Does Recent Physical and Sexual Victimization Affect Further Substance Use for Adult Drug-Involved Offenders?

Abstract

This study examined whether physical and sexual victimization experiences were related to further substance use for a sample of drug-involved adult offenders and whether this increase could be attributed to depression experienced after the victimization occurred. A total of 674 men and 284 women from the longitudinal Multisite Adult Drug Court Evaluation (MADCE) were included in analyses. The study included 23 drug court and 6 comparison sites. Study participants completed three interviews: at baseline enrollment and then at 6 and 18 months after baseline. Multilevel path modeling showed that physical and sexual victimization experiences during the year before the baseline interview were associated with further substance use at 18 months and that this relationship was mediated by depression. All relationships held for both men and women, and beyond the contribution of several control variables, including drug court program participation. Public health and criminal justice personnel working with substance-using offenders should screen individuals for victimization-related trauma and, if identified, provide assistance to evaluate and improve such individuals’ mental health and, subsequently, decrease their likelihood of using substances.

Keywords

adult offenders alcohol and drugs mental health and violence

Introduction

For decades, child psychologists, social workers, and researchers studying victimization have noted higher rates of trauma and abusive experiences among offenders than in the general population. In particular, childhood maltreatment and abuse are noted in research on juvenile delinquents (e.g., Baer & Maschi, 2003; Heck & Walsh, 2000; Maschi et al., 2010; Paton, Crouch, & Carnic, 2009), sexual offenders (e.g., Leibowitz, Laser, & Burton, 2011; Maniglio, 2011; McMackin, Leisen, Cusack, LaFratta, & Litwin, 2002), and female offenders (e.g., Stalans, 2009; Strickland, 2008), and also among older adults in correctional institutions (for a review, see Maschi et al., 2011). With regard to recent victimization experiences among incarcerated adult offenders, Wolff and Shi (2009a) noted that 40% of more than 7,500 inmates, including 6,964 men from 12 male prisons and 564 women from one female prison, experienced some form of victimization while in prison in the six months before data collection (35% of male inmates and 24% of female inmates experienced physical victimization—including being threatened or harmed with a knife or shank and being slapped, hit, kicked, or bit—and 10% of male inmates and 25% of female inmates experienced sexual victimization).¹

Past studies have also found that substance use and mental health issues are common in offender populations. For example, James and Glaze (2006) report that 42% of inmates in state prison, 29% of inmates in federal prison, and 49% of jail inmates have both mental health and substance use issues. Findings from the National Survey on Drug Use and Health show that among 5 million adults on probation in 2009, 28% reported illicit drug use that same year (Substance Abuse and Mental Health Services Administration [SAMHSA], 2010). SAMHSA (2004) also reports that the rate of serious mental illness among the probation population (19%) is more than two times that of the general population (9%).

However, literature documenting the effects of physical and sexual victimization on mental health and substance abuse outcomes has primarily focused on adolescents (e.g., Kilpatrick, Saunders, & Smith, 2003) and highly specific, nonoffender populations such as female victims of domestic violence and victims of sexual assault in the community (e.g., Jasinski, Williams, & Brewster, 1997; Kaukinen and DeMaris, 2005; Zweig, Crockett, Sayer, & Vicary, 1999).² Studies have shown that battered women are more likely than nonbattered women to have depression, anxiety, and phobias, and to be substance abusers (Roberts, Williams, Lawrence, & Raphael, 1999; Sackett & Saunders, 1999; Stark & Flitcraft, 1991) and that individuals who experience sexual victimization report more problems with depression, self-esteem, social anxiety, fear, poor relationship quality, and substance abuse, compared with nonvictims (Beckman & Ackerman, 1995; Kaukinen & DeMaris, 2005; Kilpatrick, Best, Saunders, & Veronen, 1988; Siegel, Golding, Stein, Burnam, & Sorenson, 1990; Zweig, 1997; Zweig, Barber, & Eccles, 1997; Zweig et al., 1999).

Existing literature on victims of prison violence also draws links between victimization experiences and emotional reactions/consequences but not links to substance use. More specifically, research on adult prisoners experiencing sexual violence suggests they experience adjustment problems, with many victims of such violence experiencing depression, anxiety, posttraumatic stress disorder (PTSD), rape trauma syndrome, and suicide ideation (Dumond & Dumond, 2002; Human Rights Watch, 2001; Just Detention International, 2009; Struckman-Johnson, Struckman-Johnson, Rucker, Bumby, & Donaldson, 1996). Almost 80% of sexual assault victims identified by Struckman-Johnson and colleagues (1996) reported having a significant emotional reaction to the sexual violence they experienced in prison and half experienced depression. With regard to physical victimizations in prison, Wolff and Shi (2009b) found in their sample of nearly 7,000 male inmates that the most common emotional reaction was anger (for 68% of inmate-on-inmate violence victims and 74% of staff-on-inmate violence victims); other reactions included fear, difficulty sleeping, depression, and flashbacks. Hochstetler, Murphy, and Simons (2004) reported similar findings, though they used a broader definition of victimization and their study included far less men (n = 208 male inmates). They found that that in-prison victimization (as measured by theft, robbery, assault, and threats) had direct effects on PTSD and depressive symptoms. Finally, Johnston Listwan, Colvin, Hanley, and Flannery’s (2010) survey of over 1,600 formerly incarcerated males in halfway houses found that coercion in prison (measured by inmate-on-inmate victimization, witnessed victimization of others, and threatening/coercive prison environment) was related to lower well-being in the community, and specifically, to trauma symptoms (e.g., PTSD, depression, anxiety).

Thus, an empirical gap remains in our understanding of the connection between victimization experiences, mental health issues, and substance abuse among offender populations generally, and specifically, among offenders under community supervision. To our knowledge, only one study to date shows that a history of abuse or trauma is related to both having a substance use disorder and using psychotropic medications for offenders in community corrections (Jackson, Mrug, Cook, Beidleman, & Cropsey, 2011). In the current study, we posit general strain theory as providing a succinct model for explaining the role of victimization in the development and/or persistence of mental health and substance use issues for such offender populations. Thus, we seek to examine how recent victimization experiences affect the lives of offenders on community supervision going forward, despite their status as known offenders.³

General Strain Theory

Borrowing from criminological research, Agnew’s general strain theory (GST) offers a viable framework for understanding the relationship between victimization experiences and subsequent substance use and the mechanisms by which these experiences and behaviors are connected (Agnew, 1992, 2001, 2002, 2009).⁴ GST posits that strain in the form of negative or noxious stimuli generates negative affect within individuals that, in turn, leads people to take corrective action to alleviate such feelings via internalized (substance-using) or externalized (criminal) behavioral reactions.⁵ The negative affect related to substance use may take the form of anxiety, depression, or guilt (Agnew, 2002). This meditational model reflects that the consequences of violence often can be indirect and can involve multiple variables that must be accounted for when attempting to understand the effects of such trauma (Carlson, 2005). GST further argues that some types of strain—victimization experiences, in particular—may be especially likely to lead to negative outcomes because they are seen as unjust, have a large emotional impact, and involve little personal control. Agnew (2002) found that physical victimization led to delinquency in boys and called for future research to focus on different types of victimization experiences for both boys and girls and on additional types of negative outcomes, such as substance use.

From this, two known studies to date have used a GST approach to examine the effects of different types of victimization on substance use. The first study used data from the National Youth Survey and found support for GST by documenting that childhood physical abuse led to adult substance use, and this relationship was mediated by depression (Lo & Cheng, 2007). The second study tested GST on a sample of former prisoners who were classified as serious and violent offenders to examine the effect of in-prison victimization experiences on substance use once individuals were released to the community (Zweig, Yahner, Visher, & Lattimore, manuscript submitted for publication). Results showed that in-prison victimization led to greater likelihood of substance use behaviors and that this relationship was partially mediated by feelings of depression.

These two previous studies show that GST succinctly links victimization to emotional reactions and then to behavioral outcomes for both an adolescent population and a population of former prisoners for which the noxious strain was in-prison victimization experiences. The current study attempted to understand how being victimized in the community related to further substance use for a sample of adult drug-involved offenders—nearly all of whom were under some form of community supervision—by testing the applicability of GST to these issues. More specifically, we aimed to answer the following research questions:

Research Question 1: Does adult victimization increase the likelihood of using substances, even after controlling for other known predictors of substance use, and, if so, is this path mediated by depression, as predicted by GST?

Research Question 2: Do the path models defining the relationship between victimization and substance use differ between male and female adult offenders?

We hypothesized that we would find a main effect whereby victimization experiences lead to a greater likelihood of using substances (after controlling for relevant background variables related to substance use and for participation in a drug court program) and that this relationship would be mediated by depression. Our analysis of gender differences in these relationships was exploratory.

Method

We used data from the Multisite Adult Drug Court Evaluation (MADCE), funded by the National Institute of Justice—a longitudinal, quasi- experimental study of 23 drug courts and 6 comparison jurisdictions, designed to compare drug court participants to other offenders with similar drug use, criminal history, and psychosocial profiles in jurisdictions that did not offer drug courts (Rossman, Roman, Zweig, Rempel, & Lindquist, 2011). The 23 drug courts were chosen based on variation in implementation of specific drug court policies and were located in seven geographic clusters; 6 non-drug court jurisdictions in similar locations were also selected as comparison. For drug court sites, participants were recruited and enrolled into the study at the point they were enrolled into the drug court programs. The comparison sites included several alternative modes for handling drug-involved offenders, representing the diverse set of activities used in jurisdictions that do not currently implement drug courts. Notably, some comparison sites mandated offenders to community-based treatment but without other components of the drug court model; other comparison sites involved standard probation only. The comparison sites included several Treatment Alternatives for Safe Communities (TASC) programs, a Breaking the Cycle program (BTC—which involved both treatment and judicial status hearings, and standard court-referred, probation-monitored treatment. Participants were recruited and enrolled into the study at the point they started the TASC or BTC programs. Those on standard probation were selected from a list of probationers based on having similar charges to those charges allowed in the drug court programs and a referral to substance abuse treatment.

Data Collection

The MADCE included in-person interviews with offenders across the 29 sites conducted at three different intervals: (a) baseline when participants enrolled in drug courts or in alternative conditions in the comparison sites, (b) 6 months after the baseline interview, and (c) 18 months after baseline. Baseline enrollment took place during a 16-month period from March 2005 through June 2006. Interviews were conducted using a computer-assisted personal interviewing system (CAPI). A team of field interviewers used by RTI International were extensively trained in human subjects’ regulations and administering informed consent processes, in gaining cooperation, in locating hard-to-find study participants, in implementing the CAPI system, in properly and consistently administering the interview, in interviewing in correctional settings (if necessary), and in identifying and responding to distressed respondents. The field interviewers across sites were supervised by two field supervisors, who were responsible for monitoring performance and efficiency, in addition to assisting with facility access (when respondents were incarcerated) and locating sample members. The interviews lasted between 1.5 and 2 hours and covered a variety of topics including background characteristics (e.g., demographics, drug use and treatment history, criminal history, mental health, close ties to drug users and those involved in the criminal justice system, and current victimization experiences) and outcomes (criminal behavior, drug use, and other measures of personal functioning).

Sample

Seventy-two percentage of eligible study participants were enrolled at baseline, for a total initial sample of 1,781 offenders. Subsequently, 86% of those reached at baseline also completed 6-month interviews, and 83% completed 18-month interviews. For the current study, only a subsample (n = 958) is included in the analysis given that only a portion of the sample was asked questions about victimization during the baseline interview as these items were added to the survey midway through baseline enrollment.⁶ Furthermore, we restrict analyses to those who completed the baseline, 6-, and 18-month interviews. A total of 76% of those interviewed at baseline completed both the 6- and 18-month interviews, with identical attrition rates between the drug and comparison court samples (each was 76%). All analyses were corrected for attrition as described below.

The final sample was mostly male (72%), with an average age of 34 years. Fifty-four percentage were White and one-third (34%) were Black/African American. Sixty percentage had their high-school diploma or GED equivalent, and the average annual income reported was just more than US$11,000.

Measures

We examined the outcome measure of average days of drug use per month, as reported for the 6 to 18 months after the baseline interview (covering a 12-month period). Eight drugs were included: alcohol, marijuana, cocaine, heroin, hallucinogens, amphetamines, prescription drugs, and methadone. Possible responses ranged from 0 to 30 days per month.

The hypothesized psychological mediator of the victimization—substance use relationship, a depression scale, was measured 6 months after the baseline interview and indicated the frequency of depressive symptoms reported by respondents (alpha = .81; Andresen, Malmgren, Carter, & Patrick, 1994.). The scale asked respondents how often they experienced any of ten possible feelings and behaviors in the prior seven days, including the following: being bothered by things that do not usually bother you, having trouble keeping your mind on what you were doing, feeling depressed, feeling like everything you did was an effort, feeling hopeful about the future (reverse coded), feeling fearful, having restless sleep, being happy (reverse-coded), feeling lonely, and being unable to get going. Response choices for each item included 0 = never, 1 = rarely, 2 = sometimes, 3 = often, and 4 = always; these were summed across the 10 items to compute the scale score.

The two main independent variables of interest, physical and sexual victimization, covered abuse during the 12 months prior to the baseline interview and were coded as “1” if respondents reported experiencing any abuse during that time and “0” if not. Physical victimization covered being pushed, slapped, grabbed, having one’s arm twisted or hair pulled, being restrained or shoved, having something thrown at one that could hurt, being punched or hit with something that could hurt, being kicked, being slammed against a hard surface, beaten up, choked, strangled, burned or scalded on purpose, and having a knife or gun used on one. Sexual victimization included being physically forced—by hitting, holding down, or using a weapon—to have oral sex, anal sex, or vaginal sex; having someone verbally insist on sex (oral, anal, or vaginal) when one did not want to; and insisting on sex without a condom.

We also included a number of independent variables as controls, all of which were measured at baseline, with the exception of days not on the street (described below). Importantly, we controlled for respondents’ participation in drug court (versus comparison court), which was a dichotomous variable indicating “1” if respondents had participated in a drug court and “0” if they had participated in a comparison court. Other control variables were measures found in prior research to be related to drug use, such as age, age at first drug use, race/ethnicity, gender, education, income, marital status, presence of minor children, family drug abuse, primary drug of choice, and prior arrests. We also included a dichotomized scale measuring presence of features of antisocial personality disorder;⁷ a dichotomized measure of features of depression at baseline; a frequent drug user measure that indicated respondents who reported using alcohol or drugs an average of 20 to 30 days per month for the 6 months prior to the baseline interview; a measure of prior days of drug treatment, which was the average number of days per month of clinical substance abuse treatment (excluding self-help groups) for any episodes immediately prior to the baseline interview; and finally, a measure of days not on street, which was the average days per month during the follow-up period in which respondents were either incarcerated, hospitalized, or in residential treatment settings—times during which respondents were at minimal risk of drug use.

Descriptive Statistics

Descriptive statistics are presented in Table 1, with means and standard deviations reported separately for respondents who were physically victimized (31% of sample) or sexually victimized (9% of sample), compared with those who were not. Notably, missing data were sufficiently limited that we could assume it was missing completely at random (Allison, 2001). Thus, respondents with missing data on any covariates included in the estimated models were excluded, with model Ns ranging from 923 to 924 (96% of the sample).

Table 1.

Descriptive Statistics for Variables, by Victimization Status at Baseline

				Physically Victimized?
				Yes (n = 301)	No (n = 657)	Yes (n = 89)	No (n = 869)
	Missing	Minimum	Maximum	M (SD)	M (SD)	M (SD)	M (SD)
Dependent variable
Days of drug use per month	0%	0	30	4.06 (6.60)****	2.63 (5.40)	4.05 (7.18)*	2.96 (5.66)
Psychological mediator
Depression scale	<1%	0	30	8.77 (6.28)****	6.34 (5.84)	8.87 (6.89)***	6.89 (5.96)
Control variables
Age	<1%	18	66	31.34 (10.16)****	35.16 (10.59)	31.76 (9.68)**	34.23 (10.68)
Age at first drug use	<1%	1	38	12.98 (3.90)***	13.84 (3.68)	12.71 (3.96)**	13.67 (3.74)
Antisocial personality disorder	0%	0	1	0.54 (0.50)****	0.39 (0.49)	0.59 (0.49)***	0.42 (0.49)
Black/African American	0%	0	1	0.24 (0.43)****	0.38 (0.49)	0.42 (0.50)	0.33 (0.47)
Days not on street per month	0%	0	30	7.20 (9.80)****	5.08 (8.44)	8.03 (10.89)***	5.49 (8.67)
Depressed at baseline	0%	0	1	0.45 (0.50)***	0.35 (0.48)	0.53 (0.50)***	0.36 (0.48)
Drug court (versus comparison)	0%	0	1	0.60 (0.49)	0.63 (0.48)	0.63 (0.49)	0.62 (0.49)
Family drug abuse	0%	0	1	0.88 (0.32)**	0.82 (0.39)	0.88 (0.32)	0.83 (0.37)
Frequent drug user	0%	0	1	0.37 (0.48)*	0.32 (0.47	0.35 (.48)	0.33 (0.47)
High school diploma or GED	0%	0	1	0.58 (0.49)	0.61 (0.49)	0.61 (0.49)	0.60 (0.49)
Income	0%	US$0	US$180K	US$9,885 (US$17,051)	US$11,620 (US$18,261)	US$11,485 (US$24,671)	US$11,054 (US$17,083)
Male gender	0%	0	1	0.70 (0.46)	0.73 (0.44)	0.66 (0.49)	0.73 (0.45)
Married or in relationship	0%	0	1	0.52 (0.50)	0.53 (0.50)	0.50 (0.50)	0.53 (0.50)
Minor children (<18 years)	0%	0	1	0.46 (0.50)	0.50 (0.50)	0.49 (0.50)	0.49 (0.50)
Prior arrests	3%	0	25	7.70 (7.56)	8.07 (8.23)	9.22 (8.46)	7.83 (7.98)
Prior days of drug treatment	0%	0	31	2.15 (5.10)	2.12 (5.87)	2.02 (5.24)*	2.14 (5.69)
Primary hard^a drug of choice	0%	0	1	0.62 (0.49)	0.61 (0.49)	0.69 (0.46)	0.61 (0.49)
White	0%	0	1	0.64 (0.48)****	0.50 (0.50)	0.53 (0.50)	0.55 (0.50)

Notes: All data come from MADCE interviews; N = 958 respondents from 28 courts (301 respondents were physically victimized; 89 respondents were sexually victimized). Dependent variables covered the yearlong period from 6 to 18 months after baseline, psychological mediator was measured 6 months after baseline, and victimization and control variables were measured at baseline. Results were weighted to control for drug court/comparison comparability at baseline and attrition bias (Rempel & Farole, 2011).

Measured at baseline, this includes cocaine, heroin, amphetamines, and other drugs. Comparison group is alcohol and marijuana.

p < .10. **p < .05. ***p < .01. ****p < .001.

Data Analysis

Our analytic approach consisted of multilevel structural equation modeling (MSEM) in the Mplus 6.0 statistical software program (Muthén & Muthén, 1998-2010). MSEM in Mplus allowed us to estimate the direct and indirect effects of victimization on the dependent variable, while accounting for the effects of the psychological mediator, depression, and control variables, as well as the hierarchical clustering of respondents into 28 sites (the final sample came from 22 of the 23 drug courts and 6 comparison sites). MSEM also enabled us to control for the between-courts’ effect of drug court participation (versus comparison court) so as to isolate the individual-level effects of interest, to test for possible moderation of individual-level effects by gender, and to account for the over-dispersed distribution of the dependent variable by modeling its prediction using negative binomial regression.

All models were run using maximum likelihood estimation with a sandwich estimator to compute robust standard errors, which accounted for the nonnormality of the dependent variable and the nonindependence of observations due to clustered sampling (Muthén & Muthén, 1998-2010). Measures of model fit for negative binomial path models are currently unavailable in Mplus; however, observed fit indices for the same models run using linear regression were excellent (e.g., chi-square values were not statistically significant at a p < .05; root mean square errors of approximation were less than 0.03; Bollen, 1989). In addition, inverse probability weights (IPW) were used in all analyses to adjust for comparability between drug court and comparison court respondents at baseline, as well as the relatively small amount of attrition following the baseline interview (Rempel & Farole, 2011).

To address the first research question, we estimated four MSEMs (path models) in Mplus. Model A was a negative binomial regression predicting days of drug use per month at the 18-month interview, using physical victimization in the year prior to baseline and all of the specified control variables as predictors. Model B was a repeat of this negative binomial regression, substituting sexual victimization for physical victimization as a predictor of drug use. Model C was an expansion of Model A to include the hypothesized psychological mediator—depression as of the six-month interview. This model allowed physical victimization (and depression at baseline) to predict depression at the 6-month interview using linear regression, and 6-month depression to predict 18-month drug use using negative binomial regression; thus, this model assessed the indirect effect of physical victimization on drug use. It also allowed physical victimization to continue to have a direct effect on drug use. Model D was exactly the same hypothesized mediation model, substituting sexual victimization for physical victimization as a predictor of depression and drug use.

To test the second research question, moderation by gender in the effects of victimization on depression or the dependent variable (as specified in Models A through D), we conducted chi-square difference testing of nested MSEMs (Muthén & Muthén, 2005). The nested models that we compared consisted of (a) a restricted model that held all effects of predictors on the dependent variable to be equivalent across men and women and (b) an unrestricted model that permitted the effects of victimization on depression and the dependent variable to vary by gender. By assessing the statistical significance of the chi-square difference in corrected log likelihoods obtained from both models, we determined whether allowing differentiation by gender significantly improved the fit of the model to the data—in other words, whether the hypothesized models’ effects were moderated by gender (Satorra & Bentler, 1999).

Results

We found that both physical and sexual victimization experiences in the year before the baseline interview positively increased drug use 18 months later, even after controlling for a number of other factors related to drug use, as well as participation in a drug court program. We also found support for general strain theory, in that the strain of victimization was indirectly—through higher levels of depression 6 months later—associated with increased drug use as of the 18-month interview. With regard to physical victimization, we found evidence of full mediation by depression, and with regard to sexual victimization, we found partial mediation.

In Table 2, we present the unstandardized coefficients (betas) and standard errors (SEs) associated with the victimization and control variables included as predictors of days of drug use per month. Both physical victimization (Model A) and sexual victimization (Model B) had statistically significantly associations with later drug use at the p < .05 and p < .01 levels, respectively. These relationships held true despite controlling for respondents’ participation in a drug court or comparison site court, age, age at first drug use, antisocial personality disorder, race/ethnicity, days not on the street during the follow-up period, depression at baseline, family drug abuse, status as a frequent drug user, education, income, gender, marital status, children, prior arrests, prior drug treatment, and primary drug of choice. Several of these control variables also showed statistically significant effects on drug use (see Table 2).

Table 2.

Does Victimization Increase Drug Use Among Adult Drug-Involved Offenders, After Controlling for Other Variables?

	Model A: Days of Drug Use per Month		Model B: Days of Drug Use per Month
	b	SE	b	SE
Victimization
Physical victimization	0.311**	0.154
Sexual victimization			0.542***	0.173
Control variables
Age	−0.012	0.011	−0.013	0.010
Age at first drug use	−0.020	0.015	−0.019	0.016
Antisocial personality disorder	0.212	0.135	0.242*	0.124
Black/African American	−0.012	0.169	−0.089	0.150
Days not on street per month	0.002	0.014	0.002	0.014
Depressed at baseline	0.034	0.098	−0.010	0.104
Drug court (versus comparison)^a	−0.986****	0.301	−0.997****	0.302
Family drug abuse	0.078	0.230	0.042	0.236
Frequent drug user	0.517****	0.114	0.537****	0.113
High school diploma or GED	−0.127	0.176	−0.168	0.177
Income (natural log)^b	0.012	0.033	0.015	0.032
Male gender	0.201	0.181	0.190	0.153
Married or in relationship	0.043	0.114	0.042	0.114
Minor children	−0.207	0.148	−0.198	0.154
Prior arrests	0.004	0.011	0.005	0.011
Prior days of drug treatment	−0.005	0.014	−0.003	0.014
Primary hard drug of choice	−0.446****	0.103	−0.428****	0.101
White	−0.272**	0.130	−0.285**	0.126
Between courts intercept	1.851****	0.504	1.946****	0.517
Within courts dispersion	2.765****	0.335	2.748****	0.328

Notes: All data come from MADCE interviews; N = 924 respondents from 28 courts (96% of sample). Days of drug use per month covered the yearlong period from six to 18 months after baseline, while victimization and control variables were measured at baseline. Models were estimated in Mplus 6.0 using multilevel negative binomial regression; beta coefficients are unstandardized and standard errors (SE) robust (Muthén & Muthén, 1998-2010). Results were weighted to control for drug court/comparison comparability at baseline and attrition bias (Rempel & Farole, 2011).

This is a between courts independent variable; all other independent variables were measured within courts.

For analysis purposes, this variable was transformed into its natural log to minimize skewness.

p < .10. **p < .05. ***p < .01. ****p < .001.

In Table 3, we present the unstandardized coefficients and standard errors associated with predictors in the models testing mediation of victimization’s effect on drug use via depression. Model C focuses on the effect of physical victimization on drug use and shows evidence that this relationship is fully mediated by depression at the 6-month interview.⁸ Respondents who experienced physical victimization in the year prior to the baseline interview showed significantly higher levels of depression 6 months later (beta = 2.151, p < .001), and that depression was associated with significantly more days of drug use per month as of the 18-month interview (beta = 0.022, p < .05). Furthermore, once accounting for this indirect relationship of victimization on drug use via depression, physical victimization no longer had a significant direct effect on drug use 18 months later (beta = 0.239, p not significant at .10).

Table 3.

Does Depression Mediate the Relationship Between Victimization and Drug Use Among Adult Drug-Involved Offenders?

	Model C				Model D
	Depression Scale		Days of Drug Use per Month		Depression Scale		Days of Drug Use per Month
	b	SE	b	SE	b	SE	b	SE
Psychological mediator
Depression scale			0.022**	0.011			0.023**	0.011
Victimization
Physical victimization	2.151****	0.359	0.239	0.151
Sexual victimization					1.321*	0.683	0.498***	0.178
Control variables
Age			−0.013	0.011			−0.013	0.010
Age at first drug use			−0.020	0.015			−0.019	0.015
Antisocial personality disorder			0.202	0.136			0.220*	0.130
Black/African American			−0.045	0.156			−0.112	0.141
Days not on street per month			−0.002	0.014			−0.002	0.014
Depressed at baseline	4.113****	0.467			4.210****	0.450
Drug court (versus comparison)^a			−0.948****	0.296			−0.920****	0.284
Family drug abuse			0.088	0.241			0.059	0.247
Frequent drug user			0.542****	0.120			0.557****	0.123
High school diploma or GED			−0.128	0.172			−0.159	0.171
Income (natural log)^b			0.011	0.033			0.014	0.033
Male gender			0.292	0.182			0.300*	0.165
Married or in relationship			0.049	0.107			0.049	0.107
Minor children			−0.226	0.143			−0.224	0.148
Prior arrests			0.002	0.010			0.003	0.010
Prior days of drug treatment			−0.007	0.013			−0.006	0.013
Primary hard drug of choice			−0.436****	0.109			−0.420****	0.106
White			−0.278**	0.135			−0.288**	0.132
Within courts intercept	4.792****	0.346			5.287****	0.352
Between courts intercept			1.675****	0.486			1.670****	0.485
Within courts dispersion			2.739****	0.337			2.716****	0.328

Notes: All data come from MADCE interviews; N = 923 respondents from 28 courts (96% of sample). Days of drug use per month covered the yearlong period from 6 to 18 months after baseline, the depression scale was measured six months after baseline, and victimization and control variables were measured at baseline. Models were estimated in Mplus 6.0 using multilevel structural equation modeling, with linear regression for the depression scale outcome and negative binomial regression for the drug use outcome; beta coefficients are unstandardized and standard errors (SE) robust (Muthén & Muthén, 1998-2010). Results were weighted to control for drug court/comparison comparability at baseline and attrition bias (Rempel & Farole, 2011).

This is a between courts independent variable; all other independent variables were measured within courts.

For analysis purposes, this variable was transformed into its natural log to minimize skewness.

p < .10.**p < .05.***p < .01.****p < .001.

Model D focuses on the effect of sexual victimization and shows evidence of partial mediation in its effect on drug use via depression at 6 months. Sexual victimization experiences in the year prior to baseline were associated with higher levels of depression as of the 6-month interview (beta = 1.321, p < .10 approaching significance), whereas depression subsequently was associated with drug use in the year prior to the 18-month interview (beta = 0.023, p < .05). Unlike physical victimization experiences, however, accounting for depression did not fully explain the relationship between sexual victimization and subsequent drug use. Rather, there remained a direct, significant effect of sexual victimization on drug use (beta = 0.498, p < .01) beyond its mediated indirect effect.

Importantly, all of the relationships in Models C and D held true despite inclusion of the previously described control variables; thus—regardless of respondents’ participation in a drug court or comparison site court, age, race/ethnicity, drug use and drug treatment at baseline, criminal history, marital status, education, etcetera—physical and sexual victimization experiences were associated with higher levels of subsequent depression, which in turn was associated with more days of subsequent drug use. Furthermore, with regard to predicting depression as of the 6-month interview, both physical and sexual victimization had effects on depression despite controlling for the statistically significant association of depression at baseline.

To answer our second research question, we reanalyzed the effects of interest from Models A through D—namely, those involving physical or sexual victimization, depression, drug use, and crime—while retaining all the previously described controls, to examine the possibility that these effects varied between male and female offenders. For each model, we created a restricted version that permitted no variation by gender, and an unrestricted version that allowed pathways from victimization to the mediator and dependent variable to differ between men and women. As shown in Table 4, we then compared the log likelihoods resulting from the restricted and unrestricted models to determine if their difference, distributed as a chi-square, varied significantly (Muthén & Muthén, 2005; Satorra & Bentler, 1999). As can be seen in the last column of Table 4, none of the differences between the restricted and unrestricted versions of Models A through D achieved statistical significance; rather, all chi-square values had probabilities greater than .10. From this, we conclude that the effects of physical and sexual adult victimization experiences on depression and drug use, as previously presented, hold equally true for male and female drug-involved offenders.

Table 4.

Does the Effect of Victimization on Outcomes Vary by Gender?

	Restricted Model		Unrestricted Model		Difference
	LL₀(p₀)	c₀	LL₁(p₁)	c₁	$χ_{d}^{2}$ (df_d)	c_d	p < .10
Model A	−2,328.89(23)	1.15	−2,328.71(24)	1.13	0.47(1)	0.79	No
Model B	−2,328.10(23)	1.09	−2,327.95(24)	1.10	0.23(1)	1.28	No
Model C	−5,216.53(28)	1.29	−5,215.12(31)	1.22	4.91(3)	0.58	No
Model D	−5,227.04(28)	1.27	−5,225.08(31)	1.22	4.94(3)	0.79	No

Notes: All data come from MADCE interviews; Ns ranged from 923 to 924 respondents from 28 courts. Models were estimated in Mplus 6.0 using multilevel negative binomial regression for the drug use outcome (Models A through F) and linear regression for the psychological mediator, depression (Models C and D; Muthén & Muthén, 1998-2010). Results were weighted to control for drug court/comparison comparability at baseline and attrition bias (Rempel & Farole, 2011). Tests of variation by gender were conducted via chi-square difference testing of nested models using the log likelihoods (exact methods are described in Muthén & Muthén, 2005; Satorra & Bentler, 1999). Restricted models held the direct and indirect (through depression) effects of victimization equal across men and women, whereas the unrestricted models allowed the effects to vary by gender. The difference indicates whether the unrestricted model offered a significant improvement in model fit, thereby indicating moderation of effects.

Discussion

This study demonstrated support for Agnew’s (2001, 2002, 2009) general strain theory when examining recent physical and sexual victimization experiences in an offender population under community supervision and how such experiences relate to depression and substance use. More specifically, after controlling for several factors related to substance use for drug-involved offenders, we found that experiencing physical violence in the prior year led to greater likelihood of further substance use and that this relationship was fully mediated by feelings of depression. In addition, after controlling for several factors related to substance use for drug-involved offenders, we found that experiencing sexual violence in the prior year also led to greater likelihood of further substance use and that this relationship was partially mediated by feelings of depression.

These findings support and extend similar findings based on nonadult populations (Agnew, 2002; Lo & Cheng, 2007) and on more severe criminal justice populations (Zweig et al., manuscript submitted for publication)—that is, former prisoners classified as severe and violent offenders who experienced in-prison victimization—by showing that (a) GST explains the relationship between victimization experiences and negative outcomes for adult drug-involved offenders and that (b) depression mediates the relationship between victimization and substance use outcomes in offender populations. It also demonstrates that GST holds for both adult men and women.

Two limitations to the current study are worth noting. First, like all statistical representations of causal processes, there are limitations of these models in applied research. The models we hypothesized were carefully guided by theory (GST) and practical considerations (modeling the longitudinal data such that the present cannot influence the past). However, there are other ways of modeling these data and thus, always the possibility that other models exist which may provide a comparable or even more accurate representation of the relationships among the variables we analyzed. It is also certain that we did not include all of the possible influences on drug use, as we were limited by those available to us in the data we selected. Second, offenders who reported victimization experiences were not asked who perpetrated the violence against them, so we could not examine if GST models hold for victims whose experiences were perpetrated by people with whom they know well or intimately versus those they do not know well (e.g., romantic partners, close friends, other family members versus acquaintances, strangers, etc). It is not clear whether or not processes that lead to substance use outcomes vary for offenders who are victims of violence from intimates versus victims of violence from others. Relatedly, it also may be important to further explore how different types of victimization based on frequency and severity might affect the GST model now that this study shows that the presence or absence of victimization supports this theory. Future research should examine GST in light of issues of perpetrator–victim relationships, as well as variations in types of victimization.

Despite these limitations, these findings contribute to the literature on how victimization relates to substance use by examining this relationship using longitudinal data and for a population of adult offenders. The findings also have implications for public health and criminal justice personnel who work with substance-using offenders. That is, beyond the challenges already associated with offender status, victimization experiences can detrimentally impact individuals’ mental health and general well-being, further exacerbating challenges such individuals’ face in terms of being a successful citizen.

Thus, those working in programs treating or supervising offenders should screen individuals for victimization-related trauma, and specifically depressive outcomes, and provide assistance to both male and female victims to address such issues. Researchers have noted that, for substance abuse treatment populations, unaddressed issues related to interpersonal violence and victimization may hamper treatment progress and be misinterpreted by providers as a lack of motivation to change (Quimette, Kimerling, Shaw, & Moos, 2000). Alternatively, if providers are sensitive to the existence of victimization trauma and its consequences for individuals’ mental health, even among adult offender populations, they could provide specific relevant counseling and treatment to diminish the likelihood of future substance use. Toward this end, criminal justice agencies—such as probation and parole—should develop the internal capacity to provide such treatment, themselves, or partner with community-based agencies that can do so instead. Addressing the trauma associated with victimization is important to both victims who initially report a crime and to offenders who also have been victimized (Herman, 2010); by doing so, both providers and clients can proceed with the hope of preventing future negative outcomes, including substance use.

Footnotes

Acknowledgements

The authors would like to thank Dr. Nancy La Vigne (Director, Justice Policy Center, Urban Institute) for the support to conduct analyses for this article.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: Data collection was supported by Grant Number 2003-DC-BX-1001 awarded by the National Institute of Justice, Office of Justice Programs, U.S. Department of Justice. Points of view in this document are those of the authors and do not necessarily represent the official position or policies of the U.S. Department of Justice, Urban Institute, or its trustees.

Notes

Bios

Janine M. Zweig, PhD, is a senior research associate in the Justice Policy Center at the Urban Institute and has 16 years of research experience. Her work addresses issues relating to violent victimization, substance use, and adolescent and young adult development. She has examined youth’s use of technology related to teen dating violence victimization and perpetration; the risk factors for and consequences of sexual victimization during adolescence and young adulthood; and the effectiveness of victim service, criminal justice, and social service responses to sexual assault and domestic violence, including processes by which victims receive free sexual assault forensic exams. She has also completed a project focused on examining corrections’ response to sexual violence in state prisons. She received her PhD in human development and family studies from Pennsylvania State University.

Jennifer Yahner, MA, is a Research Associate in the Urban Institute’s Justice Policy Center. She is a former national consortium on violence research fellow who has been conducting criminal justice research for more than 12 years, with experience in studies of prisoner reentry, violence, substance use, and program evaluations. She is skilled in sophisticated statistical modeling, large-scale data management, and a variety of quantitative methodologies (generalized linear modeling, structural equation modeling, multi-level analysis, social network analysis, inverse probability weighting, cost analysis) using Mplus, SAS, SPSS, LISREL, and Ucinet. Her responsibilities on projects such as the Returning Home studies, Judicial Oversight Demonstration Transitional Jobs Research Demonstration, and Hard-to-Employ evaluations have led to publications in peer-reviewed journals such as Criminology and Public Policy, Crime and Delinquency, and Justice Quarterly. She recently completed multilevel analyses of data collected from 29 drug courts in 8 states for the Multisite Evaluation Of Adult Drug Courts.

Shelli B. Rossman, MA, is a senior fellow in the Urban Institute’s Justice Policy Center and has more than 30 years of research and management experience on projects for federal/national, state, and local governments, as well as private-sector clients in the areas of 1) criminal justice, including reentry, problem-solving courts, community-based and correctional supervision, case management and comprehensive service delivery for offender populations, delinquency prevention and intervention, and victimization; 2) public health, focused on substance abuse, HIV/AIDS, and reproductive health; and 3) community safety. In both national and international settings, her projects have addressed improving the standards and monitoring of service delivery for at- and high-risk populations, as well as cultural competency and gender equity issues. She recently completed the largest problem-solving court research study ever conducted: NIJ’s Multisite Evaluation of Adult Drug Courts (MADCE)—a study of 23 courts and 6 comparison jurisdictions in 8 states, during which researchers completed nearly 5,000 in-person surveys, conducted multiple site visits documenting core activities, and performed more than 1,000 research-conducted drug screens.

References

Agnew

(1992). Foundation for a general strain theory of crime and delinquency. Criminology, 30(1), 47-87.

Agnew

(2001). Building on the foundation of general strain theory: Specifying the types of strain most likely to lead to crime and delinquency. Journal of Research in Crime and Delinquency, 38, 319-361.

Agnew

(2002). Experienced, vicarious, and anticipated strain: An exploratory study on physical victimization and delinquency. Justice Quarterly, 19, 603-632.

Agnew

(2009). General strain theory. In Krohn

M. D.

Lizotte

A. J.

Hall

G. P.

(Eds.), Handbook on crime and deviance (pp. 168-186). New York, NY: Springer.

Allison

(2001). Missing data. Thousand Oaks, CA: SAGE.

Andresen

E. M.

Malmgren

J. A.

Carter

W. B.

Patrick

D. L.

(1994). Screening for depression in well older adults: Evaluation of a short form of the CES-D (Center for Epidemiologic Studies–Depression Scale). American Journal of Preventive Medicine, 10(2), 77-84.

Baer

Maschi

(2003). Random acts of delinquency: Trauma and self-destructiveness in juvenile offenders. Child and Adolescent Social Work Journal, 20(2):85-98.

Baron

R. M.

Kenny

D. A.

(1986). The moderator-mediator variable distinction in social psychological research: Conceptual, strategic, and statistical considerations. Journal of Personality and Social Psychology, 51, 1173-1182.

Beckman

L. J.

Ackerman

K. T.

(1995). Women, alcohol, and sexuality. In Galanter

(Ed.), Recent developments in alcoholism: Women and alcoholism (Vol. 12, pp. 267-285). New York, NY: Plenum.

10.

Bollen

K. A.

(1989). Structural equations with latent variables. New York, NY: Wiley.

11.

Carlson

B. E.

(2005). The most important things learned about violence and trauma in the past 20 years. Journal of Interpersonal Violence, 20(1), 119-126.

12.

Dumond

R. W.

Dumond

D. A.

(2002). The treatment of sexual assault victims. In Hensley

(Ed.), Prison sex: Practice and policy (Ch. 5, pp. 67-88). Boulder, CO: Lynne Rienner.

13.

Heck

Walsh

(2000). The effects of maltreatment and family structure on minor and serious delinquency. International Journal of Offender Therapy and Comparative Criminology, 44, 178-193.

14.

Herman

(2010). Parallel justice for victims of crime. Washington, DC: The National Center for Victims of Crime.

15.

Hochstetler

Murphy

D. S.

Simons

R. L.

(2004). Damaged goods: Exploring predictors of distress in prison inmates. Crime & Delinquency, 50, 436-457.

16.

Human Rights Watch. (2001). No escape: Male rape in U.S. prisons. New York, NY: Author.

17.

Jackson

D. O

Mrug

Cook

Beidleman

Cropsey

K. L.

(2011). Factors predicting substance dependence and psychotropic medication use among offenders in community corrections. Addictive Behaviors, 36, 755-758.

18.

James

D. J.

Glaze

L. E.

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

19.

Jasinski

J. L.

Williams

L. M.

Brewster

A. L.

(1997). Partner violence: A 20 year review and synthesis executive summary. In United States Air Force Domestic Violence Literature Review, Synthesis and Implications for Practice, an edited book produced through the United States Department of Agriculture, Cooperative Research Education and Extension Service cooperative agreement No. 95-EXCA-3-0414, the University of Missouri and the United States Air Force.

20.

Johnston Listwan

Colvin

Hanley

Flannery

(2010). Victimization, social support, and psychological well-being: A study of recently released prisoners. Criminal Justice and Behavior, 37, 1140-1159.

21.

Just Detention International. (2009). The basics about sexual abuse in U.S. detention. Retrieved from http://www.justdetention.org/en/factsheets/TheBasics.pdf

22.

Kaukinen

DeMaris

(2005). Age at first sexual assault and current substance use and depression. Journal of Interpersonal Violence, 20, 1244-1270.

23.

Kilpatrick

D. G.

Best

C. L.

Saunders

B. E.

Veronen

L. J.

(1988). Rape in marriage and in dating relationships: How bad is it for mental health? In Prentky

R. A.

Quinsey

V. L.

(Eds.), Human sexual aggression: Current perspectives (pp. 335-344). New York: The New York Academy of Sciences.

24.

Kilpatrick

D. G.

Saunders

B. E.

Smith

D. W.

(2003). Youth victimization: Prevalence and implications. Washington, DC: U.S. Department of Justice, Office of Justice Programs, National Institute of Justice.

25.

Leibowitz

G. S.

Laser

J. A.

Burton

D. L.

(2011). Exploring the relationships between dissociation, victimization, and juvenile sexual offending. Journal of Trauma & Dissociation, 12(1), s38-s52.

26.

C. C.

Cheng

T. C.

(2007). The impact of child maltreatment on young adults’ substance use. American Journal of Drug and Alcohol Use, 33, 139-146.

27.

Maniglio

(2011). The role of childhood trauma, psychological problems, and coping in the development of deviant sexual fantasies in sexual offenders. Clinical Psychology Review, 31, 748-756.

28.

Maschi

Dennis

K. S.

Gibson

MacMillan

Sternberg

Hom

(2011). Trauma and stress among older adults in the criminal justice system: A review of the literature with implications for social work. Journal of Gerontological Social Work, 54, 390-424.

29.

Maschi

MacMillan

Morgen

Gibson

Stimmel

(2010). Trauma, world assumptions, and coping resources among youthful offenders: Social work, mental health, and criminal justice implications. Child & Adolescent Social Work Journal, 27, 377-393.

30.

McMackin

R. A.

Leisen

M. B.

Cusack

J. F.

LaFratta

Litwin

(2002). The relationship of trauma exposure to sex offending behavior among male juvenile offenders. Journal of Child Sexual Abuse, 11(2), 25-40.

31.

Muthén

L. K.

Muthén

B. O.

(1998-2010). Mplus user’s guide. Sixth edition. Los Angeles, CA: Author.

32.

Muthén

L. K.

Muthén

B. O.

(2005). Chi-square difference testing using the Satorra-Bentler scaled chi-square. Retrieved from www.statmodel.com/chidiff.shtml

33.

Paton

Crouch

Carnic

(2009). Young offenders’ experiences of traumatic life events: A qualitative investigation. Clinical Child Psychology & Psychiatry, 14(1), 43-62.

34.

Preacher

K. J.

Zyphur

M. J.

Zhang

(2010). A general multilevel SEM framework for assessing multilevel mediation. Psychological Methods, 15, 209-233.

35.

Quimette

P. C.

Kimerling

Shaw

Moos

R. H.

(2000). Physical and sexual abuse among women and men with substance use disorders. Alcoholism Treatment Quarterly, 18(3), 7-17.

36.

Regoeczi

W. C.

(2000). Adolescent violent victimization and offending: Assessing the extent of the link. Canadian Journal of Criminology, 42, 493-505.

37.

Rempel

Farole

(2011). Analytic strategy for producing unbiased estimates of drug court impact. In Rossman

S. B.

Roman

J. K.

Zweig

J. M.

Rempel

Lindquist

C. H.

(Eds.), The Multi-site Adult Drug Court Evaluation: The Impact of Drug Courts (Vol. 4, Appendix A, pp. 267-305). Washington, DC: Urban Institute.

38.

Roberts

G. L.

Williams

G. M.

Lawrence

J. M.

Raphael

(1999). How does domestic violence affect women’s mental health? Women & Health, 28(1), 117-129.

39.

Rossman

Roman

J. K.

Zweig

Rempel

Lindquist

(Eds.). (2011). The multi-site adult drug court evaluation. Washington, DC: Urban Institute.

40.

Sackett

L. A.

Saunders

D. G.

(1999). The impact of different forms of psychological abuse on battered women. Violence and Victims, 14(1), 105-117.

41.

Satorra

Bentler

P. M.

(1999). A scaled difference chi-square test statistic for moment structure analysis. Retrieved from http://preprints.stat.ucla.edu/260/chisquare.pdf

42.

Schnurr

P. P.

Lunney

C. A.

Sengupta

(2004). Risk factors for the development versus maintenance of posttraumatic stress disorder. Journal of Traumatic Stress, 17(2), 85-95.

43.

Schreck

C. J.

Stewart

E. A.

Osgood

D. W.

(2008). A reappraisal of the overlap of violent offenders and victims. Criminology, 46, 871-906.

44.

Siegel

J. M.

Golding

J. M.

Stein

J. A.

Burnam

M. A.

Sorenson

S. B.

(1990). Reactions to sexual activity: A community study. Journal of Interpersonal Violence, 5, 229-246.

45.

Stark

Flitcraft

A. H.

(1991). Spouse abuse. In Rosenberg

M. L.

Fenley

M. A.

(Eds.), Violence in America: A public health approach (pp. 123-157). New York, NY: Oxford University Press.

46.

Stalans

L. J.

(2009). Women’s offending behavior: Evidence-based review of gender differences and gender responsive programs. Victims & Offenders, 4, 405-411.

47.

Strickland

S. M.

(2008). Female sex offenders. Journal of Interpersonal Violence, 23, 474-489.

48.

Struckman-Johnson

Rucker

Bumby

Donaldson

(1996). Sexual coercion reported by men and women in prison. Journal of Sex Research, 33(1), 67-76.

49.

Substance Abuse and Mental Health Services Administration. (2010). Results from the 2009 national survey on drug use and health: Volume I Summary of national findings (Office of Applied Studies, NSDUH Series H-381, HHS Publication No. SMA-10-4586Findings). Rockville, MD: Author.

50.

Substance Abuse and Mental Health Services Administration. (2004). Results from the 2003 national survey on drug use and health: National findings (Office of Applied Studies, NSDUH Series H-25, HHS Publication No. SMA-04-3964). Rockville, MD: Author.

51.

Wolff

Shi

(2009a). Type, source, and patterns of physical victimization: A comparison of male and female inmates. Prison Journal, 89, 172-191.

52.

Wolff

Shi

(2009b). Contextualization of physical and sexual assault in male prisons: Incidents and their aftermath. Journal of Correctional Health Care, 15(1), 58-77.

53.

Zweig

J. M.

(1997). A longitudinal examination of the risk factors and consequences of sexual victimization for rural young adult women (Doctoral dissertation). The Pennsylvania State University, University Park.

54.

Zweig

J. M.

Barber

B. L.

Eccles

J. S.

(1997). Sexual coercion and well-being in young adulthood: Comparisons by gender and college status. Journal of Interpersonal Violence, 12, 291-308.

55.

Zweig

J. M.

Crockett

L. J.

Sayer

Vicary

J. R.

(1999). A longitudinal examination of the consequences of sexual victimization for rural young adult women. Journal of Sex Research, 36, 396-409.

56.

Zweig

J. M.

Yahner

Visher

C. A.

Lattimore

P. K.

(manuscript submitted for publication). Using general strain theory to explore the effects of prison victimization experiences on later offending and substance use.