]
Substance crime	624	59%	64%	63%	62%	0.88	[0.55, 1.41]	1.08	[0.71, 1.63]	1.01	[0.63, 1.63]
Prostitution	624	10%	5%	4%	4%	2.53*	[1.001, 6.39]	1.15	[0.45, 2.99]	0.96	[0.31, 3.01]
Criminal justice violation	624	3%	5%	4%	3%	0.95	[0.22, 4.04]	2.07	[0.71, 6.08]	1.57	[0.44, 5.53]
Other crime	624	9%	4%	9%	10%	0.96	[0.44, 2.12]	0.43*	[0.19, 0.99]	0.94	[0.43, 2.06]
Weeks in DWJS	621	9.91	12.41	12.63	13.13	0.97*	[0.94, 0.99]	1.00	[0.98, 1.01]	1.00	[0.98, 1.02]

Only asked if participant reported having anyone they considered to be family.

Charges from arrest at baseline are not mutually exclusive groups.

p < .05.

Next, multivariate multinomial logistic regression, also using a backward stepwise procedure, was performed to examine the unique effects of each predictor on recidivism within the three time frames (see Table 4). Participant age remained in the model predicting lower odds of recidivism in the first 3 months (OR = 0.67). Not having custody of one’s children increased the odds of recidivism in the first 3 months (OR = 3.73) and in months 13 to 36 (OR = 2.02), but not in the intervening period. The SFS predicted increased odds of recidivism in the first 3 months (OR = 1.52) and in months 4 to 12 (OR = 1.35), but not in the last time period. The PICTS Super Optimism Scale also predicted increased odds of recidivism in the first 3 months (OR = 1.28), while weeks in DWJS (not shown) predicted a decrease in odds of recidivism during the first 3 months (OR = 0.96) only. This final multivariate categorical model, however, was only able to correctly classify 38% of the cases into the correct recidivism groups and thus is not as precise as the continuous models above. As family effectiveness and the illegal activity scales were significant in the univariate runs, we ran the multivariate analysis again allowing these two items to be able to enter the analysis, but neither of these predictors remained in the final model.

Table 4:

Results of Backward Stepwise Multinomial Logistic Regressions Predicting Recidivism in Different Time Periods Following Release From Jail

	Multivariate multinomial logistic regression—backward stepwise
	1st 3 Months		4-12 Months		13-36 Months
Simplified set of risk factors	OR	95% CI	OR	95% CI	OR	95% CI
Age (in decades)	0.67*	[0.52, 0.86]	0.85	[0.69, 1.04]	0.94	[0.74, 1.20]
Custody
No children	1.00	N/A	1.00	N/A	1.00	N/A
No custody of own children	3.73*	[1.90, 7.30]	1.56	[0.85, 2.87]	2.02*	[1.02, 4.00]
Custody of any of own children	1.06	[0.59, 1.90]	1.04	[0.65, 1.64]	1.29	[0.75, 2.21]
Substance Frequency Scale	1.52*	[1.17, 1.98]	1.35*	[1.09, 1.67]	1.02	[0.80, 1.30]
Super Optimism Scale	1.28*	[1.01, 1.63]	0.98	[0.79, 1.21]	0.85	[0.65, 1.09]
Weeks in DWJS	0.96*	[0.94, 0.99]	0.99	[0.98, 1.01]	1.00	[0.98, 1.02]

Note. OR = odds ratio; CI = confidence interval; DWJS = Department of Women’s Justice Services.

p < .05.

Discussion

This study systematically examined baseline predictors of recidivism over 36 months for women released from a county jail, with a focus on determining the relative predictive value of variables derived from both gender-responsive and gender-neutral criminogenic recidivism models. The study revealed that a wide range of predictors from both the gender-responsive and criminogenic models were significant at the univariate level. However, many of these factors were correlated with each other and predicted shared variance in time to recidivism. When examined in a multivariate framework, most of the variables were not necessary or retained in the final models. However, several baseline factors were robust predictors across the cumulative 36-month follow-up period. In addition, the predictive ability of some factors varied across specific time periods, as hypothesized.

Gender-Responsive Predictors

Support for the gender-responsive model was found in the robust relationship between parental status and risk of recidivism. In the multivariate Cox regression model, women who had no custody of their children had an increased risk (by about 50%) of having new charges within 36 months. Moreover, this risk was most pronounced in the initial 90 days following release, with nearly four times the odds of recidivism for women with no custody of their children as compared with women without children. In addition, having a child in foster care in the 90 days prior to arrest (only 3% of the sample) significantly decreased time to recidivism at the univariate level in the survival analyses. Loss of child custody serves as an overall indicator of high risk for recidivism; having a child currently in out-of-home placement is less frequent, but similarly an indicator of high risk.

Hence, a woman’s child custody status might be a proxy for a range of factors that increase vulnerability to recidivism (as indicated by the narrowing of significant variables from the univariate to multivariate models). In a study of gender differences in recidivism among participants in a boot camp intervention, having a greater number of children reduced the risk of recidivism over 5 years for women, but not for men, demonstrating the greater influence of children as a protective effect for women offenders (Benda, 2005). Other studies have found that women offenders who are involved in the Child Welfare System or who have lost custody of their children have greater problem severity in multiple domains, including an earlier age of drug use initiation, unstable economic and housing status, more serious mental health problems, and lack of stable family or social relationships (Grella & Greenwell, 2006; Grella, Hser, & Huang, 2006; Surratt, 2003). However, involvement with child welfare has also been associated with higher motivation for treatment among incarcerated women, suggesting that the possibility of retaining (or regaining) custody of children may be an important motivator for women following their release to the community (Grella & Rodriguez, 2011). In addition, higher scores on the EFF Scale reduced the risk of recidivism by about one third. However, this variable was not retained in the multivariate logistic regression model, thus indicating that the child custody factor might account for variance associated with overall family functioning.

Measures of substance use frequency and problem severity were significant in the univariate models as well as in the multivariate survival analyses. Furthermore, one of these measures (substance use frequency) was retained in the final multivariate logistic regression model examining time-specific effects. That severity of substance use problems was a robust predictor may reflect the nature of the sample, which was recruited from an in-custody drug treatment program. Of note, the effects of substance use severity were attenuated after 12 months and were no longer significant in either univariate or multivariate models. Clinically, this finding suggests that the drug treatment interventions may be most effective in the immediate re-entry period, when women are most at risk for relapse.

A charge of prostitution at baseline was significantly associated with increased risk of recidivism in the initial 90-day period in the univariate logistic regression model, with odds increasing by over 2.5 times. Other research has indicated that involvement with sex work is indicative of greater instability among female offenders, and is associated with lack of housing and family support (McLean, Robarge, & Sherman, 2006; Millay, Satyanarayana, O’Leary, Crecelius, & Cottler, 2009). Thus, female offenders who have a history of sex work might face particular challenges related to their lack of economic resources and supportive relationships, higher levels of psychological distress (El-Bassel et al., 1997), and higher risks for exposure to trauma and violence. Such women are particularly vulnerable to recidivism in the absence of re-entry programs or other supportive services. Nevertheless, this variable was not retained in the multivariate models, suggesting that the risk factors associated with prostitution co-vary with other predictors entered into the models.

It is noteworthy that none of the variables related to trauma exposure, trauma severity, or mental health status reported at baseline were predictive of recidivism in this sample. This finding is in sharp contrast to much of the current literature that emphasizes these factors as correlates of criminal involvement among women that merit attention in treatment models based on women’s risk/needs. Moreover, this finding runs counter to gender-responsive models of risk among female offenders (McKeown, 2010) that have emphasized the prevalence of trauma exposure and the association of trauma history with subsequent mental health and substance use disorders, which act as precursors to women’s criminal behavior and enhance the risks of recidivism (Moloney, van den Bergh, & Moller, 2009; Tripodi & Pettus-Davis, 2013). In addition, fewer than half of the women in this sample met criteria for an Axis I mental disorder (45%), whereas 38% met criteria for borderline personality disorder (BPD) and only 10% for antisocial personality disorder (ASPD). These rates are inconsistent with those found in previous studies indicating higher prevalence for global mental health disorders, including the personality disorders that are most prevalent among offenders.

Criminogenic Predictors

With regard to criminal thinking indicators, nine of the PICTS scales were significant at the univariate level, as were the Total Score of the LCSF and the Social Rule Breaking Scale. These results provide support for the criminogenic model of risk, indicating that cognitive patterns and beliefs are associated with greater risk for recidivism. Only one of these measures (super optimism) was retained in the multivariate logistic regression model within the initial 90 days after release. This scale measures the degree of adherence to beliefs that one can avoid the negative consequences that are normally associated with criminal behavior. Thus, this risk factor indicates lack of realistic appraisal of risks associated with criminal behavior, which can propel a more rapid return to criminal behavior and re-arrest. However, given the shared variance among the related criminal thinking indicators, this finding should not be interpreted to exclude the relevance of the initially significant indicators given the possibility of a “suppression effect” within a multivariate context.

Three indicators of earlier initiation or greater severity of criminal involvement were significant in the univariate logistic regression models, thus supporting the criminogenic model. These include age at first arrest, number of times incarcerated, and the Illegal Activity Scale score, though the influence of these predictors varied depending on the time frame examined in the multivariate models.

Age was also a robust predictor of recidivism, with each additional decade of age associated with reduced risk of recidivism. This effect of age in reducing recidivism for both men and women has been found in other studies (Prendergast, Huang, Evans, & Hser, 2010). However, results from the time-specific multivariate logistic regression model showed that this effect was most pronounced in the initial 90-day period, when the odds of recidivism were reduced by one third per decade of age; this effect was nonsignificant in the two later time periods.

Conclusion

This study sought to fill a gap in the literature on the predictive factors of recidivism among female offenders released from jail. Administrative data on arrest charge, time for treatment, recidivism and self-report data from a comprehensive assessment battery comprised of standardized measures were analyzed. The study included a comprehensive assessment battery with several measures administered quarterly over 3 years, very high follow-up rates, and employed a systematic set of analyses to discern the effects of a large set of variables, both singly and combined, given the paucity of current research that has identified predictors of recidivism among women offenders with substance use problems in general and specifically among those coming out of jail. The study’s findings are limited by the specific characteristics of the study sample. Although drawn from a large, urban area, the characteristics of women arrestees who were referred to treatment in this setting might reflect selection processes unique to this setting as well as characteristics of the surrounding community. Nevertheless, the findings yield important information on predictors of recidivism that merit replication in other settings.

The study found support for incorporating both gender-responsive influences (such as parental status and prostitution) and gender-neutral criminogenic risks that are embodied in the RNR model (such as indicators of criminal thinking) in a model of recidivism among women offenders with substance abuse problems. Importantly, the study found that the effects of several variables (e.g., age, Illegal Activity Scale, social rule breaking score) were dependent on the time elapsed since release from jail, whereas others (e.g., substance use severity and parental status) had more persistent effects over time. Overall, these findings support the development of re-entry services tailored for female offenders that address both gender-responsive and gender-neutral criminogenic risk factors (Wright, Van Voorhis, Salisbury, & Bauman, 2012). Moreover, the effects of several variables in the immediate re-entry period support the development of transition planning that aims to limit gaps in services for women upon release and their referral to treatment and other services. These results, combined with the increased representation of women in the inmate population over time (Minton & Golinelli, 2014), serve as evidence of the need to develop re-entry services for female inmates.

Finally, the diminishing ability of baseline factors to predict recidivism after more than a year suggests the need for future work to examine other time-varying factors over the 3-year period to further improve the model. This is also consistent with the view of substance use as a chronic condition that needs longer term management, not just brief intervention.

Footnotes

Appendix

Acknowledgements

The authors would like to thank the women who participated in the interviews, the Cook County Jail Division 17 staff who accommodated the researchers and helped extract all of the recidivism data, and Art Lurigio and Janet Titus for feedback on the article. The opinions are those of the authors and do not reflect official positions of the government.

This article was supported by National Institute on Drug Abuse (NIDA) Grants DA021174 and DA016383.

Christy K Scott, PhD, is the director of the Lighthouse Institute and the Illinois Survey Laboratory at Chestnut Health Systems in Chicago, Illinois. Her research focuses on understanding and predicting how people move through cycles of substance use crime, treatment, incarceration, and periods of recovery, as well as how to experimentally test strategies for improving recovery management over time. She has published widely on recovery management, how to achieve better than 90% follow-up rates in longitudinal studies, intensive data collection with smartphones, and developed and tested different interventions for managing addiction over time.

Christine Grella, PhD, is professor-in-residence in the Department of Psychiatry and Biobehavioral Sciences at University of California, Los Angeles (UCLA) and associate director of the UCLA Integrated Substance Abuse Programs. Her research focuses on the intersection of multiple service delivery systems, including substance abuse treatment, mental health, child welfare, health services, HIV services, and criminal justice, and the relationship of service delivery with treatment outcomes. She has published her work widely in the areas of addiction, mental health, health services, and evaluation research.

Michael L. Dennis, PhD, is a senior research psychologist at the Lighthouse Institute and director of the GAIN Coordinating Center at Chestnut Health Systems in Normal, Illinois. His research also focuses on understanding and predicting how people move through cycles of substance use and crime, treatment, incarceration and periods of recovery, as well as how to experimentally test strategies for improving recovery management over time. He has published widely on recovery management, integrating clinical and research assessment, measurement, intensive data collection with smartphones, and evaluation research.

Rodney R. Funk is a research associate at the Lighthouse Institute of Chestnut Health Systems in Normal, Illinois. He holds SPSS’s highest certification for data mining methods and modeling. His research interests are in the analysis of complex hierarchical and intensive data analytic models, as well as using clustering techniques to interpret heterogeneous samples and trends over time.

References

Adams

Leukefeld

C. G.

Peden

A. R.

(2008). Substance abuse treatment for women offenders: A research review. Journal of Addictions Nursing, 19, 61-75. doi:10.1080/10884600802111648

Alemagno

S. A.

(2001). Women in jail: Is substance abuse treatment enough? American Journal of Public Health, 91, 798-800. doi:10.2105/AJPH.91.5.798

American Psychiatric Association. (1994). Diagnostic and statistical manual of mental disorders (4th ed.). Washington, DC: Author.

American Psychiatric Association. (2000). Diagnostic and statistical manual of mental disorders (4th ed.; text rev.). Washington, DC: Author.

Andrews

D. A.

Guzzo

Raynor

Rowe

R. C.

Rettinger

L. J.

Brews

Wormith

J. S.

(2012). Are the major risk/need factors predictive of both female and male reoffending? A test with the eight domains of the level of service/case management inventory. International Journal of Offender Therapy and Comparative Criminology, 56, 113-133. doi:10.1177/0306624X10395716

Anumba

DeMatteo

Heilbrun

(2012). Social functioning, victimization, and mental health among female offenders. Criminal Justice and Behavior, 39, 1204-1218. doi:10.1177/0093854812443489

Beck

(2012). PREA data collection activities. Washington, DC: Bureau of Justice Statistics.

Benda

B. B.

(2005). Gender differences in life-course theory of recidivism: A survival analysis. International Journal of Offender Therapy and Comparative Criminology, 49, 325-343. doi:10.1177/0306624X04271194

Binswanger

I. A.

Merrill

J. O.

Krueger

P. M.

White

M. C.

Booth

R. E.

Elmore

J. G.

(2010). Gender differences in chronic medical, psychiatric, and substance-dependence disorders among jail inmates. American Journal of Public Health, 100, 476-482. doi:10.2105/AJPH.2008.149591

10.

Blitz

C. L.

Wolff

Paap

(2006). Availability of behavioral health treatment for women in prison. Psychiatric Services, 57, 356-360. doi:10.1176/appi.ps.57.3.356

11.

Blitz

C. L.

Wolff

Pan

K. Y.

Pogorzelski

(2005). Gender-specific behavioral health and community release patterns among New Jersey prison inmates: Implications for treatment and community reentry. American Journal of Public Health, 95, 1741-1746. doi:10.2105/AJPH.2004.059733

12.

Braithwaite

Treadwell

Arriola

(2005). Health disparities and incarcerated women: A population ignored. American Journal of Public Health, 95, 1679-1681. doi:10.2105/AJPH.2005.065375

13.

Center for Advancing Correctional Excellence. (2014). Risk-Needs-Responsivity (RNR) simulation tool. Fairfax, VA: George Mason University. Retrieved from http://www.gmuace.org/research_rnr.html

14.

Costa

P. T.

McCrae

R. R.

(1992). Revised NEO Personality Inventory (NEO PI-R) and NEO Five-Factor Inventory (NEO-FFI). Odessa, FL: Psychological Assessment Resources, Inc.

15.

Cox

D. R.

Oakes

D. O.

(1984). The analysis of survival data. London, England: Chapman & Hall.

16.

Davis

T. C.

Long

S. W.

Jackson

R. H.

Mayeaux

E. J.

George

R. B.

Murphy

P. W.

Crouch

M. A.

(1993). Rapid estimate of adult literacy in medicine: A shortened screening instrument. Family Medicine, 25, 391-395.

17.

Dennis

M. L.

Titus

J. C.

White

Unsicker

Hodgkins

(2003). Global appraisal of individual needs (GAIN): Administration guide for the GAIN and related measures (Version 5). Bloomington, IL: Chestnut Health Systems. Available from http://www.gaincc.org

18.

El-Bassel

Gilbert

Ivanoff

Schilling

R. F.

Borne

Safyer

S. F.

(1996). Correlates of crack abuse among incarcerated women: Psychological trauma, social support and coping behavior. American Journal of Drug and Alcohol Abuse, 22, 41-56.

19.

El-Bassel

Schilling

R. F.

Irwin

K. L.

Faruque

Gilbert

Von Bargen

. . .Edlin

B. R.

(1997). Sex trading and psychological distress among women recruited from the streets of Harlem. American Journal of Public Health, 87, 66-70. doi:10.2105/AJPH.87.1.66

20.

First

M. B.

Spitzer

Gibbon

Williams

(1996). Structured clinical interview for DSM-IV. New York: Psychiatric Institute, Biometrics Research Department.

21.

Freudenberg

Moseley

Labriola

Daniels

Murrill

(2007). Comparison of health and social characteristics of people leaving New York City jails by age, gender, and race/ethnicity: Implications for public health interventions. Public Health Reports, 122, 733-743.

22.

General Accounting Office. (1999). Women in prison: Issues and challenges confronting U.S. correctional systems (Report No. GAO/GGD-00-22). Washington, DC: Author.

23.

Grella

C. E.

Greenwell

(2006). Correlates of parental status and attitudes toward parenting among substance-abusing women offenders. The Prison Journal, 86, 89-113. doi:10.1177/0032885505283925

24.

Grella

C. E.

Greenwell

(2007). Treatment needs and completion of community-based aftercare among substance-abusing women offenders. Women’s Health Issues, 17, 244-255. doi:10.1016/j.whi.2006.11.005

25.

Grella

C. E.

Hser

Y. I.

Huang

Y. C.

(2006). Mothers in substance abuse treatment: Differences in characteristics based on involvement with child welfare. Child Abuse and Neglect, 30, 55-73. doi:10.1016/j.chiabu.2005.07.005

26.

Grella

C. E.

Lovinger

Warda

(2013). Relationships among trauma exposure, familial characteristics, and PTSD: A case-control study of women in prison and in the general population. Women & Criminal Justice, 23, 63-79. doi:10.1080/08974454.2013.743376

27.

Grella

C. E.

Rodriguez

(2011). Motivation for treatment among women offenders in prison-based treatment and longitudinal outcomes among those who participate in community aftercare. Journal of Psychoactive Drugs, 43(Suppl. 1), 58-67. doi:10.1080/02791072.2011.602275

28.

Guydish

Chan

Bostrom

Jessup

M. A.

Davis

T. B.

Marsh

(2011). A randomized trial of probation case management for drug-involved women offenders. Crime & Delinquency, 57, 167-198. doi:10.1177/0011128708318944

29.

Harlow

C. W.

(1999). Prior abuse reported by inmates and probationers (Bureau of Justice Statistics Selected Findings NCJ 172879). Washington, DC: U.S. Department of Justice, Office of Justice Programs.

30.

Hollin

C. R.

Palmer

E. J.

(2006). Criminogenic need and women offenders: A critique of the literature. Legal and Criminological Psychology, 11, 179-195. doi:10.1348/135532505X57991

31.

James

Glaze

(2006). Mental health problems of prison and jail inmates (NCJ 213600). Washington, DC: Bureau of Justice Statistics.

32.

Jordan

B. K.

Federman

E. B.

Burns

B. J.

Schlenger

W. E.

Fairbank

J. A.

Caddell

J. M.

(2002). Lifetime use of mental health and substance abuse treatment services by incarcerated women felons. Psychiatric Services, 53, 317-325. doi:10.1176/appi.ps.53.3.317

33.

Jordan

B. K.

Schlenger

W. E.

Fairbank

J. A.

Caddell

J. M.

(1996). Prevalence of psychiatric disorders among incarcerated women. II. Convicted felons entering prison. Archives of General Psychiatry, 53, 513-519. doi:10.1001/archpsyc.1996.01830060057008

34.

Khan

M. R.

Wohl

D. A.

Weir

S. S.

Adimora

A. A.

Moseley

Norcott

. . . Miller

W. C.

(2008). Incarceration and risky sexual partnerships in a southern US city. Journal of Urban Health, 85, 100-113. doi:10.1007/s11524-007-9237-8

35.

Kilpatrick

D. G.

Acierno

Resnick

H. S.

Saunders

B. E.

Best

C. L.

(1997). A 2-year longitudinal analysis of the relationships between violent assault and substance use in women. Journal of Consulting and Clinical Psychology, 65, 834-847. doi:10.1037/0022-006X.65.5.834

36.

Kimerling

Calhoun

K. S.

Forehand

Armistead

Morse

Clark

(1999). Traumatic stress in HIV-infected women. AIDS Education and Prevention, 11, 321-330.

37.

Lewis

(2006). Treating incarcerated women: Gender matters. Psychiatric Clinics of North America, 29, 773-789. doi:10.1016/j.psc.2006.04.013

38.

Maruschak

(2008). Medical problems of prisoners. Washington, DC: Bureau of Justice Statistics.

39.

Matheson

F. I.

Doherty

Grant

B. A.

(2011). Community-based aftercare and return to custody in a national sample of substance-abusing women offenders. American Journal of Public Health, 101, 1126-1132. doi:10.2105/AJPH.2010.300094

40.

Mauer

Potler

Wolf

(1999). Gender and justice: Women, drugs and sentencing policy. Washington, DC: The Sentencing Project.

41.

McCreary

L. L.

Conrad

K. M.

Conrad

K. J.

Scott

C. K.

Funk

R. R.

Dennis

M. L.

(2013). Using the Rasch measurement model in psychometric analysis of the Family Effectiveness Measure. Nursing Research, 62, 149-159. doi:10.1097/NNR.0b013e31828eafe6

42.

McHugo

G. J.

Caspi

Kammerer

Mazelis

Jackson

E. W.

Russell

Kimerling

(2005). The assessment of trauma history in women with co-occurring substance abuse and mental disorders and a history of interpersonal violence. Journal of Behavioral Health Services and Research, 32, 113-127. doi:10.1007/BF02287261

43.

McKeown

(2010). Female offenders: Assessment of risk in forensic settings. Aggression and Violent Behavior, 15, 422-429. doi:10.1016/j.avb.2010.07.004

44.

McLean

R. L.

Robarge

Sherman

S. G.

(2006). Release from jail: Moment of crisis or window of opportunity for female detainees? Journal of Urban Health, 83, 382-393. doi:10.1007/s11524-006-9048-3

45.

Messina

Grella

C. E.

(2006). Childhood trauma and women’s health in a California prison population. American Journal of Public Health, 96, 1842-1848. doi:10.2105/AJPH.2005.082016

46.

Millay

T. A.

Satyanarayana

V. A.

O’Leary

C. C.

Crecelius

Cottler

L. B.

(2009). Risky business: Focus-group analysis of sexual behaviors, drug use and victimization among incarcerated women in St. Louis. Journal of Urban Health, 86, 810-817. doi:10.1007/s11524-009-9381-4

47.

Minton

T. D.

Golinelli

(2014). Jail inmates at mid-year 2013: Statistical tables. Washington, DC: Bureau of Justice Statistics, Office of Justice Programs, U.S. Department of Justice.

48.

Moloney

K. P.

van den Bergh

B. J.

Moller

L. F.

(2009). Women in prison: The central issues of gender characteristics and trauma history. Public Health, 123, 426-430. doi:10.1016/j.puhe.2009.04.002

49.

Mullings

J. L.

Hartley

D. J.

Marquart

J. W.

(2004). Exploring the relationship between alcohol use, childhood maltreatment, and treatment needs among female prisoners. Substance Use & Misuse, 39, 277-305.

50.

Mullings

J. L.

Marquart

J. W.

Brewer

V. E.

(2000). Assessing the relationship between child sexual abuse and marginal living conditions on HIV/AIDS-related risk behavior among women prisoners. Child Abuse and Neglect, 24, 677-688. doi:10.1016/S0145-2134(00)00127-7

51.

Oser

Knudsen

Staton-Tindall

Leukefeld

(2009). The adoption of wraparound services among substance abuse treatment organizations serving criminal offenders: The role of a women-specific program. Drug and Alcohol Dependence, 103(Suppl. 1), S82-S90. doi:10.1016/j.drugalcdep.2008.12.008

52.

Pelissier

Jones

(2005). A review of gender differences among substance abusers. Crime & Delinquency, 51, 343-372. doi:10.1177/0011128704270218

53.

Pelissier

B. M.

O’Neil

J. A.

(2000). Antisocial personality and depression among incarcerated drug treatment participants. Journal of Substance Abuse, 11, 379-393. doi:10.1016/S0899-3289(00)00034-1

54.

Prendergast

Huang

Evans

Hser

Y. I.

(2010). Are there gender differences in arrest trajectories among adult drug abuse treatment participants. Journal of Drug Issues, 40, 7-26. doi:10.1177/002204261004000102

55.

Rogers

S. M.

Khan

M. R.

Tan

Turner

C. F.

Miller

W. C.

Erbelding

(2012). Incarceration, high-risk sexual partnerships and sexually transmitted infections in an urban population. Sexually Transmitted Infections, 88, 63-68. doi:10.1136/sextrans-2011-050280

56.

Salisbury

E. J.

Van Voorhis

(2009). Gendered pathways: A quantitative investigation of women probationers’ paths to incarceration. Criminal Justice and Behavior, 36, 541-566. doi:10.1177/0093854809334076

57.

Salisbury

E. J.

Van Voorhis

Spiropoulos

G. V.

(2009). The predictive validity of a gender-responsive needs assessment: An exploratory study. Crime & Delinquency, 55, 550-585. doi:10.1177/0011128707308102

58.

Sarteschi

C. M.

Vaughn

M. G.

(2010). Double jeopardy: A review of women offenders’ mental health and substance abuse characteristics. Victims and Offenders, 5, 161-182. doi:10.1080/15564881003640728

59.

Schram

P. J.

Koons-Witt

B. A.

Williams

F. P.

McShane

M. D.

(2006). Supervision strategies and approaches for female parolees: Examining the link between unmet needs and parolee outcome. Crime & Delinquency, 52, 450-471. doi:10.1177/0011128705281845

60.

Scott

C. K.

Coleman-Cowger

V. H.

Funk

R. R.

(2014). Trauma and PTSD among substance-using women entering Cook County Jail. Women & Criminal Justice, 24, 44-61. doi:10.1080/08974454.2013.842518

61.

Scott

C. K.

Dennis

M. L.

(2012). The first 90 days following release from jail: Findings from the Recovery Management Checkups for Women Offenders (RMCWO) experiment. Drug and Alcohol Dependence, 125, 110-118. doi:10.1016/j.drugalcdep.2012.03.025

62.

Shaffer

D. K.

Hartman

J. L.

Listwan

S. J.

(2009). Drug abusing women in the community: The impact of drug court involvement on recidivism. Journal of Drug Issues, 39, 803-827. doi:10.1177/002204260903900402

63.

Singer

M. I.

Bussey

Song

L. Y.

Lunghofer

(1995). The psychosocial issues of women serving time in jail. Social Work, 40, 103-113. doi:10.1093/sw/40.1.103

64.

Smith

Cullen

F. T.

LaTessa

E. J.

(2009). Can 14,737 women be wrong? A meta-analysis of the LSI-R and recidivism for female offenders. Criminology and Public Policy, 8, 183-208. doi:10.1111/j.1745-9133.2009.00551.x

65.

Spjeldnes

Goodkind

(2009). Gender differences and offender reentry: A review of the literature. Journal of Offender Rehabilitation, 48, 314-335. doi:10.1080/10509670902850812

66.

Staton-Tindall

Garner

B. R.

Morey

J. T.

Leukefeld

Krietemeyer

Saum

C. A.

Oser

C. B.

(2007). Gender differences in treatment engagement among a sample of incarcerated substance abusers. Criminal Justice and Behavior, 34, 1143-1156. doi:10.1177/0093854807304347

67.

Steadman

H. J.

Osher

F. C.

Robbins

P. C.

Case

Samuels

(2009). Prevalence of serious mental illness among jail inmates. Psychiatric Services, 60, 761-765. doi:10.1176/appi.ps.60.6.761

68.

Surratt

H. L.

(2003). Parenting attitudes of drug-involved women inmates. The Prison Journal, 83, 206-220. doi:10.1177/0032885503083002006

69.

Tangney

J. P.

Stuewig

Mashek

Hastings

(2011). Assessing jail inmates’ proneness to shame and guilt: Feeling bad about the behavior or the self? Criminal Justice and Behavior, 38, 710-734. doi:10.1177/0093854811405762

70.

Taxman

F. S.

Rhodes

A. G.

Dumenci

(2011). Construct and predictive validity of criminal thinking scales. Criminal Justice and Behavior, 38, 174-187. doi:10.1177/0093854810389550

71.

Teplin

L. A.

Abram

K. M.

McClelland

G. M.

(1996). Prevalence of psychiatric disorders among incarcerated women: Pretrial jail detainees. Archives of General Psychiatry, 53, 505-512. doi:10.1001/archpsyc.1996.01830060047007

72.

Tripodi

S. J.

Pettus-Davis

(2013). Histories of childhood victimization and subsequent mental health problems, substance use, and sexual victimization for a sample of incarcerated women in the US. International Journal of Law and Psychiatry, 36, 30-40. doi:10.1016/j.ijlp.2012.11.005

73.

van Olphen

Eliason

M. J.

Freudenberg

Barnes

(2009). Nowhere to go: How stigma limits the options of female drug users after release from jail. Substance Abuse Treatment, Prevention, and Policy, 4, 10. doi:10.1186/1747-597X-4-10

74.

Van Voorhis

(2005). Classification of women offenders: Gender-responsive approaches to risk/needs assessment. Community Corrections Report, 12(2), 19-27.

75.

Van Voorhis

(2012). On behalf of women offenders: Women’s place in the science of evidence-based practice. Criminology and Public Policy, 11, 111-145. doi:10.1111/j.1745-9133.2012.00793.x

76.

Van Voorhis

Wright

E. M.

Salisbury

Bauman

(2010). Women’s risk factors and their contributions to existing risk/needs assessment: The current status of a gender-responsive supplement. Criminal Justice and Behavior, 37, 261-288. doi:10.1177/0093854809357442

77.

Walters

G. D.

(2002). The Psychological Inventory of Criminal Thinking Styles (PICTS): A review and meta-analysis. Assessment, 9, 278-291. doi:10.1177/1073191102009003007

78.

Walters

G. D.

(2006). Appraising, researching and conceptualizing criminal thinking: A personal view. Criminal Behaviour and Mental Health, 16, 87-99. doi:10.1002/cbm.50

79.

Walters

G. D.

(2007). Measuring proactive and reactive criminal thinking with the PICTS: Correlations with outcome expectancies and hostile attribution biases. Journal of Interpersonal Violence, 22, 371-385. doi:10.1177/0886260506296988

80.

Walters

G. D.

(2011a). Predicting recidivism with the psychological inventory of criminal thinking styles and level of service inventory-revised: Screening version. Law and Human Behavior, 35, 211-220. doi:10.1007/s10979-010-9231-7

81.

Walters

G. D.

(2011b). Screening for malingering/exaggeration of psychiatric symptomatology in prison inmates using the PICTS Confusion and Infrequency scales. Journal of Forensic Sciences, 56, 444-449. doi:10.1111/j.1556-4029.2010.01651.x

82.

Walters

G. D.

Elliott

W. N.

(1999). Predicting release and disciplinary outcome with the Psychological Inventory of Criminal Thinking Styles: Female data. Legal and Criminological Psychology, 4, 15-21. doi:10.1348/135532599167743

83.

Walters

G. D.

McDonough

J. R.

(1998). The Lifestyle Criminality Screening Form as a predictor of federal parole/probation/supervised release outcome: A 3-year follow-up. Legal and Criminological Psychology, 3, 173-181. doi:10.1111/j.2044-8333.1998.tb00357.x

84.

Wexler

H. K.

Melnick

Cao

(2004). Risk and prison substance abuse treatment outcomes: A replication and challenge. The Prison Journal, 84, 106-120. doi:10.1177/0032885503262458

85.

Wright

E. M.

Van Voorhis

Salisbury

E. J.

Bauman

(2012). Gender-responsive lessons learned and policy implications for women in prison: A review. Criminal Justice and Behavior, 39, 1612-1632. doi:10.1177/0093854812451088

86.

Zlotnick

(1997). Posttraumatic stress disorder (PTSD), PTSD comorbidity, and childhood abuse among incarcerated women. Journal of Nervous and Mental Disease, 185, 761-763.